DEQ-CFW_00001432

MEMO
DATE: November 8, 2006
TO: Alan Klimek, Director
North Carolina Division of Water Quality
FROM: Alan Clark, Chief
Planning Section
North Carolina Division of Water Quality
SUBJECT: Interim Maximum Allowable Concentration for PFOA
Enclosed, for your review, is supporting documentation to establish an Interim Maximum
Allowable Concentration (IMAC) for Perfluorooctanoic acid (PFOA or C8). The IMAC regulation
is as stated in 15A NCAC 02L.0202 (c) (Groundwater Quality Standards).
Staff, in consultation with the Division of Waste Management and with the cooperation of the
Department of Health and Human Services, has developed a health -based level for PFOA based on
the systemic threshold concentration, using published and peer -reviewed toxicological data. The
U.S. EPA does not have an IRIS, Health Advisory, or other published health risk value for PFOA.
Additionally, there is no published taste or odor threshold available.
This proposed concentration, if established, would aid the Division of Water Quality and the
Division of Waste Management in evaluating site conditions and in setting health protective ground
water and soil remediation levels. The North Carolina Science Advisory Board (NC SAB) has
begun the review of existing toxicity information. The NC SAB has had two presentations on the
PFOA risk assessment issue: one by toxicologists from Dupont, and the other by US EPA regional
staff. The Board expects that within about 6 to 9 months that they will have their assessment
completed. Should this assessment provide information to support a recalculation of this standard,
staff will pursue action to consider this information for adoption in. accordance with the 2L
regulation.
Attached is the supporting data for the calculation of an IMAC for PFOA. Please feel free to contact
Connie Brower at 733-7015 ext. 380 if you need additional clarification.
cc: Coleen Sullins
Jeff Manning
Attachments
DEQ-CFW 00001432

Recommended Interim Maximum Allowable Concentration for Perfluorooctanoic Acid
(PFOA or C8)
1. Summary
An Interim Maximum Allowable Concentration (IMAC) has been developed for PFOA:
Recommended PFOA IMAC = 0.002 mg/L or 2µg/L
Basis for level: RfD = 0.0003 mg/kg-day;
Critical endpoints = Reduction in mean body weight and body weight gain from Fl male rat pups
and Fl adult male rats (F1 generation from Fo adult rats) and increased liver weights in Fo parental
males;
Critical study = Two generation rat gavage study by York et al., 2002 and Butenhoff et al., 2004.
2. Background
The chemical name for C8 is Perfluorooctanoic acid (PFOA). CAS number 335-67-1
Molecular formula of C8HF1502. One of its salts, ammonium perfluorooctanoate (APFO;
CgF1502NH4,' CAS Number 3825-26-1) is the compound that is the most widely used in industry and
most animal toxicology studies have been carried out with this compound. Once absorbed in the
body, APFO disassociates to the PFOA anion.
3. Basis for IMAC
The first step in developing an IMAC or ground water quality standard for PFOA, in accordance
with 15A NCAC 2L .0202, requires analysis of existing data to determining whether the proposed
concentration is based on the carcinogenic or noncarcinogenic effects. 15A NCAC 2L .0202
indicates that the level must be upon the "least of a systemic threshold concentration (non -cancer
endpoint) or a concentration which corresponds to an incremental lifetime cancer risk of 1 x 10-6
(cancer endpoint). If based on carcinogenic effects, then the chemical is assumed to not exhibit a
threshold and a risk approach is used to develop the basis for the appropriate concentration. In the
risk approach, a study (usually animal, but can be human) that examined carcinogenic effects is
selected to be the basis for the assessment. Mathematical models are used to determine a cancer
slope factor based on the results seen in the study. A cancer slope factor is an upper -bound estimate
of risk per increment of dose. From the cancer slope factor, unit risk estimates are developed. Unit
risk estimates convert the cancer slope factor to units of drinking water ([.g/L) or air (µg/m3). From
the unit risk, risk -specific doses can be derived that estimate the dose associated with a specific risk
level, for North Carolina a one -in -a -million (1 x 10-6) increased lifetime risk is calculated.
If the IMAC is based on noncarcinogenic effects, then the chemical is assumed to exhibit a
threshold and a Reference Dose (RfD) is developed. The RfD is an estimate (with uncertainty
spanning perhaps an order of magnitude) of a daily exposure to the human population (including
sensitive subgroups) that is likely to be without an appreciable risk of deleterious effects during a
lifetime. The RfD is calculated to be protective against a critical effect, which also results in
protection against other effects at higher doses. The RfD is expressed in units of mg/kg (body
weight) -day. - 0
DEQ-CFW 00001433

PFOA IMAC Proposal
November 7, 2006
Supporting Documentation
Epidemiological studies in workers have not seen an increase in cancer from exposure to PFOA.
Two animal carcinogenicity studies have been carried out. One study reported increases in Leydig
cell adenomas and mammary gland fibroadenomas. The second study reported increases in tumors
in the liver, Leydig cells, and pancreas. EPA's draft risk assessment on PFOA (EPA, 2005a)
concluded "overall, based on no adequate human studies and uncertain human relevance of the
tumor triad (liver, Leydig cell and pancreatic cell tumors) from the rat studies, PFOA may be best
described as "suggestive evidence of carcinogenicity, but not sufficient to assess human
carcinogenic potential" under the draft 1999 Guidelines for Carcinogen Risk Assessment."
EPA's Science Advisory Board (EPA SAB) submitted a report on January 20, 2006 on EPA's draft
risk assessment. In this report, they stated that the majority of panel members concluded that the
experimental weight of evidence with respect to the carcinogenicity of PFOA was stronger than
proposed in the draft document, and suggested that PFOA is a "likely" carcinogen in humans. This
was based on the following:
• While human data is ambiguous, two animal studies have shown carcinogenic effects at
several sites.
• There exist too many uncertainties in the mode of action for liver tumors to say that they
are not relevant to humans.
• Mammary gland adenocarcinomas seen in the animal study should be considered related to
PFOA treatment.
• Insufficient data are available to determine the mode of action for the Leydig cell,
pancreatic, and mammary gland tumors, and thus they must be presumed to be relevant to
humans.
A few members of the EPA SAB did not find the weight of evidence sufficient to support the
"likely" descriptor and agreed with the EPA's conclusion that PFOA showed suggestive evidence of
carcinogenicity. This was based on the opinion that the mode of action for the liver tumors was not
relevant to humans and that the mammary gland tumors were not demonstrated in animals when
compared to historical controls.
In August 2002, the West Virginia Department of Environmental Protection issued its "Final
Ammonium Perfluorooctanoate (C8) Assessment of Toxicity Team (CATT) Report". This report
was the result of a consent order between the State of WV and DuPont. The consent order indicated
that a scientific team was to 1) determine risk -based human health protective screening levels of C8
in air, water, and soil; 2) provide health risk information to the public, and 3) determine an
ecological health protective screening level for C8 in surface water.
The following are the key reasons why the West Virginia Assessment of Toxicity Team for PFOA
(CATT) did not conclude that the animal carcinogenicity data indicates that PFOA is probably
carcinogenic in humans (WV DEP, 2002):
• PFOA has not been shown to be genotoxic. Genotoxic compounds bind to DNA and are
more likely to be carcinogenic than nongenotoxic compounds. Although nongenotoxic
compounds can be carcinogenic, they are usually much weaker carcinogens that pose a
lesser risk to human health.
Page 2 of 7
DEQ-CFW 00001434

