ARTICLE pubs.acs.org/est
Simple Intake and Pharmacokinetic Modeling to Characterize Exposure of Americans to Perfluoroctanoic Acid, PFOA Matthew Lorber*,† and Peter P. Egeghy‡ †
Office of Research and Development, United States Environmental Protection Agency, 1200 Pennsylvania Ave, NW, Washington, DC 20460, United States ‡ Office of Research and Development, United States Environmental Protection Agency, Mail Drop E205-04, Research Triangle Park, North Carolina 27711, United States ABSTRACT: Models for assessing intakes of perfluorooctanoic acid, PFOA, are described and applied. One model is based on exposure media concentrations and contact rates. This model is applied to general population exposures for adults and 2-year old children. The other model is a simple one-compartment, first-order pharmacokinetic (PK) model. Parameters for this model include a rate of elimination of PFOA and a blood volume of distribution. The model was applied to data from the National Health and Nutritional Examination Survey, NHANES, to backcalculate intakes. The central tendency intake estimate for adults and children based on exposure media concentrations and contact rates were 70 and 26 ng/day, respectively. The central tendency adult intake derived from NHANES data was 56 and 37 ng/day for males and females, respectively. Variability and uncertainty discussions regarding the intake modeling focus on lack of data on direct exposure to PFOA used in consumer products, precursor compounds, and food. Discussions regarding PK modeling focus on the range of blood measurements in NHANES, the appropriateness of the simple PK model, and the uncertainties associated with model parameters. Using the PK model, the 10th and 95th percentile long-term average adult intakes of PFOA are 15 and 130 ng/day.
’ INTRODUCTION With useful surfactant properties, perfluorinated compounds (PFCs) are used extensively in the engineering and chemical, electronics, and medical industries.1 They are also used in numerous common consumer products including floor wax, cleaners, pesticide products, and different kinds of paper. The most widely known PFCs are the C8-chemicals perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS), which have been found to persist in the environment and are not currently known to degrade by biotic or abiotic means.2 Potential sources of exposure to PFOA and PFOS include direct industrial releases into air and water, releases of fire-fighting foam, release from consumer products (including waterproof breathable textiles and oil-and stain-protective coatings for carpets, apparel, and food containers), and release from degradation of telomer-based polymers and other fluorinated products including polyfluoroalkyl phosphoric acids (PAPs24). Potential exposure media for PFOA, PFOS, and their precursors include air, dust, water, and food (possibly via migration from food packaging and cookware).5 Measuring trace amounts of PFOA and other PFCs in environmental and exposure media (e.g., ∼1 ng/g in food, ∼1 pg/m3 in air) presents an analytical chemistry challenge due to background contributions and matrix interferences. As a result, data on food and in the indoor environment in the United States are sparse, and consequently, understanding the pathways of human exposure remains a challenge. This article not subject to U.S. Copyright. Published 2011 by the American Chemical Society
There have been few efforts which have estimated intakes of PFOA in humans. The most comprehensive study was recently published by Fromme et al.,6 who performed an exposure assessment to perfluorinated compounds (primarily PFOS and PFOA) in western countries by combining exposure media concentrations with contact rates, while also reviewing biomonitoring data. This approach of combining concentrations with contact rates is termed “forward-based” in this study. Their exposure assessment found that average general adult population exposure to PFOA was 2.9 ng/kg-day, with an upper end estimate of 12.6 ng/kg-day, dominated by dietary exposure. Washburn et al.7 calculated an intake of 0.010.1 ng/kg-day for PFO (perfluorooctane carboxylate, the conjugate base of PFOA) based on concentrations in carpeting and cookware. Trudel et al.5 derived a central tendency North American intake estimate of 2 ng/kg-bw for PFOA. Fromme et al.8 calculated median intakes of PFOA for German study subjects based on duplicate dietary samples to be 2.9 ng/kg-bw/day. Tittlemeier et al.9 conducted a market Special Issue: Perfluoroalkyl Acid Received: November 4, 2010 Accepted: April 14, 2011 Revised: March 30, 2011 Published: April 25, 2011 8006
dx.doi.org/10.1021/es103718h | Environ. Sci. Technol. 2011, 45, 8006–8014
Environmental Science & Technology basket survey of a wide range of Canadian foods, and derived an adult exposure estimate of 70 ng/day, or 1.1 ng/kg-day assuming a 62 kg adult. A limited diet survey in the UK10 resulted in an estimate of average adult dietary intake of PFOA to be 10 ng/ kg-day (assuming nondetects equal detection limits), with a high level estimate of 20 ng/kg-day (using the 97.5% estimate for food consumption and also assuming nondetects equal detection limits). Although not a total diet survey, the authors suggest they obtained relevant samples including those that are staples in the diet (bread, milk, and potatoes), those anticipated to make the highest contribution to dietary intakes of PFOA (fish and offal), and meat and vegetables. Ericson et al.11 measured 11 PFCs in 36 composite food samples (duplicates of 18 major food types) representing the diet of Spaniards living in Catalonia, Spain. They found frequent detections of PFOS only; positive concentrations of PFOA and PFHpA were found in milk, and there were nondetects for all other PFCs at detection limits generally near or less than 0.1 ng/g fresh weight. Gulkowska et al.12 sampled 27 fish market fish samples including finfish (croaker, mackerel), shrimp, crabs, and other species. They found 100% occurrence of PFOS, ranging from 0.3 to 13.9 ng/g. They also found positives for PFOA, with infrequent positive occurrences for other perfluorinated compounds. Combining concentrations with high fish consumption rates (>100 g/day) found in dietary surveys for populations in Zhoushan and Guangzhou, they found intakes of PFOA equaling 1.1 ng/ kg-day for Guangzhou residents, and 0.4 ng/kg-day for Zhoushan residents. Haug et al.13 measured 10 PFCs, including PFOA