Supplementary Figure S1 shows the study flow chart. Table 1 shows the demographic characteristics of the study populations and the compound measured in the study. The study consisted of 106 participants with 64% females with a median age of 55 (IQR = 44–56) years and a median BMI of 37 (IQR = 27–37) kg/m2. We looked at 81 environmental chemicals consisting of twenty-one ortho-chlorinated biphenyls, four nonortho-chlorinated biphenyls, seventeen PCDD/Fs, seventeen polybrominated diphenyl ethers, seven ortho-brominated biphenyls, 3 nonortho-brominated biphenyls and 12 PBDD/Fs. Of the 81 compounds, 18 were not included in subsequent analyses because detection frequencies were below 65% in all tissues. Supplementary Table S1 and Supplementary Table S2 show the summary statistics for the concentrations of each analyte in subcutaneous fat, visceral fat and liver. Concentrations of most of these chemicals show high correlation with each other and are also highly correlated in subcutaneous fat, visceral fat and liver (Supplementary Figures S2, S3, S4 and S5).
Figure 1 shows a heatmap of the beta coefficients for associations between BMI and log10-transformed, lipid-adjusted concentrations of 63 compounds after adjustments for age and gender in subcutaneous fat, visceral fat and liver. Highly significant associations between BMI and compounds are indicated with darker shades of red while less significant associations are shown in blue. After adjustments for age and gender and correction for multiple testing (P < .007, the threshold for Bonferroni correction), BMI was significantly associated with seven ortho-chlorinated biphenyls, one nonortho-chlorinated biphenyl, four PCDD/Fs and one ortho-brominated biphenyl (Table 2). The strongest associations were seen for PCB-105 in subcutaneous fat (beta = 16.838 P-val = 1.45E-06) PCB-126 in visceral fat (beta = 15.067 P-val = 7.72E-06) and PCB-118 (beta = 14.101 P-val = 2.66E-05) in liver. All associations were positive except for 2,2',4,4',5,5'-hexabromobiphenyl (BB-153)in the liver, which showed a negative correlation with BMI.
Shows a heatmap of the beta coefficients for associations between BMI and log10-transformed, lipid-adjusted concentrations of 63 compounds after adjustments for age and gender in subcutaneous fat, visceral fat and liver [Colour figure can be viewed at wileyonlinelibrary.com]
Changes in the Concentrations Selected Compounds Following a Bariatric Surgery
Follow-up subcutaneous fat biopsies of sufficient volume to allow analysis were collected from 10 bariatric surgery group participants. The biopsies were collected following an average of 2.8 years after initial surgery (range 1–4.8 years) and participants lost an average of 35.3% of their presurgery body weight. Fat mass was estimated at the baseline surgery time point and at the follow-up sample time using a prediction formula based on age, gender and BMI. Based on these estimates, the participants lost an average of 48.6% of their body fat (range 38.1%–60.6%). Figure 2A and 2B shows the concentrations of a sum of key indicator PCBs (138, 153 and 180) and chlorinated toxic equivalency (TEQ) before and after bariatric surgery. In the 10 follow-up participants, the mean concentration of the sum of indicator PCBs increased from 111.74 μg/kg lipid to 240.6 μg/kg (P = .02) while the mean chlorinated TEQ concentration increased from 10.6 ng/kg lipid to 23.0 ng/kg lipid (P = .02). BDE-153 was also measured in two participants before and after surgery, and the mean concentrations increased 2.5 μg/kg lipid to 4.04 μg/kg lipid.
Clin Endocrinol. 2020;93(3):280-287. © 2020 Blackwell Publishing