Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 84
Human Biomonitoring for Environmental Chemicals 4 Considerations in the Design of Biomonitoring Studies In Chapter 3, the committee delineated a diverse array of uses to which biomonitoring studies are or will foreseeably be put to use (Box 3-1). Detailed discussion of the scientific approaches for each of them is beyond the scope of this report, but there are some issues in the design, conduct, and analysis of all biomonitoring studies for which the committee deemed review of scientific practice crucial to the further development of the field. The committee addresses these issues in this chapter. Although the need for attention to scientific rigor attends every emerging technology in biomedical and environmental health science, and is thus not peculiar to biomonitoring, some aspects of biomonitoring demand special attention: In developing biomarkers, there are no “gold standards” against which a result or finding can be readily evaluated; in most cases, biomonitoring will afford the first opportunity for scientists to assess even qualitatively the extent to which humans are exposed to, absorb, and might be harmed by innumerable contaminants of the human-made and natural environment. Measured concentrations of biomarkers are often extremely low (in the 1 part per billion range or below) and subject to highly uncertain causes of variation, such as individual genetic differences, age, diet, habits, weather conditions, time of day, recent activity, medication and illness, and other exposures.
OCR for page 85
Human Biomonitoring for Environmental Chemicals Compared with measures of contaminants in air, water, or food, biomonitoring results are intrinsically associated with a person and thereby have far greater potential to generate concern and action, for good or ill. The social and political climate in which the new technology of biomonitoring has emerged is itself volatile; contentious and potentially fractious policy debates and litigation surround the field and render it likely that studies will be conducted or interpreted to meet the agendas of specific parties unless great care is taken to establish uniformly agreed on scientific standards against which any study can be transparently judged. Figure 4-1 presents a schematic diagram of the various considerations in the design of a biomonitoring study addressed in the chapter as well as Chapters 5 and 6 and their relationship to one another. The four stages of any biomonitoring study are study design, study conduct, data analysis, and communication and implementation of results. Each stage incorporates several steps, which follow in chronologic order and are linked in Figure 4-1 by thick arrows. Several disciplines and processes, linked to the main steps by thin arrows, can be engaged in concurrently and are used to inform decisions made for the main steps. For example, biomarker selection and validation usually follow from study hypothesis and population selection, precede participant enrollment and consent, and are informed by statistical considerations, toxicokinetics, ethics, and communication. Study-population selection must take place before study inception. The main steps from population selection through statistical analysis are described in this chapter. Chapter 5 takes up interpretation of results, and Chapter 6 deals with communication of results. In sum, the purpose of this chapter is to lay out—for the scientific, medical, legal, and policy communities—broad guidelines aimed at guaranteeing that biomonitoring studies will lead efficiently to identification of environmental contaminants that are causing risk or harm while elucidating sufficient information regarding pathways of exposure and health effects to guide their future control and will avoid the creation of widespread anxiety or apathy about contaminants whose potential for personal or societal risk appears not to warrant that reaction. The discussion will proceed by reviewing the major issues in selection of biomarkers for study, developing the sampling strategy to answer the study questions, and assessing the communication and ethical considerations that must be addressed before the study is conducted. Next, the chapter will review the major considerations regarding the execution of the study, selection of the appropriate matrix (such as, blood or urine), collection of samples, transportation of samples to the laboratory, analysis of the samples, and banking of the specimens, when relevant, for future additional analyses. Finally, we review key considerations in the statistical analysis of the laboratory results.
OCR for page 86
Human Biomonitoring for Environmental Chemicals FIGURE 4-1 Stages of a biomonitoring study. The committee deliberately incorporated discussion of communication and ethical considerations into this chapter not only because these issues present some of the most significant challenges with respect to interpretation and use of the biomonitoring data (key considerations in the committee’s charge), but because it was the committee’s intent to prompt readers to consider these issues as intrinsic in the design of biomonitoring studies. STUDY DESIGN Design of a biomonitoring study incorporates several key components, including consideration of the study hypothesis, the properties of the bio-
OCR for page 87
Human Biomonitoring for Environmental Chemicals markers to be used, the selection of the population to be sampled, and ethical and communication issues. Each of those components will depend on the intended uses of the biomonitoring data. Relevant Considerations in the Selection of Biomarkers Several criteria must be considered in selecting a biomarker. The criteria—which include sensitivity, specificity, biologic relevance, and practicality—should be met regardless of the intended use of the biomarker (Metcalf and Orloff 2004; NRC 1991). However, rarely does a biomarker satisfy all the criteria (Metcalf and Orloff 2004). The relative strengths of the criteria for a particular biomarker should guide its applications, as discussed in Chapter 3. In addition to the criteria listed above, information on the kinetics of a biomarker is critical to its use (Bernard 1995). A description of the criteria for biomarkers follows with illustrations of how they may influence a biomarker’s use. Sensitivity A biomarker should be capable of measuring a chemical or its metabolites after exposure. It should vary consistently and quantitatively with the extent of exposure (especially at low doses) (Bernard 1995; NRC 1987). However, exposures in community settings are typically lower than exposures in the occupational setting. So, for instance, in measuring chemicals in the workplace, the required limit of detection may be much higher than that needed for assessing environmental exposures in the general population. Figure 4-2 illustrates the contribution of environmental and occupational exposure to biomarker concentrations and the effect of a method’s limit of detection, LOD (or limit of quantification), on potential uses of the biomarker. The example assumes that the biomarker’s concentration in a given biologic medium results from endogenous processes and from external exposure to a chemical. If the environmental (or community) exposure is sufficiently large compared with the endogenous contribution, then, given the variability of the latter, the environmental exposure might be reliably assessed, provided that the analytic method has sufficient sensitivity. In the graph, a method with a limit of detection LOD1 would be adequate, but not a method with a sensitivity of LOD2. If workplace exposure to the chemical is large compared with both endogenous and community exposures, the biomarker could be very useful as a tool for preventing adverse health effects, even with the less sensitive analytic LOD2 method. There might also be cases where the extent of community exposure may overlap with occupational exposure and still
