Potential to Cause Harm to Human Health

PFOA has been detected in a number of locations in Canada: in house dust in Ottawa, Ontario with a mean level of 106 ng/g (n=67; range 2.29 to 1234 ng/g; median 19.72 ng/g - Kubwabo et al. 2005); in tap water in Calgary, Alberta, and Vancouver, British Columbia, at 0.2 ng/L (one sample per location - Lien et al. 2006); in surface waters from US and Canadian locations on Lake Ontario and Lake Erie at levels ranging from 13 to 50 ng/L (mean of 8 samples from 4 locations on Lake Ontario: 42.5ng/L; mean of 8 samples from 4 locations on Lake Erie: 35.6 ng/L - Boulanger et al. 2004; median of 13 samples on Lake Ontario: 21 ng/L; median of 3 samples on Lake Erie: 15 ng/L – Sinclair et al. 2006); in lake water at 5 locations in the Arctic with means of 0.9 to 14 ng/L (range 0.5 to 16 ng/L – Stock et al. 2007); and in indoor air and dryer lint (2009 email from New Substances Assessment and Control Bureau, Health Canada, to Risk Assessment Bureau, Health Canada; unreferenced^[2]). PFOA contamination has also been detected in fish and water at airport sites in Canada where fire-fighting foams have been released (Moody et al. 2002; 2008 email from Transport Canada, to Bureau of Risk and Impact Assessment, Health Canada; unreferenced).

Similarly, FTOHs have been detected in a number of locations in Canada: in house dust in Ottawa, Ontario (n=59; mean 8:2FTOH: 55 ng/g – Shoeib et al. 2005), and in the air over rural and urban sites in Ontario and Manitoba (total FTOHs in Toronto, Ontario: range 113 to 213 pg/m³, mean 165 pg/m³; Winnipeg, Manitoba: range MDL-18 pg/m³, mean 11 pg/m³; Long Point, Ontario: MDL to 52 pg/m³, mean 26 pg/m³; n=3 samples per location -Stock et al. 2004; 8:2 FTOH in Toronto, Ontario: range 9-123 pg/m³, mean 55 pg/m³; n=4; Long Point, Ontario: range 25-40 pg/m³, mean 32 pg/m³; n=2 – Martin et al. 2002) and in the Arctic (total FTOHs: mean 28 pg/m³; range for 8:2 FTOH³ – Stock et al. 2007; 8:2 FTOH: range 5.8 to 26 pg/m³; n=20 – Shoeib et al. 2006).

In 3 of 54 composite food samples collected for the Canadian Total Diet Study from 1992 to 2004, PFOA was detectable, with the highest level quantified (3.6 ng/g) in a sample of microwave popcorn (Tittlemier et al. 2007). PFOA was also found in several species of fish (both raw and cooked samples; maximum level detected 1.59 ng/g) purchased at markets in Ontario in 2006 (Del Gobbo et al. 2008) and in some native foods (seal, duck and caribou; maximum level detected 0.8 ng/g) collected in Nunavut in 1997–1998 (Ostertag et al. 2009).

PFOA was detected in the packaging materials and in the vapours produced from the microwave cooking of both tested brands of prepackaged microwave popcorn, and 8:2 FTOH was found in the packaging and vapours of one of two brands of microwave popcorn. NeitherPFOA nor 8:2 FTOH was detected in the packaging materials or vapours of plain popcorn kernels cooked in a polyethylene bag (Sinclair et al. 2007). In studies that analyzed products purchased at US retail stores,PFOA was detected in new polytetrafluoroethylene-treated cookware following heating to normal cooking temperatures (Begley et al. 2005) and in water boiled in two of four new non-stick pans (Sinclair et al. 2007).PFOA and 8:2 FTOH were detected in off-gassing from all four tested brands of new non-stick cooking pans heated at normal cooking temperatures (Sinclair et al. 2007). PFOA was also detected in cooking oil following heating under normal cooking conditions in new non-stick pans in a study in Italy (Bononi and Tateo 2007). Other studies have reported no detectable PFOA following extraction of new treated cookware (Powley et al. 2005; Washburn et al. 2005; Bradley et al. 2007). Extracts from some fluorochemical-treated articles (apparel, carpet and upholstery) were found to containPFOA (Washburn et al. 2005).

