Making Sense of Human Biomonitoring Data: Findings and Recommendations of a Workshop

www.nature.com/jes

Making sense of human biomonitoring data: Findings and recommendations of a workshop

TINA BAHADORIa, RICHARD D. PHILLIPSb, CHRIS D. MONEYb, JAMES J. QUACKENBOSSc, HARVEY J. CLEWELLd, JAMES S. BUSe, STEVEN H. ROBISONf, COLIN J. HUMPHRISg, AMI A. PAREKHh, KIMBERLY OSBORNh AND REBECCA M. KAUFFMANh aAmerican Chemistry Council, 1300 Wilson Blvd, Arlington, Virginia 22209, USA bExxonMobil Petroleum and Chemical, Hermeslaan 2, B-1831 Machelen, Belgium cUS EPA National Exposure Research Laboratory, Exposure and Dose Research Branch, PO Box 93478, Las Vegas, Nevada 89193-3478, USA dCIIT Centers for Health Research, 6 Davis Drive, PO Box12137, Research Triangle Park, North Carolina 27709, USA eDow Chemical Company, 1803 Building, Midland, Michigan 48674, USA fProctor & Gamble, Cincannati, Ohio, USA gEuropean Chemical Council (now retired), Avenue E. van Nieuwenhuyse 4, 1160 Brussels, Belgium hICF International, 9300 Lee Highway, Fairfax, Virginia 22031, USA

The ability to measure chemicals in humans (often termed biomonitoring) is far outpacing the ability to interpret reliably these data for public health purposes, creating a major knowledge gap. Until this gap is ﬁlled, the great promise of routinely using biomonitoring data to support decisions to protect public healthcannot be realized. Researchis needed to link biomonitoring data quantitatively to thepotential for adverse healthrisks, either thr ough association with health outcomes or using information on the concentration and duration of exposure, which can then be linked to health guidelines. Developing such linkages in the risk assessment paradigm is one of the primary goals of the International Council of Chemical Associations’ (ICCA) Long-Range ResearchInitiative (LRI) program in thearea of biomonitoring. Therefore, ICCA sponsored a workshop to fac ilitate development of a coordinated agenda for researchto enable an improved interpretation of human biomonitoring data. Discussions addressed threemain topics: (1) exploration of the link between exposure, dose, and human biomonitoring data, (2) the use of computational tools to interpret biomonitoring data, and (3) the relevance of human biomonitoring data to the design of toxicological studies. Several overarching themes emerged from the workshop: (a) Interpretation and use of biomonitoring data should involve collaboration across all sectors (i.e., industry, government, and academia) and countries. (b) Biomonitoring is not a stand-alone tool, and it should be linked to exposure and toxicological dose information. (c) Effective communication is critical, because when uncertainty about the actual risks is high, the perceived risks grow in the absence of communication. (d) The scope of future biomonitoring activities encompasses a variety of researchapproaches F from advancing the science to ﬁll data gaps to advancing the accessibility of the current knowledge to enable better information sharing. Journal of Exposure Science and Environmental Epidemiology (2007) 17, 308–313; doi:10.1038/sj.jes.7500581; published online 9 May 2007

Keywords: biomonitoring, American Chemistry Council, International Council of Chemical Associations, workshop, chemicals, research.

1. Abbreviations: ACC, American Chemistry Council; ADME, absorption, distribution, metabolism, and elimination; AUC, area- under-the-curve; CDC, Centers for Disease Control and Prevention; Background Cefic, the European chemical industry council; ECETOC, European Centre for Ecotoxicology and Toxicology of Chemicals; EPA, US The ability to measure chemicals in humans (often termed Environmental Protection Agency; EU, European Union; ICCA, biomonitoring) is far outpacing the ability to reliably International Council of Chemical Associations; ILSI/HESI, Interna- interpret these data for public health purposes, creating a tional Life Science Institute Healthand Sciences Institute; JCIA, Japan major knowledge gap. Until this gap is filled, the great Chemical Industries Association; LRI, Long-Range Research Initiative; MOE, margin of exposure; NGO, non-governmental organization; promise of routinely using biomonitoring data to support NOAEL, no observable adverse effects limit; NRC, National Research decisions to protect public health cannot be realized. Council; OECD, Organisation for Economic Co-operation and Develop- Researchis needed to link biomonitoring data quantitatively ment; PBPK, physiologically based pharmacokinetic; QSAR, quantita- to the potential for adverse health risks, either through tive structure–activity relationships; QSPR, quantitative structure– association withhealthoutcomes or using information on the property relationship; RfD, reference dose concentration and duration of exposure, which can then be 2. Address all correspondence to: Dr. T. Bahadori, American Chemistry linked to healthguidelines. Developing suchlinkages in Council, Long-Range ResearchInitiative, 1300 Wilson Boulevard, the risk assessment paradigm is one of the primary goals of Arlington, VA 22209, USA. the International Council of Chemical Associations’ (ICCA) Tel.: þ 1 703 741 5214. Fax: þ 1 703 741 6214. E-mail: [email protected] Long-Range ResearchInitiative (LRI) program in thearea Received 16 April 2007; accepted 16 April 2007; published online 9 May 2007 of biomonitoring. Biomonitoring workshop findings and recommendations Bahadori et al.

