The Applicability of Biomonitoring Data for Perfluorooctanesulfonate to the Environmental Public Health Continuum

1Perfluorooctanesulfonate and its salts (PFOS) are fully fluorinated organic molecules that were produced synthetically by electrochemical fluorination or from the degradation or metabolism of other fluorochemical products produced by electrochemical fluorination. PFOS and its precursors all belong to the larger class of fluorochemicals known as perfluoroalkyl substances and were derived from perfluorooctanesulfonyl fluoride (POSF), the basic chemical building block for many sulfonyl-based fluorochemicals. POSF is used primarily as an intermediate to synthesize numerous fluorochemicals, including PFOS.

23M Company produced POSF, PFOS, and POSF-related materials for over 40 years. These materials exhibit chemical and physical properties that can be substantially different from hydrogenated analogs, and perfluorinated portions of the molecules have extreme resistance to environmental and metabolic degradation. The strength of carbon–fluorine bonds contributes to the extreme stability and unique properties of PFOS. PFOS, as a surfactant, has high surface tension–reducing properties, lower water and oil solubility, and is a relatively strong organic acid. The commercial uses of PFOS included predominantly surface treatments for soil- and stain-resistant coating on fabrics, carpets, and leather, coatings on paper and packaging products for grease and oil resistance, including food contact papers, and performance chemicals uses such as fire-extinguishing foam concentrates, mining and oil surfactants, electroplating and etching bath surfactants, household additives, chemical intermediates, coatings and coating additives, carpet spot cleaners and insecticide raw materials. Total worldwide POSF production by 3M Company in 2000 was approximately 8 million pounds. However, on 16 May 2000, 3M Company announced that it would globally phase out the perfluorooctanyl chemistry used to produce certain repellents and surfactant products, which included the manufacture of PFOS and related compounds. 3M steadily reduced their production volume and discontinued the manufacture of most PFOS and POSF-based chemicals by 31 December 2002. Manufacture of PFOS for certain uses for which no substitutes are available are continuing by non-U.S. producers.

3The purpose of this article is to provide a brief overview of the biomonitoring data for PFOS in the context of the public health paradigm as discussed in. We place emphasis on reviewing PFOS data related to those parts of the public health paradigm covering exposure, dose, and potential health outcomes in an effort to develop a view of the sufficiency of the PFOS biomonitoring data for use in public health.

4PFOS has been measured primarily in human blood serum; however, data are also available on PFOS in human whole blood, plasma, liver, cord blood, breast milk, and seminal plasma. PFOS was measured in the liver and serum of cadavers. The average PFOS serum and liver data for each of 23 paired samples (serum and liver from the same individual) showed a good correlation for both male and female donors, and the mean liver to serum ratio was 1.3:1 [95% confidence interval (CI), 0.9:1–1.7:1). Mean PFOS levels for male and female donors were similar for both serum (male = 18.2 ng/mL; female = 17.2 ng/mL) and liver (male = 19.2 ng/g; female = 28.4 ng/g).

5Quantitative analysis of PFOS from selected matrices is challenging and has been conducted using various extraction techniques and instrumentation methods. Matrix effects and isomeric forms should be taken into account. Extracted matrix calibration curves (i.e., not just water blanks) should be used to ensure the accuracy of laboratory analysis. Limits of quantitation are in the low nanogram per milliliter range; however, interlaboratory variability is significant, based on the results of the PERFORCE (PERFluorinated ORganic Chemicals in the European environment) interlaboratory study. In the PERFORCE study, 17 participating laboratories produced values for a spiked serum standard that varied with a relative standard deviation of 31.5%. For whole blood, 11 participating laboratories produced values that varied with a relative standard deviation of 56.1%. Liquid–liquid extraction techniques, or a variation thereof that involves ion pairing, have been used for extraction of PFOS from biological matrices and have involved internal standards through the complete extraction. Chromatography and quantitation have typically been accomplished using high-performance liquid chromatography–mass spectrometry (HPLC-MS) in negative ion mode, with quantitation of the parent ion (LC-MS) or transition ions (HPLC-MS/MS). Solid-phase extraction techniques and column-switching methods are examples of recent improvements in the quantitation of PFOS. Mass-labeled internal standards such as 18O2-PFOS will result in improved quantitation.

