Exploratory assessment of perfluorinated compounds and human thyroid function

Physiology & Behavior 99 (2010) 240–245

Contents lists available at ScienceDirect

Physiology & Behavior j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / p h b

Exploratory assessment of perfluorinated compounds and human thyroid function Michael S. Bloom a,⁎, Kurunthachalam Kannan a,b, Henry M. Spliethoff b, Lin Tao a,b, Kenneth M. Aldous a,b, John E. Vena c,d a

Department of Environmental Health Sciences, School of Public Health, University at Albany, State University of New York, Rensselaer, New York 12114, United States Wadsworth Center, New York State Department of Health, Empire State Plaza, P.O. Box 509, Albany, New York 12201, United States Department of Social & Preventive Medicine, 3435 Main St., 182 Farber Hall, School of Public Health and Health Professions, University at Buffalo, State University of New York, Buffalo, New York 14214, United States d Department of Epidemiology and Biostatistics, College of Public Health, Paul D. Coverdell Center for Biomedical and Health Sciences, University of Georgia, Athens, Georgia 30602, United States b c

a r t i c l e

i n f o

Article history: Received 19 November 2008 Received in revised form 3 February 2009 Accepted 4 February 2009 Keywords: Anglers Free thyroxine (FT4) Perfluorodecanoic acid (PFDA) Perfluorinated compounds Perfluorooctanoic acid (PFOA) Perfluorooctanesulfonate (PFOS) Perfluoroundecanoic acid (PFUnDA) Sportfish Thyroid stimulating hormone (TSH)

a b s t r a c t Thyroid hormones play critical roles in human neurodevelopment and adult neurocognitive function. Persistent organohalogen pollutants, such as perfluorinated compounds (PFCs), may interfere with thyroid homeostasis and thus exposures to these compounds might represent risk factors for neurologic and cognitive abnormalities. In this study, serum specimens collected from thirty-one licensed anglers in New York State were analyzed for levels of thyroid stimulating hormone (TSH), free thyroxine (FT4), perfluorodecanoic acid (PFDA), perfluorononanoic acid (PFNA), perfluoroheptanoic acid (PFHpA), perfluorohexanesulfonate (PFHxS), perfluorooctanoic acid (PFOA), perfluorooctanesulfonate (PFOS), perfluorooctanesulfonamide (PFOSA), and perfluoroundecanoic acid (PFUnDA). PFOS and PFOA occurred in the highest concentrations with geometric means of 19.6 ng/mL (95% CI 16.3–23.5) and 1.3 ng/mL (95% CI 1.2–1.5), respectively. In a cross-sectional analysis, no statistically significant associations were detected for PFCs, or their sum, with TSH or FT4 at α = 0.05. However, post hoc power analyses, though limited, suggested that moderate increases in sample size, to 86 and 129 subjects, might facilitate 80% power to detect statistically significant associations for FT4 and PFDA (β = 0.09) and PFUnDA (β = 0.08), respectively. The consumption of sportfish may have contributed to PFDA (r = 0.52, P = 0.003) and PFUnDA (r = 0.40, P = 0.025) levels. This preliminary study does not indicate associations between non-occupational PFCs exposures and thyroid function. However, the possibility for weak associations for FT4 with PFDA and PFUnDA, PFCs measured in low concentrations, is raised. Given the ubiquity of PFCs in the environment and the importance of thyroid function to neurodevelopmental and neurocognitive endpoints, a confirmatory study is warranted. © 2009 Elsevier Inc. All rights reserved.

1. Introduction Suitable circulating thyroid hormone levels are critical for proper human neurodevelopment (reviewed by [1]) and adult neurocognitive function (reviewed by [2]). Alterations of circulating thyroid hormone have been associated with adverse neurodevelopmental endpoints [3] and with behavioral and psychiatric pathologies (reviewed by [4,5]). Persistent organohalogen pollutants (POPs), such as polychlorinated biphenyls for example, may interfere with thyroid homeostasis (reviewed by [6]) and/or neurologic function in children [7] and adults [8]. Perfluorinated compounds (PFCs) comprise a distinct family of POPs which are distributed widely [9], biomagnify in aquatic food chains [10], are highly persistent [11], and may interfere with thyroid function (reviewed by [12]). These compounds are structurally homologous to free fatty acids [13], with which they compete for serum protein binding sites in vivo [14]. Albumin, a serum protein ⁎ Corresponding author. Tel.: +1 518 473 1821; fax: +1 518 402 0380. E-mail address: [email protected] (M.S. Bloom). 0031-9384/$ – see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.physbeh.2009.02.005

found in high relative concentrations in humans, serves as an important regulator of bio-availability, binding approximately 10% of circulating thyroid hormone [15]. Circulating serum albumin also tightly binds PFCs [14]. Several animal studies raise the possibility that PFCs, administered at high doses, may be thyroid toxicants [16–20]. However, criticisms regarding the use of analog methods for the assessment of free hormone in some of these latter studies have been recently raised [21]. A limited number of studies have considered ‘background’, or non-occupational, levels of human exposure to PFCs and thyroid function, and these have been null to date [22,23]. Several publications considering higher levels of human exposure to PFCs, those encountered in occupational settings, have reported statistically, though not clinically, significant associations with thyroid function [24–26]. The aim of the current study was to screen hypotheses regarding potential associations between serum concentrations for eight measured PFCs, and their sum, with levels of thyroid stimulating hormone (TSH) and free thyroxine (FT4) measured in a subsample of New York State Angler Cohort Study (NYSACS) participants. New York

