Evaluation of Child/Adult Pharmacokinetic Differences from a Database Derived from the Therapeutic Drug Literature

Gary Ginsberg*,1, Dale Hattis{dagger}, Babasaheb Sonawane{ddagger}, Abel Russ{dagger}, Prerna Banati{dagger}, Mary Kozlak{dagger}, Susan Smolenski* and Rob Goble{dagger}

* Connecticut Department of Public Health, P.O. Box 340308, Mail Stop 11CHA, Hartford, Connecticut 06134; {dagger} Clark University, Center for Technology, Environment & Development, Worcester, Massachusetts; and {ddagger} National Center for Environmental Assessment, Research and Development, U.S. EPA, Washington, DC

Received September 7, 2001; accepted December 17, 2001


    ABSTRACT
 TOP
 ABSTRACT
 INTRODUCTION
 DISCUSSION
 REFERENCES
 
Pharmacokinetics (PK) of xenobiotics can differ widely between children and adults due to physiological differences and the immaturity of enzyme systems and clearance mechanisms. This makes extrapolation of adult dosimetry estimates to children uncertain, especially at early postnatal ages. While there is very little PK data for environmental toxicants in children, there is a wealth of such data for therapeutic drugs. Using published literature, a Children's PK Database has been compiled which compares PK parameters between children and adults for 45 drugs. This has enabled comparison of child and adult PK function across a number of cytochrome P450 (CYP) pathways, as well as certain Phase II conjugation reactions and renal elimination. These comparisons indicate that premature and full-term neonates tend to have 3 to 9 times longer half-life than adults for the drugs included in the database. This difference disappears by 2–6 months of age. Beyond this age, half-life can be shorter than in adults for specific drugs and pathways. The range of neonate/adult half-life ratios exceeds the 3.16-fold factor commonly ascribed to interindividual PK variability. Thus, this uncertainty factor may not be adequate for certain chemicals in the early postnatal period. The current findings present a PK developmental profile that is relevant to environmental toxicants metabolized and cleared by the pathways represented in the current database. The manner in which this PK information can be applied to the risk assessment of children includes several different approaches: qualitative (e.g., enhanced discussion of uncertainties), semiquantitative (age group-specific adjustment factors), and quantitative (estimation of internal dosimetry in children via physiologically based PK modeling).

Key Words: children; metabolism; pharmacokinetics; risk assessment.


    INTRODUCTION
 TOP
 ABSTRACT
 INTRODUCTION
 DISCUSSION
 REFERENCES
 
The risks that children incur from environmental chemicals may differ from risks in adults as a result of a number of factors: (1) exposure may be greater in young children due to a greater inhalation rate and food ingestion rate (particularly for certain foods) per body weight, and greater contact with soil, house dust, and other media which may contain contaminants (NRC, 1993Go; U.S. EPA, 1997Go); (2) once exposure has occurred, the pharmacokinetic (PK) handling of xenobiotics is likely to differ from that in adults with respect to their metabolism, clearance, protein binding, and volume of distribution (Besunder et al., 1988Go; Morselli, 1989Go; Kearns and Reed, 1989Go); (3) pharmacodynamic (PD) differences in which the sensitivity of rapidly developing tissues/systems in neonates and young children may differ from that in adults (Faustman et al., 2000Go; Pope et al., 1991Go; Vesselinovitch et al., 1979Go).

While all 3 areas of child-adult differences need to be addressed, the focus of the current paper is the second area, the potential for PK differences to play a significant role in modifying risks to children. Risk assessments have increasingly relied upon physiologically based pharmacokinetic (PBPK) models to adjust for PK differences between test animals and humans. This acknowledges that the relationship between administered dose and effective internal dose can differ across species, with this difference having significant implications for risk assessment.

While such refinements may have removed some of the uncertainty in cross-species extrapolations, risk assessments have yet to account for child-adult differences in PK of xenobiotics. This is an important issue given the developmental program of children, in which many physiologic and metabolic systems are immature in the first months after birth and change rapidly throughout childhood. For this reason, internal dose estimates developed for adults may not apply to young children, which is an area of uncertainty in extrapolating risks from adults to children. Due to the variety of physiologic stages that children pass through over the course of the first months and years of life, it is unlikely that a single extrapolation or PBPK model would suffice to describe the range of children's PK function. Rather, it is important to consider chemical dosimetry in the various developmental stages of childhood in relation to adults.

PK differences across individuals are currently represented in the risk assessment process for noncarcinogens as part of the default (generally 10-fold) interindividual uncertainty factor. This 10x factor can be seen as comprising equally a half-log (3.16x) PK component and a similar half-log PD component (IPCS, 2001Go; Renwick, 1998Go). The 3.16x PK factor is expected to account for racial, ethnic, gender, genetic, and age differences, as well as interindividual differences due to disease states and intake of drugs. Elsewhere we have analyzed the general protectiveness of the 10-fold uncertainty factor used to represent interindividual variability in noncancer risk assessments (Hattis et al., 1999Go). The current research extends that analysis by evaluating the suitability of the 3.16-fold PK factor for capturing interindividual variability when children are taken into account. In cases where child/adult differences appear larger than this factor, the risk implications need to be examined in terms of (1) period of time for which this difference exists relative to the averaging time for the RfC; (2) whether parent compound or some metabolite is responsible for toxicity and how each is affected by child/adult PK differences; (3) the effect of interchild variability on the number of children that may be outside of the standard PK uncertainty factor; and (4) how child/adult PK factors can be taken into consideration for cancer risk assessment where uncertainty factors are traditionally not used.

Research Goals and Approach
To facilitate extrapolations of risks between adults and children, our primary goal is to compare children at different stages of development with adults in terms of PK handling of xenobiotics. To accomplish this goal, a Children's PK Database was constructed from the therapeutic drug literature (available on the web at http://www2.clarku.edu/faculty/dhattis). The pharmacokinetics of many drugs have been evaluated in children to determine the dose levels most appropriate for specific ages (Anderson et al., 1997Go; Kearns and Reed, 1989Go). When combined with adult PK studies of the same drugs, comparisons can be made between between adult and child PK measurements. By evaluating PK datasets across chemicals with differing structures and clearance mechanisms (e.g., renal excretion, Phase I oxidative metabolism, Phase II conjugative metabolism), it is anticipated that child/adult differences in key metabolism and elimination pathways will become evident. D. Hattis et al. (submitted) will evaluate PK interindividual variability to determine whether children are more variable than adults, even when grouped within relatively small age categories.

Methods Used to Develop and Analyze the Children's PK Database
This study involved the following steps: identification of therapeutic drugs having pertinent PK data for children and adults in the published literature, obtaining and evaluating the primary studies to extract key data, organizing the data into a spreadsheet database, analyzing the mean data across age groups (means analysis), and analyzing the interindividual variability within age groups (variability analysis). The key data obtained from reviewed studies included: number, gender, and age of subjects, dose route and number of doses, and PK findings describing half-life, clearance, volume of distribution, and peak and area-under-the-concentration x time curve (AUC) blood concentrations. Individual subject data were logged into the database where available; in lieu of this, the reported mean, SD, and number of subjects were entered. These efforts are further described in the following section.

While it would have been optimal to obtain comparative children's and adult PK datasets for environmental toxicants, our searches found very little useful data in this area. The therapeutic drug literature became the primary focus because it contains numerous datasets of high quality that provide PK data for a range of developmental periods. The database does include limited PK data for 3 chemicals (chloral hydrate, dichloroacetic acid, trichloroethanol) that are involved in trichloroethylene and tetrachloroethylene metabolism. Data for these chemicals stem from their clinical uses in pediatric populations. The database also includes several "lifestyle drugs" (nicotine, cotinine, alcohol), with half-life data for these coming from maternal/newborn elimination studies immediately after birth.

This article represents a summary and analysis of the Children's PK Database. Data records for individual chemicals, including a complete bibliography for all PK studies relied upon, can be found elsewhere (Hattis et al., 2000Go; http://www2.clarku.edu/faculty/dhattis).

