Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Nov 1.
Published in final edited form as: Environ Int. 2021 May 24;156:106626. doi: 10.1016/j.envint.2021.106626

BMI Modifies the Association between Dietary Intake and Serum Levels of PCBs

Tuo Lan 1, Buyun Liu 2, Wei Bao 2, Peter S Thorne 1,3
PMCID: PMC8910784  NIHMSID: NIHMS1709747  PMID: 34034117

Abstract

Polychlorinated biphenyls (PCBs) are a group of persistent organic pollutants that are carcinogenic, neurotoxic and endocrine disrupting in humans. Although diet is the primary source of exposure, there is no consensus on the association between dietary intake and serum PCBs. Additionally, body mass index (BMI) –with its inverse association with serum PCBs –may play a role in the association, which has never been studied. Therefore, we aimed to examine the association between dietary intake and serum levels of PCBs, and whether the association was modified by BMI. We used data from National Health and Nutrition Examination Survey (NHANES) 2003–2004, including 1531 participants. We estimated dietary intake of PCBs using the 24-hour diet recall, USDA Food Composition Intake Database, and PCBs content in foods from the Canada Total Diet Study. Serum PCBs were measured by high-resolution gas chromatography-mass spectrometry (HRGC/ID-HRMS). We used linear regression to examine the associations of dietary PCB intake with serum levels of seven PCB congeners and six PCB metrics. Further, we explored the role of BMI in the associations. We found that participants who were older and non-Hispanic tended to have a higher serum level of ∑37-PCB. In addition, we observed positive associations between dietary intake and serum PCBs for: PCB 105, 118, 126, 138+158, and 153 (P value ranges 0.005–0.03); seven PCB indicators (P value=0.03) and the sum of 37 PCBs (P value=0.04). Furthermore, we observed an effect modification by BMI ((P for interaction= 0.01 for ∑37-PCBs), with stronger associations in underweight or normal-weight individuals, and no association in overweight and obese individuals. In conclusion, within a cross-sectional, nationally representative sample of the US population, dietary PCB intake was positively associated with serum PCBs and the association was modified by BMI. Additional studies are warranted to replicate and confirm this effect modification.

Keywords: body mass index, diet, persistent organic pollutants, polychlorinated biphenyls, serum

1. Introduction

Polychlorinated biphenyls (PCBs) are a group of man-made chemicals that have been widely used as dielectric and heat exchange fluids in power plant and industries since the 1930s given their low electrical conductivity and high physicochemical stability (ATSDR 2010). They were also widely used in building materials such as window caulk, concrete floor treatments, and fluorescent light ballasts. However, PCBs were banned from intentional manufacture in 1978 in the United States because they were found to be environmentally persistent, carcinogenic and to have adverse health effects on the nervous system, immune system and reproductive system (ATSDR 2010; IARC 2013). Despite the ban for intentional use, PCBs are still being produced and distributed as by-products in commercial paint pigments (Anezaki and Nakano 2014; Grossman 2013; Hu and Hornbuckle 2010).

People continue to be exposed to higher chlorinated PCBs primarily through ingestion of fatty food as these PCBs are lipophilic and tend to bioaccumulate along the food chain (ATSDR 2010). The more volatile lower-chlorinated congeners are often inhaled and readily metabolized to hydrophilic hydroxy- and phase II- metabolites and excreted (Grimm et al. 2015; Hu et al. 2014). Most epidemiology studies have focused on PCB exposure from fish and shellfish (WHO 2016). Other fatty food items –for instance, dairy products and meat consumed more commonly in U.S. populations –can contribute to the dietary exposure to PCBs as much as seafood (Domingo and Bocio 2007; McKelvey et al. 2010; Schecter et al. 2001). Very few studies have evaluated the effect of other fatty food intake on the body burden of PCBs and findings were inconsistent (Berg et al. 2018; DeVoto et al. 1998; Jin et al. 2017; Laden et al. 1999; Schecter et al. 2010). One possible explanation for the discrepancy is the application of different individual-food items questionnaires. An alternative dietary assessment, as suggested by the United States Department of Agriculture (USDA) and the Environmental Protection Agency (EPA) is to use commodities data (EPA 2019; Miller 2013). Studies using commodities data have found that elevated body burdens were associated with dietary exposure to metals and dioxin-like pollutants (Awata et al. 2017; Bergkvist et al. 2008). More importantly, the levels of PCBs detected in the serum may be affected by body fat. Existing research suggests that body mass index (BMI) influences the level of PCB concentrations in serum –in particular, higher BMI is associated with lower PCBs (Koh et al. 2016; Mullerova et al. 2008). To the best of our knowledge, no study has investigated whether BMI can modify the relationship between dietary PCB exposure and serum levels of PCBs, which is important to provide a valid interpretation of the PCB exposure and biomonitoring data.

In this study, we used nationally representative data from NHANES to estimate dietary exposure to PCBs and evaluate the association between dietary intake and serum level of PCBs. Further, we examined the potential effect modification by BMI.

2. Methods

2.1. Study Design and Population

NHANES, which began in 1971, is an ongoing cross-sectional study with a complex, multistage, probability sampling strategy to assess the health and nutritional status of the U.S. non-institutionalized civilian population. NHANES conducts a household interview and a physical examination in a mobile examination center (MEC), and collects information on sociodemographic characteristics, lifestyle, diet, and medical conditions. In addition, blood and urine samples are collected for laboratory tests and analyzed in the same manner for all sampling locations within two-year cycles (Curtin et al. 2012). Since 1999, NHANES has operated continuously and data have been released to the public in two-year cycles. The study was approved by the National Center for Health Statistics (NCHS) Research Ethics Review Board and written informed consent was provided by participants.

We used NHANES 2003–2004 data because it was the most recent NHANES cycle that provided measurements of PCBs in serum at the individual level. The data can be accessed directly from the website (https://wwwn.cdc.gov/nchs/nhanes/ContinuousNhanes/Default.aspx?BeginYear=2003). We did not use PCBs data in subsequent cycles because they were measured in pooled samples after 2004. For the present analysis, we included all participants who completed two 24-hour diet recalls and had available data on serum PCBs (n= 1,711). Pregnant women (n=66) and individuals who reported implausible energy intake (<500 or >4,500 kcal/day) were excluded (n=114). As a result, 1,531 participants were included in this analysis.

2.2. Assessment of Dietary Exposure to PCBs

Two nonconsecutive 24-hour diet recall interviews were used to estimate participants’ food intake. Participants were asked to recall detailed information about types and portion sizes of all foods consumed in the past 24 hours. The first recall was collected in the MEC, and the second recall was collected through telephone 3–10 days later (Ahluwalia et al. 2016; Curtin et al. 2012). We used two databases to assess the dietary exposure to PCBs. First, we employed the Food Commodity Intake Database (FCID) recipes, developed by USDA’s Agricultural Research Service (ARS) and Economic Research Service (ERS), to convert food intake from 24-hour diet questionnaire into retail-level commodities (Ahuja et al. 2013). Next, we estimated dietary intake of substances of food reported as eaten (e.g. lasagna) at the commodities level (e.g., beef, wheat flour, tomato sauce, soybean oil, etc.). Then the Vancouver 2002-PCBs concentration database from the Canada Total Diet Study was linked with FCID 2003–2008 database. The Canada Total Diet Study is the most comprehensive study of PCBs in food in North America. Representative food samples were collected from eight cities from 1992 to 2002 (Dabeka and Cao 2013). Vancouver 2002 data were chosen because they were the most recent data and included concentrations of 40 abundant PCBs congeners for 49 food commodities (Ampleman et al. 2015).

The amount of each food consumed was multiplied by the appropriate PCB congener concentration to assess individual exposure. For one day exposure, we summed the PCB intake from each food reported in the past 24 hours. We calculated the exposure for both day 1 and day 2, the estimated dietary intake of PCBs from day 1 was only modestly correlated with intake from day 2 (Spearman rank correlation coefficient = 0.2). and then we used the mean of the two days to reduce the variability. Because PCBs are lipophilic and bioaccumulating, the concentration of PCBs was high in fatty food and fish, and low in fruit and vegetables as expected (Armitage and Gobas 2007; Llobet et al. 2003). In the present study, the contributions of PCBs from vegetables and fruits were excluded as they were negligible. Thus, we grouped PCB intake into six categories by their sources: dairy products, eggs, red meat, poultry, fish, and oil.

