Document 6mjyrxwDmgdD5LjJbnd54nvE

DO]Int Arch Occup Environ Health 10 .1007/s00420-007-0213-0 ORIGINAL ARTICL E P. 1 J Assessment of lipid, hepatic, and thyroid parameters with serum perfluorooctanoate (PFOA) concentrations in fluorochemical production workers Geary W . Olsen Larry R . Zobe l Received : 4 September 2006/Accepted : 1 June 2007 Spri nger-Verlag 2007 Abstract PFOA but not consistently by locations . There were no Objectives Perfluorooctanoic acid (PFOA) results in statistically significant associations obse rv ed between peroxisome proliferator mediated effects in rats and mice PFOA and hepatic enzymes for the three facilities com- resulting in hypolipidemia but not in monkeys . Counter- bined although some modest positive associations were intuitive modestly positive associations between PFOA and observed between PFOA and hepatic enzymes at one of cholesterol levels in production workers have been incon- the three facilities . Analyses of all locations showed no sistently repo rt ed . The purpose of this assessment was to associations with TSH or T4 and PFOA. A negative examine this association in male workers who manufac- association was observed for free T4 and positive associ- tured or used PFOA at three facilities . ation for T3 ; however, the findings were well within these Methods Subjects were male employee volunta ry partic- assays' normal reference ranges . ipants of a fluorochemical medical su rv eillance program Conclusion There was no evidence that employees' ser- who provided blood samples for serum measurement of um PFOA concentrations were associated with total cho- PFOA (perfluorooctanoate) and various lipid, hepatic, and lesterol or LDL . A negative association with HDL was thyroid parameters . Statistical analyses included multiple explained by demographic differences across the three and logistic regression and analysis of covari ance . locations . Several explanations are offered for the incon- Results A total of 506 employees, who did not take sistent tri glyceri de associations with PFOA including both cholesterol-lowering medications (93% of all male par- methodological as well as biological possibilities . ticipants), were analyzed . Serum PFOA concentration s ranged from 0 .007 to 92 .03 g/ml [arithmetic mean Keywords Perfluorooctanoate PFOA Fluorochemicals 2 .21 g/ml ( 95% confidence inte rv al 1 .66-2 .77), median Cholesterol Lipid s 1 .10 g/ml] . Adjusted for age, body mass index, and alcohol usage in regression analyses, PFOA was not sta- tistically significantly ( P > 0 .05) associated with total Introduction cholesterol or low-density lipoproteins ( LDL) . High-den- sity lipoproteins ( HDL) were significantly negatively The ammonium salt of perfluorooctanoic acid (APFO), (P < 0 .01) associated with PFOA for the three facilities which rapidly dissociates in blood to perfluorooctanoate combined but not by individual sites, indicating the overall (PFOA, C7F15COO-), has been used as a processing aid in result was likely a consequence of residual confounding the production of va ri ous fluoropolymers . The widespread due to different demographic profi les at these sites . Serum presence of PFOA in the blood of the general population tri glycerides were significantly positively associated with (Calafat et al . 2006 ) has resulted in assessments of source s of exposure including indust ri al production, indirect sour- G . W . Olsen () L. R . Zobe l Medical Depa rt ment, 3M Company , ces including atmosphe ri c production, and transpo rt of telomer precursors ( De Silva and Maybu ry -1006) . Mail Stop 220-6W-08, St Paul, MN 55144, USA As a manufacturer and user of APFO until its phase-out e-mail : gwolsen @ mmm .com beginning in 2000, the 3M Company (3M) has reported Springer p. 2 Int Arch Occup Environ Healt h several periodic medical surveillance analyses of its fluo- associated with PFOA . Although Kaplan offered no addirochemical production workers at its Antwerp, Belgium ; tional hypotheses or suggested causal models (Hernan et al . Cottage Grove, MN ; and Decatur, AL manufacturing 2002), PFOA has been shown to bind or be carried on blood facilities (Ubel et al . 1980 ; Gilliland and Mandel 1996 ; albumin (Han et al . 2003) and beta-lipoproteins (KerstnerOlsen et al . 1998, 2000, 2003a) . These analyses compared Wood et al . 2003) in the rat, monkey and human ; thus, a workers' clinical chemistry and hormone results in relation positive non-causal correlation remains a possibility . to their serum measurements of either PFOA or perfluo- Re-examination of the Olsen et al . (2003a) analyses rooctanesulfonate (PFOS, C8F17SO3) . suggested some limitations as they would relate to testing In one of the aforementioned analyses, Olsen et al . the hypothesis that PFOA is positively associated with tota l (2003a) reported a positive association between PFOA and cholesterol and its non-HDL sub-fractions . First, the analserum cholesterol and triglycerides among participants of a yses were not stratified by cholesterol lowering medication fluorochemical medical surveillance program conducted in status . A positive association could be masked by inclusion 2000 at the 3M Antwerp and Decatur facilities . They of subjects whose serum cholesterol levels have been reconsidered this a spurious finding given the toxicological duced by medication if such an increase was, in part, evidence in mice and rats that shows PFOA, a peroxisome associated with higher PFOA, concentrations . Second, LDL proliferator alpha receptor agonist (PPARa), exerts effects calculations were not restricted to those instances where including increased beta-oxidation of fatty acids, inhibition serum triglycerides were <400 mg/dl . The potential for bias of the secretion of very low-density lipoproteins and cho- in the LDL calculation steadily increases with higher trilesterol from the liver, and a reduction of cholesterol and glyceride levels (Nakanishi et al . 2000) . Third, the Antwerp triglycerides in serum and accumulation of lipids in the and Decatur facility study concentrated primarily on PFOS, liver (Maloney and Waxman 1999 ; Xie et al. 2003 ; Ken- not. PFOA . Both PFOS and PFOA have been shown to result nedy et al . 2004) . No associations were reported between in hypolipidemia in rats at high concentrations and thus the PFOA and either serum cholesterol or triglycerides for the causal model hypothesized in these earlier analyses conCottage Grove male employees who participated in the centrated on testing for this effect, not hyperlipidemia . Fi- same fluorochemical medical surveillance program in 2000 nally, there was no overall analysis of the three (Olsen et al . 2003c) . manufacturing facilities (Antwerp, Belgium ; Cottage In response to these occupational findings, Costa (2004) Grove, MN ; and Decatur, AL) as the Cottage Grove analreported preliminary cross-sectional analyses of data col- ysis has historically been reported separately (Ubel et al . lected from 2000 to 2004 of approximately 40 (per year) 1980 ; Gilliland and Mandel 1996 ; Olsen et al . 2000, 2003a) Italian fluorochemical production workers . Costa observed due to differences in fluorochemical production activities . a mode~t increase of total cholesterol in workers exposed to The purpose of this study was to analyze the 3M fluoroPFOA . There was no increase of other lipids, including chemical medical surveillance program data collected in triglycerides, but the fraction of non-high density lipopro- 2000 in order to examine the hypothesis that PFOA may be tein (non-HDL) cholesterol appeared elevated . Costa sug- positively associated with increased cholesterol, LDL, and gested his findings might be consistent with the hypothesis triglyceride levels across the three 3M fluorochemical prothat PFOA could influence cholesteryl ester transfer protein duction facilities . Analyses focused on male employees who (CETP), a plasma glycoprotein that facilitates the transfer self-reported that they were not taking cholesterol lowering of cholesteryl esters from apolipoprotein A-containing medications in order to minimize unexplained bias introlipoprotein (HDL) to apolipoprotein B-containing lipo- duced by including participants taking such medication . proteins (Brousseau et al . 