260

PFOS and PFOA Analysis

•

Olsen et al

Epidemiologic Assessment of Worker Serum
Perfluorooctanesulfonate (PFOS) and
Perfluorooctanoate (PFOA) Concentrations
and Medical Surveillance Examinations
Geary W. Olsen, DVM, PhD
Jean M. Burris, MPH
Michele M. Burlew, MS
Jeffrey H. Mandel, MD

P

Perfluorooctanesulfonyl fluoride (POSF, C8F17SO2F) is used to create
applications for surfactants and paper, packaging, and surface (eg,
carpets, textiles) protectants. Such POSF-based products or their residuals may degrade or metabolize to PFOS (C8F17SO3!). PFOS concentrates in liver and serum and results in hypolipidemia as an early effect
of cumulative dosages. Male and female employees of two perfluorooctanyl-manufacturing locations (Antwerp, Belgium and Decatur, Alabama) participated in a periodic medical surveillance program that
included hematology, clinical chemistry, thyroid hormone, and urinalysis testing. Serum concentrations of PFOS and perfluorooctanoate
(PFOA, C7F15CO2!, used as a fluoropolymer emulsifier) were measured
via mass spectrometry methods. The mean serum PFOS and PFOA
concentrations for 263 Decatur employees were 1.32 parts per million
(ppm; geometric mean 0.91, range 0.06 –10.06 ppm) and 1.78 ppm
(geometric mean 1.13, range 0.04 –12.70 ppm), respectively. Mean
concentrations were approximately 50% lower among 255 Antwerp
workers. Adjusting for potential confounding factors, there were no
substantial changes in hematological, lipid, hepatic, thyroid, or urinary
parameters consistent with the known toxicological effects of PFOS or
PFOA in cross-sectional or longitudinal analyses of the workers’
measured serum fluorochemical concentrations. (J Occup Environ
Med. 2003;45:260 –270)

From the Medical Department, 3M Company, St. Paul, Minnesota (Dr Olsen, Ms Burris, Ms Burlew,
Dr Mandel).
Address correspondence to: Dr Geary Olsen, Medical Dept., 3M Company, Mail Stop 220-3W-05,
St. Paul, MN 55144; e-mail: gwolsen@mmm.com.
Copyright © by American College of Occupational and Environmental Medicine
DOI: 10.1097/01.jom.0000052958.59271.10

erfluorooctanesulfonyl fluoride
(POSF, C8F17SO2F), which is produced by an electrochemical fluorination process, is used as the basic
building block to create unique
chemistries through the sulfonyl fluoride moiety using conventional hydrocarbon reactions. Applications include surfactants and paper,
packaging, and surface (eg, carpet,
upholstery, textile) protectants. Depending upon the specific functional
derivatization or the degree of polymerization, such POSF-based products or their residuals may degrade
or metabolize to an undetermined
degree to PFOS (C8F17SO3!). PFOS
is a stable and persistent end-product
that has the potential to bioaccumulate. Although not a major commercial product, PFOS has been used in
some products, including firefighting
foams. Another fluorochemical, the
ammonium salt of perfluorooctanoate (PFOA, C7F15CO2!) is produced
to be an emulsifier in the polymerization of fluoropolymers. In May
2000, the 3M Company announced
that it would voluntarily cease manufacturing perfluorooctanyl-related
materials after PFOS was found to be
pervasive and persistent in human
populations and wildlife.1–5 Nonoccupational exposures to PFOS or
precursors are not well understood at
this time but could include environmental sources, consumer products,
or as indirect food additives.
PFOS concentrates primarily in
the liver and, to a lesser extent, in the

 JOEM

•

Volume 45, Number 3, March 2003

plasma, of rats.6 There appears to be
significant enterohepatic circulation
of PFOS with both urinary and fecal
excretion.7 Subchronic studies in rats
and primates suggest that the toxicity
of PFOS is dependent upon cumulative dose.8 –12 Decreased serum cholesterol was the earliest clinical
chemistry response observed in primates and occurred at serum PFOS
(potassium salt) concentrations
above 100 parts per million (ppm).12
Rats fed PFOS (potassium salt) at 20
ppm in the diet over a 2-year time
period experienced a marginal increase in liver tumors, primarily adenomas.13 Although the mechanism
of toxicity in laboratory animals remains to be fully elucidated, PFOS
toxicity may involve: 1) the inhibition of HMG CoA reductase and acyl
CoA cholesterol acyl transferase activity;14 2) activation of the PPAR"
receptor;15,16 3) disturbances in fatty
acid transport and metabolism;17
and/or 4) altered membrane function
changes and mitochondrial bioenergetics.18,19
There are substantial sex and
species differences in elimination
of PFOA with the longest serum
elimination half-life reported for
humans.20 –23 Besides urinary excretion, biliary excretion and reabsorption of PFOA occurs.7 The
liver is the primary target organ for
PFOA-induced toxicity. PFOA produced hypolipidemia in rodents14,24 but not in a 6-month
capsule feeding study of cynomolgus monkeys. 22 Increased liver
weight as a result of mitochondrial
proliferation occurred in all three
monkey dose groups (3, 10, and 20
mg/kg/day), although histopathologic evidence of liver injury occurred only in the highest dose
group. Two-year lifetime bioassays
of rats fed 300 ppm PFOA in the
diet resulted in the increased incidence of tumors (adenomas) of the
liver, pancreas (acinar cell), and
testes (Leydig cell).25 The hepatocellular tumors may result from a
combination of oxygen stress and
cell proliferation that accompanies

an increase in peroxisomes. The
tumors observed in the testis may
be the consequence of sustained
increases in estradiol as a result of
aromatase induction25–27 whereas
those in the pancreas may involve
the release of cholecystokinin as a
result of cholestasis.28 It remains to
be determined whether these possible mechanisms are relevant to humans.
Serum measurements of PFOS
and PFOA have been obtained as
part of the 3M Company’s effort to
assure worker safety as related, in
particular, to lipid and hepatic parameters because of the toxicological evidence that indicates the
liver is the target organ of effect in
rats and primates. The company’s
fluorochemical medical surveillance program is offered employees
on a routine periodic basis at three
3M manufacturing facilities: Antwerp (Belgium), Cottage Grove
(Minnesota), and Decatur (Alabama). Electrochemical fluorination at the Cottage Grove site involved the production of PFOA and
its associated salts but not
POSF.29 –31 The present study focused on the company’s Antwerp
and Decatur facilities, where POSF
and PFOA have been produced.
Although a cross-sectional assessment of the 1995 and 1997 Antwerp
and Decatur medical surveillance
program data was previously reported, the data were limited for
several reasons: low voluntary employee participation, no inclusion of
female employees, inherent limitations of the cross-sectional design,
and analysis of only serum PFOS
concentrations. 32 Increased employee participation in the year 2000
medical surveillance program allowed for a much larger crosssectional analysis of both male and
female employees and also enabled a
6-year longitudinal assessment of
clinical chemistries in relation to
PFOS, PFOA, and a calculated total
organic fluorine value.

