Document o9e1LqerxBKO9R5gm8N7LBEmE

Taft/ Taft Stettinius & Hollister LLP 425 Walnut Street, Suite 1800/ Cincinnati, OH 45202-3957/Tel: 513.381.2838/Fax: 513.381.0205/www.taftlaw.com Cincinnati / Cleveland / Columbus / Dayton / Indianapolis / Northern Kentucky / Phoenix / Beijing Robert a . Bilott 513-357-9638 bilott@tafttaw.com September 7, 2010 EP A Docket Center, MC 2822T U .S. Environmental Protection Agency EP A West, Room 3334 1200 Pennsylvania Avenue, NW W ashington, D .C. 20460-0001 Re: Subm ission to IR IS and AR-226 Database For PFO A /PFO S: EPA -H Q O R D -2 0 0 3 -0 0 1 6 To IR IS Database for PFO A /PFO S: In response to the Notice issued by U SEP A on February 23, 2006, regarding U S E P A 's efforts to consider perfluorooctanoic acid ("P FO A ") and perfiuorooctane sulfonate ("P F O S ") within the Integrated Risk Information System ("IR IS "), 71 Fed. Reg. P-2 ARTICLE Perfluorooctanoic Acid, Perfluorooctanesulfonate, and Serum Lipids in Children and Adolescents Results From the C8 Health Project Stephanie J. Frisbee, MSc, MA; Anoop Shankar, MD, PhD; Sarah S. Knox, PhD; Kyle Steenland, PhD; David A. 5aviz, PhD; Tony Fletcher, PhD; Alan M. Ducam an, MD, MS B ackground: Perfluorooctanoic acid (PFOA) and per fluorooctanesulfonate (PFOS) are man-made com pounds with widespread presence in human sera. In pre vious occupational and adult studies, PFOA and PFOS were positively associated with serum lipid levels. O bjective: To interrogate associations between PFOA and PFOS and serum lipids in children and adolescents. D esign: Cross-sectional community-based study. S ettin g : M id-Ohio River Valley. P a rtic ip a n ts: A total of 12 476 children and adolescents included in the C8 Health Project, which resulted from the pretrial settlementofa class action lawsuit pursuant to PFOA contamination of the drinking water supply. M ain O utcom e M easures: Serum lipids (total, highdensity lipoprotein [HDL-Cj, and low-density lipopro tein ILDL-C1 cholesterol and fasting triglycerides). R esults: Mean (SD) serum PFOA and PFOS concentra tions were69.2(LL1.9) ng/mL and 22.7(12.6) ng/mL. re spectively. In linear regression after adjustment for co variables, PFOA was significantly associated with increased total cholesterol and LDL-C, and PFOS was significantly associated with increased total cholesterol, HDL-C, and LDL-C. Using general linear model analysis of covari ance, between the first and fifth quintiles of PFOA there was a 4.6-mg/dL and a 3.8-mg/d Lincrease in the adjusted mean levels of total cholesterol and LDL-C levels, respec tively, and an 8.5-tng/dL and a 5.8-mg/dL increase in the adjusted mean levels of total cholesterol and LDL-C, re spectively, between the first and fifth quintiles of PFOS. Increases were 10 mg/dL forsome age- and sex-group strata. Observed effects were nonlinear, with larger increases in total cholesterol and LDL-C levels occurring at the lowest range, particularly of PFOA. C oadusiun: Although the epidemiologic and crosssectional natures of this study limit causal inferences, the consistently observed associations between increasing PFOA and PFOS and elevated total cholesterol and LDL-C levels warrant further study. Arch PediairAdole.se Med. 2010;164(9):860-869 Author Affiliations: Department o i CommunityMedicine (Ms Frishee and Drs Shankar, Knox, and Ducatman) and Center for Cardiovascular and Respiratory Sciences (Ms Frisbee and Dr Shankar), West Virginia University School of Medicine, Morgantown: Department of Environmental and Occupational Health, Emory University Rollins School of Public Health, Atlanta. Georgia (Dr Steenland); Department of Community and Preventive Medicine, Mount Sinai School of Medicine, New York, New York (D r Saviiz); and Public Health and Environmental Research Unit. London School of Hygiene and Tropical Medicine, London, England (Dr Fletcher). ERFLUOROOCTANOIC ACID P (PFOA) and perfluorooctane sulfonate (PFOS) areperfluoroalkyl acids: man-made com pounds used as emulsifiers known to contain perfluoroalkyi acids gen erally and PFOA and PFOS specifically. Re cent results from the National Health and Nutrition Examination Survey*-5 reported detection of perfluoroalkyi acids in almost during the m anufacture of fluoalrlospaomlyp-les, with a US population median mers, which are chemicals that give non for PFOA of 5.2 ng/mL and 3.9 ng/mL stick heat resistance to cookware or breath (1999-2000 and 2003-2004), respectively. able yet waterproofproperties to fabrics and Identified sources of human exposure to upholstery. Perfluoroalkyi acids may also re PFOA and PFOS include drinking water, sult from the metabolism or breakdown of dust, food packaging, breast milk, cord fluorinated telom ers, w hich are com blood, microwave popcorn, ambient air, and pounds used as coatings for commercial food occupational exposure,10'3although the rela packaging, factor)'treatments for fabrics and tive contribution of each is unknown. carpets, and manufacturer pretreatment for Animal studies have identified the liver "stain-resistant" clothing. as a primary target organ for perfiuoroal- Recent reviews of the scientific litera kyl acid physiologic activity. Reported toxi ture10 regarding PFOA and PFOS empha cological effects of exposure to PFOA and/or size their environmental persistence and PFOS include hepatomegaly, reduction in presence in a variety of marine and fresh serum triglycerides and cholesterol in some water species. Human serum samples across animal species, and alterations in coen myriad age groups and geographic areas are zyme A activity. These alterations in he- ( REPRINTED) ARCH PEDLAIR ADOLESC. MED/VOL 16-1 (NO 9). SEP Z010 WWW ARCHPEDIATRIC.S COM 860 Downloaded from