2.3. Serum Level of PCBs

A subsample of serum tested for PCBs was randomly selected among one-third of participants aged 12 years or older. PCBs in serum were measured by high-resolution gas chromatography-mass spectrometry (HRGC-HRMS). Around 2–10 ml of serum sample was spiked with 13C-labeled internal standards and were extracted using a C18 solid phase extraction (SPE) procedure with hexane (CDC 2003). Because a proportion of blood PCBs can bound to plasma protein besides lipid (Grimm et al. 2015), it is reasonable that the analysis includes both fresh weight and lipid adjusted forms, which NHANES provided. Lower limits of detection (LOD) varied by PCB congener and values below LOD were assigned the congener-specific value of LOD divided by the square root of two.

After excluding congeners detected in less than 50% of samples (PCB 81, 128 and 189), a total of 37 PCBs were included in the analysis. First, we analyzed six individual congeners and one co-eluting congener pair (PCB 28, 52, 105, 118, 126, 138+158, and 153) that were recommended as PCBs indicator according to the Stockholm Convention (UNEP 2013). These individual congeners were detected in more than 90% of samples in this study. Then, we used six PCB metrics in the analysis: 1) sum of 37 PCBs ( ∑37-PCBs); 2) seven indicator PCBs (∑7-PCBs); 3) seven dioxin-like PCBs (∑7DL-PCBs) (EFSA 2018); 4) 30 non-dioxin-like PCBs (∑30NDL-PCBs); 5) 34 Aroclor PCBs (∑34Aroclor-PCBs) to indicate the manufactured mixtures (ATSDR 2000); and 6) three non-Aroclor PCBs (∑3NAroclor-PCBs) defined as congener comprising less than 0.2% of Aroclor 1016, 1211, 1242, 1248, 1254, 1260 and 1262 (Table 1).

Table 1.

Combination PCB metrics and congeners list

PCB metrics PCB congeners
7 indicator PCBs (∑7-PCBs) PCB 28, 52, 101, 138+158, 153, 180
7 dioxin-like PCBs (∑7DL-PCBs) PCB 105, 118, 126, 156, 157, 167, 169
30 non-dioxin-like PCBs (∑30NDL-PCBs) PCB 28, 44, 49, 52, 66, 74, 87, 99,101,110, 138+158, 146, 149, 151, 153, 170, 172, 177, 178, 180, 183, 187, 194, 195, 196+203, 199, 206, 209
34 Aroclor PCBs (∑34Aroclor-PCBs) PCB 28, 44, 49, 52, 66, 74, 87, 99, 101, 105, 110, 118, 138+158, 146, 149, 151, 153, 156, 157, 167, 170, 172, 177, 178, 180, 183, 187, 194, 195, 196+203,199, 206
3 non-Aroclor PCBs (∑3NAroclor-PCBs) PCB 126,169, 209
Sum of 37 PCBs ( ∑37-PCBs) PCB 28, 44, 49, 52, 66, 74, 87, 99, 101, 105, 110, 118, 126, 138+158, 146, 149, 151, 153, 156, 157, 167, 169, 170, 172, 177, 178, 180, 183, 187, 194, 195, 196+203,199, 206, 209
Open in a new tab

2.4. Sociodemographic and Lifestyle Characteristics Assessment

Sociodemographic and lifestyle characteristics including age, sex, race/ethnicity, education, family income, smoking status, alcohol consumption, and physical activity were assessed by self-reported questionnaires. In the present analysis, race/ethnicity was defined as Hispanic, non-Hispanic White, non-Hispanic Black, and other. Educational attainment was categorized as less than high school, high school, and higher than high school. For individuals younger than 20 years, the highest education in their household was used as a surrogate measure of their educational attainment. Family income-to-poverty ratio (IPR) was categorized as ≤1.30, 1.31–3.50, and >3.50 following NHANES analytic guidelines (CDC 2013). Smoking status was defined as never (smoked less than 100 cigarettes in their lifetime), former (smoked more than 100 cigarettes in their lifetime but had quit smoking at the time of the survey) and current smoker (smoke at the time of the survey) (CDC 2015). Alcohol intake was categorized as non-drinker (0 g/day), moderate drinker (0.1−27.9 g/day for men and 0.1−13.9 g/day for women), and heavy drinker (≥28.0 g/day for men and ≥14.0 g/day for women) (HHS 2017). Physical activity was defined as <600, 600−1200, and >1200 metabolic equivalents of task (MET) min per week according to the Global Physical Activity Questionnaire Analysis (WHO 2014). Total energy intake (kcal/day) was calculated using the USDA automated multiple-pass method (http://www.ars.usda.gov/ba/bhnrc/fsrg). Healthy Eating Index (HEI) was calculated as an index of diet quality –where a higher score indicates a better diet –that aligns with the key recommendations of the Dietary Guidelines for Americans (HHS 2017).

Weight and height were measured during the physical examination and used for calculating BMI (weight in kilograms divided by height in meters squared). For children and teens, BMI categories used age and gender specific percentile cutoffs (Kuczmarski et al. 2002): underweight (<5th percentile ), normal (5th −84th percentile), overweight (85th −94th percentile), and obese (≥95th percentile); for adults, BMI categories were defined as underweight (<18.5 kg/m2), normal weight (18.5–24.9 kg/m2), overweight (25.0–29.9 kg/m2), and obese (≥30.0 kg/m2). As there were only 33 underweight participants, we merged them with normal weight participants. Thus, BMI was classified into three categories in the present analysis, including underweight or normal weight, overweight, and obese.

2.5. Statistical Analysis

We followed NHANES analytic guidelines which accounted for sample weights and survey design variables for all analyses (CDC/NCHS 2011). As the distribution of estimated dietary intake of PCBs and measured serum levels of PCBs were highly right-skewed, both of them were log-transformed for evaluating the statistical significance. We examined the distribution of potential confounders by quintile of dietary PCBs exposure and by serum concentration of the sum of 37 PCBs in both lipid adjusted and fresh weight forms. We used Pearson chi-square and ANOVA to compare categorical variables and serum concentration after log transformation, respectively. We conducted linear regressions to evaluate the association of dietary intake and serum level of PCBs, and modelled dietary PCB exposures as both quintiles and as a continuous variable. We adjusted for age, race/ethnicity, and sex in the basic model. Our second set of models included BMI and total energy intake as additional variables. We evaluated all potential serum PCB-related variables by bivariate regression analysis. Then we started a model with all the variables that were associated with serum PCBs (p-value<0.2) in the bivariate analysis. By backward selection, we removed variables until all remaining variables had P-values that were less than or equal to 0.05 in the model. Finally, we included a full model with all variables that could serve as potential confounders including age, sex, race/ethnicity, BMI, education level, IPR, smoking status, alcohol intake, total energy intake and HEI. As results from our second and fully adjusted models were similar, we present only those for basic and fully adjusted models.

To further explore the independent influence of dietary exposure to PCBs and body burden of PCBs, we conducted stratified analyses by BMI categories (underweight/normal, overweight and obese). We hypothesized that a higher BMI will produce higher sequestration of PCBs in adipose tissue, and that dietary exposure will have less influence on the serum level of PCBs compared to the normal/underweight. In addition, we performed sensitivity analyses by restricting to 1) individuals over 20 years of age as they may have a different dietary intake from adolescents (HHS 2017); and 2) individuals who had complete PCBs measurement data. Finally, as PCB concentrations varied greatly based on geographic context and time period, we performed a sensitivity analysis by using Ottawa 2000-PCBs concentration from the Canada Total Diet Study (Dabeka and Cao 2013). We selected this dataset not only because it was the second most recent dataset, but it also collected food from an inland city instead of a coastal city such as Vancouver, which was used in the main analysis.

We performed all statistical analyses using survey procedures with SAS software (version 9.4; SAS Institute Inc., Cary, NC, USA).

3. Results

Of the 1,531 participants who are eligible for the analysis, the mean age was 40 and 50% were women. A total of 314 food including 54 fatty items (eggs, red meat, poultry, fish, and oil, etc.) and 260 non-fatty items (vegetables, fruits, grains, etc.) were reported. The mean estimated daily intake of total PCBs was 97 nanogram (ng) equaling 1.4 ng/kg body weight (bw); the 95th percentile was 364 ng equaling 4.8 ng/kg bw. On average, the sources of exposure were meat (29%), dairy (22%), poultry (17%), fish (16%), egg (15%), and oil (1%). Compared to individuals with low dietary PCB exposure, those with a higher estimated intake of PCBs from the fatty food were more likely to be male, overweight and obese, had higher educational attainment, a higher total energy intake and a better dietary quality as assessed by the HEI-2010 (Table 2). There was a trend toward lower PCB exposures among subjects living in poverty. No effect of smoking, alcohol consumption, or physical activity was observed on dietary PCB exposure.