and PFOS, in 21 food samples in Norway, and using dietary intakes of these food products, found that the highest intake was estimated for PFOA at 31 ng/day, followed by PFOS at 18 ng/day. Intakes of PFOA have also been determined using simple pharmacokinetic (PK) modeling in conjunction with blood concentrations. This approach is termed “backward-based” in this study. The model used is a simple one-compartment, firstorder model (where the compartment is blood serum). Assuming steady state and assigning values to key model parameters, one can infer intakes from blood concentrations. Trudel et al.5 assigned a range of values to key model parameters to determine a wide range of intakes at 1130 ng/kg-day. Fromme et al.14 assigned values to key model parameters based on the literature and estimated a median intake of PFOA of 0.5 ng/kg-day from blood measurements taken of 30 individuals. Thompson et al.15 calibrated the model and then applied it to blood concentrations from large surveys of Australians to estimate average general adult population intakes of PFOA of 1.6 and 1.3 ng/kg-day for two different survey sample years. In this study, we derive intakes of PFOA in both of the ways noted above. Pathway-specific exposure intakes are estimated based on exposure media concentrations and contact rates. Intakes are developed for an adult and for a 2-year old child. General population adult exposure intakes are backcalculated based on simple PK modeling starting from body burdens of PFOA. Exposures to other populations of concern including occupational exposures or exposures associated with a contamination situation are not considered. Finally, exposures to PFOA precursors are not considered due to a lack of appropriate data. There are very few studies in the literature in which both forward- and backward-based intake estimates are derived and then compared to each other. The validity of estimates presented is strengthened when these disparate approaches arrive at similar
ARTICLE
intakes. Of the studies reviewed above, only Trudel et al.5 and Fromme et al.14 used this approach, and applied it to PFCs, including PFOA. However, their backward PK model applications did not benefit from subsequent research which provided refinement to a critical model parameter. Specifically, the volume of distribution parameter, termed Vd, is the critical uncertain model parameter for which human data did not exist. This is not a physiological parameter, per se, such as the volume of lipid in a body. In a manner of speaking, it is instead a “dilution volume” which requires calibration on human data. Fromme et al.14 acknowledged this shortcoming and used one set of animal data in their application to a small set of measurements from German people. Trudel et al.,5 in contrast, presented back-calculated intakes as a wide range of 1130 ng/kg-day recognizing the wide range of Vd from different animal studies. The approach to develop both forward- and backward-based intakes was also applied to study exposures to PFOS.16 In that effort, a forward-calculated central tendency intake of PFOS for the general adult population of the United States was estimated as 4.2 ng/kg-day, while back-calculated intakes were presented as a range of 1.624.2 ng/kg-day. Again, these authors applied the PK model without the key refinement of Vd needed also by Trudel et al.5 and Fromme et al.14 Thompson et al.15 present a model calibration (summarized below) in which human data were used to develop a single value for this imprecise parameter. They applied the model to a national Australian biomonitoring database and were able to present a single central tendency of a backcalculated intake of PFOA. In this paper, that calibrated value is used in the PK model application to a United States national database to derive a single central tendency back-calculated intake without the need for bracketing based on model uncertainty. If this single central tendency value is reasonably close to the forward-calculated central tendency intake estimate, then confidence is gained on the quantification of exposure to PFOA. The variability in pathway intakes is characterized by developing distributions of exposure media concentrations from published data or summary statistics and assuming a log-normal distribution. The variability in long-term average intakes is assumed to mirror the variability in the body burden measurements, and hence PK modeling is surmised to provide a better indication of long-term average intakes. Uncertainties in both the pathway intake analysis and the PK modeling are discussed. As noted earlier, the extent to which the two independent approaches to deriving an intake match each other speaks to the validity of the estimates.
’ METHODS General population intakes for adults and young children, aged approximately 2 years old (as characterized by exposure factors), were modeled using a “forward” approach based on exposure media concentrations and contact rates. Intakes for adults were also modeled using a “backward” PK modeling approach. Available data on PFOA concentrations in exposure media and in human biological media were extracted mainly from published, peer-reviewed references. Table 1 provides an overview of the methods used to calculate exposure including exposure factors and sources of assumed concentrations for PFOA. Route-specific exposures and subsequent intakes were estimated by a deterministic methodology consistent with EPA’s Guidelines for Exposure Assessment.17 Recommended exposure contact rates for adults and children were obtained from EPA’s Exposure Factors Handbook18 and Child-Specific Exposure Factors Handbook.19 Instead of using point estimates to represent potential exposures to all adults or 8007
dx.doi.org/10.1021/es103718h |Environ. Sci. Technol. 2011, 45, 8006–8014
Environmental Science & Technology
ARTICLE
Table 1. Data Sources and Assumptions Used to Estimate Intakes exposure parameters
source of exposure data inhalation outdoor air, ng/day; intake = conc * R * T * AF
concentration data (conc): ng/m3
Kim et al.20
3
EPA19
inhalation rate (R): 6.8 m /day (child) 3
EPA18
13.3 m /day (adult) fraction of time performing activity (T): 5 h/day = 0.208 day (child)
EPA19
3 h/day = 0.125 day (adult)
EPA18
absorption fraction (AF): 0.5
estimate inhalation indoor air, ng/day; intake = conc * R * T * AF 20 times outdoora
concentration data (conc): ng/m3 3
EPA19
inhalation rate (R): 6.8 m /day (child) 3
