Sex-specific, non-linear and congener-specific association between mixed exposure to polychlorinated biphenyls (PCBs) and diabetes in US adults

Background: Whether and to what extent the impact of exposure to various polychlorinated biphenyls (PCBs) congeners on diabetes, as well as the important contributors, have remained unclear. Objective: We aimed to investigate the association patterns between PCBs mixture and diabetes, identify the critical congeners, and explore the potential modifiers. Methods: The present study included 5900 U.S. adults from the National Health and Nutrition Examination Survey (NHANES) conducted between 2007 and 2016. Weighted logistic regression, restricted cubic spline regression, weighted quantile sum (WQS) regression and Bayesian kernel machine regression (BKMR) were applied to estimate the linear and non-linear associations of single and mixed PCB exposure with diabetes. Subgroup analyses were also conducted to explore potential sex differences. Results: In the weighted logistic regression model, total PCBs were positively associated with diabetes (OR = 1.33, P < 0.025), and significant non-linear associations were observed using RCS analyses. The non-linear positive association between PCBs mixed exposure and diabetes was likewise found in the WQS and BKMR results. PCB180, PCB194, PCB196, and PCB167 were with the highest weights in the WQS, and PCB209 and PCB66 were with the highest posterior inclusion probabilities in the BKMR. Additionally, exposure to total PCBs and most of individual PCB congeners were significantly associated with elevated risk of in females (OR = 1.74; P for trend < 0.001), while fewer significant associations were observed in males. Conclusion: The present study highlighted the importance of the long-term surveillance of PCBs and the need to enhance protective measures against them. Notably, these associations were non-linear, congener-specific, and significantly stronger in females than males, especially at relatively high levels of PCBs exposure. Further prospective and mechanistic studies were warranted to ascertain the causal effects between PCBs mixture and diabetes.