PFOA IMAC Proposal
November 7, 2006
Supporting Documentation
• The liver tumors seen in the animals were probably caused by a mechanism of action that is
not relevant in humans.
• The Leydig cell tumors are rarely seen in humans and the mechanism of action appears to
be non -linear.
• Technical questions on whether the mammary gland adenocarcinomas and pancreatic
tumors were related to the PFOA treatment.
The EPA has not calculated a cancer slope factor for PFOA. The West Virginia CATT concluded
that they would base their water and soil screening levels on the noncarcinogenic endpoints of
PFOA (the RM). They stated that they believed that the RfD would be protective against the
possibility of liver tumors and Leydig cell tumors, since these tumors, if shown to be relevant to
humans, would operate under a nongenotoxic, non -linear mechanism (WV DEP, 2002). The EPA
Guidelines for Carcinogen Risk Assessment state that, "For cases where the tumors arise through a
nonlinear mode of action, an oral Reference Dose or inhalation Reference Concentration should be
developed in accordance with EPA's established practice. This approach expands the past focus of
such reference values (previously reserved for effects other than cancer) to include carcinogenic
effects determined to have a nonlinear mode of action" (EPA, 2005b).
As noted above, without adequate information to support a calculation of a Cancer Slope Factor,
the IMAC for PFOA has been calculated with an RfD of 0.003 mg/kg-day considering non -
carcinogenic effects. Dr. Luanne Williams, a toxicologist with the North Carolina Department of
Health and Human Services, has indicated some uncertainty in using the chronic oral reference dose
of 0.0003 mg/kg-day for calculating an IMAC for PFOA. Indicating that there is uncertainty
associated with the reference dose of 0.0003 mg/kg-day because of the serious critical data gaps in
the carcinogenicity and toxicity of PFOA (Williams L, 2006).
4. RfD Development
An RfD is calculated as follows:
• Review available human and animal studies on the chemical
Weigh studies for applicability to be used as the critical study for RfD determination. Some
of the determining factors are:
o Length of study
o Number of animals used
o Endpoints examined
o Relevance of route of exposure
o Quality of study (follows EPA or other guidelines)
o Exposure levels defined (often a problem in epidemiology studies)
o Critical effect determined
• Determine critical study
• Determine highest dose level at which a critical adverse effect does not occur (NOAEL) or
the lowest dose level at which a critical adverse effect does occur (LOAEL) from the critical
study
• Determine appropriate uncertainty factor (UF) to be applied to the NOAEL or LOAEL
• Divide NOAEL or LOAEL by UF.
Page 3 of 7
DEQ-CFW 00001435

PFOA MAC Proposal
November 7, 2006
Supporting Documentation
Table 1 presents a summary of the studies that were considered for RM derivation. The human
studies were determined to be inadequate for RM determination (all studies reviewed and -
summarized in EPA, 2005a and West Virginia DEP, 2002).
Tahle 1 _ Studies Considered for RfD Derivation
Study
Animal (sex)
Length
Doses
NOAEL
LOAEL
Effects
(route)
(mg/kg-day)
(mg/kg-
(mg/kg-
for APFO
day)
day)
and PFOA
as indicated
Thomford et
Cynomolgus
26 weeks
0, 3, 10,30
ND
3
Increased
al, 2001
monkey (M)
(oral capsule)
APFO
liver weights,
possible
mortality
Goldenthal,
Rhesus
13 weeks
0, 3, 10, 30,
ND
3
Clinical signs
1978b
monkey
(gavage)
100 PFOA
(M,F
Goldenthal,
CD rats
13 weeks
0, 0.56, 1.72,
1978a
(M)
(diet)
5.64, 17.9,
0.56
1.72
Increased
63.5 PFOA
liver weights
(F)
0, 0.74, 2.3,
7.7, 22.4,
22.4
76.5
Increased
76.5 PFOA
liver weights
Palazzolo,
ChR-CD rats
13 weeks
0, 0.06, 0.64,
0.06
0.64
Increased
1993; Perkins
(M)
(diet)
1.94, 6.50
liver weights,
et al., 2004
APFO
liver
hypertrophy
Sibinski,1987
Sprague-
2 years (diet)
Dawley rats
(M)
0, 1.3, 14.2
1.3
14.2
Increased
liver weights,
1.6, 16.1
liver
(F)
APFO
1.6
16.1
hypertrophy
Decreased
body
weights,
effects on
blood
Cook et al.,
Sprague
2 years (diet)
0, 14.2 NR*
ND
14.2
Increased
1994
Dawley rats
liver weights
(M,F)
Riker
Rats
2 years (diet)
Laboratories,
(M)
0, 1.3, 14
1.3
14
Increased
1983
0, 1.6, 16
liver weights
(F)
NR*
ND
1.6
Ovarian
hyperplasia
York et al.,
Sprague-
2-generation
0, 1, 3, 10, 30
ND
1
Increased
2002;
Dawley rats
reproductive
APFO
liver weight
Page 4 of 7
DEQ-CFW 00001436

