Understanding plaque formation in patients at risk for coronary artery disease—the leading cause of morbidity and death in the world—enables physicians to better determine whether and how to treat these individuals. We used computed tomographic angiography to quantitatively evaluate the progression of nonculprit coronary plaques along the full length of the right coronary artery in 21 patients with acute coronary syndrome. Each right coronary artery was analyzed in sequential, 3-mm-long segments, and the minimum luminal area, plaque burden, and plaque volume within each segment were evaluated at baseline and at 12-month follow-up. Serial remodeling of the right coronary artery was also evaluated. In total, 625 arterial segments were analyzed. At 12-month follow-up, the plaque burden had increased slightly by 0.34% (interquartile range [IQR], −4.32% to 6.35%; P=0.02), and the plaque volume was not significantly changed (0.33 mm3; IQR, −3.05 to 3.54; P=0.213). The minimum luminal area decreased 0.05 mm2 (IQR, −1.33 to 0.87 mm2; P=0.012), and this was accompanied by vessel reduction, as evidenced by negative remodeling in 43% of the 625 segments. We conclude that serial computed tomographic angiography can be used to quantitatively evaluate the morphologic progression of coronary plaques.
Coronary artery disease (CAD), which causes more than 20 million deaths each year, is the world's leading cause of morbidity and death.1,2 Most patients with CAD have many coronary obstructions that are in varying degrees of progression and that pose different risks.3–6 Therefore, understanding the progression of coronary plaque would help physicians make more informed decisions about whether and how to treat patients at risk for CAD.
In clinical studies, intravascular ultrasound (IVUS) has traditionally been used to investigate the composition and progression of coronary plaques in the proximal or mid segments of one or 2 major coronary vessels.7,8 However, it is not appropriate for routine serial evaluations because it carries a small risk of severe procedural complications. In contrast, computed tomographic angiography (CTA) is less expensive and noninvasive. Advances in computed tomographic (CT) technology and post-image-processing techniques have made it possible not only to detect stenosis with coronary CTA, but also to use it to quantitatively analyze coronary plaques.9–13 In recent years, a few serial studies with CTA have appeared in the literature,14–17 but most of them have focused on lesions in diseased segments of the coronary tree. Our objective in this study was to use CTA to examine the natural history of coronary atherosclerosis along the full length of the right coronary artery (RCA) tree and to investigate serial changes in the minimum luminal area (MLA), plaque burden (PB), plaque volume (PV), and arterial remodeling.
Patients and Methods
Twenty-one patients treated at Beijing Anzhen Hospital from 2010 through 2013 were studied retrospectively (Table I). Inclusion criteria were as follows: age >18 years, availability of 2 coronary CTA scans taken >6 months apart, and diagnosis of acute coronary syndrome (ACS) during the initial physical examination. Exclusion criteria included previous percutaneous coronary intervention (PCI) to the RCA or coronary artery bypass grafting, severe valvular heart disease, irregular heart rhythm, renal failure, and severe hematologic disease. The institutional review board of our hospital approved the study, and all patients gave informed consent before participation.
Image Acquisition and Quantitative Measurement
In accordance with the coronary CTA Guide provided by the Society of Cardiovascular Computed Tomography,18 coronary CTAs of the RCA tree were obtained at baseline and follow-up, with either a Somatom® Definition 64-slice dual-source CT (Siemens Healthcare GmbH; Erlangen, Germany) or an Aquilion One™ 320-slice CT (Toshiba Medical Systems Corporation; Tochigi, Japan). Before examination, all patients were given nitroglycerin sublingually. Patients with a heart rate >65 beats/min were given β-blockers. During scanning, a 70- to 90-mL bolus of intravenous contrast agent was injected. The scanning settings included collimations of 0.625 mm for 64-slice CT and 0.5 to 0.75 mm for 320-slice CT, a pitch factor of 0.2 to 0.26, a reconstruction slice thickness of 0.4 mm, a tube voltage of 120 kV, and a tube current of 300 to 650 mAs. During inspection, slice thicknesses of 0.5 to 0.75 mm—measured in 0.3- to 0.4-mm increments—and moderately smooth convolution kernels were used to rebuild axial images. Scanning was performed in the middle or at the end of diastole, when coronary artery movement was relatively slow.
