DEQ-CFW_00001657

REVISED DRAFT
HAZARD ASSESSMENT OF
PERFLUOROOCTANOIC ACID
AND ITS SALTS
U.S. Environmental Protection Agency
Office of Pollution Prevention and Toxics
Risk Assessment Division
November 4, 2002
DEQ-CFW 00001657

PREFACE
This is a preliminary assessment of the potential hazards to human health and the environment
associated with exposure to perfluorooctanoic acid (PFOA) and its salts. The majority of the
toxicology information is for ammonium perfluorooctanoic acid (APFO). This assessment
includes a review of the studies that were available as of October, 2002.
DEQ-CFW 00001658

i
Table of Contents
Executive Summary
1
1.0
Chemical Identity
7
1.1
Physicochemical Properties
7
2.0
Production of PFOA and its Salts
9
2.1
Uses of PFOA and its Salts
11
2.2
Environmental Fate
12
2.2.1
Photolysis
12
2.2.2
Volatility
12
2.2.3
Biodegradation
13
2.2.4
Hydrolysis
13
2.2.5
Bioaccumulation
14
2.2.6
Soil Adsorption
15
2.3
Environmental Exposure
15
2.3.1
Discharge to Air
15
2.3.2
Discharge to Water
15
2.3.3
Discharge to Land
16
2.3.4
Environmental Monitoring
16
2.4
Human Biomonitoring
18
3.0
Human Health Hazards
23
3.1.
Metabolism and Pharmacokinetics
23
3.1.1
Half-life in Humans
23
3.1.2
Absorption Studies in Animals
24
3.1.3
Distribution Studies in Animals
25
3.1.4
Metabolism Studies in Animals
28
3.1.5
Elimination Studies in Animals
28
3.2
Epidemiology Studies
32
3.2.1
Medical Surveillance Studies from Antwerp and Decatur Plants
32
3.2.2
Medical Surveillance Studies from Cottage Grove Plant
34
3.2.3
Mortality Studies
36
3.2.4
Hormone Study
39
3.2.5
Study on Episodes of Care (Morbidity)
40
3.3
Acute Toxicity Studies in Animals
42
3.3.1
Oral Studies
42
3.3.2
Inhalation Studies
43
3.3.3
Dermal Studies
43
3.3.4
Eye Irritation Studies
43
3.3.5
Skin Irritation Studies
44
3.4
Mutagenicity Studies
44
3.5
Subchronic Toxicity Studies in Animals
44
3.6
Developmental Toxicity Studies in Animals
60
3.7
Reproductive Toxicity Studies in Animals
64
DEQ-CFW 00001659

3.8 Carcinogenicity Studies in Animals
74
3.8.1 Cancer Bioassays
74
3.8.2 Mode of Action Studies
75
3.8.2.1 Liver Tumors
76
3.8.2.2 Leydig Cell Tumors
76
3.8.2.3 Mammary Gland Tumors
77
3.8.2.4 Pancreatic Tumors
77
3.9 Immunotoxicology Studies in Animals
78
4.0 Hazards to the Environment
80
4.1 Introduction
80
4.2 Acute Toxicity to Freshwater Species
81
5.0 References
87
ANNEX I — Robust Summaries
104
DEQ-CFW 00001660