www.arcbpediuirics.com , on September 7, 2010 2010 American Medical Association. All rights reserved. p.3 patic metabolism and function have been attributed to per fluoroalkyl acid action as a peroxisome proliferatoractivated receptot-a (PPAR-a) agonist with subsequent peroxisome proliferation.1'1 Compared with effects seen in animals, human stud ies have reported different associations between PFOA and lipid levels. In occupationally exposed employees, PFOA associations have included increased PFOA concentrations and increased total cholesterol (total-C) but not triglyc erides or other lipids9; increased total-C and low-density lipoprotein cholesterol (LDL-C) but not high-density li poprotein cholesterol (HDL-C)'0; increased total-C but not triglycerides or HDL-C" ; and increased triglycerides but no association w ith total-C or LDL-C.12Astudy of a com munity cohortwith known environmental PFOA contami nation reported no association between PFOA and totaiC.15A recent, much larger study of adults from the same and adjacent environmentally exposed communities re ported a robust, positive association between PFOA and total-C and LDL-C and a less clear association between PFOA and triglycerides.1'1 To o u r knowledge, no study has investigated potential associations between perfluoroalkyl acids and lipids in chil dren and adolescents. The importance of such studies is 3-fold. First, ifsuch associations are etiologic, exposure pre vention would become important to reduce the long-term health consequences of elevations in known cardiovascu lar disease risk factors. Second, studying potential health consequences ofan environmental exposure in children and adolescents may provide greater insight because these groups likely have fewer factors confounding underlying associations (eg, prevalent chronic or acute disease, or medi cation use) compared with adul ts. Third, given possible dif ferences in physiologic processes owing to developmental changes in children and adolescents, toxic effects may be different compared with those observed in adults. Thus, the purpose of this study is to interrogate associations be tween serum PFOA and PFOS and lipids in a large, com munity-based sample of children and adolescents. METHODS STUDY METHODS AND PARTICIPANTS The participants in this study were 12476 children and adoles cents aged 1.0 to 17.9 years at their enrollment in the C8 Health Project (hereafter referred to as the Project). The Project lias been more completely described elsewhere.15Briefly, it resulted from a pretrial settlement of the class action lawsuit Leach v E.l. du Port dc Nemours & Co,16 filed in 2002 after PFOA from the DuPont W ashington W orks facility near Parkeisburg. West Virginia, was found to have infiltrated several local drinking water supplies along the inid-O hio River Valley. Project eligibility criteria included the ability to document consumption of contaminated drinking waier(from 1 of 2 public water districts in West Virginia o r4 in Ohio, or from private water sources within the. public water districts that contained &0.05 PFOA pans per billion) for at least l year between 1950 and December 3,2004, at a primary residence, place of employment, or school. Participants were enrolled during a 13-month period between August 1,2005, and August 31,2006. Participation and enrollment included documentation of iden tity and eligibility, a seif-reported survey of demographics, per sonal health history, and lifestyle habits; self-reported height and weight; and voluntary submission of a blood sample. As we have previously reported, approximately 60% oi participants were rest dents in an eligible water district at the time of their enrollment in the Project, and an estimated 80% of people who resided in the eligible water districts at the time of the Project enrolled as participants.15W ith 69 030 total participants (12 476 children and adolescents), to our knowledge the Project represents the larg est community-based study to date investigating potential asso ciations between PFOA exposure and human health effects. BLOOD SAMPLE PROCESSING AND LABORATORY METHODS Children and adolescents voluntarily suhm itted blood samples at a maximum of 26 mL. Samples were centrifuged, divided into aliquots, and refrigerated at community-based data collection sites until being shipped to laboratories for analysis. Clinical laboratory tests were performed at an accredited clini cal diagnostic laboratory (LabCorp Inc, Burlington, North Caro lina). The lipid panel included total-C, HDL-C, LDL-C, and tri glycerides. The LDL-C was calculated using the Friedewald formula for participants with triglycerides lower than 400 mg/dL (to convert to millimoles per liter, m ultiply by 0.0113). Al though fasting was not a requirem ent for phlebotomy, the time of the last meal was reported. The primary laboratory performing perfluorocarbon analysts (Exvgen Research Inc, State College. Pennsylvania) was also used for an independent study of a smaller group or residents in l wa ter district included in the Project.17Full perfluoroalkyl acid ana lytic techniques and quality assurance protocols for the Project have been published elsewhere.15Briefly, theanalyticprotocol wasa modi fication of a previously described protocol that used a protein pre cipitation extraction together with reverse-phase high-perfonnance liquid chromatography/tandem mass spectrometry.lsSpectromctric detection was performed usinga triple quadrupole mass spec trometer in selected reaction monitoring mode, monitoring for the individual mJz (mass-to-charge) transitions for the perfluoroalkyl acid and " C-PFOA surrogate. CONSENTING PROCEDURES Project data collection administered by