Inverse Association of Longitudinal Variations in Fat Tissue Radiodensity and Area

Increased CT-derived fat tissue radiodensity has been indicated as a poor prognostic factor in oncological settings, although the reasons are not clear. One hypothesis is that increased radiodensity may reflect the loss of fat droplets within adipocytes, being a proxy of recent weight loss. This study aims to test this hypothesis by evaluating the association between longitudinal variations in fat tissue radiodensity and area in a cohort of COVID-19 patients. Baseline and 2-3-month follow-up chest CT scans of severe COVID-19 pneumonia survivors were retrospectively reviewed to measure subcutaneous, visceral, and intermuscular adipose tissue (SAT, VAT, and IMAT) areas and densities at the T7-T8 vertebrae, and longitudinal variations were computed for each variable. The associations between each compartment area and radiodensity variations (standardized values) were evaluated in univariate linear models and models adjusted by age and sex. A total of 196 COVID-19 survivors with suitable baseline and follow-up CT scans were included (mean age 65 +/- 11 years, 62 (31.6%) females, 25% with diabetes and 2.6% with morbid obesity). Longitudinal variation in SAT area was inversely associated with longitudinal variation in SAT radiodensity in univariate models (coeff -0.91, 95%CI = -1.70/-0.12, p = 0.02) and after adjustment by age and sex (coeff -0.89, 95%CI = -1.7/-0.09, p = 0.03). The effect was similar and stronger for IMAT (coeff -2.1, 95%CI = -3.06/-1.19, p < 0.01 in adjusted models), and absent for VAT. Longitudinal variations in subcutaneous and intermuscular adipose tissue areas and densities are inversely associated. Higher adipose tissue radiodensity may be due to decrease in fat area (i.e., weight loss), explaining the poor prognostic effect found in cancer patients.