Incorporating children's toxicokinetics into a risk framework.

Children's responses to environmental toxicants will be affected by the way in which their systems absorb, distribute, metabolize, and excrete chemicals. These toxicokinetic factors vary during development, from in utero where maternal and placental processes play a large role, to the neonate in which emerging metabolism and clearance pathways are key determinants. Toxicokinetic differences between neonates and adults lead to the potential for internal dosimetry differences and increased or decreased risk, depending on the mechanisms for toxicity and clearance of a given chemical. This article raises a number of questions that need to be addressed when conducting a toxicokinetic analysis of in utero or childhood exposures. These questions are organized into a proposed framework for conducting the assessment that involves problem formulation (identification of early life stage toxicokinetic factors and chemical-specific factors that may raise questions/concerns for children); data analysis (development of analytic approach, construction of child/adult or child/animal dosimetry comparisons); and risk characterization (evaluation of how children's toxicokinetic analysis can be used to decrease uncertainties in the risk assessment). The proposed approach provides a range of analytical options, from qualitative to quantitative, for assessing children's dosimetry. Further, it provides background information on a variety of toxicokinetic factors that can vary as a function of developmental stage. For example, the ontology of metabolizing systems is described via reference to pediatric studies involving therapeutic drugs and evidence from in vitro enzyme studies. This type of resource information is intended to help the assessor begin to address the issues raised in this paper. Key words: children, dosimetry, risk assessment, toxicokinetics. Environ Health Perspect 112:272-283 (2003). doi:10.1289/ehp.6013 available via http://dx.doi.org/[Online 25 November 2003]

Children Can Be a Tough Group to Figure Out

Perhaps the risk assessment community has difficulty embracing the arena of children's risk because of the imposing task. Children are difficult to study for a number of reasons, not the least of which is a practical one: it is not ethically feasible to introduce chemicals of environmental concern, even at trace amounts, into infants and children. Thus, there is very little toxicokinetic data (i.e., pharmacokinetic data for environmental toxicants) in this age group. One can gather data from "natural experiments" in which children are exposed inadvertently to a pesticide or airborne chemical. The resulting biomarker data may be useful for investigators in assessing the extent of exposure, but these are not the well-controlled toxicokinetic data one needs for building a predictive children's model. Another equally daunting problem is that children are highly diverse, stretching in definition (for the purposes of this framework) from fetal through adolescent stages and beyond. Generalizations and defaults are not possible for such a sweeping range of development. Even within a narrow age range, there can be considerable variability, given the rapid and variable rate of development in early life. Thus, improving children's risk assessments through toxicokinetics is a worthwhile endeavor but one that will be hampered by data gaps and uncertainties.

Our goal in writing this article on toxicokinetic issues is to outline the toxicokinetic questions that need to be addressed if this area of study is to contribute to a children's risk assessment. We also hope to provide the reader with some background and resources that can help shed light on these questions by showing where this information can fit into a framework for children's risk assessment. We recognize that such risk assessments also need to address critical exposure and toxicodynamic issues, and that in some ways these areas overlap with toxicokinetics (Adams et al. 2000). We point out those interfaces, but for this discussion we will keep toxicokinetics as a separate consideration in the children's risk assessment process.

Toxicokinetics as a key element in children's risk assessments. The importance of toxicokinetics in risk assessment has been increasing as we have learned more about how toxicokinetic processes (especially chemical metabolism) are involved in mechanisms of toxicity and how these processes can differ markedly across species. In numerous cases, toxicokinetic analyses have allowed replacement of the traditional dose metric--applied dose per body weight per day--by a more relevant internal dose metric that facilitates extrapolation of animal dose-response data to humans (Andersen et al. 1987; Bois et al. 1990; Hattis et al. 1993; Ran and Ginsberg 1997). Most commonly this is accomplished with physiologically based pharmacokinetic (PBPK) models, which simulate chemical uptake, distribution, metabolism, and excretion in both animals and humans.

Although progress in toxicokinetics has removed some of the uncertainty in cross-species extrapolations in risk assessment for adults, these principles have yet to be applied in a systematic manner to the risk assessment of children. This is a critical need, given that children's toxicokinetics differ from adults in a number of ways: smaller body size; different ratios of fat, muscle, and water; higher breathing and metabolic rates per body weight; and immaturity of clearance systems and enzymatic reactions (Anderson et al. 1997; Besunder et al. 1988; Ginsberg et al. 2002; Kearns and Reed 1989; Renwick 1998). Another obvious difference is that children are more diverse, undergoing a developmental program of growth and maturation that continuously alters how chemicals are processed and cleared. Thus, incorporating children's toxicokinetics into risk assessment is complicated by the need to consider many developmental stages, ranging from in utero to adolescence, and by the extensive variability that can occur within each age group.

Pharmacokinetic differences between children and adults with respect to the clearance of therapeutic drugs have been recognized for years. These differences have spawned numerous clinical pharmacokinetic studies for the purpose of better titrating drug dosage to a particular age or body size (Anderson et al. 1997; Ginsberg et al. 2002; Morselli 1989; Renwick et al. 2000). The focus on children as a pharmacological receptor has not been matched by a similar focus on children as a toxicant receptor. In large part this is because of the lack of pediatric toxicokinetic studies, as mentioned above. This leaves us with a relatively rich pharmacokinetic database for children but a nearly empty toxicokinetic database. Challenges facing the children's risk assessment framework include how to combine basic information on children's physiologic development with what can be learned from children's pharmacokinetic data sets and how to develop an analytic process that allows us to compare internal doses of environmental toxicants between adults and children.

Proposed building blocks for a children's toxicokinetic assessment. In crafting a general framework for conducting children's risk assessments, a 3-phase approach has been recommended that involves problem formulation, data analysis, and risk characterization. [See companion article by Daston et al. (2004)]. An overview of this approach is provided in Figure 1, which serves to organize the information on children's toxicokinetics into a stepwise analytical approach.

[FIGURE 1 OMITTED]

Phase I: Toxicokinetic Questions in Problem Formulation

A risk assessment of children will likely begin with a number of questions regarding age groups and exposures, toxicokinetic handling of the chemical(s), and whether there are susceptible periods or unique toxic effects (toxicodynamic issues). These questions and the preliminary answers developed in problem formulation help to scope the remainder of the analysis. A major toxicokinetic question likely to arise in problem formulation is "Can early life stages be considered as part of the overall human variability distribution and thus be within the 3.2-fold toxicokinetic uncertainty factor"? This broad question can be broken into the following more-defined areas that should help the risk assessor better evaluate the issues of children's toxicokinetics:

* What types of data, resources, and analytic approaches are needed to determine whether a) children of certain ages experience higher exposures because of higher ventilation rates, cardiac outputs, or greater absorption of chemical(s); b) children of certain ages experience a higher internal dose of parent compound or active metabolite (per unit of administered dose) because of distributional, metabolism, or elimination factors; and c) children of certain ages process the chemical(s) by novel toxicokinetic pathways in order to generate metabolite profiles that are qualitatively different from those of adults?

In this article we discuss these broad, scoping toxicokinetic questions to provide background information and to help us consider what kinds of resources and analyses will be needed for their resolution.

* Does children's toxicokinetics warrant an uncertainty factor that differs from the current practice of allowing an approximately 3.2-fold factor for intersubject variability in toxicokinetics?

Noncancer risk assessments are concerned with threshold effects, with the public health exposure level set well below the observed threshold seen in animal or high-exposure human studies [U.S. EPA (U.S. Environmental Protection Agency) 1989]. A series of uncertainty factors are used to attempt to conservatively cover what we do not know about risks; the uncertainty factors lower the allowable exposure level from the animal or human no observable adverse effect level (NOAEL) or benchmark dose. Prominent among these is the interindividual variability factor, typically a 10fold factor. This factor results from a general recognition that the animal species or worker cohort from which the NOAEL is determined rarely, if ever, represents the range of diversity present in the human population at large. This uncertainty factor should encompass variability from a wide array of sources: inherited traits and genetic polymorphisms, gender and hormonal differences, lifestyle factors such as amount of stress, exercise, tobacco and alcohol use, obesity, nutritional status, and finally, disease conditions, some of which can alter the handling of xenobiotics. Related to this last factor is the ingestion of prescription and over-the-counter medications, many of which can alter the pharmacokinetics of other chemicals. Another circumstance that this intersubject uncertainty factor is asked to cover is age, encompassing a range from prenatal to elderly.

As mentioned above, the noncancer uncertainty factor is 10, which can be understood to consist of a half-log factor (3.16-fold) for toxicokinetic variability and a similar half-log factor for pharmacodynamic variability (Renwick 1998). Thus, the interindividual variability in toxicokinetics created by the various genetic, lifestyle, physiologic state, and age factors must fit within a 3.2-fold factor for the default to be adequate. In other words, one must determine whether the upper or lower ends of the children's distribution (e.g., low-end metabolizers) are captured within the bounds created by the adult central tendency value plus or minus a 3.2-fold uncertainty factor. If not, then a special children's uncertainty factor or other more quantitative approaches to characterize variability might be needed.

It is important to point out that a similar uncertainty factor does not exist for cancer risk assessment. These assessments are generally considered conservative because of the nature of the linear low-dose modeling often used and because interspecies scaling of the unit risk factor employs a generic body weight function that increases human potency relative to that of rodents (Andersen et al. 1995; U.S. EPA 1992). However, the lack of any special consideration of children's risk in cancer assessments leaves open the question of whether the derived unit risks are appropriate for all age groups. In fact, one recent cancer risk assessment did adjust the unit risk for child-specific factors, but this adjustment was not based on toxicokinetics (U.S. EPA 2000).

Thus, in the absence of precedents for using a children's toxicokinetic analysis to adjust cancer or noncancer assessments, a framework is needed for doing this in a rational and consistent manner. The hope is that such a process will foster an understanding of the size of adult/child differences in key toxicokinetic pathways relative to the amount of interindividual variability we have come to expect in noncancer assessments.

Phase II: Data Analysis and Development of Risk Assessment Approach

In this phase, chemical-specific data regarding the fate of the chemical(s) being analyzed are combined with children's data relating to the developmental profile of toxicokinetic systems (from in utero periods through adolescence). This will foster an understanding of how children of various ages are likely to handle the chemical(s). Obviously, the best case is to have metabolism and disposition data for the chemical(s) in children. However, it will be extremely rare to find such toxicokinetic data. Following are some approaches to consider for filling these data gaps:

Use of toxicokinetic data for the chemical(s) under analysis to characterize the major fate and transport mechanisms. Typically, animal data, and in some cases, adult human data also, will be available for characterizing the chemical's fate and disposition. Once the major toxicokinetic factors governing chemical fate are known, then the approaches that follow can be evaluated for utility and feasibility.

Use of surrogate chemicals (typically therapeutic agents) for which pharmacokinetic data in children exist. Even if the target chemical(s) has not been evaluated in children, a chemical that is similarly processed may have been tested. Information for surrogate chemicals can help delineate the maturation of key pathways and how the target chemical will be handled at certain developmental stages. This section describes the kinds of data sets and resources available for this part of the phase II assessment.

Over the past several years children's pharmacokinetic databases have been developed in which therapeutic drugs tested in both children and adults are identified and key pharmacokinetic parameters are compared across ages (Ginsberg et al. 2002; Renwick 1998; Renwick et al. 2000). The metabolism and clearance pathways of many of the drugs are known, making them useful indicators for particular pathways. For example, dextromerhorphan and debrisoquine are known substrates for a particular cytochrome P450 (CYP) enzyme, CYP2D6; trimethadione, chlorzoxazone, and halothane are markers for CYP2E1 activity; morphine is predominantly processed by glucuronidation; and a host of antibiotics are not extensively metabolized but are mostly excreted unchanged by the kidneys (Bertz and Granneman 1997; Dollery 2000; Kurata et al. 1998; Tanaka 1998).

A well-known example of child-adult differences in metabolic processing is caffeine, a process in which an initial N-demethylation reaction is catalyzed by CYP1A2, and a secondary N-acetylation step is catalyzed by N-acetyhransferase (Welfare et al. 2000). The CYP1A2 reaction is the primary factor governing the half-life of this drug. Caffeine's half-life in newborns is 14-fold higher than in adults, which is likely the result of the immaturity of CYP1A2 at this age (Dome et al. 2001; Ginsberg et al. 2002; Parsons et al. 1976; Parsons and Neims 1978). The metabolic clearance of caffeine and another CYP1A2. substrate, theophylline, becomes more like the adult rate by 2-6 months of age, after which time the half-life becomes significantly shorter than in that in adults (Dorne et al. 2001; Ginsberg et al. 2002). This suggests that once CYP1A2 function approaches adult levels, the greater size and blood flow to the liver (per body weight) in young children can lead to greater enzymatic clearance of CYP substrates (Gibbs et al. 1997). However, this also means that chemicals activated to toxicants by this pathway [e.g., aromatic amines, polycyclic aromatic hydrocarbons (PAHs), aflatoxin B1] may produce more active metabolite in this age group compared with adults. Unfortunately, there are no data that can be used to directly evaluate this possibility.