M.S. Bloom et al. / Physiology & Behavior 99 (2010) 240–245

State sportfish consumption is potentially an important source of exposure to PFCs [27] and thus anglers might be considered a suitable population for preliminary studies of non-occupational exposures and thyroid function. The goal of this study was to evaluate the necessity for, and to generate specific testable hypotheses and recommendations for a comprehensive study. 2. Methods 2.1. Sample selection Described elsewhere in detail [28], the NYSACS was a prospective investigation of POPs and potential health effects among 18,082 licensed New York State sportfish anglers and their partners residing in 16 New York State counties contiguous to Lakes Erie and Ontario. The current study considers a subgroup of 31 of 38 NYSACS participants who had completed the 'Dioxin Exposure Substudy' (DES) component, which has also been described in detail in a prior publication [29]. DES participants were recruited to maximize the range of sportfish-consumption-related exposure to POPs contingent on self-reported sportfish consumption and sera concentrations of PCB IUPAC #153 [30]. Between 1995 and 1997, participants completed a detailed exposure questionnaire including queries regarding sportfish and game consumption, lifestyle and demographic factors, and medical history. In addition, a 450 mL sample of blood was donated for toxicologic analysis. Age, gender [31,32], body mass index (BMI) [33], cigarette smoking (reviewed by [34]), history of physician-diagnosed goiter or thyroid condition [35], race/ethnicity [31], the self-reported use of thyroid medication or medications that might alter thyroid hormone concentrations, and the self-reported consumption of sportfish caught from New York State waters in the approximate year prior to the study (6/1/94 to 6/1/95), were considered covariates of potential importance. Informed consent was provided by each participant and this study was approved by the Human Subjects Institutional Review Boards of the State Universities of New York (SUNY) at Albany and Buffalo, and of the New York State Department of Health (NYSDOH). 2.2. Laboratory analysis Blood specimens were collected using a standard protocol and allowed to clot, yielding serum that was used in a previously described series of toxicologic analyses [29]. Remaining specimens were divided into aliquots, labeled, and archived at − 70 °C at the Center for Preventive Medicine at the University at Buffalo. 2.2.1. Thyroid marker analysis In April 2003, following a median 6.5 years (range 5.3–7.6) of storage at − 70 °C, serum aliquots were transported on dry ice to a commercial laboratory for thyroid function marker analyses. TSH was measured using a 3rd generation immunometric chemiluminescent sandwich assay and FT 4 by a competitive chemiluminescent immunoassay. An Immulite® 2000 analyzer (Diagnostics Products Corporation, Los Angeles, CA) immunoassay system, as well as proprietary testing reagents, was employed in determinations. Three quality control samples, at three different hormone concentrations, were included in each sample run. No inter-assay CV exceeded 12.5%. Due to concerns regarding the quality of other measured thyroid hormone markers, as previously described [29], only TSH and FT4 concentrations were available for this study. Stability of these thyroid function markers has been previously reported following more than 10 years of storage at − 25 °C [36]. 2.2.2. Chemical analysis In January 2006, following a median 9.1 years (range 8.1–10.3) of storage at −70 °C, remaining serum aliquots were transported on dry

241

ice to the Wadsworth Center, NYSDOH, for the analysis of PFCs. These included five perfluoroalkyl carboxylates, perfluorodecanoic acid (PFDA), perfluoroheptanoic acid (PFHpA), perfluorononanoic acid (PFNA), perfluorooctanoic acid (PFOA), and perfluoroundecanoic acid (PFUnDA); and three perfluoroalkyl sulfonates, perfluorohexanesulfonate (PFHxS), perfluorooctanesulfonate (PFOS), and perfluorooctanesulfonamide (PFOSA). An ion-pairing extraction procedure was employed using aliquots of approximately 0.5 mL from 31 subjects with sufficient sample volume available. Determinations were conducted using a high-performance liquid chromatograph (HPLC) with electrospray tandem mass spectrometry (ES-MS/MS), as previously described [9,37]. Intra-assay CVs for perfluorobutanesulfonate (PFBS) recovery standards ranged from 5–10% and the mean internal standard recovery was 99.9% (SD 11.9%). Method blanks contained only trace levels of PFHpA and PFOA. Method blank values were subtracted prior to reporting. Limits of detection (LODs) for PFNA and PFOA were calculated as three times the concentrations detected in method blanks (PFNA=0.50 ng/mL and PFOA=0.80). LODs for other analytes were determined as the lowest concentrations on calibration curves (PFDA=0.20 ng/mL, PFHpA=0.15, PFHxS = 0.15, PFOS = 2.00, PFOSA = 0.20, and PFUnDA = 0.02). For PFHpA no values exceeded the LOD, and only 9.7% of values did so for PFOSA. These analytes were thus not further considered. Values below LODs were censored as LOD/√2 prior to data analysis, assuming nondetectable values were normally distributed [38]. 2.3. Statistical analysis 2.3.1. Univariate and bivariate analysis Statistical analysis was conducted using SAS v. 9.1.3 (SAS Institute Inc. Cary, NC). Descriptive statistics are presented for individual PFCs, and their sum (ΣPFCs), as well as TSH and FT4. Normality assessment for continuously distributed variables using the Shapiro–Wilk test indicated the necessity for log transformation of PFCs and TSH prior to parametric analysis. Pearson and Spearman correlation analysis, as appropriate, and analysis of variance (ANOVA) were employed to evaluate bivariate associations between age, BMI, gender, cigarette smoking, TSH, FT4, and PFCs as appropriate. In addition, we employed the self-reported average number of sportfish meals caught from New York State waters, in the approximate year prior to the study (6/1/94– 6/1/95), as a broad marker of sportfish exposure. This variable was presumed to be the most valid among those sportfish data collected during this study, given the short duration between the exposure interval and study completion. Statistical significance was defined as P b 0.05 for a two-tailed test. 2.3.2. Multivariable analysis PFC variables were simultaneously entered as potential predictors, using a forward stepwise selection procedure, for TSH and FT4 as dependent variates, to generate multivariable linear regression models. Separate regression models were also fitted for TSH and FT4 as dependent variates on each PFC variable as a predictor. Covariates of potential importance were included in regression models where bivariate associations were of, at minimum, ‘borderline’ statistical significance (P b 0.10) for associations with a PFC variable, and with FT4 or TSH as appropriate. Influential observations, defined as those with ‘dfbeta’ scores exceeding 2/√n = 0.36 for the PFC coefficient, were excluded and regression analysis repeated [39]. The tenability of statistical assumptions necessary for general linear modeling was evaluated graphically [40]. 2.3.3. Power analysis To facilitate interpretation of study results, post hoc power analysis was conducted for non-statistically significant individual linear regression models using the program PS Power and Sample Size Calculations [41]. Sample sizes required to generate statistically