EPA18
13.3 m /day (adult) fraction of time performing activity (T): 19 h/day = 0.792 day (child)
EPA19
21 h/day = 0.875 day (adult)
EPA18
absorption fraction (AF): 0.5
estimate b
water ingestion, ng/day; intake = conc * V * AF concentration data (conc): ng/L
Sinclair et al.,23 Nakayama et al.,24 Boulanger et al.,25 3 M 200127
volume consumed (V): 0.4 L/day (child)
EPA19
1.41 L/day (adult)
EPA18 c
GI absorption fraction (AF): 0.9
estimate dust ingestion, ng/day; intake = conc * IR * AF
concentration data (conc): ng/g dust
Strynar and Lindstrom 200827
ingestion rate (IR): 0.1 g dust/day (child)
EPA19
0.05 g dust/day (adult) GI absorption fraction (AF): 0.9c
EPA18 estimate dermal absorption of dust, ng/day; intake = conc * DL * SA * R * AF Strynar and Lindstrom28
concentration data (conc): ng/g dust 2
EPA16
dust load (DL): 0.00001 g dust/cm skin (adult) 2
0.000003 g dust/cm skin (child) surface area of skin (SA): 6000 cm2 (child)
EPA19
2
EPA18
12500 cm (adult) replenishment rate (R): 2/day
estimate
dermal absorption fraction (AF): 0.048% d
Fasano et al.30 dietary ingestion, ng/day; intake = est * AF * BW
estimate (est): ng/kg/day
Tittlemier et al.9e
body weight (BW): 13 kg (2 year old child)
EPA19
71.8 kg (adult)
EPA18 c
GI absorption fraction (AF): 0.9
estimate
a
The assumption that PFOA concentrations in indoor air are 20 times higher than in outdoor air is based on indoor and outdoor levels of perfluorinated alkyl sulfonamide reported by Shoeib et al.20 b Measurements from Lake Onondaga, a Superfund site impacted by several industries, were excluded. c High absorption based on evidence of substantial enterohepatic circulation (based on gavage studies of rodents). d Based on APFO across human skin.29 e Based on measurements of Canadian diet.9
young children, distributions for each exposure pathway were developed. These were developed by holding the exposure factors constant while developing a range representing the exposure media concentration. Specifically, log-normal distributions were created from available summary statistics utilizing the mathematical relationships among the parameters of log-normal distributions. Further detail on this procedure can be found in Egeghy and Lorber.16
Outdoor air concentrations were based on values reported for Albany, New York, characterized as an urban environment.20 Vapor- and particle-phase concentrations were both measured; vapor phase ranged from 1.9 to 6.5 pg/m3, while particle phase ranged from 0.84.2 pg/m3. Indoor air concentrations were assumed to be 20 times higher than outdoor concentrations based on data provided in Shoeib et al.21 They simultaneously 8008
dx.doi.org/10.1021/es103718h |Environ. Sci. Technol. 2011, 45, 8006–8014
Environmental Science & Technology measured perfluoralkyl sulfonamides (N-methyl perfluoroctane sulfonamidoethanol, or MeFOSE, for example) in outdoor and indoor air, and found that indoor air concentrations were 1020 times higher than those of outdoor air. The outer estimate of 20 times was conservatively used in this analysis. An earlier publication by the same group reported an indoor:outdoor ratio of about 100,22 but it was a pilot-scale study with far fewer samples over much shorter durations. The constant inhalation rate was assumed to be 13.3 m3/day for the adult and 6.8 m3/day for the child.18,19 Drinking water concentrations were estimated mainly using surface water concentrations as surrogates.2327 Sinclair et al.23 measured PFOA in 51 grab samples from 9 surface water bodies in New York State. Average water concentrations from each body ranged from 19 to 49 ng/L, with a maximum of 173 ng/L found in the Hudson River. Similar surface water concentrations have been found in other studies, including studies on the Cape Fear River Basin in North Carolina,24 in the Great Lakes,25 and in public water supply systems in New Jersey.26 Measurements submitted to EPA by 3 M of drinking water and surface water in two cities with no known significant industrial use of fluorochemicals (Cleveland, Tennessee and Port St. Lucie, Florida) were all below the limits of detection (n = 23) or quantification (n = 2).27 The constant water ingestion rate was 1.4 L/day for the adult and 0.4 L/day for the child.18,19 For concentrations of PFOA in dust, the actual measured values reported in Strynar and Lindstrom28 were used. They measured 10 perfluoroalkyl acids (including PFOA) and three fluorinated telomere alcohols in 102 dust samples collected from homes and 10 samples collected from day care centers in Ohio and North Carolina in 20002001. They found a mean concentration of PFOA in these samples of 296 pg/g, a 95th percentile of 1200 pg/g, and a maximum of 1960 pg/g. For dust ingestion, rates typically assumed for soil ingestion were used. For the 2-year-old child the rate was 100 mg/day, and for adults the rate was 50 mg/ day.18,19 For dermal absorption to dust, exposure factors include an area weighted average dust load on skin of 0.01 mg/cm2 for children and 0.003 mg/cm2 for adults,29 an assumed dust replenishment rate of twice per day, and a dermal absorption fraction of 0.00048 from a study APFO in an aqueous solution.30 It is unclear how well the APFO data pertain to transfer of PFOA from dust. However, it might be speculated, at least, that transfer from an aqueous solution might be more facilitated as compared to transfer from dust, so assignment of this factor based on APFO data might lead to an overestimate of dermal transfer. Since the dermal pathway showed to be such a small contributor to overall exposure (see Discussion of results below), it is surmised that this modeling approach and data assignments are acceptable. Data on food concentrations could not be found to characterize the American diet, so results from the Canadian Total Dietary Survey9 were used. This study by Health Canada included 54 solid food composite samples analyzed for perfluorocarboxylates (including PFOA) and PFOS. Foods analyzed included fish and seafood, meat, poultry, frozen entrees, fast food, and microwave popcorn collected from 1992 to 2004 and prepared as for consumption. Nine composites contained detectable levels of perfluorinated compounds, overall. The roast beef (2.6 ng/g wet weight), pizza (0.7 ng/g wet weight), and popcorn (3.6 ng/g) were found positive for PFOA, and from this market basket survey Tittlemeier et al.9 derived an estimate of mean adult total dietary exposure to PFOA of 70 ng/day (which assumed nondetects equal to zero for composites not found positive for PFOA). We used this mean