PFOA MAC Proposal
November 7, 2006
C-"t%nvf;na llnnnm(-.ntntinn
Butenhoff et
(M, F)
study
**
al., 2004
(gavage)
ND
1
Significant
reduction in
mean body
weight gain .
***
ND
1
Decreased
body weight
and body
weight gain
Gortner, 1981
Sprague
GD 6-15
0, 0.05, 1.5,
150
ND
No effects
Dawley rats
5,150 APFO
(M, F)
Staples, 1984
Sprague-
GD 6-15
0, 100 APFO
100
ND
No effects
Dawley rats
(M,F)
Gortner, 1982
New Zealand
GD 6-18
0, 1.5, 5, 50
5
50
Development
white rabbits
APFO
al effects,
such as extra
rib
* NR = compound (APFO or PFOA) not reported
**Effect noted in Fo parental males (shown on page 72 of EPA 2005a). However, a mode of action analysis has demonstrated
that the liver effects on rats are due to a peroxisome proliferator-activated receptor alpha or PPAR a—agonism. According to
EPA, this mode of action is unlikely to occur in humans (shown on page 8 of EPA 2005a).
***Effect noted in F, male pups (pups from Fo adult rats) (shown on page 68 of EPA 2005a).
****Effect noted in Fl adult males (pups from Fo adult rats) (shown on page 73 of EPA 2005a).
GD = gestational day
ND = not determined since effects were seen at all doses tested
APFO = Ammonium perfluorooctanoate
PFOA = Perfluorooctanoic acid
York et al., 2002; Butenhoff et al., 2004 was selected as the critical study for the derivation of the
RfD. The reasons for this include the fact that the study:
• Is of excellent quality
• Follows EPA OPPTS guidelines for conducting reproductive/developmental studies
• Examined for multiple organ effects as well as developmental effects from two rat
generations.
• Presents the lowest LOAEL of all the chronic and developmental studies (the only study
with a lower LOAEL (0.64 mg/kg-day) Palazzolo, 1993; Perkins et al. 2004 is only 13
weeks duration).
Table 2 presents the study and factors used to calculate the RfD for PFOA:
Tnhh, 7_ RM fnr PFOA
Study v
Critical
NOAEL
LOAEL
OF
RfD
Effect
Page 5 of 7
DEQ-CFW 00001437

PFOA MAC Proposal
November 7, 2006
Supporting Documentation
York et al.
Reduction in
NOAEL was
1 mg/kg-
3,000
0.0003
2002;
mean body
not
day APFO
mg/kg-day*
Butenhoff et
weight and
determined
al. 2004
body weight
since effects
gain from Fl
were seen at
male rat pups
all doses
and Ft male
tested.
adults
*According to Dr. Luanne Williams, a toxicologist with the North Carolina Department of Health and Human Services, the 1 mg/kg-
day LOAEL for APFO could be used with some caution to derive a chronic oral reference dose of 0.0003 mg/kg-day for calculating
groundwater and soil screening levels for PFOA. There is uncertainty associated with the reference dose of 0.0003 mg/kg-day
because of the serious critical data gaps in the carcinogenicity and toxicity of PFOA (Williams L, 2006).
A total uncertainty factor (UF) of 3,000 (UF = UFH (10) x UFA (10) x UFs (1) x UFL (10) x UFD (3)
= 3,000) was used, consisting of the following areas of uncertainty:
1. Intraspecies variability (UFH). This factor accounts for the natural differences that occur
between human subpopulations and for the fact that some individuals may be more sensitive
than the average population. EPA recommends values of 3-10 for this factor. UFH = 10
because have not defined the most sensitive subpopulation for C8.
2. Interspecies variability (UFA). This factor is used to account for differences in response
between animals and humans. EPA recommends values of 1-10 for this factor. UFA = 10
because no data available on quantitative differences between animals and humans in
pharmacokinetics of C8.
3. Subchronic to Chronic Extrapolation (UFs). This factor is applied when the database lacks
information on the health effects of the chemical following lifetime exposure. EPA
recommends values of 1-10 for this factor. UFs = 1 because many chronic studies available
on C8.
3. LOAEL to NOAEL Extrapolation (UFL). This factor is applied when extrapolating from a
LOAEL to a NOAEL. UFL = 10 since a LOAEL (reduction in mean body weight gain) was
used in the calculations.
4. Database (UFD). This factor is applied when there are significant data gaps on the chemical.
EPA recommends values of 1-10 for this factor. UFD = 3 since there are database gaps on
the toxicity of PFOA.
5. IMAC Calculation
A ground water quality standard is derived from the multiplication of the RfD by the assumed body
weight of an adult (70 kg) and divided by the assumed daily water consumption (2L) of an adult.
This value is then multiplied by a relative source contribution to take into account exposures from
other sources i.e., food, air etc. (RSC). The Groundwater regulations establish the RSC at 20% for
organic chemicals and 10% for inorganic chemicals. The following equation is used (NC, 2005a):
IMAC for PFOA = RfD x BW x RSC
DI
Where: Rf) = RM PFOA = 0.0003 mg/kg-day
BW = Body weight of an adult, default = 70 kg
Page 6 of 7
DEQ-CFW 00001438

PFOA IMAC Proposal
November 7, 2006
Supporting Documentation
RSC = Relative source contribution, default = 20% for organics
DI = Daily water intake for an adult, default = 2 L/day
Interim Maximum Allowable Concentration for PFOA
= 0.0003 mg/kg-day x 70 kg x 0.20 = 0.002 mg/L = 2 µg/L
2 L/day
6. References
EPA, 2005a. Draft Risk Assessment of the Potential Human Health Effects Associated with
Exposure to Perfluorooctanoic acid and its Salts. Office of Pollution Prevention and Toxics, Risk
Assessment Division. SAB Review Draft. Available at:
http://www.gpa.gov/opptipir/pfog/pfoarisk.htm
EPA, 2005b. Guidelines for Carcinogen Risk Assessment. Risk Assessment Forum, Washington,
DC. EPA/630/P-03/001F.
EPA, 2004. Users Guide and Background Technical Document for USEPA Region 9's Preliminary
Remediation Goals (PRG) Table.
North Carolina Department of Environment and Natural Resources, 2005a. Classifications and
Water Quality Standards Applicable to the Groundwaters of North Carolina. Subchapter 2L.
Division of Water Quality.
North Carolina Department of Environment and Natural Resources, 2005b. Guidelines for
Establishing Remediation Goals at RCRA Hazardous Waste Sites. Division of Waste Management,
Hazardous Waste Section.
Prevedouros K, Cousins, IT, Buck, RC, Korzeniowski, SH. Sources, Fate and Transport of
Perfluorocarboxylates. 2006. Environ Sci Technol. 40(1):32-44.
West Virginia Department of Environmental Protection, 2002. Final Ammonium
Perfluorooctanoate (C8) Assessment of Toxicity Team (CATT) Report. Available at:
ht�://www.dgp.state.wv,us/item.efin?ssid=l l&sslid=665
Williams L, 2006. Dr. Luanne K. Williams, Toxicologist with the North Carolina Department of
Health and Human Services.
Page 7 of 7
DEQ-CFW 00001439