Table 2.

Estimated dietary PCBs exposure by population characteristics in NHANES 2003–2004

Estimated Dietary PCBs Exposure Quintiles
Q1 <31 ng Q2 31–44 ng Q3 43–59 ng Q4 60–94 ng Q5 >94 ng P-value
Number of Participants 306 306 307 306 306
Age, years 42.1 (1.4) 42.1 (1.5) 41.5 (1.2) 43.0 (0.9) 45.1 (1.2) 0.114
Gender <0.001
Male 29.5 (0.7) 37.4 (0.8) 46.6 (0.8) 57.7 (0.9) 58.6 (0.8)
Female 70.5 (1.4) 62.6 (1.2) 54.4 (0.8) 42.3 (1.0) 41.4 (0.8)
Race/ethnicity 0.282
Hispanic 9.2 (0.4) 13.9 (0.5) 15.4 (0.7) 13.6 (0.8) 9.8 (0.7)
Non-Hispanic white 74.7 (1.4) 68.3 (1.0) 71.9 (1.0) 72.7 (1.0) 70.3 (1.4)
Non-Hispanic black 10.5 (0.5) 12.9 (0.6) 9.6 (0.6) 8.6 (0.4) 12.3 (0.6)
Other 5.5 (0.4) 4.9 (0.3) 3.0 (0.3) 5.1 (0.4) 7.6 (0.4)
Education 0.035
Less than high school 21.1 (0.5) 20.3 (0.6) 15.7 (0.5) 16.7 (0.4) 16.0 (0.6)
High school 18.1 (0.5) 30.8 (0.7) 26.6 (0.7) 23.1 (0.6) 22.0 (0.9)
More than high school 60.8 (0.9) 48.9 (1.0) 57.7 (0.9) 60.2 (1.0) 62.0 (1.1)
Family income to poverty ratio 0.055
<1.3 25.0 (0.7) 18.8 (0.4) 23.2 (0.8) 16.4 (0.4) 18.8 (0.6)
1.3–3.5 33.7 (0.6) 41.8 (0.8) 38.5 (0.6) 37.6 (0.8) 32.6 (0.8)
>3.5 37.9 (0.8) 35.8 (0.8) 33.5 (0.9) 39.7 (0.9) 45.9 (1.3)
Missing 3.4 (0.2) 3.6 (0.3) 4.8 (0.2) 6.3 (0.5) 2.7 (0.3)
Smoking 0.875
Never 58.4 (0.9) 59.4 (0.6) 56.2 (0.9) 55.9 (0.9) 58.7 (1.0)
Former 20.7 (0.7) 17.9 (0.6) 21.5 (0.5) 24.9 (0.7) 22.8 (0.6)
Current 20.9 (0.5) 22.6 (0.7) 22.3 (0.7) 19.2 (0.6) 18.4 (0.7)
Alcohol 0.262
Non 82.3 (1.1) 77.6 (1.0) 79.4 (1.1) 72.3 (1.0) 74.4 (1.0)
Moderate 5.2 (0.3) 4.5 (0.3) 8.1 (0.3) 9.3 (0.4) 7.7 (0.5)
Heavy 12.5 (0.7) 17.9 (0.7) 12.5 (0.5) 18.4 (0.5) 17.9 (0.7)
BMI 0.008
Normal/underweight 52.9 (1.0) 44.4 (1.3) 35.3 (1.0) 31.4 (1.0) 33.2 (1.0)
Overweight 25.7 (0.8) 29.1 (0.6) 28.5 (0.7) 34.1 (0.6) 33.3 (1.2)
Obese 20.6 (0.6) 22.9 (0.8) 35.1 (0.5) 33.8 (0.9) 31.8 (0.6)
Missing 1.6 (0.2) 3.6 (0.4) 1.0 (0.2) 0.7 (0.1) 1.7 (0.2)
Physical Activity, MET-min/week 0.415
<600 36.4 (0.9) 36.5 (0.9) 37.4 (0.7) 36.3 (0.6) 36.8 (0.9)
600–1199 13.9 (0.3) 17.8 (0.6) 13.0 (0.5) 14.1 (0.7) 14.8 (0.7)
≥1200 39.6 (0.8) 36.7 (0.6) 42.2 (0.9) 42.9 (0.7) 44.7 (0.8)
Missing 10.0 (0.2) 9.0 (0.3) 7.4 (0.2) 6.6 (0.4) 3.7 (0.2)
Total energy intake (kcal/day) 1678.6 (55.9) 1886.9 (51.9) 2089.5 (46.3) 2356.0 (53.8) 2320.4 (48.1) <0.001
HEI-2010 44.8 (1.5) 45.4 (1.1) 45.4 (1.1) 46.6 (0.8) 52.6 (1.0) <0.001
Open in a new tab

Data are presented as the weighted mean and standard error for continuous variables; and weighted percentages and standard error for categorical variables.

BMI, body mass index; HEI-2010, 2010 healthy eating index; MET; metabolic equivalent of task; IPR, family income-to-poverty ratio.

The lipid-adjusted ∑37-PCB concentration did not vary by gender, education, alcohol intake, BMI or physical activity, whereas the concentration was higher among participants who were older, non-Hispanic, former smokers, and having a higher family income in the bivariate analysis (Table 3). After adjustment for sociodemographic and lifestyle factors that influence serum PCB, non-Hispanic ethnicity was still associated with a higher level of lipid-adjusted ∑37-PCB (P=0.016). However, no statistically significant difference of ∑37-PCB levels were observed across smoking or family income groups after adjustment for age (P=0.10 and P=0.65). The whole weighted ∑37-PCB concentration had a similar distribution by population characteristics to the lipid-adjusted weight, except that the concentration was lower among participants who were female and who were normal/underweight.

Table 3.

Weighted serum concentrations of total 37 PCBs by population characteristics in NHANES 2003–2004

∑37 PCBs
ng/g lipid adjusted ng/g whole weight
N=1531 median (IQR) Geometric mean (sd) P-value median (IQR) Geometric mean (sd) P-value
Gender 0.291 0.054
Male 762 148 (82–269) 151 (5) 0.83 (0.42–1.62) 0.76 (0.04)
Female 769 149 (75–268) 143 (4) 0.81 (0.36–1.63) 0.65 (0.04)
Race/ethnicity <0.001 <0.001
Hispanic 403 86 (59–173) 102 (6) 0.47 (0.27–0.94) 0.48 (0.05)
non-Hispanic white 711 159 (88–283) 155 (5) 0.90 (0.45–1.71) 0.78 (0.04)
non-Hispanic black 353 149 (69–313) 157 (12) 0.77 (0.32–1.78) 0.65 (0.10)
Other 64 137 (88–267) 142 (19) 0.69 (0.42–1.57) 0.50 (0.17)
Age, years <0.001 <0.001
12–19 497 58 (44–86) 59 (2) 0.25 (0.18–0.38) 0.22 (0.02)
20–39 415 102 (66–145) 98 (3) 0.52 (0.34–0.79) 0.44 (0.03)
40–59 307 228 (165–310) 233 (10) 1.29 (0.95–1.96) 1.29 (0.08)
>60 312 379 (289–552) 394 (20) 2.23 (1.76–3.21) 2.06 (0.18)
Education 0.614 0.448
less than high school 432 140 (75–328) 156 (10) 0.87 (0.39–1.99) 0.74 (0.07)
High school 358 156 (72–291) 148 (7) 0.89 (0.37–1.83) 0.75 (0.05)
more than high school 741 145 (81–247) 144 (5) 0.78 (0.39–1.43) 0.67 (0.04)
Family income to poverty ratio 0.004 0.013
<1.3 474 113 (64–215) 121 (7) 0.58 (0.30−−1.36) 0.54 (0.06)
1.3–3.5 569 155 (81–291) 154 (6) 0.86 (0.38–1.77) 0.75 (0.04)
>3.5 418 160 (91–264) 154 (7) 0.89 (0.46–1.56) 0.74 (0.07)
Missing 70 169 (72–338) 150 (16) 0.91 (0.34–2.11) 0.74 (0.10)
Smoking <0.001 <0.001
Never 1009 130 (67–246) 129 (3) 0.64 (0.31−−1.36) 0.58 (0.04)
Former 286 232 (137–345) 208 (11) 1.31 (0.79–2.06) 1.03 (0.08)
Current 236 145 (84–235) 144 (7) 0.83 (0.45–1.52) 0.79 (0.06)
Alcohol 0.272 0.071
Non 1255 145 (75–264) 143 (3) 0.80 (0.36–1.62) 0.68 (0.03)
Moderate 104 136 (78–268) 153 (11) 0.82 (0.50–1.80) 0.86 (0.07)
Heavy 172 162 (91–283) 165 (14) 0.89 (0.50–1.56) 0.74 (0.11)
BMI 0.170 0.024
Normal/underweight 627 133 (76–229) 135 (4) 0.60 (0.35−−1.23) 0.60 (0.04)
Overweight 446 158 (82–283) 158 (6) 0.96 (0.45–1.74) 0.83 (0.04)
Obese 436 165 (72–289) 150 (6) 0.96 (0.39–1.73) 0.73 (0.05)
Missing 22 175 (102–401) 192 (58) 1.24 (0.43–2.40) 0.63 (0.50)
Physical Activity, MET-min/week 0.165 0.094
<600 546 166 (89–306) 164 (7) 0.95 (0.50−−1.84) 0.82 (0.05)
600–1199 175 161 (88–307) 162 (7) 0.93 (0.43–1.79) 0.79 (0.08)
≥1200 557 144 (82–258) 149 (7) 0.81 (0.41–1.54) 0.70 (0.07)
Missing 253 60 (43–93) 63 (3) 0.26 (0.17–0.41) 0.24 (0.03)
Open in a new tab