A workshop was held on July 26–27, 2006, in Minnea- data in toxicology studies that are better linked to polis, MN, USA to facilitate development of a coordinated biomonitoring data sets) to advancing the accessibility agenda for researchactivities to enable improved interpreta- and utilization of current knowledge to enable better tion of human biomonitoring data. Regulatory and public information sharing and thus better coordination on perception issues and their influence on the direction and research(e.g., by comp iling libraries of information on application of biomonitoring researchwere featured. The current analytical methods or standards). workshop was sponsored by the ICCA-LRI, which is composed of the LRIs of the American Chemistry Council (ACC), Cefic (the European chemical industry council), and Recent Efforts to Improve the Understanding of the Japanese Chemical Industry Council (JCIA). The Biomonitoring Data workshop was attended by approximately 100 representatives from industry, academia, non-government organizations The collection of human biomonitoring data is rapidly (NGOs), and various government agencies from the United growing bothnationally and internationally. Thisgrowth States, Canada, Europe, and Japan. has been facilitated by access to relatively inexpensive This report highlights research recommendations derived analytical technology that is able to readily detect from the workshop breakout sessions that will constitute a ever-greater numbers and lower concentrations of chemicals basis for informing the scope and content of future human in human biological samples. However, the rapid growth in biomonitoring researchto be sponsored at theICCA-LRI the pace of analytical biomonitoring investigations has level. In addition to these recommendations, the workshop not been matched with parallel efforts to establish the aimed at fostering partnerships to develop such research implications of human biomonitoring data to actual human plans, to review the capacity and skills available to advance health risks. the topic, and identify how improved networking across Several organizations have sought to enhance the applica- stakeholders could lead to better distribution of available tion of biomonitoring data in assessments. In the United resources. Several overarching themes emerged from the States, the National Research Council’s (NRC) committee workshop: on Human Biomonitoring for Environmental Toxicants published a report in July 2006 titled Human Biomonitoring Multiple national and international programs are currently for Environmental Chemicals (NRC, 2006), which provides a working to improve the understanding of biomonitoring; reference guide for moving the field of biomonitoring therefore, interpretation and use of biomonitoring data forward, including designing and conducting studies and should involve collaboration across all sectors (i.e., researchto enable interpretation, and reporting biomonitor- industry, government, and academia) and countries so ing results to the individuals measured and the public. In that efforts are not duplicated and resources can be Europe, the European Centre for Ecotoxicology and leveraged when there are common interests. Toxicology of Chemicals (ECETOC) established a dedicated Biomonitoring is not a stand-alone tool; it should be task force on biomonitoring in 2005, withbroad representa- linked to human exposure and animal toxicological tion from academia, institutes, NGOs, and industry. They dose information. To realize the full potential of bio- developed Guidance for the Interpretation of Biomonitoring monitoring and its ability to inform individual and public Data (ECETOC, 2005), which includes a potential frame- health decisions, the data need to be placed in context. work for the interpretation of human biomonitoring data. In It is not enoughto know if chemicalsare present in addition, the International Life Science Institute’s Health the body; it is important to be able to assess relationships and Sciences Institute (ILSI/HESI) biomonitoring committee to potential adverse health impacts and to identify aims to identify and refine effective scientific uses of potential source(s) and pathway(s) leading to exposure biomonitoring tools and/or biomonitoring data to charac- outcomes. terize exposure to chemicals. Many data gaps in biomonitoring information still exist; therefore effective communication is critical, because when uncertainty about the actual risks is high, perceived risks Placing Biomonitoring Data within the Risk Assessment grow in the absence of effective communication. Public Paradigm engagement is essential not only to inform the public but also, and perhaps even more importantly, to assure The capability of measuring chemicals in the body often application of constrained scientific resources to effective exceeds any ability to evaluate meaningfully the source(s) and health protective programs and policies. pathway(s) for exposure of a chemical, as well as how and The scope of future biomonitoring activities encompasses a whether a chemical measured will pose a health risk to an variety of researchapproaches F from advancing the individual or population. Additionally, there are many science to fill data gaps (e.g., collecting pharmacokinetic challenges concerning biomonitoring programs, such as