EXECUTIVE SUMMARY
Introduction
Perfluorooctanoic acid (PFOA) and its salts are fully fluorinated organic compounds that can be
produced synthetically or through the degradation or metabolism of other fluorochemical
products. PFOA is primarily used as a reactive intermediate, while its salts are used as
processing aids in the production of fluoropolymers and fluoroelastomers and in other surfactant
uses. In recent years, less than 600 metric tons per year of PFOA and its salts have been
manufactured in the United States or imported. Most of the toxicology studies have been
conducted with the ammonium salt of perfluorooctanoic acid, which is referred to as APFO in
this report.
Environmental Fate and Effects
PFOA is persistent in the environment. It does not hydrolyze, photolyze or biodegrade under
environmental conditions.
Groundwater samples taken near fire -training areas that used fire -fighting foams containing
perfluorinated surfactants had elevated PFOA concentrations many years after the foam use.
This demonstrates the following: (1) PFOA either existed in --or was formed via degradation of --
the surfactants, (2) PFOA or its precursors migrate through the soil, and (3) PFOA persists in
groundwater.
Several wildlife species have been sampled around the world to determine levels of PFOA.
PFOA has rarely been found in fish sampled from the U.S., certain European countries, the
North Pacific Ocean and Antarctic locations, or in fish -eating bird samples collected from the
U.S., including Midway atoll, the Baltic and Mediterranean Seas, and Japanese and Korean
coasts. PFOA was found in a few mink livers from Massachusetts, but not found in mink from
Louisiana, South Carolina and Illinois. PFOA concentrations in river otter livers from
Washington and Oregon States were less than the quantification limit of 36 ng/g, wet wt. PFOA
was not detected at quantifiable concentrations in oysters collected in the Chesapeake Bay and
Gulf of Mexico of the U.S. coast.
The concentrations of PFOA in surface water, sediments, clams, and fish collected from
locations upstream and downstream of the 3M manufacturing facility at Decatur AL have been
determined. Of the three downstream water and sediment sampling locations, the two closest to
the facility had PFOA surface water concentrations significantly greater than the two upstream
sites; the three downstream locations also had sediment concentrations significantly greater than
the upstream sites. The small sample size prevented determination of significance for fish whole
body PFOA concentrations. The average PFOA concentration in clams was not significantly
different between the upstream and downstream locations.
Based on available laboratory data, APFO does not appear to bioaccumulate in fish. In a study
of fathead minnows, the calculated BCF for APFO was 1.8. In_a study of carp, the BCF ranged
DEQ-CFW 00001661

from 3.1 to 9.1.
Several species were tested to assess the acute toxicity of APFO; these included the fathead
minnow (Pimephales promelas), bluegill sunfish (Lepornis machrochirus), water flea (Daphnia
magna), and a green algae (Selenastrum capricornutum). Comparisons of the different studies
are problematic for several reasons. The studies were conducted with different test substances.
Generally the ammonium salt or the tetrabutylammonium salt was tested. Purity of the test
material is a major concern and was not sufficiently characterized in these tests. In some tests it
appeared that 100% test chemical was used, for others a chemical of lesser purity (approximately
27 to 85%) was used. Water, a solvent (isopropanol) or a combination of both was used in other
tests, for no obvious stated reason. Finally, only nominal test chemical concentrations were
reported; the actual concentrations were not reported.
Twelve tests were conducted with fathead minnows; 96-h LC50 values (based on mortality)
ranged from 70 to 843 mg/L. It is unclear why this range is so wide. Assuming these studies are
valid, and due to the limitations discussed above, these toxicity values indicate low toxicity. The
two acute values for bluegill sunfish also indicate low toxicity (96-h LC50s of >420, and 569
mg/L).
Nine acute tests were conducted with daphnids and 48-h EC50 values (based on immobilization)
ranged from 39 to >1000 mg/L. The lower values are indicative of moderate toxicity, but the
wide range makes interpretation difficult.
Seven tests were conducted with green algae; 96-h EC50 values (based on growth rate, cell
density, cell counts, and dry weights) ranged from 1.2 to >666 mg/L (the EC50 cell density value
of 1,000 mg/L is excluded from this discussion). The lower value indicates high to moderate
toxicity, based on the acute criteria. The lower value would also be indicative of moderate
toxicity, based on the chronic moderate criterion (0.1<10 mg/L). A 14-d EC50 value of 43
mg/L, based on cell counts, for green algae was also calculated in one study. This is indicative
of low chronic toxicity, based on the chronic criterion (10 mg/L). Green algae appeared to be
the most sensitive test species in the 44% APFO test sample, daphnids were the next most
sensitive, and fathead minnows were the least sensitive.
Human Health Effects and Biomonitoring
Little information is available concerning the pharmacokinetics of APFO in humans. An
ongoing 5-year, half-life study in 9 retired workers has suggested a mean serum PFOA half-life
of 4.37 years (range, 1.50 — 13.49 years). These data provide evidence of the potential to
bioaccumulate PFOA in humans.
Animal studies have shown that APFO is well absorbed following oral and inhalation exposure,
and to a lesser extent following dermal exposure. In the past, Chemolite workers have been
exposed to large dermal doses of PFOA. It appears that dermal exposure may have played a
significant role in the absorption of PFOA in these workers. Upon recognition that PFOA could
2
DEQ-CFW 00001662

