Introduction

Polybrominated diphenyl ethers (PBDEs), persistent and bioaccumulative flame retardants, are of concern because they are ubiquitous in the United States, are potentially toxic, and have been found at rapidly rising levels in humans during the past few decades (Birnbaum and Staskal 2004; Hites 2004; Schecter et al. 2005b; Sjödin et al. 2004; Webster et al. 2005). The high level of PBDE contamination in the U.S. population and food is cause for concern because these compounds are chemically similar to polychlorinated biphenyls (PCBs) and have been shown in laboratory animal studies to be toxic in a number of ways. These include cancer in high-dose studies [National Toxicology Program (NTP) 1986], reproductive and developmental toxicity (Stoker et al. 2004), endocrine disruption (Hallgren and Darnerud 2002), and central nervous system effects (Eriksson et al. 2002; Viberg et al. 2003). PBDEs can be found in some textiles, electronics, (e.g., computers, televisions), plastics, and furniture such as sofas, chairs, and mattresses. Unlike dioxins and PCBs, these chemicals are primarily indoor pollutants and are found at high levels in household vacuum dust and other home and workplace environmental samples (Schecter et al. 2005a; Stapleton et al. 2005).

Very high levels of PBDEs have recently been found in the United States in mothers' milk (Schecter et al. 2003, 2005b), blood (Mazdai et al. 2003; Morland et al. 2005; Schecter et al. 2004, 2005b; Sjödin et al. 2004), food (Schecter et al. 2004), and adipose tissue (Johnson-Restrepo et al. 2005; She et al. 2002). U.S. blood and milk concentrations were 10- to 20-fold higher than the levels found in Europe (Bocio et al. 2003; Meironyte et al. 1999; Norén and Meironyté 2000; Ohta et al. 2002). Although levels of dioxins, dibenzofurans, and PCBs in human tissues are declining, PBDEs have been increasing substantially in blood levels in the United States during the past two to three decades (Schecter et al. 2005b; Sjödin et al. 2004).

The penta-BDE and octa-BDE commercial PBDE mixtures are no longer being produced or sold in the United States, whereas deca-BDE continues to be manufactured and sold in the United States as well as worldwide. Furthermore, because these compounds are persistent in the environment, reservoir sources are likely to be present for substantial periods of time. These reservoir sources may continue to contaminate food and dust, both of which are believed to contribute substantially to human intake of these compounds (Birnbaum and Staskal 2004; Jones-Otazo et al. 2005; Webster et al. 2005).

Even though studies have begun to estimate PBDE intake from ingestion and inhalation, the amount and percent of intake from food in the U.S. general population have not been well characterized nor have the amounts of intake from dust ingestion and inhalation been well defined (Jones-Otazo et al. 2005; Stapleton et al. 2005; Webster et al. 2005). The present study expands and complements our previous U.S. market basket survey (Schecter et al. 2004) and also estimates dietary PBDE exposure by age and sex from birth through ≥ 60 years of age. The food sample size is approximately twice the size of the earlier study and includes previously unpublished congener data from that study (Schecter et al. 2004). For the first time, we characterize the U.S. population's PBDE intake from food.

Methods

Food samples were purchased during 2003 and 2004 in Dallas, Texas, from three large supermarkets representing national chains. We chose commonly eaten food types and purchased national or store brands whenever possible. Items were frozen and shipped on dry ice to Eurofins-ERGO Laboratory for analysis. The laboratory measured 13 PBDE congeners (BDE-17, BDE-28, BDE-47, BDE-66, BDE-77, BDE-85, BDE-99, BDE-100, BDE-138, BDE-153, BDE-154, BDE-183, and BDE-209) in 62 food samples by gas chromatography-isotope dilution high resolution mass spectrometry. For quality control purposes, one laboratory blank and a quality control pool for each block of samples were run. Quantification was performed only if the sample level was at least twice the blank level. For the analysis of samples of major food types such as fish, meat, cooked egg, cheese, ice cream, and sausage, a total of 5–200 g of the sample was homogenized and mixed with sodium sulfate. Before column extraction, a mixture of internal ¹³C-labeled standards was added to each sample. A mixture of cyclohexane and dichloromethane was applied during column extraction of lipids. After solvent evaporation, gravimetric lipid determination was performed. Fish oil and human milk were used as quality control pools. Details of the analytical procedure have been described elsewhere (Päpke et al. 2004; Schecter et al. 2004). In the present study, we used half the limit of detection (LOD) to estimate levels of congeners below the LOD, whereas we previously calculated them as equal to zero (Schecter et al. 2004).