Journal of Exposure Science and Environmental Epidemiology (2007) 17(4) 309 Bahadori et al. Biomonitoring workshop findings and recommendations designing studies informative of either individual or popula- relationship to external exposures and risk assessment are tion level exposures, interpreting what the data mean in terms outlined below: of a public health context, and addressing ethical and communication issues. Characteristics of good biological markers of exposure.To Human biomonitoring studies will increasingly serve as determine what characteristics are most essential for critical drivers of environmental health decision- and policy- biological markers of exposure, several criteria for making. Bothpublic and regulatory calls for initiation of collection and use of biomonitoring data were considered, biomonitoring programs have made scientifically defensible including the persistence of the chemical being measured, interpretation of these data a growing priority for improving the time period represented or the timing of exposure human health risk evaluations. Improved methods must be vis-a` -vis the time samples were collected, sensitivity and developed that provide quantitative links of biomonitoring specificity (relative to target chemicals), characteristics of data to the potential for adverse health risks, particularly the analytical methods, feasibility for collection/analysis, through better exposure–dose associations with adverse and whether/how the method has been validated (i.e., health outcomes that are either extrapolated from animal efforts to evaluate whether and/or how biological mea- toxicology studies or established directly from human surements reflect the target chemical). The answer to what epidemiology investigations. For example, opportunities characteristics determine a good marker of exposure is clearly exist to tie the results of human biomonitoring efforts complex and depends on why the data are being collected to well-established human health risk protection standards and how they are going to be used. (e.g., reference doses derived from animal toxicity studies). If Study design. To design future biomonitoring studies, lessons adverse healthrisks are indicated from suchbiomonitoring learned from prior exposure and/or epidemiology studies analyses, that is, the aggregate biomonitored concentration (and their analyses), including those conducted in occupa- approaches or exceeds those projected from reference dose tional settings, need to be considered. Also, additional exposures, researchers, industry stewards, and public health information (e.g., data to ‘‘validate’’ biological measure- officials will be stimulated to characterize further the sources ments relative to target chemicals) may need to be collected and pathways of exposure to mitigate the risk most to improve the ability to interpret biomonitoring results effectively. Conversely, if adverse health effects appear relative to external and internal exposure. A key issue to the unlikely from suchanalyses, thepublic and policy makers design of studies is how to deal effectively with variability can focus on other priorities. and uncertainty – bothin theexposures (inter- and intra- individuals), the population, and factors such as gender, age, population/race, diet, medication, and alternative medicines. Linking Exposure, Dose, and Human Biomonitoring Once biomonitoring measurements are taken outside of the Data occupational setting, the variability increases greatly along withthecomplexity of exposure patterns. Study designs need Many health indices for chemicals (e.g., reference dose to balance the inherent properties of the compound(s) being (RfD), cancer unit risk) are described in terms of external studied, number of samples, number of subjects, and types exposures (mg/kg body weight/day, p.p.m. in air or drinking of biological media that should be measured with the water, and so on). Insofar as we can relate biomonitoring objectives, resources, and other issues, including ethics, data to external exposures, we can understand better the surrounding the study. Biomonitoring data have been used relationship between such exposures and the implied health to characterize population exposures better in epidemio- risks (e.g., whether biomonitoring is showing exposures logical studies, and they can also be used to define exposed above or below the RfD). Epidemiological studies are used versus nonexposed populations better. In addition, effects to establishrelationships between exposure and outcome biomarkers can be used insofar as they have been validated. measures, and may include a combination of environmental, Epidemiologic studies are often regarded as the ‘‘gold biomonitoring, and questionnaire/survey data to assess standard’’ for human health assessments. However, a exposures. These studies can provide a direct link between recognized weakness of many epidemiological studies is in levels measured in environmental and biological media and establishing exposure, which is done in many cases using health outcomes, potentially allowing for reduced uncertainty survey analyses or other surrogate measures. Biomonitoring and an improved estimation of risks. offers a potential approach to improving the value of these The workshop participants looked at the risk assessment studies by providing a means to validate directly the paradigm from source to environment to exposure to dose to exposure assumptions that represent the fundamental determine what additional information might be needed assumptions of the epidemiologic analyses. and used in combination withbiomonitoring data to improve Links to sources of exposures. It would be helpful to the interpretation of biomonitoring data. Key issues identify other information and/or assumptions that would associated withbiomonitoring data interpretation and its be needed to relate biomonitoring results to important