Brookmar Inc, Park ersburg, West Virginia, was conducted with the authority and supervision of the W ood County, W est Virginia, Circuit Court. Brookmar Inc used a consenting procedure approved by par ties to die Settlement that included language specific to the Project's purpose and data collection procedures and the m an datory documentation requirements to demonstrate Class eli gibility. All participants subm itting a voluntary blood sample completed the standard consent and release forms of the clini cal laboratory contracted for phlebotomy. The Project group at West Virginia University and the. C8 Science Panel obtained institutional review board approval from their own academic institutions permitting access to anonymous Project data. STATISTICAL ANALYSIS The outcom e variables were total-C, LDL-C, HDL-C. and tri glycerides. For simplicity, the same covariables were consid ered for all analytic models: age, sex (except for models strati fied by sex), body mass index (BM1) z score, duration of fast (in minutes), and whether the participant engaged in a regular exercise program. Age was included as a continuous variable a n d a s 2 stra ta (aged 1.0-11.9 years and 12.0-17.9 years) to as sess possible ageand developmental confounding. Forall analy ses using quantiles, grouping was established within age group and sex, and so quantiles are age-group and sex specific. En gagement in a regular exercise program was self-reported as part (REPRINTED) ARCH PEDIATR ADOLESC. MED/ VOL 1(H (NO. 9). SEP 2010 WWW ARCHrEDlATRICS.COM 861 Downloaded from www.archpvdiatrics.com , on September 7,2010 @2010 American Medical Association. All rights reserved. P-4 of the survey ("Do you engage in an exercise program?" with a "Yes/No" response option), as were fasting status, height, and weigh!. Analyses includingiriglycerides were conducted for only those reporting a fast of 6 hours or more. The BM1 z score was calculated using Epi Info software, 2000 reference data set (Cen ters for Disease Control and Prevention, Atlanta. Georgia). Multiple linear regression was performed to assess Tor the presence o fan overall linear association between PFOA or PFOS and lipids. Both PFOA o r PFOS and the dependent variable were, natural log transformed. General linear model analysis of covariance was performed to estimate predicted lipids (estimated marginal mean [EMMD after adjustmeni for covariables based on increasing PFOAorPFOS quintiles. The serum lipid was defined as the dependent vari able, PFOA or PFOS quintile as a fixed factor, and the other in dependent variables (age, sex, BMI t score, exercise, and lasting status) a s covariates. The covariable-adjusted EMM for each quin tile is presented graphically for die overall population (all age groups and both sexes combined). Thedifference in the covariableadjusted EMM between the fifth and first quintile was estab lished for each age and sex strata. To interpret the statistical significance of the trend in the quimile-based change in ihe covariable-adjusted EMM, linear regression analysis was used to estimate the |J coefficient (and corresponding P value) for the PFOA o r PFOS quintile, which is reported as a |i for trend. To assess the linearity or nonlinearity of any association between PFOA or PFOS and serum lipid levels. 20-group quan tiles were determined and the population median for each group was calculated. General linear model analysis of covariance was again used to determine the covariable-adjusted EMM for each quantile, which was plotted against its PFOA or PFOS median. Binary logistic regression analysis was performed to assess the odds o f abnormal lipids with increasing PFOA or PFOS quintile. For total-C, LDL-C, and triglycerides, categorization as abnormal was based on American Heart Association-endorsed cuioff val ues for "borderline" or "high" in children (total-C a 170 mg/dL, LDL-C21 10 mg/dL Ito convert to millimoles per liter, multiply by 0.0259], and triglycerides 2:150 mg/dL.1' For HDL-C, values lower than 40 mg/dL (to convert to millimoles per liter, multiply by0.0259) were classified as abnormal. Logistic regression analy sis was performed using PFOA or PFOS quintile dummy variables, in which the first quintile was considered the reference group. Interaction between PFOA and PFOS wasdually assessed. Lo gistic regression analysis, with models otherwise constructed as described previously, was performed with the PFOA quintile, the PFOS quintile, and the product o i quin tiles (interaction term) in cluded in the analytic model. The statistical significance of the in teraction tenu (P for interaction) is reported. Logistic regression analysis was also used to assess for the presence of multiplicative interaction using models as described previously. Four groups were created based on the PFOA and PFOS quintiles: group 1 (PFOA first to fourth quintile and PFOS first to fourth quintile), group 2 (PFOA fifth quintile and PFOS first to fourth quintile), group 3 (PFOA first to fourth quintile and PFOS fifth quintile), and group 4 (PFOA fifth quintile and PFOS fifth quintile). Logistic regression was then completed using 3 dum my variables, in which group 1 was considered the reference group. Sensitivity analyses were conducted for the effects of fast ing and socioeconomic status. Using models otherwise con structed as described previously, multiple linear regression was performed for models with and without household income (di chotomized at s$ 3 0 0 0 0 /y or > $30 000/y) and models ad justed for fasting status or with only participants having com pleted a fast of 6 hours or longer. The magnitude and statistical significance of p coefficients for the different models were then com