BMI, body mass index; HEI-2010, 2010 healthy eating index; MET; metabolic equivalent of task; IPR, family income-to-poverty ratio.

For individual congeners, we observed positive dose-response associations between estimated dietary intake of PCBs and the serum level of PCBs 105, 118, 126, 138+158 and 153 in basic and full model (Table 4: β=0.06–0.09, P=0.001–0.033 in basic model and P=0.005–0.034 in full model), but there was no association with PCB 28 or PCB 52. For combination serum PCB congener metrics, positive associations with dietary intake were observed for ∑37-PCBs and ∑7-PCBs in the fully-adjusted model β=0.05 and P=0.042, β=0.04 and P=0.031, respectively). In terms of the magnitude of correlation between diet and serum, the partial correlation coefficient (fully adjustment) was 0.09 for PCB 118, follow by 0.082, 0.08, 0.076, 0.061 and 0.055 for PCB 138+158, PCB 153, PCB 126, ∑7-PCBs and ∑37-PCBs, respectively (P=<0.001–0.03). The correlations between diet and serum level of PCB 28, PCB 52, ∑30NDL-PCB, ∑3NAroclor-PCB, and ∑34Aroclor-PCB were non-significant, although still in the positive direction (partial correlation coefficient=0.005–0.052 and P=0.05–0.85). With respect to fresh weight serum, the associations for dietary exposure were weaker compared to the lipid adjusted serum, and the associations with fresh weight PCB 126 and ∑37-PCBs attenuated to null (supplemental table1).

Table 4.

Association of dietary intake and lipid adjusted serum levels of individual PCB congeners (top) and congener groupings (bottom) among U.S. population: multiple linear regression results

Regression coefficient by quintile and one unit of Log transformed dietary intake of PCBs
Q1 (N=294–306) Q3 (N=295–307) Q5 (N=294–306) Per one unit increase in log(PCBs intake ng/day) P-value Per unit increase
Log PCB28
 Basic Model 0 (ref) 0.04 (0.04) 0.05 (0.06) 0.01 (0.02) 0.713
 Full Model 0 (ref) 0.04 (0.03) 0.06 (0.04) 0.01 (0.02) 0.662
Log PCB52
 Basic Model 0 (ref) 0.04 (0.05) 0.04 (0.07) −0.01 (0.03) 0.808
 Full Model 0 (ref) 0.07 (0.06) 0.08 (0.08) −0.01 (0.03) 0.815
Log PCB105
 Basic Model 0 (ref) 0.23 (0.09) 0.23 (0.07) 0.08 (0.03) 0.001
 Full Model 0 (ref) 0.24 (0.08) 0.25 (0.06) 0.09 (0.03) 0.005
Log PCB118
 Basic Model 0 (ref) 0.25 (0.08) 0.24 (0.06) 0.09 (0.03) 0.010
 Full Model 0 (ref) 0.27 (0.07) 0.26 (0.06) 0.08 (0.03) 0.009
Log PCB126
 Basic Model 0 (ref) 0.14 (0.07) 0.19 (0.06) 0.06 (0.03) 0.033
 Full Model 0 (ref) 0.16 (0.06) 0.19 (0.05) 0.06 (0.03) 0.034
Log PCB138+158
 Basic Model 0 (ref) 0.17 (0.07) 0.20 (0.06) 0.07 (0.02) 0.013
 Full Model 0 (ref) 0.17 (0.07) 0.20 (0.06) 0.07 (0.02) 0.007
Log PCB153
 Basic Model 0 (ref) 0.12 (0.06) 0.18 (0.06) 0.07 (0.02) 0.013
 Full Model 0 (ref) 0.14 (0.07) 0.19 (0.06) 0.07 (0.02) 0.005
Log ∑37-PCBs
 Basic Model 0 (ref) 0.08 (0.06) 0.11 (0.06) 0.03 (0.02) 0.161
 Full Model 0 (ref) 0.12 (0.06) 0.17 (0.06) 0.05 (0.02) 0.042
Log ∑7-PCBs
 Basic Model 0 (ref) 0.10 (0.05) 0.11 (0.05) 0.03 (0.02) 0.103
 Full Model 0 (ref) 0.11 (0.05) 0.12 (0.05) 0.04 (0.05) 0.031
Log ∑7DL-PCBs
 Basic Model 0 (ref) 0.16 (0.12) 0.17 (0.10) 0.03 (0.04) 0.464
 Full Model 0 (ref) 0.23 (0.12) 0.28 (0.11) 0.06 (0.04) 0.132
Log ∑30NDL-PCBs
 Basic Model 0 (ref) 0.07 (0.05) 0.07 (0.06) 0.02 (0.02) 0.379
 Full Model 0 (ref) 0.10 (0.05) 0.11 (0.05) 0.03 (0.02) 0.109
Log ∑3NAroclor-PCBs
 Basic Model 0 (ref) −0.16 (0.10) −0.18 (0.09) −0.06 (0.03) 0.106
 Full Model 0 (ref) −0.06 (0.10) −0.06 (0.10) −0.02 (0.03) 0.539
Log ∑34Aroclor-PCBs
 Basic Model 0 (ref) 0.08 (0.05) 0.08 (0.06) 0.02 (0.02) 0.312
 Full Model 0 (ref) 0.11 (0.05) 0.12 (0.05) 0.03 (0.02) 0.082
Open in a new tab
a.

Regression coefficient present 1 unit increase in log-transformed serum level with 1 unit increase with log-transformed estimated fatty dietary intake of PCBs

b.

The number of participants varies because the frequency of detection is different across PCB congeners.

c.

Basic model: adjusted for age, sex, and race/ethnicity.

d.

Full model: model adjusted for age, sex, race/ethnicity, BMI, education level, IPR, smoking status, alcohol intake, physical activity level, total energy intake and HEI

In stratified analysis by BMI, we observed similar and stronger positive associations among normal/underweight participants, whereas there were different patterns among overweight and obese participants. Compared to all participants, we found stronger positive dose-response associations between estimated dietary intake of PCBs and serum congeners 105, 118, 126, 138&158, 153, ∑37-PCBs and ∑7-PCBs in normal/underweight (Table 5: β=0.07–0.15, P=<0.001–0.029 in basic model and P=<0.001–0.023 in full model). The association between dietary intake and serum level of ∑7DL-PCBs was not found among all participants (Table 4), but it became significantly positive among normal/underweight participants (β=0.17, P=0.003 in basic model and β=0.2, P=0.005 in full model). In addition, a suggestive positive association was found with ∑34Aroclor-PCB ((β=0.06, P=0.049 in basic model and P=0.078 in full model) for normal/underweight participants. We observed no association of dietary intakes with serum PCB 28, PCB52, ∑30NDL-PCB and ∑3NAroclor-PCB in the stratified analysis. No association was found in overweight or obese participants with serum level of PCBs, except for the association with PCB 153 (β=0.09, P=0.044 in basic model and β=0.10, P=0.051 in full model) among the overweight.

