Association between Serum Perfluorooctanoic Acid (PFOA) and Thyroid Disease in the U.S. National Health and Nutrition Examination Survey

Background Perfluorooctanoic acid (PFOA, also known as C8) and perfluorooctane sulfonate (PFOS) are stable compounds with many industrial and consumer uses. Their persistence in the environment plus toxicity in animal models has raised concern over low-level chronic exposure effects on human health. Objectives We estimated associations between serum PFOA and PFOS concentrations and thyroid disease prevalence in representative samples of the U.S. general population. Methods Analyses of PFOA/PFOS versus disease status in the National Health and Nutrition Examination Survey (NHANES) for 1999–2000, 2003–2004, and 2005–2006 included 3,974 adults with measured concentrations for perfluorinated chemicals. Regression models were adjusted for age, sex, race/ethnicity, education, smoking status, body mass index, and alcohol intake. Results The NHANES-weighted prevalence of reporting any thyroid disease was 16.18% (n = 292) in women and 3.06% (n = 69) in men; prevalence of current thyroid disease with related medication was 9.89% (n = 163) in women and 1.88% (n = 46) in men. In fully adjusted logistic models, women with PFOA ≥ 5.7 ng/mL [fourth (highest) population quartile] were more likely to report current treated thyroid disease [odds ratio (OR) = 2.24; 95% confidence interval (CI), 1.38–3.65; p = 0.002] compared with PFOA ≤ 4.0 ng/mL (quartiles 1 and 2); we found a near significant similar trend in men (OR = 2.12; 95% CI, 0.93–4.82; p = 0.073). For PFOS, in men we found a similar association for those with PFOS ≥ 36.8 ng/mL (quartile 4) versus ≤ 25.5 ng/mL (quartiles 1 and 2: OR for treated disease = 2.68; 95% CI, 1.03–6.98; p = 0.043); in women this association was not significant. Conclusions Higher concentrations of serum PFOA and PFOS are associated with current thyroid disease in the U.S. general adult population. More work is needed to establish the mechanisms involved and to exclude confounding and pharmacokinetic explanations.


Research
The perfluoroalkyl acids (PFAAs) are a family of synthetic, highly stable perfluorinated compounds with a wide range of uses in industrial and consumer products, from stain-and waterresistant coatings for carpets and fabrics to fastfood contact materials, fire-resistant foams, paints, and hydraulic fluids (Organisation for Economic Co-operation and Development 2005). The carbon-fluoride bonds that characterize PFAAs and make them useful as surfactants are highly stable, and recent reports indicate the widespread persistence of certain PFAAs in the environment and in wildlife and human populations globally (Fromme et al. 2009;Giesy and Kannan 2001;Lau et al. 2007;Saito et al. 2004). Two of the PFAAs of most concern are the eight-carbon-chain perfluorooctane sulfonate (PFOS) and perfluorooctanoic acid (PFOA, also known as C8).
Most persistent organic pollutants are lipophilic and accumulate in fatty tissues, but PFOS and PFOA are both lipo-and hydrophobic, and after absorption will bind to proteins in serum rather than accumulating in lipids (Hundley et al. 2006;Jones et al. 2003). The renal clearance of PFOA and PFOS is negligible in humans, leading to reported half-lives in blood serum of 3.8 and 5.4 years for PFOA and PFOS, respectively ).
Human biomonitoring of the general population in various countries has shown that, in addition to the near ubiquitous presence of PFOS and PFOA in blood, these may also be present in breast milk, liver, seminal fluid, and umbilical cord blood .
Extensive laboratory studies of the toxicology of PFOA and PFOS have reported enlargement of the liver, modulation of sex hormone homeostasis, developmental and immune system toxicity, hypolipidemia, and reduced body weight in rodent and nonhuman primate models [reviewed by Lau et al. (2004Lau et al. ( , 2007]. Research interest has focused on the ability of these compounds to bind to nuclear receptors, including the peroxisome proliferator-activating receptor (PPARα), and to disrupt serum protein ligand binding (Luebker et al. 2002), highlighting PFOA and PFOS as potential endocrine disruptors (Jensen and Leffers 2008).