Compilation and description of the database.
Computerized literature searches (Toxline, Medline) were conducted to find references to publications describing pharmacokinetics in children, using search words such as neonate, infant, and children and crossing these with terms such as pharmacokinetics, metabolism, distribution, excretion. Additionally, a variety of pediatric pharmacology reviews (Anderson, 1997Go; Jacqz-Aigrain and Burtin, 1996Go;Renwick, 1998Go; Yaffe and Aranda, 1992Go) were examined to identify chemicals for which PK datasets exist for children. Through these sources, approximately 100 chemicals (mostly therapeutic drugs) were identified as having at least some PK data in children. A subset of 45 chemicals was selected for more detailed analysis based upon being able to obtain the primary data sources, these sources having a reasonable number of subjects (at least 4 per age group), the availability of data for pertinent age groups (especially very early life stages), and the goal of having a variety of clearance mechanisms represented.

The list of chemicals contained in the database and major clearance mechanisms are shown in Table 1Go. Clearance mechanisms were identified from pharmacology texts and reviews (Bertz and Granneman, 1997Go; Dollery, 1999Go; Goodman-Gilman, 1990Go), as well as from primary literature as shown in the table. Chemicals cleared primarily by cytochrome P450 (CYP) pathways are generally excreted as metabolites in urine or bile, while chemicals listed as "renal" are primarily excreted unchanged in urine. Chemicals listed under the glucuronide or other conjugation pathways undergo minimal Phase I metabolism but can be directly conjugated in Phase II reactions. Most chemicals are cleared by multiple pathways but typically 1 or 2 routes of metabolism and elimination predominate. Chemicals have been assigned to a particular pathway if the major percentage of chemical disposition (> 50%) is via that pathway. This is analogous to the approach adopted by Dorne et al. (2001) for analyzing CYP1A2 function in newborns and other groups of children.


View this table:
[in this window]
[in a new window]
 
TABLE 1 Clearance Pathways for Chemicals in Children's PK Database
 
The table indicates that of the 45 chemicals contained in the database, 8 chemicals are excreted primarily unchanged in urine and so half-life or clearance data on these chemicals can be indicative of renal function. Chemicals for which CYP Phase I metabolism predominates are well represented in the database, with 18 having some form of CYP as the primary route of disposition. The most prevalent CYP substrates are those chemicals metabolized by the CYP3A family, which is the predominant class of CYP in human liver. Other types of CYPs for which information can be gained from chemicals present in the database are CYP1A2, and the CYP2C family. For some drug substrates, the specific CYPs involved have not been sufficiently elucidated or multiple CYPs and/or other pathways appear to be involved (e.g., antipyrene, mepivacaine, bupivacaine, ropivacaine, thiopentone, amobarbital). In these cases, the substrate was assigned to 1 of 2 more generalized categories (i.e., "miscellaneous CYPs" or "unclassified") for the purpose of the present analysis.

Regarding biliary function, bromosulfophthalein (BSP) is the best indicator chemical in the current database given that at least some of dosed BSP is excreted unchanged in bile. BSP has traditionally been used as a clinical marker of hepato/biliary function. However, a portion of dosed BSP is conjugated with glutathione prior to excretion in bile so the overall BSP clearance from blood is actually a composite of these 2 pathways (GSH conjugation, biliary clearance of unchanged BSP; Fehring and Ahokas, 1989Go). Two chemicals whose clearance is indicative of alcohol dehydrogenase activity (ethanol, chloral hydrate) are also part of the database.

A number of the chemicals are cleared primarily via Phase II conjugation pathways without the need for prior Phase I metabolism. Glucuronidation substrates total 6, while sulfation, glutathione and N-acetylation substrates are also contained in the database. In some cases these assignments may best be described as general Phase II substrates rather than substrates for a single conjugation pathway. This is exemplified best by paracetamol (acetaminophen), which is a candidate for glucuronidation and sulfation. In this case, a shift in pathways occurs if the preferred cofactor is deficient (Besunder et al., 1988Go; Levy et al., 1975Go).

The children and adult PK data relating to chemical half-life, blood clearance, volume of distribution, AUC and peak concentration (normalized to administered dose) were assembled in Microsoft Excel worksheets, first on an individual study basis, and then compiled across studies into a central database. Datasets were included if they were based on measurements in at least 4 subjects, with a distinct preference for datasets that also provided some measure of dispersion such as a SD or a SE (or, even more preferred, individual values for which distributional statistics could be calculated). The central database contains 366 lines of PK data, each line presenting the arithmetic mean value for a given parameter (e.g., half-life) for a particular drug/age group combination.

The composition of the central database is further described in Table 2Go, which shows the amount of data for each age group analyzed. Age groupings were developed in an effort to capture the rapid physiological and biochemical/metabolism changes that occur in the first weeks to months of life. Given the evidence that changes occur in liver CYP enzymes and other systems within the first hours to days of life (Cresteil, 1998Go; Tanaka, 1998bGo), it was especially important to obtain data for these early stages. Further, given the fact that premature infants have many underdeveloped systems at birth, it was important to evaluate them as a separate group. The age groupings beyond the neonatal period were intended to represent a developmental continuum, although the number of groups possible and the age cutoffs were a function of the data available for these ages (i.e., there would be no point in having an age group with too few data points for statistical evaluation). A large age group encompassing 10 years was used to characterize children beyond the toddler stage up until adolescence. This appeared reasonable based upon initial judgements that there was not a large degree of variability in the PK parameters under study over this broad age range. Further, there was not much data available for older children for the chemicals included in this database. Of course, individuals within any age group are at slightly different developmental stages. These interindividual differences are evaluated in the variability analysis presented by D. Hattis et al. (submitted).


View this table:
[in this window]
[in a new window]
 
TABLE 2 Data Available for Diferent Age Groups in the Children's PK Database
 
Table 2Go indicates that the central database contains substantial amounts of data for most age groups in terms of numbers of studies, chemicals, and metabolism/elimination pathways represented. The 2 exceptions are the premature neonate and adolescent groups, which contain only 12 to 15 lines of data. This makes analysis of these 2 groups more uncertain. The comparison group, adults, contains 118 data records covering 42 chemicals and 11 different metabolism/elimination routes. The body of reference data generally comes from studies of young, healthy adult subjects. It is noted that there are 3 chemicals for which adult data could not be found. This precludes child/adult comparisons in these cases, but they still enable comparisons across several child age groups and so were kept in the database. In terms of PK endpoints (Table 3Go), half-life data are the most common in the central database (147 records encompassing 41 chemicals) with considerable data also for clearance (104 records, 27 chemicals) and volume of distribution (Vd; 70 records, 25 chemicals). However, there were too few records for the blood concentration parameters (AUC and maximum concentration) to make a detailed analysis of these parameters worthwhile.


View this table:
[in this window]
[in a new window]
 
TABLE 3 Data Available for Various PK Parameters in the Central Database
 
Analysis of PK differences across age groups using regression techniques.
While across age differences are readily apparent by examination of the records for specific chemicals, a method was needed for quantitative analysis of the broader trends across groups of chemicals in the database. Multiple regression analysis (SAS, Inc. JMP, Version 3.1.6) was used to assess relationships between age group and log mean PK parameter value (e.g., mean half-life or clearance) across chemicals.

The general regression equation used for this purpose is as follows:

In this model, the chemical-specific "B's" correct for differences among chemicals in average clearance (or other parameter) relative to a specific reference chemical (theophylline was arbitrarily chosen). The regression equation thus involved 44 B coefficients specific for each chemical in the database (excluding the reference chemical) and a dummy variable (0 or 1) was used to identify which chemical gave rise to each datapoint (e.g., if fentanyl was administered to children, then the fentanyl B coefficient would have a dummy variable set to 1 in the above equation and all other chemicals would have their dummy variable set to 0 for that data point). The resulting B coefficient for each chemical is the best estimate of the log of the ratio of the adult parameter value for the specific chemical relative to theophylline. This normalization of half-life or clearance against a reference chemical allows inherent differences among chemicals for these parameter values to have minimal impact on the analysis. This in turn facilitates our ability to detect systematic differences across age groups, as age group can then become the main predictor of log mean half-life or clearance. To accomplish this the regression equation also used dummy variables to normalize for age group relative to adults (i.e., 1 if datapoint is for the particular age group being targeted, 0 if datapoint does not correspond to that age-specific B regression coefficient).

Each regression run provided central estimates of the "B" coefficients for a particular age group together with conventional standard errors, t statistics, and p values. The regression coefficient represents that age group's composite (across all chemicals) PK parameter ratio compared to adults. Since the regression equation is performed as a Log function, the antilog of the regression coefficient provides the geometric mean child/adult ratio.

