Coronary Artery Stenosis Quantification Using Multislice Computed Tomography

Abstract

Reliable noninvasive detection of stenoses with multislice computed tomography (MSCT) is feasible. This study's aim was to analyze the agreement, correlation, and reliability of MSCT with conventional coronary angiography as the reference standard for quantification of coronary artery stenoses. A total of 118 significant (at least 50%) coronary artery stenoses with a reference vessel diameter of at least 1.5 mm in 62 patients were analyzed by MSCT using 16 detector rows (Aquilion, Toshiba, Otawara, Japan), multisegment reconstruction, and voxel sizes of 0.35x0.35x0.5 mm. The degree of stenosis on MSCT and quantitative coronary angiography (QCA) was measured by correlating the difference between the reference vessel diameter (average of 2 measurements directly proximal and distal to the stenosis) and the stenotic vessel diameter to the reference vessel diameter. Correlation between the percent diameter stenosis determined by MSCT (78.2+/-13.6%) and QCA (76.0+/-14.8%) was significant (P<0.001) but only moderately so (R=0.51). Bland-Altman analysis revealed no systematic under- or overestimation with MSCT but large limits of agreements (+/-27.6%). Also the limits of agreement for interobserver agreement (reliability) of MSCT data were considerably large (+/-24.8%). Among the 27 coronary artery stenoses with a reference diameter of at least 3.5 mm, there was improved correlation (R=0.80) and the limits of agreement between MSCT and QCA were significantly smaller (+/-17.3%, P<0.008). The agreement between MSCT and QCA was not significantly different for stenoses with no calcification or only calcium spots (+/-28.2%) as compared with those with moderate-or-severe calcifications (+/-27.3%; P=0.8). MSCT allowed correct classification of coronary stenoses into low-grade (below 75%) and high-grade stenoses (at least 75%), in 62% (73 of 118). The accuracy and reliability of coronary artery stenosis quantification with MSCT using isotropic voxel sizes and multisegment reconstruction is still too low to recommend routine clinical application because of rather low agreement, correlation, and reliability. Despite these limitations, the current results demonstrate the potential of MSCT for reliable and accurate quantification of coronary artery stenoses in the near future provided that further improvements in spatial and temporal resolution will be achieved.