pared. Both PFOA or PFOS and the dependent variable were natural log transformed. All analyses were performed using SPSS statistical software (SPSS Inc, Chicago, Illinois). RfSVlTS General characteristics of participants are reported in T able 1 . The mean (SD) age was 1L. I (4.5) years, and par ticipation was similar across both sexes (48.9% girls and 51.0% boys; data missing for 6 participants 10.1%]). Of the 12 476 children and adolescents included in this study, more than 10 000 (>80.0%) hadperfluoroalkyl acid quantifica tion and serum lipids available for analysis. Consistent with US Census Bureau estimates for the area, 11894 (96.1%) reported their ethnicity as white, 4406 (39.7%) were clas sified as overweight or obese (2 8 5 th BMI percentile), and 4576 (36.7%) reported having a regular exercise pro gram. According to participant-reported residence at the time of enrollment, a slightly higher proportion of partici pants were from Ohio compared with West Virginia (6796 [54.5%1 vs 5520 [44.2%], respectively). Mean (SD) total-C was 160.7 (29.3) mg/dL, w ith65.8% of values classified as acceptable (<170 mg/dL). Mean (SD) calculated LDL-C was 87.3 (25.2) mg/dL; w ith83 7% of values classified as acceptable (< 1 1 0 mg/dL). Mean (SD) HDL-C was 49.3 (11.3) mg/dL, with 92.2% of val ues classified as ideal (2 35 mg/dL) and 19.7% lower than 40 mg/dL. Finally, mean (SD) fasting triglycerides were 99. L(56.0) tng/dL, with 85.6% of values classified as ac ceptable (S150 mg/dL). Mean (SD) serum PFOA and PFOS concentrations were 69.2(111.9) ng/mLand 22.7 (12.6) ng/mL, respectively, with serum concentrations statisticallysignificantly higher in boys and younger children for PFOAand PFOS,particularly the former. Consistent with the environmental (drinking wa ter) contamination andas previously reported,15serum PFOA concentration for 12- to 19-year-olds in the Project popu lation was substantially higher than the reported concen tration for 12- to 19-year-okls in the 2003-2004 National Health and Nutrition Examination Survey,5whereas PFOS concentrations were similar (29.3 ng/mL vs 3.9 ng/mLand 19.1 ng/mL vs 19.3 ng/mL, respectively). Resultsfromregressionanalysis(notshowrt) demonstrated thatafter adjustment for covariables, total-Cand LDL-Cwere linearly and positively associated with PFOA and PFOS (P < .001 forallmodels). Triglycerides were also linearly and positivelyassociated with PFOA (P= .02) but not with PFOS. The HDL-C was not linearly associated with PFOA but was positively associated with PFOS. For linear regression mod els and other analyses reported in this article, virtually all covariables were statisticallysignificantly associated with the dependent variable (not shown). Associations between increasing PFOA and PFOS quin tiles and the covariable-adjusted EMM of serum lipid lev els are depicted in Figure 1 . Total-C and LDL-C demon strated a consistent increase for each increase in PFOA or PFOS quintile; a 4.6-mg/dL and 3.8-mg/dL increase in the covariable-adjusted EMM of total-Cand LDL-C between the first and fifth quintiles of PFOA and an 8.5-mg/dL and 5.8mgAlL increase in die covariable-adjusted EMM of total-C and LDL-C, respectively, between the first and fifth quintilesof PFOS. Overallassociations between PFOA and PFOS and HDL-C and triglycerideswere less clear, with no appar ent association between PFOA quintile and covariableadjusted EMM for HDL-C or between PFOS quintile and (REPRINTED! ARCH PEDIATR ADOLESC MED/VOI. 164 (NO. 9). SEP 2010 WWW, ARCHPEDIATRICS.COM 862 Downloaded from www.archpcdiairics com , on September 7, 2010 0 2 0 1 0 American Medical Association. AH rights reserved. P-5 Tab le 1. Participant Characteristics* Characteristic Age, median/mean (SD), y Agegroup, median/mean (SO), y 1.0-11.9 12.0-17.9 Regular exercise Yes No : . BMI percentile, % <5.0 (underweight) 5.0-84.9 (normal) 85.0-95.0 (overweight) >95.0(obese) Missing data Time fasting, h <6 *6 Not reported Householdincome, $/y 30000 >30000 Missing data Ethnicity . -O White Black Ail olherethnicities Missing data State residenceattime ofenrollment Ohio . WestVirginia Other Totat-C, mg/dL <170 *170 Missing data LDL-C, mg/dL <110 *110 Missing data HDL-C. mg/dL <40 *40 .... ,,, > Missing dataandnonfasting participants .1-' Fasting triglycerides) mg/dL <150- *150 Missing data . PFOA. ng/mL. mean (SD) 1.0-11:9 y 12.0-17.9 y PFOS, ng/mL, mean(SD) 1.0-11:9 y 12.0-17.9 y Boys11 11.6/11.2(4:4) 3287(51.7) 3072 (48.3) 2491,(39.1) 3872.(80,9) 272(4.8) 2947(51.7) 934(16.4) 1550(27.2) 660 4591(722) 1394(21.9) 378 (5.9} 2471 (49.3) 2540 (50.7) 1352 .6061 (96.3) - 108(17) 130(2.1) 51 3476(54.6) 2817(44.3) 70(1.1) 3729 (68.6) 1706 (31.4) 928 4581 (85.3} 788(14.7) 994 . 1313(24.2) \ 4122(75.8) 702 1244 (84.6) 227(15.4) 4892 35.1/82,1 (129.2) 30.1/69.3 (107.1) 21.7/24.6 (13.4) 20.3/23.2 (12.9) :'. 'V G lrlS^ ' i ' . 11.5/11.1 (4.5)- 3249(53.2) 2862(46.8). 2085(34.1) .-. 4028'"i65.9) . 29i(5.4) .' 3168 (58.9) . 911 (16.9) 1011 (18.8) 732 4400 (72.0) 1346(22.0) 367(6:0) 2303 (47.6) 2540 (52.4) 4270 5833 (96.0) 96(1.6) 145(2.4) 34 3320 (54.3) 2703(44.2) 90(1.5) 3207 (62.8) 1901 (37.2) 1005 4158(82.0) 915(18.0) 1040 . 260(14.9) 4348 (85.1) 1231' ~ . . ' 1201 (86.7) 185(13:3) 4727 30.7/23.1 (120.1) 22.9/53.7 (88.1) 19.9^2.6(12.6) 18.2/20.5(113)' ; - ; tptaP1 11 6/11:1 (4.5) : 6536(52.4) . 5934 (47.6) - 4576 (36:7). 7900 (63.3) 563(51) 6115(55.2) 1845 (16.6) 2561 (23.1) 1392 8991 (72.1) 2740(22.0) 745(6:0) . 4774 (48.4) 5080 (51.6) 2622 '- . 11894(96.1) 202 (1.6) 275 (2.2) 85 6796 (54.5) 5520 (44.2) 160(1.3) 6936 (65.8) 3607 (34.2) 1933 . 8739 (83:7) 1703(163) . 