Fluorochemicals have historically been manufactured via two different processes. Electrochemical fluorination (ECF) was used in the production of perfluorooctane sulfonyl fluoride [POSF; F(CF₂)₈SO₂F], which was further functionalized into the sulfonamido ethanols. Telomerization was used in the production of the fluorotelomer alcohols [FTOHs; F(CF₂)_xCH₂CH₂OH]. These two alcohols were then used to manufacture commercial surfactants and polymers. A major difference between these processes is that telomerization produces exclusively the linear isomer, whereas ECF produces fluorochemicals with a mixture of structural isomers (Kissa 2002).

A major shift in industrial fluorochemical production occurred in 2000, when 3M, the major manufacturer of the POSF-based chemistries, began phasing out its perfluorooctyl materials (3M 2000). This change in production was mirrored in human sera, as concentrations of perfluorooctane sulfonate (PFOS) and perfluorooctanoate (PFOA) began to decrease around the year 2000 (Calafat et al. 2007; Haug et al. 2009; Olsen et al. 2008). These studies generally show a larger decrease in concentrations of PFOS compared with PFOA. PFOS has a longer human serum elimination half-life (Olsen et al. 2007), thus suggesting continued exposure to PFOA without similar exposure to PFOS.

Consistent with continued exposure to PFOA, concentrations of the longer chain PFCAs—perfluorononanoate (PFNA), perfluorodecanoate (PFDA), and perfluoroundecanoate (PFUnA)—have continued to increase in human sera since the 2000 POSF phase-out (Calafat et al. 2007; Haug et al. 2009). The most plausible explanation for continued PFCA exposure, without continued exposure to PFOS, is human exposure to fluorotelomer-based commercial materials.

Exposure to perfluorinated acids from the biotransformation of precursor materials is not a new idea. The sulfonamide-based materials can be biotransformed to PFOS (Xu et al. 2004), and acetate metabolites have frequently been detected in human sera (Calafat et al. 2007; Haug et al. 2009; Olsen et al. 2008). Biotransformation from FTOH to PFCA has previously been observed in a rat model (Fasano et al. 2006; Hagen et al. 1981). Fluorotelomer-based materials are composed of a mixture of chain lengths (Kissa 2002), so biotransformation of these compounds could explain concurrent human exposure to PFCAs of several chain lengths. In addition, the PFCA isomer profile in human sera is suggestive of a fluorotelomer source, because it is almost entirely linear (De Silva and Mabury 2006), with a diminished contribution from specific ECF isomers found to be more bioaccumulative than the linear isomer (Benskin et al. 2009; De Silva et al. 2009).

One class of fluorotelomer-based commercial products with a high potential for human exposure is the polyfluoroalkyl phosphate esters (PAPs). Commercial PAP formulations contain a mixture of fluorinated chain lengths as well as phosphate mono-, di-, and triesters (Begley et al. 2008; Kissa 2002). By synthesizing individual congeners in-house, we simplified this complex mixture to four PAP diesters [diPAPs; x:2, (OH)P(O)(OCH₂CH₂(CF₂)_xF)₂, where x = 4, 6, 8, or 10] and four PAP monoesters [monoPAPs; x:2, (OH)₂P(O)(OCH₂CH₂(CF₂)_xF), where x = 4, 6, 8, or 10]. PAPs are used to greaseproof food-contact paper products and have a demonstrated ability to migrate from packaging into food (Begley et al. 2008). Biotransformation from PAP to PFCA has previously been observed in rats (D’eon and Mabury 2007) and in a microbial system (Lee et al. 2010). Human PAP exposure has also been confirmed by the detection of several diPAP congeners in human sera at microgram per liter concentrations (D’eon et al. 2009). The explicit hypothesis of this work is that PAP biotransformation contributes to the PFCA burden observed in human sera. We tested this hypothesis by determining PAP uptake, elimination, and biotransformation parameters to properly constrain PFCA production from PAP exposure.

Telomerization inherently produces a mixture of perfluoroalkyl chain lengths (Kissa 2002), and in the manufacture of fluorotelomer surfactants this process is optimized to produce predominantly the perfluorohexyl chain length (Dupont 2002). We previously found the diPAP congener profile in human sera to be dominated by 6:2 diPAP (D’eon et al. 2009). Increased exposure to the perfluorohexyl chain length, combined with increased bioavailability of the shorter chain lengths, is consistent with the higher concentrations of 6:2 diPAP observed in human sera.

