Human biomonitoring as a pragmatic tool to support health risk management of chemicals – Examples under the EU REACH programme

Regulatory Toxicology and Pharmacology 59 (2011) 125–132

Contents lists available at ScienceDirect

Regulatory Toxicology and Pharmacology journal homepage: www.elsevier.com/locate/yrtph

Human biomonitoring as a pragmatic tool to support health risk management of chemicals – Examples under the EU REACH programme Peter J. Boogaard a,⇑, Sean M. Hays b, Lesa L. Aylward c a

Shell Health, Shell International bv, P.O. Box 162, 2501 AN The Hague, The Netherlands Summit Toxicology, Allenspark, CO, USA c Summit Toxicology, Falls Church, VA, USA b

a r t i c l e

i n f o

Article history: Received 8 July 2010 Available online 7 October 2010 Keywords: REACH Risk assessment Risk management DNEL Biomonitoring Equivalent Biomonitoring

a b s t r a c t REACH requires health risk management for workers and the general population and introduced the concept of Derived No-Effect Level (DNEL). DNELs must be derived for all substances that are classified as health hazards. As with analogues to other health-risk based guidance values, such as reference doses (RfDs) and tolerable daily intakes (TDIs), risk to health is considered negligible if the actual exposure is less than the DNEL. Exposure assessment is relatively simple for occupational situations but more complex for the general public, in which exposure may occur via multiple pathways, routes, and media. For such complex or partially defined exposure scenarios, human biomonitoring gives a snapshot of internal or absorbed dose of a chemical and is often the most reliable exposure assessment methodology as it integrates exposures from all routes. For human risk management human biomonitoring data can be interpreted using the recently developed concept of Biomonitoring Equivalents (BE). Basically, a BE translates an established reference value into a biomarker concentration using toxicokinetic data. If the results of an exposure assessment using human biomonitoring indicate that the levels measured are below the DNEL-based BE (BEDNEL), it would indicate that the combined exposure via all potential exposure routes is unlikely to pose a risk to human health and that health risk management measures might not be needed. Hence, BEs do not challenge existing risk assessments but rather build upon them to help risk management, the ultimate goal of any risk assessment. A challenge in implementing this approach forms the limited availability of toxicokinetic information for many substances. However, methodologies such as generic physiologically-based toxicokinetic models, which allow estimation of biomarker concentrations based on physicochemical properties, are being developed for less data-rich chemicals. Use of BE by regulatory authorities will allow initial screening of population exposure to chemicals to identify those chemicals requiring more detailed risk and exposure assessment, assisting in priority setting and ultimately leading to improved product stewardship and risk management. Ó 2010 Elsevier Inc. All rights reserved.

1. Introduction and aims In December 2006, new chemicals legislation was adopted in the European Union that aims to evaluate and control environmental and human health risks for all substances on the European market (EC, 2006). This legislation is known under its acronym REACH (Registration, Evaluation, and Authorisation of Chemicals) and requires producers and importers to register all substances placed on the European Union market. For every substance that is classified for toxicological and ecotoxicological hazards, an evaluation of the risk for environment and human health is required. Both the environmental and health risk assessments are based on the fundamental assumption that risk is a function of hazard and expo-

⇑ Corresponding author. Fax: +31 70 37 72 840. E-mail address: [email protected] (P.J. Boogaard). 0273-2300/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved. doi:10.1016/j.yrtph.2010.09.015

sure. As a logical consequence, the risk assessment of chemicals involves at first the determination whether or not the chemical poses a toxicological or ecotoxicological hazard. If a hazard is identified, the next step is to establish the relationship between exposure (dose) and the effect. Once the actual exposure has been assessed, it can be used in combination with the exposure-effect relationship to estimate the actual risk to determine whether risk management measures are required. In this process, uncertainties associated with extrapolations, availability and quality of the hazard data, the exposure-effect relation, and the actual exposure are compensated with assessment factors to obtain a conservative estimate of the risk. With the introduction of the Existing Substances Regulation in the European Union in 1993, the concept of PNEC (predicted noeffect concentration) was introduced for ecotoxicological risk assessment (EC, 1993). The PNEC is the environmental concentration of a substance below which exposure to a substance is not

126

P.J. Boogaard et al. / Regulatory Toxicology and Pharmacology 59 (2011) 125–132

expected to cause adverse effects in the ecosystem. The PNEC is compared to the PEC (predicted exposure concentration) in the ecosystem and if the PEC is below the PNEC, the actual risk for the environment is deemed to be negligible. There are a number of fundamental differences between ecotoxicological and toxicological risk assessment. The most important being that in ecotoxicology the main focus is protection of a species population or ecosystem whilst in toxicology the main focus is to protect the individual. Despite this difference, in the REACH legislation a new concept, coined after PNEC, was introduced: DNEL. DNEL stands for derived no-effect level and is defined as the level of a substance above which a human should not be exposed (EC, 2006). In the risk characterisation process, the exposure of all human populations known to be or likely to be exposed should be compared to this DNEL. In practice, DNELs must be derived for all hazardous substances placed on the market in quantities exceeding 10 tonnes per year and should reflect route, duration and frequency of exposure. DNELs should be derived for occupational settings, for consumer use, and for the general population for indirect exposure via the environment. The approach to derive a DNEL for a given substance and toxicological endpoint is provided in the technical guidance made available by the European Chemicals Agency (ECHA, 2008). The standard methodology involves the setting of a point-of-departure, which is a modified dose descriptor, usually based on a no-observed adverse effect level from an animal study. To the point of departure a series of default assessment factors is applied to compensate for variation as well as uncertainties with regard to the hazard of the substance for which the DNEL is derived. When specific information on the substance is available, informed assessment factors should be used to derive a DNEL (ECETOC, 2010; Gade et al., 2008; Gebel et al., 2009). In addition, when there is an indicative or binding occupational exposure limit (IOEL or BOEL) as established by the Scientific Committee of Occupational Exposure Limits (SCOEL) or when there is an adopted occupational exposure limit set by national authority in one of the Member States, these can be used as DNEL values for occupational settings, provided they are health-based (EC, 2009). There is still substantial debate on the way DNELs should be derived and how OELs can be used (ECETOC, 2010; Schäfer et al., 2009) but this is beyond the scope of this paper. Once a DNEL has been derived, it needs to be compared to an actual exposure level. For the occupational situation, exposure assessment is relatively straightforward, as both the exposure source and situation are usually well defined and may be described with specific exposure scenarios. For a wide variety of industrial settings exposure scenarios are being developed by registrants. Verification of these exposure scenarios, however, may be more difficult, especially if there are other exposure routes than inhalation. The development of exposure scenarios for consumers is in most cases far more complex since there is wide variation in quantities and frequencies of use. Estimating exposure is particularly difficult if exposure occurs through multiple different exposure routes, e.g. dermal and inhalation. The assessment of the exposure of the general public via the environment is most complex as exposures are usually even less well defined and may involve inhalation, dermal and oral exposures. Human biomonitoring is most helpful in both the actual exposure assessment for complex scenarios and the validation of exposure scenarios. By definition, human biomonitoring is the determination of a chemical or its metabolites in bodily fluids (e.g. urine, blood or saliva), tissues (e.g. hair), or exhaled air. For poorly defined exposure scenarios, biomonitoring is often the most reliable exposure assessment methodology as it determines integrated exposure regardless the route of exposure. In risk characterisation, biomonitoring is also frequently superior to other methods