We used the mean PBDE concentrations in our food samples, in combination with other food consumption estimates, to estimate PBDE intake for several age and sex groups in the U.S. population. We assumed nursing to be the only source of food for nursing infants, and calculation of their intake was based on the assumption that the daily consumption of human milk is 800 g (Dewey et al. 1991; Kent et al. 1999; Neville et al. 1988); milk PBDE levels were primarily levels previously reported (Schecter AJ, unpublished data; Schecter et al. 2003, 2005b). To calculate PBDE intake relative to body weight, we estimated an average weight of 7 kg for nursing infants [Centers for Disease Control and Prevention (CDC) 2000]. For other population groups, we obtained the median amount consumed each day for each of several categories of food from the U.S. Department of Agriculture (USDA) dietary intake survey (USDA 1999a, 1999b) and from Smiciklas-Wright et al. (2003). These intake estimates were multiplied by the mean PBDE concentration of our sample foods in each category to obtain a PBDE intake estimate for a "typical" man or woman (one whose food intake is at the 50th percentile) for each age group.

Results

Table 1.

Table 2.

Table 3.

Table 4.

Table 5.

Figure 1. Daily PBDE dietary intake of U.S. population by age and food group (pg/kg body weight) as shown in Table 5. Abbreviations: F, female; M, male.

Table 6.

Tables 1–4 show the PBDE levels in U.S. food products reported on a whole weight basis. Of the 18 meat samples analyzed (Table 1), total PBDE levels varied from 39 ppt wet weight in a bacon sample to 1,426 ppt in one pork sausage sample. We observed considerable variation in levels, even between samples of the same type of meat, such as bacon, ground pork, and pork sausage. BDE-209 (deca-BDE), the major remaining congener in commercial production, was detected in 8 of the 18 meat samples and, when detected, ranged from 9.7 ppt in one ground beef sample to 245 ppt in ground turkey. Compared with PBDE levels in meat products in Spain (10–172 ppt) and Japan (6.25–63.6 ppt), PBDE levels were higher in the United States (Bocio et al. 2003; Ohta et al. 2002). Of the 24 fish samples analyzed (Table 2), total PBDE concentration ranged from 11 to 3,726 ppt. These values, although somewhat higher, are comparable with PBDE concentrations in fish of 88–1,019 ppt in Spain and 17.7–1,720 ppt in Japan (Bocio et al. 2003; Ohta et al. 2002). As with meat, we found considerable variation between samples of the same type of fish. Farm grown salmon and other fish tend to have higher PBDE concentrations than wild fish. Unfortunately, it is not always clear from the store label whether salmon or other fish were farm grown, because mislabeling appears to be common (Burros 2005). We found BDE-209 in 10 of the 24 fish samples, ranging from 4.9 ppt in a canned tuna sample to 1,269 ppt in a catfish sample. Of 15 dairy products analyzed (Table 3), the total PBDE concentration varied from 7.9 ppt in both whole and nonfat cows' milk, to 683 ppt in cream cheese. We measured BDE-209 in 7 of the 15 dairy samples, with concentrations ranging from 9.1 ppt in lowfat yogurt to 481 ppt in cream cheese. Again, this was considerably higher than concentrations in these foods in Spain (10–48 ppt) (Bocio et al. 2003). Of the five types of miscellaneous food products analyzed (Table 4), eggs were lowest at 85 ppt and calf liver highest at 2,835 ppt total PBDEs. BDE-209 was found in all of the miscellaneous samples except margarine and varied from 10.3 ppt in eggs to 288 ppt in calf liver.

When PBDE concentrations are expressed on a lipid basis, fish still contain the highest levels, followed by meat and dairy products. Table 5 shows the PBDE levels for the various food types analyzed, on a lipid basis as well as on a wet weight basis. Although these lipid-normalized values reflect animal or fish body burdens of PBDEs, they are not useful in calculating dietary intake.

Table 6 and Figure 1 present estimates of dietary intake of PBDEs subdivided by food types for the U.S. population. In all groups > 1 year of age, total PBDE intake from meat is significantly higher than from any other food. As shown in Table 6, the highest dietary intake values of PBDEs are in nursing infants (307 ng/kg body weight per day, which compares to 1.0 or 0.9 ng/kg/day at ≥ 60 years of age for men and women, respectively; these are much higher than Swedish values of 0.63 and 0.58 ng/kg/day, respectively (Lind et al. 2002).

