We compared MDCT images with IB-IVUS images and determined cutoff values of HU for discriminating lipid pool from fibrosis, and fibrosis from calcification. We showed that lipid volume measured by MDCT was moderately correlated with that measured by IB-IVUS, whereas fibrous volume was not.
Comparison between IB values and Hounsfield density
Previous IVUS studies showed that MDCT could accurately characterize the tissue components of coronary plaques by comparison with gray scale IVUS findings [7, 19–23]. These studies demonstrated that hypoechoic lesions detected by conventional IVUS had lower HU than those of hyperechoic lesions. Other investigators reported that there was a significant difference in HU between hypoechoic and hyperechoic lesions whereas there was substantial overlap of HU between plaque types . These different results may be because the gray scale IVUS images were evaluated subjectively, and the results may have been influenced by interobserver variability. Another study demonstrated that the ability of MDCT for the discrimination between lipid-rich plaques from fibrous plaques was limited . However, the previous study was based on subjective classification (hypodense, isodense or hyperdense) and relatively low inter-observer agreement of MDCT (κ =0.61). In the present study, we performed an objective quantitative analysis and compared tissue components detected by MDCT with those determined by IB-IVUS. Harada et al. reported that 64-slice MDCT was a promising approach for detection of different types of coronary plaques, but MDCT overestimated low-attenuated plaque and was limited to the determination of low-attenuated plaque volume (r =0.328, p =0.18) . However, in that report, the cutoff (60 HU) between fibrous plaque and low-attenuated plaque was based on the value used in a previous study that was conducted by comparison with gray scale IVUS . In our study, a cutoff of 60 HU resulted in overestimation of low-attenuated plaque, whereas a cutoff of 50 HU, selected by comparison with IB-IVUS, was more accurate to determine the volume of lipid pool (r =0.66, p <0.001). Tanaka et al. reported that mean HU of coronary ruptured plaques was 46.8 HU that was similar to our cutoff value in the present study (50 HU), whereas that of non-ruptured plaques was 73.4 HU . They concluded that 64-slice MDCT might provide a useful tool for the non-invasive detection of plaque rupture. The results of the present study reinforced the previous findings.
Usefulness of MDCT for tissue characterization of coronary plaques
In the present study, relative fibrous volume measured by MDCT was not correlated with that measured by IB-IVUS, whereas lipid volume measured by MDCT was moderately correlated with that by IB-IVUS. The HU of each tissue component is influenced by its surrounding substances, such as intravascular contrast medium and pericardial fat [10, 25]. Lipid pool is generally surrounded by fibrous component, whereas fibrous tissue is surrounded by multiple substances such as pericardial fat or contrast medium . This difference in environment may have contributed to the lack of a correlation of fibrous volume.
Motoyama et al. reported that CT characteristics of plaques associated with ACS included low plaque density (<30 HU) . Kashiwagi et al. reported that mean HU of thin-cap fibroatheroma of coronary plaques was 35.1 HU . This value was similar to that of high risk plaques evaluated by Motoyama et al. In the present study, the cutoff for the discrimination between lipid pool and fibrosis was 50 HU. Taking the three reports into account, a CT value of 50HU was adequate for the differentiation of lipid pool from fibrosis, but not for discrimination between plaques that were associated with ACS and were not associated with ACS. There is the possibility that lower CT values (<30HU rather than 50 HU) indicate the unstable components such as necrotic core. However, IB-IVUS is not able to discriminate between lipid pool and necrotic core. Therefore, it might be difficult to compare the results of the present study with previous studies using histology.
Kitagawa et al. reported that the optimal cutoff values of CT density for predicting hypoechoic lesions evaluated by IVUS was 39 HU, whereas that for lipid pool in the present study was 50 HU . The previous study was performed by comparing hypoechoic and non-hypoechoic lesions that were determined subjectively. In addition, the comparison was performed using relatively large ROIs (one square millimeter), whereas small ROIs (0.5 mm x 0.5 mm) were used in the present study.
Previous studies reported that CT attenuation of coronary atherosclerotic plaques was influenced by intravascular contrast medium [29, 30]. CT values of coronary atherosclerotic plaques increased in proportion to the increment of CT values of the coronary lumen up to 250 HU. In contrast, the CT values of coronary atherosclerotic plaques were relatively stable when the values of the coronary lumen were 250–400 HU . The HU of the vessel lumen in the present study was stable (326 ± 55 HU). However, the CT values of the coronary lumen are important when HU is used to characterize tissue components of coronary plaques.
We previously reported that the relative volume of lipid pool could be determined using IB-IVUS imaging . However, IB-IVUS is invasive and can only be performed during coronary catheterization, whereas MDCT is minimally invasive and can be applied to the patients who were suspected as angina pectoris. The volume of lipid pool in coronary atherosclerotic lesions can be calculated using 3D color-coded maps constructed from MDCT images. MDCT volumetric analysis of lipid pool may be useful for clinical risk assessment and to provide incremental information on the effectiveness of medications. It was reported that prediction of sudden cardiac death using measurement of coronary calcification by MDCT are distinct methods of assessing risk for sudden cardiac death . Measurement of lipid pool by MDCT may be promising methods of assessing risk for coronary artery disease.
There are several limitations of the present study. First, calcification is a perfect reflector of ultrasound, causing acoustic shadowing that is typical in IVUS images. The ultrasound signals that are not able to penetrate or pass through the calcified layer are reflected back towards the transducer. Therefore, an accurate calculation of calcified area and volume is not possible using ultrasound. Likewise, a beam-hardening action by calcification that is called “partial volume effect”, hinders rigorous evaluation of the calcified volume by MDCT. Second, we excluded the area of the artifact due to the guidewire from the IB-IVUS analyses. Therefore, the guidewire artifact and calcification interfere with a rigorous calculation of the area and volume of each component. Comparison between IB values and HU of unstable plaques including thin cap fibroatheroma in patients with acute coronary syndrome is required. Third, square shaped ROI (0.5 x 0.5 mm) might be too large to enclose homogenous plaque components in human coronary plaques. Improvement in resolution of MDCT would be expected in the future. Finally, the IB-IVUS remains a research tool which does not have clinical utility. The use of IB values that was surrogate as the gold standard did not necessarily translate into an accurate analysis of plaque components.