261

Materials and Methods
Manufacturing Sites
The manufacturing operations of
the two sites are similar. Fluorochemical production occurs in several buildings. In one building,
where the base product is manufactured, POSF, occurs via electrochemical fluorination. At another
building, the POSF starting material
is reacted to form fluorochemical
amines and then further to other fluorochemicals, including alcohols and
acrylates with subsequent polymerization. Synthetic fluoroelastomers
are also produced at both facilities.
PFOA has been routinely produced
at Antwerp but only since 1999 in
Decatur. However, PFOA can be a
byproduct of POSF-related manufacturing at both facilities.
The fluorochemical medical surveillance program is available on a
periodic voluntary basis to all Antwerp and Decatur chemical plant
employees and those employees with
site-wide responsibilities (eg, environmental, health and safety workers). In 2000, approximately 340
Antwerp and 500 Decatur chemical
plant and site employees were eligible to participate. The surveillance
program measured several serum
fluorochemicals, including PFOS
and PFOA as well as hematology,
clinical chemistries, and thyroid hormones. Urinalyses were conducted
only in Decatur.

Hematology, Clinical Chemistry,
Thyroid Function, and Urinalysis
Upon collection and shipment of
specimens, Allina Laboratory Services (St. Paul, MN) performed standard hematological and clinical
chemistry tests for both manufacturing sites. These included the following hematological tests: hematocrit
(percent), hemoglobin (gm/dL), red
blood cells (RBC, 1000/mm3), white
blood cells (WBC, 1000/mm3) and
platelet count (1000/mm3); and the
following clinical chemistry tests: alkaline phosphatase (IU/L), gamma
glutamyl transferase (GGT, IU/L),

 262

aspartate aminotransferase (AST, IU/
L), alanine aminotransferase (ALT,
IU/L), total and direct bilirubin (mg/
dL), cholesterol (mg/dL), highdensity lipoprotein (HDL, mg/dL),
triglycerides (mg/dL), blood glucose
(mg/dL), blood urea nitrogen (BUN,
mg/dL), and serum creatinine (mg/
dL). Reference ranges have been relatively constant since 1994/95, although for ALT the range declined
from 20 – 65 IU/L in 1994/95 to
1– 40 IU/L in 1997 and 2000. Six
thyroid tests were conducted by LabCorp (Kansas City, MO): thyroidstimulating hormone (TSH; #IU/
mL); serum thyroxine (T4; #g/dL);
free thyroxine (free T4; ng/dL); serum triiodothyronine (T3; ng/dL);
thyroid hormone binding ratio
(THBR, previously referred to as T3
Uptake); and free thyroxine index
(FTI). TSH, free T4, and T3 were
determined by an immunochemiluminometric assay. T4 and THBR
were determined by a cloned enzyme
donor immunoassay. FTI was calculated by multiplying T4 and THBR.
Urinalyses were only performed on
Decatur employees via the standard
urine microstick analysis, which
tested for urine glucose, albumin and
red blood cells.

Fluorochemical Analyses
In the 2000 fluorochemical medical surveillance program, the employees’ sera samples were extracted
and quantitatively analyzed for
PFOS and PFOA using high-pressure liquid chromatography electrospray tandem mass spectrometry and
evaluated versus an extracted curve
from a human plasma matrix. All
serum fluorochemical analyses were
determined by Northwest Bioanalytical Laboratory Inc. (Salt Lake City,
UT). Sera samples were extracted
using an ion-pairing extraction procedure.2 Evaluation of quality control samples injected during each analytical run indicated that the
reported quantitative results may
have varied, on average, up to 20%
although most analyses were within
$10%. For all employees, serum

PFOS and PFOA Analysis

values for PFOS and PFOA values
were above the lower limit of quantitation. Results are reported in ppm.
Five other fluorochemical analytes
were also analyzed in the 2000
medical surveillance program: perfluorohexanesulfonate (C6F13SO3!);
N-ethyl perfluorooctanesulfonamidoacetate (C8F17SO2N(CH2CH3)CH2
COO!); N-methyl perfluorooctanesulfonamidoacetate (C8F17SO2N(CH3)
CH2COO!); perfluorooctanesulfonamidoacetate (C 8 F 17 SO 2 NHCH 2
COO!); and perfluorooctanesulfonamide (C8F17SO2NH2). These fluorochemicals were measured at concentrations 1 to 3 orders of
magnitude lower than PFOS and
PFOA and are therefore not presented. A total organic fluorine
(TOF) value was determined by
calculating the percent of each of
the seven specific fluorochemical’s
molecular weight that was attributed to organic fluorine (eg, PFOS,
64.7%; PFOA, 69.0%) multiplied
by the ppm measured for each fluorochemical and then summed
across all seven fluorochemicals.
Serum fluorochemical analyses
were conducted at different laboratories in the previous surveillance
years (1994 [Decatur], 1995 [Antwerp], and 1997 [both facilities])
using slight variations of the Hansen
et al. method.2 In those years, only
PFOS and PFOA were quantified.
Therefore, TOF for the longitudinal
assessment was based only on PFOS
and PFOA.

Data Analyses
For the cross-sectional analyses,
associations between PFOS, PFOA,
or TOF and each hematological and
clinical chemistry test and thyroid
hormone assay were evaluated with
descriptive statistics, analysis of
variance, and multivariable regression. For stratified analyses, employees were divided into quartiles of
their serum PFOS distribution. Age,
body mass index, current alcohol
consumption (drinks per day) and
cigarette use (cigarettes smoked per
day), years worked at Antwerp or

•

Olsen et al

Decatur, and type of job (production
versus nonproduction) were potential
confounding factors that were considered in the analyses. Production
jobs included cell operators, chemical operators, mill operators, and
crew supervisors. Nonproduction
jobs included engineers, qualityassurance laboratory and research
workers, and administration (eg,
managers, clerical staff).
Logistic regression was used to
calculate adjusted odds ratios for the
quartile distribution of employees
who had liver function tests above
the laboratory’s reference ranges. In
addition, the individual continuous
dependent variables (lipids, hepatic
enzymes or thyroid hormones) were
fitted in multivariable regression
analyses with PFOS, PFOA, or TOF
analyzed as an independent continuous variable(s). The statistical significance of the fluorochemical coefficient was considered at P % 0.05.
Natural log transformations of the
dependent variables were performed
when necessary to normalize variables and to enhance model fit. Study
results were analyzed using the SAS
System (SAS Institute, Cary,
NC).33,34
For the longitudinal assessment,
lipid and hepatic clinical chemistry
tests were evaluated as repeated
measures incorporating the random
subject effect fitted to a mixed model
by the MIXED procedure in the SAS
statistical package.35 Restricted maximum likelihood estimates of variance parameters were computed. Adjusted regression models were built
by introducing all covariates (see
below) and testing the covariance
structure. Equal spacing was assumed given there were approximately 3 years between each medical
surveillance examinations. Based on
goodness-of-fit tests, the autoregressive was routinely considered the
best covariance structure for the
mixed models. Variables included
PFOS (or PFOA or TOF), years of
observation (ie, follow-up), the interaction term of PFOS and years of
observation, age, body mass index

 JOEM

•

Volume 45, Number 3, March 2003

263

Fig. 1. Distribution of Antwerp and Decatur employees by their serum PFOS and PFOA concentrations.