Endocrine systems that may be targets of endocrine-disrupting chemicals include the hypothalamus-pituitary-thyroid (HPT) axis (Boas et al. 2006). Thyroid hormone is essential for the normal physiologic function of nearly all mammalian tissues. Thyroid hormone status is controlled by a well-established feedback mechanism, in which thyroid-stimulating hormone (TSH) stimulates the thyroid to synthesize thyroxine (T 4 ), which is then converted to the biologically active triiodothyronine (T 3 ). The rate of release of TSH is regulated by the hypothalamus as well as by the circulating levels of T 3 and T 4 . Therefore, multiple physiologic steps, including hormone biosynthesis, transport, metabolism, and action on target cells, are required for thyroid hormone homeostasis.
Numerous studies have now shown PFAAs to impair thyroid hormone homeostasis in animal studies. Depression of serum T 4 and T 3 in PFOS-exposed rats has been reported by several authors (Lau et al. 2003;Luebker et al. 2005;Seacat et al. 2003), without the concomitant increase in TSH that would be expected through feedback stimulation. Earlier mechanistic studies of structurally related perfluorodecanoic acid showed that it could reduce serum thyroid hormone levels apparently by reducing the responsiveness of the HPT axis and by displacing circulating thyroid hormones from their plasma protein-binding sites (Gutshall et al. 1989). Although circulating hormone levels were depressed, the activities of thyroid hormone-sensitive liver enzymes were elevated, suggesting that functional hypothyroidism was not occurring. A similar mechanism for PFOS has been hypothesized (Chang et al. 2007). A recent study of the mechanisms involved in PFOS-induced hypothyroxinemia in rats has indicated that increased conjugation of T 4 in the liver, catalyzed by the hepatic enzyme uridine diphosphoglucuronosyl transferase (UGT1A1), and increased thyroidal conversion of T 4 to T 3 by type 1 deiodinase may be partly responsible for the effects (Yu et al. 2009). Taken together, these findings suggest that the effects of PFAAs on thyroid hormone physiology are multiple and complex.
Extrapolations from animal laboratory studies such as these to an estimation of the risks posed by PFOA and PFOS to thyroid function in humans are complicated by the extreme variations reported in their toxicokinetic profile between species (Johnson et al. 1979;). The extremely long half-lives of PFOA and PFOS in humans contrast with the relatively rapid elimination seen in animal models (e.g., serum half-life of PFOS in rats is around 100 days) (Hundley et al. 2006), drawing attention to the potential risks to human health. Disruption to thyroid hormone balance was not found in previous studies of community exposure to PFOA (Emmett et al. 2006;Olsen et al. 2003b) or PFOS . Modest associations between PFOA and thyroid hormones (negative for free T 4 and positive for T 3 ) were reported in 506 PFOA production workers across three production facilities (Olsen and Zobel 2007); there were no associations between TSH or T 4 and PFOA, and the free hormone levels were within the normal reference range.
Given the evidence from animal studies of thyroid hormone imbalance and the varied epidemiologic results from community and occupational exposures, we aimed to explore the hypothesis that higher serum PFOA and PFOS concentrations would be associated with thyroid disease in the general adult population. The environmental chemical biomonitoring program of the U.S. Centers for Disease Control and Prevention, using samples from the U.S. National Health and Nutrition Examination Survey (NHANES), provides large-scale data on serum PFAA concentrations in population-representative samples. In this study we used these data to estimate associations between PFOA/PFOS concentrations and thyroid disease in representative samples of the U.S. general population.

Methods
Study population. Data were from three independent cross-sectional waves of NHANES: 1999NHANES: -2000NHANES: , 2003NHANES: -2004NHANES: , and 2005NHANES: -2006. NHANES surveys assess the health and diet of the noninstitutionalized civilian population of the United States and are administered by the National Center for Health Statistics (NCHS). The study protocol for NHANES was approved by the NCHS Institutional Review Board.