OCR for page 88
Human Biomonitoring for Environmental Chemicals FIGURE 4-2 Contribution of exposures to biomarker concentrations and effect of limit of detection (LOD) on its potential uses. others where the endogenous contribution to a biomarker concentration exceeds that of community exposure to the parent compound. Specificity The biomarker should be specific for the chemical or metabolites of interest; that is, it needs to be an unambiguous marker of exposure. Measurement of the unchanged parent chemical may have greater specificity than that of the metabolite, which may be common to several substances (Bernard and Lauwerys 1986). For example, if the metabolite of the parent chemical is being measured, the result may be equivocal if the same metabolite is produced endogenously or formed after exposure to other compounds. Occupational exposure to high air concentrations of benzene was formerly monitored by testing for its metabolite, phenol, in urine. However, the use of phenol to measure small environmental exposures to benzene is problematic in that many foods contain phenol, and the normal
OCR for page 89
Human Biomonitoring for Environmental Chemicals catabolism of proteins in the body also gives rise to phenol excretion (Metcalf and Orloff 2004). An example of a nonoccupational exposure is methanol, which is formed endogenously, probably as the result of the activities of intestinal flora or enzymatic processes. It is present in a number of consumer products. Methanol may be present in low concentrations in some foods, juices, and alcoholic beverages. Methanol can also be derived from the intestinal enzymatic hydrolysis of the artificial sweetener aspartame, which results in methanol absorption from the intestine (Butchko et al. 2002). It is estimated that a 355-mL serving of aspartame-sweetened beverages and of various fruit and tomato juices may contribute about 20-100 mg of dietary methanol (Butchko et al. 2002). For comparison purposes, exposure at the current Threshold Limit Value time-weighted average of methanol (262 mg/m3) would result in a daily dose of about 1,500 mg, assuming an 8-hour inhaled volume of 10 m3 of air and absorption of 57%. Biologic Relevance The biomarker should be relevant to the exposure-disease continuum. However, as discussed in Chapter 3, depending on the information that a particular biomarker provides, it is critical to consider what is known about it with respect to exposure vs health effect. As our scientific knowledge increases, our understanding of where biomarkers lie on the continuum may change (NRC 1987). That is made clear by Schulte and Talaska (1995), who define biologic relevance of markers as the extent to which they represent the underlying biologic event. The authors state that “without demonstration of a direct relationship to exposure and outcome, each biomarker study is actually a test of the biological relevance of the marker and adds to the web of association.” However, biomarkers are being used even when there is little information on exposure or health effects. The classification of biomarkers described in Chapter 3 provides a useful framework for thinking about the characteristics of biomarkers that make them relevant and useful for various applications and provides an important assessment of potential research gaps. Practicality Several practical considerations are important in the collection and analysis of biologic samples. A sample should be readily obtainable, storable for a certain period, and capable of being analyzed. (More information on the choice of matrix and logistics of sample collection and processing will be presented later, under “Sample Collection and Processing.”) The
OCR for page 90
Human Biomonitoring for Environmental Chemicals cost of analyses is a key consideration in that it often limits the number of participants in community investigations. For instance, analytic costs can be $15-20 for a simple blood analysis or $1,000-2,000 for an analysis of dioxin congeners (Metcalf and Orloff 2004). In large population-surveillance studies, such as those used by the Centers for Disease Control and Prevention (CDC), biomonitoring is usually conducted on urine and blood samples. However, in research investigations, other matrices—such as breast milk, cord blood, and breath samples—may be used. The specific biomarker used will also depend on its intended application and the population that is being sampled. For reasons of practicality and study participant convenience, most investigators collect first-morning voids or spot urine samples (Barr et al. 2005a). One recent study in children concluded that measurements of organophosphate metabolites in the first morning void more accurately represented total daily exposure than measurements in spot urine samples collected at other times during the day (Kissel et al. 2005). However, a first morning void specimen might seriously underestimate daily workplace exposure to a rapidly metabolized chemical, whereas one collected at the end of the workday would overestimate 24-hour exposure. Exposures that are episodic over a period of days might be missed entirely with either sampling regimen, but a spot urine sample could be representative of situations involving chronic exposures and intermittent exposures occurring on time scales less than the compound’s metabolic half-life (Barr et al. 2005a). With regard to blood samples, for some analyses, such as dioxins, large samples of blood (70 mL or more) are required, and collecting this volume may eliminate some susceptible subpopulations, such as children and pregnant women (Metcalf and Orloff 2004). CDC’s National Center for Environmental Health does not collect blood samples on children less than 6 years old except to analyze lead and cadmium (and in the future mercury), because it is difficult to collect the necessary blood volume (J. Osterloh, CDC, personal commun., July 27, 2005). Pharmacokinetics A key consideration regarding the practical aspects of biomarkers is the pharmacokinetics of the chemical. The measure usually referred to is the half-life, which reflects both the affinity of the chemical for the biologic matrix and the efficiency of metabolic or elimination processes. Knowledge of half-life is important for several reasons, including its use in determining sampling time (Bernard 1995). For instance, chemicals with short half-lives (a few days or even a few hours)—including cotinine, phthalates, volatile
OCR for page 91