242

M.S. Bloom et al. / Physiology & Behavior 99 (2010) 240–245

significant slopes equivalent to the observed effect sizes and errors, were estimated at α = 0.05 and β = 0.20. 3. Results 3.1. Study sample All subjects (n = 31), four of whom were women (12.9%), reported race/ethnicity as non-Hispanic white. Mean age was 39.0 years (SD 3.6), with a range 31 to 45, and mean BMI was 25.9 kg/m2 (SD 3.7). Subjects reported a median of 1–6 sportfish meals, caught from New York State waters, from 6/1/94 to 6/1/95. Values ranged from a minimum of zero meals to a maximum of 24–36 such meals during that time interval. Twelve subjects (38.7%) reported cigarette smoking at the time of blood specimen procurement. No history of physiciandiagnosed goiter or thyroid condition was reported, nor was the use of thyroid medication reported. One subject (3.2%) reported the use of Premarin and one subject reported the use of sulfasalazine, medications that might affect circulating thyroid hormone concentrations. Neither of these subjects demonstrated thyroid marker values beyond the reference range of normal, and furthermore neither of these subjects was later identified as ‘influential’ during the linear regression analyses. We thus retained these two subjects in the data set in an effort to maximize sample size and by extension study power.

out by geometric mean values equal to 19.6 ng/mL for PFOS and 0.2 ng/mL for each PFDA and PFUnDA. PFHxS demonstrated the greatest variability with a geometric standard deviation of 2.6 ng/mL and a range of 0.2–4.6 ng/mL. Also presented in Table 1 (i.e., footnotes), age had moderate inverse correlations with PFDA (r = −0.36, P = 0.044), PFNA (r = −0.56, P = 0.001), and PFUnDA (r = −0.41, P = 0.023). The geometric mean PFOA concentration of 1.47 ng/mL (SD 1.37) for males significantly (P = 0.047) exceeded that of 1.05 ng/mL (SD 1.29) for the four female subjects. Statistically significant and positive correlations were detected between the average number of sportfish meals caught from New York State waters, from 6/1/94 to 6/1/95, and PFDA (r = 0.52, P = 0.003), and PFUnDA r = 0.40, P = 0.025). No statistically significant associations were detected between BMI or cigarette smoking and PFCs. As presented in Table 2, strong or moderate correlations were detected between concentrations for: PFOS and PFDA (r = 0.70, P b 0.0001), PFOS and PFNA (r = 0.53, P = 0.002), and PFOS and PFUnDA (r = 0.48, P = 0.006). PFDA was also highly correlated with PFNA (r = 0.65, P b 0.0001), and with PFUnDA (r = 0.75, P b 0.0001), while PFNA was highly correlated with PFUnDA (r = 0.78, P b 0.0001). Correlations for PFOA were of ‘borderline’ statistical significance, with PFNA (r = 0.34, P = 0.061) and PFOS (r = 0.35, P = 0.055). No statistically significant correlations were demonstrated for PFHxS (P N 0.410).

3.2. Thyroid markers 3.4. Regression models and power analyses Median TSH was equal to 1.39 µIU/mL (range 0.43–15.70); one subject (3.2%) exceeded the laboratory reference range for normal (0.40–5.00 µIU/mL). Median FT4 was equal to 1.14 ng/dL (range 0.90– 1.55); no subject exceeded the laboratory reference range for normal (0.80–1.80 ng/dL). A statistically significant correlation was detected between age and TSH (r = −0.45, P = 0.012) as was a ‘borderline’ significant (P = 0.071) association for cigarette smoking and TSH (geometric mean non-smokers = 1.77 µIU/mL vs. smokers = 1.07). 3.3. PFCs For six PFCs, at least 52% of values were measured above the LODs (Table 1). PFOS comprised the majority (83%) of ΣPFCs on average while PFDA and PFUnDA each contributed only a small proportion to this total (1%). This relative concentration profile for PFCs was borne

Table 1 Distribution of perfluorinated compounds (ng/mL serum) measured in 31 New York State Angler Cohort Study participants. PFCs

GM (95% CI)

GSD

Min

Max

% sum

n N LOD

% N LOD

PFDAa PFHxS PFNAb PFOAc PFOS PFUnDAd ΣPFCs

0.21 (0.18–0.26) 0.75 (0.52–1.06) 0.79 (0.66–0.96) 1.33 (1.15–1.53) 19.57 (16.30–23.50) 0.20 (0.17–0.24) 23.69 (20.13–27.89)