Assessment of PFOA/PFOS concentrations. Solid-phase extraction coupled to high-performance liquid chromatography/turbo ion spray ionization/tandem mass spectrometry with isotope-labeled internal standards was used for the detection of PFOA and PFOS, with a limit of detection of 0.2 ng/mL (Kuklenyik et al. 2005). The laboratory methods and comprehensive quality control system were consistent in each NHANES wave, and documentation for each wave is available (NCHS 2009;NHANES 2006NHANES , 2007. Serum polyfluorinated chemicals (PFCs) were measured in a one-third representative random subset of persons ≥ 12 years of age in each NHANES wave. Data from individuals < 20 years of age were excluded, because questions relating to disease prevalence were asked only for adults.
Disease outcomes. In all NHANES waves, adult respondents were asked about physiciandiagnosed diseases. Associations were examined between PFOA and PFOS concentrations and thyroid disease outcomes. Individuals were asked whether they had ever been told by a doctor or health professional that they had a thyroid problem (in the 1999-2000 survey the questions related to goiter and other thyroid conditions) and whether they still had the condition. We further defined thyroid disease by considering those people who said they currently had thyroid disease and were taking any thyroid-related medication, including levothyroxine, liothyronine, "thyroid desiccated," and "thyroid drugs unspecified" for hypothyroidism and propylthiouracil and methimazole for hyperthyroidism. No details were available on specific thyroid disease diagnosis, and the PFC samples did not overlap with the thyroid hormone measurement subsamples in NHANES.
To assess disease specificity, associations were examined between PFOA and the other NHANES disease categories elicited: ischemic heart disease (combining any diagnoses of coronary heart disease, angina, and/or heart attack), diabetes, arthritis, current asthma, chronic obstructive pulmonary disease (COPD; bronchitis or emphysema), and current liver disease.
Statistical analysis. NHANES uses a complex cluster sample design with some demographic groups (including less-privileged socioeconomic groups and Mexican Americans) oversampled to ensure adequate representation. Prevalence estimates and models were therefore survey-weighted using the NHANES primary sampling unit, strata, and population weights, unless otherwise stated.
Multivariate logistic regression modeling was used to estimate odds ratios (ORs) of thyroid disease outcomes by quartile of PFOA and PFOS concentrations, and associations of other physician-diagnosed diseases. Because thyroid disease prevalence is markedly higher in women, we used sex-specific models. Because the distribution of PFC concentrations is skewed (with most people having relatively low exposures and with considerably more variance at the higher exposure end), all available data were pooled, and PFOA and PFOS concentrations were divided into population-weighted quartiles. Using the Hsieh method (Hsieh et al. 1998), our estimated power to detect an association of OR ≥ 1.8 with current treated disease comparing the top PFOA quartile (Q4) with bottom quartile (Q1) is 67% in women. Combining the lowest two quartiles (Q1 and Q2) into a larger control group provides 80% power. The corresponding minimum detectable effect size in men is OR > 2.9. Assumptions for the power calculations include a significance level of 5% and a multiple correlation coefficient of 0.2 relating PFOA exposure to potential confounders.
Models were adjusted for the following potential confounding factors: year of NHANES study; age; sex; race/ethnicity, from self-description and categorized into Mexican American, other Hispanic, non-Hispanic white, non-Hispanic black, and other race (including multiracial); education, categorized into less than high school, high school diploma (including GED), more than high school, and unknown education; smoking (from self-reported status asked for those ≥ 20 years of age), categorized into never smoked, former smoker, smoking some days, smoking every day, and unknown smoking status; body mass index (BMI; weight in kilograms divided by the square of measured height in meters), categorized into underweight (BMI < 18.5), recommended weight (BMI = 18.5-24.9), overweight (BMI = 25.0-29.9), obese (BMI = 30.0-34.9), and unknown BMI; and alcohol consumption (in adults ≥ 20 years of age, based on responses to the question "In the past 12 months, on those days that you drank alcoholic beverages, on the average day, how many drinks did you have?"), categorized into 0, 1, 2, 3, 4, and ≥ 5 drinks per day, and unknown alcohol consumption. Regression analyses were conducted using STATA/SE (version 10.1; StataCorp LP, College Station, TX, USA).