of exposure assessment, such as personal air measurements or dermal deposition assessments, because it determines the actual exposure (body burden) by capturing individual differences in behaviour (e.g. personal hygiene), physiology (e.g. respiration rate). In addition, biomonitoring may reflect differences in metabolism and hence susceptibility. When the parameter measured is the toxic compound or a compound proportionally related to the ultimate toxicant, its biomonitoring directly reflects the interand intra-individual variation and potentially reduces the uncertainty compared to external exposure measurements considerably. However, when the parameter measured is not related to the toxicity, variability in metabolism will not reduce overall variability and may on some occasions contribute to the overall variability (Boogaard, 2009). Human biomonitoring has a long history in Europe and is fully integrated in EU legislation, including REACH (EC, 1995, 2006; SCOEL, 1999a). However, guidance on how to use human biomonitoring in risk characterisation and management is very limited (ECHA, 2008). This paper aims at providing pragmatic guidance how human biomonitoring can be applied in risk characterisation using DNELs and the concept of Biomonitoring Equivalents (BEs). Historically, human biomonitoring was, by and large, restricted to occupational settings. With the developments in analytical techniques over the past decades, population-based biomonitoring has become feasible and large human biomonitoring projects, such as the NHANES (National Health and Nutrition Examination Survey) project in the USA (CDC, 2008; Paustenbach and Galbraith, 2006) and the GerES (German Environmental Survey) project in Germany (GerES, 2008; Kolossa-Gehring et al., 2007), have generated extensive (internal) exposure databases on a variety of chemicals. These surveys address general societal concerns about contamination of food, water and the environment with chemicals. The same concerns are also addressed by setting safe exposure levels for chemicals in food and in the environment. Various regulatory bodies, such as the European Food Safety Authority (EFSA), the Joint FAO/WHO Expert Committee on Food Additives (JECFA), and the USA Environmental Protection Agency (EPA), establish reference values, which represent safe, acceptable or tolerable intake levels, such as acceptable or tolerable daily intakes (ADIs and TDIs), reference doses (RfD), or minimal risk levels (MRLs). These values may be based on epidemiology and/or animal studies and have essentially the same goal as the DNEL values for consumers as defined under REACH. Although human biomonitoring has many advantages, the understanding of biomonitoring data in terms of health risks is often not straightforward. Especially the interpretation of data obtained from consumers and the general public may be complicated (Boogaard et al., 2005; Boogaard and Money, 2008). To aid in the interpretation of human biomonitoring data, the concept of Biomonitoring Equivalent (BE) was recently developed (Hays et al., 2007). BE values represent quantitative benchmarks of safe or acceptable concentrations of a chemical or its metabolite in biological specimens that are consistent with selected reference values, such as the above mentioned ADI, TDI, MRL and RfD, using the knowledge about the toxicokinetic properties of the chemical. Basically, the BE translates a reference value into a biomarker concentration by integrating the risk assessment underlying the reference value with available toxicokinetic data for a compound in order to predict steady-state biomarker concentrations consistent with those reference values. Dependent on the available scientific information, this conversion can be done with greater or lesser reliability and a communication system was proposed to communicate the results to the general public (LaKind et al., 2008). For a variety of substances BEs have recently been derived (Aylward et al., 2008a,b,c, 2009a,b; Aylward and Hays, 2008; Hays and Aylward, 2008; Hays et al., 2008).

P.J. Boogaard et al. / Regulatory Toxicology and Pharmacology 59 (2011) 125–132

Reference values have been derived and exposure data collected for a substantial number of chemical substances. For example, the US EPA has published greater than 500 toxicity reference values in the Integrated Risk Information System (IRIS). Under REACH, however, DNELs need to be derived for all hazardous substances placed on the European market at quantities greater than 10 tonnes per year and compared to exposure levels to arrive at a risk characterisation ratio (RCR). Currently it is estimated that the requirement for DNEL derivation and scenario-specific exposure assessment will apply to approximately 10,000 substances. To meet this requirement in the foreseen time scheme, new methodologies to accomplish this task are needed. A promising approach seems to be the extension of the BE methodology by using (1) DNELs and OELs as reference values and (2) more toxicokinetic modelling tools. Basically, there are three crucial requirements to enable the setting of a BE for a given substance: (1) A reliable biomonitoring method for the substance, which allows the reliable collection of a biological specimen (e.g. blood or urine) and the subsequent analytical quantification with sufficient specificity and sensitivity. (2) A DNEL or a reliable health-based OEL for the substance that may serve as the basis to derive a BE (3) Sufficient basic data on the substance to allow toxicokinetic modelling or extrapolation on the basis of biomarker concentration to obtain a concentration in a biological specimen (blood or urine) that corresponds to the reference value. The essentials of the first requirement have been described in detail elsewhere and will not be discussed here (Boogaard, 2009; Calafat and Needham, 2009). With regard to the second requirement, DNELs can serve as the basis to derive a BE by applying the same methodology as was used to derive BEs from other reference values (Hays and Aylward, 2009; Hays et al., 2007). In addition, OELs as derived for industrial settings, e.g. Permissible Exposure Limits (PEL), Threshold Limit Values (TLV) and corresponding Biological Exposure Indices (BEIs), and Maximum Accepted Concentrations (MAC) could be used in a similar way. As mentioned above, the European Commission has already accepted health-based OELs derived by SCOEL as DNEL values for occupational settings. BEs may be derived from those OELs. Other health-based OELs set by expert committees, under the condition that they are health-based, could also serve as the basis for the derivation of a BE. Although OELs, by definition, are developed to protect workers in occupational situations from adverse health effects, one approach to developing DNELs for the general population would be to build on OELs. In this way, OELs can serve as the point of departure for limit values applicable to the general population using the approaches and numerical factors as provided by ECHA (2008). In its simplest form this means that the exposure scenario is adapted from 40 years of exposure for 5 days per week and 8 h per day (as defined for OEL setting) to life-time exposure (70 years) with 7 days per week and 24 h per day (as defined for the general public) with an additional assessment factor for inter-individual variation which is expected to be larger in the general population than in the relatively homogeneous working population. The third requirement – the availability of suitable toxicokinetic data – may be the barrier for many substances since validated models are sparse. It should be borne in mind, however, that full toxicokinetic models are not essential as long as a minimum dataset is available to make a link between the reference value and the biomarker concentration (Hays and Aylward, 2008; Hays et al., 2008). For substances where toxicokinetic data is lacking, generalised toxicokinetic models may be used to predict the internal dose.