PFOA was detected in 44 of 45 samples of human breast milk from Massachusetts, with concentrations ranging from < 0.0301 to 0.161 µg/mL (Tao et al. 2008), and cord blood of newborns in Canada was found to contain PFOA in three studies (Tittlemeir et al. 2004; Monroy et al. 2008; 2009 email from New Substances Assessment and Control Bureau, Health Canada, to Risk Assessment Bureau, Health Canada; unreferenced), indicating potential exposures in utero and through breast milk.

The available data suggest that Canadians are exposed toPFOA and its precursors present in the environment, including via air, drinking water and food; and from the use of consumer products, such as new non-stick cookware and PFC-treated apparel and materials such as carpets and upholstery. In general, the levels of PFOA observed in Canada in ambient air, food and drinking water are comparable to those measured elsewhere, including the United States, Europe and Asia (Fromme et al. 2009). However, in other countries, industrial sources not present in Canada have led to localized PFOA concentrations in drinking water that are much higher than the levels reported in Canada (e.g., Little Hocking, Ohio, mean concentration of 3.5 µg/L; Emmett et al. 2006a).

A quantitative estimate of exposure to PFOA based on levels in environmental media and through use of consumer products was not derived, as biomonitoring data – which aggregates exposure from all sourcse – was available. However, such estimates of total daily intake of PFOA have been published recently. Trudel et al. (2008) estimated the exposure of the general population in North America to PFOA from air, food, drinking water, dust and consumer products. The exposure estimates for low-, intermediate- and high-exposure scenarios ranged from 1 to 130 ng/kg-bw per day for all age groups. Diet was found to be the major source of exposure for the low- and intermediate-exposure scenarios, whereas consumer products (oral exposure to treated carpet, migration from treated paper into food and inhalation during treatment of clothing) were the major contributors in high-exposure scenarios. This study did not consider contributions from exposure to PFOA precursors. In another study, Fromme et al. (2009) estimated exposure to PFOA for the general population using data from North America, Europe and Japan. The upper bound of exposure was estimated to be 12.6 ng/kg-bw per day and was mainly due to dietary consumption. Exposure to FTOHs contributed less than 1% of the exposure to PFOA. This study did not look specifically at consumer product scenarios.

Biomonitoring data form the basis of this exposure assessment for PFOA, as serum levels represent aggregate exposure from all routes and sources, including exposure to precursors of PFOA. PFOA has been identified in blood samples from non-occupationally exposed populations worldwide, including in adults and newborns in Canada. The widespread detection of PFOA in human blood indicates that humans are environmentally exposed to PFOA and/or precursor compounds that degrade to PFOA. Once in the body, PFOA can bind to certain proteins (Han et al. 2003, 2005), but there is no evidence that it is modified by metabolism, conjugation or defluorination (Vanden Heuvel et al. 1991). PFOA has a relatively long half-life in humans, in the range of 2–13 years (Burris et al. 2002; Olsen et al. 2007).

PFOA has been measured in the blood of Canadian adults and newborns in several studies, at levels ranging from 0.00048 µg/mL up to 0.0072 µg/mL. Although biomonitoring data for Canada are limited, available Canadian data are consistent with data for US populations. For example, in the 2003–2004 National Health and Nutrition Examination Survey (NHANES), the geometric mean PFOA concentration in serum of 2094 Americans aged 12 years or older was 0.0039 µg/mL, with a 95th percentile of 0.0098 µg/mL. Serum PFOA levels in children and the elderly are comparable with those in adults. Table 7 shows the results from Canadian and US biomonitoring studies.

Table 7.PFOA biomonitoring Studies From Canada and the United States

Abbreviations: LOD, limit of detection; LOQ, limit of quantification; NA, not available

Confidence in the measure of exposure (i.e., levels of PFOA in blood) is high, because although the Canadian sampling data are limited, the mean and 95th percentile values are very similar to those reported for samples collected in the United States. Additionally, the use of biomonitoring data accounts for multiple sources of exposure and eliminates the need to calculate upper bound estimates of exposure from environmental sources and consumer products and household materials.

