Integrating Biomonitoring Exposure Data into the Risk Assessment Process: Phthalates [Diethyl Phthalate and Di(2-ethylhexyl) Phthalate] as a Case Study

1The general structure of phthalates, diesters of phthalic acid, manufactured by reacting phthalic anhydride with alcohols of desired carbon-chain length, is shown in Figure 1. R and R′ are ethyl groups for diethyl phthalate (DEP) and 2-ethylhexyl groups for di(2-ethylhexyl) phthalate (DEHP). DEP (CAS no. 84-66-2) is used as a plasticizer for cellulose acetate, as a solvent, and as a carrier for fragrances in cosmetics and other consumer products. DEHP (CAS no. 117-81-7) is used primarily as a plasticizer in flexible vinyl, which is used in consumer products, flooring and wall coverings, food contact applications, and medical devices.

2The potential for exposure is, to a certain extent, a consequence of the physical and chemical properties of each phthalate. As molecular weight increases, vapor pressure, water solubility, and dermal uptake are reduced. The major route of human exposure for most phthalates is ingestion; exposure by inhalation, through drinking water, and via dermal contact tends to be limited. After ingestion, phthalates are metabolized to their corresponding hydrolytic monoesters and may further metabolize to more hydrophilic oxidative products. These metabolites can be excreted unchanged or can undergo phase II biotransformation to glucuronide conjugates. Metabolites and not the parent diesters are likely the bioactive species.

3Evidence of human hazard associated with exposure to phthalates is limited, and risk assessments have been based primarily on results of animal studies. Administration of some phthalates to rodents caused liver effects, including increased weights, elevated enzyme levels, histologic changes, and tumors, associated with peroxisomal proliferation, that is, specifically with peroxisome proliferator–activated receptor α agonism, a process related to metabolism of cholesterol and fatty acids. Due in part to species-specific metabolic differences, the relevance of these effects to humans is questionable. Nevertheless, liver effects have been used to establish no observed adverse effect levels (NOAELs) for risk assessment. Evidence also exists that some phthalates and their metabolites affect reproduction and development, particularly in male rats (e.g., epididymal malformations or absence of the epididymis, testicular lesions, increased incidence of hypospadias, cryptorchidism, decreased anogenital distance, delayed preputial separation, and retention of thoracic nipples), apparently by a process involving inhibition of androgen biosynthesis.

4Because DEP is used in personal care products, dermal toxicity is of interest. Primary dermal irritation with undiluted DEP has not been reported in humans. DEP was not a dermal sensitizer in healthy human volunteers, although sensitization was reported in some studies, mostly involving persons with skin diseases. No reports exist of oral or inhalation toxicity of DEP or of any adverse effects in humans exposed exclusively to DEP. The oral reference dose (RfD) for DEP, 800 μg/kg/day, was derived from a NOAEL of 750 mg/kg/day based on reduced growth rate, food consumption, and increased organ weights in rats. No evidence of other effects in animals at lower NOAELs exists.

5Information on the oral toxicity of DEHP is limited to mild abdominal pain and diarrhea in two persons who ingested single large doses. No reports exist of dermal or inhalation toxicity of DEHP in adult humans, and DEHP is neither a dermal irritant nor a sensitizer. DEHP does not appear to be readily absorbed through human skin. Lung disorders, resembling hyaline membrane disease, were observed in three newborns who, as preterm infants, received ventilation therapy involving polyvinyl chloride tubing. The U.S. EPA classifies DEHP as a probable human carcinogen (B2) and, based on evidence of increased liver weight in rodents, established the RfD at 20 μg/kg/day. revised its classification from “probable” to “not classifiable” after determining that the mode of action was irrelevant to humans. Because of the controversy regarding relevance of DEHP-induced rodent liver cancer to humans, cancer risk will not be discussed in this article.