Human Biomonitoring for Environmental Chemicals organic compounds, and current-generation pesticides—are rapidly eliminated from the body. For chemicals with short half-lives, the biomonitoring result will reflect only very recent exposures, within the past several hours. As Figure 4-3 shows, the shorter the half-life, the more recent the exposure has to be for it to be detected in a biomonitoring sample. This makes the relative timing of the exposure vs the taking of the biomonitoring sample, a critical determinant of the biomonitoring result. Since the biomarker concentration decreases with time, knowledge of the time lapsed between exposure and sampling is needed to calculate the dose correctly. In practice, it is usually difficult, if not impossible, to tell with any accuracy when exposure occurred. Thus, variability in sampling time (in relation to exposure) introduces huge variability in dose estimates. Additional indexes of past exposures (such as, job classification and information from exposure questionnaires) may need to be collected (Bernard 1995). However, a biomarker with a short half-life can still provide a reliable internal dosimeter if exposures are relatively constant. Cotinine, for example, has a half-life of 15-40 hours in serum, but a single determination provides a good dosimeter of steady-state concentrations in people who have stable smoking habits (Kemmeren et al. 1994). Biological half-lives of most phthalates are also short, on the order of hours, so that urinary metabolites likely reflect exposures over only the preceding day (Hauser et al. 2004). Nonetheless, monoester metabolites of four phthalates have been detected in greater than 75% of urine samples collected in the 1999-2000 NHANES (Silva et al. 2004) indicating that exposures in the United States are commonplace and frequent. In a study of five phthalate metabolites that were measured in repeat spot urine samples collected over three months from ten men (n = 90 samples), Hauser et al. (2004) concluded that the measurements of metabolite levels in a single spot urine sample were moderately predictive of exposures, with sensitivities and specificities ranging from 0.56 to 0.90 for the various phthalate metabolites for a single urine sample to predict the highest 3-month average. The measurement of the phthalate metabolites in two spot urine samples 1-3 months apart was sufficient to capture within person variability, considering both month-to-month and day-to-day variance (Hauser et al. 2004). For chemicals with half-lives of months or years—such as dioxins, polychlorinated biphenyls (PCBs), polybrominated biphenyl ethers, and first-generation halogenated insecticides—biomarkers can detect exposures months or even years after they have occurred. Such lipophilic chemicals are usually measured in blood, and the principal exposure source is usually diet. After ingestion, they are readily absorbed into the blood supply; blood concentration then decreases rapidly as the blood supply equilibrates with
OCR for page 92
Human Biomonitoring for Environmental Chemicals FIGURE 4-3 Effect of half-life on contributions of exposures during the last pre-sampling hour, day, week, month, and half-year to biologic levels of determinants. Half-lives were calculated by a one-compartmental model. For example, if the determinant is eliminated with a half-life of ten hours, the biological level mainly reflects the exposure on the day prior to sampling (contribution of 70%); to a relatively small extent, it reflects the exposure during the previous hour and week (contributions of 10% and 20%, respectively). Source: ACGIH 1995. Reprinted with permission; copyright 1995; American Conference of Governmental Industrial Hygienists. lipid-rich tissues (Flesch-Janys et al. 1996). After the initial rapid equilibration, the concentration measured in a blood sample is related to body burden and so should not change substantially in the short term. The importance of the biologic half-life is illustrated with a hypothetical example in Figure 4-4. A biomarker with a short half-life (such as 1 day, in the upper right graph) yields information only about the most recent exposures and does so only if the time of sampling in relation to exposure is known. For a biomarker with a long half-life (such as 1 year, in bottom right graph), the concentration continues to build up over time, so the total exposure duration (and age) of the person is a key factor. In this case, it may be difficult to follow exposure trends in the same people. In geographic population surveys, it may be advantageous to have intermediate half-lives (such as 1 month, in the bottom left graph), in which case a pseudo-steady-state is reached so that the biomarker concentrations reflects continuing average exposure with little influence from sporadic exposure peaks, age, or migration of people to the area under study. The example given here is representative of many environmental pollutants that are ingested in food. The reasoning can be the same for biomonitoring of ambient air pollutants in the workplace, but the time scales would usually be minutes, hours, or days rather than days, months, or years. In conclusion, the half-life of the chemical needs to be considered in
OCR for page 93
Human Biomonitoring for Environmental Chemicals FIGURE 4-4 Influence of biologic half-life relationship between exposure level (E, upper left) and biomarker level (BM). Daily exposure levels were created by Monte Carlo sampling from auto-correlated lognormal distribution. Observation time is 400 days. Biomarker levels were calculated for three half-lives—1 day (upper right), 1 month (lower left), and 1 year (lower right)—with one-compartment pharmacokinetic model.
OCR for page 94
Human Biomonitoring for Environmental Chemicals relation to the time spans of exposure duration, temporal variations in exposure, and sampling. For the sake of simplicity, simple monophasic pharmacokinetics (one compartment and one half-life) was assumed in the above example and in many other examples in this report. In real life, most chemicals express biphasic or polyphasic pharmacokinetics (several compartments and several half-lives). Squeezing a polyphasic pharmacokinetic behavior into a one-compartment model by assuming a single half-life may lead to negligible errors for some chemicals and serious misinterpretation of biomarker concentrations for others. The same can be said about nonlinear processes, such as metabolic induction, inhibition, and saturation. A good way to check the accuracy of a simple pharmacokinetic model is to verify its performance by comparing with a physiologically based pharmacokinetic (PBPK) model that may encompass the mentioned factors. In conclusion, great effort should be made to develop a human pharma-cokinetic, preferably PBPK, model early in the study design. The likely influence of, for example, model simplification (such as assuming a single half-life), metabolic saturation (see, e.g., Liira et al. 1990), and sampling time can then be addressed before investment of vast resources in sampling and analyses. By using statistical tools (Monte Carlo simulation and population models) in the model, one can examine additional features, including variability in exposure pattern (e.g., Nihlén and Johanson 1999) and intra-individual and interindividual variability in pharmacokinetic determinants, such as workload (e.g. Droz and Fernandez 1977), body build, and metabolic genotype (e.g., Jonsson and Johanson 2001). Sampling of Populations In exploratory investigations, studies of occupational groups, and clinical applications, the choice of subjects on whom biomonitoring should be done is generally straightforward. However, for most other applications (see Box 3-1), values of exogenous chemicals in blood or urine cannot practically be obtained from every member of the group of interest. Instead, researchers or public-health authorities have no choice but to obtain specimens on a sample of the population, from which statistical inferences will later be drawn regarding the (generally much larger) group as a whole. To be successful, the strategy, similar in concept to taking a poll before an election, must be performed in a scientifically valid and efficient way. For one thing, if results are to be extended across all age groups and both sexes, adequate numbers of males and females of all ages must be included in the sample. If there is reason to suspect the importance of some special factor as a determinant of levels, such as proximity to a source or dietary preference, people with various levels of such “risk factors” should be included.