Most of the relevant toxicity studies have been conducted on the ammonium salt of PFOA (APFO); studies are also available on the anion (PFO), the sodium salt (PFOA-Na) and on PFOA itself. PFOA and its salts are expected to dissociate rapidly in biological media to form the PFO anion, and the toxicological effects of exposure to PFOA, APFO and PFOA-Na are similar. Therefore, PFOA and its salts are considered to be biologically equivalent in this report. Toxicological data on individual precursors of PFOA were not reviewed as the approach for this assessment was to consider these compounds in terms of their contribution to exposure to total PFOA. A summary of the toxicological database for PFOA and its salts is provided in Appendix A. The studies considered critical to the screening assessment are those with the lowest administered doses and those that reported the lowest serum levels of PFOA associated with effects.

The lowest LOAEL identified in experimental animal toxicity studies was 0.3 mg/kg-bw per day for APFO in a short-term (14-day) oral toxicity study in rats and mice. This dose was associated with a mean serum PFOA level of 13 µg/mL and increased liver weight in male mice; and a mean serum PFOA level of 20 µg/mL and altered lipid parameters in male rats. This dose was the lowest tested; thus, no NOAEL was obtained (Loveless et al. 2006).

In a subchronic (13-week) oral toxicity study in male rats, no effects, including histopathological changes in the liver, were observed at 0.06 mg/kg-bw per day. At the next dose (0.64 mg/kg-bw per day), increased liver weight and hepatic hypertrophy were observed. These effects were not observed after an 8-week recovery period. Corresponding serum PFOA concentrations at the NOAEL and LOAEL were 7.1 and 41.2 µg/mL, respectively, at the end of the 13-week exposure period (Palazzolo 1993; Perkins et al. 2004).

Liver effects were also observed in a 14-day inhalation study in male rats. Reversible liver weight increase, reversible increases in serum enzyme activities and microscopic liver pathology, including necrosis (not reversible), occurred at exposure levels of 8 mg/m³ and above, but not at 1 mg/m³. These exposures were calculated as equivalent to doses of 2.48 and 0.31 mg/kg-bw per day, respectively^[3] (corresponding serum PFOA levels were 47 and 13 µg/mL, respectively, at the end of the 14-day exposure period) (Haskell Laboratory 1981a; Kennedy et al. 1986).

In a 26-week oral subchronic toxicity study in male monkeys, the lowest tested APFO dose of 3 mg/kg-bw per day resulted in increased liver weight (Thomford 2001b; Butenhoff et al. 2002). The average serum PFOA level based on biweekly measurements was 77 µg/mL.

In a developmental toxicity study in mice, the oral administration of APFO at 1 mg/kg-bw per day on days 1–17 of gestation resulted in increased liver weight in the dams, alterations in fetal ossification and early puberty in male pups. The mean serumPFOA level in the dams at the LOAEL of 1 mg/kg-bw per day was 21.9 µg/mL. As this was the lowest tested dose, no NOAEL was determined (Lau et al. 2006).

Two chronic toxicity studies in laboratory animals were identified. In one, CD rats were given APFO at 0, 30 or 300 ppm in the diet for 2 years (0, 1.3 or 14.2 mg/kg-bw per day for males; 1, 1.6 or 16.1 mg/kg-bw per day for females). A LOAEL of 1.3 mg/kg-bw per day was determined based on significant increases in serum liver parameters (alanine aminotransferase [ALT], alkaline phosphatase [AP] and albumin) in males. No evidence of carcinogenic activity was seen in the females, but the males showed an increase in testicular Leydig cell adenomas. This increase was significant (p = 0.05) in the high-dose group (14.2 mg/kg-bw per day) (Sibinski 1987). In the second chronic toxicity study, male CD rats were given only one dietary dose level of APFO(300 ppm; 13.6 mg/kg-bw per day) for 2 years. The exposed males had significantly higher incidences of adenomas of the liver, hyperplasia and adenomas in Leydig cells and hyperplasia and adenomas in pancreatic acinar cells (Biegel et al. 2001). Serum PFOA levels were not reported in either of the chronic studies.