ARTICLE
together with a geometric standard deviation derived from Fromme et al.8 to develop the log-normal distribution for intake from food. A simple, single-compartment, first-order PK model which predicts PFOA concentrations in blood serum as a function of dose, elimination rate, and volume of distribution, is used: dðCPÞ=dt ¼ DPðtÞ=Vd kP CPðtÞ
ð1Þ
where CP is the serum concentration (ng/mL) of PFOA, DP is the daily absorbed dose (ng/kg bw/day), Vd is the volume of distribution (mL/kg bw), and kP is the first-order elimination rate (day1). Vd and kP are assumed to be constant in this model construct. Assuming steady state conditions exist, one can solve for blood serum concentration as follows: CP ¼ DP=ðkP VdÞ
ð2Þ
This can be rearranged to calculate DP or Vd, depending on the data and modeling objectives. Vd is defined as the total amount of a substance in the body divided by its concentration in the blood or serum (Vd [mL/kg bw] = mass in body [ng/kg bw]/ concentration in blood or serum [ng/mL]). Although Vd likely has a physiological underpinning specific to the contaminant and the animal species in which it is applied, it is best thought of as a modeling parameter which needs to be carefully assigned. Previous PFC modeling studies5,14,16,31 have relied upon Vds obtained through animal dosing studies, such as those estimated by Andersen et al.,32 Griffith and Long,33 or Seacat et al.34 These values have ranged from as low as 140 to as high as 6000 mL/kg. In this work we use a value calibrated from human serum and exposure data on PFOA. The site involved a contaminated surface water supply which was shown to elevate levels of PFOA in residents consuming water from the contaminated source. Deriving an intake from contaminated drinking water, the model’s value of Vd was calibrated so that model predictions of elevated blood levels of PFOA matched those seen in the residents. This calibration appears in the Supporting Information in Thompson et al.,15 and the calibrated value of of 170 mL/kg for Vd was used in this assessment. The elimination rate constant, kP, was assigned a value of 0.0008 day1 based on a serum half-life of 2.3 years derived from a general population study.35 As an absorbed dose, DP needs to be adjusted upward to calculate the “exposure” or “intake” dose, which is the dose contacting the body. Past exposure work has shown that ingestion exposures (food, water, house dust) most explain exposures to PFCs,5,16 and based on laboratory testing of PFCs on a variety of species,36 a value of 0.9 will be assumed for the absorption fraction. DP as solved in eq 2 is divided by 0.9 to arrive at an external intake dose. The model is applied to data from the National Health and Nutritional Evaluation Survey (NHANES) from 2003/2004, encompassing over 2000 samples. The median serum concentrations of PFOA for males and females above the age of 12 were 4.6 and 3.6 ng/mL, respectively.37
’ RESULTS Results of these route-specific intake estimates are presented in box-and-whisker plots (boxplots) in Figure 1. The boxplots display estimates for the 5th and 95th percentiles (lower and upper whiskers, respectively) and the 25th percentile, median, and 75th percentile (bottom, middle, and top of box, respectively); extreme 8009
dx.doi.org/10.1021/es103718h |Environ. Sci. Technol. 2011, 45, 8006–8014
Environmental Science & Technology
ARTICLE
Figure 1. Estimated pathway-specific intakes of PFOA under a typical environmental exposure scenario for (A) young children and (B) adults.
values (those beyond the fifth or 95th percentiles) are displayed as open circles. These values are portrayed as dots and are based on the assumption of a log-normal distribution. Dermal absorption, dietary and nondietary ingestion, and inhalation are represented as separate boxes in each panel. The median PFOA intake (i.e., the sum of the median routespecific intakes) for two-year-old children under typical exposure conditions is estimated as 26 ng/day. Ingestion of dust and of food appear to be the primary routes of exposure (see Figure 1A), providing approximately 13 and 8 ng/day, respectively. At the 95th percentile, however, intake from dust ingestion (110 ng/day) is roughly three times the intake from food ingestion (37 ng/day) due to much greater observed variability in the dust concentrations. Intake from water ingestion is estimated to be the third most important source of PFOA intake at both the median (4.8 ng/day) and the 95th percentile (28 ng/day). Intake from dermal absorption and inhalation of indoor air and of outdoor air together represent less than 2% of total intake. Among adults, the aggregate median PFOA intake is estimated to be 70 ng/day, and dietary ingestion again appears to be the primary route of exposure providing about 46 ng/day (see Figure 1B). This food intake was based on very limited positive sample results from the Health Canada Total Diet Study (discussed above), so clearly there is uncertainty around this estimate. The contribution from the ingestion of drinking water follows dietary ingestion at 17 ng/day, and its relative contribution to total intake (about 24%) is similar among adults and children. Incidental ingestion of dust (6.4 ng/day or 9.1% of total intake) is far less important among adults than among children due to differences in the amount of dust assumed to be ingested.
Given the median NHANES results as described earlier, 4.6 ng/mL for males over the age of 12 and 3.6 ng/mL for females also older than 12, the backcalculated intakes are 0.72 and 0.56 ng/kg-day. Assuming body weights of 78.1 and 65.4 kg for males and females, respectively,18 the intakes are about 56 and 37 ng/day. This is between 50 and 80% of the 70 ng/day intake surmised from exposure media concentrations and contact rates, which is certainly within a very reasonable range considering the sparseness of data and other uncertainties. The 95th percentile blood concentration for males and females was 10.4 and 8.4 ng/mL, leading to exposures a little over twice as high than the medians. The 10th percentile for males and females was 2.3 and 1.6 ng/mL, about half as much as the medians. So, as a simple estimation using the low end for females of 1.6 ng/mL and the high end for males of 10.4 ng/mL, this implies a variability in long-term average adult exposures between about 15 to about 130 ng PFOA/day.