127

Such models are currently being developed using more robust physicochemical data collected under the REACH regulation for every substance (Bartels et al., 2010, 2009; Huizer et al., 2009; submitted for publication; Jongeneelen and Ten Berge, submitted for publication). The derived BE values can be used to provide a screening level interpretation of biomonitoring data in the context of the risk assessments. Substances with biomonitoring levels far below the BEDNEL values indicate relatively low exposures through all exposure routes, and therefore, relatively low RCRs. In contrast, biomonitoring values approaching or exceeding the BEDNEL values will have high RCRs and may be prioritized for further evaluation, delineation of relevant exposure routes, and potential risk reduction measures. This paper presents case studies for several compounds illustrating the derivation of BEDNEL values and their use in the interpretation of available biomonitoring data.

2. Case studies Following are four examples of derivation of proposed DNELbased BE (BEDNEL) values in terms of biomarker concentrations based on a variety of datasets, guidance values, and approaches consistent with the DNEL framework. The examples include two volatile organic compounds (toluene and 1,1,1-trichloroethane), an antibacterial compound used in consumer products (triclosan), and an additive used in plastics manufacturing (BPA). The approaches used here could be applied to other chemicals in order to leverage available data and take advantage of the integrated exposure information provided by human biomonitoring. 2.1. Toluene An IOELV for toluene of 192 mg/m3 time-weighted average (50 ppm; DIR 2006/15/EC) has been established based on avoidance of neurological responses at elevated solvent air concentrations in humans (SCOEL, 1999b) The pharmacokinetic behaviour of toluene has been well studied in humans and laboratory animals, and a physiologically-based toxicokinetic (PBTK) model for humans has been published (Tardif et al., 1995). General population exposure to toluene can occur through exposure to products containing toluene, hydrocarbon solvents, gasoline vapours, and tobacco smoke, as well as due to trace levels of toluene in environmental media such as air or drinking water. As discussed above, the estimation of a DNEL based on an occupational exposure value requires adjustment for the occupational vs. environmental exposure regime and adjustment for the application to the full population, with potentially sensitive members, rather than the healthy working population. The Biomonitoring Equivalent for a DNEL based on the IOELV can be derived as follows: (1) The available PBTK model for toluene was used to predict blood concentration vs. time profiles under a typical occupational exposure scenario (8 h per day, 5 days per week). Using the model, the time-weighted average blood concentration during time periods of occupational exposure can be estimated (Fig. 1). An alternative approach would apply standard occupational-to-environmental adjustment factors of 8/24 h/day and 5/7 days/week to the estimated IOELV associated blood value to estimate an equivalent environmental exposure level, and then estimate the steady-state blood concentration associated with this constant environmental exposure level. The time-weighted average blood concentration from the occupational exposure scenario is exactly equivalent to the steady-state blood concentration resulting from the latter approach.

128

P.J. Boogaard et al. / Regulatory Toxicology and Pharmacology 59 (2011) 125–132 Table 1 Derivation of BEDNEL for toluene and TCE based on IOELVs and two approaches to estimating relevant blood concentrations. Derivation steps

Toluene

IOELV, TWA 192 mg/m3 Average blood concentration at TWA 174 lg/la General population duration adjustment factors 8/24 h/d NAb 5/7 d/wk NAb 40/70 yrs 0.57 Air concentration after adjustment for NA occupational to environmental duration Corresponding steady-state blood 100 lg/l concentration Intra-species uncertainty factor components UFH-PD 100.5 UFH-PK 1d Adjusted BE for general population 32 lg/l Fig. 1. Modelled blood levels of toluene over one week associated with occupational exposure for 5 days (8 h per day) at the IOELV followed by two days with no occupational exposure (solid line). Dashed line indicates time-weighted average blood concentration of toluene over 7 days. The average blood concentration is equivalent to the blood concentration that would result from continuous steadystate exposure at air concentrations equal to the IOELV after adjustment for continuous environmental versus occupational exposure regimen (IOELV  8/24 h per day  5/7 days per week). See text for discussion.

(2) This average blood concentration, in turn, can be adjusted for the occupational lifetime, by dividing the blood concentration by the ratio of standard occupational duration (40 years) by the standard lifetime of 70 years. Thus, the average blood concentration from step 1 is multiplied by this ratio (40/70 years) to obtain the adjusted lifetime environmental blood concentration. (3) Finally, appropriate uncertainty factors for extrapolation from healthy workers to the general population can be applied to the resulting blood concentration. Because the derived value is based on direct measurement of blood concentration, which is directly relevant to brain concentration resulting in neurological effects, pharmacokinetic differences in the general population are directly reflected in the measured blood concentrations. That is, persons with slower metabolism will develop higher blood concentrations for the same external exposure, and the biomonitoring data will reflect these differences. Thus, in deriving a target blood concentration corresponding to the DNEL, only the pharmacodynamic component associated with extrapolation from a healthy working population should be applied, in order to reflect potential differences in intrinsic sensitivity or response to a given blood concentration of toluene. Based on the framework proposed by Renwick (1993) and adopted by the IPCS (1994), this would correspond to an assessment factor of 100.5. These derivation steps are summarized in Table 1. The resulting blood concentration of toluene (32 lg/l) is proposed as the BEDNEL for the general public. 2.1.1. 1,1,1-Trichloroethane (TCE) We used an alternative approach for TCE to illustrate a simpler pharmacokinetic approach that does not require a fully developed and validated PBTK model. If an assumption of steady-state exposure is made, simple steady-state solutions to a generic PBTK model can be used to estimate corresponding blood concentrations (Chiu and White, 2006). These solutions require only three chemical-specific parameters (Michaelis–Menten metabolic rate parameters, Vmax and KM, and the blood:air partition coefficient, PB) which can be estimated based on simple in vitro assays (reviewed in Lipscomb and Poet (2008)). The IOELV for TCE (TWA of 555 mg/ m3) can be extrapolated to a DNEL for environmental exposure conditions on an air concentration basis, and then the correspond-