In analyses of the full sample (men and women), adjusted for age, sex, NHANES wave, and ethnicity, mean levels of PFOA were volume 118 | number 5 | May 2010 • Environmental Health Perspectives higher in men than in women (p < 0.0001), and we found significant differences among ethnic groups (Table 1). Individuals with more education had higher PFOA levels (p = 0.008). Increased alcohol consumption levels were also associated with higher PFOA concentrations (p < 0.0001). We found similar differences in PFOS concentrations. Mean levels of PFOS were higher in men (p < 0.0001), with significant differences in levels between ethnic groups, and individuals with more education had higher PFOS levels (p = 0.008).
Eight individuals did not answer questions about thyroid disease, so the sample size for this analysis was n = 3,966, with 1,900 men, and 2,066 women (Table 2). In women, overall (unweighted) reported (any) thyroid disease was n = 292, and the NHANESweighted but unadjusted prevalence was 16.18%; in men, n = 69, and weighted prevalence, 3.06%. The study-weighted prevalence of current thyroid disease taking medication was necessarily lower (women, n = 163, 9.89%; men, n = 46, 1.18%).
For PFOS concentrations, in women ORs for disease trended in a similar direction but were far from significant. However, in men we found an association comparing those with PFOS concentrations ≥ 36.8 ng/mL (Q4) versus ≤ 25.5 ng/mL (Q1 and Q2): OR for treated disease was 2.68 (95% CI, 1.03-6.98; p = 0.043).
Sensitivity analyses. For a sensitivity analysis, we computed a logistic regression model including both men and women, testing an interaction term between sex and PFOA levels for treated thyroid disease risk. The interaction term was not significant (p-value for interaction = 0.152).
For a post hoc analysis, we examined associations between chemical concentration quartile and any of the other major disease categories covered in NHANES: arthritis, asthma, COPD, diabetes, heart disease, or For PFOS, we found no "positive" associations between higher serum concentrations and higher prevalence of disease. We found one statistically significant "negative" association suggesting that people reporting having COPD may be less likely to be in the highest PFOA concentration quartile (OR = 0.58; 95% CI, 0.43-0.76; p = 0.0003).

Discussion
In this study we aimed to determine whether increased serum PFOA or PFOS concentrations were associated with thyroid disease in a general adult U.S. population sample. The prevalence of thyroid disease is markedly higher in women than in men, so we estimated sex-specific associations. We found that, across all the available data from NHANES, thyroid disease associations with serum PFOA concentrations are present in women and are strongest for those currently being treated for thyroid disease. In men, we also found a near significant association between PFOA and treated thyroid disease. An interaction term analysis suggests that the PFOA trends in men and women are not significantly different, despite the relative rarity of thyroid disease in men. In addition, we found a nominally significant association between PFOS concentrations and treated thyroid disease in men but not in women.
The presence of associations with both PFOA and PFOS raises the issue of how best to perform risk assessments for combinations of perfluorochemicals. The somewhat divergent risk patterns for the two compounds support their separate risk assessment (Scialli et al. 2007), given that current legislative advice (Minnesota Department of Health 2008) is to consider the combined effects of chemicals only when two or more chemicals in a mixture affect the same tissue, organ, or organ system.
