Abbreviations:

ADME = absorption, distribution, metabolism, excretion
ATSDR = Agency for Toxic Substances and Disease Registry
BMD = benchmark dose
CHMS = Canadian Health Measures Survey
DDE = dichlorodiphenyldichloroethylene
DDT = dichlorodiphenyltrichloroethane
GerES = German Environmental Survey
LOAEL = lowest observed adverse effect level
MRL = minimum risk level
NHANES = National Health and Nutrition Examination Survey
NOAEL = no observed adverse effect level
PBDE = polybrominated diphenyl ether
PBPK = physiologically-based pharmacokinetic
PCB = polychlorinated biphenyl
PCDD = polychlorinated dibenzo-p-dioxins
PCDF = polychlorinated dibenzofurans
RfC = reference concentration
RfD = reference dose
TCDD = tetrachlorodibenzodioxin
TDI = tolerable daily intake

1. Introduction

Guideline values for environmental contaminants describe exposure thresholds likely to have minimal risk of adverse effects in humans. The Environmental Protection Agency, for example, uses the concept of reference dose (RfD) or reference concentration (RfC), which are estimates of daily oral or inhalational exposures, respectively, in humans that are likely to have no adverse non-cancer effects over a lifetime. These estimates take into consideration effects of chemicals on sensitive human subpopulations [1, 2]. The Agency for Toxic Substances and Disease Registry (ATSDR) defines minimum risk level (MRL) of acute (1 ‒ 14 days), intermediate (15 ‒ 364 days), or chronic duration (1 year or longer) to describe oral and inhalational exposures without risk for adverse non-cancer effects [3]. Tolerable daily intakes (TDI) are used by Health Canada to describe oral exposures without adverse non-cancer effects over a lifetime [4]. Guideline values are often derived by applying uncertainty factors to estimates of no observed adverse effect level (NOAEL), lowest observed adverse effect level (LOAEL) or benchmark dose (BMD) from toxicological studies in animals. Uncertainty factors may account for uncertainties in extrapolating from animal data to humans, extrapolating from a LOAEL to a NOAEL, length of exposure (e.g. from acute to chronic), and variation among human populations [5].

The magnitude of internal exposure has often been inferred based on estimates of external intake. While this method provides a rough estimate of exposure it may not be representative of the dose that ultimately arrives in the bloodstream or at target organs where toxicity occurs [6]. Chemicals are known to undergo processes of absorption, distribution, metabolism, and excretion—these pharmacokinetic processes will transform the external dose into a chemical-specific internal dose, which is the true parameter of interest in characterizing risk. A biomonitoring equivalent is the conversion of an external guideline value, such as a RfD, MRL, or TDI, to an internal dose against which biomonitoring data can be directly compared [7, 8]. Biomonitoring equivalents have been developed for several environmental contaminants, such as polybrominated diphenyl ether (PBDE) -99 [9], dichlorodiphenyldichloroethylene (p, p’-DDE) and dichlorodiphenyltrichloroethane (DDT) [10], toluene [11], hexachlorobenzene [12], and 2, 4-dichlorophenoxyacetic acid [13].

To our knowledge, biomonitoring equivalents have not yet been developed for polychlorinated biphenyls (PCBs), a ubiquitous class of environmental contaminant. Levels of PCBs have been measured in human populations around the world. In a risk assessment context, biomonitoring equivalents for PCBs could more accurately guide the interpretation of the available human biomonitoring data. Polychlorinated biphenyls contain up to 209 congeners and each congener has different chemical properties, such as biological half-life. Therefore, the development of biomonitoring equivalents for PCBs is a challenge. In this paper we provide a broad overview of the information needed to develop biomonitoring equivalents for PCBs. We begin by describing background information and worldwide PCB biomonitoring data, followed by a summary of ADME (absorption, distribution, metabolism, excretion) processes, external guideline values, and pharmacokinetic modeling studies of PCBs in animals and humans. Lastly, we present some strategies to address anticipated research challenges in the development of biomonitoring equivalents for this complex contaminant.