Perfluorooctanoic Acid (C8) Risk Assessment
Basic Information
The chemical name for C8 is Perfluorooctanoic acid (PFOA). It has a CAS number of 335-67-1 and a
molecular formula of C8HF15O2. One of its salts, ammonium perfluorooctanoate (APFO;
C8F15O2NH4; CAS No.3825-26-1) is the compound that is the most widely used in industry and most
of the animal toxicology studies have been done with this compound. Once absorbed in the body,
APFO disassociates to the PFOA anion.
C8 is a completely fluorinated organic acid, meaning that fluorine atoms completely replace the
hydrogen atoms that are typically attached to organic hydrocarbon molecules. The typical structure
has a linear chain of 8 carbon atoms.
C8 is one of many perfluorinated compounds, in the group of chemicals: the perfluorocarboxylates
or known more simply as perfluorochemicals (PFCs). The PFCs are inherently stable, nonreactive,
and resistant to degradation, due to the very high strength of the carbon -fluorine bond.
Perfluorooctyl sulfonate (PFOS) is another chemical in the PFCs that has received a lot of attention.
This chemical was discovered to be widespread in the blood of the general population in the late
1990s.
BackLyround
Production of PFCs began in 1947 using an electrochemical fluorination process. Early uses of the
PFCs were based on the chemical's stability, surface -tension lowering properties, and ability to form
stable foams. Some of the uses included fire -fighting foams, metal plating and cleaning, coating
formulations, and polyurethane production.
C8 was produced in several States from 1947 on, with its major use as a processing agent in the
making of products that resist heat, oil, stains, grease, and water. Some common uses included the
production of nonstick cookware, stain -resistant repellants, such as Scotch -Gard, surfactants, fire
retardants, and fire -fighting foams.
C8 was produced by 3M at its Cottage Grove plant in Minnesota from the late 1940s until 2002, by
DuPont near Parkersburg, West Virginia also from the late 1940s until 2003, and is still being made
by DuPont near Fayetteville, North Carolina. The North Carolina facility is the only plant in the U.S.
currently manufacturing C8.
In 2001, a class action lawsuit was filed by Ohio and West Virginia residents whose drinking water
was contaminated with C8. Later in that year, the West Virginia Department of Environmental
Protection and DuPont signed a consent order concerning the presence of C8 in the Lubeck, WV
public water supply which is near the DuPont facility near Parkersburg, W. The consent order
established two scientific teams that were tasked with 1) determining the extent and concentration of
C8 in both groundwater and surface water and 2) investigating the toxicity of C8 and developing
provisional risk factors and health protective screening levels.
DEQ-CFW 00001440

Perfluorooctanoic Risk Assessment
November 7, 2006
In March, 2002 EPA Regions 3 and 5 signed a consent order with DuPont requiring the provision of
alternative water to any resident of West Virginia or Ohio with C8 in drinking water at levels above
14 ppb. The 14 ppb was an interim value that was in effect until the water screening level was
established under the consent order described in the above paragraph. The 14 ppb was taken from a
report by Environ Corp. (a consulting firm hired by DuPont).
In December 2004, EPA signed a memorandum of understanding (MOU) with 3M and Dyneon LLC
for monitoring in the vicinity of a fluoropolymer manufacturing facility in Dacatur, Alabama. In
November 2005, EPA signed a similar MOU concerning monitoring at its plant in Parkersburg, W.
In 2005, the class action lawsuit was settled, with DuPont agreeing to pay $107 million and agreeing
to provide a state-of—the-art water treatment system designed to reduce the level of C8 in the water
supply and to provide medical monitoring.
In 2003, the Environmental Working Group (EWG) petitioned the EPA to investigate evidence that
DuPont violated TSCA 8(e) that requires immediate reporting of findings that show a "substantial
risk of injury to health or the environment." The EWG had discovered studies that had been withheld
by DuPont for decades regarding developmental effects of C8 in laboratory animals.
In December 2005, DuPont agreed to pay $10.25 million in civil fines and fund $6.25 million in
research projects to settle allegations that it failed to report environmental contamination and high
levels of C8 found in blood samples from residents near its Parkersburg, WA plant, under TSCA and
RCRA. The EPA said that this was the largest administrative civil fine ever obtained by EPA under
any federal environmental statute. One of the research projects ($5 million) that DuPont is funding is
designed to investigate the potential of 9 of DuPont's PFCs to breakdown to form C8. The second
project will foster science laboratory curriculum changes to reduce risks posed by chemicals in
schools ($1.25 million).
On January 25, 2006, EPA announced a global stewardship program on C8 and related chemicals.
Eight U.S. companies were asked to commit to reduce C8 and related chemicals from facility
emissions and product content by 95% no later than 2010, and to work toward eliminating C8 and
related chemical content from emissions and product content no later than 2015. DuPont and several
other chemical companies have agreed to participate in the program.
On Oct 17,2006, Dupont epidemiologists released Phase 2 results of a study that found no increased
mortality risk in workers exposed to PFOA. The results showed lower mortality rates than those
found in both West Virginia and the U.S. general population. The study was examined by a review
board, composed of independent experts from Georgetown University, Harvard University, Johns
Hopkins University, the University of Massachusetts — Lowell, the University of Washington and
Yale University. The study examined the occupations of 6,027 people who had worked at Dupont's
Washington Works plant between 1948 until the end of 2002. This study included a more detailed
analysis of heart disease and determined a slight increase in heart disease but, no overall increase in
heart disease deaths. The study found no increase in overall mortality from cancer. Prostate cancer
rates among the cases studied were found to be lower than rates in three reference populations. This
contrasts with a previous non -DuPont study where an increase in prostate cancer was reported.
Across the entire study population, there was a slight, but not statistically significant, increase in the
rate of kidney cancer mortality. Most of the cases showed little exposure to PFOA, but the numbers
Page 2 of 12
DEQ-CFW 00001441