6Organically bound fluorine was reported in human blood in 1968 and before the development of analytical techniques for specific fluorochemical analytes, total organic fluorine was used to measure organic fluorine in human blood. In an attempt to speciate the organic fluorine in blood using 19F nuclear magnetic resonance, tentatively suggested the presence of perfluorooctanoate and also noted that branching or the presence of a sulfonate was a possible interpretation of their findings. As HPLC-MS methods were developed, “blank” human serum was found to contain PFOS in 1997. Subsequently, PFOS was found distributed widely in human serum and fish-eating wildlife serum and liver.

7The pharmacokinetic properties of PFOS are favorable for using serum PFOS concentration as a measure of internal dose. Good absorption, lack of known metabolism, distribution primarily in extracellular space, high serum protein binding (albumin and beta-lipoproteins), and poor elimination in all species studied combine to establish serum PFOS concentration as an integration of exposures from various sources. In addition, serum PFOS concentrations can be directly associated with effects in toxicology studies. PFOS serum concentrations can be used as a marker for total exposure in occupational medical surveillance studies.

8Animal studies indicate that PFOS is well absorbed orally and distributes mainly in the serum and liver, with liver concentrations being potentially several times higher than serum concentrations. The volume of distribution at steady state, as measured in cynomolgus monkeys, is approximately 200 mL/kg, suggesting distribution primarily in extracellular space. PFOS is highly bound to albumin and has affinity for binding to β-lipoproteins as well as albumin and liver fatty acid–binding protein. PFOS is poorly metabolized and excreted and undergoes extensive enterohepatic circulation. The elimination half-life of PFOS is approximately 100 days in rats and 100–200 days in cynomolgus monkeys and appears to be approximately 5 years in human serum.

9The mechanisms and pathways leading to the presence of PFOS in human blood are not well characterized but likely involve environmental exposure to PFOS or to precursor molecules and residual levels of PFOS or PFOS precursors in industrial and commercial products. Potential sources of human exposure to PFOS may have included manufacturing operations and waste streams of POSF-based fluorochemical products and the use or degradation of some final commercial and consumer products, including indirect food-contact applications. Other potential sources may include exposure to airborne PFOS, POSF, and N-alkyl-perfluorooctanesulfonamides, surface water, sediments and sludge, and indoor air and dust. PFOS has been identified in serum and tissue samples from both occupationally and nonoccupationally exposed human populations in various countries, in various species of wildlife in many parts of the world, and in surface waters and other environmental media in various countries. PFOS is highly persistent in the environment and has also been shown to bioconcentrate in fish and biomagnify to an extent in the food chain.

10The only two countries in which multiple biomonitoring investigations of general population PFOS levels have been undertaken are the United States (Table 1) and Japan (Table 2). The total number of samples presented in these two countries represents approximately 90% of the individual analyses published in the literature. The similar distribution of averages (as well as similar ranges) from individual studies presented in Table 1 suggests that serum PFOS concentrations are relatively comparable across geographic regions and age groups in the United States. However, the most recent pooled serum data indicate modest differences by sex as well as ethnicity. In addition, the range of serum concentrations in the U.S. populations indicates that some individuals, albeit a small number, may have had relatively higher exposure levels than the majority of individuals sampled. Average PFOS concentrations were lower among the Japanese populations presented in Table 2. In one Japanese study, observed sex-related differences in serum PFOS concentrations, with males approximately 2-fold higher than females. Furthermore, reported higher concentrations among premenopausal than postmenopausal women in another study. These findings were not seen in general populations in the United States. In the only published study of its kind, PFOS was measured in 15 pairs of maternal and cord blood (fetal) samples from Japan. PFOS concentrations in maternal samples ranged from 4.9 to 17.6 ng/mL, whereas those in fetal samples ranged from 1.6 to 5.3 ng/mL, with a high degree of correlation between pairs (r = 0.94). Only two studies in Tables 1 and 2 examined time trends. Serum PFOS concentrations increased 3-fold over a 25-year time period in Miyaga, Japan. Median PFOS concentrations increased approximately 25% between 1974 (median, 25 ng/mL) and 1989 (median, 33 ng/mL) for 58 individuals living in the vicinity of Hagerstown, Maryland. However, only a 9% increase in median PFOS concentrations occurred in two nonpaired populations (n = 120 each) residing in the same area and time period. PFOS concentrations did not appear to increase between 1989 and 2001 for this region. Individual samples from three large data sets with different age groups predominate the United States findings, as reported in Table 1. Sera from children (age 2–12 years, n = 598) in 23 states, adult blood donors (age 20–69 years, n = 645) from six municipalities in the United States, and elderly (age 65–96 years, n = 238) Seattle residents were analyzed for PFOS using identical laboratory methods with comparable findings. Geometric means were 38 (95% CI, 36–39), 35 (95% CI, 33–37), and 31 (95% CI, 29–33) ng/mL, respectively.

