RESUMO
BACKGROUND: In vitro and toxicological studies have shown that non-persistent environmental chemicals can perturb thyroid hormone homeostasis. Epidemiological studies with improved exposure assessment (i.e., repeated urine samples) are needed to evaluate effects of these compounds, individually or as a mixture, in humans. We studied the associations between prenatal exposure to non-persistent environmental chemicals and neonatal thyroid hormones. METHODS: The study population consisted of 442 mother-child pairs from the French SEPAGES mother-child cohort recruited between July 2014 and July 2017. For each participant, four parabens, five bisphenols, triclosan, triclocarban, benzophenone-3 as well as metabolites of phthalates and of di(isononyl)cyclohexane-1,2-dicarboxylate were assessed in two pools of repeated urine samples (median: 21 spot urines per pool), collected in the 2nd and 3rd trimesters of pregnancy, respectively. Thyroid stimulating hormone (TSH) and total thyroxine (T4) levels were determined in newborns from a heel-prick blood spot. Maternal iodine and selenium were assessed in urine and serum, respectively. Adjusted linear regression (uni-pollutant model) and Bayesian Kernel Machine Regression (BKMR, mixture model) were applied to study overall and sex-stratified associations between chemicals and hormone concentrations. RESULTS: Interaction with child sex was detected for several compounds. Triclosan, three parabens, and one phthalate metabolite (OH-MPHP) were negatively associated with T4 among girls in the uni-pollutant model. BKMR also suggested a negative association between the mixture and T4 in girls, whereas in boys the association was positive. The mixture was not linked to TSH levels, and for this hormone the uni-pollutant model revealed associations with only a few compounds. CONCLUSION: Our study, based on repeated urine samples to assess exposure, showed that prenatal exposure to some phenols and phthalates disturb thyroid hormone homeostasis at birth. Furthermore, both uni-pollutant and mixture models, suggested effect modification by child sex, while, to date underlying mechanisms for such sex-differences are not well understood.