’ DISCUSSION This study provides a simple assessment of aggregate exposure to PFOA in the United States for two-year-old children and for adults under the scenario of a typical environment not impacted by local activities. The available measurement data and analysis in this paper point toward dietary ingestion as the major contributor to PFOA intake for adults, and dust and dietary ingestion as the major contributors for young children, under typical residential exposure scenarios. The aggregate intake estimate for adults is somewhat lower than that recently reported by Fromme et al.6 for adults in western countries (in both Europe and North America). Our median of 70 ng/day corresponds to a mean of 8010
dx.doi.org/10.1021/es103718h |Environ. Sci. Technol. 2011, 45, 8006–8014
Environmental Science & Technology 110 ng/day, while Fromme and colleagues6 reported a mean of 171 ng/day, of which approximately 99% (169 ng/day) was attributed to diet. In contrast, we estimate a much lower intake from food and much higher contributions from drinking water and housedust ingestion. Central tendency intake is slightly lower when calculated using backward PK modeling compared to the value surmised from exposure media concentrations and contact rates. However, given the uncertainties and variabilities inherent in both exercises, it might be reasonable to conclude that the long-term average intakes of adults in the United States fall within the range implied by NHANES data in conjunction with PK modeling: 15130 ng PFOA/day. The following are key uncertainties in the pathway intake analysis and the implications for each: 1 This analysis omitted pathways involving direct contact with consumer product sources, such as transfer from contact with PFC-treated carpets or clothing.38 It could be argued that dust-related and inhalation exposures provide a surrogate for these exposures, although such an argument would not account for exposures that might result from PFCs in floor care products, cosmetics, health care products, and other consumer products. Given that these direct contact exposures are likely to occur and are in addition to dust-related and inhalation exposures, the pathway estimates derived above underestimate true exposure. No estimates for direct product contact exposure in the literature could be found. 2 This analysis omitted consideration of precursors of PFOA. Vestergren et al.39 assumed PFOA was formed in the body as a result of metabolism of fluorotelomer alcohols, FTOHs. In their analysis, they found precursors to be a minor contributor, between 2 and 8% of total exposure to PFOA, using “intermediate” assumptions for media concentrations and exposure contact rates. For the high end scenario, where total PFOA exposure (equaling the sum of exposures to precursor metabolism of FTOH to PFOA and PFOA as product) was modeled to be above 30 ng/kg-day, precursor contribution was about 60% of total. They found a similar percentage precursor contribution at the intermediate and high scenarios for PFOS. In Egeghy and Lorber,16 exposure to precursors of PFOS contributed about 40% of the total PFOS exposure at the median modeled pathway intake total of 4.2 ng/kg-day. In addition to FTOHs, another important precursor is the polyfluoroalkyl phosphoric acids, PAPs. Their presence in the products, in the environment, and in human serum has been documented,4 and the internal metabolism of these compounds to PFOA has been demonstrated in rats.3 In limited human serum sampling (pooled n = 20 including n = 10 in 2004/5 and n = 10 in 2008), levels in human serum were about 4.5 μg/L for the 4:2 through 10:2 PAP diesters, which were comparable to PFOS and PFOA found in the samples.4 This suggests exposures to PAPs comparable to these other perfluorinated compounds. Like the omitted pathways noted above, consideration of precursors would add to estimated exposure, but their contribution remains an uncertainty with little information in the literature beyond the finding of Vestergren et al.39 and D’eon et al.4 to support quantitative assessments. 3 The lack of measurements of perfluorinated compounds in food remains a key shortcoming to any assessment of exposure to these compounds. This was also identified as
ARTICLE
a major uncertainty in the comprehensive assessment of Fromme et al.,6 who reported substantially more data in air, water, and housedust. Tittlemeyer et al.,9 who provided the limited data used in this assessment to characterize dietary exposures, also make the same observation. Indeed, their estimate of 70 ng PFOA/day ingestion was based largely on the single positive measurement of 2.6 ng/g found in roast beef. The log-normal distribution fitted around this finding for the assessment in this paper has to be considered suspect. Tittlemeyer et al.9 note one industry-supported study (reference provided as laboratory report) which analyzed different raw foods collected from six cities in the southern United States for PFOS and PFOA. Four whole milk and one ground beef sample contained quantifiable levels of PFOS (0.5730.852 ng/g); two ground beef, two apple, one bread, and one green bean sample contained quantifiable levels of PFOA (0.5042.35 ng/g). Only one of these positive results (PFOS in ground beef) was confirmed in its duplicate sample. A recent study in Norway13 was able to quantify PFOA in 10 of 17 samples of food, and for the other 7 samples, provided a concentration qualified as being between the detection and quantification limits. All of these concentrations were at 0.05 ng/g fresh weight and below, suggesting that future work on these compounds in food probably need to achieve quantification limits at this concentration. The Canadian total diet studies reported in Tittlemeyer et al.9 had a detection limit of 0.5 ng/g fresh weight for PFOA. Many of these uncertainties are obviated with blood measurements, which provide a composite measure of exposure from all pathways and sources. However, extrapolating that measurement to an intake entails its own uncertainties, which are as follows: 1 The simple one-compartment, first-order model applied at steady state may not be appropriate for this class of compounds. In dosing studies using Cynomolgus monkeys, a plateau of serum concentrations were observed by Seacat et al.34 for animals given 0.75 mg/kg bw/day. Andersen et al.32 observed a similar phenomenum and hypothesized a saturable reabsorption process as responsible, suggesting that a more complicated model is warranted for PK modeling of PFCs. However, the experimental doses of Seacat et al.34 and Andersen et al.32 resulted in serum concentrations in the tens of parts per million, while those in the background populations and even the affected community where higher levels in blood and water allowed for the calibration of Vd in the PK model (see Thompson et al.15) were orders of magnitude lower. Still, given this uncertainty, careful consideration should be given to the use of this simple PK model for higher exposures such as those that might occur in an occupational setting, or for populations of particular concern (susceptible populations, e.g.). Even given this uncertainty in choice of model, the assumption of steady state introduces further uncertainty. Because PFCs bioaccumulate with half-lives on the order of years, it takes several years to reach steady state. With a half-life of 2.3 years, intakes must have been steady for possibly over 10 years or more to reach a level close to steady state. However, there is evidence of a decline in the exposure to PFCs in the 2000s, suggesting that a simple steady state approach might not be appropriate. PFOS was removed from commercial markets in the early 2000s, and there is already evidence for a decline in both PFOS and PFOA serum concentrations in 8011