The regression analysis was run in several formats to test the effects of prioritizing or weighting those data groups that had larger numbers of subjects (n) or less variability. Without such weighting, the regression technique gives equal weight to studies of varying robustness and statistical confidence. To weight according to number of subjects per study group, the square root of n was used. To weight according to variability an "inverse variance weight" (Mean/SE)2 was developed for each data group. This statistic is an indicator of the coefficient of variance and so is a useful index of the spread of the data around the mean value. Regression runs utilizing the inverse variance weighting scheme performed the best in terms of the resulting r2 and p values for the age group coefficients. Thus, the results with this method are shown. However, the results using the other weighting approaches were qualitatively very similar. Greater detail on these and other statistical approaches used is described in D. Hattis et al. (submitted), and the full results for each statistical weighting procedure are presented at the web address cited above.

The database consists primarily of drug studies involving parenteral or po dosing. Across age comparisons were made based upon studies using a similar mode of dosing so that factors such as bioavailability and first pass effects would not confound these comparisons.

Results of Child/Adult PK Comparisons
General observations.
The Central Database was constructed to compile PK data for chemicals that have data in both children and adults, with emphasis placed on organizing the data by discrete age groups. A scan of the database shows that for many chemicals early life stages (premature neonates, full-term neonates, newborns 1 week to 2 months) appear different than adults in terms of clearance, half-life, and volume of distribution. Table 4Go was extracted from the Central Database to illustrate this pattern and to also show how data records have been compiled and organized. The table shows the means and SE data for a single chemical, alfentanil, a CYP3A4 substrate.


View this table:
[in this window]
[in a new window]
 
TABLE 4 Section of the Children's PK Cental Database
 
Alfentanil clearance is slower and half-life is longer in premature and full-term neonates as compared to older children and adults. While immaturity of the CYP3A4 pathway may have produced some of the lengthening in alfentanil half-life, the Vd data at the bottom of the table suggest that this parameter may have also played a role. A large Vd would tend to make the chemical less available for elimination because it would be distributed more extensively into compartments not directly involved in elimination. As seen in Table 4Go, alfentanil has a higher Vd in premature infants and neonates than in older age groups. The potential influence of Vd on half-life comparisons across chemicals and age groups is evaluated further in the section below, Potential Influence of Vd and Hepatic Extraction on Results.

Comparison across age groups with regression analysis.
The regression analysis examined the trends exemplified in Table 4Go by analyzing group mean data simultaneously across all chemicals and age groups in the database. The alfentanil example in Table 4Go illustrates the problem that for a particular chemical there will typically be age groups not represented. This is overcome by combining data across numerous chemicals such that all age groups become part of the analysis. Figure 1Go provides an overview of the developmental changes in drug half-life for the 40 chemicals in the database for which half-life data are available. The data indicate that across this array of chemicals, premature neonates tend to have a 4-fold longer half-life than adults, with the difference somewhat smaller but still significant for full-term neonates and the 1-week to 2-month age group. The older children's groups were not different than adults although there was a trend towards shorter half-life in the 6-month to 2-year group.



View larger version (23K):
[in this window]
[in a new window]
 
FIG. 1. Half-life results for the full database (40 substrates). ***p < 0.001.

 
Figure 2Go shows the same age group comparison but for clearance/kg body weight instead of half-life. Clearance data were less commonly reported (or readily calculated) from drug pharmacokinetic studies than half-life data so that there are only 27 drugs represented in Figure 2Go as compared to 40 with half-life data in Figure 1Go. Figure 2Go shows age-related changes in total body clearance corrected for body weight. The expectation that half-life and clearance will be inversely related is borne out as the age groups that had larger t1/2 (Fig. 1Go) have lower clearance rates (Fig. 2Go). The size of the child/adult differential is somewhat smaller than for half-life, with clearance 1.5- to 2-fold lower in neonates, 1.3-fold lower in infants out to 2 months of age, and then rising to 1.7-fold greater than adult clearance in the 6-month to 2-year group. Due to the greater amount of half-life data available for individual chemicals and pathways of interest, the remaining figures and text focus upon the half-life results. Half-life and clearance results for all drugs and pathways in the database are available on our website cited above.



View larger version (30K):
[in this window]
[in a new window]
 
FIG. 2. Clearance results for the full database (27 substrates). *p < 0.01; **p < 0.0001.

 
Figures 3 through 6GoGoGoGo summarize the half-life results for the individual pathways analyzed. Across the 7 renally cleared drugs, longer half-life (approximately 3-fold greater than adult) was seen in the youngest age groups, with this difference not apparent by 2–6 months (Fig. 3Go). In fact, the half-life of these chemicals became significantly faster than adults in the 2–12 year old group. A similar profile is seen in Figure 4Go for the CYP1A2 substrates. Here there is considerable variability across the 2 drugs for which child and adult data are available. The neonate/adult differential is appreciably larger for caffeine (13x) than for theophylline. Once again, half-life results for children appear to become much like the adult value by 2–6 months, with shorter half-lives suggesting more rapid clearance in the older child age groups. The results for CYP3A are qualitatively similar, with average differences compared to adult in the 2–5 fold range in the earliest age groups, but with half-lives close to or even below adult in the later age groups (Fig. 5Go).



View larger version (23K):
[in this window]
[in a new window]
 
FIG. 3. Half-life results for for renally cleared substrates (ampicillin, cimetidine, furosemide, piperacillin, ticarcillin, tobramycin, and vancomycin). **p < 0.05; ***p < 0.01.

 


View larger version (17K):
[in this window]
[in a new window]
 
FIG. 4. Half-life results for CYP1A2 substrates (caffeine and theophylline). #, SE = 21.63. *p < 0.1; **p < 0.05; ***p < 0.01; ****p < 0.0001.

 


View larger version (14K):
[in this window]
[in a new window]
 
FIG. 5. Half-life results for CYP3A substrates (alfentanil, carbamazepine, fentanyl, lignocain, midazolam, nifedipine, quinidine, triazolam). *p < 0.1; **p < 0.05.

 


View larger version (15K):
[in this window]
[in a new window]
 
FIG. 6. Half-life results for all substrates involving CYPs (18 substrates). *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

 
The database contains 1 probe substrate for the CYP2C family, tolbutamide. This agent's metabolism involves both CYP2C9 and 2C19 (Lasker et al., 1998Go; Wester et al., 2000Go). The results for tolbutamide follow that seen for other CYP pathways with its half-life 3.4-fold longer in full-term neonates than in adults (data not shown). Barbiturates are also known to be substrates for CYP2C19 (Hadama et al., 2001Go; Kobayashi et al., 2001Go). However, a variety of other CYPs and direct conjugation pathways also contribute to the disposition of this class of sedatives, precluding their classification into a specific pathway. The half-life and clearance data for amobarbital, a representative member of the barbiturates in the current dataset, are consistent with the results found for tolbutamide in that full-term neonates have 2.5-fold longer half-life than adults (data not shown). While this adds some support to the evidence that CYP2C function is deficient in neonates, this finding is not as specific to the CYP2C family as in the case of tolbutamide.

In addition to amobarbital, there are numerous drug substrates in the database whose metabolic fate involves multiple CYPs or CYPs that are not identified (antipyrene, dichloroacetate, ketamine, bupivicaine, mepivacaine, ropivacaine). These substrates were combined with the others described above for specific CYPs to create a subclassification that is all substrates whose major route of disposition is via CYPs. This subgroup thus provides information on age-related differences in metabolism for a broad array of CYPs and drug substrates. Figure 6Go shows that this "all CYPs" group has a similar developmental pattern as seen with specific CYPs: a child/adult half-life ratio of 2- to 5-fold in the youngest age groups out to 2 months of age with the ratio approaching unity in the 2- to 6-month age group and then children having somewhat shorter half-lives than adults beyond that time. This comports with clearance rates being decreased by 1.7-fold for this "all CYPs" subgroup during the first 2 months of life and then reaching and surpassing adults within the first year of life (data not shown). Figure 7Go shows the age-related changes in glucuronidation capacity, with the trend very similar to what was seen with CYPs and renal clearance.



View larger version (27K):
[in this window]
[in a new window]
 
FIG. 7. Half-life results for glucuronidation substrates (lorazepam, morphine, oxazepam, trichloroethanol, valproic acid, zidovudine). **p < 0.05; ***p < 0.01.