Interest in human diPAP exposure is related both to the diPAPs themselves and to their potential contribution to human PFCA contamination. We observed PFCA biotransformation products in the blood of the diPAP-dosed animals, which are plotted together with their expected diPAP parent in Figure 1. Despite the observation of relatively high PFCA concentrations, we detected metabolic intermediates only intermittently [see Supplemental Material (doi:10.1289/ehp.1002409)]. This result was surprising because it contrasts the high concentrations of 8:2 FTCA observed after 8:2 FTOH gavage in the rat (Fasano et al. 2006). FTOHs are the direct product of diPAP hydrolysis [as shown in Figure 2B; (D’eon and Mabury 2007; Lee et al. 2010)]. Disparity between concentrations of metabolic intermediates observed after diPAP dosing, compared with FTOH dosing, demonstrates that diPAP exposure does not simply mimic FTOH exposure. The low concentrations of metabolic intermediates observed after diPAP exposure may explain why, in contrast to the frequent detection of sulfonamide oxidation products (Calafat et al. 2007; Haug et al. 2009; Olsen et al. 2008), intermediates in the biotransformation of fluorotelomer materials have never been reported in human sera.

Contrary to the observations after diPAP exposure, we did not observe monoPAPs in the blood of the monoPAP-dosed animals. However, we observed 6:2, 8:2, and 10:2 monoPAPs in feces samples collected in the first 24 hr after administration of the monoPAP gavage dose. Because we did not detect monoPAPs in the feces after IV dosing, the concentrations observed after oral gavage presumably result from residual dosing material. The mass of monoPAPs recovered in the feces after oral gavage could account for < 1% of the administered gavage dose, indicating that the monoPAPs did not pass through the gut unaltered but were either absorbed from the gut or biotransformed within the gut contents. It is difficult to delineate between these two possibilities; however, there is literature precedent for enzyme-mediated hydrolysis within the gut contents, because this has been observed in biotransformation studies of 8:2 fluorotelomer acrylate in rainbow trout (Butt et al. 2010).

Despite the lack of monoPAPs observed in the blood, we consistently detected PFCA biotransformation products in the monoPAP-dosed animals (Table 1). We also detected relatively high concentrations of several metabolic intermediates, a result that is similar to previous studies involving FTOH exposure in the rat (Fasano et al. 2006). This consistency between monoPAP and FTOH exposure supports the hypothesis that monoPAP phosphate ester hydrolysis occurs within the gut contents, followed by FTOH uptake and further oxidation, because this route of exposure would essentially mimic direct FTOH exposure.

Differences between diPAP and monoPAP exposure were not limited to relative concentrations of metabolic intermediates. Elimination kinetics of the short-chain PFCA products in the blood [perfluorobutanoic acid (PFBA) and PFHxA] were significantly slower after diPAP dose administration than after monoPAP (t-test, p < 0.01; Table 1). After diPAP IV dosing, we observed serum elimination half-lives of 3.3 ± 1.2 days and 1.8 ± 0.5 days for PFBA and PFHxA, respectively, whereas after monoPAP IV dosing the observed half-lives were 0.53 ± 0.15 days and 0.23 ± 0.23 days, respectively. We observed a similarly long half-life for PFHxA after diPAP oral gavage (4.2 ± 1.3 days). PFBA and PFHxA have serum half-lives on the order of hours in male rats (Chang et al. 2008; Ohmori et al. 2003); the longer serum half-lives we observed after diPAP dose administration suggest a continual source of these compounds to the animal after the initial bolus dose. This observation is consistent with the gradual biotransformation of diPAP congeners present within the body. Conversely, similarity between PFBA and PFHxA elimination half-lives in the literature (Chang et al. 2008; Ohmori et al. 2003) and those observed after bolus monoPAP exposure supports the hypothesis that most monoPAP biotransformation occurred at the time of dose administration.