310 Journal of Exposure Science and Environmental Epidemiology (2007) 17(4) Biomonitoring workshop ﬁndings and recommendations Bahadori et al.

media and pathways contributing to exposures. Some interpretation of human population biomonitoring data. This examples include kinetics/metabolism, personal behaviors researchwould include obtaining knowledge regarding relative to exposure and uptake, timing and types of the persistence of chemicals in relevant medium (external exposures (e.g., intermittent vs. continuous), and locations and internal), the persistence of biomarkers of response, and environmental concentrations (measured, unmea- the exposure frequency and duration, and the timing of sured). Additionally, pharmacokinetic modeling can serve sampling (i.e., how sampling relates to potential as a useful tool and help with understanding the linkages exposures and whether there is variability due to multiple to exposure. Integrating pharmacokinetic data in modeling timeframes). or other statistical approaches can help to predict better Additionally, there is a need for validation or verification and/or characterize exposures and sources. ‘‘Reverse of computational tools in human subjects, with special dosimetry’’ analyses in which pharmacokinetic studies attention focused on choosing the appropriate analyte, are coupled withbiomonitoring studies provide improved measuring the concentrations in the appropriate medium, understanding of potential contributions of individual and relating the dose to target tissues. Toxicology studies and sources of exposure to the aggregate exposure captured in the use of computational tools for the interpretation of data human biomonitoring studies. could be improved by considering biomarkers of exposure Major sources of uncertainty. Issues related to uncertainty and effects, conducting additional researchinto theeffect of are important, including whether uncertainty could be timing on toxicological results, developing/using alternatives addressed and reduced through additional information to animal testing, and incorporating the use of genomics and collected during a biomonitoring study, the use of more ‘‘omics’’ technologies. specific/relevant information (suchas factors or assumptions) about the population, and/or improvements in models and analytical approaches. To reduce these Toxicological Study Design uncertainties, biological samples should be collected concurrently withavailable exposure information (e.g., Animal toxicology studies can be designed to provide data diet, air samples) and information about the context for needed to characterize better the potential health significance the sample collected (e.g., time and volume of urine of human biomonitoring data. Linking toxicology studies to collection, and duration since previous void to allow biomonitoring data has several advantages, perhaps the most calculation of output rates). In addition, the formation important of which is that toxicology studies provide the of adducts (e.g., for organophosphates) could be studied foundational data sets used to define acceptable levels of to help reduce uncertainty when looking at exposure to human exposures and thus have direct applicability to parent compounds versus their metabolites. refining interpretation of biomonitoring data. Conventional risk assessment practice generally relates external effect or no- effect doses established in animal toxicity studies to estimates Using Computational Tools of external doses in exposed human populations, that is, margin of exposure (MOE) comparisons. Although exter- Computational tools, such as physiologically-based pharmaco- nally administered animal doses can be established with a kinetic (PBPK) modeling, can be used to relate human high degree of precision, the same cannot be said for human biomonitoring data to estimates of external (environmental) dose assessments for which the multiplicity of potential and internal (target tissue) exposure/dose in bothhumanand sources, timing, and quantities of actual exposures may animal studies. However, to refine these relationships dramatically influence the uncertainty of the analysis. And, effectively, efforts will be needed in the further development it must also be realized that MOE relationships are further of the existing computational tools. For instance, PBPK complicated in that the mode of action studies have clearly models are useful tools, but tend to be data intensive. revealed that simple external dose descriptors in toxicity Moreover, data are not available for many chemicals of studies are often overly simplistic for toxicity mediated interest. Additionally, a mechanism to systematically prior- through metabolite generation. Thus, establishment of blood itize chemicals for further PBPK development, such as or target organ dose of active/surrogate metabolite(s) quantitative structure property relationship (QSPR) models, represents valued refinements to dose characterization in is necessary. ‘‘Simpler’’ generic models can be an alternative many animal toxicity studies. to PBPK models and can be used to screen/evaluate large Biomonitoring offers the opportunity to develop more numbers of chemicals similar to the pharmaceutical industry meaningful MOE comparisons based on internal dose experience that looks at hazard identification and quantita- (animal)/internal dose (human) ratios that will help reduce tive structure–activity relationships (QSAR). uncertainties in human health risk assessments of chemicals. Several other research needs were identified for the In particular, suchcomparisons not only compensate effective use of computational tools and the quantitative intrinsically for intra-species metabolic and pharmacokinetic