Perfluorooctanoic Risk Assessment
November 7, 2006
were too small to draw any conclusions. The entire study population showed diabetes death rates
lower than those found in West Virginia or the United States
U.S. EPA Risk Assessment
In January 2005, the EPA's Office of Pollution Prevention and Toxics (OPPT) issued its "Draft Risk
Assessment of the Potential Human Health Effects Associated with Exposure to Perfluorooctanoic
Acid and its Salts."
Margin of Exposure
The risk assessment used a margin of exposure approach to describe the potential for human health
effects associated with C8. The MOE was calculated as the ratio of the NOAEL or LOAEL from a
specific endpoint to the estimated human exposure level. The MOE does not provide an estimate of
population risk, but describes the relative "distance" between the exposure level and the NOAEL and
LOAEL.
In this risk assessment, since there was no information available on sources or pathways of human
exposure, but there was information available on serum levels of C8 from many human
biomonitoring studies and from many of the animal toxicology studies, internal doses from animal
and human studies were compared, instead of the traditional approach which uses external exposure
estimates.
A variety of endpoints from animal toxicology studies were used to calculate MOEs. The endpoints
consisted of different species, gender, and life stages. For adults, 2 sets of MOEs were calculated as
follows:
• Based on a cynomolgus monkey study based on increased liver weight and possible
mortality, using the geometric mean for the human serum level = 16,739 (8,191 for the 90th
percentile).
• Based on rat study based on reductions in body weight; for males = 9,158 (4,481 for the 90th
percentile); for females = 398 (195 for the 90th percentile).
For developmental effects, MOEs were calculated from a 2-generation reproductive toxicity study in
rats, as follows:
• For the prenatal period, for the pregnant human female; since information on the serum levels
of C8 were not available for this study, calculated an MOE based on the maximum
concentration in serum, Cm,,x = 3,095 (1,548 for the 90th percentile); an MOE based on the
Area under Concentration curve in plasma (AUC) = 823 (412 for the 90th percentile).
For the postweaning period, MOEs were calculated for several endpoints including reductions in
body weight, mortality, and delayed sexual maturation.
• Based on the geometric mean for children, ranges from 10,484 — 78,546 (6,044 — 45,279 for
the 901h percentile).
Carcinogenicity
The risk assessment examined the mode of action and summary of weight of evidence of the
carcinogenicity for C8. All of the epidemiologic data on C8 were occupational studies, most of which
Page 3 of 12
DEQ-CFW 00001442

Perfluorooctanoic Risk Assessment
November 7, 2006
were routine biomonitoring efforts conducted by 3M. With the exception of one study, all of the
studies were cross -sectional and mostly analyzed males.
The only significantly increased cancer incidence found in these studies was an increase in prostate
cancer for workers in the Chemical Division at 3M in Cottage Grove, MN. However, an update of
this study found no significantly elevated cancer rates.
There are 2 dietary carcinogenicity studies on C8 in rats. In the first study, 50 male and 50 female
Sprague-Dawley rats were fed diets containing 1.3 and 14.2 mg/kg-day (males) or 1.6 and 16.1
mg/kg-day (females) for 2 years. There was a significant increase in the incidence of testicular
(Leydig) cell adenomas in the high dose males and a significant increase in the incidence of
mammary fibroadenomas in both groups of females. However, the investigators did not consider
mammary fibroadenomas to be treatment related on the basis of the historical control incidence.
In the rd study, male Sprague Dawley rats were exposed to 300 ppm for 2 years. There was a
significant increase in the incidence of Leydig cell adenomas in the treated rats as compared to
controls and the treated rats had an increase in the incidence of liver adenomas and pancreatic acinar
cell tumors.
EPA concluded that 2 carcinogenicity studies have shown that C8 induced liver adenomas, Leydig
cell adenomas, and pancreatic acinar cell tumors in male Sprague-Dawley rats. The evidence for
mammary fibroadenomas in the female rats was considered to be equivocal since the incidences were
comparable to some background incidences.
Mode of Action and Weight of Evidence for Carcinogenicity
C8 has not been shown to be genotoxic. Strong evidence exists that the liver toxicity and liver
adenomas seen in rats following exposure to C8 result from peroxisome proliferator-activated
receptor (PPAR)a-agonist mode of action. This mode of action involves 4 steps (see Figure 1):
1. - Activation of PPARa (which regulates the transcription of genes involved in peroxisome
proliferation, cell cycle control, apoptosis, and lipid metabolism).
2. This leads to an increase in cell proliferation and a decrease in apoptosis.
3. This further leads to preneoplastic cells and further clonal expansion.
4. Formation of liver tumors.
Only PPARa activation is highly specific for this mode of action. There are also several associative
events that are markers of PPARa agonism but are not directly involved in the etiology of liver
tumors: Peroxisome proliferation and peroxisomal gene expression.
OPPTS issued guidance in 2003 to help establish that a chemical is exhibiting liver toxicity and
tumors via a PPARa agonist mode of action: in vitro evidence of PPARa agonism, in vivo evidence
of an increase in the number and size of peroxisomes, increase in the activity of acyl CoA oxidase,
and hepatic cell proliferation.
There is sufficient evidence to demonstrate the key events for a PPARa agonist mode of action
following exposure to C8 in rats.
Several scientific groups have examined the role of PPARa-induced rodent liver tumors and its
relevance for human carcinogenesis. In 1995, a workgroup convened by IARC concluded that the
Page 4 of 12
DEQ-CFW 00001443

