Sexual dimorphism and pro-atherogenic computed tomography radiomic phenotypes of subcutaneous and visceral abdominal...

Standard body fat metrics fail to capture the unique biological characteristics of subcutaneous (SAF) and visceral (VAF) abdominal fat. We characterized computed tomography (CT) radiomic profiles of SAF and VAF, identified sex-specific and pro-atherogenic signatures, and assessed their incremental value in cardiovascular risk assessment.

In 1736 adults without known coronary artery disease, we extracted size, first- and second-order (texture) features of SAF, VAF, and total abdominal fat (TAF) from non-contrast CT. Partial least squares discriminant analysis and XGBoost models were developed to differentiate abdominal fat depots, biological sex, and coronary calcification (defined as coronary calcium score, CCS > 0).

SAF and VAF exhibited distinct radiomic profiles. SAF was characterized by lower attenuation and greater high-density structural continuity. Sexual dimorphism was evident in both depots, but was particularly pronounced in SAF; males displayed greater textural heterogeneity and radiodensity compared to females. Pro-atherogenic signatures exhibited increased textural heterogeneity in both depots. However, coronary calcification was associated with higher radiodensity in SAF (metabolic densification) and lower radiodensity in VAF (pathological hypertrophy). TAF radiomics significantly enhanced the performance of the TAF area to detect CCS > 0, and its accuracy was comparable to more complex models combining the best-performing pro-atherogenic SAF and VAF radiomic features.

CT radiomics identified texture heterogeneity as a key feature of male, visceral, and pro-atherogenic adipose tissue phenotypes, while further supporting the protective characteristics of SAF compared with VAF. Unlike conventional TAF area measurements, radiomic analysis of TAF captures underlying tissue complexity and heterogeneity, providing a robust alternative to separate abdominal fat depot assessments for individualized obesity risk stratification. These findings highlight the potential of CT radiomics to differentiate biologically distinct abdominal fat phenotypes and enhance obesity-related cardiometabolic risk profiling.