Document gaG2qvMNww8qGBaDjJNVQKwwN

J09M -Volume41,Number 9,SePtember1999 799 Serum PerfluorooctaneSulfonateand Hepatic 90 and Lipid CII.McalChemistry Tests "n fluorochemical Production Employees Geary W. Olsen, DVM, PhD Jean M. BurTis,RN, MPH JeffreyH. Mandel, MD, MPH Larry R. Zobel, MD, MPH he3M Companymanufacturpersod- T uctsthactontaifnluorochemiceail-s, therasintentiocnoamlponentosrresidualimpuritiest,hat have as a precursomroleculpeerfluorooctane The 3M Company manufacturefsluarochemicawlhsi,chhaveas a precursorperfluorooctanesuy'onylfluoyide (CsF,7SO2F). These compounds may be expectedtotransform metabolicallyt,o an undetemined degree,to peiflyoroocianseu@fonate(PFOS, C8F,7SO_3-) as an end- stage metabolite. Subchronic studies in rats and primates indicate a potential for cumulative toxicitywith PFOS with the pfimaiy effect sulfonyflluorid(eCSF17So2F). Work- placeexposuremay occurby inhala- tion, ingestion, and dermal routes. These fluorochemicalsmay u-ansform metabolicay, to an undetermineddegree, to perfluorooctane sulfonate (PFOS; CF,7SO3-) as an end-stage relatedto metabolic wasting with hypoliptdemza as a consutentfinding. metabolite.' Potassium perfluo- Biennial medical surveillance has been offered to the company's rooctanesulfonate(CsF,7SO3 K') is fluorochemzcal pmduction *workers locakd in Decatur, Alabama, and Antwerp, Belgium. In 1995, the mean serum PFOS level,as measured by high-pe@bmance liquidchromatography mass spectromet-ryofr 178 male employeeswas 2.19 partsper miuion (ppm; range, 0.00 to 12.83 ppm), and in 1997, for 149 male employees,itwas 1.75 ppm (0.10 to 9.93 ppm). Our analyses suggest that among these prod@ction employees, there were no substantial changes in -serum hepatic enzymes, cholesterolo,r lipopmteins associated with PFOS levelslessthan 6 ppm. It was not possible to derive inferencesftom thefew employees who had serum PFOS levels2: -6 ppm. These resultsmay be due to the lower Levels of serum PFOS measured among these production employees, compared to those suspected to cause effectsin laboratory animals. itselaf surfactanutsed as a wettingand foaming agentin industriaalnd commercialprocesses. PFOS isknown toconcentratperimarilyin theliverand, to a 10-fold lesser extent, in the plasma of rats.2 There, appears to be significant enterohepaticcirculatioonf PFOS with both urinary and fecal excretion2..3 Cholestyramine decreased the retention of radiolabeled PFOS in the liver and plasma and increased its eliminationvia feceS.3Subchronic studiesin ratsand primatessuggest there is cumulativetoxicitywith PFOS.4-7 Lowered serum totalcho- lesterolevelsappear to be an early consistentfindin-,with cumulative toxicitryesultinign metabolicwast- ing and ultimatelydeath.Although themechanism of toxicitiyn labora- toryanimalsisnot fullyunderstood, itmay be due to an effecton fatty Fiximthe Medical Department.3M Company, St.Paul.Minn. Address correspondence to: Geary Olsen. DVM. PhD. Medical Department. 3m 220-3W-0i. St.Paul,MN 55144. Copyri-,h0t by American Collegeof Occupationaland EnvironmentalMedicine Company. acidtransportand metabolism,membrane function.peroxisome proliferation,and/ormitochondrialbioenercetics.8-I1 800 PFOS,HepaticEnzymes,and CholesteroOllsonetal Voluntarbyienniamledicalsur- reductionwas due to a largeturnover thatwere consideredin the analyses. veillanceof*fluorochemicalproduc- of employeesat both plantlocations For stratifieadnalyses.employees tionemployees has been routinely during1996 and 1997.The surveil- were dividedintofourPFOS catego- performedsincethe late1970s at lanceconsistedof a medicalques- ries(0to<I ppm. Ito< 3 ppm.3 to 3M's Decatur.Alabama, and Ant- tionnairem:easurement of height. <6 ppm. and a-6ppm) to determine %%,erpB,elgium. locationsT.otal se- weight.and blood pressure,standard whetheran effectcouldbe detectedat rum organicfluorinelevelswere an- clinicaclhemistryand