Table 5.

Association of dietary intake and lipid adjusted serum levels of PCBs: multiple linear regression results stratified by BMI

Regression coefficient by log transformed dietary intake of PCBs (ng/day)
Normal /Underweight Overweight Obese
N=599–627 N=429–446 N=415–436
Per unit increase P-value Per unit increase P-value Per unit increase P-value P for interaction
Log PCB28
 Basic Model 0.05 (0.03) 0.140 −0.02 (0.03) 0.504 −0.02 (0.05) 0.755 0.304
 Full Model 0.03 (0.02) 0.228 −0.02 (0.03) 0.534 0.00 (0.06) 0.997 0.244
Log PCB52
 Basic Model 0.04 (0.04) 0.309 −0.04 (0.05) 0.373 −0.04 (0.07) 0.587 0.003
 Full Model 0.03 (0.04) 0.483 −0.03 (0.04) 0.439 −0.03 (0.07) 0.722 0.007
Log PCB105
 Basic Model 0.14 (0.03) <0.001 0.02 (0.06) 0.757 0.03 (0.04) 0.516 0.155
 Full Model 0.13 (0.03) <0.001 0.06 (0.06) 0.404 0.04 (0.04) 0.331 0.160
Log PCB118
 Basic Model 0.15 (0.03) <0.001 0.00 (0.06) 0.958 0.04 (0.04) 0.302 0.024
 Full Model 0.14 (0.03) 0.001 0.03 (0.07) 0.657 0.05 (0.04) 0.215 0.019
Log PCB126
 Basic Model 0.13 (0.04) 0.008 −0.03 (0.06) 0.651 0.05 (0.04) 0.226 0.380
 Full Model 0.12 (0.05) 0.027 −0.02 (0.07) 0.791 0.08 (0.04) 0.086 0.422
Log PCB138+158
 Basic Model 0.11 (0.03) 0.002 0.07 (0.04) 0.085 0.05 (0.04) 0.226 0.078
 Full Model 0.10 (0.03) 0.001 0.09 (0.04) 0.056 0.03 (0.04) 0.412 0.034
Log PCB153
 Basic Model 0.10 (0.03) 0.002 0.09 (0.04) 0.044 0.05 (0.04) 0.244 0.099
 Full Model 0.10 (0.03) 0.002 0.10 (0.05) 0.051 0.02 (0.04) 0.556 0.053
Log ∑37-PCBs
 Basic Model 0.10 (0.03) 0.007 0.01 (0.04) 0.884 0.04 (0.04) 0.375 0.006
 Full Model 0.11 (0.04) 0.019 0.03 (0.04) 0.416 0.03 (0.04) 0.496 0.010
Log ∑7-PCBs
 Basic Model 0.07 (0.03) 0.029 0.03 (0.03) 0.421 0.01 (0.03) 0.681 0.081
 Full Model 0.07 (0.03) 0.023 0.06 (0.03) 0.120 0.00 (0.03) 0.938 0.062
Log ∑7DL-PCBs
 Basic Model 0.17 (0.05) 0.003 −0.10 (0.11) 0.375 0.07 (0.07) 0.359 0.033
 Full Model 0.20 (0.06) 0.005 −0.06 (0.09) 0.508 0.05 (0.09) 0.613 0.035
Log ∑30NDL-PCBs
 Basic Model 0.06 (0.03) 0.065 0.02 (0.04) 0.570 0.01 (0.03) 0.883 0.157
 Full Model 0.05 (0.03) 0.110 0.05 (0.04) 0.220 0.00 (0.03) 0.968 0.174
Log ∑3NAroclor-PCBs
 Basic Model 0.02 (0.05) 0.652 −0.05 (0.09) 0.596 −0.08 (0.06) 0.206 0.459
 Full Model 0.05 (0.06) 0.402 −0.01 (0.09) 0.906 −0.08 (0.07) 0.273 0.427
Log ∑34Aroclor-PCBs
 Basic Model 0.06 (0.03) 0.049 0.02 (0.04) 0.619 0.01 (0.03) 0.813 0.132
 Full Model 0.06 (0.03) 0.078 0.05 (0.04) 0.242 0.00 (0.03) 0.952 0.146
Open in a new tab
a.

Regression coefficient present 1 unit increase in log-transformed serum level with 1 unit increase with log-transformed estimated fatty dietary intake of PCBs

b.

The number of participants varies because the frequency of detection is different across PCB congeners.

c.

Basic model: adjusted for age, sex, and race/ethnicity.

d.

Full model: model adjusted for age, sex, race/ethnicity, education level, IPR, smoking status, alcohol intake, physical activity level, total energy intake and HE

When we performed the sensitivity analysis for table 4 and table 5 after restricting the analyses to individuals who: 1) were older than 20 years; and 2) had complete PCBs measurement data, we observed similar findings to the main analysis with comparable regression coefficients but fewer significant associations (data not shown). The attenuated associations are likely due to the smaller number of participants. Using the Ottawa 2000-PCBs concentration dataset instead of the Vancouver 2002 data, we observed the same positive direction but weaker strength of associations with individual PCB congeners and metrics (supplemental table2).

4. Discussion

In this large and nationally representative study, the mean estimated daily intake of total PCBs was 97 ng or 1.4 ng/kg bw. Participants who were older and non-Hispanic were more likely to have a higher serum level of ∑37-PCB. After adjustment for sociodemographic and lifestyle factors influencing serum PCB, the dietary exposure to PCBs was positively associated with the serum levels of PCB 105, PCB 118, PCB 126, PCB 138+158, PCB 153, ∑37-PCBs and ∑7-PCBs in total participants. The association between dietary PCBs intake and serum levels of PCB was modified by BMI, with stronger associations in normal/underweight individuals and no statistically significant association in obese individuals.

To our knowledge, this study presents the first estimated dietary exposure to PCBs for the U.S. general population. It is difficult to compare the daily estimated PCB intake with previous studies due to different study populations, different congeners, different food measured and study year. Our daily estimated PCB intake (97ng) was lower than that of the Airborne Exposure to Semi-volatile Organic Pollutants (AESOP) Study estimation for mother-child pairs of 178–295 ng that used the same food concentration data and included 40 congeners (Ampleman et al. 2015). The discrepancy might be explained by a different distribution of age –half of the participants in the AESOP Study were adolescent children who have a higher dietary exposure to PCBs than adults and elderly individuals (WHO 2016). When comparing U.S. exposure (1.4 ng/kg bw) with worldwide studies that used the World Health Organization six indicator congeners for the general population, the daily estimated PCB intake is higher than the results from China, Korea, and France (0.5–0.9 ng/kg bw, in the years 2007 to 2012). However, it is lower compared to some European countries (2.6–24 ng/kg bw in Belgium, Netherlands and Sweden, in the years 2003 to 2013). Different congeners measured (40 vs. 6), body weight (e.g., a higher prevalence of obesity in the U.S.), consumption pattern (e.g., higher vegetable consumption in China, higher dairy and fish consumption in Nordic counties), and decreasing PCB concentrations in food over time (WHO 2016) may also contribute to the differences.