dx.doi.org/10.1021/es103718h |Environ. Sci. Technol. 2011, 45, 8006–8014
Environmental Science & Technology the U.S. from the 19992000 NHANES to the 20032004 NHANES. Specifically, Calafat et al.37 noted a decline in geometric mean concentrations (from the full population) from 5.2 to 3.9 ng/mL PFOA. Deriving intakes based on a steady state model assumes constant exposures. With declining exposures resulting in lower blood levels, the likely impact of the steady state assumption on these lower blood levels is to underestimate past exposures (they were higher in the past to result in higher past blood levels), but possibly to overestimate current exposures. Body burdens will continue to decline with declining exposures until a new steady state is reached. The magnitudes of the underestimation or overestimation are not known, but intuitively given this measured decline, both the underestimation and overestimation are likely to be within a factor of 2 of the intakes calculated at steady state. 2 Point estimates in assignment of model parameters do not consider uncertainty and/or variability. The half-life of 2.3 yr for PFOA found in a general population contrasts a half-life of 3.5 yr found in an occupational cohort.40 Use of this latter estimate would result in a calibrated Vd that is higher than 170 mL/kg, but then a lowering effect on modeled intakes starting from body burdens, so the changes would compensate and there would not have been a substantive difference in predicted intakes. The half-life used in the calibration did, in fact, come from a study on the same population whose water supply was impacted.35 Once individuals in this population were provided drinking water which did not contain PFOA, their blood levels declined, and it was this finding that was used to estimate the half-life. Studies done on this population prior to a decline in their blood levels were used to calibrate the Vd, and there is uncertainty in that exercise as well. The key assumption that the water supplied the only PFOA is critical, particularly since studies also suggested consumption of locally grown vegetables could be important for exposure to PFOA.40 If the intake term in the calibration were increased, the Vd would need to be increased by the corresponding amount to arrive at the measured blood concentration. For example, if the actual exposure to other sources was twice that by water alone, the Vd would be calibrated to 340 rather than 170 mL/kg. Given the findings of the strongest relationship to water, and the fact that water concentrations were orders of magnitude higher than typically found in other noncontaminated settings, it is doubtful that vegetables and other sources of exposure provided the same amount as the contaminated water supply. Also, others had assigned values up to 6000 mL/kg5 for Vd in sensitivity analysis testing. For these reasons, the assigned value of 170 mL/kg is judged as reasonable. Intake estimates are linearly related to this parameter—an order of magnitude increase of this parameter to 1700 mL/kg would result in increased predictions of intake also an order of magnitude higher. It is interesting to note that Butenhoff et al.41 determined different values of Vd for male and female monkeys: the Vd for females was higher at 198 mL/kg compared to 181 mL/kg for males. In NHANES, the median concentrations were higher for male adults as compared to female adults, 4.6 versus 3.6 ng/mL. Using a constant Vd led a similar disproportionate intake dose by male and female adults of the American population: 56 ng/day for males and 37 ng/day for females. However, if a female Vd were higher than a male Vd, than intake
ARTICLE
predictions would be closer. For example, if the female Vd was 200 mL/kg instead of 170 mL/kg, than a predicted intake for females would be 44 ng/day, a bit closer now to 56 ng/day. Although assignment of Vd is clearly an uncertainty even with the calibration exercise, the concept of a sex-specific Vd in this modeling structure is intriguing. Other factors could as well explain the difference in male and female blood PFOA measurements, including different metabolism leading to different elimination rates and different patterns of exposure. Harada et al.42 hypothesized that monthly menstrual loss of blood and associated PFCs might explain why women have lower body burdens than men. A range of average long-term intakes, 15130 ng/day, was generated from NHANES data using a 10th and 95th percentile value. Although there are uncertainties with portraying this as a range of exposures, it may to be superior to a range than can be determined from exposure media and contact rates. This judgment is made because (1) blood measurements provide a surrogate measure of both total exposures from all pathways, (2) blood measurements represent long-term average exposures since PFOA is long-lived in the body, and (3) unknown correlations among the various exposure routes make it difficult to estimate a range of aggregate exposures from the route-specific intake estimates. Of the uncertainties noted above, the one which is of highest concern when judging the 15130 ng/day intake range is the lack of consideration of precursor compounds. These could very well contribute half or more of what is eventually measured as PFOA in the blood. As such, this range must be viewed carefully and probably characterized as an overestimate of the intake of PFOA alone. Even with this and other identified uncertainties, the central tendency estimates of intake derived by backward pharmacokinetic modeling or forward exposure modeling are similar to each other and similar to estimates published around the world as summarized in the Introduction. While this lends credibility to the intake findings presented here, uncertainties exist and further study on exposure to PFOA and other perfluorinated compounds is warranted. Continued measurements of environmental and dietary concentrations will be useful in combination with biomonitoring data to refine these estimates and also to monitor changes in aggregate exposures corresponding to changes in the production and use of PFCs.
’ AUTHOR INFORMATION Corresponding Author
*E-mail:
[email protected].