hematology thehighestserum PFOS levelsO.ther al),zeduntilthe mid-1990s when tests:and determinationof serum categoricallevelswere used thatpro- serum PFOS determinationq,uantifi- PFOS levels. videdsimilaresultsF.or mul6vwiable able by high-performanceliquid chromatographymass spectrometry. PFOS Analysis regressioanalyses,erumPFOS and thepotentiaclonfoundersof age,BMI, became availablen.e purpose of In 1995,the analysisfor serum alcoholuse.and cigarettuese were thisstudywas to providea cross- PFOS was conductedby 3M's Envi- examined as continuousand categorisectionaalnalysisof themedicalsur- ronmentalLaboratoryin St. Paul. calexplanatorvyariableisn themod- veillancdeatain relatiotno theem- Minnesota.The method used tetrabu- els.Muldvariableregressionmodels ployees'serum PFOS levels. tylammonium toion-pairwithPFOS were fittedw,ith PFOS analyzedas a in serum.13 The ion-pairswere then continuousvariableu,sing linearas extractedwith ethylacetateT.he ab- well as non-Hneartransformationisn Methods stractiopnroductwas then analyzed orderto maxirnizethepossibiliotfy Fluorochemical Production usinghigh-performanceliquidchr6- findingassociationbsetween PFOS matography-ther7nosprmaayss spec- and thedependentvariableof interest. Fluorochemicalproductionbegan trometryI.n 1997,theserum samples Naturallogtransformatioonfsthede- inDecaturin 1961 and Antwerp in were analyzedby liquidchromatog- pendent variableswere performed, 1976. In general,perfluorinated raphy/massspectrometry,using se- when necessary,to normalizevari- chemicalsareproduced via an elec- lectedion monitoringin the nega- ablesand toenhancemodel fitT.radi- trochemicalprocess:a solutionof tive-iomnode."'-is tionalstepwiseselectionprocedures organicsubsmte iselec"lyzed in anhydroushydrogen fluorideat a LaboratoryAnalyses were alsoutilize(dselectioinnand out ofmodelwas setatP = 0.1),asweH low voltage.1T2he productsof this For both time periodsand plant astakingintoaccountothercovariates electrolysciesllreactionare highly locationsU,nited LaboratoryServices thatmay be on thebiologicpathwayof fluorinatecdompounds, with the (St.Paul,W performedthestandard effecl" We didnotexan-dnechanges end-productdefinedby the starting hematologicaalnd serum chemistry in measured PFOS betweenthe two materialP.otentiaflorworkplaceex- testsT.hese includedthe following: time periodsbecause the estimated posureto thosefluorochemicaltshat allWinephosphatase(IU/L),gamma half-lifoef PFOS in the serum isbe- metabolizeto PFOS can occurinthe glutainytlransferas(eTU/L),aspartate lievedto range between 1000 and electrochemicaclell,the reactor, aminou-ansferas(eIU/L),alanineami- 1500 days (J.H. Mandel,MD, unpub- mixing,drumming, and packaging no=mferase (ITJ/L)t,otaland direct lisheddatabasedon 4 yearsof serum areas,aswellas intheplant'squality bilirubi(nmg/dL),cholestero(lmg/ measurementsof threeretirees1,998). assuranceand researchand develop- dL),low-densitpyrotein(mg/dL), Study resultwsere analyzedusingthe ment laboratories. high-densitcyholestero(lHDL; mg/ SAS System.17 Subject Selection dL)and triglyceri(dmegs/dL).Clinical chemistry,hematology, and serum Voluntaryfluorochemicamledical PFOS detem-dnationwsere performed surveillanceexaminations are of- on overnightfastedblood samples. Results fered biennialltyo approximately 200 Antwerp and 300 Decatur pro- Data Analysis The distributioofnemployees,by serum PFOS exposure categoriza- ductionemployees.In 1995,a total Descriptives,imple,and stratifiedtion,is presentedin Table 1. of 178 male employees (Antwerp: analysesa,nalysesof variancea,nd or- Whereas 20% of the Decaturem- n = 88; Decatur:n = 90) partici- dinarymultivariablreegressiownere ployeeshad exposuresat2:3ppm for patedin themedicalsurveillanceex- used toevaluateassociationbsetween both years,thisproportionin Ant- aminationsI.n 1997, 149 male em- PFOS and each hematologicaland werp declinedfrom 25% in 