Perfluorooctanoic Risk Assessment
November 7, 2006
mode of action induced in rodents by PPARa agonists is unlikely to be operative in humans. A
workshop held by the International Life Sciences Institute (ILSI) Health and Environmental Sciences
Institute concluded that although it appeared unlikely that PPARa agonists could induce liver
tumorigenesis in humans, the possibility could not be ruled out. An analysis by Klaunig et al. (2003)
concluded that:
1. The weight of evidence in linking PPARa to the mode of carcinogenic action of C8 is high
for the liver,
2. The PPARa mode of action is plausible in humans since the PPARa is present in humans
and human livers possess PPARa at sufficient levels to mediate the hypolipidaemic
response to therapeutic fibrate drugs, many of which are PPARa agonists,
3. The weight of evidence, however suggests that this mode of action is unlikely to occur in
humans based on quantitative differences in several of the key factors, such as human livers
have been found to have 10-fold less mRNA for PPARa compared with rodents.
In 2004, the majority of a FIFRA Science Advisory Panel agreed that that are relevant data indicating
that humans are less sensitive than rodents to the hepatic effects of PPARa agonists and stated that
when liver tumors are established in long term studies in rats and mice and:
1. the data are sufficient to establish that the liver tumors are a result of PPARa agonist mode of
action, and
2. other potential modes of action have been evaluated and found not to be operative,
Then the evidence for liver tumor formation in rodents should not be used to characterize potential
human liver cancer risk.
On May 30, 2006, the SAB issued their report reviewing EPA's risk assessment of C8. One of the
issues they examined was the weight of evidence of carcinogenicity and the adequacy of the data for
PPARa agonist induced rodent liver toxicity. The SAB concluded:
• Evidence to date was consistent with an interpretation that liver tumor induction likely results
from a PPARa-agonist mode of action. However, the panel members did not agree on whether
PPARa is the sole mode of action for rodent liver tumor induction and toxicity. Most panel
members felt, based on current evidence, that it is possible that PPARa agonism may not be
the sole mode of action for C8, that not all steps in the pathway of PPARa activation -induced
liver tumors have been demonstrated, that other hepatoproliferative lesions require
clarification, and that extrapolation of the mode of action across the age range in humans is
not supported.
Regarding the weight of evidence of carcinogenicity, the risk assessment reached the conclusion that
C8 exhibited "suggestive" evidence of carcinogenicity but not sufficient to assess the carcinogenic
potential of C8. This was based on the following:
1. A PPARa-agonist mode of action for liver tumors in rodents that was not considered
relevant to humans because of their decreased sensitivity to PPARa-agonism when
compared to rodents,
2. The absence of hepatic cell proliferation in a 6-month study of C8 in cynomolgous
monkeys, the species considered closest in physiology to humans,
3. The absence of a strong association between C8 exposure and tumors in human studies,
Page 5 of 12
DEQ-CFW 00001444

Perfluorooctanoic Risk Assessment
November 7, 2006
4. The belief that the Leydig cell tumors and pancreatic acinar cell tumors produced by C8 in
rats were probably not relevant to humans based on lower levels of certain hormones in
humans and differences in quantitative toxicodynamics between rats and humans,
5. The belief that the mammary fibroadenomas reported in female rats are equivocal based on
their comparable rates of occurrence relative to a historical control group.
Most panel members of the SAB concluded that the experimental weight of evidence with respect to
the carcinogenicity of C8 was stronger than proposed in the draft document and suggested that C8
should be considered "likely to be carcinogenic in humans." This was based on the following:
• While human data are ambiguous, 2 separate feeding studies in rats have shown
cancer in several sites,
• Uncertainties still exist as to whether PPARa agonism is the sole mode of action for
C8 effects on the liver,
• It was inappropriate to exclude mammary tumors in the document based on
comparisons to historical control levels for other laboratories, since the most
appropriate control group is a concurrent control group. Using that comparison,
there was an increase in mammary gland tumors,
• Insufficient data are available to determine the mode of action for the Leydig cell
tumors, pancreatic acinar cell tumors, and mammary gland tumors. In the absence of
a defined mode of action, they must be presumed relevant to humans.
These panel members were not willing to state an associated probability for C8-induced
carcinogenicity, but most panel members believed that risk assessments for each of the C8-induced
tumors are appropriate at the current time.
A few panel members disagreed and stated that currently available evidence does not exceed the
"suggestive" evidence of carcinogenicity descriptor, based on the belief that PPARa does serve as the
sole mode of action for C8-induced rodent liver tumors and that mammary tumors were not
demonstrated in animals compared to historical controls.
West Virginia Risk Assessment
In August 2002, the WV Department of Environmental Protection issued its "Final Ammonium
Perfluorooctanoate (C8) Assessment of Toxicity Team (CATT) Report". This report was the result of
the consent order between the State of WV and DuPont that said that a scientific team was to 1)
determine risk -based human health protective screening levels of C8 in air, water, and soil; 2) provide
health risk information to the public, and 3) determine an ecological health protective screening level
for C8 in surface water.
Noncancer Risk Assessment
The CATT report developed screening levels for C8 in air, water, and soil. The panel agreed that the
human studies were not adequate to be used for quantitative dose -response determinations. They
reviewed the relevant animal studies and determined that the primary target organ for C8 is the liver.
They selected a 2-generation rat study as the critical study and selected the following:
Key study I Species, I NOAEL LOAEL I BMDL I Critical I OF
Page 6 of 12
DEQ-CFW 00001445