’ ACKNOWLEDGMENT The views expressed in this article are those of the authors and do not necessarily reflect the views or policies of the U.S. Environmental Protection Agency. Mention of trade names or commercial products does not constitute endorsement or recommendation for use. ’ REFERENCES (1) Lewandowski, G.; Meissner, E.; Milchert, E. Special applications of fluorinated organic compounds. J. Hazard Mater. 2006, 136, 385–391. (2) EPA. Revised Draft Hazard Assessment of Perfluorooctanoic Acid and Its Salts; U.S. Environmental Protection Agency, Office of Pollution Prevention and Toxics, Risk Assessment Division: Washington, DC, 2002. 8012
dx.doi.org/10.1021/es103718h |Environ. Sci. Technol. 2011, 45, 8006–8014
Environmental Science & Technology (3) D’eon, J. C.; Mabury, S. A. Production of perfluorinated carboxylic acids (PFCAs) from the biotransformation of polyfluoroalkyl phosphate surfactants (PAPS): Exploring routes of human contamination. Environ. Sci. Technol. 2007, 41, 4799–4805. (4) D’eon, J. C.; Crozier, P. W.; Furdui, V. I.; Reiner, E. J.; Libelo, E. L.; Mabury, S. A. Observation of a commercial fluorinated material, the polyfluoroalkyl phosphoric acid diesters, in human sera, wastewater treatment plant sludge, and paper fibers. Environ. Sci. Technol. 2009, 43, 4589–4594. (5) Trudel, D.; Horowitz, L.; Wormuth, M.; Scheringer, M.; Cousins, I. T.; Hungerbuhler, K. Estimating consumer exposure to PFOS and PFOA. Risk Anal. 2008, 28, 251–269. (6) Fromme, H.; Tittlemier, S. A.; V€olkel, W.; Wilhelm, M.; Twardella, D. Perfluorinated compounds Exposure assessment for the general population in western countries. Int. J. Hyg. Environ. Health 2009, 212, 239–270. (7) Washburn, S. T.; Bingman, T. S.; Braithwaite, S. K.; Buck, R. C.; Buxton, L. W.; Clewell, H. J.; Haroun, L. A.; Kester, J. E.; Rickard, R. W.; Shipp, A. M. Exposure assessment and risk characterization for perfluorooctanoate in selected consumer articles. Environ. Sci. Technol. 2005, 39, 3904–3910. (8) Fromme, H; Midasch, O.; Twardella, D.; Angerer, J.; Boehmer, S.; Liebl, B. Occurrence of perfluorinated substances in an adult German population in southern Bavaria. Int. Arch. Occup. Environ. Health 2007, 80, 313–319. (9) Tittlemier, S. A.; Pepper, K.; Seymour, C.; Moisey, J.; Bronson, R.; Cao, X. L.; Dabeka, R. W. Dietary exposure of Canadians to perfluorinated carboxylates and perfluorooctane sulfonate via consumption of meat, fish, fast foods, and food items prepared in their packaging. J. Agric. Food Chem. 2007, 55, 3203–3210. (10) FSA. Survey of Fluorinated Chemicals in Food; Information Sheet Number 05/09; Food Standards Agency: United Kingdom. October 2008. http://fluoridealert.org/re/pfoa-pfos.uk.survey.2009.pdf. Accessed Oct 12, 2010. (11) Ericson, I.; Marti-Cid, R.; Nadal, M.; Bavel, B. V.; Lindstrom, G.; Domingo, J. L. Human exposure to perfluorinated chemicals through the diet: Intake of perfluorinated compounds in foods from the Catalan (Spain) market. J. Agric. Food Chem. 2008, 56, 1787–1794. (12) Gulkowska, A.; Jiang, Q.; Ka So, M; Taniyasu, S.; Lam, P. K. S.; Yamashita, N. Persistent perfluorinated acids in seafood collected from two cities of China. Environ. Sci. Technol. 2006, 40 (12), 3736–3741. (13) Haug, L. S.; Salihovic, S.; Jogsten, I. E.; Thomsen, C.; Bavel, B.; Lindstrom, G.; Becher, G. Levels in food and beverages and daily intake of perfluorinated compounds in Norway. Chemosphere 2010doi 10.1016/j.chemosphere.2010.06.023. (14) Fromme, H.; Schlummer, M. S.; Moller, A.; Gruber, L.; Wolz, G.; Ungerwiss, J.; Bohmer, S.; Dekant, W.; Mayer, R.; Liebl, R.; Twardella, D. Exposure of an adult population to perfluorinated substances using duplicate diet portions and biomonitoring data. Environ. Sci. Technol. 2007, 41, 7928–7933. (15) Thompson, J.; Lorber, M.; Toms, L. M. L.; Kato, K.; Calafat, A. M.; Mueller, J. F. Use of simple pharmacokinetic modeling to characterize exposure of Australians to perfluorooctanoic acid and perfluorooctane sulfonic acid. Environ. Int. 2010, 36, 390–397. (16) Egeghy, P.; Lorber, M. An assessment of the exposure of Americans to perfluorooctane sulfonate: A comparison of estimated intake with values inferred from NHANES data. . J. Exposure Sci. Environ. Epidemiol. 2010doi 10.1038/jes.2009.73. (17) EPA. Guidelines for Exposure Assessment; EPA/6002-92/001 FR57; Office of Health and Environmental Assessment, U. S. Environmental Protection Agency: Washington, DC, 1992. (18) EPA. Exposure Factors Handbook; EPA/600/P-95/002B; Office of Research and Development, U.S. Environmental Protection Agency: Washington, DC, 1997. (19) EPA. Child-Specific Exposure Factors Handbook; EPA/600/R/ 06/096F; Office of Research and Development, U. S. Environmental Protection Agency: Washington, DC, 2008. (20) Kim, S. K.; Kannan, K. Perfluorinated acids in air, rain, snow, surface runoff, and lakes: Relative importance of pathways to contamination of urban lakes. Environ. Sci. Technol. 2007, 41, 8328–8334.