TCE 555 mg/m3 Not estimated 0.33 0.71 0.57 75 mg/m3 317 lg/lc

100.5 1d 100 lg/l

a Estimated based on PBTK model simulation of occupational exposure schedule (see Fig. 1) averaged over 7 days per week. b Not applicable-averaging for hours per day and days per week conducted on the basis of time-weighted average blood concentrations over a week (see text for discussion and Fig. 1). c Estimated steady-state blood concentration expected at air concentration of 75 lg/m3 based on steady-state solution to the generic PBTK model using chemicalspecific parameters for TCE (see Aylward et al. (2010a), for details). d Biomarker concentration reflects PK variability in the population, so intraspecies pharmacokinetic uncertainty factor component is replaced by the direct measure of blood concentration.

ing blood concentration can be estimated. This procedure is illustrated in Table 1 for TCE. The BE (TCE in blood) associated with the DNEL is 100 lg/l. Using this approach of applying steady-state solutions to PBTK models, target blood concentrations corresponding to risk assessment-based DNELs can be derived for many VOC compounds (details of the approach are presented in Aylward et al. (2010a)). 2.2. Triclosan Triclosan is an antibacterial agent used in a variety of consumer products including toothpaste, mouthwash, hand sanitizers, etc. The European Commission Scientific Committee on Consumer Products (EC SCCP) conducted a recent evaluation of the safety of triclosan (SCCP, 2009), and derivation of BE values based on this evaluation has recently been published (Krishnan et al., 2010a). The committee identified a NOAEL of 12 mg/kg-d from a chronic rat toxicity study to use as the basis of the risk assessment. The committee noted that blood concentrations had been measured at multiple time points in the rats in the key study and estimated plasma triclosan concentrations based on those measurements. An assessment factor of 100 was applied to the NOAEL (corresponding to inter- and intra-species assessment factors of 10 each). The resulting value, 0.12 mg/kg-d, can be regarded as a DNEL for the general public under the REACH framework. A corresponding biomarker-based BEDNEL is derived by applying the same assessment factors to the rat plasma concentrations at the NOAEL dosing level (Table 2). This would yield a BEDNEL for triclosan in plasma of 0.7 mg/ml. Use of actual human biomonitoring data for triclosan coupled with the BEDNEL for assessing potential risk has the advantage of allowing a safety assessment that accounts for use of multiple consumer products containing triclosan under real-world conditions, a challenge that the EC SCCP committee found difficult to address on the basis of estimated external exposure levels (SCCP, 2009). 2.3. Bisphenol A Bisphenol A (BPA) is used in the production of polycarbonate plastics and other resins, and low-level exposure to BPA is

P.J. Boogaard et al. / Regulatory Toxicology and Pharmacology 59 (2011) 125–132 Table 2 Summary of the derivation of BEDNEL for triclosan based on the SCCP (2009) risk assessment. See Krishnan et al. (2010b) for further details on the derivation approach and details. BE derivation step

EC (2009) risk assessment

Species, endpoint

Rats, haematological endpoint alterations 12 21,800

POD (NOAEL), external dose, mg/kg-d POD (NOAEL), plasma concentration, lg/l UF, interspecies Interspecies-adjusted plasma concentration, lg/l UF, intra-species BEDNEL, plasma concentration, lg/l

3.16 6900 10 700

widespread. The European Food Safety Authority conducted a recent risk assessment of BPA and identified a NOAEL of 5 mg/kg-d based on a comprehensive three-generation reproduction and developmental toxicity assay and a total assessment factor of 100 (EFSA, 2006), resulting in a tolerable daily intake (TDI) of 50 lg/kgd. This value can be regarded as a DNEL for the general population for BPA. Because humans excrete essentially 100% of administered BPA in urine (either as the free parent or as a glucuronide; (Volkel et al., 2005, 2008)), urinary levels of total BPA can be interpreted as a measure of daily intake. Average urinary concentrations of total BPA corresponding to chronic intake at the TDI can be estimated to derive a BEDNEL for BPA based on the EFSA TDI (Table 3, adapted from Krishnan et al. (2010b)). The BEDNEL for BPA across all age groups is 2000 lg/l or 2600 lg/g creatinine. 2.4. Evaluation of biomonitoring data using the BEDNEL values The compounds evaluated here are relatively short-lived in the body with a half-life of a few hours (toluene, TCE, BPA) to a day (triclosan). These estimated BEDNEL values derived correspond to average concentrations associated with theoretically ongoing, steady exposure at the corresponding DNELs. Thus, biomarker concentrations in a spot sample of blood or urine for an individual must be interpreted very cautiously. A single measurement in an individual only provides an indication of recent rather than chronic exposure levels, and because the compound is eliminated relatively rapidly, concentrations in urine will vary substantially over the course of hours or days since exposure. However, population-based sampling can provide general information as to whether exposures are likely reaching or exceeding the DNEL on a frequent or consis-

tent basis, although extremes in the population data must be interpreted cautiously, since they may represent fluctuations rather than typical long-term levels in any individual. Alternatively, such extremes might represent more highly-exposed populations; complementary exposure science and further, targeted sampling (for example, of individuals who volunteer to use triclosan-containing products under controlled conditions) can provide additional information on the range of likely exposures in the population and help to provide additional context to the distributions from general population sampling. Table 4 provides an illustrative example of the evaluation of human biomonitoring data for the four case study compounds using the derived BEDNEL values based on available biomonitoring data from several sources. For each of the case study substances, the actual exposures, as measured using biomonitoring, are at least an order of magnitude below the BEDNEL. As can be seen, the ratio between measured exposure to TCE and the DNEL is, with a value less than 0.005, well below the required ratio of less than 1. Even for BPA and triclosan which have both been the focus of increased scrutiny from a regulatory standpoint, the ratio between actual exposure and the DNEL is far below 1. Even considering the toxicokinetic uncertainties in the BEDNEL, these data indicate that the actual human exposure for each of the substances as reflected in these biomonitoring data sets is most likely well below the DNEL. Consequently, under REACH no risk reduction measures would be required. The comparisons and conclusions from the screening evaluation are only valid for the populations represented by the biomonitoring data sets and are only as reliable as the biomonitoring data sets. If specially-exposed populations exist or are thought to exist, additional targeted biomonitoring data may need to be developed. For example, the data on triclosan levels in the general population in Australia are based on pooled samples, which do not provide information on individual variation in measured plasma concentrations. In addition, for each of the case study compounds presented in Table 4, the available biomonitoring data comes from surveys of populations outside the European Union. An analysis would need to be done for each of the substances to assess the validity of the available biomonitoring data for the potential product applications being used or proposed in the European Union (beyond the scope of this effort).