Perfluorooctanoic Risk Assessment
November 7, 2006
sex
effect
York et al.
2002
Rat, male
None
1
0.42
Liver
100
The uncertainty factor of 100 was selected based on an individual factor of 10 for human variability
and another factor of 10 for interspecies variability.
The Rf) = 0.42/100 = 0.0042 mg/kg-day. The value was rounded to 0.004 mg/kg-day.
Cancer Risk Assessment
Liver tumors: The majority of the panel members agreed that peroxisome proliferation is the mode of
action for the liver tumors. Some of the panel members felt that the use of the liver tumor response
data for human health risk assessment cannot be totally discounted, while others felt that the rodent
liver tumors are not relevant at all to humans.
Leydig cell tumors: The panel agreed that these tumors were likely to be caused by a non-genotoxic
mechanism. They also agreed that the Leydig cell tumors are a known tumor type for other
peroxisome proliferators; however they could not reach a conclusion as to whether this tumor type is
relevant to humans.
Pancreatic tumors: The panel agreed that the evidence was not sufficient to demonstrate the mode of
action for pancreatic tumors, but enhanced cell proliferation (hyperplasia) was likely to be involved.
The mode of action appears to be nongenotoxic.
Mammary fibroadenomas: The panel agreed that the data were not adequate to demonstrate a mode
of action. Several panelists were not convinced the data demonstrated a real carcinogenic effect.
Dose -Response Assessment — The panel members agreed to recommend a dose -response approach
for each tumor type.
Liver tumors: MOE approach. Since the MOE analysis often uses the benchmark response for a
precursor as the basis for deriving a point of departure, the panel judged the Rf) for liver effects as
sufficiently protective for potential liver carcinogenicity.
Leydig cell tumors: Benchmark dose modeling was conducted. The point of departure for Leydig cell
tumors was chosen by the panel from the BMD modeling output. A BMDL of 0.32 mg/kg-day was
selected as the most appropriate basis for deriving the assessment. An oral cancer slope factor was
calculated as follows:
Slope factor = risk/dose = 0.1/0.32 = 0.31 per mg/kg-day.
Risk was expressed as 0.1 because the BMDL is the point that represents a 10% increase in tumor
incidence in accordance with EPA guidance.
ScreeningLevels
evels
Page 7of12
DEQ-CFW 00001446

Perfluorooctanoic Risk Assessment
November 7, 2006
The panel agreed that the oral RfD for liver toxicity would be the basis for determining the water and
soil screening levels for the following reasons:
• High confidence in the RfI)
• The R.fl) would be protective against the quantitatively less sensitive and questionably
relevant peroxisome proliferation -related liver cancer
• Low confidence in the Leydig tumor analysis and questionable relevance to humans
• Limitations in study design, data quality, and data interpretation made it difficult to
determine whether the increased incidence of pancreatic tumors or mammary tumors were
related to C8 treatment and did not allow for the modeling of a point of departure to be
used for quantitative risk assessment.
The following equations were used to derive screening levels (from EPA Region 9 guidance on
deriving risk based concentrations):
Air Screening Level = µg/m3 = THQ x RfDi x BW x AT x 1000
EF x ED x air IR
With RfDi (mg/kg-day) = RfC x 20 m3/dgy (IR)
70 kg (BW)
Soil Screening Level = mg/kg = THQ x AT x BW
EF x ED x [soil IR/RfD x 10-6 + SA x AF x ABS/RfD x 10-6]
Water Screening Level = µg/L = THQ x AT x BW x 1000
EF x ED x [water IR/RfD]
Where:
THQ = Target hazard Quotient, assumed to be 1
RfDi = The RfC expressed in terms of dose, mg/kg-day
The oral reference dose estimated by the panel, 0.004 mg/kg-day
RfC = The inhalation RfC estimated by the panel
BW = Body weight, assumed to be 70 kg for adults and 15 kg for children
AT = Averaging time, 10950 days, the exposure duration expressed in days
EF = Exposure frequency, 350 days/year, the average number of years people are exposed
ED = Exposure duration, 30 years, the average number of years people are exposed
IR = Inhalation rate for air screening levels, 20 m3/day; Ingestion rate for soil and, Water
Screening levels, 200 mg/day soil ingested based on child exposure and, 2 L/day ingested based on
adult exposure
SA = Surface area of exposed skin, 2800 cm2/day
AF = Adherence factor, 0.2 mg/cm2, the amount of soil that adheres to skin
ABS = Skin absorption factor, specific factor not available for C8, assumed to be 0.1 for
semi -volatile chemical per EPA guidance.
The following screening levels were calculated:
Air: 0.1— 6.0 µg/m3
Soil: 244 mg/kg residential soil, rounded to 240 mg/kg
Page 8 of 12
DEQ-CFW 00001447

Perfluorooctanoic Risk Assessment
November 7, 2006
Water: 146 µg/L, rounded to 150 µg/L.
Minnesota Number
The state of Minnesota has not carried out a risk assessment for C8. However, in 2002 they calculated
a health based value (HBV) for C8 and residential and industrial soil reference values.
MN calculated a HBV as follows:
Study: Thomford et al., 2001 (26 week study in monkeys)
Critical endpoint: Liver effects
LOAEL = 3 mg/kg-day
OF = 3000 (3 for interspecies variability, 10 for intraspecies variability, and 10 for subchronic to
chronic, and 10 for LOAEL to NOAEL).
RfD = 0.001 mg/kg-day
Relative source contribution = 20%
HBV = RfD (mg/kg-day
) (RSC) 1000 µg/mg)
Intake rate (2 L/day/70 kg)
_ (0.001 mg/kg-day)(0.2)(1000 gg/mg) = 7 µg/L
0.029 L/kg/day
MN calculated the following soil reference values (SRV) for C8:
Residential SRV = 30 mg/kg
Industrial SRV = 200 mg/kg
(The calculations were not provided for these values)
Ohio Number
The state of Ohio has not carried out a risk assessment for C8. However, they have posted a letter on
their website stating that they are relying on the U.S. EPA to study C8 and cited the West Virginia
Preliminary Action Level of 150 ppb. They said that all the detections of C8 in public water supplies
are below this level and that they will take whatever action is necessary to be certain that Ohio is not
distributing drinking water with levels of C8 above any level that EPA establishes.
New Jersey Number
The state of New Jersey has adopted an "Interim Guidance Criteria" for C8. In situations where no
specific criteria exists for a synthetic organic chemical, the New Jersey Ground Water Quality
Standards specify the establishment of an interim guidance criteria of 5 ppb for compounds identified
as having "evidence of carcinogenicity." Since C8 has been shown to demonstrate evidence of
carcinogenicity in rats, New Jersey set their interim guidance at 5 ppb.
ATSDR Health Consultation
In February, 2005, ATSDR published a health consultation for C8 and PFOS at the 3M Cottage
Grove facility in Minnesota.
ATSDR summarized the same health studies that were reviewed in the EPA and West Virginia risk
assessments. They cited the Minnesota HBV and SRVs (see above).
Page 9of12
DEQ-CFW 00001448