11Although comparable in average PFOS concentrations, a small number of individuals in each studied population had relatively higher levels than the majority of individuals sampled. The factors that would lead to higher serum PFOS concentrations in some individuals are not completely understood. Some factors that may affect serum PFOS concentrations include proximity to sources of manufacture and use, length of residence in these latter areas, potential product exposures, and possible food and environmental sources. Analysis of 54 pooled serum samples collected from 1,832 participants ≥ 12 years of age in the National Health and Nutrition Examination Survey (NHANES; ) collected in 2001–2002 suggested possible concentration differences by sex (males higher) and ethnicity, in that non-Hispanic whites had statistically significantly higher concentrations compared with non-Hispanic blacks and Mexican Americans. An association of higher PFOS serum concentrations with higher fish consumption has been noted in humans as well as wildlife. It is notable that PFOS serum concentrations did not strongly correlate with serum concentrations of metabolites of N-alkyl-perfluorooctanesulfonamide molecules known to be present in products as manufacturing residuals or from degradation of products. This lack of correlation could be due to differences in the pharmacokinetic behavior of N-alkyl-perfluorooctanesulfonamides compared with PFOS or suggest that product exposures may not be a primary source of PFOS in the body.

12PFOS data sets have been published for other countries; however, few individual samples have been analyzed. Most of these data were reported by. In general, the majority of these PFOS concentrations were less than those reported for the U.S. general populations displayed in Table 1. The highest mean PFOS concentration reported by Kannan et al. was found in samples collected from Poland [male, 55 ng/mL (n = 10); female, 33 ng/mL (n = 15)]. Lower mean PFOS concentrations were reported by Kannan et al. for Korea [male, 27 ng/mL (n = 25); female, 15 ng/mL (n = 25)]; Belgium [male, 18 ng/mL (n = 16); female, 11 ng/mL (n = 4)]; Malaysia [male, 13 ng/mL (n = 16); female, 12 ng/mL (n = 7)]; Brazil [male, 14 ng/mL (n = 10); female, 11 ng/mL (n = 17)]; Italy [male, 4 ng/mL (n = 42); female, 4 ng/mL (n = 8)]; Colombia [male, 8 ng/mL (n = 31); female, 8 ng/mL (n = 25)]; and India [male, 3 ng/mL (n = 34); female, 3 ng/mL (n = 11)]. In a pilot study, reported mean concentrations of 30 and 28 ng/mL in 21 and 35 female and male Canadians, respectively. reported geometric mean concentrations of 17 and 16 ng/mL in 40 men and 26 women, respectively, from Sweden. reported a mean concentration of 5 ng/mL in 30 adults from Sri Lanka and 0.1 ng/mL in seminal plasma, with a correlation of 0.6 for PFOS between the two matrices. reported a mean concentration of 52.76 ng/mL from 85 Chinese samples with considerable differences between nine cities (ranges of means between 3.7 and 79.2 ng/mL). The higher mean PFOS serum concentration was attributed to proximity to industrial production. Other findings reported by included several blood bank pooled serum samples obtained in 1998 from Belgium, the Netherlands, and Germany that had mean (range in parentheses) serum PFOS concentrations of 17 (5–22) ng/mL, 53 (39–61) ng/mL, and 37 (32–46) ng/mL, respectively.

13PFOS serum levels have been measured in 3M employees involved in both the manufacturing of perfluorochemicals and the processing of these compounds into products. Beginning in the late 1970s, measurements were for total organofluorine. Since the mid-1990s, measurement of serum PFOS concentrations have been performed as part of employee medical surveillance examinations at the 3M manufacturing facilities in Decatur, Alabama, and Antwerp, Belgium. Between 1994 and 2000, mean PFOS concentrations approximated 1.0–2.5 μg/mL (1,000–2,500 ng/mL) and ranged between < 0.1 and 12.8 μg/mL, depending upon the job classification. Because employee participation is voluntary, biomonitoring data from these medical surveillance programs may not have provided an adequate characterization of the distribution of serum PFOS concentrations because of possible non-response bias. A study that randomly sampled the Decatur worker population indicated that serum PFOS concentrations measured during the course of medical surveillance examinations likely presented an unbiased analysis of the overall serum PFOS distribution of the workers. This study showed that mean PFOS serum concentrations for cell and chemical operators were 2.90 μg/mL (geometric mean = 1.97 μg/mL; range, 0.33–6.84 μg/mL) and 1.78 μg/mL (geometric mean = 1.48 μg/mL; range, 0.47–7.26), respectively. These values are approximately 50 times higher than mean (arithmetic or geometric) concentrations measured in the U.S. general population.