1.67 2.64 1.68 1.48 1.65 1.64 1.56

0.14 0.16 0.35 0.57 7.25 0.14 8.65

1.14 4.60 2.08 2.58 76.88 0.92 83.40

1.0 5.2 3.7 6.1 83.0 1.0 100.0

20 31 26 29 31 16 n/a

64.5 100.0 83.9 93.6 100.0 51.6 n/a

CI, confidence interval; GM, geometric mean; GSD, geometric standard deviation; LOD, limits of detection; Max, maximum measured value; Min, minimum measured value; PFCs, perfluorinated compounds; PFDA, perfluorodecanoic acid; PFHxS, perfluorohexanesulfonate; PFNA, perfluorononanoic acid; PFOA, perfluorooctanoic acid; PFOS, perfluorooctanesulfonate; PFUnDA, perfluoroundecanoic acid; ΣPFCs, sum of 6 measured perfluorinated compounds; % sum, proportion of total sum comprised by each compound. a Significant association with age (r = − 0.36, P = 0.044), and the number of sportfish meals caught from New York State waters, 6/1/94–6/1/95 (r = 0.52, P = 0.003). b Significant association with age (r = − 0.56, P = 0.001). c Higher geometric mean concentrations (P = 0.047) in 27 males (1.47 ng/mL) vs. 4 females (1.05 ng/mL). d Significant associations with age (r = − 0.41, P = 0.023), and the number of sportfish meals caught from New York State waters, 6/1/94–6/1/95 (r = 0.40, P = 0.025).

No statistically significant predictors of TSH were selected during a forward stepwise selection procedure, even when the entry/exit criteria were relaxed from α = 0.05/0.10 to α = 0.10/0.15. The results of individual linear regression modeling for TSH are presented in Table 3. No statistically significant effects were measured for PFCs (P ≥ 0.365). Age was entered as a covariate, though it was not of statistical significance, in the linear regression models for PFDA (β = −0.04, P = 0.198), PFNA (β = −0.04, P = 0.439) and PFUnDA (β = −0.04, P = 0.338). Influential observations were excluded from individual linear regression models for TSH on PFDA (n = 2), PFHxS (n = 2), PFNA (n = 3), PFOS (n = 2), PFUnDA (n = 3), and ΣPFCs (n = 2). Post hoc power analysis suggested that sample size increases of approximately 9fold (i.e., for PFDA) to 1110-fold (i.e., PFHxS), with respect to the current number of subjects, would be required to detect statistically significant associations with 80% power at the observed effect sizes (Table 3). As in the stepwise linear regression model for TSH, no statistically significant predictors were selected for FT4, nor were statistically significant associations detected for PFCs as predictors in individual linear regression models (Table 4). The linear regression coefficient for PFDA (β = 0.09) was, however, of borderline statistical significance (P = 0.107) following the exclusion of two influential observations. Influential observations were excluded from the remaining linear regression models for FT4 on PFHxS (n = 2), PFNA (n = 2), PFOA (n = 1), PFOS (n = 2), PFUnDA (n = 1), and ΣPFCs (n = 2). No covariates qualified for entry into the regression models for FT4. Post hoc power analysis suggested that sample size increase of approximately 3-fold (i.e., for PFDA) to 408-fold (i.e., PFOA), with respect to the current number of subjects, would be required to detect statistically significant associations with 80% power at the observed effect sizes (Table 4). 4. Discussion This preliminary study explored potential associations between body burdens of six perfluorinated compounds, and their sum, with TSH and FT4 among 31 licensed New York State anglers, using a crosssectional study design. No statistically significant associations were detected although post hoc power analyses raised the possibility for

M.S. Bloom et al. / Physiology & Behavior 99 (2010) 240–245 Table 2 Pearson correlation coefficients among perfluorinated compounds (ln ng/mL serum) measured in 31 New York State Angler Cohort Study participants.

PFDA PFHxS PFNA PFOA PFOS PFUnDA

PFDA

PFHxS

PFNA

PFOA

PFOS

PFUnDA

1.00 – – – – –

− 0.01 1.00 – – – –

0.65⁎⁎ 0.15 1.00 – – –

0.07 0.15 0.34 1.00 – –

0.70⁎⁎ − 0.05 0.53⁎ 0.35 1.00 –

0.75⁎⁎ 0.08 0.78⁎⁎ 0.16 0.48⁎ 1.00

⁎P b 0.01; ⁎⁎P b 0.001. PFDA, perfluorodecanoic acid; PFHxS, perfluorohexanesulfonate; PFNA, perfluorononanoic acid; PFOA, perfluorooctanoic acid; PFOS, perfluorooctanesulfonate; PFUnDA, perfluoroundecanoic acid.

weak positive associations between FT4, PFDA and PFUnDA, to be detected with moderate increases in sample size of 57 and 99 additional subjects for 80% power, respectively. The latter estimates are, however, limited, given their post hoc nature, the high proportion of values measured below the LODs for these compounds, and the small number of subjects employed for the estimates. A limited but growing body of animal literature suggests the possibility for PFCs as thyroid toxicants. However, the biological effects reported for animal models are frequently at exposures far exceeding those measured in the current study, which are more representative of those experienced by the general U.S. population, as measured during the 2003–2004 National Health and Nutrition Examination Survey (NHANES) [31]. Prior investigators raised the possibility for hypothyroid-like effects, via suppression of thyroid hormone synthesis or increased hepatic clearance of free hormone, in association with PFCs [42]. Reported effects include decreases in T4 and T3 among male PFDA treated rats [43]; increased TSH and decreased T4, FT4, and T3 in PFOA-treated male monkeys [16]; increased TSH and decreased T4, T3 and FT3 in PFOS-treated male and female monkeys [19]; decreased T4, FT4, and T3 among gravid and non-gravid female rats and mice treated with PFOS [20]; decreased T4 and FT4 in rat pups exposed in utero to PFOS [18]; and decreased T4 and T3 among PFOS-treated male and female rats [17]. Recent work has, however, cast doubt on the validity of some of these prior results suggesting that measured decreases in FT4 were secondary to biases introduced via employment of analog assays for free hormone, and that transient increases in circulating FT4, secondary to competition for serum binding proteins, may in fact be the true underlying biologic response to PFC exposure [21].