Quantitative measurement and analysis of the coronary CTAs were conducted by an experienced specialist who used OsiriX 6.0 image-processing software (Pixmeo SARL; Bernex, Switzerland). To ensure the reliability of the results, we used a blinding method to hide the time at which each CTA was obtained. During analysis, a centerline originating from the ostium was obtained manually and was used as the reference for generating curved and straightened multiplanar reformatted images. Then, starting from the ostium, contiguous, cross-sectional reconstructions of the RCA tree were rendered in 1-mm-thick slices. Finally, the luminal area and vessel area of all cross-sectional reconstructions were measured manually (Fig. 1). To ensure accuracy in the detection of plaque and the outer vessel boundary, we set the window width (WW) to 155% and the window level (WL) to 65% of the average intensity within the lumen, as previous studies have described.11,19
Citation: Texas Heart Institute Journal 44, 5; 10.14503/THIJ-16-5805
Vessel Segmentation and Blinding
To efficiently evaluate the progression of coronary plaque at baseline and at 12-month follow-up, we divided the RCA into consecutive 3-mm-long segments, starting at the ostium. We chose 3 mm because the length was methodologically reliable, and the homogeneity of the morphologic properties of the local vascular plaque across this distance was good. Accordingly, we would be able to measure changes in the local vascular plaque as accurately as possible.20,21
To ensure that the arterial segments to be measured were identical at baseline and follow-up, and to maintain blinding of the segmentation methods, a specialist who did not participate in the subsequent data analysis segmented the coronary artery tree by using fixed anatomic landmarks—primarily multiple arterial branches—as fiducial points.
During segmentation, we excluded side branches and 1-mm segments adjacent to bifurcation of the main branches. Scans of segments with motion artifacts were also excluded.
Quantitative Measurement of Coronary Plaque
Minimum luminal area (MLA), plaque burden (PB), and plaque volume (PV) for each 3-mm segment were measured, as follows22:
and
where n is the number of cross-sections per segment (n = 4 in the current study).23 Figure 2 shows an example of these calculations at baseline and follow-up.
Citation: Texas Heart Institute Journal 44, 5; 10.14503/THIJ-16-5805
Coronary Remodeling. As recommended for longitudinal studies,22,24 remodeling was analyzed on the basis of changes in vessel area from baseline to follow-up. Positive remodeling was defined as an increase in vessel area; absence of remodeling, no substantial change in vessel area; and negative remodeling, a decrease in vessel area (Fig. 3). Positive remodeling was further classified as expansive (Δ vessel area / Δ plaque area >1) or incomplete (Δ vessel area / Δ plaque area = 0 to 1).
Citation: Texas Heart Institute Journal 44, 5; 10.14503/THIJ-16-5805
Statistical Analysis
Categorical variables are presented as number or percentage. Continuous variables are presented as mean ± SD or as median and interquartile range (IQR) (Table II). Continuous variables were analyzed by using the Kolmogorov-Smirnov test. Baseline and follow-up measurements for MLA, PB, and PV were compared by using the paired t test or the Wilcoxon signed-rank test, as appropriate. To evaluate interobserver and intraobserver variability, a subset of 8 coronary arteries was randomly selected and analyzed twice by the same observer, with a delay of one month, and once by a second observer. Intraobserver and interobserver variability were evaluated by intraclass correlation coefficients (ICC) and the Bland-Altman test. Statistical analyses were performed with use of SPSS 21.0 (IBM Corporation; Endicott, NY) and Stata® 13 (StataCorp LP; College Station, Texas). A P value <0.05 was considered statistically significant.
Results
We analyzed 625 RCA segments, each 3 mm long, from 21 patients with ACS. The median time between the baseline and follow-up scans was 12 months (IQR, 9–20 mo). Of the 21 patients studied, 8 had undergone at least one stent implantation in the left coronary artery (LCA) before baseline measurement, and 13 underwent stent implantation in the LCA a few days after baseline CTA imaging. No patient underwent stent implantation in the RCA after baseline CTA imaging. After discharge from the hospital, all patients were treated with secondary prevention methods, such as antiplatelet, lipid-lowering, and antihypertension medications.
Interobserver and Intraobserver Variability
We calculated interobserver variability for 215 randomly selected segments (Fig. 4). For MLA, the ICC was 0.96, and the Bland-Altman bias was −10% (95% limits of agreement [LOA], −42.8% to 22.8%). For PB, the ICC was 0.71, and the Bland-Altman bias was 5.6% (95% LOA, −13.4% to 24.7%). For PV, the ICC was 0.85, and the Bland-Altman bias was 3.2% (95% LOA, −38.8% to 45.1%).
Citation: Texas Heart Institute Journal 44, 5; 10.14503/THIJ-16-5805
Intraobserver variability indicated excellent correlation with good limits of agreement. For MLA, the ICC was 0.99, and the Bland-Altman bias was 0 (95% LOA, −6% to 6%). For PB, the ICC was 0.96, and the Bland-Altman bias was −2.5% (95% LOA, −7.5% to 2.5%). For PV, the ICC was 0.98, and the Bland-Altman bias was −4.6% (95% LOA, −12.2% to 3.1%).
Quantitative Analysis of Coronary CTA Images
Between baseline and follow-up, the median MLA decreased 0.05 mm2 (IQR, −1.33 to 0.87 mm2; P=0.012); the median PB increased 0.34% (IQR, −4.32% to 6.35%; P=0.02); and there was no change in PV (0.33 mm3 (−3.05 to 3.54 mm3); P=0.213) (Table II). Table III shows the coronary remodeling patterns at follow-up. Among the 625 segments, 40% showed positive remodeling, 43% showed negative remodeling, and 17% showed no substantial change.