be absorbed dermally, work practices were changed and engineering controls were adopted that
reduced dermal exposures.
PFOA distributes primarily to the liver, plasma, and kidney, and to a lesser extent, other tissues
of the body including the testis and ovary. It does not partition to the lipid fraction or adipose
tissue. PFOA binds to macro -molecules in the tissues listed above. PFOA is not metabolized
and there is evidence of enterohepatic circulation of the compound. The urine is the major route
of excretion of PFOA in the female rat, while the urine and feces are both major routes of
excretion in male rats. There are major gender differences in the elimination of PFOA in rats. In
female rats, the half-life is 24 h in the serum and 60 h in the liver; in male rats, the half-life is
105 h in the serum and 210 h in the liver. The rapid excretion of PFOA by female rats is due to
active renal tubular secretion (organic acid transport system); this renal tubular secretion is
believed to be hormonally controlled, since castrated male rats treated with estradiol have
excretion rates of PFOA similar to those of female rats. Hormonal changes during pregnancy do
not appear to change the rate of elimination in rats. This gender difference has not been
observed in primates and humans.
There are limited data on PFOA serum levels in workers and the general population.
Occupational data from plants in the U.S. and Belgium that manufacture or use PFOA indicate
that the most recent mean serum levels in workers range from 0.84 to 6.4 ppm. The highest level
reported in a worker in 1997 was 81.3 ppm. In non -occupational populations, serum PFOA
levels were much lower. In both pooled blood bank samples and in individual samples, mean
serum PFOA levels ranged from 3 to 17 ppb. The highest serum PFOA levels were reported in a
sample of children from different geographic regions in the U.S. (range, 1.9 — 56.1 ppb).
Epidemiological studies on the effects of PFOA in humans have been conducted on workers.
Two mortality studies, a morbidity study, and studies examining effects on the liver, pancreas,
endocrine system, and lipid metabolism, have been conducted to date. In addition, a cross -
sectional as well as a longitudinal study of the worker surveillance data have recently been
submitted. However, these latest 2 studies focus primarily on PFOS rather than PFOA, even
though recent PFOA levels are similar to or higher than PFOS levels in workers at these plants.
(It should be noted that PFOS levels in the sampled general population are much higher than
PFOA levels).
A retrospective cohort mortality study demonstrated a weak, although statistically significant
association between prostate cancer mortality and employment duration in the chemical facility
of a plant that manufactures PFOA. However, in a recent update to this study in which more
specific exposure measures were used, a significant association for prostate cancer was not
observed. In a morbidity study, workers with the highest PFOA exposures for the longest
durations sought care more often for prostate cancer treatment than workers with lower
exposures.
Another study reported an increase in estradiol levels in workers with the highest PFOA serum
levels; however, none of the other hormone levels analyzed indicated any adverse effects. Some
DEQ-CFW 00001663