ARTICLE
(21) Shoeib, M.; Harner, T.; Wilford, B. H.; Jones, K. C.; Zhu, J. Perfluorinated Sulfonamides in Indoor and Outdoor Air and Indoor Dust: Occurrence, Partitioning, and Human Exposure. Environ. Sci. Technol. 2005, 39, 6599–6606. (22) Shoeib, M.; Harner, T.; Ikonomou, M.; Kannan, K. Indoor and outdoor air concentrations and phase partitioning of perfluoroalkyl sulfonamides and polybrominated diphenyl ethers. Environ. Sci. Technol. 2004, 38, 1313–1320. (23) Sinclair, E.; Mayack, D. T.; Roblee, K.; Yamashita, N.; Kannan, K. Occurrence of perfluoroalkyl surfactants in water, fish, and birds from New York State. Arch. Environ. Contam. Toxicol. 2006, 50, 398–410. (24) Nakayama, S.; Strynar, M. J.; Hlefant, L.; Egeghy, P.; Ye, X.; Lindstrom, A. B. Perfluorinated compounds in the Cape Fear Drainage Basin in North Carolina. Environ. Sci. Technol. 2007, 41, 5271–5276. (25) Boulanger, B; Vargo, J; Schnoor, J. L.; Hornbuckle, K. C. Detection of perfluorooctane surfactants in Great Lakes water. Environ. Sci. Technol. 2004, 38 (15), 4064–4070. (26) NJDEP. Determination of Perfluoroactanoic Acid (PFOA) in Aqueous Samples; Final Report; New Jersey Department of Environmental Protection, Division of Water Supply, Bureau of Safe Drinking Water, P.O. Box 426, Trenton, NJ 08625, January 2007. (27) 3M Laboratories. Environmental monitoring—Multi-city study: Water, sludge, sediment, POTW effluent, and landfill leachate samples; In U.S. Environmental Protection Agency Administrative Record AR2261030A; St. Paul, MN, 2001. (28) Strynar, M. J.; Lindstrom, A. B. Perfluorinated compounds in house dust from Ohio and North Carolina, USA. Environ. Sci. Technol. 2008, 42, 3751–3756. (29) EPA. Risk Assessment Guidelines for Superfund Volume 1: Human Health Evaluation Manual (Part E Supplemental Guidance for Dermal Risk Assessment); EPA/540/R/99/005; U.S. EPA: Washington, DC, 2004. (30) Fasano, W. J.; Kennedy, G. L.; Szostek, B.; Farrar, D. G.; Ward, R. J.; Haroun, L.; Hinderliter, P. M. Penetration of ammonium perfluorooctanoate through rat and human skin in vitro. Drug Chem. Toxicol. 2005, 28, 79–90. (31) Harada, K.; Saito, N.; Sasaki, K.; Inoue, K.; Koizumi, A. Perfluorooctane sulfonate contamination of drinking water in the Tama River, Japan: Estimated effects on resident serum levels. Bull. Environ. Contam. Toxicol. 2003, 71, 31–36. (32) Andersen, M. E.; Clewell, H. J.; Tan, Y. M.; Butenhoff, J. L.; Olsen, G. W. Pharmacokinetic modeling of saturable, renal resorption of perfluoroalkylacides in monkeys probing the determinants of long plasma half-lives. Toxicology 2006, 227, 156–164. (33) Griffith, F. D.; Long, J. E. Animal toxicity studies with ammonium perfluorooctanoate. Am. Ind. Hyg. Assoc. J. 1980, 41, 576–583. (34) Seacat, A. M.; Thomford, P. J.; Hansen, K. J.; Olsen, G. W. Subchronic toxicity studies on perfluorooctane sulfonate potassium salt in Cynomolgus Monkeys. Toxicol. Sci. 2002, 68, 249–264. (35) Bartell, S. M.; Calafat, A. M.; Lyu, C.; Kato, K.; Ryan, P. B.; Steenland, K. Rate of decline in serum PFOA concentrations after granulated activated carbon filtration at two public water systems in Ohio and West Virginia. Environ. Health Perspect. 2010, 118, 222– 228. (36) Hundley, S. G.; Sarrif, A. M.; Kennedy, G. L., Jr. Absorption, Distribution, and Excretion of Ammonium Perfluorooctanoate (APFO) After Oral Administration to Various Species. Drug Chem. Toxicol. 2006, 29, 137–145. (37) Calafat, A. M.; Wong, L.-Y.; Kuklenyik, Z.; Reidy, J. A.; Needham, L. L. Polyfluoroalkyl Chemicals in the U.S. Population: Data from the National Health and Nutrition Examination Survey (NHANES) 2003 2004 and Comparisons with NHANES 19992000. Environ. Health Perspect. 2007, 115, 1596–1602. (38) Guo, Z.; Liu, X.; Krebs, K. A. Roache, N. F. Perfluorocarboxylic Acid Content in 116 Articles of Commerce; EPA/600/R-09/033; Office of Research and Development, U.S. Environmental Protection Agency: Research Triangle Park, NC, 2009. (39) Vestergren, R; Cousins, I. T.; Trudel, D.; Wormuth, M.; Scheringer, M. Estimating the contribution of precursor compounds 8013
dx.doi.org/10.1021/es103718h |Environ. Sci. Technol. 2011, 45, 8006–8014
Environmental Science & Technology
ARTICLE
in consumer exposure to PFOS and PFOA. Chemosphere 2008, 73 (10), 1617–1624. (40) Emmett, E. A.; Shofer, F. S.; Zhang, H.; Freeman, D.; Desai, C.; Shaw, L. M. Community exposure to perfluorooctanoate: Relationships between serum concentrations and exposure sources. J. Occup. Environ. Med. 2006, 48, 759–770. (41) Butenhoff, J. L.; Kennedy, G. L., Jr.; Hinderliter, P. M.; Lieder, P. H.; Jung, R.; Hansen, K. J.; Gorman, G. S.; Noker, P. E.; Thomford, P. J. Pharmacokinetics of Perfluorooctanoate in Cynomolgus Monkeys. Toxicol. Sci. 2004, 82, 394–406. (42) Harada, K.; Inoue, K.; Morikawa, A.; Yoshinaga, T.; Saito, N.; Koizumi, A. Renal clearance of perfluorooctane sulfonate and perfluorooctanoate in humans and their species-specific excretion. Environ. Res. 2005, 99, 253–261.
8014
dx.doi.org/10.1021/es103718h |Environ. Sci. Technol. 2011, 45, 8006–8014