 
Potential influence of Vd and hepatic extraction on results.
Figure 8Go provides an overview of Vd changes across the 22 chemicals in the database for which this parameter was reported. The trend is for children to have greater Vd than adults, with little apparent difference among the children's age groups. While the trend for larger Vd is fairly consistent across these age groups, none of the groups are significantly different from adults at the p < 0.05 level. However, the generally higher Vd values in children may tend to lengthen chemical half-life, thus contributing to the longer half-lives shown for a number of age groups in the earlier figures.



View larger version (24K):
[in this window]
[in a new window]
 
FIG. 8. Volume of distribution results for the full database (22 substrates). *p < 0.1.

 
To evaluate this more closely, Vd and t1/2 data were plotted for the chemical with the greatest amount of both types of data in the database, theophylline. The data in Figure 9Go show the child/adult ratio for theophylline Vd and t1/2 across 5 different age groups. The data indicate greater child/adult differences in t1/2 than in Vd through 2 months of age. Further, the t1/2 profile shows large declines towards and then below adult levels beyond 2 months, a period during which Vd changes relatively little. These data suggest that while larger Vd may have some influence on theophylline t1/2 in children, other factors (e.g., immaturity of CYP1A2) appear to play a larger role. This is consistent with Vd versus t1/2 comparisons for other chemicals in the database: caffeine has a much larger neonate/adult ratio for t1/2 (13) than for Vd (1.6), while for lorazepam, the Vd ratio is 0.66 in neonates (less distribution than in adults) and yet the lorazepam t1/2 is still longer in neonates than adults (ratio = 2.6).



View larger version (13K):
[in this window]
[in a new window]
 
FIG. 9. Theophylline Vd and t1/2 across the age groups.

 
Another factor that can affect hepatic clearance is blood flow to the liver. Chemicals with high hepatic extraction are likely to have clearance limited by the rate of hepatic blood flow. This factor can minimize interindividual differences in clearance that might otherwise occur due to genetic polymorphisms, induction of liver enzymes, or the immaturity of hepatic systems (Kedderis, 1997Go). Children in general have a larger liver size and hepatic blood flow per body weight than adults (Gibbs et al., 1997Go), which tends to increase hepatic clearance of chemicals. To determine how the results of the Children's PK Database may be influenced by hepatic blood flow, chemicals that are cleared by the liver were divided into high clearance (greater than 0.5 l/min as measured in adults), intermediate clearance (0.2 to 0.5 l/min in adults), and low clearance (less than 0.2 l/min in adults).

Table 5Go shows child/adult t1/2 ratios for low as opposed to high clearance chemicals. High and intermediate clearance chemicals were combined into a third category since both types of chemicals should experience at least some flow restriction and the addition of intermediate chemicals substantially increases the number of chemicals in the analysis. The table shows that child t1/2 values are generally 2 to 5-fold higher in the youngest age groups compared to adults, with no significant differences between low and high clearance chemicals. However, in the 1-week to 2-month group the data segregate such that low clearance chemicals are statistically different than high or high/intermediate chemicals At this age, the low clearance chemicals have larger t1/2 ratios suggesting that the lack of flow restriction for these chemicals allows for the full child/adult differential to be seen. The data gathered to date do not reveal any other significant differences across low versus high clearance chemicals in the other age groups where these comparisons are possible.


View this table:
[in this window]
[in a new window]
 
TABLE 5 Separation of Database into Low, Intermediate, and High Clearance Chemicals
 

    DISCUSSION
 TOP
 ABSTRACT
 INTRODUCTION
 DISCUSSION
 REFERENCES
 
Risk assessments for children would ideally incorporate pediatric PK data for the environmental chemical(s) under analysis. However, such PK data are typically not available in children for obvious ethical reasons. The approach taken in this study is to partially fill this data gap by utilizing the extensive child/adult PK database for therapeutic drugs. Many of these drugs can be seen as indicator chemicals for particular metabolism/clearance pathways because the majority of the dose is handled by identified CYPs or other routes (e.g., glucuronidation, renal elimination). Greater confidence about conclusions for a particular pathway has been obtained by analyzing a number of chemicals, all of which are metabolized by that pathway.

Even though the analysis for some pathways was limited by the small number of chemicals represented, the developmental pattern seen in each pathway was reinforced by what was found with other pathways and with the broader analyses of the database (Fig. 1Go: all substrates in database with half-life data, n = 40; Fig. 2Go: all substrates with clearance data, n = 27; Fig. 6Go: all substrates metabolized by CYPs, n = 18). This pattern indicates immature PK systems at birth, with premature neonates generally having the longest drug half-lives (3–5 times longer than adults on average), and full-term neonates also significantly different than adults (generally 2 to 3 times longer, although 9 times longer for CYP1A2—see Fig. 4Go). These mean neonate/adult t1/2 differences can exceed the 3.16-fold PK uncertainty factor often attributed to interindividual variability. It should be noted that the child/adult ratio data presented in Figures 1–8GoGoGoGoGoGoGoGo represent the geometric mean (gm) and thus are not as strongly influenced by high-end results as would the arithmetic mean ratio. As shown by D. Hattis et al. (submitted), exceedance of the 3.16-fold uncertainty factor is even more evident when considering the full range of variability during early life stages in handling the chemicals represented in the PK database. Thus, certain age group/drug combinations can have a large number of individuals that differ from the adult t1/2 value by more than 3.16-fold.

It should be noted that comparing the size of child/adult differences relative to the default PK uncertainty factor for noncancer assessment (3.16-fold) may not be the major determinant of whether a separate children's PK factor is warranted. This default uncertainty factor is intended to cover a wide array of interindividual differences in the population that cannot readily be accounted for but which contribute to the overall PK uncertainty (genetic polymorphisms, gender, disease states, nutritional status, coexposure to other agents, etc.). Where data exist showing that a specific subpopulation (e.g., neonates) have an identifiable PK differential relative to adults, the risk assessor may choose to utilize this information in the form of age group-specific adjustment factors (IPCS, 2001Go) that can increase or decrease internal exposures and risks. These approaches are also of relevance to cancer risk assessment as this type of assessment does not involve uncertainty factors to account for interindividual variability. Thus, the types of child/adult differences described presently and by others (Dorne et al., 2001Go; Renwick, 1998Go; Renwick et al., 2000Go) can assist cancer risk assessments address an important source of uncertainty (children's internal exposure and risk).

Neonate/adult differences are most meaningful to risk assessment for effects that can result from short-term dosing in early life. This is especially the case where there is a critical window of sensitivity that overlaps with the period of PK differences from adults. Potential examples are neurotoxicity from lead or mercury (Bellinger et al., 1992Go; Engleson and Herner, 1952Go; Grandjean et al., 1997Go) and cancer risk from vinyl chloride (Laib et al., 1989Go). Such differences may have less impact on RfDs/RfCs that are set based upon chronic effects that require years of cumulative exposure and toxicity, since altered internal dosimetry in early life would be a short-term factor in the exposure assessment.

Another way to consider variability in the database is to compare the gm child/adult ratios for a group of chemicals that share a common clearance pathway, with the extreme for that chemical group. The mean differences across age groups shown in Figures 1–7GoGoGoGoGoGoGo do not show the extremes, i.e., chemicals for which child/adult ratios surpass the gm ratio for that pathway and age group. The most obvious case is furosemide and renal clearance. The gm ratio for 1 week to 2 months of age is 2.8 across all renally cleared chemicals in the database. However the t1/2 ratio for furosemide is considerably higher (15-fold). For other pathways the contrast between the single "extreme" chemical and group mean data can be substantial but not quite so striking. For CYP1A2 the mean child/adult ratio is 4 to 9 through 2 months of age (Fig. 4Go) while the caffeine half-life ratio is 13 to 17 in these age groups. The large ratio for caffeine is in contrast to the structurally related xanthine, theophylline, another CYP1A2 substrate. In this case, the maximum child/adult half-life ratio is only 4.5. Our half-life results for the CYP1A2 pathway are consistent with a recent publication analyzing heterogeneity in CYP1A2-related drug clearance across the population (Dorne et al., 2001Go). In that study, caffeine clearance was shown to be 9 to 14-fold less in neonates than in adults, while the theophylline neonate/adult clearance differential was only 2- to 3-fold. These deficits disappeared in the infant population, similar to the developmental profile shown presently.