Our results are important because PFAAs are detectable in virtually everyone in society (Kannan et al. 2004), with ubiquitous presence across global populations (Calafat et al. 2006). Occupational exposure to PFOA reported in 2003 showed mean serum values of 1,780 ng/mL (range, 40-10,060 ng/mL) (Olsen et al. 2003a) and 899 ng/mL (range, 722-1,120 ng/mL) (Olsen et al. 2003c). Production of PFOS was halted in 2002 in the United States by its principal producer, due largely to concerns over bioaccumulation and toxicity. Since then, voluntary industry reductions in production and use of other perfluorinated compounds, such as the U.S. EPAinitiated PFOA Stewardship Program (U.S. EPA 2006), have contributed to a decreasing trend in human exposure for all perfluorinated compounds (with the notable exception of perfluorononanoic acid) (Calafat et al. 2007; Our results can be compared with previous studies of human populations and of nonhuman primates. A 6-month study of cynomolgus monkeys chronically exposed to PFOA showed no associations between PFOA and thyroid parameters, at mean serum PFOA concentrations higher than those reported in NHANES, although only male monkeys were involved . The largest human study of PFOA centers on an industrial facility in Washington, West Virginia, from which PFOA spread to the population through air, water, occupational, and domestic exposure in a point-source contamination. The C8 Health Project (Steenland et al. 2009) has measured PFOA concentrations in > 69,000 residents. Markedly high concentrations were found, with an arithmetic mean of 83 ng/mL and a median concentration in serum of 28 ng/mL (C8 Science Panel 2008), far higher than the NHANES concentrations in the general population. Preliminary analyses report associations between PFOA and total cholesterol, low-density lipoproteins, and triglyceride concentrations in multivariate models adjusting for age, BMI, sex, education, smoking, alcohol, and regular exercise. Comprehensive crosssectional and follow-on analyses of associations with thyroid disease have not yet been reported but are expected to be released in 2010-2011(C8 Science Panel 2009. Importantly, disruption to thyroid hormone balance was not found in other studies of populations exposed to PFOA, despite the considerably higher levels reported in some studies (Emmett et al. 2006;Olsen et al. 2003b). Emmett et al. (2006) studied 371 residents of a community with long-standing environmental exposure to PFOA and found a median serum PFOA concentration of 181-571 ng/mL but no association between serum PFOA and a history of thyroid disease. A study that included thyroid hormone levels reported a positive association between serum PFOA concentration and T 3 levels in occupationally exposed workers, although there were no changes in other thyroid hormones (Olsen et al. 2001). Modest associations between PFOA and thyroid hormones (negative for free T 4 and positive for T 3 ) were reported in 506 PFOA production workers across three production facilities (Olsen and Zobel 2007); there were no associations between TSH or T 4 and PFOA, and the free hormone levels were within the normal reference range. A linear extrapolation of the findings reported here would be expected to lead to associations being more evident at higher exposure levels, yet this is not supported by the literature. Nonlinearity of response is not uncommon for receptor-mediated systems such as endocrine-signaling pathways that act to amplify the original signal. Large changes in cell function can occur in response to extremely low concentrations, but which may become saturated and hence unresponsive at higher concentrations (vom Saal and Hughes 2005;Welshons et al. 2003).
The mechanisms involved in thyroid homeostasis are numerous and complex, and there are multiple potential targets for disruption of thyroid hormone homeostasis (Schmutzler et al. 2007). These include thyrotropin receptor (Santini et al. 2003), iodine uptake by the sodium iodide transporter (Schröder van der Elst et al. 2003), type 1 5´-deiodinase (Ferreira et al. 2002), transthyretin (Kohrle et al. 1988), thyroid hormone receptor (Moriyama et al. 2002), and the thyroid hormone-dependent growth of pituitary cells (Ghisari and Bonefeld-Jorgensen 2005). Depression of serum T 4 and T 3 has been reported by several authors in PFOSexposed rats (Lau et al. 2003;Luebker et al. 2005;Seacat et al. 2003). One mechanism by which PFAAs may deplete T 4 is through induction of the hepatic uridine diphosphoglucuronosyl transferase (UGT) system, which is involved in hepatic metabolism of thyroid hormone and biliary clearance of T 4 as T 4 -glucuronide (Barter and Klaassen 1994). Because PFOA is an agonist for PPARα, it is plausible that induction of hepatic UGT in PFAA-exposed rats (Yu et al. 2009) could represent a PPARα-mediated response. The involvement of another PPARα agonist, WY 14643, in enhancing the hepatic degradation of thyroid hormone has recently been shown (Weineke et al. 2009).