Table 3 Multiple linear regression models for TSH (ln µIU/mL serum) on perfluorinated compounds (ln ng/mL serum) measured in New York State Angler Cohort Study participants. PFC (n)a

β

95% CI (β) Lo

95% CI (β) Hi

P-value

Required nb

PFDAc (29) PFHxS (29) PFNAc (28) PFOA (31) PFOS (29) PFUnDAc (28) ΣPFCs (29)

0.21 0.01 0.09 − 0.06 0.04 0.19 0.06

− 0.26 − 0.24 − 0.60 − 0.78 − 0.52 − 0.42 − 0.57

0.68 0.26 0.78 0.67 0.59 0.79 0.70

0.365 0.956 0.789 0.871 0.896 0.527 0.844

259 32,211 2,971 8,201 10,158 506 815

CI, confidence interval; PFC, perfluorinated compounds; PFDA, perfluorodecanoic acid; PFHxS, perfluorohexanesulfonate; PFNA, perfluorononanoic acid; PFOA, perfluorooctanoic acid; PFOS, perfluorooctanesulfonate; PFUnDA, perfluoroundecanoic acid; ΣPFCs, sum of 6 measured perfluorinated compounds; TSH, thyroid stimulating hormone. a Sample size varies due to the exclusion of influential observations as described in the text. b Sample size necessary for the detection of the observed effect size (i.e., β) at α = 0.05 and 80% power. c Adjusted for age (years).

243

Studies of exposure to PFCs and thyroid function among humans have been sparse to date, primarily comprised of a series of reports concerning 3M Company employees with occupational exposures to PFOA. A statistically significant (P = 0.002) positive association was reported for serum PFOA and TSH in a 1995 cross-sectional evaluation of 80 male 3M Company employees [25], which was consistent with a noted earlier observed positive association of borderline statistical significance (P = 0.09), in a 1993 cross-sectional study of 111 such employees. More recently, positive associations, adjusted for age, BMI, alcohol consumption, cigarette use, years worked, and occupational category, were reported between cross-sectional serum measures for PFOA (β = 0.016, P = 0.01) and PFOS (β = 0.015, P = 0.04) with log T3 among 518 3M Company employees [24]. Most recently, statistically significant associations, both crude and adjusted, were reported for PFOA with FT4 (β = − 0.01, P = 0.01) and T3 (β = 0.01, P = 0.05) from a cross-sectional study of 506 such employees [26]. The authors of these studies emphasized the lack of clinical relevance in their results, regardless of statistical significance. Median concentrations for PFOA and PFOS (1100 ng/mL and 720 ng/mL) reported in the latter study [26] exceeded those for the current (1.4 ng/mL and 18.4 ng/mL) by approximately 786-fold and 39-fold, respectively. In contrast to associations reported in treated animals and in occupationally exposed humans, no statistically significant associations have been reported to date among subjects with lower exposures to PFCs. A study of subjects with high serum PFOA concentrations in southeastern Ohio, U.S., approximately 97-fold greater than those for the general U.S. population (4.0 ng/mL) [31], reported neither an association with serum TSH (β = 2.13 × 10− 4, P = 0.38) nor a difference (P = 0.30) in PFOA concentrations between 40 subjects with a history of thyroid disease (mean = 387 ng/mL), and 331 subjects without such history (mean = 451 ng/mL) [22]. The median serum PFOA concentration of 354 ng/mL in the Ohio study exceeded that in the current study (1.4 ng/mL) by approximately 253-fold. An earlier study of 15 mother–infant pairs in Japan reported no association between a median 2.5 ng/mL cord blood concentration of PFOS, approximately 13% of the current study median, and newborn TSH or FT4 levels [23]. No statistically significant associations were measured in the current study which is generally consistent with the few prior human studies. However, this preliminary study is the first, to our knowledge, to evaluate PFDA and PFUnDA in association with thyroid function markers as well as with an indicator of New York State sportfish consumption. Recent work suggests that environmental contamination with the latter compounds is increasing globally [44]. It is tempting to speculate that a potential positive association for PFDA and/or PFUnDA with FT4 might result from the displacement of thyroid hormone from albumin binding sites [14], with consequent

Table 4 Linear regression models for free T4 (ng/dL serum) on perfluorinated compounds (ln ng/ mL serum) measured in New York State Angler Cohort Study participants. PFC (n)a

β

95% CI (β) Lo

95% CI (β) Hi

P-value

Required nb

PFDA (29) PFHxS (29) PFNA (29) PFOA (30) PFOS (29) PFUnDA (30) ΣPFCs (29)

0.09 0.02 0.04 − 0.01 − 0.03 0.08 − 0.05

− 0.02 − 0.04 − 0.07 − 0.16 − 0.17 − 0.04 − 0.19

0.21 0.08 0.15 0.14 0.10 0.20 0.10

0.107 0.567 0.424 0.896 0.623 0.204 0.532

86 480 387 12,252 1,062 129 454

CI, confidence interval; PFC, perfluorinated compound; PFDA, perfluorodecanoic acid; PFHxS, perfluorohexanesulfonate; PFNA, perfluorononanoic acid; PFOA, perfluorooctanoic acid; PFOS, perfluorooctanesulfonate; PFUnDA, perfluoroundecanoic acid; ΣPFC, sum of 6 measured perfluorinated compounds; T4, thyroxine. a Sample size varies due to the exclusion of influential observations as described in the text. b Sample size necessary for the detection of the observed effect size (i.e., β) at α = 0.05 and 80% power.