1995 to ployees(Antwerp:ii= 65. Decatur: clinicaclhemistrytestA.ge. body mass 13% in 1997.For both years.95% of n = 84) participate(dT.herewere too index(BMI: kg/in2)c.urr-enatlcohol the employees'serum PFOS levels few femaleemployees to includein consumption(drinksperday),and cig- were below 6 ppm. There were no thedataanalysis.S)ixty-oneemploy- arettuese (cigarettsemsoked perday) PFOS measurements 2!6 ppm in ees participateidn both years.This were potentialconfounding factor% Antwerp in 1997. JOEM -Volume41,Number 9,September1999 801 TABLE1 DistributoifoEnmployeesb,y Year,Locatioann,dPerfluorooctSaunlefonat(ePFOS)ExposurLeevel(inpartsper milli[opnpm]) 1995 Data 1997 Data AllEmployees Antwerp Decatur AllEmployees Antwerp Decatur PFOS n % n % n % n % n % n % 0 to <1 ppm 45 25 34 39 11 12 60 40 31 48 29 35 1 to <3 ppm 91 51 32 36 59 66 63 43 25 38 38 45 3 to <6 ppm 35 20 19 22 16 18 21 14 9 14 12 14 @-6 ppm 7 4 3 3 4 4 5 3 0 0 5 6 Total 178 100 88 100 90 100 149 100 65 100 84 100 TABLE 2 Mean ValuesofPFOS, Demographic,Serum Chemistry,and Hematologic Parameters for Antwerp and Decatur, 1995 and 1997 Examinatione 1995 Date 1997 Date Variable PFOS.ppm Age,years BMI,kg/M2 Cigarettes,no. per day Alcohold.rinksperday AlkPhos,IU/L GGT, IU/L AST, IU/L ALT, IU/L T. bilirubinm,g/dL D.bilirubmign/,dL Cholesteromlg,/dL LDL,mg/dL HDL, mg/dL Triglyceridmegs/,dL Antwerp 1.93 37... 23.94.7* 1.37541 26* 44 0.86'@0.22 214 138 FA.- 115- Decatur 2.44 45 29.2 7.9 0.2 97 48 29 47 0.58 0.21 218 136 43 187 Antwerp 1.48 33'23.55.5 1.1... 70... 26* 27 31 0.80*0.15-* 206 134 so-* ill... Decatur 1.96 44 30.0 6.6 0.1 87 36 26 34 0.58 0.12 215 137 42 192 P < 0.05;- P < 0.01; P < 0.001. BMI, body mass index;AJk Phos, alkalinpehosphatase;GGT, gamma glutamyltransferase;AST, aspartateaminotransferasAeL;T, alanineaminotransferasTe.;,totalD;.,direct; LDL,low-densitlyipoproteiHnD;L, high-densitlyipoprotein. The mean values for PFOS, demographic, and liverand lipidtestresults,by location,are presented in Table 2. The Antwerp male employee population was sicynificantly younc,er than that at Decatur, had lower BMIS, and had higher selfreported daily consumption of alcohol.In addition,theirclinicalprofiles were differentfor several tests.For both time periods,the Antwerp employees had lower mean alkaline phosphatase and triglyceridevalues and hi-her totalbilirubinand HDL values. Table 3 liststhe mean. median, standard deviation, and range of the covariatesand hepatic enzymes, cholesterola,nd lipoproteinsby four levels of PFOS categorization(0 to < 1, I to <3, 3 to <6, and a:6 ppm) for the combined populations for each surveillanceyear. (Hematology and other clinicalchemistry data were unremarkable and are not shown.) Several observations are noteworthy. First,the mean for the @t6 ppm PFOS category was one order of magnitude higher than the lowest PFOS category (0 to < I ppm) for both years.Also, the means of the four PFOS categories were significantlydifferentfrom each other.Second, there was only one variable. totalbilirubint,hathad significant (P < 0.05)F testsfordifferenceisn means in both years of analysis. Stratification by plant location and surveillance year did not result in significantfindingsfor totalbilirubin or directbilirubin(Table 4). Third, mean serum cholesterollevels remained constant (1995) or increased (1997) with higher PFOS category levels,although the HDL mean values trended lower among those employees with the highest PFOS categorizations.Stratificatiobny plant location and surveillanceyear did not resultin significantlydifferent mean cholesterolor HDL values by PFOS categories(Table 5). Finally, itshould be noted that employees in the highest PFOS category were older and had higher BMIs than did employees in the lowest PFOS categoiy. Furthermore, in 1997, the employees with at6 ppm serum PFOS levelswere only from Decatur (Table 1). thus the analyses