6In recent years the potential reproductive and developmental effects of DEHP have received more attention than the carcinogenic effects. In particular, developing rats are more sensitive to the testicular toxicity of DEHP than are older animals. Exposure of rats to DEHP during the late gestational period affected male reproductive development with a NOAEL of 5–8 mg/kg body weight/day. This NOAEL was used to assess the potential for human reproductive risks associated with DEHP exposure. Similarly, a previously determined NOAEL of 3.7 mg/kg/day for testicular effects was used by the European Union’s Scientific Committee for Toxicity, Ecotoxicity, and the Environment (CSTEE) as the basis for a tolerable daily intake (TDI) of 37 μg/kg/day.

7Phthalates are widely used in laboratory equipment, and contamination is possible. Sample contamination problems are greatly minimized when phthalate metabolites are measured. To select the most appropriate biomarkers of exposure, understanding the toxicokinetics of individual phthalates is fundamental. Although differences in absorption of phthalates exist, we address only metabolic differences in this article.

8In rats, monoethyl phthalate (MEP) is the principal urinary metabolite of DEP; smaller amounts of phthalic acid and DEP are also found. Metabolism in humans is assumed to be similar. Elimination half-lives of DEP and MEP have not been experimentally defined but, like DEHP and its hydrolytic metabolite mono(2-ethylhexyl) phthalate (MEHP), are assumed to be a few hours. These findings suggest that MEP is the most sensitive and specific biomarker of exposure to DEP.

9More than 20 urinary metabolites of DEHP have been proposed. In rodents these consist primarily of terminal oxidation products. In humans the principal DEHP metabolites are side-chain–oxidized metabolites of MEHP. In two cancer patients receiving an infusion of a platelet concentrate containing DEHP, > 50% of the DEHP disappeared from the blood in about 30 min and appeared as DEHP derivatives in urine within 6 hr. In another study of two volunteers who received DEHP orally, the urinary elimination half-life of DEHP was estimated to be 12 hr. The urinary excretion of DEHP metabolites in one person after three oral doses of D4-DEHP followed a multiphase elimination model. For the first 4–8 hr, excretion half-lives were approximately 2 hr for mono(2-ethyl-5-hydroxyhexyl) phthalate (MEHHP), mono(2-ethyl-5-oxohexyl) phthalate (MEOHP), and MEHP. Fourteen to eighteen hours postadministration, half-lives were 5 hr (MEHP) and 10 hr (MEHHP and MEOHP). MEHHP was the major metabolite initially; other metabolites, mono(2-ethyl-5-carboxypentyl) phthalate and mono(2-carboxymethylpentyl) phthalate, were more abundant starting 12 hr after exposure. The higher urinary concentrations in humans of MEOHP and MEHHP than of MEHP suggest that oxidative metabolites may provide greater analytical sensitivity than MEHP. Furthermore, oxidative metabolites cannot be formed as a result of sampling contamination and may be more advantageous as biomarkers of exposure to DEHP than MEHP. DEHP, seldom found in blood or urine except as a consequence of contamination, is not recommended as a biomarker in studies involving these media but may be useful in studies involving other media (e.g., feces).

10Highly specific, sensitive, accurate, and precise analytical methods using isotope-dilution–high-performance liquid chromatography (HPLC) coupled with tandem mass spectrometry for measuring parts-per-billion levels of selected phthalate metabolites in biologic matrices have been described.

11Urine (as matrix) and phthalate metabolite concentrations (as biomarkers) represent the most common approach to investigating phthalate exposure in humans. Phthalate concentrations in blood have been reported, but most assessed concentrations of diesters. Data from such studies are often questionable because of the potential for diester contamination. Consequently, methods were developed to measure concentrations of metabolites in serum (Kato et al. 2004a;; ), breast milk, saliva, and human amniotic fluid. Data from media other than urine could also be used for exposure assessment, but it might be more difficult to collect the samples. Thus, these alternative media may not readily lend themselves to large screening programs but may be useful in specific situations.