This drug has also been used to show that the N-acetyltransferase step is also deficient in early life. Once caffeine is demethylated by CYP1A2, it can be N-acetylated to form 5-acetylamino-6-formylamino-3-methyluracil (AFMU) or excreted in urine in the nonacetylated form (1-methylxanthine). The ratio of AFMU to 1-methylxanthine has been used to phenotype the N-acetyltransferase trait in adults and children. In children the ratio of acetylated to nonacetylated metabolite is low in newborns through 2 months of age (12% of adult ratio), but then rises to 65% of the adult ratio during the 2- to 6-month interval (Pariente-Khayat et al. 1991). This agrees with data showing that most newborns are slow acetylators but that beyond 100 days, the underlying genetic polymorphism (fast vs. slow) becomes evident (Pans et al. 1989; Szorady et al. 1987). Such data with caffeine are examples of the information that can be obtained from pharmacokinetic data for therapeutic agents.

Table 1 is a compilation of information obtained from in viva pharmacokinetic analyses of drugs with in vitro analyses of enzyme levels from blood or liver samples. The combination of the two types of information for a given clearance pathway can provide a strong indication of how the pathway's function develops in the postnatal period. In viva pharmacokinetic data have been analyzed across chemicals that share a common mode of elimination in order to develop a more complete evaluation of the function of specific pathways (Ginsberg et al. 2002; Renwick et al. 2000). Table 1 shows data for six CYP enzymes, several phase II conjugation pathways, renal and biliary clearance, other metabolic functions (e.g., epoxide hydrolase, alcohol dehydrogenase), and two esterases involved in the detoxification of organophosphate (OP) pesticides. This compilation of pharmacokinetic data by pathway and age group indicates a fairly consistent pattern, that is, premature neonates, full-term neonates, and infants up to 6 months of age tend to have less metabolic and clearance capacity than adults. The two exceptions in the chart are for enzymes expressed primarily in the fetal and early postnatal period: CYP3A7 and glutathione S-transferase (GST) pi. These fetal forms are replaced during the first year of life by corresponding (but not enzymatically equivalent) adult forms. Beyond 6 months, many CYP enzymes are sufficiently active that clearance in viva is actually greater than that in adults. This appears to be due to the greater liver size and blood flow in children compared with adults (Gibbs et al. 1997). Table 1 provides further evidence for this phenomenon in that the in viva data indicate greater clearance capacity than suggested by the in vitro protein levels or enzyme activities for CYP1A2, CYP2E1, CYP2C9/19, CYP2D6, and CYP3A4, particularly at 6 months of age.

These overall patterns are illustrated for a specific CYP, CYP3A4, as evidenced by therapeutic drugs that depend upon this CYP for clearance (Figure 2) (Ginsberg et al. 2002). CYP3A4 is the major CYP in adult human liver, but its function is evidently deficient in early life. This pattern appears to be widely applicable; an analysis across 40 therapeutic drugs involving 11 different metabolic/excretory mechanisms shows a similar shift from immaturity in the earliest age categories to evidence of drug half-lives shorter than adults beyond 6 months of age (Figure 3).

[FIGURES 2-3 OMITTED]

The period of shorter half-lives in children relative to adults may represent a time in which there is faster removal of parent compound but greater formation of metabolites, which can be a concern if chemical metabolism leads to more toxic moieties. However, this may not lead to greater toxicity if the metabolites are also cleared more readily in children. Prediction of the health implications of shifts in xenobiotic metabolism must be chemical specific and must consider both parent compound and toxic metabolites.

The period of accelerated clearance is in contrast to the earlier periods in which hepatic metabolism is slower than that in adults, leading to the potential for prolonged retention and higher levels of parent compound. This situation may be compounded by the slower renal and biliary function at these times. In terms of detoxification systems, Table 1 shows that epoxide hydrolase is active at birth but apparently only at 50% of adult function. Although the data are very limited, it appears that two forms of GST may be deficient (40-60% of adult levels) in early life. These data may have risk implications, given that GSTmu is critical to epoxide conjugation with glutathione and that epoxide hydrolase also appears to be slow in this age group. Thus, the newborn and very young infant may have a relative deficit in detoxifying epoxides, although compensating mechanisms (e.g., other GSTs) may exist. It should be noted that further research into the developmental profile of these detoxification systems is an important research need.

Another potentially important deficit in children is glucuronidation during the first 2 months of life. Because of the relative deficiency of glucuronide and high bilirubin levels (an endogenous glucuronidation substrate), it is possible that there will be greater internal exposure early in life to xenobiotics (or their metabolites) that require glucuronide conjugation. This is the case with the antibacterial drug chloramphenicol, in which a relative lack of glucuronidation contributed to its accumulation and toxicity (anemia) in infants (Mulhall et al. 1983; Vest 1965). In contrast, some xenobiotics can undergo conjugation with alternative cofactors, which allows a shift in metabolic profile if the primary pathway is compromised or immature. This occurs with acetaminophen, where sulfation predominates early in life until glucuronidation takes over (Levy et al. 1975).

The findings described above are consistent with other pediatric pharmacokinetic summaries (Dome et al. 2001; Renwick 1998, 2000) in which clearance in childhood is slower than in adults for a variety of drugs and age groups. There are also cases in which clearance is more rapid, particularly when the pediatric group was at least several months of age (Renwick 1998). This type of age-specific and pathway-specific pharmacokinetics information in children should prove useful in predicting how children (particularly neonates) may differ from adults in internal dosimetry and ultimate risk.

Simply knowing the function of particular pathways may not be sufficient to predict in vivo handling of a xenobiotic at a particular age. Numerous toxicokinetic factors are involved in chemical processing and clearance, including partitioning into body compartments, protein binding, respiration, and organ flows. Therefore, a more comprehensive analysis, such as PBPK modeling, may be needed to integrate the various factors at work and predict xenobiotic fate in children. The information provided in Table 1, combined with basic physiologic information, may make PBPK modeling of children more feasible. Where this is pursued, the uncertainty and data gaps surrounding input parameters for children must be recognized and made transparent.

Although characterization of pathway function on the basis of indicator drugs is a useful approach, it carries the uncertainty that a drug may be processed by several clearance pathways. If one pathway is deficient at a certain age, but another is more functional, then the overall half-life or clearance rate may not be affected. Instead, parent compound may be shunted from the less active to the more active pathway, leading to a shift in metabolite profile. The assignment of chemicals to particular pathways is based on the fate of the majority (generally 60% or more) of the administered dose in adult humans, as ascertained from the literature. Shifts in metabolic processing in early life would tend to obscure child versus adult clearance differences based on overall half-life or blood clearance. Thus, in some cases, the data in Table 1 and Figures 2 and 3 may be an underestimate of the degree of child--adult difference that might actually exist for a given pathway.

Another caveat for some of the in vivo data sets cited in Table 1 and Figures 2 and 3 is that they represent clinical pharmacokinetic trials in children who are not in full health. The data sets summarized in Ginsberg et al. (2002) were screened not only for sufficient numbers of subjects per age group but also for the health of the subjects on test. Data for children who were critically ill or severely compromised, especially with hepatic or renal conditions that would affect xenobiotic handling, were excluded from the database. However, it is possible that the clinical state of the children being tested may have affected factors such as volume of distribution or other pharmacokinetic factors in certain data sets. The overall concordance between in vitro and in vivo data shown in Table 1 suggests that this factor is not a systematic issue here.

Use of in vitro data to ascertain how toxicokinetic pathways change during development. The liver bank data describing CYP protein levels and enzyme activities, as well as the serum esterase data shown in Table 1, can be used as indicators of pathway function at particular ages. These data are made more reliable for risk assessment when there are corroborating in vivo half-life or clearance data for indicator chemicals. An example is CYP2E1 function, in which in vitro liver bank studies (Figure 4A, B) generally agree with in vivo drug metabolism data for the CYP2E1 substrate trimethadione (Figure 5) (Nakamura et al. 1998; Tanaka 1998; Vieira et al. 1996). These developmental profiles can be essential for estimating the degree to which CYP2E1-mediated metabolic activation may occur for a wide variety of environmental toxicants (e.g., benzene, chlorinated solvents, ethanol).

[FIGURES 4-5 OMITTED]

Use of developmental information to evaluate how toxicokinetics can affect susceptibility during the in utero period. The maternal-fetal environment presents unique toxicokinetic considerations. The developing organism is exposed as part of the maternal system via placental transport. Although maternal factors (distribution, metabolism, excretion) govern fetal exposure, exposure to the fetus can be quantitatively or qualitatively different from that in the mother. This may result from chemical accumulation in the fetal compartment (NRC (National Research Council) 2000) or from in situ metabolism that causes fetal exposure to metabolites different from that experienced in the maternal system. The following sections are a summary of the key toxicokinetic factors during in utero development.

Placental transport. Between the maternal system and conceptus is a specialized set of membranes that chemicals must cross. The placenta has its origins in the conceptus, with its unique anatomical and functional features influencing chemical transfer to the developing conceptus. This set of membranes undergoes considerable change with development and also exhibits substantial species differences. As pregnancy progresses, there is closer contact between the fetal blood and the syncytiotrophoblast--the placental cell layer most important to placental function and maternal--embryo/ fetal exchange (Garbis-Berkvens and Peters 1987). This leads to greater exchange of nutrients and chemicals across the placenta as gestation proceeds. On the macroscopic level, the effects of gestational age can be exemplified by comparing the tremendous change in the ratio of placental/fetal weight (ratio equals 4 at 10 weeks and 0.2 at 40 weeks) (Hytten and Leitch 1971).

Any substance in the maternal circulation can be transferred, to some exent, across the placenta unless it is metabolized/eliminated before or during its placental passage (Garbis-Berkvens and Peters 1987; Slikker 1987). A useful method for characterizing placental transfer across chemicals and species is the dually perfused placenta. This is an in vitro preparation in which placental transfer is defined as a clearance ratio. Studies of nonmetabolized model compounds indicate decreases in placental transfer as the molecular weight of the model compounds increases (Illsley et al. 1985). Under these very controlled conditions without maternal or fetal involvement, differences in placental transfer between species are evident, with sheep being more different than guinea pig from human. These types of functional assessments help to predict placental transfer and fetal exposure after maternal dosing.

Additional factors that influence the placental transfer of chemicals are uterine/placental blood flow, placental permeability, and placental metabolism (Juchau 1980a, 1980b; Mihaly and Morgan 1983; Miller et al. 1976; Mirkin and Singh 1976; Waddell and Marlowe 1981). These factors are not static during pregnancy but may change as gestation progresses.

Chemical delivery to the developing conceptus relies primarily on blood flow to the placenta. Although chemicals may transfer from mother to fetus via the amniotic fluid after crossing the amnion, the majority of agents gain access to the conceptus via placental passage (Nau and Liddiard 1978). In addition to the changes in placental blood flow that occur during gestation, changes in blood flow as a result of chemical exposure must also be considered. Although experimentally induced changes in placental blood flow can alter normal development of the conceptus (Barr and Brent 1978; Greiss and Gobble 1967; Millicovsky and DeSesso 1980), the effect of such perturbations on placental transport of xenobiotics needs to be evaluated.

Placental permeability to a chemical is influenced both by placental characteristics (e.g., thickness, surface area, carrier systems, and lipid/protein content of the membranes) and also by characteristics of the chemical agent (e.g., degree of ionization, lipid solubility, protein binding, and molecular weight) (Mihaly and Morgan 1983; Mirkin and Singh 1976; Welsch 1982). The rate of placental transfer is rapid for nonionized, lipid-soluble chemicals of low molecular weight (less than 1,000 daltons) and is largely controlled by placental blood flow (Mihaly and Morgan 1983; Mirkin and Singh 1976). However, charged molecules such as tubocurarine can also enter the fetus (Kivalo and Saarikoski 1972, 1976). Similarily, chemicals highly ionized at normal blood pH, such as the salicylates, readily cross the placenta (Wilson et al. 1977). The question is thus not whether a compound crosses the placenta but rather at what rate.