While the current database is too small to assess the full distribution of child/adult ratios that can exist across chemicals, it does point out that interchemical variability can be considerable within a given metabolism/clearance pathway and for structurally related chemicals. Thus, when applying the current results from the therapeutic drug literature to environmental chemicals, one needs to be aware of the gm child/adult ratios (as presented here) and also the likelihood that certain chemicals may have ratios well above this. At this point there is no basis to predict which chemicals would be more like the extreme than the mean case within each elimination pathway. However, there are obviously cases in which the neonate/adult PK function difference can be considerably larger than the 3.16-fold PK uncertainty factor. This not only applies to the half-life results for neonates shown presently, but also in terms of drug clearance. The Dorne et al. (2001) analysis projects that nearly all of the neonatal population is outside of the 3.16-fold PK uncertainty factor for caffeine clearance and approximately 1/3 of neonates are estimated to lie outside of this uncertainty factor for theophylline clearance.

The data in Figure 5Go indicates that across 8 drug substrates, CYP3A activity in premature and full-term newborns appears deficient relative to older age groups. This is noteworthy given that there is a considerable amount of a fetal form of CYP3A, 3A7 in utero. This isozyme remains high at birth and then declines thereafter (Cresteil, 1998Go; LaCroix et al., 1997Go). The profile shown in Figure 5Go suggests that the fetal isozyme is not very active at metabolizing various drug substrates as the child/adult t1/2 ratios are highest in early life and decrease along a time-frame consistent with the onset of CYP3A4 activity. This in vivo evidence regarding CYP3A7 is consistent with in vitro evidence that the fetal isozyme has less activity towards 2 anticonvulsants, carbamazepine and zonisamide (Ohmori et al., 1998Go), than does the mature isozyme, CYP3A4. Thus, it appears that the drugs in our database are not well handled by CYP3A7 and require maturation of CYP3A4 function to be eliminated efficiently.

The across age comparisons shown in Figures 1–7GoGoGoGoGoGoGo suggest immaturity of hepatic clearance systems and renal function in early life. However, given the tendency towards larger Vd in children across a broad array of chemicals (Fig. 8Go), it is possible that some of the child/adult half-life differences are due to distributional phenomena as opposed to deficits in metabolism and renal function. This concern is somewhat reduced by the evidence shown in Figure 2Go that significant child/adult differences exist in this dataset also with respect to clearance. Clearance is unaffected by changes in Vd since it is a direct reflection of the function of metabolic and renal removal systems. The somewhat smaller neonate/adult differentials for clearance relative to half-life in these figures may reflect the lack of dependence of the former parameter on Vd. Further evidence regarding the relative influence of Vd on half-life was sought due to the relatively small clearance dataset available for these drugs and given the various mechanisms by which children can have increased Vd (less protein binding, increased permeability of blood-brain barrier, greater amount of water per body weight; Besunder et al., 1988Go; Morselli, 1989Go; Kearns and Reed, 1989Go; Renwick, 1998Go). The dissociation of Vd from half-life results for theophylline (Fig. 9Go), and as seen for caffeine and lorazepam further point out that Vd seems to be a relatively minor factor in creating child/adult half-life differences in the current database. This concept is reinforced by in vitro demonstrations with liver bank specimens that a variety of CYPs are immature at birth and develop rapidly in the first weeks and months of life (Cresteil, 1998Go; Hakkola et al., 1998Go; Tanaka, 1998bGo; Treyluyer et al., 1996). In those studies, the developmental pattern for the evolution of CYP protein content or CYP in vitro activity tends to parallel the developmental pattern shown presently in Figures 1–7GoGoGoGoGoGoGo. This in vitro/in vivo concordance suggests that the immaturity of hepatic systems can be the overriding factor, without the need for additional explanations related to larger Vd in neonates.

A factor that may tend to quench functional immaturity of the liver is blood flow limitation of chemical clearance (Kedderis, 1997Go). Rapidly cleared chemicals (as determined in adults) may be blood flow limited such that the effects of enzyme induction, inhibition, or immaturity on whole body clearance may be less than expected. In fact, this does appear to be the case (Table 5Go) in which the 1-week to 2-month age group shows rapid clearance chemicals with significantly less child/adult difference than slow clearance chemicals. Prior to this time there is some suggestion that slow and rapid clearance chemicals exhibit different child/adult t1/2 ratios, but these differences were not statistically significant. One possible explanation for these findings is that within the first week of life, hepatic and renal systems are immature to the point where most chemicals are cleared slowly and thus flow limitation is not the primary factor governing hepatic extraction. This would allow the immaturity of hepatic systems to be observed in the half-life data for both slow and rapidly extracted chemicals. However, by 2 months of age, metabolic and clearance pathways may have matured sufficiently to cause some degree of flow limitation for rapidly cleared chemicals that does not occur for slower clearance chemicals. Beyond 2 months of age there were no significant differences between low and high clearance chemicals, with child/adult ratios tending to be less than unity. This is consistent with the maturation of hepatic enzymes during this time with children's larger liver weight per body weight (Gibbs et al., 1997Go) becoming an important factor in decreasing drug t1/2 relative to adults. The implications of Table 5Go for risk assessment is that the concept of hepatic blood flow limiting interindividual differences (Kedderis, 1997Go) may apply to certain windows of PK development but not to others.

The current findings are consistent with the pediatric PK summaries provided by Renwick and coworkers (1998, 2000; Dorne et al., 2001Go). They show that there are a variety of drugs for which clearance in neonates is slower than in adults. There are also cases in which they show clearance is more rapid, particularly when the pediatric group was at least several months of age (Renwick, 1998Go). Further, the relative size of the neonate/adult differences shown presently is similar to that seen in a composite analysis of 36 drugs with pediatric data (Renwick et al., 2000Go). The current analysis extends the earlier observations by providing more discrete age groupings and pathway-specific analyses.

Uncertainties in the Analysis
Uncertainties in the analysis come from trying to provide quantitative descriptions of child/adult differences across a wide variety of pediatric and adult PK studies that entail differing protocols, measurement techniques, dosing regimens, and subjects (e.g., health status, other medications). Thus, in addition to the across age group variability that we endeavored to assess, we were also confronted with across study variability, interindividual variability within an age group, and interchemical variability since the results are compiled across a diverse array of drugs. While the analysis controlled for some of this variability (e.g., normalizing half-life results to a reference chemical's values to decrease interchemical differences in absolute values), there was still much variability in the analysis that had little to do with the developmental pattern that was being studied. In spite of these sources of variability, the analysis was able to detect a clear and consistent pattern of immature drug elimination kinetics that showed steady maturation during the first year of life. This suggests that the PK developmental pattern shown presently can influence the clearance of many pharmaceuticals in predictable ways, and therefore, should also be applicable to certain environmental chemicals. Such applications may involve PBPK modeling or other approaches (see below) to develop PK estimates for environmental chemicals in children.

The manner in which children's age categories were constructed is somewhat arbitrary given that PK functions rapidly evolve during the first 3 to 6 months of life. Thus, during this period any age group created will be highly heterogenous, containing individuals on one end or the other of the developmental spectrum for the PK parameter. In general, age groups were constructed based upon physiologic considerations, and also upon the amount of PK data available for different age ranges. It is possible that different age cutpoints would have provided less variability in the results but it would appear that the overall profile of PK functional development would not differ markedly from an adjustment in age cutpoints (Hattis et al., submitted).

Another uncertainty that would also tend to conceal across age differences is the fact that the indicator drugs in the database may have important secondary clearance pathways. If the primary 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 upon the fate of the majority (50% 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 PK differences when judged by blood half-life data alone, which may have caused us to underestimate child-adult differences in certain cases. The implications of this type of underestimation for environmental toxicants are unclear since they may also have compensatory alternative pathways of disposition in children. The effect of such alternative pathways on risk needs to be evaluated within the context of toxic mechanism on a case-by-case basis.

It is important to note that some of the children's datasets represent clinical PK trials in children who were not in good health. The datasets were screened not only for sufficient numbers of subjects per age group, but also for clinical conditions such as hepatic or renal dysfunction that would affect the fate of xenobiotics. Studies were excluded that involved highly compromised subjects. However, it is possible that the clinical state of the children on test may have affected the PK results in certain datasets. The overall concordance mentioned above between in vitro and in vivo data suggest that this factor is not a major complicating issue in the current database.

In considering the relevance of the current findings to risk assessment, it is noteworthy that the database consists mostly of therapeutic drugs whose PK profiles may not be representative of certain types of environmental toxicants. For example, the adult half-life of all drugs in the database is less than 1 day, while certain classes of environmental chemicals (e.g., PCBs, dioxins, other organochlorines) can have adult half-lives that are considerably longer. In such cases, partitioning into lipid or other tissue depots may occur to a degree not seen with the drug substrates in this analysis. Thus, the current findings are most relevant for those environmental toxicants whose PK is governed by the pathways represented in the Children's PK Database.