2004) . Because PFOS concentrations were not previously associ- Kaplan (2004) summarized a cross-sectional analysis of ated with lowered serum cholesterol (Olsen et al . 2003a), 782 male and 243 female (combined eligible population PFOS was not considered an important potential con1,863) workers with potential exposure to PFOA at the Du- founding variable in these PFOA analyses . Three covariates Pont de Nemours Co . (DuPont) Parkersburg, West Virginia were : age, BMI and alcohol . In addition to the lipids assayed, facility . The highest measured PFOA concentration reported hepatic enzyme and thyroid tests were also analyzed . Serum was approximately 10 g/ml with the average less than 1 g/ triglyceride levels were also considered a potential conml . Statistically significant positive associations were iden- founder with hepatic enzyme analyses (Clark et al . 2003) . tified for total cholesterol, low-density lipoproteins (LDL) and triglycerides with serum PFOA concentrations althoug h the percent variation explained was also low . No association, Materials and methods however, was seen with the HDL fraction thereby not supporting a role between PFOA and CETP . Worker EKG The 2000 fluorochemical medical surveillance program at results and C-reactive protein measurements were also not 3M was available on a voluntary basis to all Antwerp, ~ Springer p. 3 Int Arch Occup Environ Healt h Decatur, and Cottage Grove chemical plant employees and standard (tetra hydro-PFOS) were added . Samples were those with site-wide responsibilities (e .g ., environmental, analyzed by liquid chromatography/tandem mass spec- health and safety workers) . Informed consent was obtained trometry using a PE Sciex API3000 (Applied Biosystems, to measure specific fluorochemical concentrations in the Foster City, CA) . The instrument was operated in the blood of program participants, including PFOA . The multiple reaction monitoring mode under optimized con- number of individuals eligible .(i .e ., potential to have had ditions for detection of PFOA ions formed by turbionspray occupationally related exposure) for the program included ionization . All PFOA measurements were reported above approximately 340 Antwerp, 400 Decatur, and 200 Cottage the lower limit of quantitation (5 .8 ng/ml) . Grove employees . - Statistical analyses included analysis of variance, anal- Clinical chemistry variables considered in the analyses ysis of covariance, logistic regression, and multiple were : alkaline phosphatase (IU/I), gamma glutamyl trans- regression using JMP (Cary, NC) and Stata (College Staferase (GGT, IU/1), aspartate aminotransferase (AST, IU/1), tion, TX) software . Age, BMI and alcohol (drinks per day) alanine aminotransferase (ALT, IU/I), total and direct bili- were considered covariates in the multivariable models . rubin (mg/dl), blood glucose (mg/dI:), cholesterol (mg/dl), For analysis of hepatic enzymes, serum triglycerides were LDL (mg/dl), HDL (mg/dl), and, triglycerides (mg/dl) . also considered a covariate . Distributions were examined to These measurements were-performed at Allina Laboratories assess normality using non-transformed variables . and their (St Paul, MN) . LDL was an indirect calculation using the transformations (log, square root, inverse) . In general, log Friedewald formula (Friedewald et al: 1972) [LDL = total transformations improved normality assumptions of recholesterol - HDL - (triglycerides/5)] when serum trigly= sponse and explanatory variables and , thus multiplicative cerides did not exceed 400 mg/dl . Thyroid "stimulating models (log transformations of explanatory and response hormone (TSH; lU/ml) ; serum thyroxine :(T4; l.tg/dl) ; free variables) were considered applicable . For the log transthyroxine (free T4 ; ng/dl) and serum triiodothyronine (T3 ; formation of alcohol, 0.1 was-added to drinks per day to ng/dl) were also analyzed . These thyroid related hormones prevent the log of 0 . Goodness of,,fit statistics, multicol- were measured by LabCorp (Kansas-City, MO) : T. .hyroidt, linearity, and residual plots were, examined to detect model hormone analyses excluded-five individuals who were al-,, inadequacies . Crude and adjusted odds ratios for PFOA ready diagnosed with thyroid-related conditions and likely categorized by deciles were , determined via logistic taking medication . Results, however, did not diffe .r if these regression analyses,for reference range values of the re- individuals were' included in the analyses . sponse variable . PFOA concentrations were also compared to frequency, A comprehensive report is publicly available elsewhere of the metabolic syndrome in . the study participants . Clin- that details the present study methods and results (Olsen ical identification of the metabolic syndrome .in men in- and Zobel 2006) . cludes any three of the following : waist circumference >40 in . (surrogate BMI >- 30) ; triglycerides -!,150 mg/dl ; HDL < 40 mg/dl ; blood pressure > 130/85 mmHg ; and . Results fasting glucose >- 110 mg/dl (Expert Panel 20.01) . The present study definition of the metabolic syndrome in- A total of 552 employees (Antwerp = 206 ; Cottage cluded any three of-the- following four : BMI > 30 ; trigly- Grove = 131 ; Decatur = 215) participated from these cerides 150 mg/dl ; HDL < 40 mg/dl ; , and fasting facilities representing approximately 60% of Antwerp, glucose 110 mg/dl . I .1 . 50% of Decatur and 65% of Cottage Grove eligibl e At the time of the analyses that were done in 2000-2001 employees . Demographic characteristics of those eligible by Tandem Labs (formerly Northwest Bioanalytical, Salt employees who did not participate are not known although Lake City, UT), the analytical methods used were com- substantial differences between participants and non-par- parable to those reported by Hansen et al . (2001) . The ticipants are unlikely given the routine practice of offering analytical method consisted of a liquid :liquid extraction voluntary medical surveillance programs at these facilities procedure followed by evaporation and reconstitution of for more than 20 years . the extract residue with 30 :70 20 mM ammonium acetate Of the 552 employee participants at these 3 facilities, in water :20 mM ammonium acetate in methanol (v/v) . 506 (92%) self-reported that they did not take cholesterolEight or more calibration standards were prepared on the lowering medications : 196 (95%) for Antwerp, 122 (93%) day of each run by adding 100 l of blank human serum forCottage Grove, and 188 (87%) for Decatur . Results will and 400 l of 50 mM ammonium acetate in water to focus on these 506 participants . polypropylene tubes . After a brief vortex mixing, 10 .0 N1 Presented in Table I are measures of central tendency of the appropriate spiking solution was added . In addition, for the 506 participants for PFOA, PFOS, demographic blank serum samples, both with and without internal factors and the clinical chemistries . The number and per- 1L) Springer p. 4 Int Arch Occup Environ Heath 0 r- r ^ N M O ~ 00 ~ N C7, 10 O~ O 00 r-~ N d " r~~ ~, M m O O N N M N ^ O O O O 4 O O R C ~ N --~ O V N O O N ^ V a, N CM O N C C ~ ~ m _ N ~ O C~ d' M N O Q ' Z ~ ^ O O, V O V M ^ N o0 O N 7 O C VI o~ ^ N NN O N 7 a a a a rj y A 01 Q, cN 00 N ~ N 00 C1 Oo ~M ~D O~ 'p 00 d' O O\ N V h N M O m O O oo 9 N ^ c m O ON 00 N ~ M N V Iq 00 0 ,4 rl~ N V'i .