(BMI), cigarettes smoked per day,
alcohol drinks per day, location, year
at first entry, and baseline years
worked. Serum triglycerides was
also considered a potential confounding factor for all hepatic clinical chemistry analyses (cross sectional and longitudinal).

Results
Cross-sectional Analysis
Altogether, there were 255 (75%)
Antwerp employees (206 male and

49 female) and 263 (52%) Decatur
employees (215 male, 48 female)
who voluntarily participated in the
2000 fluorochemical medical surveillance program. Seventy-three
percent of the Antwerp male employee participants and 75% of the
Decatur employee participants
worked in production activities. Only
12% of the Antwerp female employees worked in production activities
compared with 63% of the Decatur
female employees.
Presented in Fig. 1 are the distri-

butions of serum PFOS and PFOA
concentrations among the Antwerp
and Decatur employee participants.
The arithmetic mean serum PFOS
concentration among all Antwerp
subjects was 0.80 ppm (range 0.04 –
6.24 ppm) with a geometric mean of
0.44 ppm (95% CI & 0.38 – 0.51).
For PFOA, the arithmetic mean was
0.84 ppm (range 0.01–7.04 ppm)
with a geometric mean of 0.33 ppm
(95% CI & 0.27– 0.40). The arithmetic mean serum PFOS concentration among all Decatur participants

 264

PFOS and PFOA Analysis

•

Olsen et al

TABLE 1
Mean and Standard Deviation Comparisons Between Antwerp and Decatur Employees’ Serum PFOS, PFOA, TOF,
Demographic and Clinical Chemistry Results, by Gender
Males
Antwerp (N ! 206)

PFOS
PFOA
TOF
Age
BMI
Years worked
Alcoholic drinks/day
Cholesterol
HDL
Triglycerides
Alk Phos
GGT
AST
ALT
Total Bilirubin

Females
Decatur (N ! 215)

Antwerp (N ! 49)

Decatur (N ! 48)

Mean

SD

Mean

0.13d
0.07d
0.17d
36
22.8d
12 a
0.5d
208
68 a
94 d
46 a
12 d
18
13 d
0.8b

0.10
0.17
0.20
7
3.2
7
0.4
36
16
45
13
7
6
6
0.3

0.93
1.23
1.76
42
27.7
13
0.1
200
59
133
65
18
20
19
0.6

Mean

SD

Mean

0.96d
1.03d
1.60d
37 d
24.8d
13 c
1.1d
218
55 d
124 d
60 d
23 d
23 c
23 d
1.0d

0.97
1.09
1.34
9
3.0
8
1.1
41
15
87
15
17
6
10
0.3

1.40
1.90
2.65
43
28.8
16
0.1
215
44
191
74
31
26
35
0.7

SD
1.15
1.59
2.00
9
4.4
13
0.3
42
10
124
20
18
8
16
0.2

SD
0.81
1.18
1.50
9
5.9
10
0.1
39
12
152
18
15
7
10
0.2

a

p % .05 compared to Decatur (student t test).
p % .01 compared to Decatur (student t test).
c
p % .001 compared to Decatur (student t test).
d
p % .0001 compared to Decatur (student t test).
b

was 1.32 ppm (range 0.06 to 10.06
ppm) with a geometric mean of 0.91
ppm (95% CI & 0.82–1.02). For
PFOA, the arithmetic mean was 1.78
ppm (range 0.04 –12.70 ppm) with a
geometric mean of 1.13 ppm (95%
CI & 0.99 –1.30).
Antwerp male employees compared with their Decatur counterparts
had lower mean serum PFOS and
PFOA concentrations; were younger;
had lower BMIs; drank more alcoholic beverages per day; and had
higher mean HDL and total bilirubin
and lower triglyceride and hepatic
clinical chemistry values (Table 1).
Similar findings were observed between Antwerp and Decatur female
employees (Table 1).
Presented in Table 2 are the mean,
standard deviation, and range of demographic, clinical chemistry, and
thyroid hormone results stratified by
the serum PFOS quartile distribution
for the combined 421 Antwerp and
Decatur production and nonproduction male employees. (Note: The median values [data not shown in Table
2] by PFOS quartile, for PFOS,
PFOA, and TOF were, respectively:

quartile 1: [0.29, 0.25, 0.43 ppm];
quartile 2 [0.59, 0.86, 1.14 ppm];
quartile 3 [1.17, 1.20,1.88 ppm]; and
quartile 4 [2.46, 2.43, 4.06 ppm].) As
noted in the footnote to Table 2, the
demographic and clinical values reflect the higher percentage of Antwerp
employees in the lowest PFOS quartile
and the higher percentage of Decatur
employees in the upper PFOS quartiles. The fourth quartile had statistically significant higher mean values
than the first quartile for triglycerides,
alkaline phosphatase, ALT and T3.
The fourth quartile had significantly
lower mean values for drinks per day
and total bilirubin compared to the first
quartile. There were no significant
mean differences between quartiles for
cigarettes smoked, hematology, blood
glucose, BUN, serum creatinine or urinalyses (data not shown).
Similar to Table 2, presented in
Table 3 are the serum PFOS quartile
distributions for Antwerp and Decatur production and nonproduction female employees. Again, the Antwerp
employees predominated in the lowest serum PFOS quartile and Decatur
female employees predominated in

the highest quartile. [Note: The median values [data not shown], by
quartile, for PFOS, PFOA and TOF
were, respectively: quartile 1: [0.08,
0.02, 0.09 ppm]; quartile 2 [0.13,
0.05, 0.14 ppm]; quartile 3 [0.37,
0.36, 0.59 ppm]; and quartile 4 [1.34,
1.39, 2.66 ppm].) The fourth quartile
had significantly higher mean values
than the first quartile for age, BMI,
alkaline phosphatase, and GGT. The
fourth quartile had significantly
lower mean values for alcoholic drinks
per day and total bilirubin. There were
no significant mean differences for
cigarettes smoked, hematology, blood
glucose, BUN, serum creatinine, or
urinalyses (data not shown).
Summarized in Table 4 are the
number of Antwerp and Decatur employees (and percentages) who had
hepatic enzyme tests above reference
range values, stratified by the quartiles of the serum PFOS distribution.
Among male employees, 12% of the
employees in the fourth quartile had
above reference range values for
ALT and GGT compared with 4 to
8% in the first through third quartiles. For the total liver panel, 23% of