3. Discussion Under the new EU chemicals regulation (REACH) that is currently implemented, the new concept of DNEL was introduced.

Table 3 Assumptions for bodyweight, average 24-h urinary volume and creatinine excretion, and estimates of average volume-based and creatinine-adjusted urinary concentration of total BPA (free and conjugated species) consistent with exposure at TDI (50 lg/kg-d) for different age groups.

a

Age group

Body weight, kga

Daily intake and excretion of BPA at the TDI (lg/d)

Average 24 h urinary volume, lb (creatinine excretion, gc)

BPA average urinary concentration, lg/l (lg/g creatinine) at steady-state exposure at the TDI

Children, 6–11

32

1600

Adolescents, 11–16

57

2850

Men, >16

70

3500

Women, >16

55

2750

0.66 (0.50) 1.65 (1.20) 1.70 (1.50) 1.60 (1.20) Average, lg/l Average, lg/g cr

2424.24 (3200.00) 1727.27 (2375.00) 2058.82 (2333.33) 1718.75 (2291.67) 2000 (2600)

Estimated from Tables 8–1 of USEPA (2008). Urinary volumes for children from Remer et al. (2006). Volumes for adults from Perucca et al. (2007). Adolescents were assumed to have urinary volumes similar to average values for adults. c Creatinine excretion for children and adolescents estimated from Remer et al. (2002); average creatinine excretion for boys and girls under age 13, 17 mg/kg BW per day; average creatinine excretion for adolescents, 22 mg/kg BW per day. Creatinine excretion for adults estimated based on equations from Mage et al. (2004), average US height, and specified bodyweights. b

129

130

P.J. Boogaard et al. / Regulatory Toxicology and Pharmacology 59 (2011) 125–132

Table 4 Comparison of biomonitoring data to BEDNEL values for four case study compounds to illustrate use of the BE concept in a screening-level assessment of biomonitoring data. Compound Toluene TCE

Biomonitoring data description

Matrix, units

Measured concentrations

BEDNEL

Comment

General population, US (NHANES 2003–2004)

a

Blood, lg/l

Median: 0.096 95th%ile: 0.68

32 lg/l

General population, US (NHANES 2003–2004)

a

Blood, lg/l

100 lg/l

Measured concentrations >50-fold below BEDNEL No. detected levels at LOD >2000-fold below BEDNEL Measured concentrations in pooled samples >30-fold below BEDNEL Concentrations >20-fold below BEDNEL

Triclosan

Pooled samples, Australian general population

Blood serum, lg/l

Not detected in 1345 samples at LOD of 0.048 4–19 lg/l in various pools

Plasma, mg/l

30 lg/l

BPA

Samples from volunteers following 3 weeks of using multiple triclosan-containing productsc General population, US (NHANES 2003–2004)a

Urine, lg/l

Median: 2.8 95th%ile: 16

b

700 lg/l

2000 lg/l

Measured concentrations in urine >100-fold below BEDNEL

LOD Limit of detection. a CDC (2008). b Allmyr et al. (2008). c SCCP (2009).

Despite the fact that under some circumstances an existing OEL can be used as a DNEL and notwithstanding the fact that a DNEL can have the same units as an OEL, DNELs do not have the same status as OELs nor do they serve the same purpose. DNELs are merely meant to be compared to exposure levels as determined in exposure scenarios that have been developed for specific substances and uses to derive a risk characterisation ratio (RCR). If the exposure level is higher than the DNEL, that is if the RCR >1, risk reduction measures should be applied. Developing and validating exposure scenarios for a specific chemical via all potential sources and routes is a complex, cumbersome, and sometimes, extremely uncertain exercise. The REACH regulations thus set a high burden of proof for these exposure assessments. For many compounds, it may be difficult to assess all the potential exposures reliably. Human biomonitoring, which integrates all exposures regardless the route of exposure, may help to overcome these challenges and increase the confidence associated with complex exposure assessments. This applies in particular for non-occupational exposure scenarios where population-based biomonitoring surveys may be extremely helpful. Biomonitoring data can be interpreted quantitatively in the context of existing health risk assessments for many compounds using BEs. If the results of an exposure assessment using human biomonitoring indicate that the levels measured are below the BE it would indicate that the combined exposure via all potential exposure routes poses negligible human health risk. Exposure guidance values, including DNELs, can be the basis for deriving BEs as explained in this paper. Acceptance of the use of BE by regulatory authorities will not only make repetitious assessments for different exposure scenarios superfluous, but may also provide a more accurate estimate of exposure and hence lead to improved product stewardship and risk management. It must be noted, however, that the BE values derived based on DNELs are only translations of these DNELs; limitations or uncertainties in the underlying DNELs and the risk assessments on which they are based are not remedied by the translation into BE values. Limitations associated with biomonitoring data must also be acknowledged. For example, biomonitoring data cannot inform evaluations of the source or pathway of exposure, nor does biomonitoring data provide information on how long the chemical has been present in the body. In addition, it should be emphasised that interpretation on an individual level is not well possible for substances with short half lives or where exposure is highly intermittent. The latter may be the case when exposure is related to intake of certain food stuffs or associated with specific activities. In such cases, the data should rather be interpreted on a group level. In contrast, for substances with a sufficiently long half life to reach