If most of the observed diPAP biotransformation occurred within the body, then we can potentially use diPAP concentrations in the blood to predict the resulting PFCA burden. Percent biotransformation will be predicted for the diPAP congeners using the AUC for both the parent diPAP and product PFCA after either IV or oral gavage dose administration using Equation 2:

Because this calculation involves conversion between two chemical species, it was performed using blood concentrations expressed in moles as opposed to mass [see Supplemental Material (doi:10.1289/ehp.1002409)]. These biotransformation yields should be considered conservative because they assume the production of 2 mol PFCA from 1 mol diPAP.

Applying this definition of percent biotransformation to the diPAP experiments resulted in < 1% biotransformation for 4:2 diPAP, with this increasing to about 1% for 6:2 diPAP and to about 10% for 8:2 and 10:2 diPAP (Table 1). These percentages do not represent ratios of parent to product concentrations observed in the blood. Instead, they describe the cumulative PFCA burden imparted to the animal from the cumulative diPAP exposure observed in the blood. These predictions are species specific because they are heavily dependent on the pharmacokinetics of both parent and product. Pharmacokinetic parameters for fluorinated acids are not easily scaled between species (Andersen et al. 2008), so caution must be used when comparing predictions made in the rat with situations in other species, including humans.

Comparison between contaminant temporal trends in human sera with chemical production can delineate between exposures to current-use materials and legacy chemical use. For both PFOS and PFOA, a concurrent decline in the levels of these contaminants in human sera has been reported with the phase-out of the POSF-based materials (PFOA temporal trend shown in Figure 3B) (Calafat et al. 2007; Haug et al. 2009; Olsen et al. 2008). This trend suggests that POSF production was intimately linked to human PFOS and PFOA exposure. The correlation between POSF production and the PFOS found in human sera is clear, because PFOS can be produced from the biotransformation of the sulfonamide commercial materials (Xu et al. 2004), and sulfonamide oxidation products have been observed in human sera (Calafat et al. 2007; Haug et al. 2009; Olsen et al. 2008). The correlation between POSF production and human PFOA contamination is not as clear. The issue is further complicated by the observation that the decrease in PFOS concentrations was greater than that of PFOA (Calafat et al. 2007; Haug et al. 2009; Olsen et al. 2008), despite experimental data indicating faster human serum elimination kinetics for PFOA than for PFOS (Olsen et al. 2007). The best explanation for this phenomenon is that the observed PFOA elimination kinetics are offset by continued PFOA exposure. The same temporal trends also indicate continued exposure to PFNA, PFDA, and PFUnA (Calafat et al. 2007; Haug et al. 2009; Olsen et al. 2008). Although it is difficult to know whether changes in industrial practice by other fluorochemical producers may have also contributed to human PFOA exposure, the temporal trends in human sera are consistent with POSF-based materials being a significant source of human exposure to PFOS and PFOA until 2000. After 2000, fluorotelomer production continued to increase (Figure 3A), and exposure to fluorotelomer-based materials may explain continued exposure to PFOA, together with increasing exposure to the longer chain PFCAs, such as PFNA, PFDA, and PFUnA.

Some attempts have been made to model human PFCA and PFSA exposure via the biotransformation of fluorotelomer and sulfonamide materials and have found this exposure source to be negligible (Fromme et al. 2009; Vestergren et al. 2008). However, these modeling studies are premature because they are limited by a paucity of data regarding concentrations in exposure media as well as a lack of agreement regarding biotransformation yields. In addition, both studies considered human exposure only to residual FTOHs and sulfonamides, and not exposure to the actual commercial materials, which represent the largest potential source of exposure. Using the diPAP commercial surfactants as an example, the high concentrations observed in paper fiber samples (≤ 1,400 ng/g) and wastewater treatment plant sludge (≤ 860 ng/g) indicate that these fluorochemicals are likely present at high concentrations in the indoor environment (D’eon et al. 2009). In migration tests, Begley et al. (2008) observed migration of about 0.5 μg 8:2 diPAP/g butter, implying exposure to about 6 μg 8:2 diPAP for every tablespoon (15 mL) of butter consumed after contact with PAP-treated paper. These exposure estimates demonstrate that exposure to commercial fluorochemicals cannot be ignored and must be better characterized to properly understand human PFCA contamination.