~ O I~ N O_ 7 v1 O~ v1 M N M N h [~ O~ M N O~ M l 00 w M [~ O1 Ir 00 00 M O Cl O N ~ kn ~ M N N --~ I~ O O O V O ~ P O O O O U O CM T N 't~ v'1 v'1 v7 O O OV N O ~D v1 N O~ N ! ~O N N C7, N ^ N O N O C O -' N N t- ^ W) 00 O~ N > ~ ~ ~ ~ M O~ N ^ M ~ ~ O N M O V ~ ^ O V] O O, C' O N M f+l oo O M O O O M OA CIS m \0 4 ti o. U O O IR oq :-. O O oro 'n v1 O~ ,.-, ~O~ ^ V O~ N U U G d' O ~ N O N ' ~O N M O M O O N h ^ ~ U V (` .,M 1 'LJ O N 00 V M ,_ nM Cv O V1 M M N .- .-- -l~ ~ O l11 .rl ^ a0 [~ .- 7_ ~_ M ~ U N 00 I (`II pI L 00 ' ~G d O M N "' r C cy O ~ O N N ~ ~ MN O N I M N o^o 00 L~ O O V O [Y O O O O ~ 'A 00 d ` ~O 7 C M O N_ ^ O orN O^ t~ O N m pp O O M N OCO N ~!1 ~D N N ^ ^-~ O --~ O p b II Q O O~ Uj Z c/] --~ O ~ m O O ~ M --~ M ~00 O ~O ~D V N O ., O O O ^ ^ O o ,5 ~ a ~ p r ~ ~ r_ o c a ~ a o a o r- er N O cn cn m : a N =M C7 00 N Vl \,o N N ^ N -O N coOMN -L ~Co a 3 O M C O N M m A A O , .-- ~n ^ ~D O ~O Q~O` cn ~ _O _ M M r Np + Lrl ~No ~ OC d U A C U ~NOI ~ vNi ~p~I N M N N m o'o I O ~O d ~ r~ O N O oo L '[7 cCy O O N ~ ~ M O M--+ N N ^ ~ rj O O V O r O C ~e w b ~ ~ w 7 N a7 G) w 7 '0 O N E a ~ 7 c ~ t ~ M N M r"6- M .-. .-0 . ~ ~0 O N V' V ~O ~O~ N o0 6~ N O C7, N ~ ID N N O N O O - M N OOO ^ Q O O VV V C t- _ O ~Y Q~ O\ ~ ~ ~ pp m ^ ^ N M O V V4 N M 4 p II Q ~D O O~ ~'1 O M ^ l~ O N O O M O N ~ ~, ~., O b v z n p F cO N O O r ~O .4 cV ^ d' N O CN .-. 'IT ~o C\ O~ N V ~D N CN 00 N ~ UU 0 M O N O C N oo ^ pC'q pCp pC'p C d) T _T O O a tl ~ F ^_ U U T a U r U Ln O N U f (yd Yd O O ' Q Q . C k vFi .Fa vF ~ o~~i v v M CIA cin v a ~ d W d C7 U1 x E- d d C7 E- A F E- F y a ~ Springer ~ p. 5 Int Arch Occup Environ Healt h cent of employees by specific reference points for demo- and 13% among the Cottage Grove and Decatur employgraphic factors and clinical chemistry findings are pre- ees, respectively . sented in Table 2 . Besides their younger age, there were Serum PFOA concentrations for the 506 employees substantially fewer Antwerp employees (5%) who would ranged from 0 .007 to 92 .03 g/ml [arithmetic mean be categorized as obese (BMI ? 30) compared to Cottage 2 .21 g/ml (95% confidence interval 1 .66-2 .77), median Grove (44%) or Decatur (32%) employees . More than 40% 1 .10 g/ml] . PFOA results are categorized by deciles in of Antwerp and Cottage Grove employees reported drink- Table 3 . Mean PFOA decile concentrations ranged from ing one alcoholic beverage per day or more compared to the lowest decile of 0 .06 g/ml (range 0 .007-0.13 g/ml) only 1% of Decatur employees . There were regional cul- to the highest decile of 112 .15 g/ml (range 3 .71-92 .03 g/ tural/lifestyle practices that likely account for this differ- ml) with corresponding median values of 0 .06 and 4 .94 g/ ence in alcohol consumption . Other statistically significant ml, respectively . differences between the three locations were observed for The distributions of demographic factors by PFOA blood glucose, HDL, triglycerides, alkaline phosphatase, decile are presented in Table 4 . There were higher perALT, total bilirubin, and the `metabolic syndrome' . The centages of Antwerp employees in the first three deciles higher percentage of Antwerp male employees with total and lower percentages in the highest two deciles (9 and bilirubin >1 .5 mg/dl has been attributed to the likely 10) . Likewise, there were lower percentages of Decatur greater prevalence of Gilbert's . syndrome (Olsen et al . employees in the first three deciles and higher percentages 1999) . Only 1% of the Antwerp employees were catego- in the uppAr deciles (6--10), in particular decile 9 . Because rized as having the `metabolic syndrome' compared to 20 of these disparities, mean BMI values were greater in the Table 2 Number of employees (%) by location for reference Antwerp Cottage Grove Decatur P points of demographic factor s (N = 196) (N = 122) (N = 188) valuea and clinical chemistry results Age < 40 years 127 (65) 57 (47) 76 (40) 0 .0001 BMI ? 30 9 (5) 54 (44) 61 (32) 0 .0001 Alcohol ,2: I drink/day 92 (47) 49 (40) 2(1) 0 .0001 Glucose ? 110 mg/dl 10(5) 22(18) 16(9) 0 .001 Cholesterol ? 200 mg/dl 119 (61) 69 (57) 114 (61) 0 .72 Cholesterol >_ 240 mg/dl 60 (31) 32 (26) 50 (27) 0 .60 HDL < 40 mg/dl 16 (8) 36 (30) 72 (38) 0 .0001 LDL >- 130 mg/dl 106 (55) 53 (48) 100 (55) 0,40 Triglycerides >- 150 mg/dl 46 (23) 58 (48) 103 (55) 0 .0001 Alkaline phosphatase ? 120 IU/I 0(0) 0(0) 5 (3) 0 .01 AST ? 40 IU/1 4(2) 8 (7) 11 (6) 0 .09 ALT ? 40 IU/1 14 (7) 34 (28) 51 (27) 0 .0001 GGT ? 40 IU/1 25 (13) 27 (22) 34 (18) 0 .09 Total bilirubin > 1 .5 mg/dl 13 (7) 1 (1) 0(0) 0 .0001 Direct biliri3bin > 0 .4 mg/dl 0(0) 0(0) 2(1) 0 .18 TS H <0 .35 IU/ml 1 (1) 1 (1) 1 (1) 0.92 >5 .5 IU/ml 5 (3) 5 (4) 8 (4) 0 .60 T4 <4 .5 g/dl 1 (1) 0(0) 0(0) 0 .46 >12 .0 g/dl 0(0) 0(0) 0(0) 1 .0 Free T 4 <0 .70 ng/dl 1 (1) 0(0) 2 (1) 0 .49 >1 .53 ng/dl 3 (2) 1(1) 1(1) 0 .60 T3 Percentage values are given in <60 ng/dl 0 (0) 0(0) 0(0) 1 .0 parentheses >181 ng/dl 2 (1) 3 (3) 2 (1) 0 .46 Chi square test of significance Metabolic syndromeb 2 (1) 24 (20) 25 (13) 0 .0001 See text for description Springer p. 6 Int Arch Occup Environ Healt h Table 3 Number of employees, arithmetic mean, 95% confidence significant coefficient in the regression models for total interval (95% CI), median and range, by PFOA decile cholesterol or LDL . In the regression model for HDL that PFOA PFOA contained all three locations, the adjusted PFOA coefficien t was statistically significant (P = 0 .01) but only explained Decile N Mean 95% Cl Median Range 1% of the variance in the model . Separate models analyzed 1 51 0 .06~' 0 .05-0 .07 0 .06 0 .007-0 .13 for each of the three locations showed no statistically sig- 2 51 0 .20' 0 .19-0 .21 0 .19 0 .13-0 .29 nificant findings between PFOA and HDL . The model for 3 51 0 .36" 0 .35-0 .37 0 .36 0 .30-0.44 triglycerides for the combined three locations indicated a 4 50 0 .55" 0 .53-0.57 0 .54 0 .44-0.71 statistically significant positive coefficient (P < 0 .0001) for 5 51 0 .91' 0 .87-0.94 0.91 0.72-1 .10 PFOA which explained approximately 4% of the variance 6 49 1 .26' 1 .23-1 .28 1 .25 1 .11-1 .40 of the response variable . Stratified by location showed 7 50 1 .64i 1 .60-1 .67 1 .63 1 .42-1 .85 Antwerp with a statistically significant positive PFOA 8 54 2 .17' 2 .12-2 .23 2 .18 1 .86-2 .50 coefficient (P < 0 .0004) and Decatur had a marginally 9 49 3 .00" ' 2 .90-3 .10 2 .96 2 .51-3 .69 positive coefficient for PFOA (P = 0 .07) . No statistically 10 50 12.15a-' 7 .21-17 .10 4 .94 3 .71-92 .03 significant coefficient for PFOA (P = 0 .38) was found with triglycerides for the Cottage Grove location . '' Statistic.a,ll1y s0ig,nirfiecasnptlye(cPt<iv0 .e05l)ydifPferrenet stheannPtFeOdA diencileT(s)able 6 are the mean and 95% confidence 1, 2, 3, . . intervals (CI) values for cholesterol, LDL, HDL, and tri- glycerides by PFOA deciles, adjusted for age, BMI, and upper deciles and alcohol consumption was lower, alcohol . Mean serum triglyceride levels were highest in the reflecting the demographic differences seen across the upper three PFOA deciles . three facilities in Tables I and 2 . Table 7 presents adjusted odds ratios and 95% confiPresented in Figs . 