OCR for page 121
Human Biomonitoring for Environmental Chemicals testing, provides some assurance that test methods are documented and that laboratory staff are appropriately qualified to perform them. STATISTICAL ANALYSIS The first principle of analysis is that it follows design. That is particularly true of biomonitoring data. At the beginning of this chapter, we emphasized the importance of describing the selection of the sampled population. Ideally, specific analyses (based on specific hypotheses) have been outlined at the beginning of the study. The statistical jargon phrase for characterizing the sampled population is the target of inference. A dataset may have different targets of inference—with resulting differences in the statistical analysis. For example, an occupational physician faced with a specific biomarker concentration in a specific worker is faced with the inferential question of whether the worker belongs to the group of workers “exposed” or “not exposed.” The inference will be drawn by bringing in other factors, such as a careful work history. If the same physician is interested in the worker’s result for a study of occupational disease, the target of inference is different, and different statistical analyses will be required. Statisticians would couch the first question in terms of fixed-effect analyses and the second in terms of random-effects analyses. Closely tied to the target of inference is the measurement process and its precision. As mentioned previously, if only a single population of inference is of interest, the precision of the measurement is the key driver and sample size determinations should reflect the required precision. If comparisons between two populations are desired, then rather than the precision of the measurements, the magnitude of the difference between the two populations becomes the key driver in the sampling effort. Before complicated statistical models are constructed and run—increasingly easy with more and more powerful statistical computing packages—it is absolutely necessary to describe the basic characteristics of each variable—number of observations, mean, standard deviation, minimum, and maximum. That will reveal which data are below the limits of detection, are missing, are miscoded, and are outliers. If the study involves three or four key variables, associations among the variables should also be examined. Histograms and scatterplots will reveal data structures unanticipated from the numerical summaries. If the study is exploratory, it is not uncommon to have a multitude of creative ideas about the nature of the biomarker response and its relationship to the exposure. Ideally, some comparisons of particular interest are specified in advance. The analyses can then be divided into confirmatory and exploratory phases. Two key considerations in such exploratory studies are selection bias and confounding; both were discussed in the section on the sam-
OCR for page 122
Human Biomonitoring for Environmental Chemicals pling of populations, and they should be addressed specifically at the analytic stage. Often, they appear as a throwaway paragraph in the “Study Limitations” section of a paper or a thesis. They deserve a more prominent place. Longitudinal data are increasingly common. Some of the larger national surveys are considering incorporating longitudinal features. The primary reason is that such approaches can detect shifts more precisely. The Achilles heel of longitudinal studies is missing data. Causes of missing data include death, moving from the area, and refusal to continue with the study. Statisticians have developed a useful terminology for patterns of missingness. Longitudinal data are usually analyzed with fairly complicated statistical models, such as mixed-effect regression models, generalized estimating equations, and logistic-regression models. The validity of a particular model depends heavily on the missing-data pattern. Given that an appropriate class of models has been identified, secondary concerns enter, such as precision of the estimates. The models are not for the statistically naïve, and a professional statistician or someone well versed in the field needs to be consulted. A specific example of longitudinal data involves the monitoring of pesticide handlers in the eastern part of Washington state. The pesticides of concern are cholinesterase inhibitors. A baseline acetylcholinesterase (AChE) concentration is obtained from each handler before the spraying season begins, and handlers are monitored after every 30 hours of handling activity. The end point is AChE depression from baseline expressed as a percentage. If the depression is greater than 20%, specific regulatory actions are begun. The data collected are relevant for each handler but also for the agricultural industry. Issues of false-positive and false-negative depressions also have to be dealt with. In this example, the nature of the missing data becomes very important. Suppose that workers become too sick from the pesticide exposure and simply quit work. This would introduce selection bias. The growers tend to focus on false positives, the workers on false negatives. Resolution of these conflicting aims requires a careful understanding of the within-subject variability and measurement error. The analyst needs to understand that statistical models vary in their inferential utility. Linear models of some type or other are the most common and the most easily analyzed with statistical computing packages, but they may be only rough approximations of the real world. An oft-quoted aphorism of G.E.P. Box is that “all models are wrong, some are useful.” That is no doubt true, but it misses another level of detail of such models as follows. At the simplest level, a model fits the data. At the next level, a model predicts the data. At its most useful level, a model shows unanticipated features of the data and the research, and this is the ideal especially for biomarker research. The most exquisite characterization of association
OCR for page 123
Human Biomonitoring for Environmental Chemicals does not necessarily constitute evidence of causation. As new biomarkers are investigated, the causal link to exposure is a first requirement. It is usual whenever biologic samples are obtained in medical practice or in any health encounter for results to be expressed in relation to “normal.” In fact, a relationship to “normal” is the question almost every subject asks first when presented with such information. For biomonitoring results, it is a generally meaningless question—“normal” for human-made chemicals or chemicals that did not enter the environment except for human activity, is actually zero. Despite that, many laboratories report results, such as blood lead concentrations, in just this way—a practice likely to result in confusion. For example, it is not rare to see reports of blood concentrations that contain the phrase “normal for industrial workers.” In truth, such concentrations may be common in industrial workers, or even typical, but hardly normal in the usual meaning of that word in relation to health. Such comparisons are especially problematic for biomarkers whose relationship to health and environmental sources is less well studied. The discussion below touches briefly on the scientific issues raised by such implicit or explicit comparisons of results with reference ranges. Although the notion of “normal” is not appropriate to biomonitoring, there are two axes of potential reference around which comparisons may be pertinent and indeed important. The first is in relation to comparable subjects. For example, where adequate information exists, it would be entirely appropriate to express values in relation to one or more groups of subjects. For example, it would be reasonable to report an individual or group of blood lead concentrations in children as “comparable with those measured (extensively) in populations of children not known to have a specific exposure source” or “suggestive of an identifiable source of exposure.” In adults, it might be reasonable to express lead concentrations as “comparable with concentrations seen in the population not occupationally exposed to lead” “within the range of concentrations seen in workers exposed to lead under well-controlled conditions,” or the like. Those are not assessments of normality or health effects, merely comparisons with other potentially relevant populations that could lead to research inferences or clinical or public-health interventions. A second axis for interpretation of results, relevant only in the small number of situations where information is available to link a biomarker concentration to risk or clinical effects, is comparison with such risk or health-effect concentrations, for example, at or below the concentrations previously associated with developmental delays in children. The point is that the health comparison must be explicit and referenced to the source of the inference, such as an epidemiologic study or reference publication.