14Because multiple sources and routes of exposure were probable, estimating external worker PFOS exposure is problematic. Employees may have been exposed to POSF and/or other perfluorochemicals in the manufacturing environment by one or more routes. The primary route of exposure may have varied among employees and depended on several factors, including process conditions, job tasks, work location, personal hygiene, personal habits, and general work practices. Biomonitoring allows for the assessment from all routes of exposure.

15The toxicological profile of PFOS has been studied extensively. Several repeat-dose toxicology studies with PFOS in rodents and nonhuman primates have indicated the potential to reduce body weight and bodyweight gain, increase liver weight, and reduce serum cholesterol. The dose–response curve for mortality in repeat-dose studies is very steep for sexually mature rats and primates as well as neonatal rats and mice exposed in utero. Microscopic changes attributable to PFOS include hepatocellular hypertrophy and vacuolation in rats and monkeys, and hepatic necrosis at lethal doses in rats. PFOS has been tested for genotoxic activity in a battery of microbial and mammalian systems that included assays for induction of gene mutation (Salmonella typhimurium and Escherichia coli), gene conversion (Saccharomyces cerevisiae D4), chromosomal aberrations (human lymphocytes and mouse bone-marrow micronuclei), and unscheduled DNA synthesis (primary rat hepatocytes). PFOS did not show genotoxic activity in any of these assay systems. In a 2-year dietary study of PFOS in Sprague-Dawley rats, hepatocellular adenoma was increased in the high-dose (20 μg potassium PFOS/g feed) males and females. Thyroid follicular cell adenomas were observed in male rats in the high-dose, stop-dose group for which PFOS was eliminated from the diet after 1 year. Decreased serum estradiol was noted in a 6-month monkey study at treatment levels that had a corresponding reduction in serum total cholesterol. Alterations in serum thyroid hormones have been reported and may be due in part to negative bias in analog measurements and competition between PFOS and thyroid hormones for binding in serum.

16In a two-generation reproduction study in rats, mating and fertility were not affected; however, neonatal survival, pup birth weight, and growth of pups in lactation were decreased, and developmental delays were noted. The no observed adverse effect level for these effects was 0.1 mg/kg/day. Reduced postnatal survival and body weight gains may result from in utero exposure. Prenatal developmental toxicity studies of PFOS have been conducted in rats, mice, and rabbits. Prenatal effects in rats administered PFOS during gestation included statistically significant decreases in fetal body weight and statistically significant increases in external and visceral anomalies, delayed ossification, and skeletal variations. Maternal toxicity in rats exposed to PFOS during gestation included clinical signs of toxicity and reduction in body weight and food consumption. In rabbits administered PFOS during gestation, statistically significant reductions in fetal body weight and statistically significant increases in delayed ossification were observed; signs of maternal toxicity consisted of abortions and reductions in body weights and food consumption. On the whole, the prenatal developmental effects noted in these studies are consistent between studies, and their significance is somewhat mitigated by the fact that they occur in the presence of maternal deficits in weight gain and feed consumption.

17A number of studies have been conducted to investigate the possible modes of action of PFOS. Induction of peroxisome proliferation and associated peroxisomal enzymes, activation of nuclear receptors, interference in lipid metabolism and decreases in serum cholesterol, interference in mitochondrial bioenergetics, delays in lung maturation, inhibition in gap junctional intercellular communication processes, alterations in calcium channels, and alterations in thyroid hormone homeostasis have all been investigated as possible modes of action; however, at present, the mechanisms of action related to the toxicity of PFOS are still not clearly understood. Gene array studies have also been conducted in the rat to investigate changes in mRNA transcriptional responses to PFOS treatment. Gene expression studies have demonstrated up-regulation of transcription products primarily related to fatty acid metabolism, e.g., β-oxidation pathways, as well as cytochromes P450 and some hormonal regulatory gene transcripts.