2. Background on polychlorinated biphenyls

Although the production, use, and trade of PCBs was prohibited under the Stockholm Convention since 2004 [14] and international production stopped since 1993 [15], these chemicals persist in air, water, soil, biota, and human tissues. PCBs are synthetic materials that were produced for coolant and lubricant properties in electrical equipment, such as capacitors and transformers, and in plasticizers, oils, inks, paints, adhesives, and waxes [15]. Breivik et al. estimated that total global production of PCBs amounted to about 1.3 million tons, of which greater than 70 percent were tri-, tetra-, and pentachlorinated homologues [16]. About half of this production came from the United States and nearly all consumption occurred in the Northern hemisphere [16].

Individual PCB congeners vary in the degree and position of chlorine atoms, resulting in 209 possible congeners. The industrial manufacture of PCBs was mostly in the form of complex commercial PCB mixtures defined by degree of chlorination patterns, such as Aroclors in the United States, Kanechlors in Japan, and Clophens in Germany. The number and positioning of chlorine atoms on the biphenyl ring will determine potential for atmospheric volatilization, environmental degradation, bioaccumulation, and metabolism. Generally, the higher chlorinated PCB congeners are more resistant to environmental breakdown and bioaccumulate in biota [5]. In addition, the positioning of the chlorine atoms will determine toxic effects. Dioxin-like congeners, for example, have a coplanar structure with no or maximum of one chlorine atom in the ortho positions. These congeners act through the aryl hydrocarbon receptor and cause toxicity similar to 2, 3, 7, 8-tetrachlorodibenzodioxin (TCDD) [17].

Exposure to PCBs has been shown to have many adverse effects in wildlife and humans and, therefore, their persistence in the environment is of high concern. Most effects in humans have been observed in the settings of occupational exposures or poisoning incidents and include increase in liver enzymes, gastrointestinal symptoms, increased thyroid gland volume and risk for goiter, upper respiratory tract symptoms, joint pain, skin irritation and chloracne, hematological and immunological effects, neurological effects, reproductive effects, and cancer [5]. The International Agency for Research on Cancer has classified PCBs as Group 2A, probably carcinogenic to humans [18].

3. Worldwide biomonitoring data

PCB biomonitoring data for children, adolescents, and adults are available from several countries through national surveys, such as the National Health and Nutrition Examination Survey (NHANES) in the United States, the Canadian Health Measures Survey (CHMS), and the German Environmental Survey (GerES). Table 1 presents data from United States, Canada, Australia, Germany, Spain, and Belgium. In addition, monitoring of PCBs in remote populations, such as the Inuit, has been conducted [20, 21]. Table 2 shows the sum of PCB congeners detected in breast milk samples around the world. Biomonitoring equivalents could be used to interpret this worldwide PCB biomonitoring data.

4. Absorption, distribution, metabolism, and excretion (ADME)

The ATSDR contains comprehensive data on the ADME of PCB congeners in animals and humans. The information described in this section comes mostly from that report [5, 22]. The primary route of human exposure to PCBs is through oral ingestion from contaminated foods, drinking water, and breast milk. Other minor routes of exposure are inhalational and dermal absorption [22]. PCBs are lipophilic chemicals and, therefore, absorption from the gastrointestinal tract to blood lipids occurs passively. Once PCBs are absorbed they distribute preferentially to adipose tissue and liver where they may remain for years. Commonly detected congeners in human tissues are PCB-138, 153, and 180 [22].

PCBs can be metabolized by cytochrome P450 isozymes to polar metabolites which can then undergo phase 2 metabolism by conjugation with glutathione and glucuronic acid. Arene oxides of PCBs are formed by CYP 1A1, CYP 1A2, CYP 2B1, CYP 2B2, and CYP 3A and are transformed to hydroxylated aromatic compounds or methylsulfonyl metabolites. The rate of metabolism decreases with higher chlorinated congeners. Phenolic metabolites predominate, although other metabolites such as trans-dihydrodiols, polyhydroxylated congeners, and methyl ether derivatives may also be formed. Hydroxylated metabolites accumulate in lung, liver, and kidneys. Due to metabolism of parent compounds and selective retention of certain congeners in body tissues (e.g. PCB-153), the original mixture of PCBs ingested will not be the same as the congeners subsequently detected in serum, adipose, and breast milk.