OCR for page 124
Human Biomonitoring for Environmental Chemicals SUMMARY AND CONCLUSIONS Biomonitoring challenges laboratory methods, ethical considerations, and communications strategies. It also challenges the technical limits of epidemiology and biostatistics. This chapter has provided guidelines for the conduct of biomonitoring studies, including study design, study conduct, and data analysis. Because not all biomonitoring studies are conducted with the same rigor, it is important that the guidelines presented be followed to ensure, to the extent possible, that biomonitoring studies will lead to the identification of chemicals that are causing risks or health effects, will provide information on exposure pathways and health effects to guide future control efforts, and will avoid anxiety or apathy about chemicals where personal or societal risk appears not to warrant that reaction. Wherever blood or other matrices are being collected from a sample of any population—whether for surveillance, etiologic study, or clinical evaluation—the highest standards of sampling theory should be adhered to, and the approach should be explicitly described in publications and reports so that biases may be recognized. If it is feasible, results should be expressed in terms not only of age groups, sex, and race, but also in relation to quantifiable lifestyle factors, such as occupation, income, and education. A great deal of effort and time is involved in getting studies approved. Hence, it is absolutely essential to address questions of ethics at the design stage. Informed consent and IRB approval became especially important when more than one research site or jurisdiction is involved, as this can introduce a problematic cycle of approval at one level and modification at another. Because the value of biomonitoring in the long run is the ability to link the sample result back to human health and risk factor information, the links between the biomonitoring and health data should not be severed, while at the same time ensuring confidentiality of the data. A topic linked to ethics but with its own issues and strategies is communication. The committee is convinced that communication starts with the study proposal and continues through study design. Each of the many constituencies associated with a proposed study requires careful consideration in planning communication strategy and content. Most biomonitoring studies will eventually have applications, and researchers need to anticipate potentially affected communities and plan for communication with them. A carefully, comprehensively designed study facilitates study conduct. Two key elements of study conduct are sample collection and laboratory analysis. The choice of matrix depends on both theoretical and practical considerations. Those include the tissue most likely to represent the biomarker route of exposure and the ease of obtaining samples. Careful attention to collection and shipping from field to laboratory is a prerequisite for valid laboratory analysis. The choice of analytic laboratory methods is
OCR for page 125
Human Biomonitoring for Environmental Chemicals determined by cost, sensitivity, and specificity. Quality control surrounds both the collection and the laboratory analysis of samples. Analysis of biomonitoring data, following the strategies initiated in study design, begins with description of the data in ways that are transparent. Sources of a priori importance and those introduced by random variation must be explicitly addressed in any statistical model, as must the context for tests of statistical significance. In every case, diligent attention to noncausal relationships within the data—associations due to bias and confounding— demand the greatest professional attention because such spurious associations are likely when so many chemicals are studied, and hypotheses remain so open-ended. Data analysts should approach results mindful that for the vast majority of biomarker measurements, results can neither be claimed to be normal in the usual sense—zero is probably the original human condition for most—nor be assigned value judgments, such as “high.” Any such designations should emerge from the interpretation of the data in context, not from the results of a single individual study or survey. RECOMMENDATIONS A biomonitoring-study report should contain a detailed description of the origin of the sample of subjects selected for study. The vast majority of biomonitoring studies are not based on a CDC-like probabilistic sample of the population. Investigators should state explicitly the population that their results apply to. Editors of journals should insist that this information be included in any submission. Any analysis of biomonitoring data should include an assessment of the importance of sources of variation. Basic sources to be considered are laboratory, intra-individual, interindividual, and variability attributable to groups. Great effort should be made to develop a human pharmacokinetic, preferably PBPK, model early in the study-design process. The likely influence of, for example, model simplification (for example, assuming a single half-life), metabolic saturation, and influence of sampling time may then be addressed before spending vast resources on sampling and analyses. By adding statistical tools (Monte Carlo simulation, population models) to the model, additional features may be examined including variability in exposure pattern and intra- and interindividual variability in pharmacokinetic determinants. Investigators, including CDC, that conduct surveys of biomarkers in the population should routinely collect detailed information about SES, lifestyle, and other cofactors on each subject and should routinely present results organized in a way that addresses the issue of whether biomarker concentrations vary as a function of each. Epidemiologic analyses of bio-
OCR for page 126
Human Biomonitoring for Environmental Chemicals markers in relation to health should routinely include appropriate adjustments for such covariates. Incorporating communication in the design of a biomonitoring study is essential. The type of communication needed will depend on the goals of the biomonitoring study. Careful attention to planning effective communication at the beginning of a biomonitoring study will make communication at the end of the study easier and may make the technical aspects of the study proceed more smoothly. Biomonitoring funders should require communication planning in any application for support. Research is needed to develop and disseminate effective methods for evaluating communication of study results. With respect to providing information to study participants, the committee considers that failing to provide detailed information about all biomarkers to be measured, no matter how many chemicals are involved in the study, raises ethical questions. Blanket consent for use of biomonitoring samples at some future time has the potential to result in abuse. However, there are practical difficulties in repeated tissue sampling that the absence of blanket consent reinforces. Research is needed to develop new approaches for obtaining consent for future uses of biomonitoring data. Biologic and environmental specimens should be carefully collected with rigorous QA-QC provisions and should be processed and banked in multiple aliquots with comprehensive characterization of their origin and history. Laboratory analysis of human samples for trace contaminants or their metabolites inevitably produces results that deviate quantitatively from the actual concentrations. Such deviations can, for example, complicate exposure classifications in epidemiologic studies, detection of time trends in exposure, and comparison of studies that use data produced with different analytic methods. Individual laboratories can use standard QA-QC methods to minimize and define the magnitude of the variations. However, federal agencies and statutes, such as CDC, the National Institute of Standards and Technology, and statutes such as CLIA, could play important roles in improving the overall quality of biomonitoring laboratory data and their utility in health-related applications. NHANES and other studies are rapidly expanding the list of chemicals for which population-based reference values are available. Health practitioners will increasingly seek to compare data from potentially occupationally or environmentally exposed patients with those reference values. That application requires consistency between data generated from different laboratories. CMS, through the CLIA program, regulates clinical-laboratory testing for nonresearch-related purposes. However, the program does not emphasize testing for chemicals associated with environmental or occupational