18In addition to toxicologic studies, epidemiological and medical surveillance studies of exposed 3M Company fluorochemical workers have been conducted by 3M Company for over 25 years. This set of POSF production workers (with potential exposure to PFOS) is the only group reported on from a human health perspective. Clinical tests in medical surveillance examinations in workers have not shown consistent patterns of associations between PFOS serum levels and hematology, hormonal, and clinical chemistry parameters. A cohort mortality study of the 3M Decatur (Alabama) manufacturing facility showed no statistically significant excess mortality for most types of cancer and for nonmalignant causes. However, bladder cancer mortality was elevated (three observed vs. 0.2 expected; standardized mortality ratio 16.12; 95% CI, 3.32–47.14) among male workers who had worked in high PFOS exposure jobs for a minimum of 1 year. It was unclear whether PFOS or other fluorochemicals contributed to the excess of bladder cancer deaths. To further investigate this association, mailed a questionnaire to all living members of the original cohort with validation of reported bladder cancers through medical record review when permitted by the subject. A total of 11 cases of primary bladder cancer were identified for the cohort compared with 8.6 expected (standardized incidence ratio 1.28; 95% CI, 0.64–2.29) based on U.S. National Cancer Institute incidence rates. Analyses by duration worked showed no definitive trend.

19Worker insurance claims data categorized as episodes of care have also been evaluated. For a priori interests, the observed to expected episodes of care experience were comparable for the Decatur fluorochemical and a neighboring film plant (control) employee population for liver tumors, bladder cancer, thyroid and lipid metabolism disorders, and reproductive, pregnancy, and perinatal disorders, and higher for biliary tract disorders and cystitis recurrence. Non-a priori associations among the fluorochemical plant workers included benign colon polyps, malignant colorectal tumors, and malignant melanoma. Research is currently being conducted to further investigate these associations.

20Biomonitoring data can be used to examine regional differences and time trends. PFOS concentrations in Charlotte, North Carolina, were the highest of the geographical regions investigated by 3M, and a preliminary screening study conducted by Centers for Disease Control and Prevention to validate methodology showed slightly higher values in Atlanta, Georgia. Of six North American areas tested, reported the largest mean concentration in the troposphere of N-methyl perfluorooctanesulfonamidoethanol (359 pg/m3) was from the Griffin, Georgia, location. Although the southeastern United States is an area of high carpet and fabric production, it was not possible to establish that this was the reason for the somewhat higher values measured. In addition few data are available that can describe exposure trends. It is assumed that with the voluntary discontinuation of manufacturing of PFOS and POSF-based chemicals by 3M Company, PFOS exposures will eventually diminish.

21A review of the data for PFOS related to the exposure, dose, and potential health outcomes parts of the public health paradigm suggest that the PFOS biomonitoring data are relevant for use in public health. Pharmacokinetic data are available to relate serum PFOS concentrations to toxicity and, therefore, to potential health outcomes that may occur across the spectrum of development from in utero through adulthood as depicted in the Public Health Paradigm. Analytical methods are sufficiently sensitive, precise, and accurate for the measurement of general-population serum PFOS as a biomarker of exposure. The database for potential health effects is reasonably robust and includes most toxicologic end points as well as several studies of individuals occupationally exposed to PFOS.

22The biomonitoring and toxicology data related to PFOS were presented at the International Biomonitoring Workshop on 21–22 September 2004 at Research Triangle Park, North Carolina. Based on the outcomes of that workshop and a general review of the data available for PFOS, the following types of information are considered to be of additional value in understanding the distribution of PFOS in general populations.

23Strengthen the database to allow conversion of whole blood and plasma PFOS concentration values to serum PFOS concentration. Various reports give whole blood measurements converted to estimate serum concentrations by making the assumption that all PFOS is in serum.

24Strengthen the relationship between serum and liver concentrations of PFOS. Although this information is available from toxicology studies, the human data are very limited.

25Bank samples of blood, plasma, or serum for future needs, including early screening investigations (e.g., NHANES).

26Obtain additional data for children, especially children < 2 years of age [e.g., ].

27Obtain matched serum and urine samples from humans and research animals, where possible (e.g., NHANES), to better understand toxicokinetics.

28Obtain more placental, cord blood, and milk samples [e.g., ].

29Confirm potential ethnic and sex differences in distributions of blood PFOS concentrations.

30Biomonitor additional populations with specific potential occupational, dietary, or consumer exposures to PFOS and related materials; e.g., high consumers of fish, workers with downstream exposure to POSF-based products, and consumers of POSF-based products.

31Build a physiologically based pharmacokinetic (PBPK) model for human handling of PFOS that could explain uptake, clearance, and distribution and exposure pathways by reverse PBPK modeling.

32This article has been modified from the original manuscript published online. In the section “Exposure Assessment,” the number of pooled samples has been changed from 50 to 54, and the number of participants ≥ 12 years of age has been changed from 1,836 to 1,832. In Table 1, column 5, line 23, “13 (pooled)” has been changed to “14 (pooled).”