Computed Tomography-Based Radiomics Provides New Insights Into Associations Between Pericoronary Fat Characteristics...

Pericoronary adipose tissue (PCAT) is an established imaging biomarker of coronary inflammation; however, the influence of low-density lipoprotein cholesterol (LDL-C) remains unclear. This study aimed to explore the associations between PCAT and LDL-C using coronary computed tomography angiography (CCTA)-based radiomics.

This retrospective study stratified 150 patients undergoing CCTA into two groups according to serum LDL-C levels (≥3.4 mmol/L vs. <3.4 mmol/L). A total of 288 radiomic features were extracted from the PCAT surrounding the left anterior descending artery, left circumflex artery, and right coronary artery. After the initial filtering using the Wilcoxon rank-sum test, univariate logistic regression and Pearson correlation analyses were applied to identify features associated with elevated LDL-C levels. Key features were further validated using a gradient boosting machine (GBM) ensemble model combined with Shapley Additive Explanations (SHAP) analysis.

A total of 11 radiomic features were significantly associated with elevated LDL-C levels (p < 0.05), including both first-order and texture-based features. Mantel correlation analysis revealed that the gray level size zone matrix (GLSZM)-derived feature, GLSZM.LCXLargeAreaHighGrayLevelEmphasis, demonstrated the strongest association (Mantel's r ≈ 0.15; p < 0.01). The GBM model achieved the best performance, with an area under the receiver operating characteristic curve (AUROC) of 0.889 in the training set and 0.724 in an internal hold-out test set. SHAP analysis identified first-order energy and large-area high gray-level features as the most important contributors to the discrimination of high LDL-C status.

Elevated LDL-C levels are significantly associated with increased spatial heterogeneity and high gray-level clustering in PCAT, thereby providing imaging-based evidence supporting the association between LDL-C and PCAT.