1 and 2 are scatterplots of the natural dence intervals by reference range cutoff points listed in logs of the blood lipids and PFOA . Pearson correlation Table 2, using the lowest PFOA decile as the reference . coefficients and P values (in parentheses) between PFOA Serum PFOA concentrations were not associated, posi- and the lipid parameters were : cholesterol (r = 0 .05, tively or negatively, with cholesterol or LDL . For P = 0 .32), LDL (r = 0 .006, P = 0 .90), HDL (r =--0 .17, HDL < 40 mg/dl, the non-adjusted odds ratio for the P < 0 .0001), and triglycerides (r = 0 .21, P < 0 .0001) . highest decile was 3 .0 (95% CI 1 .2-7 .5) . Adjusted for age, Unadjusted and adjusted coefficients for PFOA when re- BMI and alcohol, the highest decile odds ratio became 2 .6 gressed with total cholesterol, LDL, HDL, and triglycerides (95% CI 1 .0-6 .8) (Table 5) . Adjusted also for location, this are presented in Table 5 . Analyses are also presented for od.ds ratio became 1 .8 (95% CI 0 .7-4 .8) . Adjusted odds each location in Table 5 . PFOA was not a statistically ratios for triglyceride levels _>_150 mg/dl were also highest Table 4 Distribution of demographic factors by PFOA decil e PFOA Antwerp Cottage Grove Decatur Age BMI BMI >- 30 Drinks/day >_l drink/day Decile N(%o) N(%) N(%) Mean (SD) Mean (SD) N(%) Mean (SD) N(%) 1 28 (55) 17 (33) 6(12) 39 (11) 27 .1 (4 .8) 13 (25) 0 .7 (0 .7)' 21 (41) 2 25 (49) 13 (26) 13 (25) 41 (9) 26 .7 (4.1)' 7 (14) 0 .7 (0 .9) 17 (33) 3 28 (55) 11 (22) 12 (23) 38 (8) 26 .4 (3 .9)'' 8(16) 0 .7 (1 .0)' 15 (29) 4 20 (40) 13 (26) 17 (34) 38 (10) 27 .8 (4 .3) 13 (26) 0 .7 (1 .0)' 13 (26) 5 22 (43) 11 (22) 18 (35) 40 (10) 27 .2 (5 .0) 8 (17) 0.7 (1 .0)' 16(32) 6 21 (43) 7 (14) 21 (43) 40 (9) 27 .4 (4 .8) 12 (24) 0 .7 (1 .0)' 14 (29) 7 20 (40) 9 (18) 21 (42) 39 (9) 26 .9 (3 .9)' 10 (20) 0 .7 (0 .9)' 14 (28) 8 18 (33) 11 (20) 25 (46) 41 (10) 27 .5 (5 .4) 17 (31) 0 .6 (0 .9)' 16 (30) 9 9 (18) 7 (14) 33 (67) 39 (9) 28 .9 (4 .9)e' g 21 (43) 0 .3 (0 .5)-h 5 (10) 10 5 (10) 23 (46) 22 (44) 39 (9) 28 .3 (4 .0)` 15 (30) 0 .5 (0 .7) 12 (24 ) P<0 .0001k P < 0 .03' P<0 .14` '-' Statistically significantly (P < 0 .05) different than PFOA decile 1, 2, 3, . . ., 10, respectivel y k' 1, ` P value associated with Chi square tesi for location, BMI distribution and ?i drink/day, respectively 1 Springer f/ p. 7 Int Arch Occup Environ Health L Ne r f':?ftt7r.' jd ;, . ' . '_ ; s rs .~ , .''t.:t'.~ ~~ j~._ ~:'s'i'' _ .. . 6.0 J .. x : . 'to ' t S . ,tr~: h ..;: . ,, . . :' ~~~~"~: .ts~ ~ iN w~ . ~ , , :~ , , -6 .0 -4 .0 -2.0 0.0 In 6. 9 a~~ , 2 .0 4.0 6 .0 PFOA -6.0 -4.0 2 .9 -2.0 0.0 In PFOA 2 .0 4 .0 6 .0 i:~: ~ r~~,~ ~_ ~ ~' . ,~~ . . - , . 1~~.; H , . . . J ~ 4 .0 D ,1 , 6~ A .~ , , ~ ta~ C . . ' T = N = y~ ~_ `j . . Ol ',~ Z 1'-~ .I M . ~ ~,~ ~ ~ J1'~ s ` ~~ , - . I .~ ~: , I~~ } Tr . -6.0 -4 .0 -2 .0 0 .0 2 .0 4 .0 6 .0 In PFOA -6 .0 -4 .0 -2 .0 0.0 2 .0 4 .0 6.0 Fig. 1 Scatterplots of the natural log PFOA (ng/m.l) log total cholesterol and I,D]~ - by the A natural,,, In PFO Fig, 2 Scatterplots of the natural log PFOA (ng/ml) by the natural log HDL and triglycerides , for PFOA deciles 8-10 as seen with adjusted mean tri- . No discernable_ trends in adjusted mean hepatic enzyme glyceride values reported in Table 7 . . . values were apparent when analyzed by deciles (Table 9) . Presented in Figs . 3 and 4 are scatterplots of natural logs There were no statistically significant odds ratios (non- of hepatic clinical chemistries by PFOA . Pearson correla- adjusted or adjusted) for the reference points AL- tion coefficients and P values (in parentheses) between T>- 40 IU/1 or GGT >_ 40 IU/1 when compared to decile I PFOA and the hepatic parameters were : alkaline phos- (Table 10) . Odds ratios for alkaline phosphatase, AST, and phatase (r = 0 .08, P = 0 .06), AST (r = -0 .01 P = 0 .83), total bilirubin are not shown because of the failure of the ALT (r = 0 .005, P < 0 .005), and GGT (r = 0.02, logistic models to converge because of the very few data P < 0 .02) . Analyses presented in Table 8 of the regression points that were out-of-reference range (Table 2) for these models showed marginal statistically significant coeffi- clinical parameters . cients for PFOA when adjusted for age, BMI, and alcohol Presented in Figs . 5 and 6 are scatterplots of natural logs for ALT (P = 0.06) and GGT (P = 0 .05) but not when . of thyroid hormones by PFOA . Pearson correlation coefadjusted for age, triglycerides, and alcohol for ALT ficients and P values (in parentheses) between PFOA and (P = 0 .40) and GGT (P = 0 .55) . Total bilirubin was sta- the thyroid hormones were : TSH (r = 0 .08, P = 0 .07), T4 tistically significantly negatively associated with PFOA (r = -0 .04, P = 0 .32), free T4 (r =-0 .14, P < 0 .002), and regardless of covariates used . Stratified by location, T3 (r = 0 .07, P = 0 .11) . For all locations combined, there Decatur had marginally statistically significant positive were no statistically significant adjusted coefficients for coefficients, regardless of covariates adjusted, ranging be- PFOA in the models except for Free T4 (negative coeffi- tween 0 .01 < P < 0 .07 for alkaline phosphatase, ALT, cient for PFOA) and T3 (positive coefficient for PFOA) GGT, and total bilirubin . The amount of variance of the (Table 11) . However, the full models only explained 5 and hepatic response variables explained by PFOA in these 2% of the variance of free T4 and T3, respectively . The models was minimal ranging from <1 to 3% . No statisti- - thyroid results were considered not clinically relevant as cally significant PFOA coefficients for the hepatic enzymes results from these models were well within normal refer- were observed in either the Antwerp or Cottage Grove ence ranges for the four thyroid parameters when serum regression models . PFOA concentrations were predicted in a model to rang e ~L) Springer p. 8 Int Arch Occup Environ Hea ;~n Table 5 Non-adjusted and Ln PFOA Ln PFO A adjusted natural log (In) PFOA coefficients with In lipid Non-adjusted Adjusted " measurements Coefficient SE P value Coefficient SE P value Ln cholestero l All locations 0 .0059 0 .0060 0 .32 0 .0076 0 .0059 0 .20 Antwerp 0 .0051 0 .0106 0 .63 0 .0130 0 .0096 0 .18 Cottage Grove 0 .0034 0 .0089 0 .70 0 .0021 0 .0100 0 .83 Decatur 0 .0221 0 .0139 0 .11 0 .0266 0.0141 0 .06 Ln LDL All locations 0 .0012 0 .0089 0 .89 0 .0021 0.0090 0 .81 Antwerp -0.0037 0 .0157 0 .81 0 .0106 0.0147 0 .47 Cottage Grove -0 .0022 0 .0139 0 .87 0 .0049 0.0145 0 .73 Decatur 0 .0258 0 .0199 0 .20 0 .0302 0.0200 0 .13 Ln HDL All locations -0 .0307 0 .0079 0 .0001 -0 .0183 0.0069 0 .01 Antwerp -0 .0057 0 .0136 0 .68 -0 .0095 0.0131 0 .47 Cottage Grove -0 .0153 0 .0122 0 .21 -0 .0192 0.0120 0 .11 Decatur -0 .0256 0 .0149 0 .09 -0 .0207 0.0141 0 .14 Ln triglycerides All locations 0 .0892 0 .0185 0 .0001 0 .0711 0.0169 0 .0001 Antwerp 0.0840 0 .0288 0 .004 0.0980 0.0270 0 .0004 Cottage Grove 0.0343 0 .0316 0 .28 ' See study- methods. Adjusted Decatur 0 .0715 0 .0400 0 .08 for Ln age, Ln BMI, Ln alcoho l 0.0280, 0.0314 0 .3 8 0 .0689 0.0376 0 .07 between 0 .005 and 100 g/ml (Table 12) . Presented in adjusted mean for the highest decile for free T4 was staTable 13 are the mean thyroid hormone-related values tistically significantly lower than that of the first decile . adjusted for age, BMI and alcohol . The mean TSH value in There were no statistically significant differences -between the fourth decile of Table 13 is influenced by one subject decile-adjusted means for T3 in Table 13 . Because so few whose TSH value was 65 .3 IU/ml (see overall results in thyroid values were out-of-reference range (Table 2), the Table 1) who was not diagnosed at the time as hypothy- findings from the logistic analyses are not presented beroid . If removed, the mean of the fourth decile became cause of the models' lack of convergence . 2 .15 lU/ml and the decile . statistically significant differ- PFOA was not associated with the metabolic syndrome . ences with this fourth decile in Table 13 disappear . The Age-adjusted odds ratios (95% CI in parentheses) for the Table 6 Adjusted mean and 95% confidence intervals (95% CI) for lipid clinical chemistry results, by PFOA decil e PFOA Cholesterol LDL HDL Triglycerides Decile Mean 95% Cl Mean 95% Cl Mean 95% Cl Mean 95% Cl 1 214 203-225 137 127-147 50' 46-53 145' 116-173 2 211 199-222 135 125-145 51' 48-54 124''"' 95-153 3 209 198-220 128 118-138 51i 48-54 153i 124-182 4 210 199-222 133 123-143 50' 46-53 145' 116-175 5 217 206-228 140 130-150 48 . 