2034 ; 2073(19.7) 8470(80.3) 1933 2445 (85.6) 412 (14,4) 9619 32.6/77.7(124.9) 26.3/61.8 (983) 20.7/23.6 (13.1) 19.3/213 (12.2) Abbreviations: SMI. bodymass index (calculated as weightinkilograms divided by height in meters squared): HDL-C. high-density lipoprotein cholesterol: LDL-C, low-density lipoprotein cholesterol: PFOA, perfluorooctanoicacid; PFOS, perfluorooctanesulfonate; total-C, total cholesterol. SI conversion factors: ToconvertHDL-C. LDL-C, andtolal-Ctomillimoles per(ter. multiplyby 0.0259:fasting triglycerides tomillimoles perliter, multiply by0.0113. aOataare presented as number (percentage) unless otherwise indicated. Denominatorsvary because of missing data. Percentages maynottotal 100 because of rounding, are based oncategorytotals, and reflectthe "valid percentage" (ie. missing data arenot included in the denominator). "n=8359. cn=6111. aN=12 470. Data reguarding sex weremissing for 6 participants. covariable-adjusted EMM for triglycerides. There was a small Increase in the covariable-adjusted EMM of HDL-C and the first to th ird quintiles of PFOS, but not for the third to fifth quintiles of PFOS. These, associations are m ore fully characterized in Table 2, w hich reports differences betw een the fifth and first quintiles and 3 for trend for each of these age/sex groups. For PFOA and total-C and LDL-C, there w as a (REPRINTED) ARCH PED1ATR ADOLESC MED/VOL lfi-t (NO. 9). SEP 2010 WWW.ARCHPEDIATRICS.COM 863 Downloaded from www.archpcdiatrics.com, on September 7 2010 2010 American Medical Association. All rights reserved. P-6 Figure 1. Changes in covariabie-adjusledestimatedmarginal means (general linearmodelanalysis) across perfluorooctanoicacid (PFOA) and pertluorooctanesudonate (PFOS) quintiles. A,Totalcholesterol (totaTC). B, High-densitylipoproteincholesterol (HDL-C). C. Low-densitylipoproteincholesterol (LDL-C). D.Fastingtriglycerides. Lipidvalues are presentedas mean (SE). Toconvert total-C. HOL-C. and LOL-C tomillimolesperliter, multiply by0.0259; tasting triglyceridestomillimoles per liter, multiply by0.0113. trend toward a larger increase in the covariable-adjusted EMM in the younger compared with the older age group (5.8 mg/dL vs 4.2 mg'dL, respectively, for total-C and 4.9 mg/dL vs 3.2 mg/dL, respectively, for LDL-C) and in boys compared w ith girls (within each age group). The {3 for trend for each of these age groups was statistically signifi cant (P < .05). For PFOS, there was a trend toward a larger increase in the covariable-adjusted EMM in older com pared with younger age groups (9.5 mg/dL vs 5.5 mg/dL, respectively, for total-C and 7.5 mg/dL vs 3.4 mg/dL, respectively, for LDL-C) and a trend toward larger increases for boys compared with girls (within each age group). The (3 for trend for each of these age groups was also statistically significant (P<.05). In contrast, each p for trend forage group and sex strata for the associations between PFOA and HDL-C and tri glycerides was not statistically significant (except in 1 group), making differences in the covariable-adjusted EMM difficult to interpret. For associations between PFOS and HDL-C, each f3for trend was statistically significant for boys and both sexes combined (but not girls) for both age groups, although marginal increases in the covariableadjusted EMM were small (1.1-2.6 mg/dL). Similarly, each 3 for trend for age group and sex strata for the associa tions between PFOS and triglycerides was not statisti cally significant (except in 1 group), making differences in the covariable-adjusted EMM difficult to interpret. In Figvr* 3 and Figure 3 , the covariable-adjusted EMM for serum lipids for the 20-group quantiles of PFOA or PFOS are plotted against the median of PFOA or PFOS for the quantile. For PFOA and PFOS, the results dem onstrated a nonlinear association between increasing PFOA or PFOS concentration and total-C and LDL-C. For PFOA, the largest increases in the covariable-ad justed EMM of total-C and LDL-C were seen at the low est range of PFOA concentrations; the slope attenuated at higher serum concentrations. Although population exposure and corresponding serum concentrations of PFOS were lower, the relationship between increasing PFOS quantiles and the covariable-adjusted EMM of total-C and LDL-C was similar across the spectrum of serum PFOS concentrations. The HDL-C level also demonstrated a small, nonlinear association with in creasing PFOS quantile. Logistic regression results are given in T able 3. In creasing PFOA and PFOS quintiles were positively as sociated with an increased risk of abnormal total-C (ad justed odds ratio, 1.2 [95% confidence interval, 1.1-1.4] and 1.6 [1.4-1.9], respectively) and LDL-C (1.4 [1.21.7] and 1.6 [1.3-1.9], respectively). Increasing PFOS quintiles were also associated with decreased risk of low HDL-C (adjusted odds ratio, 0.7 [95% confidence inter val, 0.6-0.9]). Neither PFOA nor PFOS was associated with an increased risk of abnormal triglycerides. (REPRINTED) ARCH PEDIATR ADOLESC MED/VOL IM (NO. 9). SEP 2010 WWW.ARCKPEDIATRICS.COM 864 Downloaded from www.archpediairics.com , on September?, 2010 0 2 0 1 0 American Medical Association. All rights reserved. p.7 Table 2; Differences in PFOA and PFOS Between First and Fifth Quintile EMMs (GLM Analysis) and Assessment of Quintile Trend (Regression Analysis) Age Group, y PfOA : 1.0-11.9 Bothsexes Girts '; . Boys 12.0-17.9 Both sexes Girts. ; '. Boys pros : 1.0-11.9, Both sexes Girls Boys 12.0-17.9 Both sexes Girts Boys t t Difference in No. (Fasting)* EMM, mg/dL 3857 (803) 1886(397) , . 1971 (406) 5293 (1428) 2520(687) . 2773(741) 5.8 . .... 5.8 , 6.3 4.2 3.9 4.8 3857(803) 1886 (397) 1971 (406) 5293(1428) 2520 (687) 2773 (741) 5.5 4.6 6.2 9.5 9.4 9.3 Tot!