 JOEM

•

Volume 45, Number 3, March 2003

265

TABLE 2
Antwerp and Decatur Production and Non-Production* (N & 421) Male Employees’ PFOS, PFOA, TOF, Demographic,
Clinical Chemistry and Thyroid Hormone Results by Quartile of Serum PFOS Distribution*
Quartile 1 (N ! 105)
Mean
PFOS
PFOA
TOF
Age
BMI
Years worked
Drinks/day
Cholesterol
HDL
Triglycerides
Alk Phos
GGT
AST
ALT
Total Bilirubin
TSH
T4
Free T4
T3

0.272,3,4
0.542,3,4
0.622,3,4
38 3
25.8
12 3
0.9 3,4
214
54
131 4
61 3,4
24
25
26 4
1.03,4
2.0
8.3
1.1
124 4

SD

Range

Quartile 2 (N ! 105)
Mean

SD

0.11 0.04 – 0.42 0.601,3,4 0.12
0.77 0.01– 4.03 1.211,4
1.19
0.58 0.05–3.03 1.401,3,4 0.89
10
23– 60 41
10
4.0
19.2– 40.8 26.9
4.0
10
1–38 15
12
1.0
0 –5
0.6
0.9
41
140 –331 214
43
15
31–121 47
11
95
32–527 155
102
16
26 –98 67
18
16
7–111 29
22
8
13–58 25
6
13
10 –91 28
11
0.3
0.5–2.0
0.9
0.3
1.2
0.03–5.7
3.1
6.6
1.4
0.5–11.5 8.2
1.4
0.2
0.9 –1.5
1.1
0.1
17
94 –164 128
20

Range

Quartile 3 (N ! 106)
Mean

SD

Range

0.43– 0.81
1.191,2,4 0.24
0.06 –7.04
1.451,4
1.10
0.38 –5.69
2.121,2,4 0.87
21– 63
42 1
9
19.0 –37.3 27.3
4.5
2–38
16 1
11
0–4
0.51
0.9
121–308 215
39
29 – 80
48
13
35– 633 169
123
30 –142 69 1
21
7–144 26
15
16 – 49
24
7
10 – 63
28
14
0.3–2.0
0.81
0.3
0.5– 65.3
2.1
2.0
4.2–12.0
8.3
1.5
0.6 –1.4
1.1
0.2
86 –186 127
21

Quartile 4 (N ! 105)
Mean

SD

Range

0.82–1.68
2.691,2,3 1.09 1.69 –10.06
0.12–7.48
2.701,2,3 1.63 0.25–12.70
0.98 – 6.61
4.411,2,3 1.72 1.92–12.23
22– 61
40
9
27– 60
17.2–50.1 27.2
4.5 17.8 – 45.5
1–38
15
10
2–38
0–6
0.51
0.9
0 –5
105–303 222
44
122–384
24 –100
48
15
26 –119
32–731 177 1
123
39 –796
30 –160
70 1
19
21–126
6 – 89
30
17
7– 85
7–51
25
9
13– 69
6 –103
33 1
19
8 –99
0.4 –2.0
0.81
0.3
0.4 –2.2
0.2–18.8
2.5
2.8
0.5–21.5
3.3–12.9
8.4
1.4
4.7–11.4
0.4 –1.6
1.1
0.2
0.8 –1.6
91–196 132 1
22
87–190

* Number of male employees by location, production (P) and non-production (NP) category and quartile (percent in parenthesis).
Q1

Antwerp
Decatur
Total
1

Q2

Q3

Q4

P

NP

P

NP

P

NP

P

NP

38
7
45 (43)

38
22
60 (57)

38
40
78 (74)

12
15
27 (26)

38
51
89 (84)

4
13
27 (16)

36
63
99 (94)

2
4
6 (6)

Mean is significantly different (P % .05, Bonferroni (Dunn) t test) from the mean of the 1st quartile; 2 2nd quartile; 3 3rd quartile; 4 4th quartile.

the male employees had one or more
liver clinical chemistry tests above
the reference range value compared
with 14 to 16% of the male employees in the lower three quartiles. Odds
ratios were calculated for each quartile for those employees above or
below reference ranges as listed in
Table 4 (quartile one reference odds
ratio & 1.0). The odds ratios were
adjusted for the potential confounding effects of age, BMI, alcohol,
cigarettes, and location. For ALT,
the odds ratios for the second, third,
and fourth quartiles were (95% CI in
parentheses) 0.6 (95% CI & 0.1–
2.8), 1.2 (0.3– 4.8), and 2.1 (0.6 –
7.3), respectively. For GGT, the odds
ratios were 1.3 (0.4 – 4.1), 0.9 (0.3–
3.1), and 2.0 (0.7–5.8), respectively.
For the total liver panel, the odds
ratios were: 1.1 (0.5 to 2.3), 1.1
(0.4 –2.3), and 1.6 (0.7–3.3), respectively. None of the odds ratios were

statistically significant (P % 0.05).
Logistic models did not meet satisfactory convergence criteria because
of the few male subjects with values
above the reference ranges for alkaline phosphatase or AST or for female employees for any liver function analysis in Table 4.
Using multivariable regression analysis and adjusting for potential confounders, serum PFOS was positively
associated with the natural log of serum cholesterol (PFOS coefficient &
0.020, P value & 0.04) and triglycerides (PFOS coefficient & 0.066, P
value & 0.01), although these associations contributed minimally (partial R2
% 0.01 and 0.03, respectively) to the
variation explained in the models (cholesterol model adjusted R2 & 0.06;
triglyceride model adjusted R2 &
0.27). PFOA and TOF were also positively associated with cholesterol and
triglycerides with similar variations

explained. These modest positive associations for PFOS or PFOA, however,
were opposite the expected direction
based on the known toxicity of these
compounds. HDL was not significantly associated with PFOS or PFOA.
Adjusting for their potential confounders, the hepatic enzyme and bilirubin
analyses were not significantly associated with PFOS or PFOA. Multivariable regression analyses of the thyroid
hormones resulted in no significant
associations with PFOS or PFOA except for a positive association with the
natural log of T3 (PFOS coefficient &
0.015, P value & 0.04; PFOA coefficient & 0.016, P value & 0.01), which,
again, contributed minimally to the
variation explained in the model (partial R2 & 0.01).

