UCSF

BACKGROUND: Atopic disease may be influenced by prenatal and early life exposure to endocrine disrupting chemicals, including bisphenols, but results from epidemiological studies have been mixed. This study aimed to extend the epidemiological literature, hypothesizing that children with higher prenatal bisphenol exposure are more likely to have childhood atopic disease. METHODS: Urinary bisphenol A (BPA) and S (BPS) concentrations were measured in each trimester from 501 pregnant women in a multi-center, prospective pregnancy cohort. Ever asthma, current asthma, wheeze, and food allergy) were assessed at age six via standardized ISAAC questionnaire. We constructed generalized estimating equations to examine BPA and BPS exposure jointly at each trimester for each atopy phenotype. BPA was modeled as a log-transformed continuous variable, whereas BPS was modeled as detected versus not detected. We also modeled pregnancy-averaged BPA values and a categorical indicator for number of detectable BPS values over pregnancy (0-3) in logistic regression models. RESULTS: First trimester BPA was associated with inverse odds of food allergy among the entire study sample (OR = 0.78, 95% CI = 0.64-0.95, p = 0.01) and females only (OR = 0.69, 95% CI = 0.52-0.90, p = 0.006). The inverse relationship persisted in pregnancy-averaged models of BPA among females (OR = 0.56, 95% CI = 0.35-0.90, p = 0.006). Second trimester BPA was associated with greater odds of food allergy in the entire sample (OR = 1.27, 95% CI = 1.02-1.58, p = 0.03) and among males only (OR = 1.48, 95% CI = 1.02-2.14, p = 0.04). Odds of current asthma increased among males in the pregnancy-averaged BPS models (OR = 1.65, 95% CI = 1.01-2.69, p = 0.045). CONCLUSION: We saw opposite effects of BPA on food allergy that were trimester- and sex-specific. These divergent associations warrant further investigation. There is some evidence to suggest that prenatal BPS is associated with asthma among males, but further research is required in cohorts with a greater proportion of prenatal urine samples with detectable BPS to validate these results.