density lipoproteins (LDL), and high -density lipoproteins (HDL). All of the participants were
placed into five categories of total serum fluorine levels: <1 ppm, 1-3 ppm, >3 - 10 ppm, >10 -
15 ppm, and > 15 ppm. The range of the serum fluorine values was 0 to 26 ppm (mean 3.3
ppm). Approximately half of the workers fell into the > 1 - 3 ppm category, while 23 had serum
levels < 1 ppm and 11 had levels > 10 ppm.
There were no significant differences between exposure categories when analyzed using
univariate analyses for cholesterol, LDL, and HDL. In the multivariate analysis, there was not a
significant association between total serum fluorine and cholesterol or LDL after adjusting for
alcohol consumption, age, BMI, and cigarette smoking. There were no statistically significant
differences among the exposure categories of total serum fluorine for AST, ALT and GGT.
However, increases in AST and ALT occurred with increasing total serum fluorine levels in
obese workers (BMI = 35 kg/m2). This result was not observed when PFOA was measured
directly in serum of workers in 1993, 1995, or 1997 surveillance data of employees of the
Cottage Grove plant (Olsen, et al., 2000).
Since PFOA was not measured directly and there is no exposure information provided on the
employees (eg. length of employment/exposure), the results of the study provide limited
information. The authors state that no adverse clinical outcomes related to PFOA exposure have
been observed in these employees; however, it is not clear that there has been follow-up of
former employees. In addition, the range of results reported for the liver enzymes were fairly
wide for many of the exposure categories, indicating variability in the results. Given that only
one sample was taken from each employee, this is not surprising. It would be much more helpful
to have several samples taken over time to ensure their reliability. It also would have been
interesting to compare the results of the workers who were known to be exposed to PFOA to
those who were originally thought not to be exposed to see if there were any differences among
the employees in these groups. There were more of the "unexposed" employees (n = 65)
participating in the study than those who worked in PFOA production (n = 48).
3.2.3 Mortality Studies
A retrospective cohort mortality study was performed on employees at the Cottage Grove, MN
plant which produces APFO (Gilliland and Mandel, 1993). At this plant, APFO production was
limited to the Chemical Division. The cohort consisted of workers who had been employed at
the plant for at least 6 months between January 1947 and December 1983. Death certificates of
all of the workers were obtained to determine cause of death. There was almost
complete follow-up (99.5%) of all of the study participants. The exposure status of the workers
was categorized based on their job histories. If they had been employed for at least 1 month in
the Chemical Division, they were considered exposed. All others were considered to be not
exposed to PFOA. The number of months employed in the Chemical Division provided the
cumulative exposure measurements. Of the 3537 (2788 men and 749 women) employees who
participated in this study, 398 (348 men and 50 women) were deceased. Eleven of the 50 women
and 148 of the 348 men worked in the Chemical Division, and therefore, were considered
exposed to PFOA.
36
DEQ-CFW 00001664