Longitudinal Analysis
A total of 174 Antwerp and Decatur male employees participated in

 266

PFOS and PFOA Analysis

•

Olsen et al

TABLE 3
Antwerp and Decatur Production and Non-Production* (n & 97) Female Employees’ PFOS, PFOA, TOF, Demographic,
Clinical Chemistry and Thyroid Hormone Results by Quartile of Serum PFOS Distribution*
Quartile 1 (n ! 24)
Mean
PFOS
PFOA
TOF
Age
BMI
Years Worked
Drinks/day
Cholesterol
HDL
Triglycerides
Alk Phos
GGT
AST
ALT
Total Bilirubin
TSH
T4
Free T4
T3

0.073,4
0.043,4
0.093,4
34 4
22.84
11
0.44
207
66
93
50 4
11 4
19
13
0.83,4
2.2
10.2
1.1
145

SD

Range

Quartile 2 (n ! 24)
Mean

0.02 0.04 – 0.10
0.134
0.04 0.01– 0.23
0.074
0.04 0.05– 0.26
0.173,4
9
24 –52
37 4
2.7
18.4 –28.3 23.94
8
1–29
15
0.4
0 –1
0.44
39
132–274 203
16
46 –121
65
48
26 –248
91
16
22– 81
44 3,4
7
2–32
13
5
11–31
18
5
8 –35
16
0.2
0.5–1.2
0.83,4
1.2
0.03– 4.9
2.2
2.0
6.6 –13.8
9.8
0.1
0.8 –1.3
1.2
28
98 –191 147

SD
0.03
0.07
0.07
7
4.3
7
0.4
39
16
41
11
8
7
11
0.3
1.5
3.1
0.7
53

Range

Quartile 3 (n ! 25)
Mean

SD

0.10 – 0.19
0.391
0.02– 0.34
0.611
0.09 – 0.35
0.801,2,4
25–52
39
17.3–32.3 25.5
3–29
12
0 –1
0.3
138 –302 200
33–104
63
24 –172 107
20 – 65
59 2
5– 41
14
9 – 43
19
6 –58
16
0.2–1.7
0.61,2
0.03– 6.7
2.5
4.6 –18.3
9.9
0.7– 4.6
1.1
81–345 133

0.15
0.74
0.61
9
6.1
9
0.4
32
15
53
16
6
5
6
0.2
1.4
2.3
0.1
31

Range

Quartile 4 (n ! 24)
Mean

SD

0.20 – 0.70
1.511,2,3 0.76
0.04 –3.50
1.881,2,3 1.20
0.21–3.02
2.771,2,3 1.44
25–58
44 1,2
6
18.3– 45.3 28.71,2
5.7
2–27
14
10
0 –2
0 1,2
0.1
139 –271 208
42
38 –104
60
13
32–233 164
206
32–91
69 1,2
18
7–30
22 1
21
11–33
19
7
7–36
19
10
0.3–1.0
0.51,2
0.1
0.7– 6.5
2.3
1.0
5.8 –15.1
9.1
2.1
0.9 –1.3
1.0
0.1
86 –228 127
28

Range
0.77–3.62
0.25–5.41
0.86 –7.81
30 –52
20.3– 41.5
3–32
0 –1
129 –313
36 –91
42–1049
41–100
6 –97
7–39
6 – 47
0.3– 0.8
1.0 –5.2
5.8 –14.2
0.7–1.2
86 –196

* Number of female employees by location, production (P) and non-production (NP) category and quartile (percent in parenthesis).
Q1

Antwerp
Decatur
Total
1

Q2

Q3

Q4

P

NP

P

NP

P

NP

P

NP

3
0
3 (12)

20
1
21 (88)

2
1
3 (12)

17
4
21 (88)

1
7
8 (32)

6
11
17 (68)

0
22
22 (92)

0
2
2 (8)

Mean is significantly different (P % .05, Bonferroni (Dunn) t test) from the mean of the 1st quartile; 22nd quartile; 33rd quartile; 44th quartile.

the year 2000 medical surveillance
program and at least one of the two
previous program years: 64 (37%)
participated in both 1994/95 and
2000 (Antwerp & 45, Decatur & 19);
69 (39%) of the longitudinal group
participated in both 1997 and 2000
(Antwerp & 34, Decatur & 35); and
41 (24%) participated in all 3 years
(Antwerp & 21, Decatur & 20). For
purposes of brevity, these three subpopulations are referred to as subgroups A, B, and C, respectively.
Presented in Table 5 are the mean
PFOS, PFOA, and TOF concentrations for these three subgroups by
manufacturing site and year of the
medical surveillance program. Serum PFOS declined over the 6-year
time period whereas mean serum
PFOA concentrations increased for
subgroups B and C.
Serum PFOS was not a significant
predictor of cholesterol or triglycer-

ides in the longitudinal analyses (Table 6). PFOA and TOF, however,
were positively associated with cholesterol as well as triglycerides. This
association was primarily attributed
to the 21 Antwerp employees represented in subgroup C whose mean
serum PFOA levels increased over
the 6-year period from 1.32 ppm to
2.06 ppm whereas their serum PFOS
levels declined from 2.10 ppm to
1.53 ppm. During the same time
period, their mean cholesterol values
increased from 208 mg/dL to 229
mg/dL and their triglyceride levels
increased from 85 mg/dL to 123
mg/dL. Their BMIs increased from
23.4 to 24.3. There were no significant PFOS, PFOA, or TOF coefficients associated with changes in
HDL or the various liver function
tests (data not shown) adjusting for
the potential confounders.

Discussion
Medical surveillance programs in
the workplace, including the present
fluorochemical program, are usually
voluntary. High participation rates in
these voluntary programs are important to identify employees with abnormal test results, minimize participation bias, and increase the
statistical power to detect subtle abnormalities.36 Also, nonparticipation
may not allow for an adequate characterization of the distribution of
chemical-specific serum concentrations in employee biomonitoring
analyses. We previously addressed
this latter question by conducting a
random sample biomonitoring study
of Decatur employees and 80% of
those eligible participated.37 The serum distributions of PFOS and
PFOA concentrations were comparable to previous data collected

 JOEM

•

Volume 45, Number 3, March 2003

267

TABLE 4
Number of Participants (Percent in Parenthesis) by Employee Population Who Had Above Reference Range Values for
Hepatic Clinical Chemistry Tests by Quartile of Serum PFOS Distribution
Alkaline
Phosphatase
Q1

Q2

Q3

AST

Q4

Q1

Q2

ALT

Q3

Q4

Q1

Q2

Q3

GGT
Q4

Q1

Q2

Q3

Total Liver Panel*
Q4

Q1

Q2

Q3

Q4

Antwerp & Decatur
Male Employees
Production and1 0 (0) 1 (1) 3 (3) 2 (2) 3 (3) 1 (1) 1 (1) 4 (4) 4 (4) 4 (4) 7 (7) 13 (12) 6 (6) 8 (8) 6 (6) 12 (12) 15 (14) 17 (16) 17 (16) 24 (23)
Non-Production
Female Employees
Production and2 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 1 (4) 0 (0) 0 (0) 0 (0) 1 (4) 0 (0)
0 (0) 0 (0) 0 (0) 0 (0)
2 (8)
0 (0)
2 (8)
0 (0)
2 (8)
Non-Production
* Includes Alkaline Phosphatase, AST, ALT, GGT, Total and Direct Bilirubin (at least one of these tests were above upper reference range).
See Table 2 PFOS quartile distribution.
2
See Table 3 PFOS quartile distribution.
1

TABLE 5
Mean and Standard Deviation of PFOS, PFOA, TOF for Three Employee Subgroups Who Participated in Fluorochemical
Medical Surveillance Program Between 1994 and 2000
Subgroup A (N ! 64)
1994/1995