A growing body of data describes the in vitro binding affinity of PFOA to human serum-binding proteins (Chen and Guo 2009) and to PPARα, -β, and -γ and other nuclear receptors (Vanden Huevel et al. 2006), but the contribution of these mechanisms to PFOA's thyroid-mediating effects in humans remains to be established. Many cellular and metabolic processes, including lipid metabolism, energy homeostasis, and cell differentiation, are controlled by PPARα. Early studies of the effects of PFAAs in rodents showed that a single dose lowered heart rate and body temperature and depressed T 4 and T 3 . Replacement of T 4 did not reverse the clinical symptoms of hypothermia (Gutshall et al. 1988;Langley and Pilcher 1985). Although circulating thyroid hormone levels were low, liver enzymes responsive to thyroid hormone levels were elevated, suggesting that thyroidal homeostasis was not functionally compromised. Chang et al. (2007) found that exposure to PFOS for up to 3 weeks did not affect functional thyroid status, because free T 4 , TSH, and various thyroidresponsive liver enzymes were all unaffected. These findings and later results have led to proposals that displacement of circulating thyroid hormones from plasma protein-binding sites and a reduced responsiveness of the HPT axis contribute significantly towards PFOA's hypothyroid-inducing effects ). Whatever the mechanisms involved, it is clear that more research is merited to clarify the pathways involved.
The feedback mechanism by which the rate of release of TSH and the circulating levels of T 3 and T 4 are regulated tends to show a low level of individual variation (Felt-Rasmussen et al. 1980). Therefore, subtle disruption of the HPT axis within normal reference ranges may have negative health consequences for the individual while remaining within normal reference values, highlighting the importance of including both clinical and laboratory end points in such studies. The NHANES data do not allow specification of the precise type of thyroid disease present, because NHANES does not report on individual hormone levels. PFOA concentration was positively associated with free T 4 and negatively associated with T 3 levels in a cohort of 506 exposed workers, with a near significant association with TSH levels ), although all effects were regarded as modest.
The limitations of these analyses should be noted. We based the PFOA and PFOS measures on a single serum sample. Although PFOA has a half-life of 4 years , so a single sample is likely to represent medium-term internal dose, samples taken at several time points might be more accurate in classifying exposure. Any misclassification from single measures would tend to decrease power and underestimate the real strengths of association. Second, the PFOA concentrations were measured at the same time as disease status, making attribution of causal direction difficult. This raises the possibility of reverse causation. One might hypothesize that after onset of thyroid disease, changes in the nature of exposure or in the pharmacokinetics of PFOA might occur [including patterns of absorption, distribution (including protein binding) or excretion]. Because the associations we report were present in people who were on thyroid hormone replacements, which effectively mimic normal thyroid function, a mechanism for reverse causation through changes in pharmacokinetics is difficult to imagine. Confounding by unmeasured factors is also possible, but it is unlikely that confounding could explain similar findings reported from some of the diverse experimental and observational studies discussed above.
Post hoc association testing with other common diseases (necessarily involving multiple statistical testing) did not identify other robust associations of higher PFC concentration with increased disease prevalence, suggesting specificity of our findings for thyroid disease. An apparent association between higher PFOS concentrations and lower prevalence of COPD requires replication, to exclude a false-positive result from multiple testing.
In addition to the limitations of our analyses, the strengths should also be noted: This is the first large-scale analysis in a nationally representative general adult population of directly measured serum concentrations of PFOA and PFOS. In addition, the associations present are strongest for the most specific identification of thyroid disease, based on reported diagnosis with current use of thyroid-specific medication. The NHANES study also supported adjustment of models for a range of potential confounding factors, which in fact made relatively minor differences to the key estimates, suggesting that the associations are robust.
Further work is clearly needed to characterize the PFOA and PFOS associations with specific thyroid diagnoses and thyroid hormone levels in the general population and to clarify whether the associations reflect pathology, changes in exposure, or altered pharmacokinetics. Longitudinal analyses are also needed to establish whether high exposures predict future onsets of thyroid disease, although concurrent alteration of thyroid functioning would still be a cause for concern.

Conclusions
Higher PFOA and PFOS concentrations are associated with thyroid disease (and being on thyroid-related medication) in the NHANES U.S. general adult population representative study samples. More work is needed to establish the mechanisms underlying this association and to exclude confounding and pharmacokinetic explanations.