45-51 162i 133-191 6 218 206-229 141 130-151 48 45-51 160i 131-190 7 214 203-225 133 123-143 50 47-53 158' 128-187 8 215 204-226 136 126-146 47 44-50 172 144-201 9 221 210-233 140 130-150 48 44-52 165i 135-194 10 216 204-227 133 123-144 44a' 41-47 208-, 179-238 Adjusted for age, BMI and alcoho l ' Statistically significantly (P < 0 .05) different than PFOA decile(s) 1, 2, 3, . . ., 10, respectively 1~ Springer p. 9 Int Arch Occup Environ Healt h Table 7 Adjusted odds ratios (OR) and 95% confidence intervals (95% CI) for lipid clinical chemistry reference points, by PFOA decile PFOA Chol ? 200 mg/dl LDL ? 130 mg/dl HDL <- 40 mg/dl Triglycerides ? 150 mg/dl Decile ORa 95% Cl ORb 95% CI OR' 95% Cl OR 95% Cl ORa 95% CI ORb 95% Cl ORa 95% CI ORb 95% Cl 1 1 .0 - 1 .0 - 1 .0 - 1 .0 - 1 .0 - 1 .0 - 1 .0 - 1 .0 2 0 .4 0.2-1 .0 0 .4 0 .2-1 .0 0 .7 0 .3-1 .6 0 .7 0 .3-1 .7 1 .1 0 .4-3 .2 1 .0 0 .4-3 .0 0 .7 0.3-1 .8 0 .6 0 .2-1 .6 3 0 .9 0.4-2.0 0 .9 0 .4-2 .0 0 .7 0 .3-1 .6 0 .7 0 .3-1 .6 0 .5 0 .1-1 .5 0 .4 0 .1-1 .3 1 .0 0.4-2 .4 0 .9 0 .4-2 .2 4 0 .9 0.4-2 .1 0 .9 0 .4-2 .0 0 .8 0 .4-1 .8 0 .8 0 .4-1 .9 2 .0 0 .8-5 .4 1 .7 0 .6-4 .7 1 .3 0.5-3 .1 1 .1 0 .4-2 .6 5 1 .7 0.7-4 .0 1 .6 0 .7-3 .9 1 .4 0 .6-3 .1 0 .4 0 .6-3 .2 1 .0 0 .4-2 .9 0 .8 0 .3-2 .4 1 .2 0.5-3 .0 1 .0 0 .4-2 .5 6 1 .0 0 .4-2 .2 0 .9 0 .4-2 .1 0 .9 0 .4-2 .0 0 .9 0 .4-2 .1 1 .7 0 .7-4 .7 1 .4 0 .5-3 .9 1 .7 0.7-4 .0 1 .3 0 .5-3 .2 7 0 .8 0 .4-1 .9 0 .8 0 .3-1 .8 0 .7 0 .3-1 :5 0 .7 0 .4-1 .6 0 .6 0 .2-1 .8 0 .5 0 .1-1 .4 0 .9 0.4-2 .2 0 .7 0 .3-1 .8 8 1 .0 0 .4-2 .2 0 .9 0 .4-2 .1 1 .1 0 .5-2 .4 1 .1 0 .5-2 .5 1 .4 0 .5-3 .7 1 .0 0 .4-2 .8 2 .7 1 .2-6 .5 2 .1 0 .9-5 .2 9 1 .4 0 .6-3 .3 1 .3 0 .6-3 .1 1 .2 0 .5-2 .7 1 .2 0 .5-2 .8 0 .9 0 .3-2 .4 0 .6 0 .2-1 .7 2 .4 1 .0-5 .9 1 .7 0 .7-4 .3 10 1 .1 0 .5-2 .6 1 .1 0 .5-2 .6 1 .2 0 .5-2 .8 1 .4 0 .6-3 .3 2 .6 1 .0-6 .8 1 .8 0 .7-4 .8 2 .4 1 .0-5 .8 1 .8 0 .8-4 .4 Adjusted for age, BMI and alcohol b Adjusted for age, BMI, alcohol and locatio n 6th, 7th, 8th, 9th and 10th PFOA deciles were 1 .0 (0 .3- vs . 2 .21 g/ml) but were statistically significantly older (49 3 .5), 0 .5 (0 .1-2 .0), 0.9 (0 .3-3 .0), 1 .1 (0 .3-3 .6) and 1 .0 vs . 40) and had higher mean BMI (28 .8 vs . 27 .4), serum (0 .3-3 .6), respectively . glucose (103 vs . 91), triglycerides (226 vs . 159), and sev- Other analyses included those subjects (n = 46) who eral liver enzyme results ( ulkaline phosphatase 73 vs . 66, self-reported cholesterol lowering medication usage . These ALT 36 vs . 30, GGT 36 vs : 28) . Cholesterol (221 vs . 214), 46 subjects, compared to the 506 non-prescribed subjects ; LDL (134 vs . 136), and H.DL (47 vs . 49) were not statishad comparable mean PFOA concentrations (1 .98 g/ml tically significantly different between the two groups . No - ~ . . . - ~: . 6 .0 . i j_s . '".. } M~ LA ~f I' : r ~~' ~ y , i Y r' _ . ~ j . .44 ~ .r ' ~ a ~~~2ti '~~t f wI. ' " . ~ ; A :i '~~'` '`4 . ~ .0 . ~ . T' i ' 2 2 .0 = ' 6 .0 -4 .0 -2.0 0.0 2 .0 4.0 6 .0 In PFOA ' -6 .0 -4 .0 -2.0 0.0 In PFO A 2.0 4 .0 6.0 H < _ ~ * ~ N~ . . . , ~ ,g .o ~. . .. .. . o~..i+. ~~;N~i = . ' ~' . , _ ~ 7 '% 4i~1~ ~?.>1~ . . 00 ~~ ~_ ` '`': ~.. . ~ ~! ~}+i ~. 2 .0 ~ ~' 24 4 . . ' -6.0 -4.0 -2 .0 .0 0 0 .0 2 .0 4 .0 EM) 6 .0 -6 .0 -4 .0 -2 .0 0.0 In PFOA In PFOA 2 .0 4 .0 6 .0 Fig. 3 Scatterplots of the natural log PFOA (ng/ml) by the natural Fig . 4 Scatterplots of the natural log PFOA (ng/ml) by the natural log alkaline phosphatase and AST log ALT and GGT ~ Springer ~ P . 10 Int Arch Occup Environ Health Table 8 Non-adjusted and Non-adjusted Adjusted' b adjusted natural log (In) PFOA Ln PFOA Ln PFOA coefficients with In hepati c clinical chemistry Coefficient SE P value Coefficient SE measurements P value Ln alkaline phosphatase All locations 0.0155 0 .0082 0 .06 0 .0093a 0 .0081 0 .25 0.0037b 0 .0081 0 .65 Antwerp -0,0025 0 .0137 0.85 -0 .0060 0 .0139 0 .6 7 -0.0170 0 .0140 0 .22 Cottage Grove -0 .0141 0 .0113 0 .21 -0 .0127 0 .0117 0 .28 -0 .0140 0 .0117 0 .2 4 Decatur 0 .0394 0.0191 0 .04 0 .0460 0 .0192 0 .02 0 .0394 0 .0192 0 .04 Ln AST All locations -0 .0018 0 .0086 0 .83 -0 .0051 0.0086 0 .55 -0 .0089 0.0087 0 .31 Antwerp -0 .0048 0 .0137 0 .73 -0 .0029 0.0138 0 .83 -0 .0066 0.0142 0 .64 Cottage Grove -0 .0281 Decatur 0 .0205 0 .0141 0.0200 0 .05 -0 .0258 -0 .0271 0.31 0.0114 0.0062 0.0146 0.0145 0.0203 0.0203 0 .08 0 .07 0 .57 0 .76 Ln ALT All locations 0 .0402, 0.0143 0 .005 0.0249 0 .0132 0 .06 0 .0115 0 .0136 0 .40 Antwerp -0.0122 Cottage Grove -0.0131 0 .0220 0 .58 0 .0215 0 .54 -0 .0085 -0 .0293 .-0 .0096 -0 .0008 0 .0222 0 .7 0 0 .0222 0 .19 0 .0209 0 .65 0 .0208 0 .6 9 Decatur 0 .0954 0 .0300 0 .002 0 .0704 0 .0287 0 .02 0 .0581 0 .0287 0 .04 Ln GGT All locations 0 .0409 0 .0174 0 .02 Antwerp 0 .0170 0.0307 0 .58 0 .0326 0.0166 0 .05 0 .0097 0.0163 0 .55 0 .0269 0 .0294 0 .3 6 -0 .0047 0 .0295 0 .87 Cottage Grove -0 .0088 0 .0292 0 .76 -0 .0198 0 .0286 0 .49 -0.0233 0 .0270 0 .39 Decatur 0 .0754 . 0 .0344 0 .03 0 .0800 0 .0344 0 .02 0 .0599 0 .0329 0 .07 Ln total bilimbi n All locations -0 .0406 0 .0101 0 .0001 -0 .0325 0 .0099 0 .001 -0 .0267 0 .0101 0 .01 Antwerp --0 .0117 0 .0178 0 .51 -0 .0122 0 .0182 0 .5 0 --0 .0093 0 .0188 0 .62 Study methods . Adjusted for Cottage Grove -0 .0060 0 .0138 0 .66 -0 .0098 Ln age, Ln BMI, Ln alcohol -0 .0067 b Study methods . Adjusted for Decatur -0 .0528 0 .0203 0 .01 -0 .0537 Ln age, Ln triglycerides, Ln -0 .0462 alcohol 0 .0142 0 .0141 0 .0209 0 .0206 0 .49 0 .64 0 .01 0 .03 4~ Springer /~ p .11 Int Arch Occup Environ Healt h Table 9 Adjusted mean and 95% confidence intervals (95% CI) for hepatic clinical chemistry results, by PFOA decil e Total Direct PFOA Alk Phos AST ALT GGT Bilirubin Bilirubi n Decile Mean 95% CI Mean 95% Cl Mean 95% CI Mean 95% Cl Mean 95% Cl Mean 95% C I 1 66 61-71 26'f 24-28 29 25-33 32f 27-38 0 .9' 0.86-1 .02 0 .1 0 .09-0 .12 2 599''i 54-64 25 23-27 29 25-33 25 19-31 0 .9' 0.84-1 .00 0 .1 0 .09-0 .12 3 64 9 59-69 26'f 24-28 30 26-34 27 21-33 1 .01-i 0.89-1 .06 0.1 0 .09-0 .12 4 65 61-70 24 22-26 28''j 24-32 28 22-34 0.9' 0.82-0.99 0.1 0 .08-0 .11 5 649 60-69 22a'`,g-' 20-25 27'j 23-31 24' 18-30 0 .9 0.79-0.95 0.1' 0 .10-0 .13 6 66 61-71 22a` .8-' 20-24 25g''j 21-28 21J 15-27 0 .9` 0.78-0.95 0.1 0,08-0 .11 7 72h'`''h 67-77 26"f 24-28 31f 27-35 28 22-34 0.9` 0.77-0.94 0.1 0 .08-0 .11 8 65g 60-70 26`'t 24-28 29 26-33 29 23-35 0 .9c 0.77-0.93 0.1' 0 .10-0 .13 9 70' 66-75 26`'r 24-28 34d-r 30-38 33`'f 27-39 0 .8a~ 0.69-0.85 0.1`h 0 .07-0 .10 10 68b 63-72 24 22-26 34-f 30-38 30 24-36 0 .8` 0.75-0.92 0.1 0 .08-0 .12 Adjusted for age, BMI, and alcoho l Statistically significantly (P < 0.05) different than PFOA decile(s) 1, 2, 3, . . ., 10, respectively significant differences were observed with thyroid hor- facilities, only further demonstrated the lack of an associ- mones . Twenty percent of these 46 subjects were catego- ation between PFOA and cholesterol in this workforce . rized as having the metabolic syndrome compared to 7% of Unlike rats and mice, there' was no reduction in serum those not prescribed cholesterol-lowering medication . Not cholesterol in cynomolgus monkeys dosed with PFOA unexpectedly based on the results already presented, a (ammonium salt) for 6 months (Butenhoff et al . 