With respect to sociodemographic and lifestyle factors, the finding that age was positively associated with total PCB concentration has been consistently observed in previous studies (Berg et al. 2018; Jensen 1989; Laden et al. 1999; Stellman et al. 1997). This is because older people grew up in an environment with higher PCB concentrations decades ago, and they had longer time to accumulate PCB. Our finding that Hispanics had lower serum PCB compared to non-Hispanics was also observed in the AESOP study (Koh et al. 2016). In addition, a study in New York showed that Hispanics also had lower PCB concentrations in adipose tissue than other races/ethnicities (Johnson-Restrepo et al. 2005). One possible reason is that some Hispanics were born and raised in places outside the U.S., where PCBs were less manufactured and used. As for smoking and family income, the observation that former smokers and higher family income tended to have higher serum PCBs was confounded by age. Those who had quit smoking or had a higher family income also tended to be older than current smokers and non-smokers or those who had a relatively lower income. To date, the literature has no consensus on the relation between smoking and PCB. Female smokers were found to have lower blood PCB in NHANES 2001–2002 (no difference for males) (Ferriby et al. 2007); and lower blood PCB during pregnancy and lower breast milk PCB postpartum among women of childbearing age (Chen et al. 2006; Uehara et al. 2007). In contrast, a Korean study found that smoking was positively associated with blood PCB among females (Moon et al. 2017); and a Czech study found the positive association was among all smokers but the study did not adjust for age (Cerná et al. 2008). Additionally, no association between smoking and PCB was observed in an Italian cohort among the general population or among residents in the most polluted area (Apostoli et al. 2005; Donato et al. 2006). The overall inconsistency suggested that the association varied across different age and gender groups, and smoking is a potential confounder that needs to be considered in evaluating serum PCB. Our bivariate analysis that higher family income was associated with a higher PCB concentration in blood was in line with previous studies that did not adjust for age (Borrell et al. 2004; Cao et al. 2011). Our null finding for income after adjustment for age differed from a prenatal study conducted near a PCB contaminated Superfund site. The prenatal study observed that high income was associated with lower cord blood PCB after adjustment for maternal age (Choi et al. 2006). However, household income in this prenatal study can be an indicator of the distance from the contaminated site. The inhalation exposure to semi-volatile PCBs and wind-blown dust carrying PCBs near the contaminated site may explain this discrepancy.

When evaluating the association between dietary PCB intake and serum concentration of PCBs, most studies focus on PCB indictors such as PCB 28, 52, 101, 138, 153 and 180 (Furberg et al. 2002; Gonzalez-Alzaga et al. 2018; WHO 2016). Our findings of the positive associations between estimated PCBs intake and serum concentration of PCBs confirmed the results in Furberg’s and Gonzalez-Alzag’s studies for commonly detected PCBs including PCB 105, 118, 138 (coeluted with 158), and 153. In addition, we found dietary exposure was primarily responsible for the elevated serum level of PCB 126, which had the highest toxic equivalence factor among all PCBs (Van den Berg et al. 2006). We observed a null association for PCB 28 likely due to its low degree of chlorination (bearing three chlorines on the biphenyl group) and higher volatility. Thus, people are more likely exposed to PCB 28 through inhalation than ingestion (Ampleman et al. 2015; Grimm et al. 2015). For PCB metrics, we found positive associations with ∑37-PCBs and ∑5-PCBs, but non-association with dioxin-like or non-dioxin like PCBs, Arochor or non-Arochor PCBs. This suggested that dietary exposure was responsible for the increase of total PCBs in serum, and the structure and the source of PCBs had an impact on their toxicokinetics (Hu et al. 2010; Hu et al. 2014; Matthews and Dedrick 1984).

This study for the first time showed that BMI modified the association between dietary intake and serum level of PCBs. In participants with underweight or normal-weight, dietary exposure was positively associated with serum concentrations of higher chlorinated PCB congeners, ∑37-PCBs, ∑5-PCBs, and ∑7DL-PCBs. However, we did not find the same association among individuals who were overweight or obese. Although we found a positive association of dietary exposure with PCB 153 in overweight individuals, the association was much weaker compared to normal/underweight individuals. Our findings were consistent with our previous AESOP study and other work showing that there was an inverse association between BMI and serum level of PCBs (Koh et al. 2016; Mullerova et al. 2008). PCBs accumulate in lipid-rich tissues, particularly adipose tissues, because of their lipophilicity. We conjecture that if participants are overweight and obese, recent dietary exposure will have less influence on the serum PCBs considering their higher body mass and adiposity. In addition, rodent studies have found that the concentrations of PCBs and their hydroxylated metabolites were higher in liver, stomach, kidney, skin and adipose tissues than in blood (ATSDR 2010; Hu et al. 2010; Hu et al. 2014; Hu et al. 2015; Thorne et al. 2015). As PCBs accumulate largely in other tissues instead of blood, serum levels of PCBs may not be an accurate biomarker of the body burden of PCBs, especially for individuals with high BMI. On the other hand, metabolic homeostasis between adipose tissue and blood is also associated with BMI–circulation of PCBs that were released from adipose tissue in blood is higher among individuals who have lower BMI (Chakaroun et al. 2012; Domazet et al. 2020; Xue and Ideraabdullah 2016). Another factor that needed to be considered for the role of BMI in PCB toxicokinetics is birth cohort effect. PCB concentration in food has been decreasing since its ban (Saktrakulkla et al. 2020), whereas BMI has increased in the past decades (Flegal et al. 1998; Ogden et al. 2004). Simulation studies of PCB 153 suggested that during time of declining exposure, elimination rate is a more determinant factor of PCB in body and individuals with higher BMI eliminate PCB from their body more slowly compared to those with lower BMI (Dzierlenga et al. 2019; Wood et al. 2016). Thus, the ongoing dietary exposure has less impact on PCB in blood while PCBs are decreasing, particular among individuals with higher BMI. Future longitudinal studies are needed to evaluate whether the effect modification observed depends on our study period (year 2003–2004).

Our study had several strengths. First, the study population was nationally representative, which allows us to generalize the findings to the U.S. population. Second, we assessed dietary exposure to PCBs at commodity levels which is more appropriate for environmental chemical risk assessment. Finally, we modelled PCB congeners in serum based on their toxicity (dioxin-like vs. non-dioxin-like) and source (Aroclors vs. non-Aroclors) instead of total PCBs. There were also several limitations in our study. Although 24-hour diet recalls can provide accurate and detailed information on food over a short-term, it may not reflect participants’ usual dietary intake because of day-to-day variation of diet. Especially for higher-chlorinated PCBs that have long half-lives, long-term diet is better to capture the eating patterns. Thus, it increased the likelihood of misclassification of dietary exposure. However, the misclassification was likely to be non-differential and of less concern, because participants were unaware of their body burden of PCBs. An additional limitation was that we used a Canadian database of PCBs in foods due to a lack of a total diet study of PCBs in the U.S. Nevertheless, this concern can be reduced for two reasons: 1) Canada and U.S. were classified into the same food consumption cluster for the similarity of commodity consumption (WHO 2016), and 2) our sensitivity analysis for geographic variability yielded similar findings using the datasets from Vancouver and Ottawa, which represent coastal and continental diets, respectively. Finally, our study used one cycle of NHANES, thus we cannot evaluate the period and birth cohort effect.

5. Conclusions

In conclusion, in this nationally representative study, we found that participants who were older and non-Hispanic tended to have a higher serum level of PCB. We also found that a significant association between dietary PCB intake and serum levels of several PCB congeners and BMI modified the associations, with a stronger association among normal/underweight individuals. Future prospective studies are needed to replicate and confirm these findings.

Supplementary Material

1

Highlights.

  • This is the first U.S. nationally representative estimate of dietary PCB exposure.

  • We found positive associations between dietary intakes and serum levels of PCBs.

  • This association was strongest among normal weight individuals.

  • Among overweight and obese individuals, no such association was observed.

Acknowledgement

We thank Dr. Linda Snetselaar and Dr. Tyler Titcomb for assistance with food commodity data, and Sarah Perry for assistance with statistical analysis. This analysis was funded by the National Institute of Environmental Health Science through the Iowa Superfund Research Program (NIH P42ES013661). Facilities support was funded through the Environmental Health Sciences Research Center (NIH P30 ES005605).

