Noninvasive assessment of coronary atherosclerotic plaque may be useful for risk stratification and treatment of atherosclerosis.
Materials and Methods
We studied 47 patients to investigate the accuracy of coronary plaque volume measurement acquired with 64-slice multi-slice computed tomography (MSCT), using newly developed quantification software, when compared with quantitative intracoronary ultrasound (QCU). Quantitative MSCT coronary angiography (QMSCT-CA) was performed to determine plaque volume for a matched region of interest (regional plaque burden) and in significant plaque defined as a plaque with ≥50% area obstruction in QCU, and compared with QCU. Dataset with image blurring and heavy calcification were excluded from analysis.
In 100 comparable regions of interest, regional plaque burden was highly correlated (coefficient r = 0.96; P < 0.001) between QCU and QMSCT-CA, but QMSCT-CA overestimated the plaque burden by a mean difference of 7 ± 33 mm3 (P = 0.03). In 76 significant plaques detected within the regions of interest, plaque volume determined by QMSCT-CA was highly correlated (r = 0.98; P < 0.001) with a slight underestimation of 2 ± 17 mm3 (P = not significant) when compared with QCU. Calcified and mixed plaque volume was slightly overestimated by 4 ± 19 mm3 (P = ns) and noncalcified plaque volume was significantly underestimated by 9 ± 11 mm3 (P < 0.001) with QMSCT-CA. Overall, the limits of agreement for plaque burden/volume measurement between QCU and QMSCT-CA were relatively large. Reproducibility for the measurements of regional plaque burden with QMSCT-CA was good, with a mean intraobserver and interobserver variability of 0% ± 16% and 4% ± 24%, respectively.
Quantification of coronary plaque within selected proximal or middle coronary segments without image blurring and heavy calcification with 64-slice CT was moderately accurate with respect to intravascular ultrasound and demonstrated good reproducibility. Further improvement in CT resolution is required for more reliable measurement of coronary plaques using quantification software.
Since its introduction, the inability to accurately quantify the degree of stenosis as well as plaque volume has been one of the major limitations of non-invasive coronary angiography with Multi-Slice Computed Tomography (MSCTA). Previous investigations have attempted semi-quantitative approaches but failed to demonstrate adequate correlation with quantitative coronary angiography. Moreover, wide variation between individual measurements was reported, indicating high observer variability (1;2). Possibly a more automated approach may result in more accurate and importantly, reproducible results. In the current study therefore, recently developed dedicated quantification software was evaluated against intravascular ultrasound (IVUS)(3).
In a total of 47 patients MSCTA by means of 64-slice MSCT was performed as part of a larger ongoing protocol. In addition, IVUS was performed in preferably all 3 vessels. A region of interest (ROI) was defined on the MSCT images within the proximal and middle segments of the coronary arteries. Subsequently, the ROI was matched to the IVUS data using anatomical landmarks. Of note, segments with stents or extensive calcium were excluded. On IVUS, the length of the ROI, lumen volume, vessel volume and plaque volume were derived.
An independent observer evaluated all MSCT images using dedicated software. On orthogonal L-mode views lumen-plaque border and outer vessel border were manually traced, followed by manual adjustment of the borders on the corresponding cross-sectional views. Subsequently, the same parameters as for IVUS were derived. In addition, intra- and interobserver variability were assessed.
After exclusion of ROIs with suboptimal image quality, 100 ROI remained for analysis. No significant difference was observed for the mean length of the ROI between MSCT and IVUS. Strong correlations were observed between the 2 techniques for lumen volume, vessel volume and plaque volume (Pearson’s correlation coefficients of respectively 0.98, 0.98 and 0.96, P<0.001). Nevertheless, lumen volume was significantly underestimated by quantitative MSCT, resulting in a significant overestimation of plaque volume. Importantly, excellent inter- and intraobserver reproducibility was noted.
Accordingly, the authors conclude that quantification of coronary plaque volume using dedicated software for MSCT is feasible. However, one should realize that the current observations were obtained in highly selected regions of interest. Indeed, the limited spatial resolution of MSCT remains an important drawback of the technique and is unlikely to equal invasive techniques. Future studies should demonstrate whether reliable quantification may also become feasible in smaller as well as more heavily calcified segments. Nevertheless, the current study is an important step towards quantitative MSCTA.