Maternal considerations. A chemical must pass through and may interact with several anatomical compartments on its journey from the site of maternal exposure to the organ systems of the developing conceptus. Maternal factors act to either enhance or diminish the concentration of an active chemical in the conceptus. Maternal detoxification decreases the amount of parent compound available for placental transport, whereas maternal bioactivation may make more active chemical available. The entire set of maternal toxicokinetic factors (e.g., absorption, distribution, serum binding, and elimination) also influences the concentration of active agent at the target site. Because of the physiologic changes that occur during pregnancy, the influence of these maternal factors on chemical delivery may also change during gestation (Bogaert and Thiery 1983; Cummings 1983; Juchau 1995; Juchau and Faustman-Watts 1983; Noschel et al. 1980). These issues have been reviewed (Brock-Utne et al. 1980; Juchau and Faustman-Watts 1983; Krauer et al. 1980; Levy 1981; Slikker and Miller 1994).

For chemicals that bind avidly to plasma proteins, this factor can retard placental transfer. In general, only the free drug crosses the membranes of the placenta (Krauer et al. 1980; Welsch 1982). Protein binding is usually reversible, and there are a finite number of binding sites. Thus, binding is saturable, and equilibrium may be described by the law of mass action (Krauer et al. 1980; Miller et al. 1976). As long as binding is reversible, it does not prevent the chemical from crossing membranes but only slows the rate at which the transfer occurs (Levine 1973).

Placental biotransformation. This may be the most critical factor in influencing the delivery of chemicals to the developing conceptus. Placental biotransformation of a chemical prior to fetal delivery may dramatically alter the chemical profile in the conceptus from that in the mother (Slikker et al. 1982). Equilibrium factors, which influence the rate of placental transfer, can result in quantitative differences of exposure. Placental metabolism, however, can qualitatively alter the exposure of the conceptus to potentially toxic chemicals. Placental metabolism is less well characterized than hepatic metabolism, but existing data suggest that the placenta has considerably less metabolic capacity than adult liver (Juchau 1980b; Mirkin and Singh 1976). A listing of some of the human placental xenobiotic and hormone-metabolizing enzymes or isoenzymes is presented in Table 2.

Embryo/fetal considerations. As with most organ systems, the various chemical-metabolizing systems undergo quantitative if not qualitative changes during development. Chemical biotransformation by the developing conceptus has been extensively reviewed (Dutton 1978; Dutton and Leakey 1981; Dvorchik 1981; Eltom et al. 1993; Juchau and Faustman-Watts 1983; Leakey 1983; Nan and Neubert 1978; Neims et al. 1976; Pelkonen 1977, 1980; Rane and Tomson 1980; Slikker 1987, 1994; Slikker and Miller 1994). Despite that data have been collected using a variety of techniques and some data gaps exist because of technical or ethical reasons, several general conclusions may be drawn from the literature.

* During prenatal development, the activities of most enzymes that catalyze phases I and II reactions are lower than those in adults.

* As in the adult, the conceptus exhibits substrate specificity in its ability to metabolize chemicals, suggesting the existence of several sets of enzymes or isozymes, which may or may not be the same as those in the adult.

* These enzyme systems may be inhibited or induced by maternal pretreatment with a variety of chemicals.

* Enzyme activity generally increases with gestational age.

* The ontogeny of each enzyme may be different, and the controlling mechanisms of maturation of enzyme activity are incompletely understood.

* Prenatal human and nonhuman primates exhibit higher levels of many metabolizing enzymes (especially P450s) than do commonly used laboratory species.

* As in the adult, the liver of the conceptus appears to have the greatest capacity for chemical metabolism. The fetal adrenal, kidney, lung and brain also exhibit metabolic capabilities.

Table 3 summarizes recent data concerning some CYP isozymes and their occurrence in human fetal tissues.

Fetal distribution and elimination. The majority of chemicals entering the fetal circulation do so via the umbilical vein after passage through the placenta. A portion of the blood flow entering the liver of the conceptus is shunted via the ductus venosus directly to the inferior vena cava and to the heart for total body distribution (Rudolph and Heymann 1967). The remaining umbilical flow enters hepatic tissue and exits to the vena cava via the portal vein (Dawes 1968). There is potential for a significant first-pass effect from chemical passing through the fetal liver prior to other sites; however, a substantial fraction passes directly to the remaining tissues (Power and Longo 1975). As in the adult, approximately 16% of the fetal cardiac output is directed toward the fetal brain (Behrman et al. 1970).

Just as placental transfer of chemicals is the predominate pathway from the maternal system to the conceptus, placental transfer is also the predominate route for embryo/fetal elimination of chemicals. The same toxicokinetic rules apply for fetal-to-maternal transport across the placenta: chemicals that are nonionized and lipid soluble will diffuse across the placenta according to the concentration gradient from conceptus to mother. However, if a chemical has been conjugated by the fetus (e.g., glucuronidation, sulfation) or otherwise metabolized to a more polar form, the rate of return to the maternal circulation will be slower than that for the parent compound (Dancis et al. 1958; Goebelsmann et al. 1968, 1972; Levitz et al. 1960). PBPK models to predict the fate of toxicants in fetal tissues as a function of development are emerging (Kim et al. 1996; Sandberg et al. 1996).

Use of animal toxicokinetic information to evaluate how development can affect internal dosimetry. Laboratory animals can be useful in a number of respects in providing toxicokinetic data and related information for children's risk assessments of environmental pollutants (Harroff et al. 1997). The discussion here will be limited primarily to mice and rats, as most toxicokinetic and toxicology studies employ these two rodent species.

Diversity is one of the more daunting problems anticipated when studying children because of the numerous variability factors described earlier. Intersubject variability is substantially lower in homogeneous rodent populations provided by major animal suppliers. Uniform groups of animals of the same established genetic and husbandry backgrounds can be maintained under defined and carefully controlled conditions. Therefore, it is possible with animal studies to control more variables and to better focus on age-dependent differences in chemical metabolism and disposition.

Animal toxicokinetics and susceptibility to xenobiotics. Series of studies of OPs in rats have clearly shown that the relatively low detoxification capacity of weanlings places them at increased risk of acute, high-dose OP poisoning. Benke and Murphy (1975) concluded that the lower acute toxicity of parathion and methyl parathion in adult rats was due to the adults' greater detoxification capacity. More recent studies have confirmed and extended these findings. The maturational profiles of P450-catalyzed desulfuration (activation) and dearylation (inactivation) as well as carboxylesterase and arylesterase (inactivation) activities have been characterized in the liver and/or blood of rats (Atterberry et al. 1997; Moser et al. 1998). Despite very low carhoxylesterase and arylesterase activities, rat fetuses and sucklings appear to be protected from chlorpyrifos because of limited transfer of the compound from the mother (Mattsson et al. 2000). However, this protective effect would not be expected if neonates were exposed directly (i.e., not via the mother) to OPs. On the basis of somewhat anecdotal information of groups of people poisoned by these pesticides, the suggestion is that young children (0-4 years of age) are at increased risk (Diggory et al. 1977; Zwiener and Ginsburg 1988). However, the exposure information is not detailed suffciently in these cases to confirm the sensitivity of any particular person or age group. Ecobichon and Stephens (1972) reported that blood arylesterase activity is relatively low in young children (1-2 years of age), with this group having pronounced intersubject variability. No more recent information could be located on the time course of maturation of arylesterase or related enzymes in human blood or liver.

Comparative studies show that neonatal rodents are frequently more susceptible to toxicants than adult animals, but such findings should be interpreted with caution when extrapolating to humans. Done (1964) and Goldenthal (1971) compiled the results of L[D.sub.50] (median lethal dose) studies of several hundred chemicals in neonatal and mature rodents. The neonatal animals were more sensitive to many but not all the compounds. Almost all the age-dependent differences in L[D.sub.50] values were less than an order of magnitude; indeed, most varied no more than 2- to 3-fold. More pronounced interage differences were seen for a few drugs, some of which (e.g., chloramphenicol, diazepam) are poorly metabolized and accumulate to toxic levels in human newborns. As full-term human newborns are more mature than their rodent counterparts with respect to liver metabolism, interage differences might be less pronounced in humans. However, maturation is much more rapid in rodents, such that even a few days of growth can result in marked disparity in chemical metabolism, disposition, and effects. Results from the Done (1964) acute toxicity compilation (Figure 6) suggest dramatic differences in sensitivity in early postnatal life. These types of examples from the animal literature may be excellent research opportunities for exploring the relative importance of toxicokinetic versus toxicodynamic mechanisms in determining susceptibility in early life. Although the animal research may be able to point out mechanisms of toxicokinetic susceptibility, extrapolation of temporal trends across species is difficult because organs and their associated functions mature at different rates in different species. This makes the choice of an appropriate animal model for children complex and cross-species extrapolations more uncertain than what we are accustomed to in adult risk assessments [Bruckner 2000; NAS (National Academy of Sciences) 1993].

[FIGURE 6 OMITTED]

Despite the dissimilarities in maturation mentioned previously, rodent studies can provide valuable information on mechanisms and specific immaturities that may be broadly applicable to infants and children. Heavy metals are a good example. It is widely recognized that dietary antigens, maternal antibodies, and other macromolecules may be absorbed through the immature gastrointestinal mucosa. Increased levels of relatively polar molecules in the cerebrospinal fluid of infants suggest that such compounds penetrate the blood-brain or blood-cerebral spinal fluid barrier more readily in infants than in children and adults (Adinolfi 1985; Dziegielewska et al. 2000). Therefore, it has been hypothesized that equivalent exposures of infants and adults to heavy metals such as mercury and lead will result in greater oral absorption and greater central nervous system deposition (and toxicity) in the infants. Kostial et al. (1978) demonstrated substantially greater gastrointestinal absorption and whole-body retention of lead, mercury, and cadmium by suckling rats than by adults. A similar phenomenon has been reported for lead in juvenile monkeys (Pounds et al. 1978) and human infants (Ziegler et al. 1978). Retention of intravenously injected lead was 8 times higher in the brains of suckling rats than in the brains of adults (Kostial et al. 1978). Brain retention of injected mercury was 19 times higher during the perinatal period. Rodents have been very useful in delineation of the toxicokinetic basis of age-dependent differences in heavy-metal toxicity, but subhuman primates are generally the animals of choice to examine toxicodynamic factors (Davis et al. 1990).

Relatively few in vivo data are available from kinetics studies of drugs or other chemicals in preadolescent animals because of technical difficulties in working with such small subjects and prior lack of interest in immature populations. Similarly, there is a paucity of information on the maturation of many physiologic processes in small animals. The maturation of hepatic xenobiotic metabolism in rats, in contrast, has been relatively well characterized (Imaoka et al. 1991; Renwick 1998; Watanabe et al. 1993). Although rat liver is immature at birth with respect to many metabolic functions, certain CYP functions, epoxide hydrolase, and glucuronidation function reach adult levels within the first week to 10 days. Other functions such as glutathione transferase and aryl hydrocarbon hydroxylase take longer to develop (Renwick 1998).

Despite these known metabolic differences during development, findings of increased susceptibility in juvenile rodents have not been followed up to determine whether toxicokinetic mechanisms underlie these susceptibility differences. For example, Yoo et al. (1987) observed that CYP2E1-catalyzed metabolic activation of N-nitrosodimethylamine to a mutagen was substantially greater by hepatic microsomes of weanlings (25 days of age) than those of adult rats. Carbon tetrachloride, a chemical that also undergoes CYP2E1-catalyzed metabolic activation, was more hepatotoxic in 15-day-old than in 60-day-old male rats (Jahn et al. 1993). However, no accounts were located of animal studies relating susceptibility to injury by these CYP2E1 substrates to the time course of maturational changes in the chemical's metabolism.

Data gaps and challenges in applying animal data to children's toxicokinetic assessments. It is possible, under certain well-defined conditions, to make reasonable predictions of the disposition of drugs in adult humans on the basis of animal studies (Lin 1995). For example, the oral absorption of lipophilic compounds is relatively species independent, as is the volume of distribution of unbound drug (Fichtl et al. 1991; Lin 1995). Methods to extrapolate other parameters such as renal excretion of filtered drugs across species provide reasonable estimates for human pharmacokinetic function (Lin 1995). However, across species, differences in protein binding and drug metabolism can be substantial, leading to greater uncertainty in relying on pharmacokinetic data generated in animals (Nedelcheva and Gut 1994; Stevens et al. 1993). Nevertheless, metabolic clearance of low doses of flow-limited drugs should be less dependent on species-inherent rates of metabolism than on relative liver blood flow rates (Lin 1995). Thus, data from animal studies often can be useful in forecasting toxicokinetics in humans when their limitations are considered. However, this conclusion is based on comparison of adult animals with human adults. The development of toxicokinetic functions in early life is sufficiently different in rodents compared with humans to make direct extrapolation from juvenile animal studies difficult.