Possible Applications for Environmental Chemicals
The demonstrated utility of PBPK analyses in extrapolating from animals to humans suggests that this approach would also be useful in extrapolating from adults to children. Such an endeavor requires obtaining the key parameter inputs for PBPK models across children's developmental stages. These parameters include respiratory rate, cardiac output, tissue compartment volumes and blood flows, Michaelis-Menten rate constants for saturable processes (e.g., CYP-mediated metabolism), and first order rate constants for processes known to be nonsaturable (e.g., certain Phase II metabolic reactions, renal elimination). The types of data gathered in the Children's PK Database (e.g., half-life, clearance, Vd) are what was available from the pediatric drug literature. These endpoints are not direct inputs into PBPK models, but instead can be thought of as the end result of the PK processes that the models try to simulate.

There are at least 2 ways in which this type of information can be used to assist risk assessors in understanding how children's PK may alter internal exposure to environmental chemicals and ultimate risk. These can be thought of as qualitative or quantitative assessment options as follows:

Qualitative option.
Use of child/adult ratios as rough indicators of PK throughput across specific pathways. The pathways analysis (Figs. 3–7GoGoGoGoGo) shows the developmental profile for chemical removal via CYPs, glucuronidation, and renal elimination. Additional information is available for a CYP not analyzed in the current database, CYP2E1. This CYP is involved in the activation of numerous xenobiotics including nitrosamines, chlorinated solvents, ethanol, and benzene (Wormhoudt et al., 1999Go). Its developmental profile is generally similar to that shown presently for other CYPs: immature at birth, rapid onset postnatal, and full maturation by 6 months to 1 year (Cresteil, 1998Go; Kurata et al., 1998Go; Nakamura et al., 1998Go; Tanaka, 1998bGo; Vieira et al., 1996Go). CYP2E1 was not part of the current assessment because clearance and half-life data were not available for indicator drug substrates for this pathway (trimethadione, halothane). However, as cited above, other types of in vivo and in vitro data support this developmental profile for CYP2E1. Aside from the CYPs, there are children's data, mostly from in vitro studies, for other important clearance and detoxification pathways including glutathione transferases, epoxide hydrolase, N-acetylation, and plasma carboxylesterases (Ecobichon and Stephens, 1972Go; Pariente-Khayat et al., 1991Go; Ratanasavanh et al., 1991Go; Strange et al., 1989Go).

Therefore, the child/adult comparisons exemplified in the current study can be used to assess whether children will be more or less exposed to parent compound and metabolites relative to adults. When combined with chemical-specific information on PK, and mechanism or mode of action, it may be possible to make a qualitative assessment of how children's PK affects risk for specific chemicals and at specific age groups. For example, knowing that CYP1A2 is deficient by roughly 3- to 10-fold in full-term neonates suggests that aromatic amine activation, a CYP1A2 function, would be well below adult levels in this group, thus protecting neonates from carcinogenic risk. However, the somewhat greater rate of CYP1A2 activity at 6 months of age and beyond (Fig. 4Go) suggests greater metabolic activation and risk in these stages of childhood. Additionally, chemicals that require CYP1A2 for metabolic detoxification and clearance may have excessively high internal doses in neonates. In certain cases, the implications of across age PK differences may be sufficiently clear to allow a semiquantitative approach involving age group-specific adjustment factors. Such assessments may help the assessor understand how large a role children's PK might play in determining risk. This in turn can help determine whether a more detailed analysis, such as PBPK modeling, is needed for children.

Quantitative (PBPK) approach.
As described above, the endpoints in the Children's PK Database are more akin to the outputs rather than the inputs to PBPK models. Therefore, the Children's PK Database may be very useful for calibration and validation of children's PBPK models. This effort requires building a base physiological model for children of various ages from datasets that describe children's body size, water and fat content, energy expenditure and respiratory rate, and growth profiles for various organs (Besunder et al., 1988Go; DHHS, 1996Go; Kearns and Reed, 1989Go; Pelekis, 2001Go; Thurlbeck, 1982Go). The base children's model could then be adapted for therapeutic drugs contained in the Children's PK Database to allow prediction of chemical concentrations in blood and blood t1/2 for select chemicals and datasets. The Michaelis-Menten or first order metabolic rates can be backfit to match the actual blood concentration and half-life data found in children. A parallel effort of calibrating an adult PBPK model for the same drug substrate would then show how the metabolic constants need to be adjusted when going from adults to children of different ages to make the age-specific PBPK models fit. The age adjustments could then be validated with another drug case study for the same clearance pathway. Such age- and pathway-specific child metabolism adjustment factors could then be used as part of the physiologic model to predict chemical metabolism and fate for environmental chemicals whose major fate pathways are covered by the Children's PK Database and which have validated adult human models. The uncertainty and data gaps surrounding input parameters for children need to be recognized and made transparent. However, this approach may allow the pediatric drug PK literature to be used to develop a predictive tool that would allow adjustment for children's PK in risk assessments for environmental chemicals.

A simpler quantitative approach may also be possible for children 6 months and beyond where it appears that hepatic and renal function have matured. Child/adult PK differences appear to be most dependent on the larger liver size per body weight in this age group, which scales based upon a surface area correction (body weight to the 3/4 power; Gibbs et al., 1997Go). Thus, the faster removal rates in children over 6 months of age (Figs. 2–5GoGoGoGo) can be predicted simply by scaling hepatic and renal function in this manner. This may serve to be a useful correction for many chemicals cleared by these systems.

Summary and Conclusions
A Children's PK Database has been assembled that contains 45 chemicals covering a wide range of chemical structures, mechanisms of action, and metabolism and clearance pathways. The dataset has been broken into discrete children's age categories to facilitate comparisons across postnatal development through adolescence. Regression results show that age group has a significant impact on drug half-life and clearance with immaturity of metabolic and clearance systems evident over the first weeks to months of life for all pathways analyzed. Beyond 6 months of age, most PK functions analyzed were comparable to or faster than adult function. These findings are consistent with in vitro studies of liver enzyme function during development. The concordance of in vivo and in vitro data suggest that the across age group PK differences seen in this dataset of mostly therapeutic drugs are likely to also exist with environmental chemicals; in particular, those which rely upon the same metabolism/clearance pathways as analyzed in this study. The across age comparisons suggest that the size of neonate/adult differences in PK function can be larger than the 3.16-fold PK uncertainty factor often attributed to interindividual variability. The range of values and extent of variability in the children's PK database is more explicitly documented in Hattis et al. (submitted).

Thus, the current results point out potentially important child/adult differences in PK function that could affect the way children's risks are understood. The results can be directly useful in qualitative assessments of children's risks by pointing out how key activation and detoxification/removal pathways differ across age groups. Additionally, the data compiled in the Children's PK Database can be a resource in constructing age-specific and pathway-specific PBPK models that predict internal dosimetry of environmental toxicants or their metabolites in children.


    ACKNOWLEDGMENTS
 
This research was supported by a U.S. EPA/State of Connecticut Assistance Agreement #827195-0.


    NOTES
 
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, U.S. EPA, or Clark University.