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Reference

  1. Ahluwalia N, Dwyer J, Terry A, Moshfegh A, Johnson C. 2016. Update on NHANES Dietary Data: Focus on Collection, Release, Analytical Considerations, and Uses to Inform Public Policy. Adv Nutr 7:121–134. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Ahuja JK, Moshfegh AJ, Holden JM, Harris E. 2013. USDA food and nutrient databases provide the infrastructure for food and nutrition research, policy, and practice. J Nutr 143:241s–249s. [DOI] [PubMed] [Google Scholar]
  3. Ampleman MD, Martinez A, DeWall J, Rawn DF, Hornbuckle KC, Thorne PS. 2015. Inhalation and dietary exposure to PCBs in urban and rural cohorts via congener-specific measurements. Environ Sci Technol 49:1156–1164. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Anezaki K, Nakano T. 2014. Concentration levels and congener profiles of polychlorinated biphenyls, pentachlorobenzene, and hexachlorobenzene in commercial pigments. Environ Sci Pollut Res Int 21:998–1009. [DOI] [PubMed] [Google Scholar]
  5. Apostoli P, Magoni M, Bergonzi R, Carasi S, Indelicato A, Scarcella C, et al. 2005. Assessment of reference values for polychlorinated biphenyl concentration in human blood. Chemosphere 61:413–421. [DOI] [PubMed] [Google Scholar]
  6. Armitage JM, Gobas FA. 2007. A terrestrial food-chain bioaccumulation model for POPs. Environ Sci Technol 41:4019–4025. [DOI] [PubMed] [Google Scholar]
  7. ATSDR. 2000. Chemical and physical information. Accessed 10 January 2020. https://www.atsdr.cdc.gov/toxprofiles/tp17-c4.pdf.
  8. ATSDR. 2010. Toxicological Profile for Polychlorinated Biphenyls. Agency for Toxic Substances and Disease Registry. [PubMed] [Google Scholar]
  9. Awata H, Linder S, Mitchell LE, Delclos GL. 2017. Association of Dietary Intake and Biomarker Levels of Arsenic, Cadmium, Lead, and Mercury among Asian Populations in the United States: NHANES 2011–2012. Environ Health Perspect 125:314–323. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Berg V, Nost TH, Sandanger TM, Rylander C. 2018. Predicting human plasma concentrations of persistent organic pollutants from dietary intake and socio-demographic information in the Norwegian Women and Cancer study. Environ Int 121:1311–1318. [DOI] [PubMed] [Google Scholar]
  11. Bergkvist C, Oberg M, Appelgren M, Becker W, Aune M, Ankarberg EH, et al. 2008. Exposure to dioxin-like pollutants via different food commodities in Swedish children and young adults. Food Chem Toxicol 46:3360–3367. [DOI] [PubMed] [Google Scholar]
  12. Borrell LN, Factor-Litvak P, Wolff MS, Susser E, Matte TD. 2004. Effect of socioeconomic status on exposures to polychlorinated biphenyls (PCBs) and dichlorodiphenyldichloroethylene (DDE) among pregnant African-American women. Arch Environ Health 59:250–255. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Cao LL, Yan CH, Yu XD, Tian Y, Zhao L, Liu JX, et al. 2011. Relationship between serum concentrations of polychlorinated biphenyls and organochlorine pesticides and dietary habits of pregnant women in Shanghai. Sci Total Environ 409:2997–3002. [DOI] [PubMed] [Google Scholar]
  14. CDC. 2003. Laboratory Procedure Manual: PCBs and Persistent Pesticides in Serum. Part accessed 19 January 2020. https://www.cdc.gov/nchs/data/nhanes/nhanes_03_04/l28_c_met_%20PCBs_and_Persistent_Pesticides.pdf:Atlanta,GA:Centers for Disease Control and Prevention (CDC)National Center for Environmental Health. [Google Scholar]
  15. CDC. 2013. Centers for Disease Control and Prevention. National Health and Nutrition Examination Survey: Analytic Guide- lines. Accessed 10 may 2020 https://wwwn.cdc.gov/nchs/nhanes/analyticguidelines.aspx.
  16. CDC. 2015. Centers for Disease Control and Prevention, National Center for Health Statistics. Adult Tobacco Use Informa- tion: Glossary. Accessed 10 may 2020 https://www.cdc.gov/nchs/nhis/tobacco/tobacco_glossary.htm.
  17. CDC/NCHS. 2011. Variance Estimation. Part accessed 17 January 2020. http://www.cdc.gov/nchs/tutorials/nhanes/surveydesign/VarianceEstimation/intro.htm Centers for Disease Control and Prevention/ National Center for Health Statistics. [Google Scholar]
  18. Cerná M, Malý M, Grabic R, Batáriová A, Smíd J, Benes B. 2008. Serum concentrations of indicator PCB congeners in the Czech adult population. Chemosphere 72:1124–1131. [DOI] [PubMed] [Google Scholar]
  19. Chakaroun R, Raschpichler M, Kloting N, Oberbach A, Flehmig G, Kern M, et al. 2012. Effects of weight loss and exercise on chemerin serum concentrations and adipose tissue expression in human obesity. Metabolism 61:706–714. [DOI] [PubMed] [Google Scholar]
  20. Chen JW, Wang SL, Yu HY, Liao PC, Lee CC. 2006. Body burden of dioxins and dioxin-like polychlorinated biphenyls in pregnant women residing in a contaminated area. Chemosphere 65:1667–1677. [DOI] [PubMed] [Google Scholar]
  21. Choi AL, Levy JI, Dockery DW, Ryan LM, Tolbert PE, Altshul LM, et al. 2006. Does living near a Superfund site contribute to higher polychlorinated biphenyl (PCB) exposure? Environ Health Perspect 114:1092–1098. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Curtin LR, Mohadjer LK, Dohrmann SM, Montaquila JM, Kruszan-Moran D, Mirel LB, et al. 2012. The National Health and Nutrition Examination Survey: Sample Design, 1999–2006. Vital Health Stat 2:1–39. [PubMed] [Google Scholar]
  23. Dabeka R, Cao XL. 2013. The Canadian total diet study design: 1992–1999. Food Addit Contam Part A Chem Anal Control Expo Risk Assess 30:477–490. [DOI] [PubMed] [Google Scholar]
  24. DeVoto E, Kohlmeier L, Heeschen W. 1998. Some dietary predictors of plasma organochlorine concentrations in an elderly German population. Arch Environ Health 53:147–155. [DOI] [PubMed] [Google Scholar]
  25. Domazet SL, Grontved A, Jensen TK, Wedderkopp N, Andersen LB. 2020. Higher circulating plasma polychlorinated biphenyls (PCBs) in fit and lean children: The European youth heart study. Environ Int 136:105481. [DOI] [PubMed] [Google Scholar]
  26. Domingo JL, Bocio A. 2007. Levels of PCDD/PCDFs and PCBs in edible marine species and human intake: a literature review. Environ Int 33:397–405. [DOI] [PubMed] [Google Scholar]
  27. Donato F, Magoni M, Bergonzi R, Scarcella C, Indelicato A, Carasi S, et al. 2006. Exposure to polychlorinated biphenyls in residents near a chemical factory in Italy: the food chain as main source of contamination. Chemosphere 64:1562–1572. [DOI] [PubMed] [Google Scholar]
  28. Dzierlenga MW, Yoon M, Wania F, Ward PL, Armitage JM, Wood SA, et al. 2019. Quantitative bias analysis of the association of type 2 diabetes mellitus with 2,2’,4,4’,5,5’-hexachlorobiphenyl (PCB-153). Environ Int 125:291–299. [DOI] [PubMed] [Google Scholar]
  29. EFSA. 2018. Panel on Contaminants in the Food Chain (CONTAM). Risk for animal and human health related to the presence of dioxins and dioxin-like PCBs in food and feed. EFSA J, 16(11):e05333 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. EPA. 2019. Commonly Consumed Food Commodities. Part accessed 26 February 2020. https://www.epa.gov/minimum-risk-pesticides/commonly-consumed-food-commodities.
  31. Ferriby LL, Knutsen JS, Harris M, Unice KM, Scott P, Nony P, et al. 2007. Evaluation of PCDD/F and dioxin-like PCB serum concentration data from the 2001–2002 National Health and Nutrition Examination Survey of the United States population. J Expo Sci Environ Epidemiol 17:358–371. [DOI] [PubMed] [Google Scholar]
  32. Flegal KM, Carroll MD, Kuczmarski RJ, Johnson CL. 1998. Overweight and obesity in the United States: prevalence and trends, 1960–1994. Int J Obes Relat Metab Disord 22:39–47. [DOI] [PubMed] [Google Scholar]
  33. Furberg AS, Sandanger T, Thune I, Burkow IC, Lun E. 2002. Fish consumption and plasma levels of organochlorines in a female population in Northern Norway. J Environ Monit 4:175–181. [DOI] [PubMed] [Google Scholar]
