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

Area of Science:

• Cardiovascular Imaging
• Radiomics
• Biomarker Discovery

Background:

• Pericoronary adipose tissue (PCAT) is a known imaging biomarker for coronary inflammation.
• The specific influence of low-density lipoprotein cholesterol (LDL-C) on PCAT characteristics remains incompletely understood.
• Coronary computed tomography angiography (CCTA) is a key imaging modality for cardiovascular assessment.

Purpose of the Study:

• To investigate the association between PCAT radiomic features and serum LDL-C levels.
• To utilize CCTA-derived radiomics for quantifying PCAT alterations related to dyslipidemia.

Main Methods:

• Retrospective analysis of 150 patients undergoing CCTA, stratified by LDL-C levels (≥3.4 mmol/L vs. <3.4 mmol/L).
• Extraction of 288 radiomic features from PCAT around major coronary arteries.
• Statistical analysis including Wilcoxon rank-sum test, logistic regression, Pearson correlation, and validation with a gradient boosting machine (GBM) and SHAP analysis.

Main Results:

• Eleven radiomic features, including first-order and texture-based metrics, showed significant association with elevated LDL-C (p < 0.05).
• The GLSZM.LCXLargeAreaHighGrayLevelEmphasis feature exhibited the strongest correlation (Mantel's r ≈ 0.15, p < 0.01).
• The GBM model achieved high performance (AUROC 0.889 training, 0.724 testing), with first-order energy and large-area high gray-level features identified as key discriminators.

Conclusions:

• Elevated LDL-C levels are linked to increased spatial heterogeneity and specific gray-level clustering within PCAT.
• CCTA-based radiomics provides imaging evidence supporting the relationship between LDL-C and PCAT characteristics.
• These findings may enhance the understanding of PCAT as a biomarker in the context of hypercholesterolemia.