1 To whom correspondence should be addressed. Fax: (860) 509-7785. E-mail: gary.ginsberg{at}po.state.ct.us. Back


    REFERENCES
 TOP
 ABSTRACT
 INTRODUCTION
 DISCUSSION
 REFERENCES
 
Anderson, B. J., McKee, A. D., and Holford, N. H. G. (1997). Size, myths and the clinical pharmacokinetics of analgesia in paediatric patients. Clin. Pharmacokinet. 33, 313–327.[ISI][Medline]

Bellinger, D. C., Stiles, K. M., and Needleman, H. L. (1992). Low-level lead exposure, intelligence and academic achievement: A long-term follow-up study. Pediatrics 90, 855–861.[Abstract]

Bertz, R. J., and Granneman, G. R. (1997). Use of in vitro and in vivo data to estimate the likelihood of metabolic pharmacokinetic interactions. Clin. Pharmacokinet. 32, 210–258.[ISI][Medline]

Besunder, J. B., Reed, M. D. and Blumer, J. L. (1988). Principles of drug biodisposition in the neonate. A critical evaluation of the pharmacokinetic-pharmacodynamic interface (Part I). Clin. Pharmacokinet. 14, 189–216.[ISI][Medline]

Blanco, J. G., Harrison, P. L., Evans, W. E., and Relling, M. V. (2000). Human cytochrome P450 maximal activities in pediatric versus adult liver. Drug Metab. Dispos. 28, 379–382.[Abstract/Free Full Text]

Cohen, J., and Royston, D. (2001). Remifentanil. Curr. Opin. Crit. Care 7, 227–231.[Medline]

Cresteil, T. (1998). Onset of xenobiotic metabolism in children: Toxicological implications. Food Addit. Contam. 15(Suppl.), 45–51.[ISI][Medline]

Dempsey, D., Jacob, P., 3rd, and Benowitz, N. L. (2000). Nicotine metabolism and elimination kinetics in newborns. Clin. Pharmacol. Ther. 67, 458–465.[ISI][Medline]

DHHS (1996). Third National Health and Nutrition Examination Survey, 1988–1994, NHANES III. File Documentation #76200. U.S. Dept. of Health and Human Services, National Center for Health Statistics, Centers for Disease Control and Intervention, Hyattsville, MD. Available from NTIS.

Dollery, C. (Ed.) (1999). Therapeutic Drugs, 2nd ed. Churchill Livingstone, Edinburgh.

Dorne, J. L. C. M., Walton, K., and Renwick, A. G. (2001). Uncertainty factors for chemical risk assessment. Human variability in the pharmacokinetics of CYP1A2 probe substrates. Food Chem. Toxicol. 39, 681–696.[ISI][Medline]

Ecobichon, D. J., and Stephens, D. S. (1972). Perinatal development of human blood esterases. Clin. Pharm. Ther. 14, 41–47.[ISI]

Ekstrom, G., and Gunnarsson, U. B. (1996). Ropivacaine, a new amide-type local anesthetic agent, is metabolized by cytochromes P450 1A and 3A in human liver microsomes. Drug Metab. Dispos. 24, 955–961.[Abstract]

Engel, G., Hofmann, U., Heidemann, H., Cosme, J., and Eichelbaum, M. (1996). Antipyrene as a probe drug for human oxidative drug metabolism: Identification of the cytochrome P450 enzymes catalyzing 4-hydroxyantipyrene, 3-hydroxymethylantipyrine, and norantipyrine formation. Clin. Pharmacol. Ther. 59, 613–623.[ISI][Medline]

Engleson, G., and Herner, T. (1952). Alkyl mercury poisoning. Acta Paediatr. Scand. 41, 289–294.

Faustman, E. M., Silbernagel, S. M., Fenske, R. A., Burbacher, T. M. and Ponce, R. A. (2000). Mechanisms underlying children's susceptibility to environmental toxicants. Environ. Health Perspect. 108(Suppl. 1), 13–21.

Fehring, S. I., and Ahokas, J. T. (1989). Effect of the glutathione-S-transferase inhibitor, tienilic acid, on biliary excretion of sulphobromopththalein. Chem. Biol. Interact. 69, 23–32.[ISI][Medline]

Gantenbein, M., Attolini, L., Bruguerolle, B., Villard, P. H., Puyoou, F., Durand, A., Lacarelle, B., Hardwigsen, J., and Le-Treut, Y. P. (2000). Oxidative metabolism of bupivacaine into pipecolylxylidine in humans is mainly catalyzed by CYP3A. Drug Metab. Dispos. 28, 383–385.[Abstract/Free Full Text]

Gibbs, J. P., Murray, G., Risler, L., Chien, J. Y., Dev, R., and Slattery, J. T. (1997). Age-dependent tetrahydrothiophenium ion formation in young children and adults receiving high dose busulfan. Cancer Res. 57, 5509–5516.[Abstract]

Goodman-Gilman, A., Rall, T. W., Nies, A. S. and Taylor, P. (1990). The Pharmacological Basis of Therapeutics, 8th ed. Pergamon Press, New York.

Grandjean, P., Weihe, P., White, R. F., Debes, F., Araki, S., Yokoyama, K., Murata, K., Sorensen, N., Dahl. R., and Jorgenson, P. J. (1997). Cognitive deficit in 7-year-old children with prenatal exposure to methylmercury. Neurotoxicol. Teratol. 19, 417–428.[ISI][Medline]

Guitton, J., Buronfosse, T., Desage, M., Lepape, A., Brazier, J.-L., and Beaune, P. (1997). Possible involvement of multiple cytochrome P450s in fentanyl and sufentanil metabolism as opposed to alfentanil. Biochem. Pharmacol. 53, 1613–1619.[ISI][Medline]

Hadama, A., Ieiri, I., Morita, T., Kimura, M., Urae, A., Irie, S., Kaneda, T., Mamiya, K., Tashiro, N., Higuchi, S., and Otsubo, K. (2001). P-Hydroxylation of phenobarbital: Relationship to (s)-mephenytoin hydroxylation (CYP2C19) polymorphism. Ther. Drug Monit. 23, 115–118.[ISI][Medline]

Hakkola, J., Tanaka, E., and Pelkonen, O. (1998). Developmental expression of cytochrome P-450 enzymes in human liver. Pharmacol. Toxicol. 82, 209–217.[ISI][Medline]

Hamaoka, N., Oda, Y., Hase, I., and Asada, A. (2001). Cytochrome P4502B6 and 2C9 do not metabolize midazolam: Kinetic analysis and inhibition study with monoclonal antibodies. Br. J. Anesth. 86, 540–544.[Abstract/Free Full Text]

Hattis, D., Banati, P., and Goble, R. (1999). Distributions of individual susceptibility among humans for toxic effects. How much protection does the traditional tenfold factor provide for what fraction of which kinds of chemicals and effects? Ann. N.Y. Acad. Sci. 895, 286–316.[Abstract/Free Full Text]

Hattis, D., Russ, A., Banati, P., Kozlak, M., Goble, R., and Ginsberg, G. (2000). Development of a Comparative Child/Adult Pharmacokinetic Database Based Upon the Therapeutic Drug Literature. Report to USEPA, Assistance Agreement 827195–0, Oct. 2000..

Henderson, G. N., Yan, Z., James, M. O., Davydova, N. and Stacpoole, P. W. (1997). Kinetics and metabolism of chloral hydrate in children: Identification of dichloroacetate as a metabolite. Biochem. Biophys. Res. Commun. 235, 695–698.[ISI][Medline]

IPCS (2001). Guidance Document for the Use of Chemical-Specific Adjustment Factors (CSAFs) for Interspecies Differences and Human Variability in Dose/Concentration—Response Assessment. International Program on Chemical Safety. Available at: http://www.ipcsharmonization.org. Accessed December 4, 2001.

Jacqz-Aigrain, E., and Burtin, P. (1996). Clinical pharmacokinetics of sedatives in neonates. Clin. Pharmacokinet. 31, 423–443.[ISI][Medline]

Kawashiro, T., Yamashita, K., Zhao, X. J., Koyama, E., Tani, M., Chiba, K., and Ishizaki, T. (1998). A study on the metabolism of etoposide and possible interactions with antitumor or supporting agents by human liver microsomes. J. Pharmacol. Exp. Ther. 286, 1294–1300.[Abstract/Free Full Text]

Kearns, G. L., and Reed, M. D. (1989). Clinical pharmacokinetics in infants and children. A reappraisal. Clin. Pharmacokinet. 17(Suppl. 1), 29–67.[ISI][Medline]

Kedderis, G. L. (1997). Extrapolation of in vitro enzyme induction data to humans in vivo. Chem. Biol. Interact. 107, 109–121.[ISI][Medline]

Kobayashi, K., Kogo, M., Tani, M., Shimada, N., Ishizaki, T., Numazawa, S., Yoshida, T., Yamamoto, T., Kuroiwa, Y., and Chiba, K. (2001). Role of CYP2C19 in stereoselective hydroxylation of mephobarbital by human liver microsomes. Drug Metab. Dispos. 29, 36–40.[Abstract/Free Full Text]

Kurata, N., Nishimura, Y., Iwase, M., Fischer, N. E., Tang, B. K., Inaba, T., and Yasuhara, H. (1998). Trimethadione metabolism by human liver cytochrome P450: Evidence for the involvement of CYP2E1. Xenobiotica 28, 1041–1047.[ISI][Medline]

LaCroix, D., Sonnier, M., Moncion, A., Cheron, G., and 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.[Abstract]