Testing of PFOA or its ammonium or sodium salts (see Appendix A for specific compound details and references) for genotoxic potential produced no evidence of activity in three in vivo bone marrow micronucleus tests in mice, several Ames bacterial mutation assays and three in vitro chromosomal aberration tests (two in hamster cells and one in human cells). Positive results were obtained in one chromosome damage test in hamster cells and one micronucleus assay in human cells in vitro. PFOA caused oxidative DNA damage in human hepatoma cells in culture and in the liver of rats treated by the oral or intraperitoneal route. The genotoxicity database indicates that PFOA compounds are not mutagenic.

In rodents, PFOA triggers a peroxisome proliferation response, which is mediated by PPARa. Activation of PPAR acauses changes in liver and affects lipid metabolism and transport and other biochemical processes. The triad of benign tumours observed in PFOA-exposed male rats (liver, testicular Leydig cell and pancreatic acinar cell adenomas) is typical of PPAR agonists, including clofibrate, 2,2-dichloro-1,1,1-trifluoroethane (HCFC 123) and pirinixic acid (WY-14,643) (Cook et al. 1999; Kennedy et al. 2004). It has been proposed that the activation of hepatic PPARa, rather than direct genotoxic action, is the critical event in the induction of these tumours. The confidence in the PPARamode of action is high for liver tumours, moderate for Leydig cell tumours (LCTs) and low for pancreatic acinar cell tumours (PACTs) (reviewed in Klaunig et al. 2003).

PFOA has been shown to activate PPARa in vitro and causes hepatic peroxisome proliferationin vivo. Temporal and dose–response data support a PPARa mode of action for rat liver tumours. In the chronic dietary study by Biegel et al. (2001), PFOA caused an increase in liver weight and peroxisome proliferation in the liver of treated rats at every time point studied, starting at 1 month of treatment, and the first occurrence of liver tumours was after 12 months. In a 14-day study in male rats, Liu et al. (1996) determined a NOEL of 0.2 mg/kg-bw per day and a LOEL of 2 mg/kg-bw per day for increased liver weight and hepatic peroxisome proliferation (measured by hepatic ß-oxidation), key endpoints in liver tumour initiation. Evidence of liver carcinogenicity was observed at 13.6 mg/kg-bw per day in the chronic study by Biegel et al. (2001), but not at 1.3 mg/kg-bw per day in the chronic toxicity study by Sibinski (1987), which is in agreement with the dose–response relationship for early key events. In a subchronic study in monkeys, there was no increase in peroxisome proliferation in response to PFOA, although liver effects, including liver weight increases, were observed (Butenhoff et al. 2002).

PFOA did not induce peroxisome proliferation or cell proliferation in male rat Leydig cells (Biegel et al. 2001), suggesting a mechanism for tumour induction other than direct PPAR activation in the testes. It has been proposed that PFOA-induced LCTs in the male rat are the result of an increase in serum estradiol due to hepatic PPARaactivation and subsequent changes in activities of enzymes involved in steroid biosynthesis. Studies have shown that treatment with PFOA causes increased hepatic aromatase activity and serum estradiol levels (Liu et al. 1996; Biegel et al. 2001). In a 14-day study in male rats, Liu et al. (1996) demonstrated that hepatic aromatase activity and serum estradiol were increased only at doses that caused hepatic peroxisome proliferation. A NOEL of 0.2 mg/kg-bw per day and a LOEL of 2 mg/kg-bw per day for increased hepatic aromatase activity and serum estradiol were determined. In the chronic toxicity study by Sibinski (1987), LCTs were increased significantly at the high dose of 14.2 mg/kg-bw per day but not at the low dose of 1.3 mg/kg-bw per day. PFOA caused an increase in serum estradiol in exposed rats at 1, 3, 6, 9 and 12 months, prior to the first occurrence of LCTs, in the chronic dietary study by Biegel et al. (2001). Thus, the proposed PPARa mode of action is consistent with dose–response and temporal associations with increased serum estradiol. Within the testis, estradiol modulates growth factor expression, which could lead to hyperplasia and adenomas (Biegel et al. 1995). PFOA and other PPARa agonists, including bezafibrate, monoethylhexyl phthalate (MEHP) and WY-14,643, have also been shown to inhibit testosterone production in Leydig cells in vitro (Klaunig et al. 2003), although serum testosterone was not decreased in the 2-year rat bioassay with PFOA (Biegel et al. 2001). In non-human primates, treatment with PFOA did not cause an increase in serum estrogen or a decrease in serum testosterone, and no abnormal histopathology was noted in the testes in the 26-week study (Butenhoff et al. 2002).

