Association of Endocrine Active Environmental Compounds With Body Mass Index and Weight Loss Following Bariatric Surgery

Harshal Deshmukh; Lesa L. Aylward; Martin Rose; Alwyn Fernandes; Peter Sedman; Natalie J. Thatcher; Stephen L. Atkin; Thozhukat Sathyapalan


Clin Endocrinol. 2020;93(3):280-287. 

In This Article


Study Population and Sample Collection

The study population consisted of patients undergoing Roux-en-Y gastric bypass surgery for weight loss or patients who were undergoing abdominal surgery for nonbariatric reasons (mainly Nissen fundoplication for gastro-oesophageal reflux disease) from April 2010 until March 2015. Informed consent for the study was obtained and National Research Ethics Committee, Yorkshire and Humber—South Yorkshire, UK (REC reference 10/H1304/13), approved the study. Anthropometric parameters including height and weight were measured on the day of surgery. During surgery, tissue samples were taken including liver (500 mg) and adipose tissue (visceral and subcutaneous: 40 g each). Opportunistic contacts for follow-up sampling with the participants were also made when they attended the bariatric surgery follow-up clinic. For those who consented, subcutaneous fat biopsies were taken at follow-up. An incision was made in the anterior abdominal wall under local anaesthesia after sterile precaution and subcutaneous fat biopsies were taken. Samples were stored at −80ºC until packed in dry ice for transport to the UK National Reference Laboratory for Contaminants in Food and Feed (FERA) laboratory for analysis. After receipt at the FERA laboratory, samples were stored at −20ºC until analysis. These dioxins are considered very stable compounds and are highly persistent in vivo and vitro with an estimated half-life of 9–15 years on soil surfaces.[22]

Sample Analysis

The method[23] used for the preparation, extraction and analysis of samples is comprehensively validated, formally accredited and forms part of the modular CEN method EN16215:2012. In brief, samples were fortified with 13C-labelled analogues of target compounds and exhaustively extracted using mixed organic solvents. PBDEs and ortho-substituted PCBs/PBBs were separated from nonortho-substituted PCBs/PBBs, PCDD/Fs and PBDD/Fs by fractionation on activated carbon. The two fractions were further purified using adsorption chromatography on alumina. Analytical measurement was carried out using high-resolution gas chromatography-high-resolution mass spectrometry (HRGC-HRMS) for all analytes apart from the ortho-substituted PCBs which were analysed by high-resolution gas chromatography-unit resolution mass spectrometry (HRGC-LRMS). Lipid content was measured using a method based on BS:4401: Part 4 1970 (Werner-Schmidt Method).

The analysis is accredited (UKAS) to ISO 17 025 standards, with the inclusion of an in-house reference material and method blanks which were evaluated prior to reporting of sample data and used to determine the limits of detection. Further quality assurance measures included the successful participation in international intercomparison exercises such as Dioxins in Food-2011 to 2014, and EURL-run PT exercises on dioxins, dioxin-like PCBs, ICES-6 PCBs and PBDEs. Additionally, quality control evaluation for the accompanying data follows the criteria specified for chlorinated dioxins and PCBs (Commission Regulation 252/2012, (European Commission, 2012). Not every targeted sample was collected, and for a variety of reasons, the number of analysed samples varied by matrix and chemical compound group. For example, in some cases, particularly for liver samples, the collected sample was of insufficient mass to allow analysis for all of the target chemical groups. In some cases, the analytical results were flagged as 'indicative' and were inconsistent with other data collected for the same compounds, and the decision was made to exclude those analytical results from the statistical analyses. Finally, for brominated dioxins and furans, 17 out of 21 contributing to toxic equivalency (TEQ) were measured due to the lack of available standards. Specifically, octa-brominated dioxin and furan were not measured and only one hepta-brominated (a hepta-brominated furan) was measured. Although these have low toxic equivalency factor (TEF) values they are likely to contribute to total brominated TEQ, so it is likely that the brominated toxic equivalency (TEQ) values will be somewhat underestimated.

Statistical Methods

Summary statistics for each measured congener in each tissue (detection frequency, geometric mean, geometric standard deviation, minimum and maximum) were generated in R 3.5.0.

Several studies have shown very significant association of age with distribution of dioxin chemicals in the population.[24,25] A simple bivariate correlation which will be confounded by age is therefore not useful and can be misleading. Hence, we assessed the association of these environmental chemicals using regression analysis adjusting for both age and gender. Linear regression analysis implemented in R was used to test the association of BMI with the log10-transformed concentration of each compound, adjusted for age and gender. Bonferroni correction was used to identify a threshold for statistical significance (0.05/number of tests) to account for multiple testing. Since the environmental chemicals showed a high degree of correlation in subcutaneous tissues, visceral fat and liver, we used a conservative Bonferroni correction of 0.05/63 where 63 is the number of chemicals measured across all the tissues.

Chlorinated PCBs and PCDD/Fs were analysed in baseline and follow-up samples from 10 participants. Sensitivity analysis was performed by (a) imputing missing data with random forest algorithm imputation in R and (b) by using raw values on environmental chemicals (rather than log-transformed).

To assess changes in analyte concentrations following weight loss, changes in the sum of three persistent indicator PCBs (138, 153 and 180) and in lower bound estimates of chlorinated ED were examined. Baseline and follow-up fat samples were analysed for selected brominated compounds in 10 individuals. BDE 153 was selected as a marker to examine changes in persistent brominated compound concentrations in these two participants. Shapiro-Wilk tests were used to compare the median concentrations of the analytes before and after the bariatric surgery.