Effects of Contrast Medium and Vertebral Measurement Levels on Computed Tomography-Based Body Composition Parameters: Skeletal Muscle and Adipose Tissue Analysis

OBJECTIVE: This study aims to evaluate the effects of contrast medium administration and vertebral measurement level (L3 and L1) on computed tomography (CT)-derived body composition parameters and to establish conversion equations to enable the normalization and harmonization of these metrics across heterogeneous imaging protocols, which would aid in oncology prognostication or opportunistic screening. METHODS: A total of 203 dynamic contrast-enhanced CT examinations, including unenhanced (phase 0) and early arterial, late arterial, portal, and equilibrium phases (phases 1-4, respectively), were retrospectively enrolled in the study. Skeletal muscle area (SMA) and mean skeletal muscle density (MSMD), subcutaneous adipose tissue area (SATA) and mean subcutaneous adipose tissue density (MSATD), and visceral adipose tissue area (VATA) and mean visceral adipose tissue density (MVATD) were measured at the L3 and L1 levels across five phases (phases 0-4). Measurement changes among phases and levels were assessed statistically. RESULTS: The percentage changes in SMA, SATA, and VATA in phases 1/2/3/4 from phase 0 were as follows: L3 SMA, +1.1%/+2.1%/+2.8%/+3.5%; L1 SMA, +2.0%/+2.7%/+3.0%/+3.7%; L3 SATA, -0.4%/-1.9%/-2.9%/-4.7%; L1 SATA, -0.3%/-2.2%/-3.5%/-6.1%; L3 VATA, -7.5%/-17.8%/-20.0%/-22.2%; and L1 VATA, -8.5%/-20.1%/-22.6%/-23.7%. Differences in MSMD, MSATD, and MVATD in phases 1/2/3/4 compared with phase 0 were as follows (Hounsfield units): L3 MSMD, +2.1/+5.8/+7.8/+9.6; L1 MSMD, +2.9/+7.7/+9.1/+9.8; L3 MSATD, +1.2/+3.7/+4.8/+6.2; L1 MSATD, +1.2/+3.8/+4.7/+5.6; L3 MVATD, +1.5/+4.1/+5.2/+6.1; and L1 MVATD, +1.9/+4.9/+5.8/+6.3. Values between L3 and L1 showed a strong linear correlation (coefficient of determination, 0.950-0.999), suggesting interchangeability. CONCLUSION: The effects of contrast phase and measurement level on CT body composition parameters were comprehensively characterized. These findings provide robust conversion equations that enable normalization of data from cohorts with mixed imaging protocols, thereby strengthening the prognostic value of body composition analysis.