PFOS
PFOA
TOF
a

Subgroup B (N ! 69)

2000

1997

Subgroup C (N ! 41)

2000

1994/1995

1997

2000

Mean

SD

Mean

SD

Mean

SD

Mean

SD

Mean

SD

Mean

SD

Mean

SD

2.13
1.36
2.32

2.16
1.63
2.09

1.36
1.59
1.98

1.24
1.72
1.69

1.36
1.22
1.73

1.37
0.97
1.47

1.20
1.49a
1.80

0.91
1.05
1.16

2.13
1.41
2.35

1.46
1.09
1.36

2.09
1.90a
2.66a

1.54
1.87
1.72

1.65
1.77a
2.29

1.62
1.22
1.60

comparison between 1994/1995 and 2000 mean data, P % .05

through past, as well as the present
fluorochemical medical surveillance
programs; thus, we have concluded
that this voluntary fluorochemical
medical surveillance program has
likely presented a nonbiased assessment of the employees’ serum fluorochemical distribution.
The company’s 2000 fluorochemical medical surveillance program
had three times the number of participants compared to previous surveillance years. The voluntary participation rates for the present assessment
were 53% (Decatur) and 75% (Antwerp) and were higher than those in
previous surveillance years. This increase in employee participation was
likely the consequence of at least two
factors: 1) heightened employee
awareness about their serum fluorochemical values; and 2) the company’s announcement that it would voluntarily cease production of
perfluorooctanyl chemistry in certain

repellents and surfactants because of
the pervasive and persistent nature of
PFOS.
We observed several demographic
and lifestyle differences between the
Antwerp and Decatur employees. In
particular, Antwerp male employees
were younger, had lower BMIs, and
drank more alcoholic beverages than
their Decatur counterparts. These
differences can be important confounding variables when analyzing
lipid and hepatic clinical chemistries.
In the cross-sectional analysis, the
positive association in the multivariable models between PFOS and serum cholesterol is contrary to the
substantial body of toxicological literature that suggests a negative association
in
laboratory
animals.12,13,18,38 In a 6-month primate
PFOS capsule study, triglycerides
were unaffected but decreased serum
cholesterol was an early toxicological response that occurred at serum

PFOS levels above 100 ppm.12 This
serum PFOS concentration is approximately ten times greater than
the highest employee value (10.06
ppm) measured in the present study.
Accordingly, the positive association
we observed between measured serum PFOS concentrations and total
cholesterol appears spurious.
The serum PFOA concentrations
measured in these 518 employees
were lower than those measured
among the company’s manufacturing
employees at its Cottage Grove facility (the primary site of PFOA manufacture) whose serum PFOA levels
have been assayed as high as 114
ppm (mean & 5 ppm; median & 1
ppm).31 Mean serum trigylceride
levels were greater among the Cottage Grove APFO production workers with the highest (!10 ppm)
PFOA serum concentrations although adjustment for potential confounders have provided inconclusive

 268

PFOS and PFOA Analysis

•

Olsen et al

TABLE 6
Longitudinal Analyses of Serum Cholesterol or Triglycerides Levels by PFOS, PFOA or TOF and Other Covariates for 174
Male Employees
Cholesterol#
Coefficient
PFOS Model*
PFOS
Years Observed
PFOS ( Years Obs
PFOA Model*
PFOA
Years Observed
PFOA ( Years Obs
TOF Model*
TOF
Years Observed
TOF ( Years Obs

Triglycerides#
95% C.I.

Coefficient

95% C.I.

0.010
0.0009
!0.0004

(!0.005)– 0.025
(!0.008)– 0.010
(!0.004)– 0.003

0.025
!0.004
0.006

(!0.015)– 0.065
(!0.029)– 0.021
(!0.004)– 0.015

0.032
0.005
!0.005

0.013– 0.051
(!0.004)– 0.014
(!0.001)–(!0.002)

0.094
0.007
!0.008

0.045– 0.144
(!0.017)– 0.031
(!0.018)– 0.002

0.021
0.004
!0.003

0.006 – 0.035
(!0.005)– 0.014
(!0.005)– 0.0003

0.053
!0.0005
!0.0005

0.014 – 0.093
(!0.027)– 0.026
(!0.008)– 0.007

#
Natural log
* Adjusted for age, BMI, drinks/day, cigarettes/day, location, entry period and baseline years worked.

results.31 The positive associations
observed between PFOA and triglycerides in the present study are inconsistent with the known hypolipidemic effect of this compound in rats
that is thought to be associated with
activation of the nuclear receptor,
peroxisome proliferator-activated receptor (PPAR").14,16,39 Haughom
and Spydevold have suggested that
this hypolipidemic effect results
from an overall decrease in lipoprotein formation due to decreased activities of hydroxymethylglutaryl
CoA reductase and acyl-CoA:cholesterol acyltransferase as well as a
reduction in fatty acid synthesis.14
The relevance of the PPAR"-mediated response in humans has been
debated because of the greatly decreased expression of PPAR" in humans and other nonresponsive species.40,41 On the other hand, PPAR',
a nuclear receptor expressed mainly
in adipose tissue, has been shown to
be activated by the antidiabetic thiazolidinendiones, which upregulates
glycerol kinase activity stimulating
increased hepatic triglycerides.42
Whether this mode of action is plausible for PFOA in humans is not
known. Mouse and human PPAR'1
were unresponsive to PFOA when
tested at a range of 0.5 to 40 #M in
a cell transfection assay.43 In a
6-month oral toxicity study of PFOA

(ammonium salt) in male cynomolgus monkeys, PFOA was significantly associated with triglycerides
in the high-dose group.22 This association was observed in measurements taken after one month of dosing at which time group mean
triglyceride was significantly higher
than control values as well as within
group pretreatment values. At the
end of the study mean triglyceride
was elevated compared to time related controls but not to the animals’
pretreatment values. However, only
two primates were evaluated in the
high-dose group at the end of study.
Inspection of individual values for
PFOA serum concentration and serum triglyceride values did not reveal a meaningful association between these two parameters (John
Butenhoff, personal communication). Therefore, although the possibility cannot be discounted that
PFOA may activate the PPAR' nuclear receptor with a resulting increase in triglycerides, the evidence
for this mechanism remains equivocal.
Adjusting for potential confounding factors, there were no substantial
associations between hepatic enzymes and the employees’ serum
PFOS concentrations. This observation is consistent with results from
the 6-month PFOS capsule feeding

study where no overt hepatic toxicity
was observed in the two lower dose
groups (0.03 or 0.15 mg/kg/day)
whose mean serum concentrations
measured 16 and 83 ppm (males) and
13 and 67 ppm (females), respectively.12 The 0.75 mg/kg/day treatment
group did show hepatocellular hypertrophy and lipid vacuolation
along with other signs of toxicity and
two of six male monkeys died (no
female deaths). No significantly increased differences from control or
pretreatment values were observed at
end-of-study for serum levels of alkaline phosphatase or ALT in the
high dose group. Mean serum concentrations for male and female
monkeys at this dose was 170 ppm.
The liver:serum concentration ratio
in the primate was approximately 1:1
to 2:1 for all treatment groups.
We observed a positive association between PFOS and T3 in the
longitudinal assessment. There is unlikely any clinical relevance for this
association. No other thyroid hormone was associated with PFOS and
the observation is contrary to the
primate study, which showed lowered T3 values without an indication
of a hypothyroid compensatory increase in TSH, hypolipidemia or thyroid gland histological changes.
Retrospective cohort mortality
studies have not reported statistically