OCR for page 127
Human Biomonitoring for Environmental Chemicals exposure. The quality of such data would be improved if CMS created a new chemistry subspecialty related to environmental and occupational medicine and expanded the array of regulated analytes supplied by approved proficiency-test providers. It is essential to develop and use noninvasive and ultrasensitive specimen-collection techniques for the biomonitoring of children and other groups that can provide only small samples. Researchers should anticipate maintenance of samples in biorepositories (banks) for future generations of researchers. Such banking requires attention, at the planning and design stages, to ethical considerations, communication strategies, location and storage of samples, and—last but not least—the costs involved. As soon as practicable, researchers should standardize and harmonize biomonitoring measurements to make scientific communication feasible. That may involve the development of reference materials that can be shared among laboratories. Standardization and harmonization are particularly important for international collaboration. REFERENCES ACGIH (American Conference of Governmental Industrial Hygienists). 1995. Topics in Biological Monitoring: A Compendium of Essays. Cincinnati, OH: American Conference of Governmental Industrial Hygienists. Adler, P.S. 2002. Science, politics, and problem solving: Principles and practices for the resolution of environmental disputes in the midst of advancing technology, uncertain or changing science, and volatile public perceptions. Penn. St. Environ. Law Rev. 10(2):323-343. Andrews, C.J. 2002. Humble Analysis: The Practice of Joint Fact-Finding. Westport, CT: Praeger. Balch, G.I., and S.M. Sutton. 1995. Putting the first audience first: Conducting useful evaluation for a risk-related government agency. Risk Anal. 15(2):163-168. Barr, D.B., R.Y. Wang, and L.L. Needham. 2005a. Biologic monitoring of exposure to environmental chemicals throughout the life stages: Requirements and issues for consideration for the National Children’s Study. Environ. Health Perspect. 113(8):1083-1091. Barr, D.B., L.C. Wilder, S.P. Caudill, A.J. Gonzalez, L.L. Needham, and J.L. Pirkle. 2005b. Urinary creatinine concentrations in the U.S. population: Implications for urinary biologic monitoring measurements. Environ. Health Perspect. 113(2):192-200. Beecher, N., E. Harrison, N. Goldstein, M. McDaniel, P. Field, and L. Susskind. 2005. Risk perception, risk communication, and stakeholder involvement for biosolids management and research. J. Environ. Qual. 34(1):122-128. Bernard, A.M. 1995. Biokinetics and stability aspects of biomarkers: Recommendations for application in population studies. Toxicology 101(1-2):65-71. Bernard, A.B. and R. Lauwerys. 1986. Present status and trends in biological monitoring of exposure to industrial chemicals. J. Occup. Med. 28(8):558-562. Bernard, A., and R. Lauwerys. 1987. General principles for biological monitoring of exposure to chemicals. Pp. 1-16 in Biological Monitoring of Exposure to Chemicals: Organic Compounds, M.H. Ho, and H.K. Dillon, eds. New York: Wiley.
OCR for page 128
Human Biomonitoring for Environmental Chemicals Butchko, H.H., W.W. Stargel, C.P. Comer, D.A. Mayhew, C. Benninger, G.L. Blackburn, L.M. de Sonneville, R.S. Geha, Z. Hertelendy, A. Koestner, A.S. Leon, G.U. Liepa, K.E. McMartin, C.L. Mendenhall, I.C. Munro, E.J. Novotny, A.G. Renwick, S.S. Schiffman, D.L. Schomer, B.A. Shaywitz, P.A. Spiers, T.R. Tephly, J.A. Thomas, and F.K. Trefz. 2002. Aspartame: Review of safety [review]. Regul. Toxicol. Pharmacol. 35(2 Pt 2):S1-S93. Comfort, L.K. 1985. Action research: A model for organizational learning. J. Pol. Anal. Manage. 5:100-118. Daston, G., E. Faustman, G. Ginsberg, P. Fenner-Crisp, S. Olin, B. Sonawane, J. Bruckner, W. Breslin, and T.J. McLaughlin. 2004. A framework for assessing risks to children from exposure to environmental agents. Environ. Health Perspect. 112(2):238-256. Droz, P.O., and J.G. Fernandez. 1977. Effect of physical workload on retention and metabolism of inhaled organic solvents. A comparative theoretical approach and its applications with regards to exposure monitoring. Int. Arch. Occup. Environ. Health. 38(4):231-246. European Commission. 2004. Baseline Report on “Biomonitoring for Children” in the Framework of the European Environment and Health Strategy (COM(2003)338 final). Prepared by the Technical Working Group on Integrated Monitoring. January 9, 2004 [online]. Available: http://www.brussels-conference.org/Download/baseline_report/BR_Biomonitoring_final.pdf [accessed June 3, 2005]. ECETOC (European Centre for Ecotoxicology and Toxicology of Chemicals). 2005. Guidance for the Interpretation of Biomonitoring Data. Document No. 44. Brussels, Belgium: ECOTOC. Ehrmann, J.R., and B.L. Stinson. 1999. Joint fact-finding and the use of technical experts. Pp. 375-400 in The Consensus Building Handbook: A Comprehensive Guide to Reaching Agreement, L. Susskind, S. McKearnan, and J. Thomas-Larmer, eds. Thousand Oaks, CA: Sage. EPA (U.S. Environmental Protection Agency). 2002. Community, Culture and the Environment: A Guide to Understanding a Sense of Place. EPA 842-B-01-003. Office of Water, U.S. Environmental Protection Agency, Washington, DC [online]. Available: http://www.epa.gov/ecocommunity/pdf/ccecomplete.pdf [accessed Dec. 1, 2005]. Eskenazi, B., A. Bradman, E.A. Gladstone, S. Jaramillo, K. Birch, and N.T. Holland. 2003. CHAMACOS, A longitudinal birth cohort study: Lessons from the fields. J. Child. Health 1(1):3-27. Eskenazi, B., E. Gladstone, G. Berkowitz, C. Drew, E. Faustman, N.T. Holland, S. Lanphear, S. Meilsel, F. Perera, V. Rauh, A. Sweeney, R. Whyatt, and K. Yolton. 2005. Methodologic and logistic issues in conducting longitudinal birth cohort studies: Lessons learned from the Center of Children’s Environmental Health and Disease Prevention Research. Environ. Health Perspect. 113(10):1419-1429. Fairchild, A.L., and R. Bayer. 2004. Ethics and the conduct of public health surveillance. Science 303(5658):631-632. Fisher, A., M.T. Pavlova, and V.T. Covello, eds. 1991. Evaluation and Effective Risk Communications: Workshop proceedings. EPA/600/9-90/054. Interagency Task Force on Environmental Cancer and Heart and Lung Disease, Center for Environmental Research Information. U.S. Environmental Protection Agency, Washington, DC. Flesch-Janys, D., H. Becher, P. Gurn, D. Jung, J. Konietzko, A. Manz, and O. Papke. 1996. Elimination of polychlorinated dibenzodioxins and dibenzofurans in occupationally exposed persons. J. Toxicol. Environ. Health Part A 47(4):363-378. Ginsberg, G., D. Hattis, and B. Sonawane. 2004. Incorporating pharmacokinetic differences between children and adults in assessing children’s risks to environmental toxicants. Toxicol. Appl. Pharmacol. 198(2):164-183.