12Defining human exposure to phthalates requires measuring concentrations of parent compounds or their metabolites in urine and other biomatrices as well as understanding the pharmacokinetics of individual phthalates. The Centers for Disease Control and Prevention (CDC) collects urinary metabolite data for the general population, primarily through the National Health and Nutrition Examination Survey (NHANES), an ongoing national survey designed to evaluate the health and nutritional status of the U.S. population. NHANES is unique in its ability to examine public health issues that can be addressed through physical and laboratory examinations. NHANES 1999–2000 and 2001–2002 provided nationally representative population-based urinary phthalate metabolite data, based on one specimen per participant, for selected demographic groups in the United States. However, young (i.e., < 6 years of age) and older individuals (i.e., > 60 years of age) were not represented in the population sampled, and no data on prenatal exposures were collected.

13Data from NHANES and other studies conducted in the United States and abroad have confirmed that human exposure to phthalates is widespread (Tables 1–3). Some situations, not specifically addressed by large surveys such as NHANES, may lead to phthalate exposures well above those found in the general population. Examples include the use of certain medications with enteric coatings containing phthalates [e.g., DEP, dibutyl phthalate (DBP)] or related to using DEHP in medical devices.

14Studies of specific health effects with environmental phthalate exposures using urinary metabolite concentrations as exposure surrogates exist. However, these epidemiologic data are limited and drawing firm conclusions has been difficult.

15Previous exposure assessments for phthalates have been indirect, that is, relying on surveys of product use, measuring phthalates in various media, estimating human contact, and pharmacokinetic assumptions based on animal data. In contrast, direct methods using urinary metabolite concentrations as biomarkers for phthalate exposure may provide the most accurate assessments because these concentrations represent an integrative measure of exposure from multiple sources and routes and can be used to calculate phthalate exposure in the general and specific populations.

16For phthalate metabolite data, two calculation methods produced similar results. For illustrative purposes, we show the method of as expressed by:

17in which DI is the daily intake in milligrams per kilogram per day; UE is the creatinine-corrected urinary metabolite concentration in micrograms per gram; CE is the creatinine clearance rate, normalized for body weight, in milligrams per kilogram per day; Fue is the molar conversion factor that relates urinary excretion of metabolite to diester ingested; and MWd and MWm are the molecular weights of diester and metabolite, respectively. For these calculations, we set CE at 20 mg/kg/day for adults, 11 mg/kg/day for children, and 9.8 mg/kg/day for infants. We set Fue at 0.69 mg/kg/day for DEP (as MEP), 0.13 mg/kg/day for DEHP (as MEHP), 0.23 mg/kg/day (as MEHHP), and 0.15 mg/kg/day (as MEOHP).

18An Fue value for DEP has not been determined experimentally but is assumed to be similar to the value determined for DBP. By contrast, urinary excretion of DEHP metabolites has been studied after oral and intravenous administration. The earliest reports of Fue for DEHP metabolites came from studies that had either analytical limitations or small sample sizes. Subsequently, an MEHP Fue value was determined by HPLC–mass spectrometry from a study involving seven individuals dosed orally with both 13C-DEHP and 13C-diisooctyl phthalate. Because the 13C-MEHP and 13C-monooctyl phthalate signals co-eluted, Fue for these species could not be determined separately, and the MEHP value of 0.13 is the average. We used Fue for the oxidative DEHP metabolites (MEHHP, 0.23; MEOHP, 0.15) from a study of one adult man given three single oral doses of D4-DEHP; the estimated Fue for MEHP was 0.06, about half the value used in the calculations in this case study.