Discussion

This larger U.S. market basket survey confirms that PBDE contamination levels in U.S. food are currently higher than previously reported in other countries (Bocio et al. 2003; Huwe and Larsen 2005; Ohta et al. 2002). Fish are highest in PBDE contamination on a whole weight basis, followed by dairy products and meat. Meat is the major source of PBDEs in the U.S. diet after nursing ends, followed by dairy products and fish, unlike some other countries where fish intake predominates (Bocio et al. 2003; Darnerud et al. 2001; Ohta et al. 2002). Men, with larger daily intakes of food, have a larger dietary intake of PBDEs than do women.

A large variation of PBDE levels exist, even for the same type of food (Huwe and Larsen 2005). Although the present study is the largest PBDE food survey in the United States to date to the best of our knowledge, we cannot claim that these new data are a representative sampling of the U.S. diet. Like other published surveys from other countries, the sample size needs to be increased and the samples need to be representative of the diet(s) of the country. Until this is done, uncertainty in estimates of food levels will exist, and as a result, intake estimates will be somewhat imprecise.

The comparatively higher PBDE levels in food cannot however be the only explanation for the 10- to 20-fold higher levels in blood and milk from the U.S. general population compared with European and Canadian levels (Bocio et al. 2003; Mazdai et al. 2003; Meironyte et al. 1999; Morland et al. 2005; Norén and Meironyté 2000; Ohta et al. 2002). Our total daily PBDE intake from dietary sources for adults is only 0.9–1.2 ng/kg body weight, which compares to Spain's 1.2–1.4 ng/kg/day (Bocio et al 2003) and the United Kingdom's approximately 1.5 ng/kg/day, assuming an average adult weight of 70 kg (Harrad et al. 2004), but is higher than Sweden's 0.58–0.63 ng/kg/day (Lind et al. 2002). Although there is a great deal of uncertainty on half-lives of PBDEs, assuming a maximum half-life of 2 years (Geyer et al. 2004; Thuresson et al. 2006) and an American body composition of approximately 25% adipose tissue, PBDE intake from food would lead to a steady-state body burden of < 10 ppb lipid. Given that the median lipid-adjusted levels in the United States from recent blood, milk, and adipose specimens exceed 30 ppb lipid, and that those of the top 5% of the population are 10–100 times greater, it appears unlikely that diet is the sole or even major source of exposure to PBDEs. This is in direct contrast to the situation with PCBs and dioxins in which > 95% of the exposure of the general population comes from food (U.S. Environmental Protection Agency 2004). This suggests that other routes of intake might be more significant for PBDEs than is the case for dioxins and PCBs.

The trends in dietary intake of PBDEs show a decreased intake per kilogram of body weight with age, with the highest dietary intake during nursing in the first year of life, 307 ng/kg body weight. This is due to the high level of PBDEs in human milk (median, 1,056 pg/g wet weight), assuming that human milk was the only food consumed. Children 2–5 years of age have higher PBDE dietary intake per kilogram of body weight than do older persons because of higher food intake per kilogram of body weight.

PBDE congeners 47, 99, 100, 153, and 154, and in some cases 209, are major contributors in both food concentration and dietary intake estimates. This reflects the previously reported findings on the congener distribution in human blood (Schecter et al. 2005b).

As is true for dioxins and PCBs, human breast milk is a major source of daily exposure to PBDEs for infants. Based on lactational exposure to dioxins, the body burden of the infant does not exceed 3–5 times that of the mother, in spite of the 50–100 times greater daily intake (Abraham et al. 1996; Lorber and Phillips 2002). A similar situation may exist for PBDE exposure of nursing infants, in which case human milk can be a significant route of exposure for babies. We join Wu et al. (2005) in suggesting that, as well as food, routes of exposure such as house dust ingestion and inhalation are likely important pathways of PBDE intake for children as well as adults.

Extrapolating from rodent studies, MacDonald (2005) hypothesized that health risks are possible for more highly exposed persons in the U.S. general population. Although the health effects of the levels we report are not clear, it seems reasonable from a public health standpoint to reduce the levels of these chemicals in the environment.