PCBs are primarily eliminated in feces as parent form and as metabolites in urine and bile. Elimination half-lives vary greatly depending on congener; for example, half-life for PCB-28 is estimated at 1.4 years whereas the half-life for PCB-163 is more than 20 years. Other congeners have been found to have infinite half-lives, indicating repeated exposures or absence of decline over the experimental duration.

5. Guideline values

Guideline values for PCBs were collected from sources of the World Health Organization, European and North American jurisdictions (e.g. French Food Safety Agency, Health Canada, Environmental Protection Agency), and occupational organizations (e.g. National Institute for Occupational Safety and Health) (Tables 3 and Table 4) [5, 22, 24, 25, 26, 27, 28, 29, 30]. The reference values pertain to total PCBs, total non-dioxin like PCBs, dioxin-like PCBs, or PCB mixtures (e.g. Aroclor 1254) and ingestion through food, water or breast milk. The maximum reference value for ingestion is 5 µg/kg/day based on toxic effects of Phenochlor-DP6 in rats (CSHPF 1991). More recent sources have lowered ingestion references values, ranging from a low of 0.01 µg/kg/day for seven PCB indicators in food (AFSSA 2007) to a high of 0.13 μg/kg/day for total non-dioxin like PCBs (Health Canada 2010). For dioxins, a reference value of 2.33 pg TEQ/kg/day has been specified by the Joint FAO/WHO Expert Committee on Food Additives (JECFA 2001). A theoretical TDI for dioxin-like PCBs is 1.63 pg TEQ/kg/day, given that these PCBs constitute 70% of dioxin mixtures [29]. The European Food Safety Authority 2005 indicated benchmark dose lower confidence limits in breast milk of 0.63-0.71 µg/g lipid for total PCBs based on cognitive outcomes in children exposed prenatally and 65 pg TEQ/g lipid for dioxin-like PCBs, polychlorinated dibenzo-p-dioxins (PCDD) and polychlorinated dibenzofurans (PCDF) based on neurological and immune effects with perinatal exposures. Three occupational reference standards for inhalational and dermal exposures were identified for Aroclor 1242 and Aroclor 1254 of either 0.5 mg/m³ or 1 mg/m³.

6. Pharmacokinetic modeling

The conversion of an external PCB reference value to a corresponding internal dose requires the use of a pharmacokinetic model that describes the absorption, distribution, metabolism and excretion of a chemical in the body. The model description can vary from simple relationship between dose exposure and internal metrics to highly complex structures with saturable processes. The level of detail in the model structure will depend on the quality and quantity of data available.

6.1. Modeling in animals

Several pharmacokinetic modeling studies of PCBs have been conducted in wildlife species, such as marine mammals, birds, and polar bears [31, 32, 33, 34, 35, 36, 37, 38, 39]. Pharmacokinetic modeling of PCBs has also been carried out in experimental animals, such as rats, mice and fish [40, 41, 42, 43, 44, 45, 46, 47, 48, 49] and in swine and sheep [49]. PCB-153 has been most commonly modeled, although some work has also been done with other congeners and with PCB mixtures, such as Aroclor 1242, 1254, and 1260. One study focused on dioxin-like PCBs and their transfer from feed and soils to eggs of laying hens [50].

