Epicardial Adipose Tissue Shows No Significant Incremental Value for Discriminating Coronary Artery Disease in Adults Aged 80 Years and Older: A Machine Learning Analysis.

Background

Epicardial adipose tissue (EAT) has been associated with coronary artery disease (CAD) in middle-aged and mixed-age populations, but its value in adults aged 80 years and older remains uncertain. This study investigated whether computed tomography (CT)-derived EAT volume provides incremental value for discriminating CAD in this very elderly population.

Methods

We retrospectively included 155 patients aged ≥80 years who underwent coronary imaging evaluation between 2020 and 2022. EAT volume was quantified from non-contrast chest CT. Logistic regression, CatBoost modeling, and SHapley Additive exPlanations (SHAP) were used to assess the association of EAT with CAD, its discriminative value, and its relative contribution.

Results

Forty-five patients (29.0%) had CAD. Median EAT volume did not differ between groups (p = 0.716) and was not associated with CAD after adjustment for age, sex, body mass index, diabetes, and glycated hemoglobin A1c (HbA1c). Adding EAT did not improve discrimination in logistic regression [area under the curve (AUC) 0.698 vs. 0.697; likelihood-ratio p = 0.945] or in cross-validated CatBoost analysis (AUC 0.623 vs. 0.614). SHAP ranked EAT eighth of eighteen predictors, below lipoprotein(a), total cholesterol, left atrial diameter, HbA1c, low-density lipoprotein cholesterol, diabetes, and atrial fibrillation.

Conclusions

In adults aged 80 years and older, CT-derived EAT volume was not independently associated with CAD and added no discriminative value beyond conventional clinical risk factors. Total EAT volume may have limited utility as a CAD marker in very elderly adults.