Of the 1175 participants in the study, 53% were women, the mean age was 59±9 years, and the mean BMI was 28.2±5.2 kg/m2. The mean pericardial and intrathoracic fat volumes were 124(±51) cm3 and 114(±62) cm3, respectively. See for study sample characteristics.
Age- and sex-adjusted Pearson’s correlation coefficients for pericardial fat with inflammatory marker concentrations revealed statistically significant positive correlations of pericardial fat with CRP, fibrinogen, intracellular adhesion molecule-1, interleukin-6, isoprostanes, MCP-1, P-selectin, and tumor necrosis factor receptor-2. The same markers, as well as myeloperoxidase, had significant positive Pearson’s correlation coefficients for intrathoracic fat ().
Age- and Sex-adjusted Pearson’s Correlation Coefficients for Inflammatory Markers and Fat Volumes.
Results of sex-pooled multivariable linear regression models with backwards elimination of inflammatory markers are presented in . For pericardial fat, backwards elimination in Model 1 revealed significant correlations with concentrations of CRP (P<0.0001), interleukin-6 (P<0.02), and MCP-1 (P<0.01); the overall R2 of Model 1 was 0.244. After further adjusting for BMI and waist circumference (Model 2), interleukin-6 was no longer significant; and after further adjusting for abdominal visceral fat volume (Model 3), none of the inflammatory markers were retained in the model. In Model 4, which adjusted for additional covariates but not abdominal visceral fat volume, CRP (P<0.01) and MCP-1 (P<0.05) remained correlated with pericardial fat.
Multivariable-adjusted Backwards Elimination Linear Regression Models for Relations Between Inflammatory Markers & Fata.
Similarly for intrathoracic fat, backwards elimination in Model 1 revealed significant correlations with CRP (P<0.0001), interleukin-6 (P=0.03), and MCP-1 (P<0.01) (R2 of Model 1 was 0.425); with loss of significance of interleukin-6 after further adjusting for BMI and WC (Model 2). However, CRP remained significantly correlated with intrathoracic fat even after additional adjustment for abdominal visceral fat in Model 3 (P<0.0001) and the additional covariates in Model 4 (P<0.0001). Examining age- and sex-adjusted mean CRP by tertiles of intrathoracic and abdominal visceral fat, the trend for higher CRP concentrations across intrathoracic fat tertiles was significant (P<0.05) within all 3 abdominal visceral fat tertiles ().
Figure 1 Mean log-transformed circulating C-reactive protein (CRP) concentrations across tertiles of intrathoracic fat and abdominal visceral adipose tissue. Trend across tertiles of intrathoracic fat were significant within all tertiles of abdominal visceral (more ...)
We examined the addition of tumor necrosis factor-alpha and urinary isoprostanes to the marker panel in the subset of subjects with these markers. Backwards elimination revealed that urinary isoprostanes were significantly correlated with intrathoracic fat even after adjustment for BMI, waist circumference, and abdominal visceral fat (P=0.02), as well as the covariates in Model 4 (P<0.001). Examining age- and sex-adjusted mean urinary isoprostanes concentrations by tertiles of intrathoracic and abdominal visceral fat, the trend for increased urinary isoprostanes across intrathoracic fat tertiles was significant (P<0.05) in the middle abdominal visceral fat tertile ().
Figure 2 Mean log-transformed urine isoprostanes concentrations across tertiles of intrathoracic and abdominal visceral adipose tissue. Trend across tertiles of intrathoracic fat were significant within the middle tertile of abdominal visceral fat (*P<0.05). (more ...)
For intrathoracic fat, the test for statistical interaction between sex and marker concentrations was significant for CRP (P<0.0001), fibrinogen (P=0.0007), and isoprostanes (P=0.0001). Effect sizes were larger for men than women. There was no statistical interaction between intrathoracic fat and age or obesity. For pericardial fat, the test for statistical interaction between sex and marker concentrations was significant for fibrinogen (P=0.007) and isoprostanes (P=0.009); and that between obesity and marker levels was significant for MCP-1 (P=0.008) and P-selectin (P=0.004). Effect sizes were larger for men than women, and for obese versus non-obese participants. There was no statistically-significant interaction between pericardial fat and age.