244

M.S. Bloom et al. / Physiology & Behavior 99 (2010) 240–245

increases in circulating free (i.e., biologically available) hormone. A reported increase in mRNA transcripts and activity for the thyroid responsive hepatic malic enzyme element, among PFOS-treated rats, is consistent with the latter premise [21]. However, our post hoc power analysis also indicated the absence of any association for PFDA and PFUnDA with TSH (i.e., required sample size increases of at least 230 and 478 additional subjects for 80% power, respectively). We might speculate that this observation, consistent with prior human literature for other PFCs, indicates accommodation of mild excesses in circulating free thyroid hormone by compensatory mechanisms in the hypothalamus–pituitary–thyroid axis and thus the absence of clinically relevant endpoints [45]. However, the cross-sectional study design, such as that employed here, provides only a ‘snapshot’ in time, and precludes consideration of temporality between exposure and end-point. The data reported in this manuscript, though preliminary in nature, raise the possibility for common source exposures which may contribute particular PFC ‘profiles’. Although the identification of such potential exposure sources is beyond the scope of the current study, the detection of positive, and statistically significant (P b 0.025), correlations of moderate strength (r = 0.40–0.52) between the number of sportfish meals caught from New York State waters, between 6/1/94 and 6/1/95, and PFUnDA and PFDA indicate a potentially important source of PFC exposure. However, this observation may merely reflect our participant sampling strategy and thus this result must be interpreted with caution. Moderate to strong (r = 0.48–0.78), and statistically significant (P ≤ 0.006), correlations among PFDA, PFNA, PFUnDA and PFOS are suggestive of a common exposure pathway for these compounds. There were no statistically significant intercorrelations demonstrated for PFOA or PFHxS, and furthermore relevant point estimates were generally weak (r = 0.02– 0.35). The latter observation raises the possibility of distinct exposure pathways such as, the indoor environment vs. sportfish consumption for PFOA and PFHxS among study subjects. While environmental PFCs may have the potential to affect human thyroid function, there also exist other halogenated environmental compounds, such as the diphenyl ethers, that may alter thyroid hormone synthesis, action, and/or metabolism. We have previously reported a non-statistically significant (P N 0.05), though plausible, positive association between the sum of nine lipid-adjusted polybrominated diphenyl ether congeners (ΣPBDEs) and FT4, sampled from 36 subjects participating in the DES [46]. These data indicated a required sample size increase to 318 subjects to facilitate the detection of a statistically significant association (P b 0.05), with 80% study power, equal to the measured point estimate (β = 0.013). In comparing this prior result with those from the current study, it is notable that point estimates from the latter exceed the prior by several-fold (i.e., βPFDA = 0.09; βPFUnDA = 0.08), and that sample size requirements for adequate power to detect a statistically significant association in the latter are only a fraction of the former (i.e., nPFDA = 89; nPFUnDA = 129). This observation may be indicative of a greater potential for environmental PFC exposures to affect circulating thyroid hormone concentrations in humans than that which exists for environmental PBDE exposures, at the levels experienced by participants in the DES. However, this observation may also reflect differences in laboratory precision when measuring these two disparate families of chemical compounds. Thus the comparison of the current PFC results to the prior PBDE results is only suggestive at best. The DES included only 29 subjects for whom both serum PFC and plasma PBDE data were available, which precluded the simultaneous consideration of these compounds in a single linear regression model with sufficient precision. Several factors, most notably the aforementioned small number of subjects, limit the conclusions that can be drawn from this preliminary study. The availability of only 31 subjects did not permit simultaneous consideration of PFCs in multivariable linear regres-

sion models or the consideration of additional environmental compounds of interest and their potential interactions with PFCs; a general rule of thumb being that approximately 10 observations are required for the unbiased estimation of each predictor [40]. We were thus unable to evaluate the weighted sum of considered PFCs (i.e., in which a regression coefficient is estimated for each PFC) on TSH and FT4, and were limited to the uniform weighting of PFCs in the form of a sum variable (i.e., ΣPFCs). This uniform weighting may be inappropriate given the high intercorrelations demonstrated among many of the measured PFCs. In addition, the censoring of a substantial proportion of values below the LODs for PFDA (35.5%) and PFUnDA (48.4%) may have introduced bias into the linear regression model results and consequently the post hoc power estimates for these compounds [47]. These compounds proved most interesting in terms of post hoc power analysis, however our approach assumes that a larger sample size will demonstrate both the effect estimate, and the variance in the effect estimate generated using the current sample, an unlikely scenario. In addition, we did not measure levels of inorganic ions of relevance to this investigation in biologic specimens, such as fluoride and iodide, thus limiting our ability to interpret the thyroid marker data in the context of potential alterations in iodide metabolism due to fluorinated, or other exogenous compounds. In summary, no statistically significant associations were observed among concentrations of six measured PFCs, with TSH or FT4 in this preliminary study of sportfish anglers. The results of a post hoc power analysis raise the possibility that weak positive associations are extant between FT4 and PFDA and PFUnDA that may be detected employing a larger sample size; although this is only a suggestion and by no means conclusive. Furthermore, exposure pathways to these compounds may be interrelated, with the consumption of sportfish caught from New York State waters potentially providing an important contribution, as well as be related to exposure pathways for other emerging POPs of concern, such as PBDEs. However, given the ubiquity of exposure to PFCs, the possibility that PFDA and PFUnDA levels may be increasing, and the critical role played by thyroid function in human neurodevelopment and neurocognitive function, a larger confirmatory study is warranted to confirm or refute these possibilities. Role of funding sources This work was completed at the University at Albany, State University of New York in collaboration with the New York State Biomonitoring Program at the Wadsworth Center, New York State Department of Health [Centers for Disease Control and Prevention (CDC) National Center for Environmental Health (NCEH) grant U59CCU22339202], and the New York State Angler Cohort Study at the University at Buffalo, State University of New York [Agency for Toxic Substances and Disease Registry (ATSDR), grant H75-ATH 298338, and the Great Lakes Protection Fund, grant RM 791-3021]. The protocol for this study was reviewed and approved by the Institutional Review Boards for Protection of Human Subjects of the State Universities of New York (SUNY) at Albany and Buffalo, and of the New York State Department of Health. Acknowledgements This research was funded in part by the Agency for Toxic Substances and Disease Registry (ATSDR), Grant H75-ATH 298338, the Great Lakes Protection Fund, Grant RM 791-3021, and the Centers for Disease Control and Prevention (CDC) National Center for Environmental Health (NCEH), Grant U59CCU22339202. The authors would like to thank Dr. Adriana Verschoor for technical review of this manuscript and Dr. Edward F. Fitzgerald for his insightful comments.