A hepatic PPAR-mediated pathway may also be involved in the induction of PACTs in male rats, although the confidence in this mode of action is lower than for liver tumours or LCTs (reviewed in Klaunig et al. 2003). In vivo mechanistic evidence is mainly indirect, from other PPAR-agonists that have been demonstrated to cause rat PACTs. Pancreatic acinar cell (PAC) hypertrophy, hyperplasia and adenomas in rat have been shown to be modified by steroid hormones, including estradiol and cholecystokinin. PFOA exposure could cause reduced bile flow or changes in bile composition resulting from downstream effects of hepatic PPARa activation. This may, in turn, cause an increase in cholecystokinin (CCK) levels and stimulation of the PACs . Although no mechanistic data are available for PFOA itself, decreased bile flow and bile acid concentration as well as a sustained increase in plasma CCK occurred in rats treated with the PPARaagonist WY-14,643. These changes were observed as early as 3 months, earlier than the time point at which PACTs were first observed. PFOA-treated rats had increased pancreatic cell proliferation at 15, 18 and 21 months, and acinar hyperplasia was significantly increased in the chronic dietary study by Biegel et al (2001). No abnormal histopathology was noted in the pancreas of PFOA-treated monkeys in the 26-week study (Butenhoff et al. 2002).

In a draft risk assessment, the US EPA stated that “there is strong evidence to conclude that the liver toxicity and liver adenomas that are observed in rats following exposure toPFOA result from a PPARa-agonist mode of action,” which is unlikely to occur in humans. They also conclude that although the LCTs and PACTs may be relevant to humans, they probably do not represent a significant cancer hazard due to quantitative differences in receptor expression and other toxicodynamic factors (US EPA 2005). However, the Science Advisory Board reviewed the US EPA’s risk assessment and concluded that there may be other modes of action for liver tumours and that as the modes of action for LCTs and PACTs are unknown, they should be considered relevant to humans (US EPA 2006b).

Recent studies in mice lacking expression of the PPARa gene (PPARa-null) suggest that some of the effects associated with exposure to PFOA are independent of the peroxisome proliferation pathway. Developmental effects, such as delayed eye opening, deficits in postnatal weight gain and reduced postnatal survival, were observed in wild-type but not PPARa-null mice, suggesting that these effects are dependent on PPARa expression. However, early pregnancy losses were observed in both strains of mice, indicating involvement of pathways other than peroxisome proliferation (Abbott et al. 2007). Yang et al. (2002) noted that reductions in spleen weight and number of splenocytes in wild-type mice treated with PFOA in the diet for 7 days did not occur in treated PPARa-null mice. PFOA-exposed mice also had reduced thymus weight and number of thymocytes, effects that were attenuated in the PPARa-null mice. However, significant liver weight increase occurred in both wild-type and PPARa-null mice. Gene profiling studies also suggest that PFOA can alter mouse liver genes independent of PPARa (Rosen et al. 2008a, b);however, the toxicological relevance of this is not known.

Available epidemiological studies on the adverse health effects of exposure to PFOA include cross-sectional general population studies, studies in populations exposed to higher levels of PFOA through contaminated drinking water and occupational exposure studies. However, no causal relationships between PFOA and adverse health effects have been established due to numerous confounding factors including exposure to multiple chemicals.