may be confounded because Decatur employees were generally older and heavier than Antwerp employees. For example, in 1997 the mean HDL level for the a6 ppm group was 40 mg/dL (Table 3). However. this mean value was solely from Decatur employees. Linear and nonlinear relationships between PFOS and the dependent variablesof interest,takinc,into account the potentialconfoundina affectsof age. BMI. alcohol,and cigarettes.resulted in many analyses. Total bilirubinshowed a significant 802 PFOS, HepaticEnzymes, and Cholesterol-Olsonatal TABLE3 Mean,MedianS,tandarDdeviatio(nSD)ofMean andRangeofPFOS,byDemographiacndSerumChemistrifeosrAntwerp andDecatuErmployeesCombined,for1995(n= 178)and1997(n= 147) 1996Date 1997Data PFOS'(ppm) PFOS,ppm 0 to<1 1 to<3 3 to <6 2--6 Age,years 0 to <1 1 to <3 3 to <6 a6 Alcohol,drinksperday 0 to<1 1 to<3 3 to<6 a6 BMI, kg/m2 0 to <1 1 to <3 3 to <6 2:6 Cigarettesn,o.per day 0 to<1 1 to<3 3 to<6 2:6 Alkalinpehosphatase,IU/L 0 to<1 1 to<3 3 to<6 2:6 GGT, U/L 0 to <1 1 to <3 3 to <6 2:6 AST, U/L 0 to<1 1 to<3 3 to<6 a6 ALT. IU/L 0 to <1 1 to <3 3 to <6 a6 T. bilirubimng,/dL 0 to<1 1 to<3 3 to<6 a6 Mean Median SD 0.491 1.82' 4.121' 8.171 F value 0.50 0.27 1.77 0.58 3.97 0.81 7.73 2.52 321.9,P < 0.0001 37 422 40 45 F value 36 a 41 9 40 7 43 7 3.7,P - 0.02 0.8 0.5 1.2 0.7 F value 0.6 0.9 0.1 0.7 0.3 1.9 0.0 1.1 4.0.P - 0.009 25.5 27.7 24.93 27.7 F value 24.8 4.2 26.3 5.8 25.0 3.8 29.4 4.2 3.7,P = 0.02 2.6 6.8 10.6 0.4 F value 0.0 6.4 0.0 11.3 8.0 12.4 0.0 1.1 4.8,P = 0.003 80 89 86 88 F value 78 89 85 85 1.3,P 22 27 21 24 0.28 43 47 40 43 F value 31 36 39 33 O.S.P 28 39 15 18 0.71 27 29 25 33 F value 48 46 42 51 F value 25 26 24 33 1.8,P 13 12 5 6 0.14 I 43 20 42 21 41 7 49 17 1.0,P - 0.38 0.88 0.70 0.50 0.662 0.642 0.60 0.30 0.60 0.28 0.76 0.70 0.23 F value= 4.4.P = 0.005 Range Mean Median SD Range 0.00to 0.90 1.00to 2.91 3.00to 5.80 6.06 to 12.83 21 to 58 25 to60 26 to55 37 to56 0 to3.6 0 to3.6 0 to6.0 0 to2.9 17.9to38.7 19.6to60.7 17.9to32.5 20.6 to 33.0 0.0to 25.0 0.0to 40.0 0.0to 40.0 0.0to 3.0 31 to158 49 to191 32 to124 63 to136 16 to 155 2 to293 21 to80 28 to 79 15 to 96 14 to 90 13 to 37 26 to 43 27 to 118 18 to 183 30 to59 29 to82 0.40to 2.90 0.20to 1.50 0.20to 1.40 0.50 to 1.20 0.52' 1.781 3.871 7.201 F value 0.52 0.27 1.64 0.56 3.59 0.70 6.68 1.59 367.6.P < 0.0001 36 422 41 42 F value 34 41 42 45 5.1,P 11 9 5 9 0.002 0.5 0.5 1.0 0.2 F value 0.1 0.1 0.1 0.1 1.8,P 0.8 0.8 1.8 0.2 0.15 26.0 27.7 27.3 30.8 F value 24.9 26.4 27.9 29.7 2.1.P 4.9 5.7 4.4 4.0 0.10 4.7 8.2 4.1 6.0 F value 0.0 0.0 O.D 0.0 1.5,P 9.4 11.3 8.3 13.4 0.23 77 73 83 79 76 74 88 84 F value 1.2,P 17 23 22 18 0.32 28 36 28 33 F value 22 25 27 37 1.1,P 20 33 14 12 0.34 27 26 25 29 F value 25 25 23 28 0.5.P 7 7 7 3 0.67 31 33 34 41 F value 30 29 31 45 0.9,P 11 16 18 10 = 0.46 0.77 0.612 0.63 0.58 F value 0.60 0.40 0.60 0.21 0.50 0.31 0.50 0.24 2.9.P = 0.04 0.10to 0.97 1.02to 2.89 3.09to 5.30 6.05to 9.3 21 to62 24 to63 32 to54 29 to52 0 to4.3 0 to5.0 0 to7.1 0.1to 0.8 20.1to41.7 18.1to 48.5 19.1to 36.0 26.1to 36.2 0 to40 0 to40 0 to30 0 to 30 49 to 132 41 to 163 29 to 120 65 to 114 10 to 142 10 to 179 13 to71 17 to48 13 to53 15 to56 14 to43 26 to34 13 to60 10 to89 14 to82 25 to49 0.30to2.30 0.30to 1.30 0,40to 1.30 0.40to 1.00 JOEM oVolume41,Number 9,SeptemberIM TABLE3 Continued. PFOS* (ppm) D. bilirubimng,/dL 0 to <1 1 to <3 3 to <6 a:6 Cholesterol.mg/dL t 0 to <1 1 to <3 3 to <6 2--6 LDL, mg/dL 0 to <1 1 to <3 3 to <6 2:6 HDL. mg/dL 0 to <1 1 to <3 3 to <6 a6 Triglyceridesm,gtdl 0 to <1 1 to <3 3 to <6 a-6 Mean 1995 Data Median SD 0.22 0.21 0.21 0.20 F value 0.20 0.05 0.20 0.06 0.20 0.04 0.20 0.02 0.6,P = 0.58 219 216 214 213 F value 215 213 214 221 0.1, P 47 43 35 36 0.96 140 134 137 142 F value 137 134 135 136 0.2,P 43 40 34 32 0.87 53 48 45 45 F value 53 47 46 46 2.9,P 13 13 11 9 0.04 129 161 isa 132 F value 96 133 142 151 1.1.P 98 107 