M.S. Bloom et al. / Physiology & Behavior 99 (2010) 240–245

References [1] Porterfield S, Hendrich C. The role of thyroid hormones in prenatal and neonatal neurological development—current perspectives. Endocr Rev 1993;14:94–106. [2] Bauer M, Goetz T, Glenn T, Whybrow PC. The thyroid–brain interaction in thyroid disorders and mood disorders. J Neuroendocrinol 2008;20:1101–14. [3] Haddow J, Palomaki G, Allan W, Williams J, Knight G, Gagnon J, et al. Maternal thyroid deficiency during pregnancy and subsequent neuropsychological development of the child. N Engl J Med 1999;341:549–55. [4] Whybrow PC, Bauer M. Behavioral and psychiatric aspects of hypothyroidism. In: Braverman LE, Utiger RD, editors. Werner & Ingbar's, the thyroid: a fundamental and clinical text. 9th ed. Philadelphia: Lippincott Williams & Wilkins; 2005. p. 842–9. [5] Whybrow PC, Bauer M. Behavioral and psychiatric aspects of thyrotoxicosis. In: Braverman LE, Utiger RD, editors. Werner & Ingbar's, the thyroid: a fundamental and clinical text. 9th ed. Philadelphia: Lippincott Williams & Wilkins; 2005. p. 644–50. [6] Hagmar L. Polychlorinated biphenyls and thyroid status in humans: a review. Thyroid 2003;13:1021–8. [7] Jacobson J, Jacobson S. Evidence for PCBs as neurodevelopmental toxicants in humans. Neurotoxicology 1997;18:415–24. [8] Fitzgerald EF, Belanger EE, Gomez MI, Cayo M, McCaffrey RJ, Seegal RF, et al. Polychlorinated biphenyl exposure and neuropsychological status among older residents of upper Hudson River communities. Environ Health Perspect 2008;116:209–15. [9] Kannan K, Corsolini S, Falandysz J, Fillmann G, Kumar K, Loganathan B, et al. Perfluorooctanesulfonate and related fluorochemicals in human blood from several countries. Environ Sci Technol 2004;38:4489–95. [10] Kannan K, Tao L, Sinclair E, Pastva SD, Jude DJ, Giesy JP. Perfluorinated compounds in aquatic organisms at various trophic levels in a Great Lakes food chain. Arch Environ Contam Toxicol 2005;48:559–66. [11] Olsen GW, Burris JM, Ehresman DJ, Froehlich JW, Seacat AM, Butenhoff JL, et al. Half-life of serum elimination of perfluorooctanesulfonate, perfluorohexanesulfonate, and perfluorooctanoate in retired fluorochemical production workers. Environ Health Perspect 2007;115:1298–305. [12] Lau C, Anitole K, Hodes C, Lai D, Pfahles-Hutchens A, Seed J. Perfluoroalkyl acids: a review of monitoring and toxicological findings. Toxicol Sci 2007;99:366–94. [13] Luebker DJ, Hansen KJ, Bass NM, Butenhoff JL, Seacat AM. Interactions of fluorochemicals with rat liver fatty acid-binding protein. Toxicology, 2002;176: 175–85. [14] Jones PD, Hu W, De Coen W, Newsted JL, Giesy JP. Binding of perfluorinated fatty acids to serum proteins. Environ Toxicol Chem 2003;22:2639–49. [15] Benvenga S. Thyroxine binding to members and non-members of the serine protease inhibitor family. J Endocrinol Investig 2002;25:32–8. [16] Butenhoff J, Costa G, Elcombe C, Farrar D, Hansen K, Iwai H, et al. Toxicity of ammonium perfluorooctanoate in male cynomolgus monkeys after oral dosing for 6 months. Toxicol Sci 2002;69:244–57. [17] Chang S, Thibodeaux JR, Eastvold ML, Ehresman DJ, Bjork JA, Froehlich JW, et al. Thyroid hormone status and pituitary function in adult rats given oral doses of perfluorooctanesulfonate (PFOS). Toxicology 2008;243:330–9. [18] Lau C, Thibodeaux J, Hanson R, Rogers J, Grey B, Stanton M, et al. Exposure to perfluorooctane sulfonate during pregnancy in rat and mouse. II: postnatal evaluation. Toxicol Sci 2003;74:382–92. [19] Seacat AM, Thomford PJ, Hansen KJ, Olsen GW, Case MT, Butenhoff JL. Subchronic toxicity studies on perfluorooctanesulfonate potassium salt in cynomolgus monkeys. Toxicol Sci 2002;68:249–64. [20] Thibodeaux J, Hanson R, Rogers J, Grey B, Barbee B, Richards J, et al. Exposure to perfluorooctane sulfonate during pregnancy in rat and mouse. I: maternal and prenatal evaluations. Toxicol Sci 2003;74:369–81. [21] Chang SC, Thibodeaux JR, Eastvold ML, Ehresman DJ, Bjork JA, Froehlich JW, et al. Negative bias from analog methods used in the analysis of free thyroxine in rat serum containing perfluorooctanesulfonate (PFOS). Toxicology 2007;234:21–33. [22] Emmett EA, Zhang H, Shofer FS, Freeman D, Rodway NV, Desai C, et al. Community exposure to perfluorooctanoate: relationships between serum levels and certain health parameters. J Occup Environ Med 2006;48:771–9. [23] Inoue K, Okada F, Ito R, Kato S, Sasaki S, Nakajima S, et al. Perfluorooctane sulfonate (PFOS) and related perfluorinated compounds in human maternal and cord blood samples: assessment of PFOS exposure in a susceptible population during pregnancy. Environ Health Perspect 2004;112:1204–7. [24] Olsen G, Burris J, Burlew M, Mandel J. Epidemiologic assessment of worker serum perfluorooctanesulfonate (PFOS) and perfluorooctanoate (PFOA) concentrations and medical surveillance examinations. J Occup Environ Med 2003;45:260–70.