steady state levels in the body under the conditions of exposure, interpretation of the data on a personal level may be more feasible (Boogaard, 2009; Boogaard and Money, 2008). The examples here are presented in terms of the REACH framework and legislation, but other regulatory applications are clearly possible. In the agricultural pesticide arena, regulations aim to minimise worker and bystander exposures during use and to control residues remaining on foods. Biomonitoring can provide information to evaluate whether regulations are succeeding in the intended goals. For example, the herbicide 2,4-dichlorophenoxyacetic acid (2,4-D) is widely used in North America and Europe for weed control on crops, and has been the subject of numerous biomonitoring studies for applicators, farm family members, and members of the general public including children (reviewed in Aylward et al. (2010b)). Urinary BE values corresponding to the US reference dose (RfD) and occupational exposure targets have been derived (Aylward and Hays, 2008). Comparison of the various targeted and general population biomonitoring data for 2,4-D to these BE values suggest that ‘‘current usage patterns and risk management efforts by industry and government are likely keeping average exposure to 2,4-D for the general population and in farm family members, and likely other persons potentially exposed due to proximity during usage of this herbicide, to levels well below current non-cancer reference values established both by the US EPA’s Office of Pesticide Programs and by Canada’s PMRA” (Aylward et al., 2010b). The conclusions drawn from the 2,4-D example were strong, partly because of the wealth and breadth of biomonitoring data available for 2,4-D in North America. In the case of 2,4-D, biomonitoring studies were available to help verify the exposure scenarios associated with the product use, in people residing near the product uses, and exposures in the general population resulting from product uses and/or exposures from the food chain from residuals in foods. Large population based surveys can be extremely valuable to help assess the extent and pervasiveness of exposures among the general population. There are always concerns that more highly exposed subpopulations may not be captured in these general population based surveys (which is also true of a conventional exposure assessment). Having additional studies to help elucidate the extent of exposures in potentially highly-exposed populations (for instance associated with product uses such as the studies available for triclosan) helps to answer this important question. The degree to which these ancillary studies are needed (or are helpful) is partly dependent upon the types of products a chemical is in, the likely frequency of exposures and the half-life of the compound. The types of supplementary studies are most helpful for those compounds in which exposures are likely to be dominated from product use (as opposed to exposures being dominated by

P.J. Boogaard et al. / Regulatory Toxicology and Pharmacology 59 (2011) 125–132

ambient exposures), use of the product is infrequent, and the halflife in the body is very short (less than the frequency of exposure). To this degree, assessments under the REACH program (and any other regulatory program) will need to balance these issues to gauge whether enough biomonitoring data (and the right types of data) are available to make confident conclusions. In the USA, the chemicals regulation framework under the Toxic Substances Control Act (TSCA) is currently being considered for reform and updating. Biomonitoring has been discussed for inclusion as an important component of the updated legislation and regulatory framework. However, the simple collection of biomonitoring data will not inform the safety of the chemicals in commerce unless quantitative criteria based on health risk assessment are available for interpretation of these data. In the absence of such criteria, all human biomonitoring data can provide is a qualitative indication of exposure which is entirely dependent upon analytical detection limits (Boogaard et al., 2005; Boogaard and Money, 2008). If health risk-based criteria such as BEs can be derived, biomonitoring data become much more valuable for evaluation of the safety of chemical usage patterns, as in the 2,4-D example presented above. Whilst the possible value of biomonitoring data in evaluating chemical safety under REACH or in an update of the TSCA framework is apparent, there are significant challenges associated with conducting large population-based biomonitoring studies, particularly as the list of chemicals of interest grows. One possible approach to refining and focusing biomonitoring efforts might be to use pooled population-based samples collected from blood banks or other resources in order to screen for the presence of chemicals or groups of chemicals at levels approaching a level of interest based on BE assessments. If chemicals are detected in pooled serum samples at levels that are a significant fraction of the BE values, this might suggest the need for more detailed, population-based sampling efforts or targeted sampling of individuals based on likely exposure patterns. This approach has been used in Australia to characterise population average biomarker concentrations of dioxins, polybrominated diphenyl ethers, perfluorinated compounds, and triclosan (Allmyr et al., 2008; Harden et al., 2007; Karrman et al., 2006; Toms et al., 2009a,b). Biomonitoring is likely most useful for chemicals with wide exposure and use patterns. Human biomonitoring data can reduce uncertainties attendant on estimates of theoretical exposures that could occur from the presence of such chemicals in products or the environment. However, properly designed and conducted, ethically approved targeted studies could allow assessment of relevant internal exposures under actual exposure scenarios for specific chemicals or products. Targeted studies could also inform the exposures of people with unusual exposure conditions, such as residence in an area with a source, or through use of a specific product type. Again, interpretation of such data in a health risk context requires understanding of toxicokinetics or of the relevant internal concentrations of interest for a chemical. Simple detection does not inform safety assessment. Toxicokinetic data, however, tend to be available primarily for widely used, high volume substances and for substances of particular concern. As indicated in the case studies, sophisticated PBTK models are not always necessary to make a reliable prediction of a BEDNEL. Currently, a number of arithmetic toxicokinetic models and generalised PBTK models are being developed that require a minimal amount of data on a substance to predict biomonitoring values that can be used to calculate BEs (Bartels et al., submitted for publication; Bartels et al., 2010, 2009; Huizer et al., 2009; Jongeneelen et al., submitted for publication). The recently developed generic PBTK model IndusChemFate is available as freeware (http://www.cefic-lri.org/lri-toolbox/induschemfate). Depending on the available data, more or less sophisticated models can be