2002) . greater percentage of those prescribed cholesterol-lowering A weakly negative association was observed with HDL medications were Cottage Grove and Decatur employees that was possibly due to uncontrolled (i .e ., residual) con- (79%) than those not prescribed medications (61%) . founding; based on lower HDL values observed among the Analyses with the clinical chemistries were similar whether Cottage Grove and Decatur workers than the Antwerp they excluded or included these 46 employees who self- workers and their markedly different demographic factors reported cholesterol lowering medications (data not (e .g ., BMI) . When the analyses were stratified by location, shown) . no statistically significant associations were observed be- tween HDL and PFOA . In another occupational study, Kaplan (2004) did not report an association between HDL Discussion and serum PFOA concentrations . Neither did Emmett et al . (2006a) in a community-based exposure study whose Based on an analysis of 506 maie participants of the 2000 median PFOA concentrations (approximately 0 .354 g/ml) fluorochemical medical surveillance program offered at the were 60 times higher than that of general population 3M Antwerp, Cottage Grove, and Decatur fluorochemical studies (Calafat et al . 2006 ; Olsen et al . 2003b) . To further production facilities who self-reported that they did not clarify any possible association with HDL, the A apolipo- take cholesterol lowering medications, there was no evi- proteins, which form the major proteins found in HDL, dence that these employees' serum PFOA concentrations could be measured although this was not part of the present were positively, or negatively, associated with serum total study . Apolipoprotein Al was not associated with serum cholesterol or LDL . These data analyses support the PFOA of comparable concentrations in a small analysis of argument that the PFOA and cholesterol association ob- Italian production workers (Costa 2004) . served in an analysis by Olsen et al . (2003a) of Antwerp Serum triglyceride levels were positively associated and Decatur workers in this database could have been a with PFOA . Although a biological association cannot be spurious finding or may have been the result of residual ruled out, it is also possible that this association might have confounding between the different demographic popula- been due, at least partially, to residual confounding as a tions as seen in Tables I and 2 . Addition of the Cottage consequence of the disproportionate number of Cottage Grove workforce, that was actively engaged in the manu- Grove and Decatur employees with higher PFOA concen- facture of the ammonium salt of perfluorooctanoic acid and trations than the Antwerp employees . The same positive had the highest serum PFOA concentrations of the three association was also observed between PFOS and seru m 12 IL Springer p . 12 Int Arch Occup Environ Healt h U N N t~ --~7 -^ 11C! N 1 ~ N A N d (V M ~ O O O O O O~ "-_- O O O '4. 0 PG O t~ [~ G~ [~ V O vl V O , 2.Q y . 2 ye .~~ U M M ~ N ^ (~ T l~ N N N N c!1 i el1 M c E t : ~w~ s ~ ~~ r < d kf) N N M M N ~n ~n g.O -4 .d .Z .Q U I O O O O O O O O O V ~ ~ 2 .0 4 .0 BA c4 0 1Z oo a oo a 1a v, ri O '- o 0 0 0 o { U M N O n M ! , l~ "t 'c e- N N M 4 ~ I O O O O O O O O O NI ~h v -2 . 0 In PFOA ........___......_ . ._ .____. ..____ ._ ._.5.6 .. . .... . ........_.. .. ...._ ..... . . . . ._ ._ . ._ ._. __._ . ._ _. ., L1'., O I~ O~ ~O ~O N M ':, O .-, O O -. p .-. O .-. .. .-. U O. 7 0o ct .D vl O v-i v~ ` ~ ^ N c1 4 A I M N~~ co rn o o~ o 0 0 0 00 ~ kj ~ = . ~ . ~~'~~~~ ~ ZI.. . . . . w' o C o0 0 oo n ry r ~ c^ 0 ~ c o<~ ci f . ..1~ as.'~.'~~i~~.tt..y ~~ . ~ tt.'~" .~ . . i. . s' T , v-, - O~ N O l NO t VUNM IT 0 I II N CL rn I O O O O o Ci O O O -6 .0 -4.0 -2 .0 0 .0 2 .0 4.0 0U A l In PFOA r C7 o~ r o rn v N r n o w C7 O - o 0 0 0 o Fig. 5 Scatterplots of the natural log PFOA (ng/ml) by the 6 .0 natural ~-,zl: r,, o, Q, r log TSH and T 3 U ~ 4 N N O V11 N O+ I O O O O O O O O O F triglycerides (Olsen et al . 2003c) . This prior association o o 0 0 0~~~ o with PFOS is inconsistent with the well-established hyp- olipidemia reported in PFOS-related toxicological studies v M N N q~~~~ with concentrations much higher than in the present study a o 0 0 0 0 0 0 0 0 (Seacat et al 2002, 2003) . On the other hand, PFOA was statistically significantly positively associated with trigly- xOo o-o4 R ro ` N0Nv0, N0Moc,~ cerides in the high dose (30/20 mg/kg) monkey group whose steady state serum PFOA concentration was c U M o v o o ~~~ U 158 10 g/ml (range 20 467 g/ml) (Butenhoff et al . 8 s 0 o I o 0 0 0 0 0 0 0 0 = '002) . This association was observed in measurements "Mr taken after 1 month of dosing at which time the group a -- O o U 0 -0 -- o0 d triglym ceridee leve a l wnas significantly higher than con- trol values as well as within group pretreatment values . At N Cq C4 N ~~~ o L~ the end of the study the mean triglyceride was elevated o I o 0 0 0 0 0 0 0 0 0~ o o compared to time related controls but not to the animals ' C pre-treatment values . However, only two primates were a ^C,~ 0~ 1^ N^~'' N o evaluated in the high-dose group at end of study . Inspection O -4 o 0 0 0- ~~^ of individTual values for PFOA serum concentration and T ^ u ~ ^ ~ ` ~ + .~ serum triglyceride values did not reveal any meaningful N N N i O M M I1 M P~ Pa . .`" 10 v o , rl~ " 11~ 1 C1 1 10 6 as 6 associations between these two parameters . zt C, I o 0 0 0 0 0 0 0 o en cp to to Al Of the four lipid measurements assessed in this study, aOoo~CO,~ooo~-M - cM q 0M'0,:'t0N'0v.ser~ umt rig~ lycerides have three times more intra-individual o ro a biological variation than either total cholesterol, LDL, or ~= HDL (Steinz anads Myers 1994) . Serum triglycerides are Q, q ~c,, m rn influenced by obesity, alcohol intake, and inattention to Springer p .13 Int Arch Occup Environ Health fasting requirements for blood collection, including caffe- inated drinks . It can also be hypothesized that the associ- ; , ~:!~ :.~' % ~ ; . t. ; Y~ ~r,~ '~,~~ t.; e . = i i :.,,, ~,~ ~ N, ' ations observed for serum triglycerides and PFOA might be the consequence of the non-adherence to fasting requirements by some shift production workers and/or the effect of postprandial metabolic responses in shift workers . Several studies have indicated postprandial serum triglyceride ' levels are higher among night shift workers than da y workers (Al-Naimi et al . 2004 ; Morgan et al . 1998 ; Karlsson et al . 2001 ; Lund et al . 2O01) . If a subset of subjects -6.0 -4 .0 -2 .0 0 .0 2.0 4.0 6 .0 (i .e ., production workers with higher PFOA serum con- In PFOA centrations) who worked night shift were less likely to .. ... . . .... .... . .. .. .. ................... . ............ ... .... .. . . i.a. ........... .......... ..- .. .. .- .. .... ..-....... ........-.- .. ................... ..... adhere to the fasting requirements and/or have postprandial metabolic profiles similar to other night shift workers, then F' a non-causal positive association between PFOA concen- . ~ :::,;: ; 't {~y . . trations and serum triglycerides could be observed when all subjects are included in the anaIysis . Unfortunately, at the ~' '- 46.0 '~ ~ time of data blood collection shift status information was .0 . =k W.*+ - 2 . s . ~ 4 .0 6. . not obtained; consequently'_ with this database it-is no t .~ possible to further address this methodological . question . Countering this possible hypotheSis is the-fact that blood glucose, also requiring a fasting-sample, was not associated with PFOA at any site . However, hyperglycemia has not In PFOA been consistently - associated with night shift workers Fig . 6 Scatterplots of the natural log PFOA (ng/ml) by the natural (Al-Naimi et al . 2004 ; Karlsson ,et al .