-C* 0 tor Trend ; (SE) 1.3 (0.3) 11(0.4) 1.6 (0.4) 1.1 (0.3) 10(0.4) 11 (0.4) 13(0.3) 13(0.5) 12(0.5) 2.1 (0.3) 19(0.4) 2.1 (0.4) I P . ' 'lV'->:' -D1ifference Irt EMM, mg/dL ; HDLrC". 0 lor,Trend , (SE) .<.001 i < 10 T: -0)02(0.1) <001 <10 ; 0.02.(0:2) <.001 ..J.v'';-<1;p I T ; -0.06(0.2) <001 .02 .005 <1.0 <1.0 <10 0.1 (0.1) 0.3 (0)2) 0.03 (0.1) <.001 .004 .01 <001 <001 <.001 1.6 . <10 2.6 1,5' 1.8 11 0.3 (0.1) 0.1 (0.2) 0.5 (0.2) 0.8 (0.1) 0.3 (03) 0.4 (0.1) 1 f, .88 .90 : .70 .20 .09 . .80 ,007 50 .003 .001 .06 .003 Age Group, y PFOA 1.0-11.9 , Both sexes Girls Boys ,12.0-17.9 Both sexes Girls Boys PFOS 1.0-11.9 Both sexes Girls Boys 12.0-17.9 Both sexes Girts Boys No. (Fading)3 3857(803) 1886(397) . 1971(406) 5293(1428) 2520(687) 2773(741) 3857(803) 1886 (397) 1971 (406) 5293(1428) 2520(687) 2773(741) J Difference In EMM, mg/dL LDL-C"-' 0 forTrend 0 (SE) 4.9 10(0.3) 5.4 0.8 (0A) 4.8 11 (014) 3.2 0.7 (0.2) 3.2 0.7 (0.4) 3.5 0.7 (0.3) . Fasting Triglycerides" 1 . Differencein 0 torTrend 0 P i . : -|MMi:ipg/dl. v;. (SE) . "'J'. .001 2.0 .04 16.2 .004 5.3 2.0(13) 4.0 (19) 0.4 (1.9) .004 3.8 .05 1.8 103 5,9 15(1.1) 0.8(14) 2.4(16) 3.4 0.9 (03) , ; .002 2.8 , 0.1 (14) 2.6 0.8 (0.4) .04 .7 .6 0.6 (19) 4.1 0.9 (0.4) ,03 ,-1.4 ; ; -03(2.0) 75 17(0.2) <001 <1.0 -0.1(10) 6.9 1.5 (0A) <.001 -13.4 ; -3.0(13) 7.9 13(0.3) <001 111 2.2 (1.6) I P. .10 .04 30 .10 .60 .10 .99 .70 .90 .90 .02 .20 Abbreviations: EMM. estimated marginal mean; GLM, general linear model; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; PFOA, perfluorooctanoic acid; PFQS, perfluoroactanesuifonate;total-C. total cholesterol. SI conversion factors; Toconvert HDL-C. LDL-C. and total-C tomillimoles per liter,multiply by0.0259; tasting triglycerides to millimoles per liter, multiply by 0.0113. 1The first number is thetotal tasting and nonfasting participantsand so is the sample size fortheanalyses of total-C, HDL-C, and LDL-C; thenumber in the parentheses is fasting participants onlyandso is thesample size for analysis offasting triglycerides. bModelswereadjusted for age. estimated time of fasting, body mass index zscore, sex, and regular exercise; sex-stratified models were notadjustedfor sex. 'Calculated tor participants witha triglyceride level <400 mg/dL regardless of fasting status. ' Defined as self-reported fasting ==6hours before phlebotomy. Results of analyses assessing interaction appear in Tetbla 4 . For total-C, LDL-C. and triglycerides, results do not support an interaction between PFOA and PFOS in the prediction of these lipids. There is some evidence of mul tiplicative interaction between PFOA and PFOS in the re duction of risk of low HDL-C (the odds ratio for group 4 exceeds the product of the odds ratios for groups 2 and 3). Sensitivity analyses were conducted to assess the sta bility of the association between PFOA or PFOS and lipid levels after the inclusion of household income and for fasting-only participants. Results (not shown) sug gested thatafter adjustment for the same covariables, mod els were stable and associations were unaltered. The posi tive, statistically significant association between both PFOA and PFOS and total-C and LDL-C was not altered after the inclusion of household income (a proxy for socio economic status) or when analysis was performed only on the subset of participants who had completed a fast of 6 hours or more. Likewise, the positive, statistically sig nificant association between PFOA and fasting triglyc- (REPR1NTED) ARCH PEDIATR ADOLESC. MED/VOL 64 (NO. 9). SEP 2010 863 WWW.ARCHPEDIATRICS.COM Downloaded from vwvw.arohpedialrics.com , on September 7,2010 2010 American Medical Association. All rights reserved. p.8 figure 2. Nonlinear changes incovariable-adjusted estimated marginal means [general linear model analysis) across perfluorooctanoicadd (PFOA) 20 group quantiles. Models wereadjusted for age, estimated time ot tasting, body massindex7 score, sex. and regular exercise, lipid values are presentedas mean (SE). Values for PFOA are presentedas themedian value for the quantilegroup inquestion (axis truncated because ofvery large range in median values * PFOA serum concentration).A. Total cholesterol (total-C). B. High-densitylipoprotein cholesterol (HDl-C). C. Low-density lipoprotein cholesterol (LDL-C). D, Fasting triglycerides. To convert total-C, HDL-C, and LDL-C to millimolespertiter, multiplyby 0.0259; fasting triglycerides to millimoles perliter, multiplyby0.0113. erides was not altered after the inclusion of household income. Perfluorooctanesulfonatc was not associated with fasting triglycerides, with or without the inclusion of household income. The association between PFOA and HDL-C was not statistically significant, with or without the inclusion of household income or for the Easting or nonfasting participants. The positive, linear association between PFOS and HDL-C was statistically significant, with and w ithout the inclusion of household income and for the fasting and nonfasting participants. COMMENT To our knowledge, this study reports the first assess m ent of associations between PFOA and PFOS and se rum lipids in children and adolescents from the largest com m unity-based study of the effects of PFOA expo sure to date. Across several types of analyses, results con sistently provided evidence for a positive association be tween PFOA and PFOS and serum lipids, specifically an increase in total-C and LDL-C with increasing PFOA and PFOS serum concentrations. Dose-response relation ships were nonlinear, with larger increases in lipids at the lower range of PFOA concentration in particular. Ad ditionally, results suggested that the magnitude of asso ciation between PFOS and total-C and LDL-C was higher than that between PFOA an d total-C and LDL-C. Fi nally, there was a statistically