Models of various complexities have been developed with the simplest using a single compartment model that represented the whole organism [48]. In the single compartment model, bioaccumulation and elimination of Aroclor 1254 was studied in striped bass and parameters for absorption rate constant and elimination rate constant were estimated. Several studies employed a two-compartment model to represent, for example, blubber and rest of body in pilot whales [30], gut and rest of body in ringed seal [33], central and fat in laying hens [50], and yolk + albumen and embryo in herring gulls [36]. The most complex scheme was a physiologically-based pharmacokinetic (PBPK) model that incorporated multiple compartments to represent lungs, blood, perfused adipose, deep adipose, bone, brain, muscle, kidney, liver, and intestine in East Greenland polar bears [38]. A risk quotient analysis was conducted by estimating critical body residues of many contaminants, including PCBs, based on reproductive endpoints and then comparing observed contaminant levels measured in polar bears with the critical body residues to determine the proportion at risk for reproductive toxicity [38]. Other studies introduced greater complexity by modeling pregnancy, birth, and lactational processes in females [35] and incorporating time-dependent variations in pharmacokinetic parameters [38, 43, 46, 52].

The pharmacokinetic models were used to estimate lifetime exposure and bioaccumulation [31, 32, 33, 36, 37, 52], influence of blubber temperature gradients on PCB concentrations [33], mother-offspring transfer of PCBs [35, 36, 41], protein binding [39], interactions between congeners [41, 43], and the effect of chlorine position on PCB kinetic behaviour [43].

6.2. Modeling in humans

Pharmacokinetic modeling of PCBs has been conducted in pregnant women [52], lactating women [54, 55, 56], infants and children [57, 58, 59], women with breast cancer [60, 61], Inuit [56, 58, 59, 62, 63, 64]workers in a contaminated building [64], and general populations [66, 67, 68, 69](Table 5). As with animal modeling studies, the human pharmacokinetic models varied in complexity from simple one-compartment models [55, 63, 66, 68, 69], two-compartment models representing maternal and fetal lipids or liver and fat [53, 64], to a 22 compartment model in lactating women [53]. The most commonly modeled congener was PCB-153, although other congeners, such as PCB-28, PCB-99, PCB-138, PCB-180, and PCB-199 were also examined. In breast cancer studies, pharmacokinetic models were used to estimate lifetime exposure or exposure during susceptible time windows, which may more accurately represent exposures at the time of cancer diagnosis [60, 61]. Similarly, pharmacokinetic modeling was used to simulate exposures during specific time periods (i.e. prenatal, lactational, postnatal) and their association with mental and psychomotor development or behaviour in infants and children [57, 58].

The modeling studies conducted for Inuit estimated PCB-153 intake through breast milk using reverse dosimetry [55], simulation of PCB-153 prenatal and postnatal exposures [57], infant exposure to PCB-138, 153, and 180 through placental transfer and breast feeding [58], fate of PCB-99 and PCB-153 in blood and other tissues using a generic PBPK model [61], estimation of PCB-153 cancer risk based on prenatal, postnatal, and lifetime exposures [62], and estimation of daily intake doses of PCB-77, PCB-126, PCB-153, and PCB-169 with comparison to a reference value [63].

7. Addressing challenges for the development of biomonitoring equivalents

The development of biomonitoring equivalents using pharmacokinetic modeling is complex for PCBs because of the presence of multiple congeners with different pharmacokinetic and toxicological properties. In addition, there are multiple pathways of exposure (e.g. ingestion, inhalation, dermal absorption, placental and lactational transfer), species differences in kinetics, and adverse endpoints which must be taken into consideration when modeling PCB pharmacokinetics.

The following is a list of procedural steps that needed to be followed to develop useful biomonitoring equivalents based on existing information:

1. Deciding which congeners to model: A list of PCB congeners that are abundant in biota and/or have high potential for toxicity, are shown in Box 1. These congeners fall under five homologue groups (i.e. tri-, tetra-, penta-, hexa-, and hepta-). Prioritization should be given to developing biomonitoring equivalents for these congeners as they are most relevant for risk assessments.

Box 1. Abundant and/or toxic PCB congeners.