Standardized Mortality Ratios (SMRs), adjusted for age, sex, and race were calculated and
compared to U.S. and Minnesota white death rates for men. For women, only state rates were
available. The SMRs for males were stratified for 3 latency periods (10, 15, and 20 years) and 3
periods of duration of employment (5, 10, and 20 years).
For all female employees, the SMRs for all causes and for all cancers were less than 1. The only
elevated (although not significant) SMR was for lymphopoietic cancer, and was based on only 3
deaths. When exposure status was considered, SMRs for all causes of death and for all cancers
were significantly lower than expected, based on the U.S. rates, for both the Chemical Division
workers and the other employees of the plant.
In all male workers at the plant, the SMRs were close to 1 for most of the causes of death when
compared to both the U.S. and the Minnesota death rates. When latency and duration of
employment were considered, there were no elevated SMRs. When employee deaths in the
Chemical Division were compared to Minnesota death rates, the SMR for prostate cancer for
workers in the Chemical Division was 2.03 (95% CI.55 - 4.59). This was based on 4 deaths
(1.97 expected). There was also a statistically significant association with length of employment
in the Chemical Division and prostate cancer mortality. Based on the results of proportional
hazard models, the relative risk for a 1-year increase in employment in the Chemical Division
was 1.13 (95% Cl 1.01 to 1.27). It rose to 3.3 (95% Cl 1.02 -10.6) for workers employed in the
Chemical Division for 10 years when compared to the other employees in the plant. The SMR
for workers not employed in the Chemical Division was less than expected for prostate cancer
(.58).
An update of this study was conducted to include the death experience of employees through
1997 (Alexander, 2001a). The cohort consisted of 3992 workers. The eligibility requirement
was increased to 1 year of employment at the Cottage Grove plant, and the exposure categories
were changed to be more specific. Workers were placed into 3 exposure groups based on job
history information: definite PFOA exposure (n = 492, jobs where cell generation, drying,
shipping and packaging of PFOA occurred throughout the history of the plant); probable PFOA
exposure (n = 1685, other chemical division jobs where exposure to PFOA was possible but with
lower or transient exposures); and not exposed to fluorochemicals (n = 1815, primarily non -
chemical division jobs).
In this new cohort, 607 deaths were identified: 46 of these deaths were in the PFOA exposure
group, 267 in the probable exposure group, and 294 in the non -exposed group. When all
employees were compared to the state mortality rates, SMRs were less than 1 or only slightly
higher for all of the causes of death analyzed. None of the SMRs were statistically significant at
p = .05. The highest SMR reported was for bladder cancer (SMR = 1.31, 95% Cl = 0.42 — 3.05).
Five deaths were observed (3.83 expected).
A few SMRs were elevated for employees in the definite PFOA exposure group: 2 deaths from
cancer of the large intestine (SMR = 1.67, 95% CI = 0.02 — 6.02), 1 from pancreatic cancer
37
DEQ-CFW 00001665

(SMR = 1.34, 95% CI = 0.03 — 7.42), and 1 from prostate cancer (SMR = 1.30, 95% CI = 0.03 —
7.20). In addition, employees in the definite PFOA exposure group were 2.5 times more likely
to die from cerebrovascular disease (5 deaths observed, 1.94 expected; 95% CI = 0.84 — 6.03).
In the probable exposure group, 3 SMRs should be noted: cancer of the testis and other male
genital organs (SMR = 2.75, 95% CI = 0.07—15.3); pancreatic cancer (SMR — 1.24, 95% CI =
0.45 — 2.70); and malignant melanoma of the skin (SMR = 1.42, 95% CI = 0.17 — 5.11). Only 1,
6, and 2 cases were observed, respectively. The SMR for prostate cancer in this group was 0.86
(95% CI = 0.28 — 2.02) (n = 5).
There were no notable excesses in SMRs in the non -exposed group, except for cancer of the
bladder and other urinary organs. Four cases were observed and only 1.89 were expected (95%
CI = 0.58 — 5.40).
It is difficult to interpret the results of the prostate cancer deaths between the first study and the
update because the exposure categories were modified in the update. Only I death was reported
in the definite exposure group and 5 were observed in the probable exposure group. All of these
deaths would have been placed in the chemical plant employees exposure group in the first
study. The number of years that these employees worked at the plant and/or were exposed to
PFOA was not reported. This is important because even 1 prostate cancer death in the definite
PFOA exposure group resulted in an elevated SMR for the group. Therefore, if any of the
employees' exposures were misclassified, the results of the analysis could be altered
significantly.
The excess mortality in cerebrovascular disease noted in employees in the definite exposure
group was further analyzed based on number of years of employment at the plant. Three of the 5
deaths occurred in workers who were employed in jobs with definite PFOA exposure for more
than 5 years but less than 10 years (SMR = 15.03, 95% CI = 3.02 — 43.91). The other 2 occurred
in employees with less than 1 year of definite exposure. The SMR was 6.9 (95% CI = 1.39 —
20.24) for employees with greater than 5 years of definite PFOA exposure. In order to confirm
that the results regarding cerebrovascular disease were not an artifact of death certificate coding,
regional mortality rates were used for the reference population. The results did not change.
When these deaths were further analyzed by cumulative exposure (time -weighted according to
exposure category), workers with 27 years of exposure in probable PFOA exposed jobs or those
with 9 years of definite PFOA exposure were 3.3 times more likely to die of cerebrovascular
disease than the general population. A dose -response relationship was not observed with years of
exposure.
It is difficult to compare the results of the first and second mortality studies at the Cottage Grove
plant since the exposure categories were modified. Although the potential for exposure
misclassification was certainly more likely in the first study, it may still have occurred in the
update as well. It is difficult to judge the reliability of the exposure categories that were defined
without measured exposures. Although serum PFOA measurements were considered in the
exposure matrix developed for the update, they were not directly used. In the second study, the
38
DEQ-CFW 00001666

