Persistent organic pollutants (POPs) are a group of organic chemicals potentially toxic to human health. Induction of oxidative stress is one of the suspected mechanisms of action. The aims of this study were to model exposure-response functions between blood levels of a mixture of POPs and biomarkers of oxidative stress, to identify potential interactions between POPs, and to estimate the overall effect of the mixture. Levels of POPs and oxidative stress biomarkers were measured in the blood of 467 women from the French E3N cohort study, aged 45-73 years, collected between 1994 and 1999. In total, 41 POPs quantified in at least 75% of samples, and 3 antioxidant enzymes (Superoxide Dismutase 1 (SOD1), Superoxide Dismutase 2 (SOD2), and alpha-Glutathione S-Transferase (GST alpha)) were included. A Bayesian Kernel Machine Regression (BKMR) model was fitted for each oxidative stress biomarker, including the 41 POPs as exposure variables and adjusting for potential confounders identified using a directed acyclic graph. Additionally, linear regression models including each POP biomarker separately adjusted for potential confounders were run. With the BKMR models, only two POPs biomarkers were found associated to SOD1 (PFPeS) and SOD2 (PCB- 156). A greater number of POPs appeared associated to GST alpha (oxychlordane, dieldrin, PFUnDA, PFHpA, PCB-28, PCB-153, PCB-180, PBDE-47, PBDE-100, and PBDE-153). Single-pollutant linear models also highlighted statistically significant associations in the same direction as the BKMR model. The BKMR models also highlighted non-linear cumulative effects, with overall negative trends for SOD1 and SOD2 and a positive trend for GST alpha. These findings support that oxidative stress may be involved in the mechanisms linking the exposure of mixtures of POPs and related health effects. Further epidemiological studies on larger populations, as well as toxicological studies, are necessary to confirm these results.