As mentioned above, one of the most promising approaches to quantitative estimation of internal doses of chemicals in children is PBPK modeling. Models have been developed to describe the disposition of a number of chemicals in pregnant rodents and their fetuses as well as in the lactating rat and nursing pup (Krishnan and Andersen 1998). Luecke et al. (1994) and Welsch et al. (1995) have adapted such models to human pregnancy to forecast potential teratogenic events. O'Flaherty developed PBPK models that accurately predicted time courses of lead in the blood and its deposition in bones of developing rats (O'Flaherty 1991) and children (O'Flaherty 1995). These models incorporated age-dependent changes in body weight, tissue volumes and blood flows, and bone formation and resorption rates. Pelekis et al. (2001) have made assumptions about distributions of physiologic and biochemical parameters in children and have estimated their effects on the disposition of volatile organic chemicals. Gentry et al. (2002) have modeled a variety of different chemicals across a wide range of ages by building functions into their model that adjust for metabolism and renal function changes during development.

Knowledge gained from PBPK modeling efforts with animals can be of significant benefit in the understanding of the kinetic behavior of chemicals in humans and in developing PBPK models for humans. Toxicokinetic studies in animals can yield a number of important parameters, including in vivo and in vitro partition coefficients, metabolic rate constants, and identity and stoichiometric yield of metabolites. Such chemical- and species-specific data are needed to construct an appropriate model. The model may be used to generate simulations of the time course of parent compound and/or metabolite(s) in blood, tissues, and urine. This information is often useful in design of the in vivo kinetics experiments needed to assess the accuracy of the model's predictions. When physiologic parameters are scaled allometrically to humans, the animal-based model becomes a starting point for making predictions regarding human pharmacokinetics. However, development of PBPK models for children would require extra effort to scale the model to the physiology and functional immaturity that occurs at specific developmental stages. Although it is tempting to use juvenile animal data to inform such children's PBPK modeling efforts, the across-species differences in physiologic and biochemical maturation add substantial uncertainty to this approach. Data in juvenile subhuman primates may be more promising because of their similarity to growing children (Conrad et al. 1995), but this would involve more expense.

Use of basic physiologic differences across age groups to ascertain how factors such as renal clearance, protein binding, and lipid partitioning may vary with age. Renal clearance and protein binding capacity are immature at early postnatal stages and develop over the course of the first 6 months to year of life (Table 1). Glomerular filtration rate as well as transporter (secretory) systems in the proximal convoluted tubule are deficient at birth (Kearns and Reed 1989; Morselli 1989), leading to relatively slow clearance of a wide array of antibiotics and other renally cleared drugs and metabolites. Compounding the chances for greater chemical effect in the first months of life is a deficient protein binding status due to lower levels of albumin (which binds weakly acidic chemicals) and alpha-1-glycoprotein (which binds basic chemicals). Another issue is the relative deficiency to conjugate and excrete bilirubin early in life, which leads to greater occupation of serum binding sites by bilirubin. These factors may contribute to a greater ratio of free compared with bound chemical, leading to a higher potential for uptake into tissues and toxic action. Diminished protein binding can also lead to shorter duration of action because of greater availability of chemical for metabolism and elimination. These factors would likely be important only for chemicals that exhibit a high degree of plasma protein binding (e.g., trichloroacetic acid).

Summary of phase II toxicokinetic analysis. As information in the areas described above is gathered and processed, we can evaluate from an internal dose perspective which age group(s) appears to be most exposed. Age groups that appear to differ most from adults can be prioritized for more detailed analysis as the process continues. The following questions may be helpful:

Does the pharmacokinetic profile shift at natural age breakpoints? An early step in analyzing children's toxicokinetic data is deciding whether and how to create bins across ages. The rapid development of children requires that the population be broken into relatively small age groups, particularly in the early postnatal period. If larger bins are used, the data may become highly variable, with age specificity becoming a causality of the desire to simplify the assessment. Thus, bins must be constrained to reflect key developmental stages. Nevertheless, it is useful to make them as large as possible, as this maximizes the number of subjects per group and enhances the power of cross-age comparisons. It should be noted that similar age binning may occur in the exposure and toxicodynamic areas to reflect critical changes in behavior or windows of heightened vulnerability. At some point the children's kineticist may be asked to adjust his bins to match those created in these other areas for the sake of harmony and to allow risk calculations for neatly defined age groups. Age bins were developed to organize the data shown in Table 1 and Figures 2 and 3. These bins are somewhat arbitrary and were based on the overall availability of data for a variety of pathways and chemicals. For specific chemicals and pathways, alternative bins can certainly be considered.

Which pharmacokinetic parameters are most important in determining internal dose? Although PBPK models aim for the simplest approach that yields predictive results, there are inevitably many parameters that must be estimated to run the model. These parameters include compartment size, partition coefficients, cardiac output, organ blood flows, metabolic rates, lung ventilation rates, and urinary elimination rates. Because these models contain multiple tissue compartments and we may be dealing with several different age groups, there is a potentially large number of parameter values that would require elucidation. However, the model may not be equally sensitive to all the inputs, making those parameters that have the greatest influence on model performance the most important to calibrate properly. Efforts to simplify pharmacokinetic modeling in both adults and children have relied on a series of algebraic equations that highlight key pharmacokinetic parameters while not including others that may be model insensitive (Pelekis et al. 2001). Such approaches point out the importance of identifying those pharmacokinetic parameters that drive internal dose. Which pharmacokinetic factors are most influential will depend on the properties of the chemical under investigation. The clearance of chemicals with high hepatic extraction (rapid liver metabolism) will depend primarily on hepatic blood flow, whereas chemicals with low hepatic extraction depend most on the rate of liver metabolism relative to other competing rates (urinary elimination or exhalation of parent compound). To determine the focus of a children's toxicokinetic analysis, the most important toxicokinetic parameters in a validated adult model must be determined. Additional effort can then be put into identifying how these parameters change as different child age groups are considered. Ultimately, the fate of any chemical will be determined by the interaction of all pharmacokinetic factors acting simultaneously. This kind of sensitivity analysis will ensure that those parameters with the greatest influence are given the attention they warrant.

Regardless of whether one is conducting a full PBPK assessment across ages or a less rigorous approach, it is important to know how key parameters change with age. As discussed above, various data resources and analytic approaches may be used to determine how key parameter values differ between children and adults. Similar to Table 1, it may be helpful to develop a table of key physiologic and metabolic parameters that affect the fate of the chemical in question and list what is known about those pathways for the in utero and postnatal periods. If the parameter values are not materially different between children and adults or if it appears that certain factors negate one another, one might conclude that it is not important to conduct further toxicokinetic assessment for that age groups(s). However, for those chemicals and age groups where children appear to receive a greater internal dose of active toxicant, further analytic steps may be considered as follows:

Semiquantitative approaches. Investigators should use information on the predominant disposition pathways to evaluate child/adult dosimetry differences. It may be possible to base conclusions on one or more particularly useful surrogates from the drug literature. In cases where pathways that govern chemical activation or clearance appear to be substantially different across early life and adults, an age group-specific adjustment factor may be considered. This may be based on the size of the child/adult ratio in that parameter value, using comparisons based on central tendency values or on bounding percentile in children versus central tendency in adults (e.g., 5th percentile of child metabolic function vs. adult median percentile). Such adjustment factors provide the opportunity for the risk assessment to consider child-specific toxicokinetic information readily available. However, it is a simplistic approach that does not consider the various ways in which changes in one pathway's function may be modified by other physiologic factors, leading to an overall influence greater or less than expected. A more complete analysis involving PBPK modeling is needed to integrate child/adult differences in a variety of parameters into across age-group comparisons of internal dosimetry.

Quantitative approaches--PBPK analysis. This involves constructing a child age group-specific model that incorporates the base physiologic parameter values specific to the age group(s) being analyzed as well as modifications in metabolism and renal clearance to reflect the major functional immaturities in early life. This model would then incorporate chemical-specific parameter values to simulate how children at specific developmental stages are likely to handle the chemical. It is essential to carefully consider data needs for such an analysis and whether sufficient data exist or can be generated/estimated.

Ideally, phase II of the children's toxicokinetic risk assessment would result in the following:

* An understanding of the toxicokinetic mechanisms that govern the fate of the target chemical

* Data resources that describe how these mechanisms change during development

* An initial prioritization of age groups for more detailed analysis based on likelihood for internal dosimetry differences relative to adults

* An analytic strategy (semiquantitative or quantitative techniques) for incorporating these data into the risk assessment process.

Phase III: Risk Characterization

The analytic approaches developed in phase II would be followed through so that phase III can attempt to answer the following questions:

* How similar/dissimilar are in utero and childhood stages to adults in terms of internal dose received per unit of administered dose?

* How variable are young age groups versus adults?

* Do all age groups fall within the default interindividual variability factor of 3.2?

* Do child:adult differences warrant the development of chemical- and age-specific uncertainty factors?

* Can internal dosimetry estimates for different age groups be developed and used in quantitative risk assessment?

To place the analytical results in perspective, the risk characterization will usually try to describe the degree of uncertainty and variability in the analysis. The following questions may be particularly pertinent to such a discussion.

Are children more variable in their pharmacokinetic responses than adults? How do we assess and apply this information? The role of variability in the risk assessment process can range from purely qualitatitive (e.g., people are different from one another, so we are not as certain as we would like to be about the degree of risk) to semiquantitative (e.g., people vary considerably in metabolizing chemical X, so it seems prudent to use a half-log toxicokinetic uncertainty factor) to fully quantitative (e.g., we have sufficient individual data to plot the population distribution of effective internal doses per unit of external dose, so let us show the risk manager the cleanup options that protect the 50th, 75th, 90th, and 95th percentiles). The fully quantitative mode of variability analysis is used in lead uptake/biokinetic modeling to enable risk assessors to calculate the amount of lead reduction in water, soil, air, or diet needed to bring 95% of the population of exposed children below the blood lead target of 10 [micro]g/dL (U.S. EPA 1994). It is logical that children are more toxicokinetically variable than adults because of their variable growth and maturation rates, as well as the genetic, nutritional, disease, body composition, and prescription (and other) drug factors that create toxicokinetic variability in adults. Children also present the problem of variable growth rates, which can make even small age bins relatively heterogeneous. Greater variability can affect risk conclusions, especially if one is concerned about protecting the tails (e.g., 90th percentile) of the distribution. With greater variability, it is also more likely that a substantial fraction of a certain age group will lie outside the half-log toxicokinetic variability range we normally allocate to the adult defaults. Thus, it is critical that a children's toxicokinetic assessment characterizes in some way the degree of variability present in each age group's data set and determines whether that variability is greater than in the adult case.

By addressing these questions, the risk characterization will endeavor to determine whether toxicokinetic differences are likely between children and adults and whether these differences will be important to the overall risk assessment conclusions for this chemical/ scenario. If so, the risk characterization can describe the advantages and disadvantages of applying qualitative (e.g., professional judgment), semiquantitative (e.g., modified uncertainty factors), or quantitative approaches (PBPK modeling with distributional analyses). The major caveats and data gaps should be elaborated so that the limitations of the assessment are clear and critical research needs are identified. Finally, the characterization should describe the toxicokinetic factors that appear most influential in creating child/adult differences so that the key risk drivers for children are explained.

Table 1. Postnatal developmental profile of pharmacokinetic functions.