PCB-28 (2, 4, 4I trichlorobiphenyl)	PCB-105 (2, 3, 3I, 4, 4I pentachlorobiphenyl)	PCB-156 (2, 3, 3I, 4, 4I, 5 hexachlorobiphenyl)
PCB-52 (2, 2I, 5, 5I tetrachlorobiphenyl)	PCB-118 (2, 3I, 4, 4I, 5 pentachlorobiphenyl)	PCB-163 (2, 3, 3I, 4I, 5, 6 hexachlorobiphenyl)
PCB-74 (2, 4, 4I, 5 tetrachlorobiphenyl)	PCB-125 (2I, 3, 4, 5, 6I pentachlorobiphenyl)	PCB-168 (2, 3I, 4, 4I, 5I, 6 hexachlorobiphenyl)
PCB-77 (3, 3I, 4, 4I tetrachlorobiphenyl)	PCB-126 (3, 3I, 4, 4I, 5 pentachlorobiphenyl)	PCB-169 (3, 3I, 4, 4I, 5, 5I hexachlorobiphenyl)
PCB-100 (2, 2I, 4, 4I, 6 pentachlorobiphenyl)	PCB-128 (2, 2I, 3, 3I, 4, 4I hexachlorobiphenyl)	PCB-170 (2, 2I, 3, 3I, 4, 4I, 5 heptachlorobiphenyl)
PCB-101 (2, 2I, 4, 5, 5I pentachlorobiphenyl)	PCB-138 (2, 2I, 3, 4, 4I, 5I hexachlorobiphenyl)	PCB-180 (2, 2I, 3, 4, 4I, 5, 5I heptachlorobophenyl)
PCB-104 (2, 2I, 4, 6, 6I pentachlorobiphenyl)	PCB-153 (2, 2I, 4, 4I, 5, 5I hexachlorobiphenyl)	PCB-190 (2, 3, 3I, 4, 4I, 5, 6 heptachlorobiphenyl)

2. Choosing a guidance value: As described in Tables 3 and Table 4, different organizations have formulated various guideline values based mostly on mixtures of PCB congeners rather than individual congeners. Therefore, it will be necessary to make assumptions about the constitution of these mixtures so that biomonitoring equivalents for individual congeners can be derived.

3. Parameterization of pharmacokinetic models: A full PBPK model requires input of several parameters. Where possible, these parameters should be based on human data to minimize species differences. However, in the absence of such data, animal data or theoretical structural equations, both of which introduce greater degrees of uncertainty in the models, can be used. The ATSDR contains data on the absorption, distribution, metabolism, and excretion of PCB congeners as described previously [5, 22]. In addition, Parham et al. devised structural regression formula to calculate blood-adipose partition coefficients and metabolic rates for any of the 209 congeners based on structural characteristics [66, 67]. An important component of the modeling process will be to evaluate and transparently declare the confidence in the inputted parameters.

4. Minimizing complex scenarios: Depending on the research question, initial modeling attempts should focus on the most common routes of exposure. In general populations, ingestion of contaminated foods will be the primary exposure pathway. For lactating women, elimination of PCBs through breast milk should additionally be considered.

8. Conclusion

Biomonitoring equivalents for PCBs will be useful from a risk assessment standpoint because they can be used to interpret the biomonitoring data collected from populations around the globe. The percentage within and exceeding guidance values can be calculated for full survey samples, as well as for sensitive sub-populations, such as children, pregnant women, women of childbearing age and the elderly. These comparisons will help to contextualize the results of biomonitoring surveys and provide evidence on which to base public health interventions for the percentage of the population exceeding guidelines. In this paper we have reviewed key information and challenges for deriving biomonitoring equivalents for PCBs. Modelers, environmental epidemiologists, and risk assessors can make use of this information to evaluate worldwide PCB biomonitoring data and develop international guidelines.

Acknowledgments

We would like to thank Stella Chen for assisting with literature acquisition, organization, and data retrieval. We also acknowledge support from the Northern Contaminants Program (NCP), Canada Research Chair, and Health Canada.

Conflict of interest

All authors declare no conflicts of interest in this paper.