Objective: After menopause, body composition changes with body fat accumulation, and an increase in cardiometabolic risk factors. Total fat mass, regional fat mass, and visceral adipose tissue (VAT) may be estimated with anthropometric measures, bioelectrical impedance analysis (BIA), and dual-energy X-ray absorptiometry (DXA). The aim of our study was to assess which measurement correlated best with cardiometabolic risk factors in healthy nonobese postmenopausal women.
Methods: The CoLaus/OsteoLaus cohort included 1,500 postmenopausal women (age range 50-80). We analyzed correlations between: 1) measurements of body composition assessed by anthropometric measures, BIA, and DXA and 2) these measurements and different selected cardiometabolic risk factors, such as blood pressure, lipid markers (cholesterol subtypes and triglycerides), and metabolic markers (glucose, insulin, adipo- nectin, and leptin). Spearman correlation coefficient, stepwise forward regression, and linear regression analyses were used to determine association between anthropometric measurements and cardiometabolic risk factors.
Results: In the 803 included participants (mean age 620±7.1 y, mean body mass index 25.6 kg/m2±4.4), correlations between total fat mass measured by BIA and total fat mass, android fat, gynoid fat, or VAT measured by DXA are very strong (from r=0.531, [99% confidence interval (CI), 0.443-0.610] to r=0.704, [99% CI, 0.640-0.758]). Body mass index and waist circumference have a higher correlation with VAT (r=0.815, [99% CI, 0.772-0.851] and r=0.823 [99% CI, 0.782-0.858], respectively) than BIA (r=0.672 [99% CI, 0.603-0.731]). Among the anthropometric measurement and the measurements derived from DXA and BIA, VAT is the parameter most strongly associated with cardiometabolic risk factors. VAT better explains the variation of most of the cardiometabolic risk factors than age and treatment. For example, nearly 5% of the variability of the diastolic blood pressure (9.9 vs 4.9), nearly 15% of the variability of high-density lipoprotein cholesterol (20.3 vs 3.8) and triglyceride (21.1 vs 6.5), 25.3% of the variability of insulin (33.3 vs 8.1), and 37.5% of the variability of leptin (37.7 vs 1.1) were explained by VAT. Conclusions: BIA seems not to be a good tool to assess VAT. At the population level, waist circumference and body mass index seem to be good tools to estimate VAT. VAT measured by DXA is the parameter most correlated with cardiometabolic risk factors and could become a component of the cardiometabolic marker on its own.