 JOEM

•

Volume 45, Number 3, March 2003

significant standardized mortality ratios for all cancer or liver cancer
deaths at either the Decatur or Cottage Grove manufacturing sites.44 – 46
The Decatur mortality study did observe three deaths from bladder cancer compared to 0.2 expected (standarized mortality ratio & 12.8; 95%
CI & 2.6 –37.4) in the subgroup of
workers with the highest potential
exposure to perfluorooctanesulfonyl
fluoride (POSF)-based materials.44
Alexander et al. did not determine
whether this association was fluorochemical-related or possibly due to
other nonfluorochemical occupational exposures, nonoccupational
exposures or chance. PFOS did not
result in any bladder tumors in the
high dose group (20 ppm) of a 2-year
bioassay of rats.13 The present study
found no differences among Decatur
employees’ serum PFOS concentrations and their urinalyses. Additional
epidemiologic research is ongoing to
further understand the bladder cancer
mortality association among the Decatur workforce.
Although we were able to perform
a longitudinal assessment of the
medical surveillance data, several
limitations remained in our analyses.
Only 41 of the 175 employees in the
longitudinal analysis participated in
all three surveillance years. Insufficient numbers prevented longitudinal analyses of the female employees. Because 3M announced a
phase-out of the production of perfluorooctanyl chemistry-related
materials, it is doubtful whether
there will be more employees who
can be included in future longitudinal assessments. Given the variability inherent in the analytical
method and the different laboratories used, serum measurements of
PFOS and PFOA may have varied
$20% although most were $10%.
This experimental error may have
masked associations with lipid or
hepatic clinical chemistries although the range of PFOS and
PFOA measured was relatively
consistent throughout the study
time period. Our findings suggest

that employees’ measured serum
PFOS concentrations have either
remained constant or declined
slightly during this 6-year time period. However, serum PFOA levels
trended slightly upwards during the
study period, which may have been
the result of increased production.
In summary, a cross-sectional
analysis of 421 male and 97 female
POSF and PFOA production and
nonproduction employees, as well as
a longitudinal analysis over a 6-year
time period of 174 male employees,
have not shown substantial changes
in lipid or hepatic clinical chemistry
test results that are consistent with
the known toxicological effects of
these compounds. This finding was
not unexpected because these employees’ average serum concentrations were considerably lower than
those known to cause the earliest
clinical effects in laboratory animals.

Acknowledgment
The authors gratefully acknowledge the
assistance of Dr John Butenhoff.

References
1. 3M Company. SIDS Initial assessment
report: Perfluorooctane sulfonic acid and
its salts. St. Paul, MN: 3M Company;
2000. U. S. EPA Docket AR-226.
2. Hansen KJ, Clemen LA, Ellefson ME,
Johnson JHO. Compound-specific, quantitative characterization of organic fluorochemicals in biological matrices. Environ Sci Technol. 2001;35:766 –770.
3. Giesy JP, Kannan K. Global distribution
of perfluorooctane sulfonate in wildlife.
Environ Sci Technol. 2001;35:1339 –
1342.
4. Kannan K, Koistenen J, Beckmen K, et
al. Accumulation of perfluorooctane sulfonate in marine mammals. Environ Sci
Technol. 2001;35:1593–1598.
5. Kannan K, Franson JC, Bowerman WW,
Hansen KJ, Jones PD, Geisy JP. Perfluorooctane sulfonate in fish-eating water
birds including bald eagles and albatrosses. Environ Sci Technol. 2001;35:
3065–3070.
6. Johnson JD, Gibson SJ, Ober RE. Extent
and route of excretion and tissue distribution of total carbon-14 in rats after a
single i.v. dose of FC-95-14C. St. Paul,
MN :Riker Laboratories, Inc.; 1979. U. S.
EPA Docket AR-226.

269
7. Johnson JD, Gibson SJ, Ober RE. Cholestyramine-enhanced fecal elimination
of carbon-14 in rats after administration
of ammonium [14C] perfluorooctanoate
or potassium [14C] perfluorooctanesulfonate. Fundam Appl Toxicol. 1984;4:
972–76.
8. Grasty RC, Grey BE, Thibodeaux J, Lau
C, Rogers JM. Critical period for increased neonatal mortality induced by
perfluorooctane sulfonate (PFOS) in the
rat (abstract 118). Toxicologist. 2002;66:
25.
9. Lau C, Rogers JM, Thibodeaux JR, et al.
Maternal and developmental toxicity of
perfluorooctane sulfonate (PFOS) in the
rat (abstract 119). Toxicologist. 2002;66:
25.
10. Butenhoff J, York R, Seacat A, Luebker
D. Perfluorooctanesulfonate-induced
perinatal mortality in rat pups is associated with a steep dose-response (abstract
120). Toxicologist. 2002;66:25.
11. Butenhoff JL, Seacat AM. Comparative
sub-chronic toxicity of perfluorooctane
sulfonate (PFOS) and N-ethyl perfluorooctanesulfonamido-ethanol (N-Et
FOSE) in the rat (abstract 1655). Toxicologist. 2001;60:348.
12. Seacat AM, Thomford PJ, Hansen KJ,
Olsen GW, Case MT, Butenhoff JL. Subchronic toxicity studies on perfluorooctanesulfonate potassium salt in cynomolgus monkeys. Toxicol Sci. 2002;68:
249 –264.
13. Seacat AM, Thomford PJ, Butenhoff JL.
Terminal observations in Sprague Dawley rats after lifetime dietary exposure to
potassium perfluorooctanesulfonate (abstract 906). Toxicologist. 2002;66:185.
14. Haughom B, Spydevold O. The mechanism underlying the hypolipemic effect
of perflurooctanoic acid (PFOA), perfluorooctanesulphonic acid (PFOSA) and
clofibric acid. Biochemica et Biophysica
Acta. 1992;1128:65–72.
15. Ikeda T, Fukuda K, Mori I, Enomoto M,
Komai T, Suga T. Induction of cytochrome P-450 and peroxisome proliferation in rat liver by perfluorinated octanesulfonic acid. In: Fahmi HD, Sies H, eds.
Perixosmes in Biology and Medicine.
New York: Springer Verlag; 1987:304 –
308.
16. Berthiaume JM, Wallace KB. Perfluorooctanoate, perfluorooctanesulfonate
and N-ethyl perfluorooctanesulfonamido
ethanol: peroxisome proliferation and mitochondrial biogenesis. Toxicol Lett.
2002;129:23–32.
17. Luebker DJ, Hansen KJ, Bass N, Butenhoff JL, Seacat AM. Interactions of fluorochemicals with rat liver fatty acid-

 270

18.