88 45 0.35 Range 0.02 to 0.40 0.10 to 0.40 0.20 to 0.30 0.10 to 0.30 100 to 340 118 to 315 128 to 278 160 to 251 29 to 261 44 to 234 65 to 190 95 to 178 31 to 94 26 to 94 23 to 74 34 to 61 41 to 622 41 to 651 34 to 413 64 to 187 Mean 1997 Date Median SD 0.15 0.122 0.12 0.10 F value 0.10 0.07 0.10 0.04 0.10 0.04 0.10 0.00 3.5.P = 0.02 198 216 2292 229 F value 197 219 224 238 4.3,P 40 42 29 26 0.006 124 141 1482 145 F value 128 134 142 156 3.7.P 34 38 24 26 0.01 46 44 48 40 F value 48 45 47 38 1.1,P 11 10 10 4 0.34 148 156 166 220 F value= 107 122 158 191 0.5,P 162 108 92 83 0.67 Range 0.10 to 0.40 0.10 to 0.20 0.10 to 0.20 0.10 to 0.10 110 to 280 116 to 365 192 to 321 186 to 250 50 to 205 61 to 290 111 to 196 103 to 164 19 to 74 28 to 69 32 to 69 37 to 45 38 to 1209 44 to 534 45 to 394 149 to 352 See Table I forsample sizeby year. Mean issignificantdliyfferen(tP < 0.05,Bonforroni(Dunn)test)from the mean 2 Mean issignificantdliyfferen(tP < 0.05,Bonferron!(Dunn)test)from the mean 3 Mean issignificantlydifferent(P < 0.05, Bonferroni (Dunn) test)from the mean of the otherPFOS categories. of 0 to <1 ppm PFOS category. of 1 to <3 ppm PFOS category. quadraticassociatiownith PFOS for Decaturemployees.That is,totalbilirubinlevelsinitialldyeclinedb.ut subsequentlyincreasedwith increasing PFOS levels.This association was not observed in the Antwerp employee population,which had highertotalbilirubilnevelsI.n 1997, serum cholesterolevelswere positivelyassociated(linearlyw)ith serum PFOS, afteradjustmentfor potentialconfounding factorsamong Decaturemployees (datanotshown). This associationwas not observed with serum PFOS and cholesterolin 1995 among Decatur employees or in eithertime periodwith the Antwerp employees.Stratificatioofnthe databy plant(Table5) and multiva- riableanalyses(data not shown) showed no consistentassociations between PFOS and HDL. For both plantscombined, HDL levelswere significantlnyegativelyassociated (linearlwyi)th PFOS in 1995,after adjustmentfor possibleconfounding factorsb,ut thiswas not observed in 1997. Discussion We conducted two cross-sectional analysesof surveillancdeata to examine the associationsbetween serum PFOS levelsand severalhematologicaalnd clinicaclhemistrytests in male fluorochemicalproduction employees.For both years,95% of the measured serum PFOS levels were below 6 ppm. Because theAntwerp and Decatur employees were dissimilabry age, BMIS, and selfreportedalcoholuse,we conducted combined as well as separateanalyses by plantlocation.These three demographic factorsexplain,atleast partiallyw,hy the Antwerp employees had lower mean serum triglyceridesand higherHDL levelsthan the Decatur employees.113-22 Clinicalhepaticenzyme testswere not associatedwith the employees' serum PFOS levels.This was an importantquestionto address becausePFOS has been reported(1)to be a peroxisome proliferatoirn ratsg*lo(;2) to resultin increased plasma livertransaminasetestsin a 804 PFOS,HepaticEnzymes,and CholesterolOlsonatal TABLE4 Mean,Median(Med)S,tandarDdeviati(oSnD)ofMeanandRangeofPFOS,andTotalandDirecBtilirubLienvelsb,yYear and PlantLocation Antwerp Decatur PFOS* (ppm) 1995 Data T. bilirubimng,/dL 0 to<1 1 to<3 3 to <6 a6 D. bilirubimng,/dL 0 to <1 1 to <3 3 to <6 a6 1997 Data T.bilinibimng,/dL 0 to<1 1 to <3 3 to <6 a6 D. bilirubimng,/dL 0 to <1 1 to <3 3 to <6 a6 Mean Med SD Range Mean Med SD Range 0.96 0.83 0.75 0.93 F value 0.80 0.55 0.80 0.26 0.70 0.30 0.90 0.25 1.2,P 0.31 0.23 0.22 0.21 0.20 F value 0.20 0.06 0.20 0.04 0.20 0.03 0.20 0.00 0.7.P 0.55 0.90 0.68 0.79 - F value 0.80 0.46 0.70 0.23 0.70 0.40 - - 2.3,P = 0.11 0.16 0.13 0.14 - F value 0.20 0.08 0.10 0.05 0.10 0.05 - 2.2,P = 0.12 0.40to 2.90 0.40to 1.30 0.30 to 1.40 0.70 to 1.20 0.20to 0.40 0.20to 0.30 0.20to 0.30 0.20to 0.20 0.40to 2.30 0.30to 1.30 0.30to 1.30 0.10 to 0.40 0.10 to0.20 0.10to 0.20 0.65 0.57 0.51 0.63 F value 0.60 0.50 0.50 0.65 0.7.P 