Exploring pericoronary adipose tissue differences between calcified and non-calcified plaques on coronary CT angiography with fat attenuation index and radiomics analysis

Abstract

Abstract Funding Acknowledgements Type of funding sources: Public grant(s) – National budget only. Main funding source(s): Ministry of Science and Technology. Introduction Coronary CT angiography (CCTA)-derived pericoronary adipose tissue (PCAT) is a novel marker for coronary inflammation. The relationship between the extravascular PCAT and the vascular lesion itself is particularly important to establish its use in monitoring the disease status. Past studies investigated lesion-level PCAT differences in vessels with acute myocardial infarction. However, lesion-level PCAT differences between calcified and non-calcified plaques in patients with stable coronary artery disease were not well understood. Purpose To investigate the differences between PCAT adjacent to calcified and PCAT adjacent to non-calcified plaques using fat attenuation index (FAI) and radiomics analysis. Material and Methods In a cohort of 100 consecutive patients who underwent CCTA, 446 plaques were identified in 89 patients. Among these plaques, 213 were calcified plaques and 49 were non-calcified plaques, with the rest classified as mixed by our expert cardiologist and radiologist. The PCAT of these 262 plaques were then extracted. Following prior literature, adipose tissue was defined as voxels with attenuation ranging −190 to −30 HU. PCAT was extracted within 3 mm distance from the vessel wall, along the middle 10 mm of each plaque. FAI was defined as the mean attenuation of the PCAT volume. For radiomics analysis, we used PyRadiomics to calculate the radiomic features and the random forest classifier from Scikit-Learn to distinguish PCAT volumes of calcified and non-calcified plaques. Patient-level 5-fold cross validation was used to develop the classifier. Result FAI were higher in PCAT adjacent to non-calcified plaques compared to PCAT adjacent to calcified plaques (−81.4 [IQR: −91.8 to −73.5] vs −86.7 [IQR: −94.1 to −78.4], p = 0.008). The classifier using radiomics features achieved better discrimination (AUROC=0.738 [0.715–0.761]) than FAI alone (AUROC=0.607 [0.605–0.609]). Conclusion Significant differences in PCAT characteristics were found between calcified and non-calcified plaques either with direct comparison of FAI or radiomics analysis. Radiomics analysis revealed further discrimination between PCAT characteristics other than FAI may exist between calcified and non-calcified plaques. It is also important to note that the FAI of calcified plaques could be affected by beam hardening. Further investigation is needed to explore the underlying cause of such differences and their clinical implications.