Perfluorooctanoic Risk Assessment
November 7, 2006
ATSDR also summarized the available environmental data, stating that PFOA does not
bioconcentrate through the food chain, while PFOS does. Bald eagles from the Midwest showed the
highest levels of PFOS in plasma (up to 2,570 ng/mL) and mink from the Midwest showed the
highest levels in tissue (in liver, up to 3,680 ng/g). Concentrations of C8 in wildlife samples are
typically approximately 10 times lower and are much less widely distributed.
This report did not reach any conclusions, "The potential impacts on public health from
perfluorchemical releases at the 3M Cottage Grove facility cannot be fully assessed at this time,
because there are not sufficient environmental data available regarding PFC impacts from the facility
in soil, groundwater, surface water, sediments, and biota. At this time perfluorochemical releases
from the site represent an indeterminate public health hazard."
C8 and Related Compounds in Blood
Mean C8 Levels in the General Population:
Studies on 3 separate age groups across the U.S. — mean C8 levels = approximately 4-5 ppb C8 in
serum (EPA, 2005).
C8 in Workers:
3M biomonitoring in Decatur, Alabama plant, mean C8 levels, 1997 =1,400 ppb; 1998 =1,540 ppb;
2000 =1,780 ppb; 2002 = 1,497 ppb (EPA, 2005).
3M biomonitoring in Cottage Grove, MN plant, mean C8 levels, 1995 = 6,800 ppb; 1997 = 6,400
ppb; 2000 = 4,510 ppb; 2002 = 4,300 ppb (EPA, 2005).
DuPont biomonitoring in its Washington Works, WV plant, mean C8 levels, 1989-90 = 1,960 ppb;
1995 =1,560 ppb; 2000 =1,530 ppb (EPA, 2005).
DuPont biomonitoring in its Fayetteville, NC plant, mean C8 levels, 2002 = 11 ppb; 2003 = 217 ppb
(DuPont Biennial Report for the Manufacture of APFO letter, October 26, 2004).
Mean PFOS Levels in the General Population:
Mean serum levels of PFOS in the nonoccupational general population in the U.S. = 30-40 ppb in
serum (Olsen et al., 2005).
PFOS in Workers:
Mean serum levels of PFOS in workers have averaged between 500 and 2,000 ppb in serum (Olsen et
al., 2005).
C8 Levels in the Environment
Concentrations in water
U.S.
Analyzed 16 Great Lakes water samples for C8: Concentrations ranged from 27-50 ng/L (Boulanger
et al., 2004).
Page 10 of 12
DEQ-CFW 00001449

Perfluorooctanoic Risk Assessment
November 7, 2006
Monitored for C8 in surface waters in the Great Lakes, and lakes in Michigan, Wisconsin, and
Minnesota. Found C8 in all samples; levels in remote areas ranged from 0.14 to 0.66 ng/L and levels
in urban surface waters ranged from 0.45 to 19 ng/L. The authors concluded that the primary source
of the C8 found in Lake Michigan was most likely from wastewater treatment effluents, and was not
from the air (Simcik and Dorweiler, 2005).
Outside the U.S.
Japan: Most major rivers and several lakes in Japan have been sampled. Most C8 concentrations were
as low as 0.1 ng/L in remote areas and about 2-10 ng/L in urban areas (Prevedouros et al., in press).
Japan: Analyzed rivers and lakes in Japan with C8 concentrations ranging from a mean of 0.97 — 21.5
ng/L. C8 in drinking water in Osaka City measured at 40 ng/L, significantly higher than in other
cities measured (Saito et al., 2004).
The Orient: Analyzed for C8 in coastal seawaters of Hong Kong, The Pearl River Delta, and Korea,
with concentrations of 0.73-5.5, 0.24-16, and 0.24-320 pg/mL, respectively (So et al., 2004).
Concentrations in sediment
U.S. and other countries: Sediment samples from a small U.S. city, the Netherlands, several
Scandanavian countries, the San Francisco Bay, and the Niagara River reported C8 levels of 24 pg/g
to 18 ng/g. Samples with concentrations over 1 ng/g were atypical with the majority of the reported
concentrations in the hundreds pg/g range (Prevedouros et al., in press).
Concentrations in air from precipitation and in snow
U.S.
C8 was detected as high as 50 ng/L, but generally less than 10 ng/L, in precipitation samples from 3
different U.S. sites in 1998 (Prevedouros et al., in press).
Outside the U.S.
Finland and Sweden: C8 found in precipitation with a concentration range of 8-17 ng/L (Prevedouros
et al., in press).
High Arctic: C8 found in snow and ice caps at 2-3 ng/L (Prevedouros et al., in press).
Concentrations in wildlife
U.S.
C8 found in loggerhead turtles (mean conc. = 3.20 ng/mL) and Kemp's ridley turtles (mean conc. _
3.57 ng/mL) in inshore waters of Core Sound, North Carolina and offshore waters of South Carolina,
Georgia, and Florida (Keller et al., 2005).
C8 detected in some samples (PFOS found in all samples) of the livers of mink and river otters
collected across the U.S., with a maximum concentration of 27 ng/g (Kannan et al., 2002a).
Outside the U.S.
South America: C8 found in bile of fish (mullets) in Colombia at concentrations ranging from 1.27 —
713 ng/mL (Olivero-Verbel et al., 2005).
Page 11 of 12
DEQ-CFW 00001450

Perfluorooctanoic Risk Assessment
November 7, 2006
Japan and Korea: C8 detected in 5-10% samples of livers of birds, with a maximum concentration of
21 ng/g (Kannan et al., 2002b).
Baltic and Mediterranean Seas and coast: C8 detected sporadically (PFOS detected in all samples) in
liver and blood of fish, dolphins, seals, whales, and eagles. Highest concentration of 95 g/g reported
(Kannan et al., 2002c).
Concentrations in house dust
Canada: C8 found in house dust in Ottawa (Kubwabo et al., 2005).
Japan: C8 detected in vacuum cleaner dust in all homes sampled at 69 — 3,700 ng/g (Moriwaki et al.,
2003).
Figure 1
Key Events in the Mode of Action for PPARa Agonist Induced Rodent Liver Tumors
Causative Events
PPARa Agonist — Activation of PPARa Associative Events —
y Expression of Peroxisomal genes
Increase in Peroxisomes (number and size)
Cell Proliferation
Decreased Apoptosis
I
Preneoplastic foci
Clonal expansion
Liver tumors
Page 12 of 12
DEQ-CFW 00001451