Perfluoroalkyl Substances and Thyroid Function in Pregnant Women and their Offspring

Background: Normal thyroid function during pregnancy is critical for fetal development. Experimental studies suggest that certain perfluoroalkyl substances (PFAS) may disrupt thyroid function but findings from observational studies have been inconsistent.Aim: To examine the association between maternal PFAS concentration in pregnancy with 1) maternal thyroid function in pregnancy and 2) offspring thyroid function 20 years later.Methods: In a cohort of Danish pregnant women who gave birth in 1988-89 concentrations of PFAS, thyroid stimulating hormone (TSH) and free thyroxine (FT4) were quantified in 876 women using samlpes from week 30 of gestation. TSH and FT4 concentrations were also quantified in 411 offspring 20 years later.Results: The median concentrations of PFHxS, PFOS, PFOA, PFNA and PFDA were 0.35 ng/mL, 21.5 ng/mL, 3.7 ng/mL, 0.34 ng/mL and 0.05 ng/mL, respectively. In pregnancy concentrations of PFNA were associated with around 14% higher maternal TSH concentrations (95% CI: 10%, 30%) when comparing the highest to lowest quartile of exposure (p for trend=0.04). Non-significant but positive associations were observed for the other compounds and no associations were observed with maternal FT4. Prospectively maternal concentrations of PFOA were associated with around 24% higher (95%CI: 8%, 43%) TSH concentrations in the offspring at 20 years when comparing the highest to lowest quartile of maternal PFOA exposure (p for trend = 0.01). A modest non-significant association was observed for PFOS (p for trend = 0.13). No associations were observed for the other compounds and no associations were observed with offspring FT4 concentrations.Conclusion: Except for PFNA no clear link was observed between circulating PFAS concentrations and thyroid function in pregnancy. Prospectively, there were some indications that maternal PFOA concentrations may be associated with higher offspring TSH concentrations at 20 years but replication is needed to exclude chance findings.