Discussion
This study shows the feasibility of using CTA to quantitatively evaluate the progression of nonculprit coronary plaque along the RCA in patients with ACS. The main finding was that, in patients who underwent PCI to the LCA and continued to be treated with conventional therapy after discharge from the hospital, the MLA of the RCA was reduced, whereas there was a slight increase in PB during the 12-month follow-up period. Our results also show that CTA yields good intra- and interobserver variability for MLA, PB, and PV.
When CTA is used for quantitative analysis, WW and WL settings can affect the measured values of vessel and luminal areas. Investigators have shown that setting WW and WL based on the mean luminal density is better than using a fixed threshold,25 and that CT provides optimal matching with IVUS in the quantitative measurement of plaques when appropriate WW and WL settings are used.11,19 On the basis of these findings and the results of our preliminary tests, we set the WW to 155% and the WL to 65% of the average luminal density. In addition, we found that the ICC was better for intraobserver variability than for interobserver variability; this was caused by differences between the 2 observers in the estimation of the mean luminal density.
The segmentation method is very important in evaluating plaque progression within the RCA tree. Until recently, most studies focused on only a small segment of the coronary tree or a specific subset of lesions.9–13 In 2012, Papadopoulou and colleagues9 evaluated plaque progression in the entire coronary tree, which they divided into 17 segments on the basis of the coronary CTA analysis guidelines provided by the Society of Cardiovascular Computed Tomography Guidelines Committee.18 However, when long segments are used, characterizing the morphology of the coronary arteries and the parameters we examined in our study—MLA, PB, and PV—is difficult. Dividing the arteries into shorter, 3-mm segments enables us to obtain a better morphologic representation.
The use of CTA to evaluate coronary plaque progression is likely to increase, particularly with further improvements in hardware and software. Whereas IVUS is considered the gold standard for evaluating plaque progression, it is invasive, and it cannot measure severe stenosis (>90%) in segments and distal small vessels with a diameter >2 mm). Previous studies comparing IVUS and CTA support the feasibility of using the latter to evaluate atherosclerotic plaque size, PB, remodeling, eccentricity, and plaque progression.14,21,26,27 To ensure the accuracy of the evaluation, we have found that adhering to a standard CTA protocol at baseline and at follow-up is necessary; preferably, the studies should be performed on the same scanner. Moreover, standards for evaluating plaque are essential to reduce the difference between intraobserver and interobserver variability.
Our study has some limitations. First, it was a single-center trial with a small sample size of only 21 patients. Therefore, multivariate analysis to evaluate the impact of baseline variables on plaque progression at follow-up was not possible. Second, because this was a retrospective pilot study, identifying patients who had undergone CTA at 2 different time points was difficult. Of 1,028 patients with initial CTA images, only 21 were eligible; thus, this study did not include a comparison group with stable angina pectoris. Third, although the evaluation of coronary plaque composition is beneficial in predicting acute coronary events, we did not take it into consideration because the accuracy of CTA for this purpose remains questionable. Finally, we used a manual method for the quantitative measurement of CTA images. Because this process takes a long time and requires experienced clinicians, using it in large-scale studies or the clinic is not practical.
Despite these limitations, we think that CTA is a feasible method of quantitatively evaluating coronary plaque progression, provided that appropriate WW and WL settings are used.
Computed tomographic angiograms (cross-sectional reconstruction images) show quantitative measurements of A, B, C) luminal areas (no-color look-up table mode) and D, E, F) the corresponding vessel areas (GE color mode) perpendicular to the centerline of the coronary artery.
Example of the vessel analysis protocol. Straightened multiplanar reformatted (MPR) images from the right coronary artery A) at baseline and B) at 12-month follow-up show a 3-mm segment, divided into 4 sequential cross-sections. Starting from the ostium, the vessel area and luminal area of all contiguous, 1-mm-thick cross-sectional reconstructions of the RCA tree were measured (area graphs). Cross-sections of the 3-mm segments are shown C, D, E, F) at baseline and G, H, I, J) at follow-up. The luminal area and vessel area of each cross-section were measured directly. The plaque area was calculated by subtracting the luminal area from the vessel area, and the plaque burden was calculated as the sum of all 4 contiguous plaque areas, divided by the total vessel area. In the segment shown, the plaque burden increased from baseline to follow-up.
MLA = minimum luminal area
Illustration shows coronary remodeling patterns. Incomplete or expansive changes were considered to be positive remodeling.
Bland-Altman plots show A, B, C) interobserver and D, E, F) intraobserver variability for the A, D) minimum luminal area (MLA), B, E) plaque burden, and C, F) plaque volume.
Contributor Notes
From: Departments of Biomedical Engineering (Drs. He, Lin, X. Liu, Shu, Wu, and Xu), Cardiology (Dr. Y. Liu), and Radiology (Dr. Zhang), Beijing Anzhen Hospital, Capital Medical University, Beijing 100029, People's Republic of China
This study was supported by the National Science Foundation of China (81670371) and the Capital Public Health Project (Z161100000116086).
