Real-world cumulative exposure of various endocrine-disrupting chemicals and association with breast cancer risk via common carcinogenic pathway.

Abstract

630 Background: Convergent evidence suggests carcinogenicity of endocrine disrupting chemicals (EDCs). But there is a lack of large-scale epidemiological evidence concerning real-world EDCs mixture exposure and breast cancer (BC). A dataset of emerging knowledge of toxic pathways provides a valuable toolkit to infer the carcinogenicity of chemicals, but current studies of this thus filed are rare. We test whether real-world EDCs exposure, independently or in mixture, increases BC risk in the population, through carcinogenic toxic pathways. Methods: A total of 234,273 women from the UK Biobank cohort were followed for a median of 13.85 years. The annual monitoring record from UK Water Quality Sampling Harmonised was used to estimate the exposure to 13 EDCs (e.g., lead and chlorpyrifos-methyl) for each woman by Kriging interpolation model. The association between EDCs and breast cancer was analyzed by Cox proportional hazard model (for single EDC) and weighted quantile sum (WQS) regression (for mixture). The EDCs were also added to the modified Gail model to test their additional contribution. Twenty-seven Olink proteins were selected via EDC-BC toxic pathway dataset and BC genetic propensity score model to examine their correlation with EDCs. Interaction between EDCs and these proteins and SNPs within their sequence were analyzed by multiplicative model. Results: During the follow-up period, 9,282 women developed BC, 984 of whom died. All the 13 EDCs showed association with increased BC risk. Each unit increase of the 13-EDC mixture was associated with 2.15 fold (95% CI, 2.01-2.31) of BC risk and 4.18 fold of BC-related mortality (95% CI, 3.39-5.17), and 34.38 % of total morbidity and 59.6% of BC-related mortality in the population were attributable to EDC mixture. The risk of hormone receptor-positive BC was higher than that of hormone receptor-negative BC (HR: 2.87 vs HR: 2.03, p = 0.002). The EDCs provided 2.1 % improvement to the modified Gail model of BC prediction. For the 27 Olink proteins referring to BC toxic pathway and genetic propensity score, 26 (96.3%) correlated with at least one EDC (average number ≥8), among which NACC1 protein mediated up to 98.8% (chlorpyrifos-methyl) of the individual EDC’s effect on BC. The EDC mixture showed interaction with multiple proteins, including NACC1 and estrogen receptor 1, on BC-related mortality. Conclusions: Various EDCs in the real world may increase BC morbidity and mortality via common carcinogenic pathways. Future policies of novel EDC emission and safety evaluations must consider the cumulative EDC stressor in the environment because they may act as a whole.