significantly increased risk of abnonnal total-C and LDL-C with increasing PFOA and PFOS serum concentrations. Results reported in this article are consistent with those of previous studies in adults, which have generally shown a trend toward a positive association between PFOA and cholesterol. Our observations are also consistent with a study of adults (2 18 years) from the same Project popu lation, in which authors reported an 11- to 12-mg/dL in crease in total-C from the lowest to highest decile of se rum PFOA (vs our finding of a 4.6-mg/dL increase from the lowest to highest quintile) and a corresponding 8to 9-mg/dL increase in LDL-C (vs 3.8 mg/dL reported in this article), and a 10- to 12-mg/dL increase, in total-C from the lowest to highest decile of serum PFOS (vs our finding of an 8.5-mg/dL increase from the lowest to high est quintile) and a corresponding 11- to 12-mg/dL in crease in LDL-C (vs 5.8 mg/dL reported in this article).14 Evidence from animal studies has suggested that acti vation of PPAR-a is an important mechanism through which PFOA and PFOS exert biological effects, although it is unclear whether this animal-based evidence can be extrapolated to humans. The elimination time for PFOA and PFOS in humans is substantially longer than in ro dents, resulting in a longer duration to reach a steady- (REPRINTED) ARCH PED1ATR ADOLESf. MED/VOL 164 (NO. 9). SEP 2010 WWW.ARCHPEDIATRTCS.COM 866 Downloaded from www archpediavrics.com , on September 7,2010 @2010 American Medical Association. All rights reserved. P-9 Figure 3. Nonlinearchanges incovariable-adjusted estimated marginal means (general linear model analysis) across pertluorooctanesulfonate (PFOS) 20-group quantiles. Modelswereadiusled for age, estimated timeoffasting, body mass index?score, sex, and regular exercise. Lipidvaluesare presented as mean (SE). Valuesfor PFOS arepresentedas the medianvalve tor the quantile groupin question. A, Iota! cholesterol (totai-C). B, High-density lipoprotein cholesterol (HDL-C). C. Low-density Hpoproteincholesterol (LDL-C). D, Fasting triglycerides. Toconverttotal-C, HDL-C. and LOL-C to millimoles perliter, multiply by0.0259; fasting triglycerides tomillimoles per liter, multiply by0.0113. TablB3.RiskolbnonnalBlodSerumlJ|iids(LogisPRegr8anAnalyss)BasedonlncreasingPFOAandPfOSQuiiiUles OddsHatio (^onfldenMtnterval) ; Quintile " :; PFOA First ..Second . ; Third Fourth .Fifth. . pros; ' . ; Rrtt ; Second . Third1 ' Fbyrth, ' .Fifth-" ' -t'T ria t-C * . " ,, HDL-C.*' ; (DL-C*1 c " 1.0 [Reference} -, - . -- 1.0 [Reference]'-' - i-0.(0.8-?t:a) 'T2<1<M 4) i.oiQ.:t)=, ; 1.2 (in -1:4) ' v ' : 1.0(0.0-12)- - 09(0.8-1.1) -- " ,; '-T'v-j.Vf t.o [Reference] 1 2 (1.0-1.5) > 1.2(L0;1.4) -:v T2:(1.0-1;4) - 1.4(1.2-17} ,, - 1.[Reference] 1.0 (Reference) .,t& (0 ,& U ):- ' -T-..c=:."; r ::-`,.T:-"2(i J' :: -7 ;(i.-2 -i:5 ) ` '0.8(07-10) ' V '-1.2 (1,0-15) ^o.BLCO./^J.) : - ! S ` '^ I 3 ( y - 1 0 ) .// 16(12-1,9) ' : Fasting Trlglyeerfd?^ 1.0 [Reference] 1.0'(0.7-1.5) 1.3 (.9-1.9) 1;6 (1.1-2.3) 1.0(07-1.6} 1.0 [Reference) 1.3 (0.9-1.8) 1,0(07-1.4) 11(07-16) 1,2(0.8-15) Abbreviations; HDL-C, bigh-density lipoproteincholesterol; LDL-C. low-densi1y lipoprotein cholesterol; PFOA perftuorooctanoic acid; PFOS. pertluorooctanesulfonate;total-C, total cholesterol. ` Models wereadjusted forage. estimated timeof fasting, body massindex/score, sex, and regular exercise. ``Calculated for participants with atriglycride level <400 mg/dL (to convertto millimolesper liter, multiply by 0.0113) regardless offasting status. cDefinedas self-reported fasting a6 hours before phlebotomy. state dose.20 In addition, PPAR-a is expressed in human liver tissue at approximately 10% of rodent levels, and hu mans and other primates are refractory or less responsive to PPAR-a agonists compared with rodents.21 Thus, as sociations reported in this article are etiologically plau sible if a non-PPAR-a mechanistic pathway operates in humans in addition to or instead of a PPAR-a pathway; some studies of hum ans and other primates have re- (REPRINTED) ARCH PEDIATR ADOLESC. MED/VOL 164 (N. 9 ). SEP 2010 WWW.ARCHPED1ATRICS.COM 867 Downloaded from www.archpediatrics.com , on September 7, 2010 2010 American Medical Association. AL! rights reserved. Financial Disclosure: Ms FrisbeeandDrs Shankar, Knox, and Ducatman were engaged in the Project pursuant to a contractual relationship between Brookmar Inc and West Virginia University. This study was funded in part by that contract. Drs Steenland, Savitz, and Fletcher are mem bers of the separate C8 Science Panel, whose Wood County Circuit Court-appointed obligation is to deter mine "probable links" between PFOA exposure and health outcomes. Panel membership was jointly agreed to by the Plaintiffs and DuPont and then approved by the Court. As such, these authors are in receipt of funding from the C8 Class Action Settlement Agreement. Funding/Support: The C8 Health Project was funded by the Settlement Agreement in the case of Leach v E.I. du Pont dc Nemours & Co., Civil Action No. 01-C-608 (W. Vo. Circuit Court of Wood County, 2004), which resulted from drinking water contamination by the chemical perfluorooctanoic acid (PFOA, or C8). The Court authorized the creation of Brookmar Inc, an independent company cre ated solely to design, publicize, and implement the C8 Health Project, including all enrollment and data collec tion. Brookmar Inc received funding exclusively from the Settlement, administered through a Court-approved fi nancial professional (the Health Project Administra tor). Brookmar Inc made data from theC 8 Health Project available to the C8 Science Panel, also created pursuant to the Settlement Agreement, and to a group