Pharmacokinetics             Premature                Full-term
pathway                       neonates                 neonates

Renal clearance                                 35% - in vivo
                                                  [t.sub.1/2]
                                                  for 7 drugs (a)
                                                GFR ~20% adult (b)

CYP1A1                         -- (c)                   -- (c)

CYP1A2                                          2% - PL (e)
                                                2% - in vitro EA (f)
                                                11% - in viva
                                                  [t.sub.1/2]
                                                  for 2 drugs (d)

CYP2A6                                          20% - Nicotine
                                                  [t.sub.1/2] (ff)

CYP2E1                                          13% - PL (h)
                                                27% - in vitro EA (i)
                                                27% - Serum
                                                DMO/TMO (j)

CYP2C9/19              21% - PL (k)             30% - in vivo
                       33% - in vitro EA (l)      [t.sub.1/2]
                                                  for 1 drug (m)

CYP2D6                                          13% - PL (n)

CYP3A4                 19% - in vivo            17% - in vitro EA (p)
                         [t.sub.1/2]            50% - in vivo
                         for 8 drugs (o)          [t.sub.1/2]
                                                  for 8 drugs (o)

CYP3A7                                          1100% - in vitro EA
                                                  (q)

Epoxide hydrolase                               42% - PL (f)
                                                50% - CBZ-E/CBZ (s)

Glucuronidation        23% - in vivo            34% -in vivo
                         [t.sub.1/2]              [t.sub.1/2]
                         for 6 drugs (t)          for 6 drugs (t)

Sulfation/                                      84% - APAP clearance
 glucuronidation                                  (u)

Acetylation                                     83% slow phenotype (v)

GSH transferase                                 GST[[alpha].sub.B1] ~
                                                  100% - PL (x)
                                                GST[[alpha].sub.B2] ~
                                                  40% - PL (x)
                                                GSTmu ~ 60% - PL (x)
                                                GSTpi>2100% - PL (x)

Biliary excretion      57% - BSP clear (z)      70% - BSP clear (z)

Serum protein          Few binding              Few binding
                         sites (aa)               sites (aa)

Serum cholinesterase   50% - EA (bb)            57% - EA (bb)

Serum arylesterase                              28% - EA (cc)

Alcohol                                         55% - in vivo alcohol
  dehydrogenase                                   clearance (dd)

Pharmacokinetics
pathway                   1 Week-2 months             2-6 Months

Renal clearance        28% - in vivo            88% - in vivo
                         [t.sub.1/2]              [t.sub.1/2]
                         for 7 drugs (a)          for 7 drugs (a)
                                                GFR - 100% adult (b)

CYP1A1                         -- (c)                   -- (c)

CYP1A2                 4% - PL (e)
                       3% - in vitro EA (f)     16% - PL (e)
                       4% - in vitro EA (g)     9% - in vitro EA (f)
                       23% - in vivo            13% - in vitro EA (g)
                         [t.sub.1/2]            81% - in vivo
                         for 2 drugs (d)          [t.sub.1/2]
                                                  for 2 drugs (d)

CYP2A6

CYP2E1                 22% - PL (h)             30% - PL (h)
                       39% - in vitro EA (l)    47% - in vitro EA (i)
                       27% - in vivo serum
                       DMO/TMO (j)

CYP2C9/19              29% - PL (k)             38% - PL (k)
                       30% - in vitro EA (l)    45% - in vitro EA (l)

CYP2D6                 22% - PL (n)             34% - PL (n)

CYP3A4                 29% - in vitro EA (p)    37% - in vitro EA (p)
                       55% - in vivo
                         [t.sub.1/2]
                         for 8 drugs (o)

CYP3A7                 600% - in vitro EA (q)   300% - in vitro EA (q)

Epoxide hydrolase      50% - CBZ-E/CBZ (s)      50% - CBZ-E/CBZ (s)

Glucuronidation        47% - in vivo            102% - in vivo
                         [t.sub.1/2]              [t.sub.1/2]
                         for 6 drugs (t)          for 6 drugs (t)

Sulfation/                                      149% - APAP
 glucuronidation                                  clearance (u)

Acetylation            12% - in vivo caffeine   65% - in vivo caffeine
                         N-acetylation (w)        N-acetylation (w)

GSH transferase        GST[[alpha].sub.B1] ~    GST[[alpha].sub.B1] ~
                         100% - PL (x)            100% - PL (x)
                       GST[[alpha].sub.B2] ~    GST[[alpha].sub.B2] ~
                         40% - PL (x)             40% - PL (x)
                       GSTmu ~ 60% - PL (x)     GSTmu ~ 60% - PL (x)
                       GSTpi>2100% - PL (x)     GSTpi>2100% - PL (x)

Biliary excretion      90% - BSP clear (z)

Serum protein          Few binding              Increasing # sites
                         sites (aa)               (aa)

Serum cholinesterase   101 % - EA (bb)          91% - EA (bb)

Serum arylesterase     34% - EA (cc)            52% - EA (cc)

Alcohol                15% - in vitro EA (ee)
  dehydrogenase

Pharmacokinetics
pathway                   6 Months-1 year             1-2 Years

Renal clearance        137% - in vivo           137% - in vivo
                         [t.sub.1/2]              [t.sub.1/2]
                         for 7 drugs (a)          for 7 drugs (a)

CYP1A1                         -- (c)                   -- (c)

CYP1A2                 25% - PL (e)             175% - in vivo
                       15% - to vitro EA (f)      [t.sub.1/2]
                       13% - in vitro EA (g)      far 2 drugs (a)
                       175% - in vivo
                         [t.sub.1/2]
                         for 2 drugs (d)

CYP2A6

CYP2E1                 36% - PL (h)             92% - in vivo serum
                       41% - in vitro EA (i)    DMO/TMO (c)
                       97% - Serum
                       DMO/TMO (j)

CYP2C9/19              36% - PL (k)             36% - PL (k)
                       83% - in vitro EAI       83% - in vitro EA (l)
                       182% - in vivo           182% - in vivo
                         [t.sub.1/2]              [t.sub.1/2]
                         for 1 drug (m)           for 1 drug (m)

CYP2D6                 45% - PL (n)

CYP3A4                 46% - in vitro EA (p)    110% - in vitro EA (p)
                       200% - in vivo           200% - in vivo
                         [t.sub.1/2]              [t.sub.1/2]
                         for 8 drugs (o)          for 8 drugs (o)

CYP3A7                 200% - in vitro EA (q)

Epoxide hydrolase      50% - CBZ-E/CBZ (s)      50% - CBZ-E/CBZ (s)

Glucuronidation        84% - in vivo            84% - in vivo
                         [t.sub.1/2]              [t.sub.1/2]
                         for 6 drugs (t)          for 6 drugs (t)

Sulfation/                                      70% - APAP
 glucuronidation                                  clearance (u)

Acetylation

GSH transferase        GST[[alpha].sub.B1] ~
                         112% - PL (x)
                       GST[[alpha].sub.B2] ~
                         62% - PL (x)
                       GSTmu ~ 93% - PL (x)
                       GSTpi>2100% - PL (x)

Biliary excretion

Serum protein          Increasing # sites       Adult level (aa)
                         (aa)

Serum cholinesterase   104% - EA (bb)

Serum arylesterase     84% - EA (cc)            80% - EA (cc)

Alcohol                32% - in vitro EA (ee)
  dehydrogenase

Pharmacokinetics
pathway                        > 2 Years

Renal clearance        185% - in vivo [t.sub.1/2]
                         for 7 drugs (a)

CYP1A1                           -- (c)

CYP1A2                 54% - PL (e)
                       35% - in vitro EA (f)
                       185% - in vivo [t.sub.1/2]
                         for 2 drugs (d)

CYP2A6

CYP2E1                 82% - PL (h)
                       83% - in vitro EA (b)
                       92% - Serum
                       DMO/TMO (c)

CYP2C9/19              130% - in vivo [t.sub.1/2]
                         far 1 drug (m)

CYP2D6                 88% - PL (n)

CYP3A4                 189% - in vivo [t.sub.1/2]
                         for 8 drugs (o)

CYP3A7

Epoxide hydrolase      65% - CBZ-E/CBZ (s)

Glucuronidation        74% - in vivo [t.sub.1/2]
                         for 6 drugs (t)

Sulfation/
 glucuronidation

Acetylation            48% slow
                         phenotype (v)

GSH transferase        121% - in vivo [t.sub.1/2]
                         of busulfan (y)

Biliary excretion

Serum protein          Adult level (aa)

Serum cholinesterase

Serum arylesterase

Alcohol                91 % - in vitro EA (ee)
  dehydrogenase

Abbreviations: APAP, acetaminophen; BSP, bromosulfophthalein; CBZ,
carbamazepine; CBZ-E, carbamazepine-epoxide; DMO, dimethadione; EA,
enzyme activity; EH, epoxide hydrolase; GFR, glomerular filtration
rate; PL, protein level; TMO, trimethadione.

Pathway function is shown as % adult. (a) Ginsberg et al. (2002)
analyzed renally cleared drugs: ampicillin, cimetadine, furosemide,
piperacillin, ticarcillin, tobramycin, vancomycin. (b) GFR in mL/min/kg
(Besunder at al. 1988; Morselli 1989). (c) CYP1A1 protein or enzyme
activity was not detectable microsomes from liver bank samples at any
age, indicating very low constitutive levels (Sonnier and Cresteil
1998). (d) ginsberg et al. (2002) analysis of drugs cleared primarily
via CYP1A2: caffeine, theophylline, (e) Sonnier and Cresteil (1998)
measured  CYP1A2 protein levels via immunochemical methods in
microsomes from liver bank samples (n = 6-23 per age group).
(f) Sonnier and Cresteil (1998) measured CYP1A2 activity in microsomes
from liver bank samples using methoxyresorufin as substrates (n = 6-23
per age group). (g) Cazeneuve et al. (1994) measured CYP1A2 activity
in microsomes from liver bank samples via caffeine N-demethylation.
(h) Vieira et al. (1996) measured CYP2E1 protein levels in liver bank
microsomes by immunochemical means (n = 4-23 per age group). (i) Vieira
et al. (1996) measured CYP2E1 activity levels in liver bank microsomes
using chlorozoxazone as substrate (n = 2 for group 1-10 years of age;
otherwise, n = 9-21 per age group). (j) Tanaka (1998) and Nakamura et
al. (1998) report the ratio of oxidized metabolite DMO to parent drug
TMO in serum of children dosed at the following ages: < 4 weeks
(n = 5); 1 month-1 year (n = 5); 1-10 years (n = 21); adult (n = 20).
CYP2E1 converts TMO to DMO. (k) Treluyer at al. (1996) measured CYP2C
protein levels in microsomal preparations from liver bank samples with
immunochemical techniques. (l) Treluyer at al. (1996) measured CYP2C19
enzyme activity in microsomal preparations from liver bank samples
using diazepam as substrate. (m) Ginsberg at al. (2002) analyzed
tolbutamide, a drug cleared primarily via CYP2C. (n) Treluyer et al.
(1991) measured CYP2D6 protein levels in microsomal preparations from
liver bank samples with immunochemical techniques. (o) Ginsberg et al.
(2002) analyzed drugs cleared primarily by CYP3A: afentanil,
carbamazepine, fentanyl, lidocaine, midazolam, nifedipine, quinidine,
triazolam. (p) LaCroix at al. (1997) measured CYP3A4 EA in microsomes
from liver bank samples using testosterone as substrate (n = 12
newborns; 9 at 1 week-1 month; 13 at 1-3 months; 11 at 3-12 months; 2
at > 1 year; 11 adults). (q) LaCroix at al. (1997) measured CYP3A7
EA in microsomes from liver bank samples using dehydroepiandrosterone
as substrate (n= 12 newborns; 8 at 1 week-1 month, 20 at 1-3 months;
14 at 3-12 months, 0 at > 1 year, 12 adults). (r) Ratanasavanh
et al. (1991) measured epoxide hydrolase protein levels by
Western blot on liver bank microsomal preparations (n= 5 for
1 day-1 week of age; n= 5 for adults). (s) EH levels were
estimated by CBZ-E to CBZ ratio in blood in different age groups at
steady state after repeat drug administration in epileptic
patients. Data were pooled across 2 studies (Korinthenberg at al.
1994; Pyonnonen at al. 1977). Higher CBZ-E/CBZ ratios in children
are indicative of slower EH activity, as CBZ-E formation and CBZ
clearance is slower in young children (Eichelbaum at al. 1975,
1985; Kuhnz et al. 1983; Ohmori at al, 1998). (t) Ginsberg at al.
(2002) analysis of drugs cleared directly via glucuronidation:
lorazepam, morphine, oxazepam, trichloroethanol, valproic acid,
idovudine. (u) Ginsberg at al. (2002) compilation of kinetic data
far APAP across 5 five studies; n=7-24 per age group. APAP is a
substrate for various conjugation reactions, with sulfation
predominating in early life (Besunder at al. 1988; Levy at al.
1975). (v) Pariente-Khayat at al. (1991) measured the ratio of
acetylated to nonacetylated metabolite in urine after caffeine
administration in children (n=10-26). Younger age groups were
compared against an older age group (6 months-2 years), as
adult data were not available and onset of rapid phenotype
occurs byt his age. (w) Szorady et al. (1987) Thenotyped 100
newborns 2-3 days old as well as 100 or more subjects in older
age groups, using acetylation of sulfadimidine (as appears in
urine). (x) Strange at al. (1989) measured GST protein levels by
radimmmunoassay in liver bank tissue cytosols. GST[[alpha].sub.B1]
levels were 70% of adult in utero and 112% of adult at 55-10 months
postnatal. Therefore, the estimate is that newborn through 6-month
levels would be approximately equal to adult. GSTmu protein
levels were 25% of adult during the in utero period and 62% of
adult at 5-10 months postnatal. Therefore, the estimate is that
newborn through 6-month levels are 40% of adult. GSTmu is 22% of
adult during in utero period and 93% of adult at 55-10 months.
Therefore, the estimate is that newborn through 6-month levels are
60% of adults. GSTpi is 5300% of adult during in store period and
2100% of adult at 55-10 months. (y) Gibbs et al. (1997) measured
busulfan half-life in a group at 14 chil-dren (average age = 3
years) relative to adults. Busulfan is metabolized predominately by
GST[[alpha].sub.B1]. (z) Jusko (1972) measured clearance of BSP in
groups of 5 or more children; values in young children were compared
against another child group (3 months-14 years of age), as adult data
were not available. (aa) Serum protein binding sites (albumin,
[[alpha].sub.1],-acid glycoprotein) are law in newborns but increase
to all levels by 1 year (Besunder et al. 1988). (bb) Data were pooled
across 4 studies (Augustinsson and Barr 1963; Ecobichon and Stephens
1972; Lehmann at al. 1957; Zsigmond and Downs 1971) in which serum
cholinesterase levels were measured with acetylchoune, benzoylcholine,
butyrylcholine, or procaine as substrate. (cc) Augustinsson and Barr
(1963) and Ecobichon and Stephens (1972) measured serum
arylesterase activity with phenyl acetate as substrate. (dd)
Idanpaan-Heikkila at al. (1972) followed the removal of ethanol from
mother and newborn blood (n= 6 far each age); newborns were exposed
transplacentally and born with equal blood alcohol concentration
as mother. (ee) Pikkarainen at al. (1967) measured alcohol
dehydrogenase activity in liver samples from newborns 1 week-2
months (n = 2); infants 6 months -2 years (n=2); older children 5-15
years (n =3); and adults (n=3). Data from 9 fetuses were
approximately 25% of the new born levels. (ff) Dempsey at al. (2000)
meas-ured nicotine elimination from newborns (n = 5) exposed
transplacentally.