131

used to predict the BE corresponding to the DNEL. The initial validations indicate that predictions made using this approach are within an order of magnitude of the actual value. This is promising as it would allow application of the calculated BDDNEL values in a tiered approach to determine whether or not there is a reason for concern. If the actual exposures as determined through biomonitoring surveys are one or more orders of magnitude less than the derived BEDNEL it is unlikely that there is a realistic health risk requiring risk reduction measures. On the other hand, if the biomonitoring levels would be close to or greater than the BEDNEL, a more accurate risk assessment or risk reduction measures would be triggered. In this approach the vast task of risk characterisation of about 10,000 substances over the next few years as required under REACH may be brought back to realistic proportions. In conclusion, human biomonitoring data provides significant advantages over an external dose-based exposure assessment, and is being increasingly widely used to detect and assess chemical exposure in the environment and from consumer products (Boogaard, 2009; Schulz et al., 2007; Sexton et al., 2004). In parallel, there has been an increasing focus on understanding and applying mode of action, internal dose assessment, and chemical-specific assessment factors in risk assessment (Boobis et al., 2006, 2008). Coupling human biomonitoring data with a dose–response assessment based on internal and/or absorbed dose via use of a Biomonitoring Equivalent (BE) provides an extremely powerful and scientifically robust approach to conducting a risk assessment (Hays and Aylward, 2009). Conducting chemical exposure and risk assessments using both of these advances is the next logical step in this progression and provides a pragmatic tool in the context of risk management as required by REACH and other imminent chemical regulations. 4. Funding sources and competing financial interests statements The authors did not receive specific funding for preparation of the manuscript and had complete freedom to design, implement and report the analyses and views presented herein. Examples given in this paper do not necessarily represent the point of view of any consortium or SIEF. References Allmyr, M. et al., 2008. The influence of age and gender on triclosan concentrations in Australian human blood serum. Sci. Total Environ. 393, 162–167. Aylward, L.L., Hays, S.M., 2008. Biomonitoring Equivalents (BE) dossier for 2,4dichlorophenoxyacetic acid (2,4-D) (CAS No. 94–75-7). Regul. Toxicol. Pharmacol. 51, S37–S48. Aylward, L.L. et al., 2008a. Biomonitoring Equivalents (BE) dossier for toluene (CAS No. 108-88-3). Regul. Toxicol. Pharmacol. 51, S27–S36. Aylward, L.L. et al., 2008b. Biomonitoring Equivalents (BE) dossier for trihalomethanes. Regul. Toxicol. Pharmacol. 51, S68–S77. Aylward, L.L. et al., 2008c. Derivation of Biomonitoring Equivalent (BE) values for 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) and related compounds: a screening tool for interpretation of biomonitoring data in a risk assessment context. J Toxicol Environ Health A 71, 1499–1508. Aylward, L.L. et al., 2009a. Derivation of Biomonitoring Equivalents for di(2ethylhexyl)phthalate (CAS No. 117-81-7). Regul. Toxicol. Pharmacol. 55, 249– 258. Aylward, L.L. et al., 2009b. Derivation of Biomonitoring Equivalents for di-n-butyl phthalate (DBP), benzylbutyl phthalate (BzBP), and diethyl phthalate (DEP). Regul. Toxicol. Pharmacol. 55, 259–267. Aylward, L.L. et al., 2010a. Chemical-specific screening criteria for interpretation of biomonitoring data for volatile organic compounds (VOCs) – application of steady-state PBPK model solutions. Regul. Toxicol. Pharmacol. 58, 33–44. Aylward, L.L. et al., 2010b. Biomonitoring data for 2,4-dichlorophenoxyacetic acid in the United States and Canada: interpretation in a public health risk assessment context using Biomonitoring Equivalents. Environ. Health Perspect. 118, 177– 181. Bartels, M.J., Loizou, G., Price, P., Spendiff, M., Arnold, S., Cocker, J., Ball, N., 2009. Development of a tiered set of modelling tools for derivation of biomonitoring guidance values. Toxicol. Lett. 189 (S1), 154.

132

P.J. Boogaard et al. / Regulatory Toxicology and Pharmacology 59 (2011) 125–132

Bartels, M.J., Loizou, G., Price, P., Rick, D., Spendiff, M., Arnold, S., Cocker, J., Ball, N., 2010. Development of PK- and PBPK-based modelling tools for derivation of biomonitoring guidance values. Toxicol. Lett. 196 (S), 42. Bartels, M., Loizou, G., Rick, D., Price, P., Spendiff, M., Arnold, S., Cocker, J., Ball, N., submitted for publication. Development of PK- and PBPK-based modeling tools for derivation of biomonitoring guidance values. Boobis, A.R. et al., 2006. IPCS framework for analyzing the relevance of a cancer mode of action for humans. Crit. Rev. Toxicol. 36, 781–792. Boobis, A.R. et al., 2008. IPCS framework for analyzing the relevance of a noncancer mode of action for humans. Crit. Rev. Toxicol. 38, 87–96. Boogaard, P.J., 2009. Biomonitoring of the Workplace and Environment, third ed.. In: Ballentyne, B., Marrs, T., Syversen, T. (Eds.), General and Applied Toxicology, vol. 4 Wiley Chichester, UK, pp. 2559–2589. Boogaard, P.J., Money, C.D., 2008. A proposed framework for the interpretation of biomonitoring data. Environ Health 7 (Suppl. 1), S12. Boogaard, P.J., Farmer, P.B., Knudsen, L.E., Onyon, L., Robison, S.H., Schoeters, G., Stropp, G.D., Wilks, M.F., Will, W., Holt, M., 2005. Guidance for the interpretation of biomonitoring data. ECETOC Document 44. ECETOC, Brussels. Calafat, A.M., Needham, L.L., 2009. What additional factors beyond state-of-the-art analytical methods are needed for optimal generation and interpretation of biomonitoring data? Environ. Health Perspect. 117, 1481–1485. CDC, 2008. National Report on Human Exposure to Environmental Chemicals (4th Report 2003–2004). Available from: . Chiu, W.A., White, P., 2006. Steady-state solutions to PBPK models and their applications to risk assessment I: route-to-route extrapolation of volatile chemicals. Risk Anal. 26, 769–780. EC, Council Regulation (EEC) No. 793/93 of 23 March 1993 on the evaluation and controls of existing substances, vol. 793/93, 1993. EC, Commission decision of 12 July 1995 setting up a Scientific Committee for Occupational Exposure Limits to Chemical Agents (95/320/EC), vol. 95/320/EC, 1995. EC, Regulation (EC) No. 1907/2006 of the European Parliament and of the Council of 18 December 2006 concerning the Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH), establishing a European Chemicals Agency, amending Directive 1999/45/EC and repairing Council Regulation (EEC) No. 793/93 and Commission Regulation (EC) No. 1488/94 as well as Council Directive 76/769/EEC and Commission Directives 91/155/EEC, 93/67/EEC, 93/ 105/EC and 2000/21/EC, vol. 1907/2006, 2006. EC, Communication from the Commission on the reviews of Annexes I, IV and V to regulation (EC) No. 1907/2006 of the European Parliament and of the Council concerning the Registration, Evaluation and Authorisation and Restriction of Chemicals (REACH) – Text with EEA relevance [SEC(2009)447], vol. C(2009)2482. Brussels, Commision of the European Communities, 8 April 2009, 2009. ECETOC, 2010. Guidance on Assessment Factors to Derive DNELs. ECETOC Technical Report 110. ECETOC, Brussels. ECHA, 2008. Guidance on information requirements and chemical safety assessment. Chapter R.8: characterisation of dose [concentration]-response for human health. European Chemicals Agency (ECHA), Helsinki. Available from: . Gade, A.L. et al., 2008. Testing REACH draft technical guidance notes for conducting chemical safety assessments – the experience of a downstream user of a preparation. Regul. Toxicol. Pharmacol. 51, 168–180. Gebel, T. et al., 2009. About hazard and risk assessment: regulatory approaches in assessing safety in the European Union chemicals legislation. Reprod. Toxicol. 28, 188–195. GerES, Health and Environmental Hygiene – German Environmental Survey (GerES). Available from: . 2008. Harden, F.A. et al., 2007. Evaluation of age, gender and regional concentration differences for dioxin-like chemicals in the Australian population. Chemosphere 67, S318–S324. Hays, S.M., Aylward, L.L., 2008. Biomonitoring Equivalents (BE) dossier for acrylamide (AA) (CAS No. 79-06-1). Regul. Toxicol. Pharmacol. 51, S57–S67. Hays, S.M., Aylward, L.L., 2009. Using Biomonitoring Equivalents to interpret human biomonitoring data in a public health risk context. J. Appl. Toxicol. 29, 275–288. Hays, S.M. et al., 2007. Biomonitoring equivalents: a screening approach for interpreting biomonitoring results from a public health risk perspective. Regul. Toxicol. Pharmacol. 47, 96–109. Hays, S.M. et al., 2008. Biomonitoring Equivalents (BE) dossier for cadmium (Cd) (CAS No. 7440-43-9). Regul. Toxicol. Pharmacol. 51, S49–S56.