at West Vir ginia University, pursuant to a contractual relationship with Brookmar Inc. The design of the Project was devel oped in consultation with, although not subject to, the wishes of the settling parties. All authors declare that their ability to design, conduct, interpret, and publish this re search study was unimpeded and fully independent of the Court and settling parties, who had no role in pre paring, reviewing, or approving the manuscript. Additional Contributions: We gratefully acknowledge the contributions of Cathy Lally, MPH, and Jessica MacNeil, MPH, from the Emory University Rollins School of Public Health for their expertise and assistance in data cleaning and the personnel at Exygen Research Corpo ration and Axys Analytics for ensuring accurate descrip tions oflaboratory methods. Finally, because of the scale of the C8 Health Project, many individuals provided valu able assistance during its development and implemen tation. We gratefully acknowledge these contributions and thank all individuals who supported the execution of the project.*12 REFERENCES 1. LauC. AnitoleK. ModesC. LaiD. Pfahles-HulchensA,SeedJ. Perfluoroalkylac ids: a reviewol monitoring and toxicological findings. Toxicol Sci. 2007:99 <2):366-394 2. KennedyGLJr,ButenhotfJL, OlsenGW.etal.Thetoxicologyofgerfluorooctanoate. Crit Rev Toxicol. 2004:34(4):351-384. 3. KudoN. KawashimaV. Toxicityaodtoxicokineticsof perfluorooctanoicacid in humans andanimals. J ToxicolSo. 2003;28(2):49-57. 4. CaiafatAM,KuldenyikZ,RekJyJA,CaudillSP, TollyJS, NeedhamLL. Serumcon centrationsof 11 polyfiuoroalkyl compounds intheUSpopulation: datafromthe NationalHealthandNutritionExaminationSurvey(NHANES) EnvironSd Techno! 2007:41(7):2237-2242. 5. CaiafatAM.WongL-Y,KuklenyikZ, ReidyJA. NeedhamLL. Polyfiuoroalkylchemi cals IntheUS population:dataIrontPieNationalHealthandNutritionExamina tion Survey (NHANES) 2003-2004andcomparisons withNHANES1999-2000. Environ HealthPerspect. 200?;115(11):1596-1602. 6. Klrrman A. Ericson I, vanSavel B. etal. Exposure ofperfluorinated chemicals throughlactation: levels ofmatchedhumanmilkandserumandatemporal trend, 1996-2004. inSweden. EnvironHealthPerspect 2007:115(2):226-230. 7. InoueK, OkadaF, ItoR,etal. Perftuorooctanesulfonate(PFOS)andrelatedper- ftuorinatedcompounds in humanmaternal and cord bloodsamples: assess mentof PFOS exposureina susceptiblepopulation during pregnancy. Environ Health Perspect 2004;112(11):1204-1207. 8. OlsenGW, Logan PW, HansenKJ, etal.Anoccupationalexposureassessment ofaperftuorooclanesulfonylfluorideproductionsite:tamonttoring.AIHAJ (Fair fax. /a;. 2003:64{5):651-659. 9. Sakr CJ. Leonard RC, Xreckmann KH, Slade M0. Cullen MR. Longitudinal study of serumlipids andliver enzymes in workers with occupational expo sure to ammonium perfluorooctanoate. J Occup Environ Med. 2007:49(8): 872-879. 10. SakrCJ, KreckmannKH,GreenJW,GilliesPJ, ReynoldsJL. LeonardRC.Crosssectional studyol lipidsandliverenzymes relatedtoaserumDiomarkerol ex posure(ammoniumperfluorooctanoateor APFO)as partofageneral healthsur veyinacohortofoccupationallyexposedworkers. J OccupEnvironMed.2007: 49(10):1086-1096. 11. CostaG.SartoriS,ConsonniD.Thirtyyearsofmedicalsurveillanceinpetfluooclanoicacidproductionworkers. J OccupEnvironMed. 290931(3):364-372. 12. OlsenGW. ZobelLR. Assessmentoflipid,hepatic,andthyroidparameterswith serumperfluerooctanoate(FPOA)concentrations infhrorochemicalproduction workers. IniArchOccupEnviron Health. 2007;81(2):231-246. 13. EmmettEA.ZhangH.ShoferFS,etal.Communityexposuretoperfluorooctano ate:relationshipsbetweenserumlevelsandcertainhealthparameters.J Occup EnvironMed. 2006:48(8):771-779. 14. SteenlandK, TinkerS, FrisbeeS, DucatmanA.VaccarinoV.Associationof per- fluoroocsnoicacidandpsriluorooctanesulfonatewithserumlipidsamongadults living nearachemicalplant. AmJEpidemiol 2009:170(10):A268-1278. 15. Frisbee SJ, BrooksAP Jr, MaherA, etal. TheC8 Health Project design, meth ods, andparticipants. Environ Health Perspect 2009;117(12):1873-1882. 16. JackLeachel olv El du PontdeNemoursandCompany, 01C-608 (WV2002). 17. EmmeitEA,ShoferFS,ZhangH,FreemanD,Desai C,ShawLNLCommunityex posureto perfluorooctanoate: relationships betweenserumconcentrationsand exposuresources. J OccupEnvironMed. 2006;48(8):759-770. 18. FlahertyJM. Connolly PD, Decker ER, etal. Duantitative determinationol per fluorooctanoicacidinserumandplasmabyliquidchromatographytandemmass spectrometry. J Chromatogr 8 Ana/yt Techno! BiomadLife Sci. 2005:819(2): 329-338. 19. AmericanHeartAssociation.AmericanHeartAssociationscientificposition:cho lesterol and atherosclerosis inchildren, httpyrwwwjmericanheart.org/presenter .jhtml?identitier-4499.AccessedOctober 19,2009. 20. HundleySG, SarrifAM, KennedyGLJr. Absorption, distribution,andexcretion or ammoniumperfluorooctanoate (APFO) afterora! administration tovarious species. Drug Chem Toxicol. 2006:29(2):137-14S. 21. Klaunig JE. Babich MA, Baetcke KP. etal. PPARn agonist-induced rodenttu mors: modes of action and human relevance. Crit Bev Toxicol. 2003:33(6): 655-780. 22. OkochiE, Nishimaki-MogamiT, SuzukiK. TakahashiA Perfluorooctanoicacid, a peroxisome-proliferatinghypoipidemicagenLdissociatesapoipoprotein848from lipoproteinparticlesanddecreasessecretion ofverylowdensitylipoproteins by culturedrathepatocytes. BiochimBiophysActa. 1999:1437(3):393-401. 23. PublicHealthService,AgencylorToxicSubstancesandDiseaseRegistry.Draft ToxicologicalProfile forPeriluoroalkyts.Atlanta,GA:DepartmentofHealthand HumanServices; 2009. (REPRINTED) ARCH PEDIATR ADOLESC MED/VOL 164 (NO. 9). SEP 2010 WWW.ARCHPED1ATR1CS.COM 869 Downloaded from www.archpediatrics.coin , on September 7,2010 2010 Am erican Medical Association. All rights reserved.