Table 2. Partial list of human placental xenobiotic- and
hormone-metabolizing enzymes or isoenzymes. (a)

Phase        Type                    Reaction (gene)

I       MFO              0-Deethylase (CYP1A1)
I       MFO              Aryl hydrocarbon hydroxylase (CYP1A1)
I       MFO              Hydroxylase (CYP3A7)
I       MFO              Aromatase (CYP19)
I       MFO              Cholesterol side-chain cleavage (CYP11A)
I       MFO              Estrogen catecholformation,
                         2-Hydroxylation (CYP1A1)
                         4-Hydroxylation (CYP1B1)
I       MFO              25-Hydroxycholecalciferol hydroxylase
I       Oxidoreductase   17[beta]-Hydroxydehydrogenase
                         Type 1
                         Type 2
I       Oxidoreductase   11[beta]-Hydroxydehydrogenase
I       Oxidation        Dehydrogenase
I       Oxidation        Monoamine
II      Sulfatase        Sulfate cleavage
II      Conjugation      GST
II      Conjugation      Catechol-0-methyltransferase

Phase   Type             Substrate                   Constitutive

I       MFO              7-Ethoxycoumarin                (+)
I       MFO              PAH                             (?)
I       MFO              Cortisol                         +
I       MFO              Androgens                       (+)
I       MFO              Cholesterol                     (+)
I       MFO              Estrogens                       (+)

I       MFO              25-Hydroxycholecalciferol
I       Oxidoreductase                                   (+)
                         Estrone to estradiol
                         Estradiol to estrone
I       Oxidoreductase   Cortisol/cortisone              (+)
I       Oxidation        Alcohol/acetaldehyde            (+)
I       Oxidation        Norepinephrine                  (+)
II      Sulfatase        Steroid sulfates                (+)
II      Conjugation      Epoxides                        (+)
II      Conjugation      Catecholamines, catechol        (+)
                           estrogens

Phase   Type                 Inducer                Inhibitor

I       MFO              Cigarette smoke        Aminoglutethimide
I       MFO              Cigarette smoke       Alpha-naphthoflavone
I       MFO                    --                       --
I       MFO                    --               Aminoglutethimide
I       MFO                    --               Aminoglutethimide
I       MFO              Cigarette smoke                --

I       MFO                    --                       --
I       Oxidoreductase         --             16-Methylene estradiol

I       Oxidoreductase         --                       --
I       Oxidation              --                       --
I       Oxidation              --          Monoamine oxidase inhibitors
II      Sulfatase              --                       --
II      Conjugation            --                       --
II      Conjugation            --                       --

Abbreviations: MFO, mixed-function oxidase; (+), the enzyme is
constitutively expressed inhuman placenta; (?), it is unclear
whether the enzyme is constitutively expressed;-, no known inducer
or inhibitor for the placental enzyme. (a) Data modified from Slikker
and Miller (1994), with additional data from Arcuri et al. (1999);
Juchau (1995); Moghrabi et al. (1997); Rasheed et al. (1997); and
Zusterzeel et al. (1999).

Table 3. Some CYP isozymes and their occurrence
in human tissues. (a)

            Adult       Adult       Fetal       Fetal
CYP       liver (b)   brain (b)   liver (b)   liver (b)

1A1           +           +           +
1A2          +++          -           -
1B1           +                       +           +
2A6          ++                       -
2A7           +                       -
2B1/2B2       +           +
2B6/2B7       +                       -
2C           +++                  [+ or -]
2C8-19        +                       +           +
2D6           +           +       [+ or -]
2E1           +           +       [+ or -]
2F1           -                       -
3A4          +++          +       [+ or -]
3A5          ++                       +
3A7           +                      +++
461           -                       -

(a) Data modified from Carpenter et al. (1997); Chen et al.
(1999); Farm and Omiecinski (1993); Hakkola et al. (1994,
1997, 1998a, 19956); LaCroix et al. (1997); Maenpaa et al.
(1993); Murray et al. (1992); Ravindranath and Boyd (1995);
Slikker (1997); and Yang et al. (1995). (b) [+ or -], possibly present
in small quantities, or mRNA observed but no protein confirmation;
-, not detected; +, present in low levels; ++, present
in moderate levels; +++, present in high levels.

REFERENCES

Adams J, Barone S Jr, LaMantia A, Philen R, Rice DC, Spear L et al. 2000. Workshop to identify critical windows of exposure for children's health: neurobehavioral work group summary. Environ Health Perspect 108(suppl 3):S535-S544.

Adinolfi M. 1985. The development of the human blood-CSF-brain barrier. Dev Med Child Neurol 27:532-537.

Andersen ME, Clewell HJ III, Gargas ML, Smith FA, Reitz RH. 1987. Physiologically based pharmacokinetics and the risk assessment process for methylene chloride. Toxicol Appl Pharmacol 87:185-205.

Andersen ME, Clewell H III, Krishnan K. 1995. Tissue dosimetry, pharmacokinetic modeling, and interspecies scaling factors. Risk Anal 15:533-537.

Anderson BJ, McKee AD, Holford NH. 1997. Size, myths, and the clinical pharmacokinetics of analgesia in pediatric patients. Clin Pharmacokinet 33:313-327.

Arcuri F, Sestini S, Cintorino M. 1999. Expression of 11[beta]-hydroxysteroid dehydrogenase in early pregnancy: implications in human trophoblast-endometrial interactions. Semin Reprod Endocrinol 17:53-61.

Atterberry TT, Burnett WT, Chambers JE. 1997. Age-related differences in parathion and chlorpyrifos toxicity in male rats: target and nontarget esterase sensitivity and cytochrome P450-mediated metabolism. Toxicol Appl Pharmacol 147:411-418.

Augustinsson KB, Barr M. 1963. Age variation in plasma arylesterase activity in children. Clin Chim Acta 8:568-573.

Barr M, Brent RL. 1978. Uterine vascular interruption and combined radiation and surgical procedures. In: Handbook of Teratology (Wilson JG, Fraser C, eds). New York:Plenum, 275-304.

Behrman RE, Lees MH, Peterson EN, De Lannoy CW, Seeds AE. 1970. Distribution of the circulation in the normal and asphyxiated fetal primate. Am J Obstet Gynecol 108:956-969.

Benke GM, Murphy SD. 1975. The influence of age on the toxicity and metabolism of methyl parathion and parathion in male and female rats. Toxicol Appl Pharmacol 31:354-269.

Bertz RJ, Granneman GR. 1997. Use of in vitro and in vivo data to estimate the likelihood of metabolic pharmacokinetic interactions. Clin Pharmacokinet 32:210-258.

Besunder JB, Reed MD, Blumer JL. 1988. Principles of drug biodisposition in the neonate. A critical evaluation of the pharmacokinetic-pharmacodynamic interface (part I). Clin Pharmacokinet 14:189-216.

Bogaert MC, Thiery M. 1983. Pharmacokinetics and pregnancy. Eur J Obstet Gynecol Reprod Biol 16:229-235.

Bois FY, Zeise L, Tozer TN. 1990. Precision and sensitivity of pharmacokinetic models for cancer risk assessment: tetrachloroethylene in mice, rats, and humans. Toxicol Appl Pharmacol 102:300-315.

Brock-Utne JG, Downing JW, Mankowitz E. 1980. Drugs and the fetomaternal unit. S Afr Med J 58:366-369.

Bruckner JV. 2000. Differences in sensitivity of children and adults to chemical toxicity: the NAS panel report. Regul Toxicol Pharmacol 31:280-285.

Carpenter SP, Savage DD, Schultz ED, Raucy JL. 1997. Ethanolmediated transplacental induction of CYP2E1 in fetal rat liver. J Pharmacol Exp Ther 282:1028-1036.

Cazeneuve C, Pons G, Rey E, Treluyer JM, Cresteil T, Thiroux G, et al. 1994, Biotransformation of caffeine in human microsemes rom foetuses, neonates, infants and adults. Br J Clin Pharmacol 37:405-412.

Chen H, Brzezinski MR, Fantel AG, Juchau MR. 1999. Catalysis of drug oxidation during embryogenesis in human hepatic tissues using imipramine as a model substrate. Drug Metab Dispos 27:1306-1308.

Conrad S, Ha J, Lohr C, Sackett G. 1995. Ultrasound measurement of fetal growth in Macaca nemestrina. Am J Primatol 36:19-35.

Cummings AJ. 1983. A survey of pharmacokinetic data from pregnant women. Clin Pharmacokinet 8:344-354.

Dancis J, Money WL, Condon GP, Motimer L. 1958. The relative transfer of estrogens and their glucuronides across the placenta in the guinea pig. J Clin Invest 37:1373-1378.

Daston G, Faustman E, Ginsberg G, Fenner-Crisp P, Olin S, Sonawane B, et al. 2004. A framework for assessing risks to children from exposure to environmental agents. Environ Health Perspect 112:238-256.

Davis JM, Otto DA, Weil DE, Grant LD. 1990. The comparative developmental neurotoxicity of lead in humans and animals. Neurutoxicol Teratol 12:215-229.

Dawes GS. 1968. Umbilical blood flow and vascular resistance: changes with age. In: Foetal and Neonatal Physiology (Dawes GS, ed). Chicago:Year Book Medical Publications.

Dempsey D, Jacob P III, Benowitz NL. 2000. Nicotine metabolism and elimination kinetics in newborns. Clin Pharmacol Ther 67:458-465.

Diggory HJ, Landrigan PJ, Latimer KP, Ellington AC, Kimbrough RD, Liddle JA, et al. 1977. Fatal parathion poisoning caused by contamination of flour for international commerce. Am J Epidemiol 106:145-153.

Dollery C. 2000. Therapeutic Drugs. 2nd ed. Edinburgh:Churchill Livingstone.

Done AK. 1964. Developmental pharmacology. Clin Pharmacol Ther 5:432-479.

Dome JL, Walton K, Renwick AG. 2001. Uncertainty factors for chemical risk assessment: human variability in the pharmacokinetics of CYP1A2 probe substrates. Food Chem Toxicol 39:681-696.

Dutton GJ. 1978. Developmental aspects of drug conjugation, with special reference to glucuronidation. Annu Rev Pharmacol Toxicol 18:17-35.