19The first data on urinary phthalate metabolite concentrations, including MEP and MEHP, reported in a U.S. population of 289 adults from NHANES III, were used to calculate exposures to the corresponding phthalate diesters. Subsequently, the CDC reported U.S. nationally representative urinary concentrations of seven phthalate metabolites in 2,540 participants of NHANES 1999–2000 and of 10 phthalate metabolites in 2,782 participants of NHANES 2001–2002. The frequencies of detection of individual phthalate metabolites were similar. However, the median concentration of MEP was almost 2-fold lower in NHANES 1999–2000 and 2001–2002 than in NHANES III. These differences may have reflected reduced exposures to DEP or have been related to differences in sample sizes. In contrast, the MEHP concentrations remained essentially constant, although they were highest in NHANES 2001–2002 (Table 1). MEHHP and MEOHP were only measured in NHANES 2001–2002. Their median concentrations were 5-fold (MEHHP) and more than 3-fold (MEOHP) higher than the median MEHP concentration. The NHANES 1999–2000 and 2001–2002 data, stratified by age, gender, or ethnicity, indicated some differences in urinary concentrations of phthalate metabolites. For MEHP, MEHHP, and MEOHP, children exhibited higher urinary concentrations than adults, although when accounting for creatinine clearance, the calculated external exposures were similar (Table 2).

20Urinary concentrations of DEP and DEHP metabolites in other smaller groups were largely consistent with the NHANES 1999–2002 data (Table 3). In general, differences between various segments of the population were smaller than the differences across the population, that is, from lowest to the most highly exposed individuals. The underlying explanation for the range of exposures is unknown but may be related to individual lifestyle choices. However, selection of study subjects (at least for NHANES) did not exclude those occupationally exposed, and specific situations may contribute to higher exposures for some individuals. Median urinary MEP concentrations in 85 German children and adults were approximately half those in NHANES 1999–2002. By contrast, median urinary MEHP, MEHHP, and MEOHP concentrations were approximately twice those in NHANES 1999–2002, but 95th percentile values were similar (Table 1). Whether these findings reflect differences in sampling (e.g., first morning vs. non-first morning voids, nonrepresentative nature of the population examined in Germany) or in exposure patterns between the United States and Germany is unknown.

21Estimates of DEP exposure resulting from its use in personal care products, based on conservative assumptions, were not realistic (730 μg/kg/day from fragrances and 100 μg/kg/day from personal care products). With food as the largest identified contributor to exposure for most individuals, calculated median DEP exposure ranges were 2–6 μg/kg/day for most of the population, with somewhat higher estimates for toddlers and lower estimates for infants. For DEHP, relying heavily on a previous study, estimated DEHP exposure ranges within the general population were 3–30 μg/kg/day, with higher exposures likely in occupational settings and the highest associated with certain medical procedures. Estimates from other researchers also fall in this range.

22For DEP a comparison of the biomarker-based and indirect approaches indicates that, in adults, mean estimates derived by indirect methods were about half the mean exposures calculated from bio-marker-based data (Table 4). Because this indirect approach did not consider DEP exposure from cosmetics use, these differences are expected. The 95th percentile exposures calculated from biomonitoring data were above these indirect estimates but far below unrealistic estimates of exposures from cosmetic and personal care products. One might hypothesize that exposure from sources other than personal care products accounts for approximately half the mean total DEP exposure, with exposure from personal care products comprising the remainder. In agreement with this hypothesis, children have lower exposures to DEP than adults (Tables 1, 2). Three overall conclusions emerge from this example: a) indirect methods can provide realistic estimates of exposure only if reasonable assumptions are used; b) use of biomonitoring data can yield precise exposure estimates because it does not require overly conservative assumptions; and c) it may identify situations in which not all potential sources of exposure were considered.

23For DEHP, urinary MEHP data produced estimates of mean exposure that were lower than those using the indirect methods, although the 95th percentile values were similar (Table 4). Using DEHP oxidative metabolite data, the estimated DEHP exposures are about twice those calculated from MEHP data (Tables 1, 2, 4). That mean DEHP exposures within the general population, calculated from urinary metabolite data, are approximately 4-fold lower than the indirect estimates may be due, in part, to reliance on older measurements of phthalates in various media, particularly food, as the basis for indirect estimates. Conservatism may also be introduced by assumptions about absorption based on results of animal studies. Nevertheless, this comparison suggests that, for DEHP, all relevant sources of exposure were taken into consideration when using the indirect approach.