- 2003) . Also, the log T4 and free T4 association between serum triglycerides and PFOA was not Table 11 Non-adjusted and :Non-adjusted Adjusted ' adjusted natural log (In) PFOA Ln PFOA Ln PFOA coefficients with in thyroid- related hormone measurements Coefficient SE P value Coefficient SE P value Ln TS H All locations 0 .0395 0.0204 0 .05 0 .0360 0 .0207 0 .08 Antwerp 0 .0509 0.0329 0 .12 0 .0391 0 .0333 0 .24 Cottage Grove -0 .0016 0.0310 0 .96 -0 .0111 0 .0322 0 .73 Decatur 0 .0343 0.0497 0 .49 0 .0365 0 .0513 0 .48 LnT4 All locations -0 .0037 0.0054 0 .50 Antwerp -0 .0022 0.0099 0 .83 Cottage Grove -0 .0124 0 .0072 0 .09 Decatur -0 .0012 0 .0126 0 .92 Ln free T4 -0 .0057 0.0054 0 .29 -0 .0041 0.0099 0 .68 -0 .0093 0 .0072 0 .20 -0 .0083 0 .0127 0 .51 All locations -0 .0138 0 .0044 0.002 -0 .0117 0 .0043 0 .01 Antwerp -0 .0108 0 .0078 0.17 -0 .0140 0 .0078 0 .07 Cottage Grove -0 .0093 0 .0058 0.11 -0 .0071 0 .0059 0.23 Decatur -0 .0138 0 .0103 0.18 -0 .0184 0 .0105 0.08 Ln T3 All locations 0 .0107 0 .0052 0.04 0 .0105 0 .0053 0 .05 Antwerp 0 .0222 0 .0077 0.005 0 .0216 0 .0077 0 .006 Cottage Grove 0.0026 0 .0096 0:79 0 .0006 0 .0099 0 .9 5 a See study methods . Adjusted Decatur 0.0317 0 .0117 0.008 0 .0271 0 .0119 0 .02 for Ln age, Ln BMI, Ln alcohol /4- 1L) Springer p . 14 Int Arch Occup Environ Healt h Table 12 Predicted thyroid-related measurements (reference range prevalence in the US population, as indicated by the third in parentheses) based on regression models for 40 year-old male with National Health and Nutrition Examination Survey BMI = 28, and consumes 0 .5 alcoholic drinks per day (NHANES) (Clark et al . 2003) . Measured hepatic clinical PFOA Predicted Predicted Predicted Predicted chemistry enzymes in the present study were not consis- (g/ml) TSH (IU/ml) T4 (g/dl) free T4 T3 (ng/dl) tently associated with employees' serum PFOA concen(0 .25-5 .5) (4 .5-12 .0) (ng/dl) (60-181) trations adjusted for age, alcohol, and either BMI or serum (0 .70-1 .53) triglycerides . A weakly positive asscciation between 0 .005 1 .58 8 .22 1 .15 119 alkaline phosphatase, ALT and GGT with PFOA was ob- 0 .01 1 .62 8 .19 1 .15 119 served among the Decatur male participants . Several epi- 0 .10 1 .76 8 .08 1 .11 121 demiologic research studies of the Decatur workforce have 0 .50 1 .87 8 .01 1 .09 124 not reported associations with liver disease (malignant or 1 .00 1 .91 7 .98 1 .09 125 non-malignant conditions) using a variety of data sources 5 .00 2 .03 7 .90 1 .06 128 including death certificates (Alexander et al . 2003), epi- 10.00 2 .07 50.00 2 .20 100 .00 2 .26 7 .87 7 .80 7 .77 1 .06 128 sodes of care (Olsen et al . 2004) and self-reports (Alex1 .04 131 ander and Grice 2006) . Neither a worker population study 1 .03 132 (Kaplan 2004) nor a community-based PFOA-expose d population (Emmett et al . 2006a, b) has reported associa- tions between PFOA and hepatic clinical chemistries . In observed at the Cottage Grove site which had both pro- the Emmett et al . (2006a) study, there was no relationship duction and non-production workers . Nevertheless, a po- between the blood levels of PFOA and the results for sitive association between serum triglycerides and PFOA cholesterol, hepatic-related tests (serum protein, albumin, was reported by Kaplan (2004) whose study population bilirubin, serum alkaline phosphatase, AST, ALT and also consisted of production (shift workers) and non-pro- GGT), or being treated for or informed by a physician that duction (non-shift workers) participants . To further clarify a community participant had liver disease (cirrhosis, hepa possible positive association ., between workers' .PFOA atitis, and any other liver condition) . concentrations and serum triglyceride levels, adjustment In the present study, there were no consistent associa- for shift work becomes a methodological necessity for tions between TSH, T4, free T4 or T3 with PFOA across subsequent occupational research analyses . the individual facility locations . Overall, T4 was negatively Although hepatic enzymes, in particular GGT, AST and associated with PFOA and T3 was positively associated but ALT are known to be elevated with heavy alcohol con- these trends were well within normal reference ranges . The sumption, these liver enzymes can also be elevated due to lack of thyroid related hormone associations is also con- obesity and dyslipidemia (Collantes et al . 2004 ; Mofrad sistent with results from Kaplan (2004) and Emmett et al . and Sanyal 2003 ; Ruhl and Everhart 2003) as non-alco- (2006a) . In a 6-month oral capsule PFOA dose study of holic fatty liver disease has substantially increased in cynomolgus monkeys, Butenhoff et al . (2002) reported no Table 13 Adjusted mean and 95% confidence intervals for thyroid measurements, by PFOA decile PFOA TSH T4 Free T4 T3 Decile Mean 95% Cl Mean 95% Cl Mean 95% Cl Mean 95% Cl 1 2 .07a 1 .13-3 .01 2 2 .00d 1 .04-2 .96 3 4 1 .89a 0.94-2 .84 3 .43"- ' 1 2 .48-4 .38 5 2 .61 1 .65-3 .55 6 2 .07`' 1 .11-3 .03 7 2 .21 1 .26-3 .16 8 2 .36 1 .46-3 .32 9 2 .84 1 .87-3 .81 10 2 .40 1 .46-3 .37 8 .29 8 .45 8.08 8 .04 7 .98 f 8 .53"i 8 .07 8 .40 8 .42 7 .94f 7 .92--8 .66 1 .15e'g1 1 .11-1 .19 124 118-131 8 .07-8 .83 1 .11 1 .07-1 .15 125 118-131 7 .70-8 .46 1 .14`j 1 .10-1 .18 124 118-131 7 .67-8 .41 1 .12 1 .08-1 .16 126 120-133 7 .60-8 .35 1 .07a'` 1 .03-1 .11 126 t20-133 8 .15-8 .91 1 .10 1 .05-1 .14 128 121-134 7 .70-8 .44 1 .09 1 .06-1 .14 129 123-135 8 .03-8 .76 1 .08 1 .06-1 .14 130 124-136 8 .03-8 .80 1 .11 1 .06-1 .15 129 123-136 7 .56-8 3 2 1 .07 " ` 1 .03-1 .11 128 123-136 Adjusted for age, BMI and alcoho l " Statistically significantly (P < 0 .05) different than PFOA decile(s) 1, 2, 3, . . ., 10. respectively Springer p .15 Int Arch Occup Environ Health clear changes in thyroid hormone homeostasis for TSH, T4 of free thyroxine in rat serum containing perfluorooctanesulfoor free T4 . Three high-dose monkeys '\ serum concentra- nate (PFOS) . Toxicology 234 :21-3 3 Clark JM, Brancati FL, Diehl AM (2003) The prevalence and etiology tions mentioned above) that were removed from dosing due of elevated aminotransferase levels in the United States . Am J to toxicity had T3 values that trended downward compared Gastroenterol 98 :960-967 to pretreatment measures . This is opposite the observation Collantes R, Ong JP, Younossi ZM (2004) Nonalcoholic fatty liver in the present study . Finally, a negative bias in thyroid- disease and the epidemic of obesity . Cleve Clin J Med 71 :657-664 Costa G(2004) Letter. US EPA docket AR226-1868 . US Environmental Protection Agency, Washington, D D (PFOS), has recently been reported .by Chang et al . ( 2007) . De Silva AO, Mabu ry SA ( 2006) Isomer distribution of perfluorocarboxylates in human blood : potential correlation to source . Environ Sci Technol 40 :2903-2909 Emmett EA, Zhang H, Shofer FS, Freeman D, Rodway NV, Desai C, Conclusion Shaw LM ( 2006a) Community exposure to perfl uorooctanoate: relationships between serum levels and certain health parame. JOEM 48 :771-77 9 PFOA concentrations measured in this workforce of 506 ters Emmett EA, Zhang H, Shofer FS, Freeman D, Rodway NV, Desai C, male fluorochemical production workers were not associ- Shaw LM (2006b) Community exposure to perfluorooctanoate : ated with total cholesterol, LDL, hepatic enzymes or thy- relationships between serum concentrations and exposure roid hormones . Residual confounding likely explained an sources : JOEM 48 :759-77 0 association between PFOA and HDL . Several hypotheses Expe rt Panel (2001) Executive summa ry of the third report of the National Cholesterol Education Program (CN.CEP) Expe rt Panel are offered for the inconsistent associations observed be on detection, evaluation, and treatment of high blood cholesterol tween PFOA and tri glyceri des . Although a biological in adults (Adult Treatment Panel ITI) . JAMA 285 :2486=2497 explanation cannot be ruled out, other hypotheses are also' F riedewald WT, Levy RI, Fred ri ckson DS (1972) Estimation of the plausible. These include non adherence to fasting recjuire concentration, of .low-density lipoprotein