Maternal Perfluoroalkyl Substances, Thyroid Hormones, and DIO Genes: A Spanish Cross-sectional Study

Abstract

Results of studies on perfluoroalkyl substances (PFASs) and thyroid hormones (THs) are heterogeneous, and the mechanisms underlying the action of PFASs to target THs have not been fully characterized. We examined the relation between first-trimester maternal PFAS and TH levels and the role played by polymorphisms in the iodothyronine deiodinase 1 (DIO1) and 2 (DIO2) genes in this association. Our sample comprised 919 pregnant Spanish women (recruitment = 2003–2008) with measurements of perfluorohexanesulfonic acid (PFHxS), perfluorooctanoic acid (PFOA), perfluorooctanesulfonic acid (PFOS), perfluorononanoic acid (PFNA), thyroid-stimulating hormone (TSH), total triiodothyronine (TT3), and free thyroxine (FT4), and we genotyped for single-nucleotide polymorphisms in the DIO1 (rs2235544) and DIO2 (rs12885300) genes. We performed multivariate regression analyses between PFASs and THs and included the interaction term PFAS–genotypes in the models. PFHxS was associated with an increase in TSH (% change in outcome [95% CI] per 2-fold PFAS increase = 6.09 [−0.71, 13.4]), and PFOA and PFNA were associated with a decrease in TT3 (−7.17 [−13.5, −0.39] and −6.28 [−12.3, 0.12], respectively). We found stronger associations between PFOA, PFNA, and TT3 for DIO1–CC and DIO2–CT genotypes, although interaction p-values were not significant. In conclusion, this study found evidence of an inverse association between PFOA and TT3 levels. No clear effect modification by DIO enzyme genes was observed.

Comment on “Maternal Perfluoroalkyl Substances, Thyroid Hormones, and DIO Genes: A Spanish Cross-sectional Study”: Predictability of Multiple Imputations for Large Amounts of Missing Data

“…One of these factors, estimated glomerular filtration rate (eGFR), can reportedly alter blood PFAS concentrations during pregnancy, potentially causing confounding with health outcomes . The authors reported that the association between perfluorooctanoic acid and TT3 did not change after eGFR was included in the statistical model (Figure S7 in the original article) . A previous INMA project study on PFASs also found that eGFR did not affect the studied association and did not correlate with blood concentrations of PFAS …”

“…Sarzo et al analyzed associations between first-trimester maternal perfluoroalkyl substances (PFASs) and thyroid hormone concentrations and the role played by genetic polymorphisms of the Iodothyronine Deiodinase 1 ( DIO1 ) and Iodothyronine Deiodinase 2 ( DIO2 ) genes in the INMA (Infancia y Medio Ambiente) project . After adjusting for maternal age, cohort, country of birth, season of blood collection, educational level, smoking habit, and previous breastfeeding when calculating the effects on triiodothyronine (TT3) concentrations, they identified a possible inverse association between perfluorooctanoic acid and total TT3 concentrations.…”

“…One review states that complete case analysis (CCA) is preferred over multiple imputations if the percentage missing is <5% and that multiple imputations are invalid if the percentage is >40% . In the study, we are here questioning, not only eGFR, but also 4,4′-dichlorodiphenyldichloroethylene, ∑ polychlorinated biphenyls, and β-hexachlorocyclohexane, were imputed in 288 patients (31.3%) with missing values. The authors did not clarify whether the missing data were missing not at random.…”

Response to “Comment on Maternal Perfluoroalkyl Substances, Thyroid Hormones, and DIO Genes: A Spanish Cross-sectional Study: Predictability of Multiple Imputations for Large Amounts of Missing Data”

“…We appreciate the comments by Kobayashi and Harada on our work. The authors were concerned about the number and the nature of the missing data in some of the sensitivity analyses performed in our study “Maternal Perfluoroalkyl Substances, Thyroid Hormones, and DIO Genes: A Spanish Cross-sectional Study” by Sarzo et al Specifically, they referred to the analyses performed with the estimated glomerular filtration rate (eGFR) and some organochlorine compounds (OCs): 1,1-dichloro-2,2-bis­(4-chlorophenyl)­ethylene (4,4′-DDE), the sum of three polychlorinated biphenyls (∑(PCBs): PCB-138, −153, and −180), and beta-hexachlorocyclohexane (b-HCH).…”

“…Specifically, in the main model (i.e., the model without genes) we included alternatively several variables reported in the literature − and available for this cohort which could confound the PFAS and TH associations, such as urinary iodine, salt intake, iodine intake from diet and supplements, energy-adjusted seafood intake, lipids plus OCs, mercury (Hg), the four PFASs simultaneously, prepregnancy body mass index, and glomerular filtration rate (eGFR). Hence, we performed a total of 156 sensitivity analyses (4 PFASs × 3 THs × each of the above-mentioned variables) and presented them in Figure S7 in Sarzo et al…”