OCR for page 129
Human Biomonitoring for Environmental Chemicals Goodman, G.E., M.D. Thornquist, C. Edelstein, and G.S. Omenn. 2006. Biorepositories: Let’s not lose what we have so carefully gathered! Cancer Epidemiol. Biomarkers Prev. 15(4):599-601. Grill, K., and S.O. Hansson. 2005. Epistemic paternalism in public health. J. Med. Ethics 31(11):648-653. Gunter, E.W. 1997. Biological and environmental specimen banking at the Centers for Disease and Control and Prevention. Chemosphere 34(9/10):1945-1953. Hauser, R., J.D. Meeker, S. Park, M.J. Silva, and A.M. Calafat. 2004. Temporal variability of urinary phthalate metabolite levels in men of reproductive age. Environ. Health Persp. 112(17):1734-1740. Heinrich-Rumm, R., G. Lehnert, and J. Angerer. 1996. The German external quality assessment scheme in occupational and environmental medicine. Ann. Ist. Super Sanita. 32(2):2 47-251. Holland, N.T., M.T. Smith, B. Eskenazi, and M. Bastaki. 2003. Biological sample collection and processing for molecular epidemiological studies. Mutat. Res. 543 (3):217-234. Holland, N.T., L. Pfleger, E. Berger, A. Ho, and M. Bastaki. 2005. Molecular epidemiology biomarkers: Sample collection and processing considerations. Toxicol. Appl. Pharmacol. 206(2):261-268. IOM (Institute of Medicine). 1999. Toward Environmental Justice: Research, Education, and Health Policy Needs. Washington, DC: National Academy Press. ISBER (International Society for Biological and Environmental Repositories). 2005. Best practices for repositories I: Collection, storage, and retrieval of human biological materials for research. CPT 3(1):5-48. Jonsson F., and G. Johanson. 2001. A Bayesian analysis of the influence of GSTT1 polymorphism on the cancer risk estimate for dichloromethane. Toxicol. Appl. Pharmacol. 174(2):99-112. Kaneko, A., J.K. Lum, L. Yaviong, N. Takahashi, T. Ishizaki, L. Bertilsson, T. Kobayakawa, and A. Bjorkman. 1999. High and variable frequencies of CYP2C19 mutations: Medical consequences of poor drug metabolism in Vanuatu and other Pacific islands. Pharmacogenetics 9(5):581-590. Kass, N.E. 2005. Research Ethics and Human Biomonitoring Research. Presentation at the Second Meeting on Human Biomonitoring of Environmental Toxicants, April 28, 2005, Washington, DC. Kemmeren, J.M., G. van Poppel, P. Verhoef, and M.J. Jarvis. 1994. Plasma cotinine: Stability in smokers and validation of self-reported smoke exposure in nonsmokers. Environ. Res. 66(2):235-243. Kissel, J.C., C.L. Curl, G. Kedan, C. Lu, W. Griffith, D.B. Barr, L.L. Needham, and R.A. Fenske. 2005. Comparison of organophosphorous pesticide metabolite levels in single and multiple daily urine samples collected from preschool children in Washington State. J. Expo. Anal. Environ. Epidemiol. 15(2):164-171. Landi, M.T., and N. Caporaso. 1997. Sample collection, processing and storage. Pp. 223-236 in Application of Biomarkers in Cancer Epidemiology, P. Toniolo, P. Boffetta, and D.E.G. Shuker, eds. IARC Scientific Publications 142. Lyon: IARC. Lasker, R.D., and E.S. Weiss. 2003. Broadening participation in community problem-solving: A multidisciplinary model to support collaborative practice and research. J. Urban Health 80(1):14-60. Liira, J., G. Johanson, and V. Riihimäki. 1990. Dose-dependent kinetics of inhaled methyl ethyl ketone in man. Toxicol. Lett. 50(2-3):195-201. Lof, A., A. Johanson, M. Rannug, and M. Warholm. 2000. Glutathione transferase T1 phenotype affects the toxicokinetics of inhaled methyl chloride in human volunteers. Pharmacogenetics 10(7):645-653.