0.16 0.28 0.18 0.10 0.54 0.20 0.20 0.20 0.20 F value 0.20 0.00 0.20 0.06 0.20 0.04 0.20 0.08 0.4,P = 0.74 0.63 0.56 0.51 0.58 F value 0.60 0.50 0.50 0.50 1.0,P 0.30 0.18 0.16 0.24 0.41 0.13 0.11 0.11 0.10 F value= 0.10 0.10 0.10 0.10 1.3.P 0.06 0.03 0.03 0.00 0.28 0.40to 0.90 0.20to 1.50 0.20to 1.00 0.50 to 0.70 0.20to 0.20 0.10to 0.40 0.20to 0.30 0.10 to 0.30 0.30to 1.40 0.30to 1.00 0.30to 0.90 0.40to 1.00 0.10 to 0.30 0.10 to 0.20 0.10 to 0.20 0.10 to 0.10 See Table1 forsample sizesby year. subchronicrat study4;and (3) to increaseserum aspartateaminotransferaseand lower alkalinephosphatase levelsin a subchronicrhesus monkey study.5 There appears to be significant enterohepaticirculationof PFOS with bothurinaryand fecalexcretion in the rat.2*A3lthough we observed a quadraticrelationbetween PFOS and totalbilirubiinn our multivariable analyseswith the Decatur employee population,interpretationf these data is difficulgti,ven the narrow range of PFOS serum levelsassociatedwith the totalbilirubivnalues. Interestingltyh,e Antwerp employees'totalbilirubilnevelswere significantlyhigher than the Decatur employees'levels.We suspect there may be a greaterprevalenceof Gilbert'ssyndrome2"-23among the Antwerp employees.In 1995. 15 (17%) Antwerp employees had totalbilirubin values2ti.2 mg/dL, compared with three(3%) Decaturemployees. Comparable percentagesalsoexisted in 1997.We are uncertainwhether post-collectifoanctorsi,ncludinghemolysis,light,and heat,could have contributedto the differenceisn total bilirubilnevelsbetween theAntwerp and Decatur employees.18Unfortunately,totalbilirubiwnas not analyzedintwo subchronicrhesusmonkey studies,which resultedindeath of alla.nimalsin the 4.5 mg/kg/dAy PFOS dose group or higher.5.O6ngoing toxicologystudiesin ratsand primatesshould provide additional perspectivesregardingany biologicaleffectbetween PFOS and bilirubin levels. Our datado not suggesta reduction in totalserum cholesterowlith PFOS at the serum levelsmeasured among theseproductionemployees.PFOS isa peroxisomeprolifemtoirn therat.and hypolipidemiaha.,b;een consistently observedinsubchronicritand primate toxicologystudies.-8"-1-('Rhesus monkeys thathad been administeredPFOS at 4.5 mg/kgtday had mean serum cholesterovlaluesthatdecreasedfrom 183 mg/dL to 99 mg/dL within30 days.5Rhesus monkeys administered 1.5 mgtkgtday had mean cholesterol levelsthatdecieasedfrom 195 mg/dL to III mg/dL within90 days.A noobservable-effelcetvelwas seen for the 0.5 mg/kg dose group at 90 days. Serum PFOS measurements were not determined in theserhesus monkeys. However, datafrom a recentcynomolgus monkey dose-range-findinsgtudy suggestedthathypolipidemiamay be initiallayssociatedwith serum PFOS levelsin therangeof 100 to 200 ppM.7 Haughom and Spydevold suggested thatthe hypolipidemiceffectof PFOS may be due to reducedliveractivitoyf hydroxymethyl glutariaccid-CoA reductase and acyl-CoA cholesterol acyltransferaswei,th enhanced fatty acidoxidationin the liver.N'abbefeld et al observedthatPFOS has a high affinitfyorthe fattyacidcarrierproteinsalbumin and i.-fatatcyid-binding JOEM Volume41,Number 9,September1999 TABLE5 Mean,Median(Med)S,tandarDdeviatio(nSD)ofMeanandRangeofPFOS,andTotalandHDL Cholesterboyl,Yearand PlantLocation Antwerp Decatur PFOS* (ppm) 1995 Data Cholesterolm,g/dL 0 to <1 1 to <3 3 to <6 a6 HDL, mg/dL 0 to<1 1 to <3 3 to <6 a6 1997 Data Cholesterolm,g/dL 0 to <1 1 to <3 3 to <6 a6 HDL, mg/dL 0 to<1 1 to<3 3 to<6 2--6 Mean Mod SD 220 206 217 223 F value 219 50 211 49 215 30 221 16 0.6,P - 0.61 56 53 50 53 F value 57 13 51 13 49 11 49 7 1.1.P = 0.35 193 213 228 - F value 190 41 205 48 223 38 - 2.9,P = 0.07 51 48 51 - F value 50 12 46 10 so 10 - 0.8.P = 0.47 Range Mean Mod SD 100 to340 118 to315 178 to266 208 to240 31 to94 33 to79 31 to74 48 to61 110 to277 116 to365 192 to321 19 to74 34 to68 39 to69 215 221 209 206 F value 208 39 218 39 213 42 206 47 O.S.P = 0.69 43 45 39 39 F value 41 9 44 12 39 9 39 5 1.4,P = 0.25 204 218 230 229 F value 208 38 226 39 230 23 238 26 2.0.P = 0.13 42 41 9 42 41 10 45 45 10 40 38 4 F value= 0.5,P = 0.67 Range lr>4to276 132 to300 128 to278 160 to 251 31 to59 26 to94 23 to51 34 to46 145 to277 152 to290 197 to280 186 to250 26 to59 28 to69 32 to62 37 to45 See Table 1 forsample sizesby year. proteinin vivo.Thiscouldpotentially This was due,inpart,toa largeturn- alterfattyacidtransportb,iochemistry, over of employees at both plantsbe- and metabolism,which conceivably tween examinations.Sixth,therecould couldthen leadto a decreasein cho- be measurement errorin important lesteroelsterificatiaonnd metabolic confoundingvariablesA.nalysisof the wasting.11 data of the 61 subjectswho pardci- Several methodological issues patedin both yearsshowed thatthere shouldbe consideredin evaluatingthe was excellentcorrelatiofnor the con- resultfstom our study.Fint@thecross- foundingfactorsof BNE (r = .92,P = sectionadlesigndoes not allow for a 0.0001),self-reporteadspectsOf alco- directanalysisof the temporalityof an hol consumption (r = .88, P = associationS.econd, thevoluntarypar@ 0.0001),and cigarettesmoking (r = ticipatiornatesin the fluorochen-dcal .79,P = 0.0001).As expected,these medical surveillancweere not ideal 61 employees'serum PFOS levelsfor amon-A-t-heeligiblfeluorochemicaplro- the2 yearswere highlycorrelate(dr= ductionemployees.Third.the serum .92,P = 0.0001).Theirresultdsidnot levelsof PFOS thatwere measured differfrom thoseof the entirestudy may be below the no-effecltevelin population.Finally,the qualityof laboratoraynimals.Fourth.PFOS con- medicalsurveillancdeata,priorto its centratepsrimarilyintheliverof lab- use forstudyingan a priorihypothesis, oratoryanimals.Serum measurements can oftenbe evaluatedby whether of PFOS may not adequatelyreflect known positiveassociationasre ob- body burden.Fifth,the two cross- served.In thisregard,we did observe sectionaalnalysescannotbe viewed as variousexpectedassociationss,uch as independent populations.Sixty-one cigarettsemokinc,and elevatedwhite employees were studiedin both years. blood cellcounts and largeBMIS, associatedwith elevatedliverawl%aminase levels.24*25 In summary, our findingssuggest thatamong theseAntwerp and Decatur male fluorochemicalproduction employees, therewere no substantial changes in serum hepaticenzymes, cholesterolo,r lipoproteinsassociated with PFOS levelsless than 6 ppm. It is not possibleto derive inferencesfrom the few employees with serum levels a:6 ppm. Limitationsof thisstudy includeitscrosssectionaldesign,the low voluntary participatiornatesamong the employees,and the lower levelsof serum PFOS measured among these employees.compared with thosesuspectedto cause effectsin laboratory animals. ACknowledgment The authorsgratefullaycknowledgethe assistancoef MicheleBuriew. 806 PFOS,HepaticEnzymes,and CholesteroOllsonatal References I. (jibst)iSiI. J(ihii%t)JnD. Ober RE. AbN- /.iiiiF-OI.VE iiiiil'rissiIo)ti,sti-il)tilciiot"ill h*lii)iipttioifitC)(titt-hiii1i4-A.lic-Arthiiiiiivir(itiottitfN-EthvlFO.SF.-"'CIn Feeti.St. Paul.MN: Rike'r Ljtx)ratorieIsnc.1983. 2. John%on JD. Gibson SJ. Ober RE. Extent and Route of Excretionand TissueDistribiitito)fn Total Carbon-14 in Rats Aftera Single iv. Dose of FC-95-1"C ProjectNo. 8900310200. St. Paul,MN: Riker LaboratoriesInc;1979. 3. Johnson JD. Gibson SJ. Ober RE. Cholestyramine-enhancefdecaleliminadon of carbon-14in mts afteradministratioonf ammonium ["Cl peffluorooctanoatoer potassium (14CI perfluorooctanesulfonate. Fund Appi ToricoL 1984.4:972-976. 4. GoidenthalEl.JessupDC, Geil RG, JeffersonND, Arceo RJ. Ninety-Day SubacuteRat ToxicirSvtue(vS.tu4vNo. 137095. Mattawan, MI: InternationaRlesearch Development Corp. November 1978. 5. GoidenthalEl,JessupDC, Geil RG. JeffersonND. Ninety-DaySubacuteRhesus Monkey Toxicity Study. Stu4i, No 137085. Mattawan, MI: InternationaRlesearch Development Corp; December 1978. 6. GoidenthalEl.JessupDC, Geil RG, JeffersonND. Ninety-Dav Subacute Rhesus Monkey ToxicityStud'y.Study No. 137087. Mattawan, MI: InternationaRlesearchDevelopment Corp-.January 1979. 7. lbomford PJ. Four-Week Capsuk Toxicity Snidy With PerfliiorooctaneSu#'onk Acid PotassiumSalt(PFOS) in Cynonwlogus Monkeys.CovanceStudy6329-222.Madison.W'I:Covance Co; August 1998. 8. IkedaT, Fukuda Y, Mori 1.Enomoto M, KornaiT, Sup T. Inductionof cytochrome P450 and peroxisomeproliferatiionnmt liverby perfluorinateodctanesulfoniaccid. In:FahiraiHD, SiesH. eds.Peroxisomesin Biology and Medicine. New York: SpringerVeriag: 1997:304-308. 9.H;itigh4)Bi.iSipydevi)l0d. The niecha- 16. GreenlandS. M(wJclingand variablese- iiistiujiiticrlyintghe hyrx)lipemiceffect leciit)nin epidemic)logicanalysis.Am J 4)1p'erIlLitirt)ocianoaiccid (PFOA). per- Piil)iiH(i-,(tlfh1.989-.79:340-349. nuor4)ctiisiteilpht)nziiccidIPFOSA) and 17. SAS Institute.Inc. .5.,ISUver.T Gititle: citiribricacid. Bi(it-heitBtioltli.vsAt-t(i. Vervioti 6. Cary. NC: SAS 1992; 1128:65-72. InstitutIen.c;19tX). 10. SohleniusAK. ErikssonAM. Hogstrom 18. Davem TJ. Scham-hmidt BF. Biochemi- C. Kimiand M. EX-PierrJeW. Perfluo- callivertestsI.n:Feldman M. Sleisengcr rooctanesulfonicacidisa potentinducer MH, Scharschmidt BF, eds.Gastrointes- of peroxisomal fattyacid beta-oxidation tinaland LiverDisease,6th ed.Philadel- and otheractivitieksnown to be affected phia, PA: W.B. Saunders Co.-.1993: by peroxisome proliferatorisn mouse 1112-1122. liverP.harmacol Toxicot 1993;72:90-93. 19. Lewis JA. IllustratedGuide to Diagnos- I].Nabbefeld J,Butenhoff,J.Bass N, Seacat ticTests.Springhouse. PA: Springhouse A. Displacementof a fluorescentllya- Corp; 1994:100-107,194-207. beled fattyacidanalogue from fattyacid 20. Friedman LS. Liver,biliarytractand carrierproteinsby Wyeth- 14.643,ammo- pancreas.In:Tierney LM. McPhee SJ. nium perfluorooctanoa,tpeo,tassiumper- Papadakis WA. eds. Medical Diagnosis fluorooctanesulfonateand otherknown and Treatment,37th ed. Stamford, CT: peroxisome prolifemtor(sabstmct)T.ox- Appleton and Lange; 1998:628-630. icologist1.998;42:395. 21. Fu PC. Lipids.lipoproteinsa,nd apoli- 12. Bryce H. Industriaalnd utilitariaasnpects poproteins.In:Howanitz IH, Howanitz of fluorinechemistry.In:Simons J.ecL PJ.eds.LaboratoryMedicine:TestSelec- Fluorine Chemistrv. New York: Aca- tion and Interpretation.New York: demic Press;1964:297-492. ChurchillLivingstoneInc; 1991:173- 13. Johnson JD. Wolter JT. ColaizyGF. Rethwill PA., Nelson RM. Quantificationof 198. 22. Wolf PL. Liver function.In:Howanitz Perfluroocianoaatned Perfluoroocranesul- JH, Howanitz PJ,eds.LaboratoryMedi- fonatein Human Serwn Using Ion-Pair cine.TestSelectionand Interpretation. Extractioannd High Perfomowe Liquid -New York: ChurchillLivingstoneInc-, Chromatography-Thermosprov Mass 1991:67-84. Spectrometry With Automated Sample 23. Lidofsky S, ScharschmidtBF. Jaundice. PreparatioiLSL Paul.MN: 3M Environ- In:Feldman M. SleisengerMH. Schar- mental Laboratory-1.996. schmidt BF. eds. Gastrointestinaland 14. Anderson DJ, Mulvana DE. Analytical Liver Disease, 6th ed. Philadelphia: W.B. Report for the Determination of Perfluo- Saunders Co; 1998:220-224. rooctanoate and Perfiuoroocianesm4fon- 24. Olsen GW, Kusch GD, StaffordBA. atein Human Serum by LCIMS. Ithaca, Oudmundsen SL. CurrierMF. The posi- NY: Advanced BioanalyticalServices tiveknown associatiodnesign:a quality Inc;August 1997. assurancemethod foroccupationalhealth 15. Anderson DJ, Mulvana DE. Analytical surveillancdeata.J Occup Med. 199 1;33: Reportfor theDeterminationofPerfluo- 998-1000. roocianoateand Perfluoroocianesulfon- 25. Bums CJ. Boswell JM. Olsen GW. Liver ate in Human Serum by LCIMS. Ithaca, enzyme activityand body mass index. NY: Advanced BioanalyticaSlervices J Occup Environ Med. 1997;38:1248- Inc-.September 1997. 1252.