Dutton GJ, Leakey JE. 1981. The perinatal development of drug-metabolizing enzymes: what factors trigger their onset? Prog Drug Res 25:189-273.

Dvorchik BH. 1981. Nonhuman primates as animal models for the study of fetal hepatic drug metabolism. In: Drug Metabolism in the Immature Human (Soyka LF, Redmond GP, eds). New York:Raven Press, 146-162.

Dziegielewska KM, Knott GW, Saunders NR. 2000. The nature and composition of the internal environment of the developing brain. Cell Mol Neurobiol 20:41-56.

Ecobichon DJ, Stephens DS. 1972. Perinatal development of human blood esterases. Clin Pharmacol Ther 14:41-47.

Eichelbaum M, Ekbom K, Bertilsson L, Ringberger VA, Rane A. 1975. Plasma kinetics of carbamazepine and its epoxide metabolite in man after single and multiple doses. Eur J Clin Pharmacol 8:337-341.

Eichelbaum M, Tomson T, Tybring G, Bertilsson L. 1985. Carbamazepine metabolism in man induction and pharmacogenetic aspects. Clin Pharmacokinet 10:80-90.

Eltom SE, Babish JG, Schwark WS. 1993, The postnatal development of drug-metabolizing enzymes in hepatic, pulmonary and renal tissues of the goat. J Vet Pharmacol Ther 16:152-163.

Farin FM, Omiecinski CJ. 1993. Regiospecific expression of cytochrome P-450s and microsomal epoxide hydrolase in human brain tissue. J Toxicol Environ Health 40:317-335.

Fichtl B, Nieciecki AV, Walter K. 1991. Tissue binding versus plasma binding of drugs: general principles and pharmacokinetic consequences. In: Advances in Drug Research, Vol 20 (Testa B, ed). London:Academic Press, 117-166.

Garbis-Berkvens JM, Peters PW. 1987. Comparative morphology and physiology of embryonic and fetal membranes. In: Pharmacokinetics in Teratogenesis, Vol 1 (Nau H, Scott WJ, eds). Elkins Park, PA:Franklin Book Company, 14-44.

Gentry R, Teeguarden J, Sarangapani R, Covington T, Lawrence G, McDonald T, et al. 2002. Evaluation of the potential impact of age- and gender-specific pharmacokinetic differences on tissue dosimetry [Abstract]. Toxicol Sci 66:1-S.

Gibbs JP, Murray G, Risler L, Chien JY, Dev R, Slattery JT. 1997. Age-dependent tetrahydrothiophenium ion formation in young children and adults receiving high dose busulfan. Cancer Res 57:5509-5516.

Ginsberg G, Hattis D, Sonawane B, Russ A, Banati P, Kozlak M, et al. 2002. Evaluation of child/adult pharmacokinetic differences from a database derived from the therapeutic drug literature. Toxicol Sci 66:185-200.

Goebelsmann U, Roberts JM, Jaffe RB. 1972. Placental transfer of [sup.3]H-oestriol-3-sulphate-16-glucosiduronate and [sup.3]H-oestriol-16-glucosiduronate-[sup.14]C. Acta Endocrinol (Copenh) 70: 132-142.

Goebelsmann U, Wiqvist N, Diczfalusy E. 1968. Placental transfer of estriol glucosiduronates. Acta Endocrinol (Copenh) 59:426-432.

Goldenthal EI. 1971. A compilation of L[D.sub.50] values in newborn and adult animals. Toxicol Appl Pharmacol 18:185-207.

Greiss FC Jr, Gobble FL Jr. 1967. Effect of sympathetic nerve stimulation on the uterine vascular bed. Am J Obstet Gynecol 97:962 967.

Hakkola J, Pasanen M, Pelkonen O, Hukkanen J, Evisalmi S, Anttila S, et al. 1997. Expression of CYP1B1 in human adult and fetal tissues and differential inducibility of CYP1B1 and CYP1A1 by Ah receptor ligands in human placenta and cultured cells. Carcinogenesis 18:391-397.

Hakkola J, Pasanen M, Purkunen R, Saarikoski S, Pelkonen O, Maenpaa J, et al. 1994. Expression of xenobiotic-metabolizing cytochrome P450 forms in human adult and fetal liver, Biochem Pharmacol 48:59-64.

Hakkola J, Pelkonen O, Pasanen M, Raunio H. 1998a. Xenobiotic-metabolizing cytochrome P450 enzymes in the human fetoplacental unit: role in intrauterine toxicity. Crit Rev Toxicol 28:35-72.

Hakkola J, Tanaka E, Pelkonen O. 1998b. Developmental expression of cytochrome P-450 enzymes in human liver. Pharmacol Toxicol 82:209-217.

Harroff HH Jr. 1997. Use of animal models in pediatric research. Adv Pediatr 44:453489.

Hattis D, White P, Koch P, 1993. Uncertainties in pharmacokinetic modeling for perchloroethylene. II. Comparison of model predictions with data for a variety of different parameters, Risk Anal 13:599-610.

Hytten FE, Leitch I. 1971. The product of conception. In: The Physiology of Human Pregnancy. Oxford, UK:Blackwell Scientific Publications, 322.

Idanpaan-Heikkila J, Jouppila, P, Akerblom HK, Isoaho R, Kauppila E, Koivisto M. 1972. Elimination and metabolic effects of ethanol in mother, fetus, and newborn infant. Am J Obstet Gynecol 112:387-393.

Illsley NP, Hall G, Penfold P, Stacey TE. 1985. Diffusional permeability of the human placenta. Contrib Gynecol Obstet 13:92-97.

Imaoka S, Fujita S, Funae Y. 1991. Age-dependent expression of cytochrome P-450s in rat liver. Biochim Biophys Acta 1097:187-192.

Jahn F, Router A, Karge E, Danz M, Klinger W. 1993. Age dependent different influence of carbon tetrachloride on biotransformation of xenobiotics, glutathione content, lipid peroxidation and histopathology of rat liver. Exp Toxicol Pathol 45:101-107.

Juchau MR. 1980a. Extrahepatic drug biotransformation in the placenta. In: Extrahepatic Metabolism of Drugs and Other Foreign Compounds (Gram TE, ed). New York:SP Medical and Scientific Books, 211-238.

--. 19805. Drug biotransformation in the placenta. Pharmacol Ther 8:501-524.

--. 1995. Placental enzymes: cytochrome P400s and their significance. In: Placental Toxicology (Rama Sastry BV, ed). Boca Raton, FL:CRC Press, 197-212.

Juchau MR, Faustman-Watts E. 1983. Pharmacokinetic considerations in the maternal-placental-fetal unit. Clin Obstet Gynecol 26:379-390.

Jusko WJ. 1972. Pharmacokinetic principles in pediatric pharmacology. Pediatr Clin North Am 19:81-100.

Kearns GL, Reed MD. 1989. Clinical pharmacokinetics in infants and children. A reappraisal. Clin Pharmacokinet 17(suppl 1): S29-S67.

Kim CS, Binienda Z, Sandberg JA. 1996. Construction of a physiologically based pharmacokinetic model for 2,4-dichlorophenoxyacetic acid (2,4-D) dosimetry in the developing rabbit brain. Toxicol Appl Pharmacol 136:250-259.

Kivalo I, Saarikoski S. 1972. Placental transmission and foetal uptake of [sup.14]C-dimethyltubocurarine. Br J Anaesth 44: 557-561.

--. 1976. Placental transfer of [sup.14]C-dimethyltubocurarine during Caesarean section. Br J Anaesth 48:239-242.

Korinthenberg R, Haug C, Hannak D. 1994. The metabolism of carbamazepine to CBZ-10-11-epoxide in children from the newborn age to adolescence. Neuropediatrics 25:214-216.

Kostial K, Kello D, Jugo S, Rabar I, Maljkovic T. 1978. Influence of age on metal metabolism and toxicity, Environ Health Perspect 25:81-86.

Krauer B, Krauer F, Hytten FE. 1980. Drug disposition and pharmacokinetics in the maternal-placental-fetal-unit, Pharmacol Ther 10:301-328.

Krishnan K, Andersen ME, 1998. Physiological pharmacokinetic models in the risk assessment of developmental neurotoxicants. In: Handbook of Developmental Neurotoxicology (Sikker W, Chang L, eds). New York:Academic Press, 709-725.

Kuhnz W, Jager-Roman E, Rating D, Diechl A, Kunze J, Hedge H, et al. 1983. Carbamazepine and carbamazepine-10, 11-epoxide during pregnancy and postnatal period in epileptic mothers and their nursed infants: pharmacokinetics and clinical effects. Pediatr Pharmacol (New York) 3:199-208.

Kurata N, Nishimura Y, Iwase M, Fischer NE, Tang BK, Inaba T, et al. 1998. Trimethadione metabolism by human liver cytochrome P450: evidence for the involvement of CYP2E1. Xenobiotica 28:1041-1047.

LaCroix D, Sonnier M, Moncion A, Cheron G, Cresteil T, 1997. Expression of CYP3A in the human liver. Evidence that the shift between CYP3A7 and CYP3A4 occurs immediately after birth. Eur J Biochem 247:625-634.

Leakey JEA. 1983. Ontogenesis. In: Biological Basis of Detoxication (Caldwen J, ed). New York:Academic Press, 77-103.

Lehmann H, Cook J, Ryan E. 1957. Pseudocholinesterase in early infancy. Proc R Sac Med 50:147-150.

Levine RR. 1973. Pharmacology Drug Actions and Reactions. Boston:Little, Brown, 103.

Levitz M, Condon GP, Money WL, Dancis J. 1960. The relative transfer of estrogens and their sulfates across the guinea pig placenta: sulfurylation of estrogens by the placenta. J Biol Chem 235:973-977.

Levy G. 1981. Pharmacokinetics of fetal and neonatal exposure to drugs. Obstet Gynecol 58(suppl 5):S9-S16.

Levy G, Khanna NN, Soda DM, Tsuzuki O, Stern L. 1975. Pharmacokinetics of acetaminophen in the human neonate: formation of acetaminophen glucuronide and sulfate in relation to plasma bilirubin concentration and o-glucaric acid excretion. Pediatrics 55:818-825.

Lin JH. 1995. Species similarities and differences in pharmacokinetics. Drug Metab Dispos 23:1008-1021.

Luecke RH, Wosilait WD, Pearce BA, Young JF. 1994. A physiologically based pharmacokinetic computer model for human pregnancy. Terstology 49:90-103.

Maenpaa J, Bane A, Raunio H, Honkakoski P, Pelkonen O. 1993. Cytochrome P450 isoforms in human fetal tissues related to phenobarbital-inducible forms in the mouse. Biochem Pharmacol 45:899-907.

Mattsson JL, Maurissen JP, Nolan RJ, Brzak KA. 2000. Lack of differential sensitivity to cholinesterase inhibition in fetuses and neonates compared to dams pretreated perinetally with chlorpyrifos. Toxicol Sci 53:438-446.

Mihaly GW, Morgan DJ. 1983. Placental drug transfer: effects of gestational age and species. Pharmacol Ther 23:253-266.

Miller RK, Koszalka TR, Brent RL. 1976. The transport of molecules across placental membranes. In: The Cell Surface in Animal Embryogenesis and Development (Post G, Nicolson GL, eds). New York:Elsevier/North-Holland Biomedical Press, 145-223.

Millicovsky G, DeSesso JM. 1980. Differential embryonic cardiovascular responses to acute maternal uterine ischemia: an in viva microscopic study of rabbit embryos with either intact or clamped umbilical cords. Teratology 22:335-343.

Mirkin BL, Singh S. 1976. Placental transfer of pharmacologically active molecules. In: Perinatal Pharmacology and Therapeutics (Mirkin BL, ed). New York:Academic Press, 1-69.

Moghrabi N, Head JR, Andersson S. 1997. Cell type-specific expression of 17[beta]-hydroxysteroid dehydrogenase type 2 in human placenta and fetal liver. J Clin Endocrinol Metab 82:3872-3878.

Morselli PL. 1989. Clinical pharmacology of the perinatal period and early infancy. Clin Pharmacokinet 17(suppl 1):13-28.

Maser VC, Chanda SM, Mortensen SR, Padilla S, 1998. Age- and gender-related differences in sensitivity to chlorpyrifos in the rat reflect developmental profiles of esterase activities. Toxicol Sci 46:211-222.

Mulhall A, de Louvois J. Hurley R. 1983. Chloramphenicol toxicity in neonates: its incidence and prevention. Br Med J (Clin Res Ed) 287:1424-1427.