Journal of Exposure Science and Environmental Epidemiology (2007) 17(4) 311 Bahadori et al. Biomonitoring workshop findings and recommendations variabilities, but also for human exposures, and represent to confirm that the animals’ metabolic profile is relevant to more precisely cumulative exposure to environmental humans. Other issues associated with improving inter- chemicals. As noted above, animal toxicology studies are pretation of animal to human biomonitoring evaluations important in the development of internal dose biomarkers include the following: (1) greater variability in timing and representative of external chemical exposure, and are a rates of human exposures (e.g., humans typically ingest valuable tool in the identification of potential biomarkers and food more intermittently than rodents, and human diets of the conditions under which they can be used. are extremely heterogeneous); (2) animal dietary toxicity

Toxicology studies can and should be modified to collect studies generally produce relatively stable Cmax, Cmin, pharmacokinetic data to enable improved extrapolation to and area-under-the-curve (AUC) values over the entire human biomonitoring data. The workshop participants time of chemical treatment, while human biomonitored identified the following challenges for modifying toxicology values would be expected to be significantly more variable studies to collect pharmacokinetic data and linking them to owing to lifestyle factors impacting exposures; (3) Owing human biomonitoring data: to desire for noninvasive sampling, human biomonitoring samples may be often collected from urine, thus making Study design. Pharmacokinetic samples must be collected direct comparison to animal blood pharmacokinetic without compromising results of core toxicity evaluations parameters more complex; and (4) within humans, (e.g., for rats, no more than three samples over 24-h polymorphisms and other exposure variables not present period). Early assessment in vitro metabolism screens with in animal studies need to be considered. animal and human hepatocytes should be encouraged, because they will provide information on the metabolic relevance of different animal models to humans, support Recommendations for Research selection of relevant metabolites to measure, and inform The workshop provided an opportunity for interdisciplinary cross-species extrapolation. Since toxicology tests often and international exchange on how interpretation of human use multiple exposure routes (e.g., dietary, drinking water, biomonitoring data can be improved and what research could and inhalation), absorption, distribution, metabolism, and help fill the knowledge gaps. The ICCA LRI used the elimination (ADME) data should be collected under the workshop discussions to form the basis for its research same administration conditions as the bioassay, particu- agenda that will be regionally implemented. The recommen- larly if the animal study represents a primary data set for dations from the workshop, however, are broadly applicable risk assessment. Since existing regulatory agency ADME across government, industry, and academia, and can be most guidelines call for oral gavage characterizations, efforts to effectively implemented through interdisciplinary and multi- update the OECD and other regulatory agency ADME stakeholder collaboration and dialogue. guidelines to promote collection of dosimetry data under Recommendations for future researchin theinterpretation the exposure conditions of animal bioassays should be of biomonitoring data include the following: encouraged. Real-world exposure varies greatly over time, but dose is constant in toxicology studies; therefore, and Design and conduct studies that help relate biomonitoring thus toxicology studies can be used to explore the impacts data to relevant sources and pathways contributing to of dose rate and exposure intervals on pharmacokinetic exposures. Studies should identify and address other behavior. In addition, such studies may also illuminate the information that would be needed and that should be appropriate timing of biological samples collection (e.g., collected concurrent withthebiological monitoring in the time to reach maximum excretion). Other factors such order to relate the biomonitoring data to external as diet should also be considered in study design for their exposures. potential to impact biomarker excretion (e.g., observed Develop suitable guidance values that serve to improve the quantitative impact of fiber diet can be comparable to basis for the interpretation of human biomonitoring data differences due to enzyme polymorphism). The group and whichalign withexisting effect/risk indicators (e.g., expressed an overarching concern that interpretation of RfDs, cancer unit risks) and formulate a process for their biomonitoring data will likely be a resource-intensive consistent development. For example, studies need to be chemical-by-chemical effort, but that ADME investments developed exploring how internal dose/internal dose will nonetheless contribute significant value to those refinements to MOE risk assessment calculations can be efforts. implemented suchthatdifferential contributions of cross- Uncertainties. Information on intra- and inter-species species ADME behaviors can be compensated more factors that impact pharmacokinetic behavior is necessary, effectively, i.e., refining simple default assumptions in because interpretation of biomonitoring data will differ future risk assessments. A potential starting place for such for classes of chemicals that vary in these parameters. To investigations might include development of methods extrapolate from animals to humans, researchers need allowing comparison of the AUCs associated with animal