Recently, two studies (one cross-sectional study in the United States and one cohort study in Denmark) have given limited suggestions of a weak association between gestational exposure to PFOA and reduced birth weight (Apelberg et al. 2007b; Fei et al. 2007). However, the magnitude of the effect was small, given normal variations in the parameters measured, and all of the children were within the normal range of variation. No associations between maternal serum PFOA levels and birth weight were observed in other general population studies in Canada (Monroy et al. 2008; Hamm et al. 2009), Japan (Washino et al. 2009) or a US community with a mean serum PFOA level 10 times higher than that of the general US population (Stein et al. 2009).

Confidence in the effects assessment for PFOA is moderate to high, as the toxicological database covers a wide range of endpoints and life stages, several species and both sexes.

The available data are sufficient to derive margins of exposure (MOEs) based on comparisons of serum PFOA levels in laboratory animals at the critical effect levels with serum PFOA levels in humans from biomonitoring studies. The MOEs between the serum PFOA levels associated with the most sensitive effects in laboratory animals and the serum PFOA levels in adult Canadians range from approximately 3800 to 22 600 for mean serum values and from 2100 to 12 600 for conservative 95th percentile values. Using 95th percentile serum concentrations for US adults or children, MOEs range from about 1300 to 7900 (Table 8).

Table 8. Margins of Exposure

¹MOE was calculated as the ratio of the serum PFOA level associated with the critical effect to the serum PFOA level in humans.
² Kubwabo et al. (2004).
³ Calafat et al. (2007).
⁴ Olsen et al. (2004b).

The experimental animal studies that were used for generation of MOEs were those with the lowest administered doses and serum PFOA levels associated with effects. A 14-day oral study in mice and rats (Loveless et al. 2006) had the lowest LOAEL identified in any study (0.3 mg/kg-bw per day), and MOEs were calculated based on the serum PFOA levels in mice and rats associated with this dose. Although no NOAEL was determined in the critical study, there is supporting evidence from other studies, such as a 13-week oral rat study and a 14-day rat inhalation study, in which no effects were observed at 0.06 and 0.31 mg/kg-bw per day, respectively (Haskell Laboratory 1981a; Kennedy et al. 1986; Palazzolo 1993; Perkins et al. 2004).

MOEs were also calculated based on serum levels in mice dosed at the LOAEL for developmental toxicity (1 mg/kg-bw per day) (Lau et al. 2006). Although no NOAEL was determined in this study, in a follow-up study designed to examine the mechanism of action of PFOA-induced developmental effects, a NOAEL for developmental toxicity of 0.3 mg/kg-bw per day was determined, based on decreased neonatal survival at 0.6 mg/kg-bw per day and delayed eye opening at 1 mg/kg-bw per day. As serum PFOA levels were not measured until 22 days post-partum, effect levels from this study were not used for determining an MOE (Abbott et al. 2007).

A 26-week toxicity study in monkeys was also selected for MOE estimation, as primates are considered to be a better surrogate for humans than are rodents (Thomford 2001b; Butenhoff et al. 2002). The steady-state serum PFOA level of monkeys dosed at the LOAEL of 3 mg/kg-bw per day was used for the calculations. No NOAEL was determined in this study.

The use of serum levels for the MOE calculations significantly reduces uncertainties associated with interspecies and intraspecies differences in pharmacokinetics. In laboratory animals, PFOA distributes primarily to the serum and liver (Vanden Heuvel et al. 1991; Butenhoff et al. 2004a; Hundley et al. 2006; Kudo et al. 2007). Data on PFOA tissue distribution in humans are limited, but it is assumed that the pattern would be similar to that observed in experimental animals (including male rats, male and female mice and non-human primates) (Hundley et al. 2006). PFOA has been detected in human post-mortem liver samples in two studies (Olsen et al. 2003b; Maestri et al. 2006). Since data on PFOA levels in human tissues are limited, the availability of data on PFOA concentrations in serum makes it the most appropriate measure of internal exposure. Due to the long half-life of PFOA and its lack of metabolism, human adult serum levels of PFOA represent cumulative (lifetime) exposure to PFOA and any contributions from its precursors. Serum levels of PFOA are comparable in children, adults and the elderly. Although the critical toxicology studies in experimental animals were conducted over less than lifetime exposure durations, pharmacokinetic data indicate that the reported serum PFOA levels would represent steady state (Vanden Heuvel et al. 1991; Butenhoff et al. 2004a; Lau et al. 2006).