Huizer, D., Jongeneelen, F., Ten Berge, W., 2009. Use of a Generalized Toxicokinetic Model for the Prediction of Internal Dose for Occupational Exposure Assessment: The Example of Pyrene. Sixth International Conference on Innovations in Exposure Assessment (X2009), Cambridge, MA, 2009, pp. 9. Available from: . IPCS, 1994. Assessing human health risks of chemicals: derivation of guidance values for health-based exposure limits. Environmental Health Criteria, No. 170. World Health Organisation, Geneva. Jongeneelen, F., Ten Berge, W., submitted for publication. A generic, cross-chemical predictive PBTK-model with multiple entry routes running as application in MS-Excel; Design of the model and simulation of experimental observations. J. Exp Sc. Environ Epidemiol. Karrman, A. et al., 2006. Levels of 12 perfluorinated chemicals in pooled Australian serum, collected 2002–2003, in relation to age, gender, and region. Environ. Sci. Technol. 40, 3742–3748. Kolossa-Gehring, M. et al., 2007. German environmental survey for children (GerES IV) – first results. Int. J. Hyg. Environ. Health 210, 535–540. Krishnan, K. et al., 2010a. Biomonitoring equivalents for triclosan. Regul. Toxicol. Pharmacol. 58, 10–17. Krishnan, K. et al., 2010b. Biomonitoring equivalents for bisphenol A (BPA). Regul. Toxicol. Pharmacol. 58, 18–24. LaKind, J.S. et al., 2008. Guidelines for the communication of Biomonitoring Equivalents: report from the Biomonitoring Equivalents Expert Workshop. Regul. Toxicol. Pharmacol. 51, S16–S26. Lipscomb, J.C., Poet, T.S., 2008. In vitro measurements of metabolism for application in pharmacokinetic modeling. Pharmacol. Ther. 118, 82–103. Mage, D.T. et al., 2004. Estimating pesticide dose from urinary pesticide concentration data by creatinine correction in the Third National Health and Nutrition Examination Survey (NHANES-III). J. Expo. Anal. Environ. Epidemiol. 14, 457–465. Paustenbach, D., Galbraith, D., 2006. Biomonitoring: is body burden relevant to public health? Regul. Toxicol. Pharmacol. 44, 249–261. Perucca, J. et al., 2007. Sex difference in urine concentration across differing ages, sodium intake, and level of kidney disease. Am. J. Physiol. Regul. Integr. Comp. Physiol. 292, R700–R705. Remer, T., Fonteyn, N., Alexy, U., Berkemeyer, S., 2006. Longitudinal examination of 24-h urinary iodine excretion in school children as a sensitive, hydration status-independent research tool for studying iodine status. Am. J. Clin. Nutr. 83 (3), 639–646. Remer, T., Neubert, A., Maser-Gluth, C., 2002. Anthropometry-based reference values for 24-h urinary creatinine excretion during growth and their use in endocrine and nutritional research. Am. J. Clin. Nutr. 75 (3), 561–569. Renwick, A.G., 1993. Data-derived safety factors for the evaluation of food additives and environmental contaminants. Food Addit. Contam. 10, 275–305. SCCP, 2009. Opinion on triclosan. European Commission, Directorate General for Health and Consumer Protection. Brussels. Schäfer, C., Dorn, S.B., Jakupoglu, C., Mostert, V., 2009. OEL vs. DNEL or expert judgement vs. default factors – reference values under REACH exemplified by styrene. Toxicol. Lett. 189S, S270–S271. Schulz, C. et al., 2007. The german human biomonitoring commission. Int. J. Hyg. Environ. Health 210, 373–382. SCOEL, 1999. Methodology for the derivation of occupational exposure limits. Key documentation. EUR 19253, Luxembourg. SCOEL, 1999. Recommendation from Scientific Committee on Occupational Exposure Limits for Toluene. In: 18216, E., (Ed.), vol. SCOEL/SUM/18-final. Sexton, K., Needham, L.L., Pirkle, J.L., 2004. Human biomonitoring for environmental chemicals. Am. Sci. 92, 38–45. Tardif, R. et al., 1995. Physiologically-based pharmacokinetic modeling of a mixture of toluene and xylene in humans. Risk Anal. 15, 335–342. Toms, L.M. et al., 2009a. Polyfluoroalkyl chemicals in pooled blood serum from infants, children, and adults in Australia. Environ. Sci. Technol. 43, 4194– 4199. Toms, L.M. et al., 2009b. Serum polybrominated diphenyl ether (PBDE) levels are higher in children (2–5 years of age) than in infants and adults. Environ. Health Perspect. 117, 1461–1465. Volkel, W. et al., 2005. Quantitation of bisphenol A and bisphenol A glucuronide in biological samples by high performance liquid chromatography–tandem mass spectrometry. Drug Metab. Dispos. 33, 1748–1757. Volkel, W. et al., 2008. Determination of free and total bisphenol A in human urine to assess daily uptake as a basis for a valid risk assessment. Toxicol. Lett. 179, 155–162.