Contribution of Risk Factors to the Development of Coronary Atherosclerosis as Confirmed via Coronary CT Angiography: A Longitudinal Radiomics-based Study

Abstract

Background Various cardiovascular risk factors are thought to modify atherosclerosis in a similar fashion (ie, by increasing the magnitude of coronary artery disease [CAD]). However, coronary CT angiography allows precision phenotyping of plaque characteristics through use of radiomics. Purpose To assess whether different cardiovascular risk factors have distinctive contributions to the changes in plaque morphologic features over time. Materials and Methods Individuals with or without HIV infection and cocaine use and without cardiovascular symptoms underwent coronary CT angiography between May 2004 and August 2015. In the current HIPAA-compliant study, the effects of cocaine use, HIV infection, and atherosclerotic cardiovascular disease (ASCVD) risk on the temporal changes (mean ± standard deviation, 4.0 years ± 2.3 between CT angiographic examinations) in CAD structure were analyzed by using radiomic analysis. The changes in radiomic features were analyzed by using linear mixed models, with correction for factors that may change plaque structure: high-sensitivity C-reactive protein level, statin use, positive family history of CAD, and total plaque volume to account for any potential intrinsic correlation between volume and morphologic features. Clusters among significant radiomic features were identified by using hierarchical clustering. Bonferroni-corrected P values less than .00004 (.05 divided by 1276) were considered to indicate significant differences. Results Of 1429 participants, 300 with CAD confirmed at coronary CT angiography were randomly selected (mean age, 48 years ± 7; 210 men, 226 people infected with HIV, 174 people who use cocaine) and 1276 radiomic features were quantified for each plaque. Cocaine use was significantly associated with 23.7% (303 of 1276) of the radiomic features, HIV infection was significantly associated with 1.3% (17 of 1276), and elevated ASCVD risk was significantly associated with 8.2% (104 of 1276) (P < .00004 for all). Parameters associated with elevated ASCVD risk or cocaine use and HIV infection did not overlap. There were 13 clusters among the 409 parameters, eight of which were affected only by cocaine use and three of which were affected only by ASCVD risk. Conclusion Radiomics-based precision phenotyping indicated that conventional risk factors, cocaine use, and HIV infection each had different effects on CT angiographic morphologic changes in coronary atherosclerosis over 4 years. © RSNA, 2021 Online supplemental material is available for this article. See also the editorial by Schoepf and Emrich in this issue.

Graphical abstract

Figure 1:

Manhattan plots of P values for associations between cocaine use, HIV infection, and elevated atherosclerotic cardiovascular disease (ASCVD) risk and each radiomic parameter. A–C, P values for univariate associations between each radiomic feature and cocaine use, HIV infection, and elevated ASCVD risk in univariate models. D–F, P values for associations between each radiomic feature and cocaine use, HIV infection, and elevated ASCVD risk in multivariate models corrected for high-sensitivity C-reactive protein level as the most common marker of inflammation, positive family history of coronary artery disease (CAD) as an indicator of potential genetic predisposition for CAD progression, statin use because it is known to modify the composition and development of coronary plaques, and the plaque volume itself because we wished to correct for any potential intrinsic correlation between volume and morphologic characteristic. Radiomic parameters are situated on the x-axis in the same order of each subplot, and the corresponding P values are located on the y-saxis. Points above the red line (P = .00004) indicate radiomic features in which the given predictor showed a significant association. There was no overlap between radiomic features associated with cocaine use or elevated ASCVD risk, potentially implying different pathways of plaque progression.

Figure 2:

A, Hierarchical cluster dendrogram of radiomic features significantly associated with cocaine use, HIV infection, and/or elevated atherosclerotic cardiovascular disease (ASCVD) risk. Clusters are color-coded depending on risk factor with which features were associated. B, Heatmap of R² values for linear regressions between each pair of significant radiomic features (n = 409). Elements of the heatmap are color coded depending on risk factor with which features were associated. Clusters are outlined in yellow. C, Corresponding P values for cocaine use, HIV infection, and increased ASCVD for each radiomic feature. Features are reordered based on hierarchical clustering to correspond to the dendrogram. Bars extending farther than the red line (P = .00004) indicate significant associations. Results from hierarchical clustering indicate that there are distinct morphologic feature sets that are associated with only specific risk factors. Furthermore, P values for cocaine use among clusters associated with cocaine use were magnitudes lower than for HIV infection and especially elevated ASCVD risk. In addition, P values for elevated ASCVD risk for the three clusters containing only radiomic features associated with elevated ASCVD risk were magnitudes lower than for cocaine use in HIV infection. These results potentially imply distinct pathways of coronary atherosclerosis progression because modifying effects of cocaine use and conventional cardiovascular risk factors are clearly separable.

Figure 3:

Corresponding P values for associations between cocaine use, HIV infection, elevated atherosclerotic cardiovascular disease (ASCVD) risk and the significant radiomic features stratified by sex and age. A, B, Corresponding P values for associations between cocaine use, HIV infection, and increased ASCVD for each radiomic feature stratified by sex. C, D, Corresponding P values for associations between risk factors and each significant radiomic feature stratified by age based on the median age of 51 years. The features are reordered according to hierarchical clustering. Bars extending farther than the red line (P = .00004) indicate significant associations. The sex-based results indicate sex-specific contributions of the different risk factors on coronary atherosclerosis morphologic features. Furthermore, age stratification indicates that different risk factors may have different contributions to atherosclerosis depending on the individual’s age.

Figure 4:

Bar chart of P values for radiomic features affected by cocaine use or elevated atherosclerotic cardiovascular disease (ASCVD) risk stratified by disease subgroups. A, Corresponding P values for cocaine use among the total population, elevated ASCVD risk, and low ASCVD risk subgroups, shown in different shades of blue. B, Corresponding P values for ASCVD risk among the total population and among cocaine user and nonuser subgroups, shown in different shades of green. Bars above the red line (P = .00004) indicate significant associations. Results indicate that modifying effects of cocaine use may require a susceptible environment (increased ASCVD risk) to occur. However, once it is present, it modifies the morphologic features of atherosclerosis differently than ASCVD risk. In addition, ASCVD risk may have a more profound effect among cocaine nonusers, which may imply that the effects of cocaine use on morphologic changes overwhelm the effects of ASCVD.