Where an MOE is estimated on the basis of effects in rodents, the possible role of peroxisome proliferation should be acknowledged. Rats and mice are sensitive to the effects of peroxisome proliferators, whereas monkeys and humans are relatively non-responsive at similar doses (reviewed in Klaunig et al. 2003; Kennedy et al. 2004). As humans are generally less susceptible than rodents to peroxisome proliferators, MOEs based on peroxisome proliferation–dependent effects in rodents would be conservative. As the margins were calculated based on the most sensitive effects and species, they are considered protective of both PPAR-dependent and PPAR-independent effects.

APFO has induced tumours in exposed rats, but PFOA compounds have not been tested for carcinogenic potential in any other laboratory animal species. The PPAR-agonist mode of action proposed for rat liver, testes and pancreatic tumours may not be relevant for humans, but human relevance has not been definitively determined according to established frameworks (Meek et al. 2003; Boobis et al. 2006).

PFOA has been shown to activate the human PPARa in cell culture (Takacs and Abbott 2007; Wolf et al. 2008).PPARaactivation has a wide range of effects, including regulating the expression of genes involved in cell growth and survival. In fact, some PPARa ligands have been shown to possess anti-tumorigenic properties, such as suppression of growth of several types of human cancer cells in vitro and inhibition of carcinogenesis in vivo (reviewed in Pozzi and Capdevila 2008), makingPPARa a potential target for cancer therapy. It should also be noted that other PPARa ligands, such as fibrates, which cause a high incidence of tumours in rodents, are commonly used therapeutically in humans, with no evidence of carcinogenicity in epidemiological studies (reviewed in Peters et al. 2005).

Although the modes of action for tumour induction have not been demonstrated conclusively and the relevance of the rat tumours to carcinogenicity in humans is uncertain, the genotoxicity database indicates that PFOA is not directly mutagenic. Thus, as the tumours observed in male rats are not considered to have resulted from direct interaction with genetic material, a threshold approach is used to characterize risk to human health. The MOEs generated for non-neoplastic effects are protective with respect to the increased incidence of benign tumours observed in the chronic studies of PFOA in rats, since a) the tumours were observed only at PFOA doses higher than those that induced non-neoplastic effects, b) the genotoxicity database indicates that PFOA is not mutagenic, c) rat liver tumours are likely induced via a mode of action that is not relevant to humans and d) there were no PFOA-related effects in non-human primates that have been associated with the development of pancreatic or testicular tumours in rats.

The use of serum levels reduces the uncertainty associated with a determination of the upper-bound estimate of human intake of PFOA, owing to the limited available data on levels of PFOA and precursors in air, foodstuffs, drinking water and breast milk and resulting from contact with household materials treated with perfluorinated substances. Moreover, the levels of PFOA in human serum provide a measure of aggregate exposure from multiple sources and exposure routes. The use of serum levels also significantly reduces uncertainties associated with interspecies and intraspecies differences in pharmacokinetics.

Use of 95th percentile serum values is conservative. Furthermore, measures have recently been taken to reduce global facility emissions and product content of PFOA and related chemicals, including in Canada (US EPA 2006c; EPA 2009). Two recent biomonitoring studies in the United States have already shown declining serum PFOA levels (Calafat et al. 2007; Olsen et al. 2008). Exposure to PFOA in Canada is also expected to decrease over time, although serum data are currently not available to confirm this. The MOEs presented here are based on serum levels from 1994 to 2004, and the MOEs may increase over time as exposure levels are predicted to decrease in the future.

Uncertainty remains regarding the mode of action for tumour induction; however, as the database for genotoxicity suggests that PFOA is not mutagenic, the MOEs based on non-neoplastic effects in the most sensitive species are considered protective of any potential carcinogenic effects in humans.

^[2] Preliminary data from Chemicals, Health and Pregnancy (CHirP) study (http://www.cher.ubc.ca/chirp/)
^[3] Doses were calculated using the following reference values: rat inhalation rate = 0.11 m³/day; rat body weight = 0.35 kg (Health Canada 1994).