OCR for page 130
Human Biomonitoring for Environmental Chemicals Lu, C., R. Bravo, L.M. Caltabiano, R.M. Irish, G. Weersekera, and D.B. Barr. 2005. The presence of dialkylphosphates in fresh fruit juices: Implications for organophosphorus pesticide exposure and risk assessments. J. Toxicol. Environ. Health A 68(3):209-227. Marmot, M.G., G.D. Smith, S. Stansfeld, C. Patel, F. North, J. Head, I. White, E. Brunner, and A. Feeney. 1991. Health inequalities among British civil servants: The Whitehall II study. Lancet 337(8754):1378-1393. Metcalf, S.W., and K.G. Orloff. 2004. Biomarkers of exposure in community settings. J. Toxicol. Environ. Health Part A 67(8-9):715-726. Mizutani, T. 2003. PM frequencies of major CYPs in Asians and Caucasians. Drug Metab. Rev. 35(2-3):99-106. Nebert, D.W. 2000. Suggestions for the nomenclature of human alleles: Relevance to ecogenetics, pharmacogenetics and molecular epidemiology. Pharmacogenetics 10(4): 279-290. Needham, L.L., and R.Y. Wang. 2002. Analytic considerations for measuring environmental chemicals in breast milk. Environ Health Perspect. 110(6):A317-A324. Needham, L.L., H. Ozkaynak, R.M. Whyatt, D.B. Barr, R.Y. Wang, L. Naeher, G. Akland, T. Bahadori, A. Bradman, R. Fortmann, L.J. Liu, M. Morandi, M.K. O’Rourke, K. Thomas, J. Quackenboss, P.B. Ryan, and V. Zartarian. 2005. Exposure assessment in the national Children’s Study: Introduction. Environ. Health Perspect. 113(8):1076-1082. Nihlén, A., and G. Johanson. 1999. Physiologically based toxicokinetic modeling of inhaled ethyl tertiary-butyl ether in humans. Toxicol. Sci. 51(2):184-194. NRC (National Research Council). 1987. Biologic markers in environmental health research. Environ Health Perspect. 74:3-9. NRC (National Research Council). 1991. Monitoring Human Tissues for Toxic Substances. Washington, DC: National Academy Press. NRC (National Research Council). 1993. Pesticides in the Diets of Infants and Children. Washington, DC: National Academies Press. Özkaynak, H., R.M. Whyatt, L.L. Needham, G. Akland, and J. Quackenboss. 2005. Exposure assessment implications for the design and implementation of the National Children’s Study. Environ. Health Perspect. 113(8):1108-1115. Pflugh, K.K., J.A. Shaw, and B.B. Johnson. 1994. Establishing Dialogue: Planning for Successful Environmental Management; A Guide to Effective Communication Planning. Trenton, NJ: New Jersey Department of Environmental Protection and Energy. Risk Policy Report. 2001. Chemicals in breast milk could act as indicator, shape regulatory policy. Inside EPA’s-Risk Policy Report 8(2):24-26. Rivers, P.A., A. Dobalian, and F.A. Germinario. 2005. A review and analysis of the clinical laboratory improvement amendment of 1988: Compliance plans and enforcement policies. Health Care Manage Rev. 30(2):93-102. Rohrmann, B. 1992. The evaluation of risk communication effectiveness. Acta Psychol. 81(2):169-192. Rohrmann, B. 2000. A socio-psychological model for analyzing risk communication processes. Australasian J. Disaster Trauma Studies [online]. Available: http://www.massey.ac.nz/~trauma/issues/2000-2/rohrmann.htm [accessed April 21, 2006]. Santos, S.L., and C. Chess. 2003. Evaluating citizen advisory boards: The importance of theory and participant-based criteria and practical implications. Risk Anal. 23(2):269-280. Schulte, P.A. 2005. Application and Design Issues in Human Biomonitoring. Presentation at the Second Meeting on Human Biomonitoring of Environmental Toxicants, April 28, 2005, Washington, DC.
OCR for page 131
Human Biomonitoring for Environmental Chemicals Schulte, P.A., and M.H. Sweeney. 1995. Ethical considerations, confidentiality issues, rights of human subjects, and uses of monitoring data in research and regulation. Environ. Health Perspect. 103(suppl. 3):69-74. Schulte, P.A., and G. Talaska. 1995. Validity criteria for the use of biological markers of exposure to chemical agents in environmental epidemiology. Toxicology 101(1-2): 73-88. Schulte, P.A., D. Hunter, and N. Rothman. 1997. Ethical and social issues in the use of biomarkers in epidemiological research. IARC Sci Publ. (142):313-318. Silva, M.J., D.B. Barr, J.A. Reidy, N.A. Malek, C.C. Hodge, S.P. Caudill, J.W. Brock, L.L. Needham, and A.M. Calafat. 2004. Urinary levels of seven phthalate metabolites in the U.S. population from the National Health and Nutrition Examination Study (NHANES) 1999-2000. Environ. Health Perspect. 112(3): 331-338. Soskolne, C.L. 1997. Ethical, social, and legal issues surrounding studies of susceptible populations and individuals. Environ. Health Perspect. 105(suppl. 4):837-841. Stern, P.C., and H.V. Fineberg, eds. 1996. Understanding Risk: Informing Decisions in a Democratic Society. Washington, DC: National Academy Press. Stinson, B.L., and J.R. Ehrmann. 1998. Joint fact-finding in consensus-building processes. Inside EPA’s-Risk Policy Report 5(11):36-38. Thompson, K.M., and D.L. Bloom. 2000. Communication of risk assessment information to risk managers. J. Risk Res. 3(4):333-352. Tinker, T. 1994. Case Studies of Applied Evaluation for Health Risk Communication: Workshop Proceedings, June 9, Annapolis, MD. U.S. Department of Health and Human Services, Washington, DC. Veach, P.M., D.M. Bartels, and B.S. LeRoy. 2001. Ethical and professional challenges posed by patients with genetic concerns: A report of focus group discussions with genetic counselors, physicians, and nurses. J. Genet. Couns. 10(2):97-119. Warholm, M., A.K. Alexandrie, J. Hogberg, K. Sigvardsson, A. Rannug. 1994. Polymorphic distribution or glutathione transferase activity with methyl chloride in human blood. Pharmacogenetics 4(6):307-311. Webler, T. 1998. Beyond science: Deliberation and analysis in public decision making. Human Ecol. Rev. 5(1):61-64. Weinstein, N.D., and P.M. Sandman. 1993. Some criteria for evaluating risk messages. Risk Anal. 13(1):103-114. Wessels, D., D.B. Barr, and P. Mendola. 2003. Use of biomarkers to indicate exposure of children to organophosphate pesticides: Implications for a longitudinal study of children’s environmental health. Environ. Health Perspect. 111(16):1939-1946. Winn, D.M., M.E. Reichman, and E. Gunter. 1990. Epidemiologic issues in the design and use of biologic specimen banks. Epidemiol. Rev. 12:56-70.
Representative terms from entire chapter: