Associations of serum phthalate metabolites with thyroid hormones in GraMo cohort, Southern Spain

Abstract

The general population is continuously exposed to phthalates via various consumer products. Epidemiological research relating phthalate exposure to thyroid function during non-developmental periods is limited. This study aimed to investigate the associations between specific serum phthalate metabolites and indicators of thyroid function in adults. We measured 10 serum phthalate metabolites and thyroid hormones - total triiodothymnine (TT3), free thyroxine (FT4) and thyroid stimulating hormone (TSH) - in a subsample of 207 adults from the GraMo cohort. This subsample was made up of men and women (in equal proportions) of middle age (49 +/- 17 years) and from Southern Spain (province of Granada). Data on age, sex, body mass index, residence area, tobacco use, alcohol consumption and attained education were obtained from a questionnaire. Phthalate metabolites were log-transformed and categorized into tertiles. Cross-sectional associations of each metabolite with thyroid hormones were analyzed using multivariable-adjusted linear regression models. The mixture effect of metabolite phthalates was assessed using weighted quantile sum regression. After multivariable-adjustment, the following phthalate metabolites were significantly associated with TT3 in a dose-response manner: MMP (beta = 0.90: 95% confidence interval 0.68,1.12), MEP (beta = 0.67: 0.44, 0.90), MiBP (beta = 0.49: 0.21, 0.77), MiDP (beta = 0.27: 0.03, 0.52), MBzP (beta = 0.51: 0.28, 0.73), MEHP (beta = -0.59: -0.82, -0.35) and MiNP (beta = -0.43: -0.71, -0.14), when comparing highest vs. lowest exposed. The sum of all metabolites was also linked to FT4 levels. No significant associations were observed for TSH except for MiNP. Although phthalate metabolites with different molecular weight showed opposite associations, overall metabolite concentrations seem to associate with increased TT3 and FT4 serum levels. The cross-sectional nature of this analysis limits causal inference.