245

[25] Olsen GW, Gilliland FD, Burlew MM, Burris JM, Mandel JS, Mandel JH. An epidemiologic investigation of reproductive hormones in men with occupational exposure to perfluorooctanoic acid. J Occup Environ Med 1998;40:614–22. [26] Olsen GW, Zobel LR. Assessment of lipid, hepatic, and thyroid parameters with serum perfluorooctanoate (PFOA) concentrations in fluorochemical production workers. Int Arch Occup Environ Health 2007;81:231–46. [27] Sinclair E, Mayack DT, Roblee K, Yamashita N, Kannan K. Occurrence of perfluoroalkyl surfactants in water, fish, and birds from New York State. Arch Environ Contam Toxicol 2006;50:398–410. [28] Vena J, Buck G, Kostyniak P, Mendola P, Fitzgerald E, Sever L, et al. The New York Angler Cohort Study: exposure characterization and reproductive and developmental health. Toxicol Ind Health 1996;12:327–34. [29] Bloom M, Vena J, Olson J, Moysich K. Chronic exposure to dioxin-like compounds and thyroid function among New York anglers. Environ Toxicol Pharmacol 2006;21:260–7. [30] Bloom M, Vena J, Swanson M, Moysich K, Olson J. Profiles of ortho-polychlorinated biphenyl congeners, dichlorodiphenyldichloroethylene, hexachlorobenzene, and mirex among male Lake Ontario sportfish consumers: the New York State Angler Cohort Study. Environ Res 2005;97:177–93. [31] Calafat AM, Wong LY, Kuklenyik Z, Reidy JA, Needham LL. Polyfluoroalkyl chemicals in the U.S. population: data from the National Health and Nutrition Examination Survey (NHANES) 2003–2004 and comparisons with NHANES 1999–2000. Environ Health Perspect 2007;115:1596–602. [32] Hollowell J, Staehling N, Flanders W, Hannon W, Gunter E, Spencer C, et al. Serum TSH, T(4), and thyroid antibodies in the United States population (1988 to 1994): National Health and Nutrition Examination Survey (NHANES III). J Clin Endocrinol Metab 2002;87:489–99. [33] Pinkney J, Goodrick S, Katz J, Johnson A, Lightman S, Coppack S, et al. Leptin and the pituitary–thyroid axis: a comparative study in lean, obese, hypothyroid and hyperthyroid subjects. Clin Endocrinol (Oxf) 1998;49:583–8. [34] Bertelsen JB, Hegedus L. Cigarette smoking and the thyroid. Thyroid 1994;4: 327–31. [35] Canaris G, Manowitz N, Mayor G, Ridgway E. The Colorado Thyroid Disease Prevalence Study. Arch Intern Med 2000;160:526–34. [36] Mannisto T, Surcel H, Bloigu A, Ruokonen A, Hartikainen A, Jarvelin M, et al. The effect of freezing, thawing, and short- and long-term storage on serum thyrotropin, thyroid hormones, and thyroid autoantibodies: implications for analyzing samples stored in serum banks. Clin Chem 2007;53:1986–7. [37] Hansen KJ, Clemen LA, Ellefson ME, Johnson HO. Compound-specific, quantitative characterization of organic fluorochemicals in biological matrices. Environ Sci Technol 2001;35:766–70. [38] Hornung R, Reed L. Estimation of average concentration in the presence of nondetectable values. Appl Occup Environ Hyg 1990;5:46–51. [39] Belsley DA. Regression diagnostics: identifying influential data and sources of collinearity. Hoboken: Wiley-Interscience; 2004. p. 292. [40] Kleinbaum DG. Selecting the best regression equation. In: Kleinbaum D, Kupper L, Muller K, Nizam A, editors. Applied regression analysis and other multivariable methods. Pacific Grove: Duxbury Press; 1998. p. 386–422. [41] Dupont W, Plummer W. Power and sample size calculations for studies involving linear regression. Control Clin Trials 1998;19:589–601. [42] Gutshall DM, Pilcher GD, Langley AE. Mechanism of the serum thyroid hormone lowering effect of perfluoro-n-decanoic acid (PFDA) in rats. J Toxicol Environ Health 1989;28:53–65. [43] Langley AE, Pilcher GD. Thyroid, bradycardic and hypothermic effects of perfluoro-n-decanoic acid in rats. J Toxicol Environ Health 1985;15:485–91. [44] Hart K, Kannan K, Isobe T, Takahashi S, Yamada TK, Miyazaki N, et al. Time trends and transplacental transfer of perfluorinated compounds in melon-headed whales stranded along the Japanese coast in 1982, 2001/2002, and 2006. Environ Sci Technol 2008;42:7132–7. [45] Jahnke G, Choksi N, Moore J, Shelby M. Thyroid toxicants: assessing reproductive health effects. Environ Health Perspect 2004;112:363–8. [46] Bloom M, Spliethoff H, Vena J, Shaver S, Addink R, Eadon G. Environmental exposure to PBDEs and thyroid function among New York anglers. Environ Toxicol Pharmacol 2008;25:386–92. [47] Schisterman E, Vexler A, Whitcomb B, Liu A. The limitations due to exposure detection limits for regression models. Am J Epidemiol 2006;163:374–83.