312 Journal of Exposure Science and Environmental Epidemiology (2007) 17(4) Biomonitoring workshop ﬁndings and recommendations Bahadori et al.

no-observed adverse effect level (NOAEL) doses to similar the best characteristics to select and prioritize substances blood AUC values observed in humans. forthetypeofadditionalresearchneeds. Derive methods for using internal dose implications for Develop a tiered framework for pharmacokinetic data refinement of risk assessments. Researchis also needed to collection for chemicals depending on toxicology screening describe how best to make comparisons of internal doses in results. animals and humans. Develop a library of past, recent/current or planned biomonitoring researchand results as well as a library of Summary validated analytical methods or standards (at both the public health and occupational levels). This will help those In summary, the workshop discussions recognized the value withan interest to assess whatwe have learned already in and potential that biomonitoring has for improving toxico- the field of biomonitoring from measurement techniques to logy, risk assessment, public health decisions, and policy- interpretation to applications. Better sharing of informa- making. Continued advancement in the interpretation and tion, suchas core information thatmay be contained in application of biomonitoring data will require coordinated product use registers and time diary databases, will help to researchand collaborative efforts on thepart of all relevant break down the barriers of institutional ownership, high- stakeholders. This workshop and the recommendations light the gaps in current biomonitoring data, and promote derived from it form the basis for the development regional research that will improve the science and understanding programs in interpretation of human biomonitoring data that of biomonitoring data and its interpretation. include other workshops and as well as requests for proposals, Develop model protocols for more effective data collection inviting submissions of interdisciplinary researchproposals in human biomonitoring studies, including when and how that further the interpretation of biomonitoring data. oftentocollectsamples.Theseprotocolsshouldconsider whether it is more advantageous to collect multiple biomarkers or multiple samples when resources are Acknowledgements limited. Develop case studies on the interpretation of biomonitor- We thank the entire Scientific Organizing Committee, ing data based on existing health data/guidelines and including Rick Becker, American Chemistry Council; Ben template descriptions of what pharmacokinetic informa- Blount, Centers for Disease Control and Prevention; tion would be most helpful for PBPK modeling of Laurent Bontoux, European Union Joint ResearchCentre; different classes of chemicals. Peter Boogaard, Shell Health Services – Shell International Develop alternatives, or less intensive models, to the ‘‘gold BV; Akira Fukushima Lion Corporation, Japan; Annette standard’’ PBPK modeling, suchas alternative screening- Guiseppi-Elie, Dupont; Doug Haines, HealthCanada; level approaches (e.g., single compartment models, Masatoshi Kumamoto, Japan Chemical Industry Associa- correlations). tion; Marsha Morgan, US EPA; Moiz Mumtaz, Centers for Develop approaches/methods for obtaining (realistic) Disease Control and Prevention; Larry Needham, Centers human pharmacokinetic data rapidly on chemicals for for Disease Control and Prevention; and Miles Okino, US use in PBPK modeling (e.g., microdosing, noninvasive EPA. This meeting report summarizes recommendations methods). Important ethical issues need to be considered that were generated by the workshop participants. The views in developing these methods. expressed do not necessarily represent positions of the Develop or apply quantitative structure property relation- participating organizations, nor do they represent an ship (QSPR) approaches for developing pharmacokinetic endorsement of any particular product or vendor. models and human kinetic data for various chemicals. Design and conduct animal and human studies to address the impacts of fundamental differences between species References

dosimetry and dose rates on the interpretation of European Centre for Ecotoxicology and Toxicology of Chemicals (ECETOC). biomonitoring data. The signiﬁcance of these differences Guidance for the Interpretation of Biomonitoring Data, 2005. Document needs to be quantiﬁed. This may involve working through number 44. November 30, 2005. National ResearchCouncil (NRC). Human Biomonitoring for Environmental a case study (or a small number of case studies) on Chemicals. NRC Committee on Human Biomonitoring for Environmental relatively data-richsubstances to helpidentify and sort out Toxicants, 2006. July 2006.

Journal of Exposure Science and Environmental Epidemiology (2007) 17(4) 313