Murray GI, Foster CO, Barnes TS, Weaver RJ, Snyder CP, Ewen SW, et al. 1992. Cytochrome P4501A expression in adult and fetal human liver. Carcinogenesis 13:165-169.

Nakamura H, Tanaka E. Ishikawa A, Fukao K, Tsuji K, Ohkawa H, 1998. Age-related changes in hepatic drug-oxidizing activity using trimethadione as a probe drug in humans. Hepatol Res 12:85-92.

NAS. 1993. Pesticides in the Diets of Infants and Children. Washington, DC:National Academy of Sciences.

NRC (National Research Council). 2000. Scientific Frontiers in Developmental Toxicity and Risk Assessment. Washington, DC:National Academic Press.

Nau H, Liddiard C, 1978. Placental transfer of drugs during early human pregnancy. In: Role of Pharmacokinetics in Prenatal and Perinatal Toxicology (Neubert D, Merker H-J, Nau H, Langman J, eds). Stuttgart, West Germany:Theime, 465-481.

Nau H, Neubert D. 1978. Placental transfer of drugs during early human pregnancy. In: Role of Pharmacokinetics in Prenatal and Perinetal Toxicology (Neubert D, Merker H-J, Nau H, Langman J, eds). Stuttgart, West Germany:Theime, 13-44.

Nedelcheva V, Gut I. 1994. P450 in the rat and man: methods of investigation, substrate specificities and relevance to cancer. Xenobiotica 24:1151-1175.

Neims AH, Warner W, Loughnan PM, Aranda JV. 1976. Developmental aspects of the hepatic cytochrome P-450 monooxygenase system. Annu Rev Pharmacol Toxicol 16:427-445.

Noschel N, Peiker G, Voigt R, Meinhold P, Muller B, Schroder S, et al. 1980. Research on pharmacokinetics during pregnancy. Arch Toxicol Suppl 4:S380-S384.

O'Flaherty EJ. 1991. Physiologically based models for bone-seeking elements. II. Kinetics of lead disposition in rats. Toxicol Appl Pharmacol 111:313-331.

--.1995. Physiologically based models for bone-seeking elements. V. Lead absorption and disposition in childhood. Toxicol Appl Pharmacol 131:297-308.

Ohmori S, Nakasa H, Asanome K, Kurose Y, Ishii I, Hosokawa M, et al. 1998. Differential catalytic properties in metabolism of endogenous and exogenous substrates among CYP3A enzymes expressed in COS-7 cells. Biochim Biophys Asta 1380:297-304.

Pariente-Khayat A, Pans G, Rey E, Richard MO, D'Athis P, Moran C, et al. 1991. Caffeine acetylator phenotyping during maturation in infants. Pediatr Res 29:492-495.

Parsons WD, Aranda JV, Neims AH, 1976. Elimination of transplacentally acquired caffeine in fullterm neonates. Pediatr Res 10:333.

Parsons WD, Neims AH. 1978. Effect of smoking on caffeine clearance. Clin Pharmacol Ther 24:40-45.

Pelekis M, Gephart LA, Lerman SE. 2001. Physiological-model-based derivation of the adult and child pharmacokinetic intraspecies uncertainty factors for volatile organic compounds. Regul Toxicol Pharmacol 33:12-20.

Pelkonen O. 1977. Transplacental transfer of foreign compounds and their metabolism by the foetus. Prog Drug Metab 2:119-161.

--.1980. Biotransformation of xenobiotics in the fetus. Pharmacol Ther 10:261-281.

Pikkarainen PH, Raiha NC. 1967. Development of alcohol dehydrogenase activity in the human liver. Pediatr Res 1:165-168.

Pons G, Rey E, Carrier O, Richard MO, Moran C, Badoual J, et al. 1989. Maturation of AFMU excretion in infants. Fundam Clin Pharmacol 3:589-595.

Pounds JG, Marlar RJ, Allen JR. 1978. Metabolism of lead-210 in juvenile and adult rhesus monkeys (Macaca mulatta). Bull Environ Contam Toxicol 19:684-691.

Power GG, Longo LD. 1975. Fetal circulation rimes and their implications for tissue oxygenation. Gynecol Invest 6:342-355.

Pyonnonen S, Sillanpaa M, Fray H, Iisalo E. 1977. Carbamazepine and its 10,11-epoxide in children and adults with epilepsy. Eur J Clin Pharmacol 11:129-133.

Rane A, Tomson G. 1980. Prenatal and neonatal drug metabolism in man. Eur J Clin Pharmacol 18:9-15.

Rag HV, Ginsberg GL. 1997. A physiologically-based pharmacokinetic model assessment of methyl t-butyl ether in groundwater for a bathing and showering determination. Risk Anal 17:583-597.

Rasheed A, Hines RN, McCarver-May DG. 1997. Variation in induction of human placental CYP2E1: possible role in susceptibility to fetal alcohol syndrome. Toxicol Appl Pharmacol 144:396-400.

Retanasavanh D, Beaune P, Morel F, Flinois JP, Guengerich FP, Guillouzo A. 1991. Intralobular distribution and quantitation of cytochrome P-450 enzymes in human liver as a function of age. Hepatology 13:1142-1151.

Ravindranath V, Boyd MR. 1995. Xenobiotic metabolism in brain. Drug Metab Rev 27:419-448.

Renwick AG. 1998. Toxicokinetics in infants and children in relation to the ADI and TDI. Food Addit Contain 15(suppl):S17-S35.

Renwick AG, Borne JL, Walton K, 2000. An analysis of the need for an additional uncertainty factor for infants and children, Regul Toxicol Pharmacol 31:286-296.

Rudolph AM, Heymann MA. 1967. The circulation of the fetus in utero. Circ Res 21:163-184.

Sandberg JA, Binienda Z, Lipe G, Rose LM, Parker WB, Ali SF, et al. 1996. Placental transfer and fetal disposition of 2',3'-dideoxycytidine (ddC) and 2',3'-dideoxyinosine (ddl) in the rhesus monkey. Drug Metab Dispos 23:881-884.

Slikker W Jr. 1987. The role of metabolism in the testing of developmental toxicants. Regul Toxicol Pharmacol 7:390-413.

--. 1994. Placental transfer and pharmacokinetics of developmental neurotoxicants. In: Principles of Neurotoxicology (Chang LW, ed), New York:Marcel Dekker, 659-680.

--. 1997. The Developing Nervous System, Comprehensive Toxicology (Sipes IE, McQueen CA, Gandolfi AJ, eds). New York:Pergamon, 295-307.

Slikker W Jr, Hill DE, Young JF. 1982. Comparison of the transplacental pharmacokinetics of 17[beta]-estradiol and diethylstilbestrol in the subhuman primate. J Pharmacol Exp Ther 221:173-182.

Slikker W Jr, Miller RK. 1994. Placental metabolism and transfer role in developmental toxicology. In: Developmental Toxicology (Kimmel CA, Buelke-Sam J, eds). New York:Raven Press, 249-283.

Sonnier M, Cresteil T. 1998. Delayed ontogenesis of CYP1A2 in the human liver. Eur J Biochem 251:893-898.

Stevens JC, Shipley LA, Cashman JR, Vandenbranden M, Wrighton SA. 1993. Comparison of human and rhesus monkey in vitro phase I and phase II hepatic drug metabolism activities. Drug Metab Dispos 21:753-760.

Strange RC, Howie AE, Hume R, Metharoo B, Bell J, Hiley C, et al. 1989. The developmental expression of alpha-, mu- and pi-class glutethione S-transferases in human liver. Biochim Biophys Asta 993:186-190.

Szorady I, Santa A, Veress I. 1987. Drug acetylator phenotypes in newborn infants. Biol Res Pregnancy Perinatol 8:23-25.

Tanaka E. 1998. In viva age-related changes in hepatic drug-oxidizing capacity in humans. J Clin Pharm Ther 23:247-255.

Treluyer JM, Cheron B, Bonnier M, Cresteil T. 1996. Cytochrome P450 expression in sudden infant death syndrome. Biochem Pharmacol 92:497-504.

Treluyer JM, Jacqz-Aigrain E, Alvarez F, Cresteil T. 1991. Expression of CYP2D6 in developing human liver. Eur J Biochem 202:583-588.

U.S. EPA. 1989. Risk Assessment Guidance for Superfund, Vol I. Human Health Evaluation Manual, Part A. EPA/540/1-89/002. Washington, DC:U.S. Environmental Protection Agency.

--. 1992. Draft Report: A Cross-Species Scaling Factor for Carcinogen Risk Assessment Based on Equivalence of mg/kg/day. Fed Reg 57:24152-24173.

--. 1994. Guidance Manual for the Integrated Exposure Uptake Biokinetic Model for Lead in Children. EPA 540-R-93081. PB93-963510, Washington, DC:U.S. Environmental Protection Agency.

--. 2000. IRIS file for Vinyl Chloride and Toxicology Review of Vinyl Chloride in Support of Summary Information on IRIS. EPA/635R-00/004, Washington, DC:U.S. Environmental Protection Agency.

Vest MF. 1965. The development of conjugation mechanisms and drug toxicity in the newborn. Biol Neonat 8:258-266.

Vieira I, Sonnier M, Cresteil T. 1996. Development expression of CYP2E1 in the human liver. Hypermethylation control of gene expression during the neonatal period. Eur J Biochem 238:47-483.

Waddell WJ, Marlowe C. 1981. Transfer of drugs across the placenta. Pharmacol Ther 14:375-390.

Watanabe J, Asaka Y, Kanamura S. 1993. Postnatal development and sublobular distribution of cytochrome P-450 in rat liver: a microphotometric study. J Histochem Cytochem 41:397-400.

Welfare MR, Bassendine MF, Daly AK. 2000. The effect of NAT2 genotype and gender on the metabolism of caffeine in nonsmoking subjects. Br J Clin Pharmacol 49:240-243.

Welsch F, 1982. Placental and fetal uptake of drugs. J Vet Pharmacol Ther 5:91-104.

Welsch F, Blumenthal GM, Conolly RB. 1995. Physiologically based pharmacokinetic models applicable to organogenesis: extrapolation between species and potential use in prenatal toxicity risk assessments. Toxicol Lett 82-83:539-547.

Wilson JG, Ritter EJ, Scott WJ, Eradkin R. 1977. Comparative distribution and embryotoxicity of acetylsalicylic acid in pregnant rats and rhesus monkeys. Toxicol Appl Pharmacol 41:67-78.

Yang HL, Namkung MJ, Juchau MR. 1995. Expression of functional cytochrome P-450A1 in human embryonic hepatic tissues during organogenesis. Biochem Pharmacol 49:717-726.

Yoo J-SH, Ning SM, Patten C J, Yang CS. 1987. Metabolism and activation of N-nitrosodimethylamine by hamster and rat microsomes: comparative study with weanling and adult animals. Cancer Res 47:992-998.

Ziegler EE, Edwards BB, Jensen RL, Mahaffey KR, Fomon SJ. 1978. Absorption and retention of lead by infants. Pediatr Res 12:29-34.

Zsigmond EK, Downs JR. 1971. Plasma cholinesterase activity in newborns and infants. Can Anesth Sac J 18:279-285.

Zusterzeel PL, Peters WH, De Bruyn MA, Knapen MF, Merkus HM, Steegers EA. 1999. Glutathione S-transferase isoenyzmes in decidua and placenta of preeclamptic pregnancies. Obstet Gynecol 94:1039-1038.

Zwiener RJ, Ginsburg CM 1988. Organophosphate and carbamate poisoning in infants and children. Pediatrics 81:121-126.

Gary Ginsberg (1), William Slikker, Jr. (2), James Bruckner (3), and Babasaheb Sonawane (4)

(1) Connecticut Department of Public Health, Hartford, Connecticut, USA; (2) National Center for Toxicological Research, U.S. Food and Drug Administration, Jefferson, Arkansas, USA; (3) University of Georgia, Athens, Georgia, USA; (4) National Center for Environmental Assessment, Office of Research and Development, U.S. Environmental Protection Agency, Washington, DC, USA

This article is part of the mini-monograph "Assessing Risks in Children from Exposure to Environmental Agents."

Address correspondence to G. Ginsberg, Connecticut Dept. of Public Health, 410 Capitol Ave., Mail Stop 11CHA, Hartford, CT 06134 USA. Telephone: [phone]. Fax: [phone]. E-mail: [e-mail]

The views expressed in this article are those of the authors and do not necessarily reflect the views or policies of the State of Connecticut, the National Center for Toxicological Research, the University of Georgia, or the U.S. Environmental Protection Agency.

The authors declare they have no competing financial interests.

Received 23 September 2002; accepted 19 May 2003.