19.

20.

21.

22.

23.

24.

25.

26.

binding protein. Toxicology. 2002;176:
175–185.
Starkov A, Wallace KB. Structural determinants of fluorochemical-induced mitochondrial dysfunction. Toxicol Sci. 2002;
66:244 –252.
Giesy JP, Huy WY, Jones P, Upham B,
Trosko J. Comparisons among perfluorinated compounds of effects on gap junctional intercellular communication (abstract 208). Toxicologist. 2002;66:43– 44.
Johnson JD, Ober RE. Extent and route
of excretion and tissue distribution of
total carbon-14 in male and female rats
after a single IV dose of FC-143-14C. St.
Paul, MN: Riker Laboratories, Inc.; 1980.
U. S. EPA Docket AR-226.
Hanjijarvi H, Ylinen M. A proposed species difference in the renal excretion of
perfluorooctanoic acid in the beagle dog
and rat. In: Beynen A, Sollveld H, eds.
New Developments in Biosciences: Their
Implications for Laboratory Animals Sciences. Dordrecht: Martinus Nijhoff;
1988:409 – 412.
Butenhoff JL, Costa G, Elcombe C, et al.
Toxicity of ammonium perfluorooctanoate (APFO) in cynomolgus monkeys after
26 weeks of oral dosing. Toxicol Sci.
2002;69:244 –257.
Ubel FA, Sorenson SD, Roach DE.
Health status of plant workers exposed to
fluorochemicals—a preliminary report.
Am Ind Hyg Assoc J. 1980;41:584 –589.
Pastoor TP, Lee KP, Perri MA, Gillies
PJ. Biochemical and morphological studies of ammonium perfluorooctanoateinduced hepatomegaly and peroxisome
proliferation. Exp Mol Pathol. 1987;47:
98 –109.
Biegel LB, Hurt ME, Frame SR,
O’Connor JL, Cook JC. Mechanisms of
extrahepatic tumor induction by peroxisome proliferators in male CD rats. Toxicol Sci. 2001;60:44 –55.
Cook JC, Murray SM, Frame SR, Hurtt
ME. Induction of Leydig cell adenomas
by ammonium perfluorooctanoate: A
possible endocrine-related mechanism.
Toxicol Appl Pharmacol. 1992;113:209 –
217.

PFOS and PFOA Analysis
27. Liu RCM, Hurtt ME, Cook JC, Biegel
LB. Effect of the peroxisome proliferator,
ammonium perfluorooctanoate (C8) on
aromatase activity in tissues of male
Crl:CD BR (CD) rats. Fundam Appl
Toxicol. 1996;30:220 –228.
28. Obourn JD, Frame SR, Bell RH, Longnecker DS, Elliott GS, Cook JC. Mechanisms for the pancreatic oncogenic effects of the peroxisome proliferator,
Wyeth-14, 643. Toxicol Appl Pharmacol.
1997;45:425– 436.
29. Gilliland FD, Mandel JS. Serum perfluorooctanoic acid and hepatic enzymes, lipoproteins and cholesterol: a study of
occupationally exposed men. Am J Ind
Med. 1996;129:560 –568.
30. Olsen GW, Gilliland FD, Burlew MM,
Burris JM, Mandel JS, Mandel JH. An
epidemiologic investigation of reproductive hormones in men with occupational
exposure to perfluorooctanoic acid. J Occup Environ Med. 1998;40:614 – 621.
31. Olsen GW, Burris JM, Burlew MM,
Mandel JH. Plasma cholecystokinin and
hepatic enzymes, cholesterol and lipoproteins in ammonium perfluorooctanoate
production workers. Drug Chem Toxicol.
2000;23:603– 620.
32. Olsen GW, Burris JM, Mandel JH, Zobel
LR. Serum perfluorooctane sulfonate and
hepatic and lipid clinical chemistry tests
in fluorochemical production employees.
J Occup Environ Med. 1999;41:799– 806.
33. Burlew MM. SAS Macro Programming
Made Easy. Cary, NC: SAS Institute Inc.;
1998:1–280.
34. SAS Institute, Inc. SAS/STAT User’s
Guide, Version 8. Cary, NC: SAS Institute Inc.; 1999.
35. Littell RC, Pendergast J, Natarajan R.
Modeling covariance structure in the
analysis of repeated measures data. Stat
Med. 2000;1793–1819.
36. Rempel D. Medical surveillance in the
workplace: overview. Occup. Med. 1990;
5:435– 438.
37. Olsen GW, Logan PW, Hansen KJ, et al.
An occupational exposure assessment of
a perfluorooctanesulfonyl fluoride pro-

38.

39.

40.

41.

42.

43.

44.

45.

46.

•

Olsen et al

duction site. Biomonitoring. Am Ind Hyg
Assoc J. In press.
Sohlenius AK, Eriksson AM, Hogstrom
C, Kimland M, DePierre JW. Perfluorooctane sulfonic acid is a potent inducer of peroxisomal fatty acid Boxidation and other activities known to
be affected by peroxisome proliferators
in mouse liver. Pharmacol Toxicol.
1993;72:90 –93.
Just WW, Gorgas K, Hartl FU, Heinemann P, Salzer M, Schimassek H. Biochemical effects and zonal heterogeneity
of peroxisome proliferation induced by
perfluorocarboxylic acids in rat liver.
Hepatology. 1989;9:570 –581.
Cattley RC, DeLuca J, Elcombe C, et al.
Do peroxisome proliferating compounds
pose a hepatocarcinogenic hazard to humans? Reg Toxicol Pharmacol. 1998;27:
47– 60.
Bentley P, Calder I, Elcombes C, Grasso
P, Stringer D, Wiegand JH. Hepatic peroxisome proliferation in rodents and its
significance for humans. Fd Chem Toxicol. 1993;31:857–907.
Guan HP, Li Y, Jensen MV, Newgard
CB, Steppan CM, Lazar MA. A futile
metabolic cycle activated in adipocytes
by antidiabetic agents (advance online
publication). Nat Med. 2002:1–7.
Maloney EK, Waxman DJ. Transactivation of PPAR" and PPAR' by
structurally diverse environmental chemicals. Toxicol Appl Pharmacol. 1999;161:
209 –218.
Alexander BH, Olsen GW, Burris JM,
Mandel JH, Mandel JS. Mortality of
workers employed at a fluorochemical
production facility. Occup Environ Med.
In press.
Gilliland FD, Mandel JS. Mortality
among employees of a perfluorooctanoic
acid production plant. J Occup Med.
1993;35:950 –954.
Alexander BH. Mortality study of workers employed at the 3M Cottage Grove
facility. Minneapolis, MN: University of
Minnesota; April, 2001. U. S. EPA
Docket AR-226.