chemical plant employees were sub -divided into PFOA-exposed groups, and the film plant
employees essentially remained in the "non -exposed" group. This was an effort to more
accurately classify exposures; however, these new categories do not take into account duration of
exposure or length of employment. Another limitation to this study is that 17 death certificates
were not located for deceased employees and therefore were not included in the study. The
inclusion or exclusion of these deaths could change the analyses for the causes of death that had
a small number of cases. Follow up of worker mortality at Cottage Grove (and Decatur) needs to
continue. Although there were more than 200 additional deaths included in this analysis, it is a
small number and the cohort is still relatively young. Given the results of studies on
fluorochemicals in both animals and humans, further analysis is warranted.
3.2.4 Hormone Study
Endocrine effects have been associated with PFOA exposure in animals; therefore, medical
surveillance data, including hormone testing, from employees of the Cottage Grove, Minnesota
plant were analyzed (Olsen, et al., 1998a). PFOA serum levels were obtained for volunteer
workers in 1993 (n = 111) and 1995 (n = 80). Sixty-eight employees were common to both
sampling periods. In 1993, the range of PFOA was 0-80 ppm (although 80 ppm was the limit of
detection that year, so it could have been higher) and 0-115 ppm in 1995 using thermospray mass
spectrophotometry assay. Eleven hormones were assayed from the serum samples. They were:
cortisol, dehydroepiandrosterone sulfate (DHEAS), estradiol, FSH, 17 gamma-
hydroxyprogesterone (17-HP), free testosterone, total testosterone, LH, prolactin, thyroid -
stimulating hormone (TSH) and sex hormone -binding globulin (SHBG). Employees were
placed into 4 exposure categories based on their serum PFOA levels: 0-1 ppm, 1- < 10 ppm, 10-
< 30 ppm, and >30 ppm. Statistical methods used to compare PFOA levels and hormone values
included: multivariable regression analysis, ANOVA, and Pearson correlation coefficients.
PFOA was not highly correlated with any of the hormones or with the following covariates: age,
alcohol consumption, BMI, or cigarettes. Most of the employees had PFOA serum levels less
than 10 ppm. In 1993, only 12 employees had serum levels > 10 ppm, and 15 in 1995.
However, these levels ranged from approximately 10 ppm to over 114 ppm. There were only 4
employees in the >30 ppm PFOA group in 1993 and only 5 in 1995. Therefore, it is likely that
there was not enough power to detect differences in either of the highest categories. The mean
age of the employees in the highest exposure category was the lowest in both 1993 and 1995
(33.3 years and 38.2 years, respectively). Although not significantly different from the other
categories, BMI was slightly higher in the highest PFOA category.
Estradiol was highly correlated with BMI (r = .41, p < .001 in 1993, and r = .30, p < .01 in
1995). In 1995, all 5 employees with PFOA levels > 30 ppm had BMIs > 28, although this effect
was not observed in 1993. Estradiol levels in the >30 ppm group in both years were 10% higher
than the other PFOA groups; however, the difference was not statistically significant. The
authors postulate that the study may not have been sensitive enough to detect an association
between PFOA and estradiol because measured serum PFOA levels were likely below the
observable effect levels suggested in animal studies (55 ppm PFOA in the CD rat). Only 3
39
DEQ-CFW 00001667