  34. Gonzalez-Alzaga B, Lacasana M, Hernandez AF, Arrebola JP, Lopez-Flores I, Artacho-Cordon F, et al. 2018. Serum concentrations of organochlorine compounds and predictors of exposure in children living in agricultural communities from South-Eastern Spain. Environ Pollut 237:685–694. [DOI] [PubMed] [Google Scholar]
  35. Grimm FA, Hu D, Kania-Korwel I, Lehmler HJ, Ludewig G, Hornbuckle KC, et al. 2015. Metabolism and metabolites of polychlorinated biphenyls. Crit Rev Toxicol 45:245–272. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Grossman E. 2013. Nonlegacy PCBs: pigment manufacturing by-products get a second look. Environ Health Perspect 121:A86–93. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. HHS. 2017. US Department of Health and Human Services. Dietary guidelines for Americans 2015–2020. [Google Scholar]
  38. Hu D, Hornbuckle KC 2010. Inadvertent polychlorinated biphenyls in commercial paint pigments. Environ Sci Technol 44:2822–2827. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Hu X, Adamcakova-Dodd A, Lehmler HJ, Hu D, Kania-Korwel I, Hornbuckle KC, et al. 2010. Time course of congener uptake and elimination in rats after short-term inhalation exposure to an airborne polychlorinated biphenyl (PCB) mixture. Environ Sci Technol 44:6893–6900. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Hu X, Adamcakova-Dodd A, Thorne PS. 2014. The fate of inhaled (14)C-labeled PCB11 and its metabolites in vivo. Environ Int 63:92–100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Hu X, Adamcakova-Dodd A, Lehmler HJ, Gibson-Corley K, Thorne PS. 2015. Toxicity Evaluation of Exposure to an Atmospheric Mixture of Polychlorinated Biphenyls by Nose-Only and Whole-Body Inhalation Regimens. Environ Sci Technol 49:11875–11883. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. IARC. 2013. IARC monographs on the evaluation of the carcinogenic risk of chemicals to humans. Lyon: International Agency for Reasearch on Cancer. 107: Polychlorinated biphenyls and polybrominated biphenyls. . [Google Scholar]
  43. Jensen AA. 1989. Background levels in humans. Halogenated biphenyls, terphenyls, naphthalenes, dibenzodioxins and related products:345–364. [Google Scholar]
  44. Jin W, Otake M, Eguchi A, Sakurai K, Nakaoka H, Watanabe M, et al. 2017. Dietary Habits and Cooking Methods Could Reduce Avoidable Exposure to PCBs in Maternal and Cord Sera. Sci Rep 7:17357. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Johnson-Restrepo B, Kannan K, Rapaport DP, Rodan BD. 2005. Polybrominated diphenyl ethers and polychlorinated biphenyls in human adipose tissue from New York. Environ Sci Technol 39:5177–5182. [DOI] [PubMed] [Google Scholar]
  46. Koh WX, Hornbuckle KC, Wang K, Thorne PS. 2016. Serum polychlorinated biphenyls and their hydroxylated metabolites are associated with demographic and behavioral factors in children and mothers. Environ Int 94:538–545. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Kuczmarski RJ, Ogden CL, Guo SS, Grummer-Strawn LM, Flegal KM, Mei Z, et al. 2002. 2000 CDC Growth Charts for the United States: methods and development. Vital Health Stat 11:1–190. [PubMed] [Google Scholar]
  48. Laden F, Neas LM, Spiegelman D, Hankinson SE, Willett WC, Ireland K, et al. 1999. Predictors of plasma concentrations of DDE and PCBs in a group of US women. Environ Health Perspect 107:75–81. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Llobet JM, Bocio A, Domingo JL, Teixido A, Casas C, Muller L. 2003. Levels of polychlorinated biphenyls in foods from Catalonia, Spain: estimated dietary intake. J Food Prot 66:479–484. [DOI] [PubMed] [Google Scholar]
  50. Matthews HB, Dedrick RL. 1984. Pharmacokinetics of PCBs. Annu Rev Pharmacol Toxicol 24:85–103. [DOI] [PubMed] [Google Scholar]
  51. McKelvey W, Chang M, Arnason J, Jeffery N, Kricheff J, Kass D. 2010. Mercury and polychlorinated biphenyls in Asian market fish: a response to results from mercury biomonitoring in New York City. Environ Res 110:650–657. [DOI] [PubMed] [Google Scholar]
  52. Miller D. 2013. Food Commodity Consumption Data and New Tools used by US EPA’s Office of Pesticide Programs. In: ISEE Conference, Vol. 2013. [Google Scholar]
  53. Moon HJ, Lim JE, Jee SH. 2017. Association between serum concentrations of persistent organic pollutants and smoking in Koreans: A cross-sectional study. J Epidemiol 27:63–68. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Mullerova D, Kopecky J, Matejkova D, Muller L, Rosmus J, Racek J, et al. 2008. Negative association between plasma levels of adiponectin and polychlorinated biphenyl 153 in obese women under non-energy-restrictive regime. Int J Obes (Lond) 32:1875–1878. [DOI] [PubMed] [Google Scholar]
  55. Ogden CL, Fryar CD, Carroll MD, Flegal KM. 2004. Mean body weight, height, and body mass index, United States 1960–2002. Adv Data:1–17. [PubMed] [Google Scholar]
  56. Saktrakulkla P, Lan T, Hua J, Marek RF, Thorne PS, Hornbuckle KC. 2020. Polychlorinated Biphenyls in Food. Environ Sci Technol 54:11443–11452. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Schecter A, Cramer P, Boggess K, Stanley J, Papke O, Olson J, et al. 2001. Intake of dioxins and related compounds from food in the U.S. population. J Toxicol Environ Health A 63:1–18. [DOI] [PubMed] [Google Scholar]
  58. Schecter A, Colacino J, Haffner D, Patel K, Opel M, Papke O, et al. 2010. Perfluorinated compounds, polychlorinated biphenyls, and organochlorine pesticide contamination in composite food samples from Dallas, Texas, USA. Environ Health Perspect 118:796–802. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Stellman S, Djordjevic M, Muscat J, Citron M, White A, Kemeny M, et al. Adipose and serum levels of organochlorinated pesticide and PCB residues in Long Island women: Association with age and body mass. In: Proceedings of the AMERICAN JOURNAL OF EPIDEMIOLOGY, 1997, Vol. 145JOHNS HOPKINS UNIV SCHOOL HYGIENE PUB HEALTH 111 MARKET PLACE, STE 840 …, 81–81. [Google Scholar]
  60. Thorne P, Ampleman M, Hu X, Adamcakova-Dodd A, Hornbuckle K. 2015. Uptake of inhaled polychlorinated biphenyls (PCBs) in a human longitudinal cohort study and animal inhalation studies. Eur Respiratory Soc 46: PA4096. [Google Scholar]
  61. Uehara R, Nakamura Y, Matsuura N, Kondo N, Tada H. 2007. Dioxins in human milk and smoking of mothers. Chemosphere 68:915–920. [DOI] [PubMed] [Google Scholar]
  62. UNEP. 2013. Guidance on the global monitoring plan for persistent organic pollutants. Stockholm Convention on Persistent Organic Pollutants. Accessed 10 December 2020. http://chm.pops.int/Implementation/GlobalMonitoringPlan/Overview/tabid/83/Default.aspx. [Google Scholar]
  63. Van den Berg M, Birnbaum LS, Denison M, De Vito M, Farland W, Feeley M, et al. 2006. The 2005 World Health Organization reevaluation of human and Mammalian toxic equivalency factors for dioxins and dioxin-like compounds. Toxicol Sci 93:223–241. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. WHO. 2014. Surveillance and Population-Based Prevention, Preven- tion of Noncommunicable Diseases Department, World Health Organization. Global Physical Activity Question- naire (GPAQ) Analysis Guide. Geneva, Switzerland. Accessed 10 may 2020 https://www.who.int/ncds/surveillance/steps/resources/GPAQ_Analysis_Guide.pdf. [Google Scholar]
  65. WHO. 2016. Safety evaluation of certain food additives and contaminants. In: Supplement 1: Non-dioxin-like polychlorinated biphenyls. Geneva:World Health Organization. Prepared by the eightieth meeting of the Joint FAO/WHO Expert Committee on Food Additives (JECFA). [Google Scholar]
  66. Wood SA, Xu F, Armitage JM, Wania F. 2016. Unravelling the Relationship between Body Mass Index and Polychlorinated Biphenyl Concentrations Using a Mechanistic Model. Environ Sci Technol 50:10055–10064. [DOI] [PubMed] [Google Scholar]
  67. Xue J, Ideraabdullah FY. 2016. An assessment of molecular pathways of obesity susceptible to nutrient, toxicant and genetically induced epigenetic perturbation. J Nutr Biochem 30:1–13. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS1709747-supplement-1.docx (68.2KB, docx)

RESOURCES