24Biomonitoring data can also be used to address the exposure component of risk assessment. In risk assessment, exposure estimates are compared with NOAELs that for phthalates were from studies in rats. Important and controversial issues relating to these hazard data include choice of species, identification of critical end points, and relevance to humans. Discussing those issues in detail is beyond the scope of this article. Rather, this section relates results of risk assessments based on phthalate exposures calculated from urinary metabolite data to conclusions of previous risk assessments.

25The most reasonable indirect estimates of mean exposure to DEP were 2–6 μg/kg/day, depending on the ages of the groups considered and neglecting consideration of cosmetics and personal care products. Estimates of DEP exposure in the general population, based on biomonitoring data, are 5.4 μg/kg/day, with a 95th percentile of 64.7 μg/kg/day (Table 4). Thus, indirect and biomarker-based methods produced comparable estimates and indicated that within the United States most individuals are exposed to DEP levels well below the RfD (800 μg/kg/day).

26For DEHP, indirect estimates of mean exposure were 5.8–8.2 μg/kg/day and a range of 3–30 μg/kg/day. From urinary metabolite data, estimated mean exposures are in the range of 1–2 μg/kg/day, with a 95th percentile of 7–17 μg/kg/day depending on the metabolite used (Table 4). This comparison suggests that both indirect and biomarker-based methods produced mean estimates below the RfD (20 μg/kg/day) and TDI (37 μg/kg/day), although the upper ranges of exposure approximated the RfD. As another example, the National Toxicology Program (NTP) CERHR determined that the NOAEL for reproductive effects in rats was 5–8 mg/kg/day and expressed concern over the potential for reproductive risk among infants younger than 1 year, if their exposures were significantly higher than those of the general population (1–30 μg/kg/day). Biomonitoring data are unavailable for healthy infants younger than 1 year, so this specific question cannot be addressed from the available data. However, for those 6 or more years of age, biomonitoring data indicate that ambient exposures to DEHP within the United States are lower than estimates used by the NTP-CERHR and that children’s and adults’ exposures are comparable. Some medical interventions may result in higher exposures to DEHP. These medical treatments entail risk–benefit calculations that make risk assessments substantially different from those relating to ambient exposures and are beyond the scope of this exercise. Note that for children and adults, exposure estimates calculated from the oxidized DEHP metabolites were approximately twice those calculated from MEHP (Tables 1, 2). However, for premature neonates the differences were in the range of an order of magnitude (Table 2), presumably from differences in metabolism and/or excretion in these preterm infants.

27We make the following recommendations for future research:

28Improve the understanding of human metabolism and pharmacokinetics. The most relevant urinary metabolites and appropriate metabolites for other matrices that provide the greatest analytical sensitivity must be measured. Differences in metabolic patterns among phthalates are important both toxicologically and in exposure assessment, especially when comparing relative exposures to different phthalates because the complex metabolism of high-molecular-weight phthalates leads to additional metabolic products (e.g., oxidative metabolites).

29Refine molar conversion factors to relate external phthalate exposure to urinary metabolite concentrations. Based on available data, the largest uncertainties appear in premature neonates.

30Determine the biologic media best suited for biomarker studies. If media other than urine are evaluated, methodologic issues must be considered.

31Improve the understanding of the mechanisms of action of phthalates in humans.

32Determine whether more highly exposed groups can be identified and, if so, identify the sources of exposure. Potentially vulnerable segments of the population (e.g., children, women of reproductive age, minorities) should be evaluated.

33Determine whether use of urinary metabolite data as an adjunct to epidemiology studies is possible. One specific issue relates to categorizing exposure from a limited number of urine samples.