Laib, R. J., Bolt, H. M., Cartier, R., and Bartsch, H. (1989). Increased alkylation of liver DNA and cell turnover in young versus old rats exposed to vinyl chloride correlates with cancer susceptibility. Toxicol. Lett. 45, 231–239.[ISI][Medline]

Lasker, J. M., Wester, M. R., Aramsombatdee, E., and Raucy, J. L. (1998). Characterization of CYP2C19 and CYP2C9 from human liver: Respective roles in microsomal tolbutamide, S-mephenytoin, and omeprazole hydroxylations. Arch. Biochem. Biophys. 353, 16–28.[ISI][Medline]

Levy, G., Khanna, N. N., Soda, D. M., Tsuzuki, O., and Stern, L. (1975). Pharmacokinetics of acetaminophen in the human neonate: Formation of acetaminophen glucuronide and sulfate in raltion to plasma bilirubin concentration and D-glucaric acid excretion. Pediatrics 55, 818–825.[Abstract]

Mayers, D. J., Hindmarsh, K. W., Sankaran, K., Gorecki, D. K. J., and Kasian, G. F. (1991). Chloral hydrate disposition following single-dose administration to critically ill neonates and children. Dev. Pharmacol. Ther. 16, 71–77.[ISI][Medline]

Messina, E. S., Tyndale, R. F., and Sellers, E. M. (1997). A major role for CYP2A6 in nicotine C-oxidation by human liver microsomes. J. Pharmacol. Exp. Ther. 282, 1608–1614.[Abstract/Free Full Text]

Morselli, P. L. (1989). Clinical pharmacology of the perinatal period and early infancy. Clin. Pharmacokinet. 17(Suppl. 1), 13–28.[ISI][Medline]

Nandi, V., and Soine, W. H. (1997). HPLC analysis for amobarbital N-glycosides in urine. J. Pharm. Biomed. Anal. 15, 1187–1195.[ISI][Medline]

Nakajima, M., Yamamoto, T., Nunoya, K., Yokoi, T., Nagashima, K., Inoue, K., Funae, Y., Shimada, N., Kamataki, T., and Kuroiwa, Y. (1996). Characterization of CYP2A6 involved in 3'hydroxylation of cotinine in human liver microsomes. J. Pharmacol. Exp. Ther. 277, 1010–1015.[Abstract]

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

NRC (1993). Pesticides in the Diets of Infants and Children. National Research Council. National Academy Press, Washington, DC.

Ohmori, S., Nakasa, H., Asanome, K., Kurose, Y., Ishii, I., Hosokawa, M., and Kitada, M. (1998). Differential catalytic properties in metabolism of endogenous and exogenous substrates among CYP3A enzymes expressed in COS-7 cells. Biochem. Biophys. Acta 1380, 297–304.[ISI][Medline]

Pariente-Khayat, A., Pons, G., Rey, E., Richard, M.-O., D'Athis, P., Moran, C., Badoual, J., and Olive, G. (1991). Caffeine acetylator phenotyping during maturation in infants. Pediatr. Res. 29, 492–495.[Abstract]

Pelekis, M., Gephart, L. A., and Lerman, S. E. (2001). Physiological-model-based derivation of the adult and child pharmacokinetic intraspecies uncertainty factors for volatile organic compounds. Regul. Toxicol. Pharmacol. 33, 12–20.[ISI][Medline]

Pope, C. N., Chakraborti, T. K., Chapman, M. L., Farrar, J. D. and Arthun, D. (1991). Comparison of in vivo cholinesterase inhibition in neonatal and adult rats by three organophosphorothioate insecticides. Toxicology 68, 51–61.[ISI][Medline]

Ratanasavanh, D., Beaune, P., Morel, F., Flinois, J. P., Guengerich, F. P., and Guillouzo, A. (1991). Intralobular distribution and quantitation of cytochrome P-450 enzymes in human liver as a function of age. Hepatology 13, 1142–1151.[ISI][Medline]

Renwick, A. G. (1998). Toxicokinetics in infants and children in relation to the ADI and TDI. Food Addit. Contam. 15(Suppl.), 17–35.[ISI][Medline]

Renwick, A. G., Dorne, J. L. and Walton, K. (2000). An analysis of the need for an additional uncertainty factor for infants and children. Regul. Toxicol. Pharmacol. 31, 286–296.[ISI][Medline]

Stacpoole, P. W., Henderson, G. N., Yan, Z., Cornett, R. and James, M. O. (1998). Pharmacokinetics, metabolism, and toxicology of dichloroacetate. Drug Metab. Rev. 30, 499–539.[ISI][Medline]

Strange, R. C., Howie, A. F., Hume, R. Matharoo, B., Bell, J., Hiley, C., Jones, P., and Beckett, G. J. (1989). The developmental expression of {alpha}-, µ- and {pi}-class glutathione S-tranferases in human liver. Biochem. Biophys. Acta 993, 186–190.[ISI][Medline]

Tanaka, E. (1998a). Clinically important pharmacokinetic drug-drug interactions: Role of cytochrome P450 enzymes. J. Clin. Pharm. Ther. 23, 403–416.[ISI][Medline]

Tanaka, E. (1998b). In vivo age-related changes in hepatic drug-oxidizing capacity in humans. J. Clin. Pharm. Ther. 23, 247–255.[ISI][Medline]

Thurlbeck, W. M. (1982). Postnatal human lung growth. Thorax 37, 564–571.[Abstract]

Treluyer, J. M., Cheron, G., Sonnier, M., and Cresteil, T. (1996). Cytochrome P-450 expression in sudden infant death syndrome. Biochem. Pharmacol. 52, 497–504.[ISI][Medline]

U.S. EPA (1997). Exposure Factors Handbook. EPA/600/P-95/002Fb.

Vesselinovitch, S. D., Rao, K. V. N., and Mihailovich, N. (1979). Neoplastic response of mouse tissues during perinatal age periods and its significance in chemical carcinogenesis. Natl. Cancer Inst. Monogr. 51, 239–250.[Medline]

Vieira, I., Sonnier, M., and Cresteil, T. (1996). Developmental expression of CYP2E1 in the human liver. Hypermethylation control of gene expression during the neonatal period. Eur. J. Biochem. 238, 476–483.[Abstract]

Von Moltke, L. L., Greenblatt, D. J., Harmatz, J. S., Duan, S. X., Harrel, L. M., Cotreau-Bibbo, M. M., Pritchard, G. A., Wright, C. E. and Shader, R. I. (1996). Triazolam biotransformation by human liver microsomes in vitro: Effects of metabolic inhibitors and clinical confirmation of a predicted interaction with ketoconazole. J. Pharmacol. Exp. Ther. 276, 370–379.[Abstract]

Wandel, C., Bocker, R., Bohrer, H., Browne, A., Rugheimer, E., and Martin, E. (1994). Midazolam is metabolized by at least three different cytochrome P450 enzymes. Br. J. Anaesth. 73, 658–661.[Abstract]

Wester, M. R., Lasker, J. M., Johnson, E. F., and Raucy, J. L. (2000). CYP2C19 participates in tolbutamide hydroxylation by human liver microsomes. Drug Metab. Dispos. 28, 354–359.[Abstract/Free Full Text]

Wormhoudt, L. W., Commandeur, J. N. M., and Vermeulen, N. P. (1999). Genetic polymorphisms of human N-acetyltransferase, cytochrome P450, glutathione-S-transferase, and epoxide hydrolase enzymes: Relevance to xenobiotic metabolism and toxicity. Crit. Rev. Toxicol. 29, 59–124.[ISI][Medline]

Yaffe, S. J., and Aranda, J. V. (Eds.) (1992). Pediatric Pharmacology, 2nd ed. W. B. Saunders, Philadelphia, PA.

Yamazaki, H., Nakano, M., Imai, Y., Ueng, Y.-F., Guengerich, F. P., and Shimada, T. (1996). Roles of Cytochrome b5 in the oxidation of testosterone and nifedipine by recombinant cytochrome P450 3A4 and by human liver microsomes. Arch. Biochem. Biophys. 325, 174–182.[ISI][Medline]

Yanagihara, Y., Kariya, S., Ohtani, M., Uchino, K., Aoyama, T., Yamamura, Y., and Iga, T. (2001). Involvement of CYP2B6 in n-demethylation of ketamine in human liver microsomes. Drug Metab. Dispos. 29, 887–890.[Abstract/Free Full Text]