cholesterol in plasm a without use of the preparati ve ultracentrifuge. Clin Chem ments for morn ing blood collection by night shift produc- 18 :499-502 tion workers and/or increas ed triglyceri de levels that have' Gilliland FD, Mandel IS (i996) Serum perfl uorooctanoic acid and been found in night shift workers, in general : Because hepatic enzymes, lipoproteins and cholesterol : a study of PFOA has been shown to ' bind or be carried to blood occupationally exposed men . Am J Ind Med 129 :560-56 8 Han X, Snow TA, Kemper RA, Jepson GW (2003) Binding of albumin and beta-lipoproteins in the rat, monkey and hit- perfl uorooctan oic acid to rat and human plasma proteins . Chem man, a non-causal positive Correlation also remains another Res Toxicol 16 :775-78 1 possible explanation . Hansen KJ, Clemen LA, Ellefson ME, Johnson HO (2001) Compound-specific quantitative characte ri zation of organic ft,uprochemicals in biological matri ces . Environ Sci Technol 35 :766-77 0 References Hernan MA, Hemandez- Diaz S . Werler MM, Mitchell AA (2002) Causal knowledge as a prerequisite for confounding evaluation : Alexander BH, Gri ce M (2006) Self-reported medical conditions in _ an application to bi rth defects epidemiology . Am J Epidemiol perfluorooctanesulfonyl fluori de manufacturi ng workers . Uni- 155 :176-18 4 versity of Minnesota, Minneapolis, 1 Februa ry , 2006. US EPA Kaplan AM ( 2004) Ammonium perfluorooctanoate. US EPA docket docket AR-226-3677 . US Environmental Protection Agency, AR-226-1867 and AR 226-1868 . US Environmental Protection Washington, DC Agency, Washington, DC Alexander BH, Olsen GW, Bur ri s JM, Mandel JH, Mandel IS (2003) Karlsson B, Knutsson A, Lindahl B (2001) Is there an association Mo rt ality of employees of a perfluorooctanesulphonyl fluo ri de between shift work and having a metabolic syndrome? Results manufacturing facility . Occup Environ Med 60 :722-729 from a population based study of 27,485 people . Occup Environ Al-Naimi S, Hampton SM, Richard P, Tzung C, Morgan LM (2004) Med 58 :747-75 2 Postprandial metabolic profiles following meals and snacks eaten Karlsson BH, Knutsson AK, Lindahl BO, Alfredsson LS (2003) du ri ng simulated night and day shift work . Chronobiol Int - Metabolic disturbances in male workers with rotating three-shift 21 :937-947 work . Results of the WOLF study . Int Arch Occup Environ Brousseau ME, Schaefer EJ, Wolfe ML, Bloedon LT, Digenio AG, Health 76 :424-43 0 Clark RW, Mancuso JP, Rader DJ ( 2004) Effects of an inhibitor Kennedy GL, Butenhoff JL, Olsen GW, O'Connor JC, Seacat AM, of cholestery l ester transfer protein on HDL cholesterol . N Engl J Perkins RG, Biegel LB, Murphy SR, Farrar DG (2004) The Med 350 :1505-1515 toxicology of perfluorooctanoate . C ri t Rev Toxicol 34 :351-38 4 Butenho ff J, Costa G, Elcombe C, Farrar D, Hansen K, Iwai H, Jung Kerstner-Wood C, Coward L, Gorman G (2003) Protein binding of R, Kennedy G, Lieder P, Olsen G, Thomford P (2002) Toxicity perfluorobutane sulfonate, perfluorohexane sulfonate, perfluo- of ammonium perfluorooctanoate in male cynomolgus monkeys rooctane sulfonate and perfluorooctanoate to plasma (human, rat, after oral dosing for 6 months . Toxicol Sci 69 :244-257 and monkey), and various human-deri ved plasma protei n Calafat AM, Kuklenyik Z, Caudill SP, Reidy JA, Needham LL (2006) fractions . Southern Research Institute . Study 9921 .7 . US EPA Perfluorochemicals in pooled serum samples from United States docket AR-226-1354 . US Environmental Protection Agency, residents in 2001 and 2002 . Environ Sci Technol 40 :2128-2134 Washington, DC Chang SC, Thibodeaux JR, Eastvold ML, Ehresman DJ, Bjork JA, Lund J, Arendt J, Hampton SM, English J, Morgan LM (2001) Froehlich JW, Lau CS, Singh RJ, Wallace KB, Butenhoff JL Postprandial hormone and metabolic responses amongst shift (2007) Negative bias from analog methods used in the analysis workers in Antarctica . J Endoc ri nol 171 :557-564 ~ Springer ~~ p . 16 Int Arch Occup Environ Healt h Maloney EK, Waxman DJ (1999) Transactivation of PPARa and Olsen GW, Church TR, Miller JP, Burris JM, Hansen KJ, Lundberg PPARy by structurally diverse environmental chemicals . Toxicol JK, Armitage JB, Herron RM, Medhdizadehkashi Z, Nobiletti Appl Pharmacol 161 :209-218 JB, O'Neill EM, Mandel JH, Zobel LR (2003b) Perfluorooc- Mofrad PS, Sanyal AJ (2003) Nonalcoholic fatty liver disease . tanesulfonate and other fluorochemicals in the serum of AmerMedGenMed 5 :1-13 ican Red Cross adult blood donors . Environ Health Perspect Morgan L, Arendt J, Owens D, Folkard S, Hampton S, Deacon S, 111 :1892-190 1 English J, Ribeiro D, Taylor K (1998) Effects of the endogenous Olsen GW, Butenhoff JL, Mandel JH (2003c) Assessment of lipid, clock and sleep time on melatonin, insulin, glucose, and lipid hepatic and thyroid function in relation to an occupational metabolism . J Endocrinol 157 :443-451 biologic limit value for perfluorooctanoate . 3M Company, S t Nakanishi N, Matsuo Y, Yoneda H, Nakamur K, Suzuki K, Tatara K Paul . US EPA docket AR-226-1351 . US Environmental Protec(2000) Validity of the conventional indirect methods including tion Agency, Washington, DC Friedewald method for determining serum low-density lipopro- Olsen GW, Burlew MM, Marshall JC, Burris JM, Mandel JH (2004) tein cholesterol level : comparison with the direct homogeneous Analysis of episodes of care in a perfluorooctanesulfonyl fluoride enzymatic analysis. J Occup Health 42 :130-137 production facility . JOEM 46 :837-84 6 Olsen GW, Zobel LR (2006) An analysis of the 2000 fluorochemical Ruhl CE, Everhart JE (2003) Determinants of the association of (perfluorooctanoate, PFOA) medical surveillance program at 3M overweight with elevated serum alanine aminotransferase activCompany's Antwerp (Belgium), Cottage Grove (Minnesota), ity in the United States . Gastroenterology 124 :71-7 9 and Decatur (Alabama) facilities. 3M Company, St Paul, 6 May, Seacat AM, Thomford PJ, Hansen KJ, Olsen GW, Case MT, 2006 . US EPA docket AR-226-3678 . US Environmental Pro- Butenhoff JL (2002) Subchronic toxicity studies on perfluorooctection Agency, Washington, DC tanesulfonate potassium salt in cynomolgus monkeys . Toxicol Olsen GW, Gilliland FD, Burlew MM, Burris JM, Mandel IS, Mandel Sci 68 :249-264 JH (1998) An epidemiologic investigation of reproductive Seacat AM, Thomford PJ, Hansen KJ, Clemens LA, Eldridge SR, hormones in men with occupational exposure to perfluoroocta- Elcombe CR, Butenhoff JL (2003) Sub-chronic dietary toxicity noic acid . JOEM 40 :614-622 of potassium perfluorooctanesulfonate in rats . Toxicology Olsen GW, Burris JM, Mandel JH, Zobel LR (1999) Serum 183 :117-13 1 perfluorooctane sulfonate and hepatic and lipid clinical chem- Stein EA, Myers GL (1994) Lipids, lipoproteins, and apolipoproteins . istry tests in fluorochemical production employees . JOEM In : Burtis CA, Ashwood ER (eds) Tietz textbook of clinical 42 :799-806 chemistry . WB Saunders Co ., Philadelphia, pp 1002-1093 Olsen GW, Bums JM, Burlew MM, Mandel JH (2000) Plasma Ubel FA, Sorenson SD, Roach DE (1980) Health status of plant cholecystokinin and hepatic enzymes, cholesterol and lipopro- workers exposed to fluorochemicals-a preliminary report . Am teins in ammonium perfluorooctanoate production workers . Drug Ind Hyg Assoc J 41 :584-589 Chem Toxicol 23 :603-620 Xie Y, Yang Q, Nelson BD, DePierre JW (2003) The relationship Olsen GW, Burris JM, Burlew MM, Mandel JH (2003a) Epidemi- between liver peroxisome proliferation and adipose tissue ologic assessment of worker serum perfluorooctanesulfonate atrophy induced by peroxisome proliferators exposure and (PFOS) and perfluorooctanoate (PFOA) concentrations and withdrawal in mice . Biochem Pharmacol 66:749-75 6 medical surveillance examinations . JOEM 45 :260-270 9 Springer ~