Similarities and Dissimilarities in Physiology, Metabolism, and Disease States and Responses to Therapy in Children and Adults
The pediatric population often responds to drugs and other therapeutics differently than adults do. Generally, the guidelines that practitioners use when they prescribe drugs have not been based on biologic or pharmacologic principles when they extrapolate the drug doses used for adults to infants and children. Not only have the guidelines tended to be simplistic in that they assume a linear relationship between children and adults, but they have also not made allowances for the complex changes in growth and development that take place during childhood. A number of quantitative and qualitative differences in the anatomy and physiology of the infant and developing child can affect the absorption, distribution, metabolism, and excretion of various drugs and other xenobiotic compounds. Additionally, children differ from adults not only in anatomical and physiological ways but also in the types of diseases from which they suffer and in the manifestations of those diseases that they do have in common with adults. These factors can determine the types of therapies developed for children as well as the design of studies that evaluate new therapeutic agents.
Adverse experiences to drugs that have been tolerated by adults have been observed in children when the drugs have not been adequately studied before their use in the pediatric population. A structured analysis of children's responses to drugs that are tolerated by adults could strengthen the safety component of the study design. The ontogeny of drug-metabolizing enzymes in humans, for instance, may explain several differences in the responses to drugs by adult and pediatric populations. Understanding the relative role of absorption in drug biotransformation may also lead to better understanding of the drug-
metabolizing abilities of the different populations, Phenotyping studies of drug biotransformation could also help explain differential responses to drugs.
The results of studies with adults thus cannot typically be extrapolated to children because growth and development issues add a range of complicated variables to the already intricate realm of drug metabolism and pharmacodynamics. Studies of drugs for use in children must be designed to account for the complex developmental changes that can affect drug biodisposition and pharmacodynamics. The following summaries of the presentations at the workshop examine the uniqueness of the pediatric population and describe how infants and children are similar or dissimilar to adults with regard to physiology, drug metabolism, immunology, cognitive effects, and response to disease states and therapy.
Presented by Sumner Yaffe, M.D.
Director, Center for Research for Mothers and Children,
National Institute for Child Health and Human Development
Since passage of the Food and Drug Modernization Act of 1997 and subsequent revision of regulations by the U.S. Food and Drug Administration (FDA), children as a class of individuals have come a long way toward receiving better and more appropriately dosed therapeutics. One has to look back only to 1938 when sulfanilamide (called an elixir) appeared on the market as an antibacterial drug. Everyone was anxious to use it for infants and children, particularly for the treatment of infectious diseases prevalent among children at that time. The elixir, made by mixing sulfanilamide with diethylene glycol, was distributed for 1 month before 107 children died of renal toxicity following its administration.
In 1938 the Food and Drug and Cosmetic Act was amended to require pre-marketing clearance for new drug applications (NDAs). Drugs had to be safe but not necessarily efficacious. In 1960 the sedative thalidomide dramatically raised new issues about the importance of testing drug safety in a variety of populations and established the need to have standards for effectiveness. Additional requirements were enacted. Those requirements mandated that those submitting NDAs weigh the benefit versus the risk and conduct adequate and well-controlled clinical trials with full and informed consent. Despite the evolution of strict standards for the testing of drugs with adults, 80 percent of drugs that are on the market have not been tested in studies with children. Studies of drugs with infants and children are urgently needed. The agenda of this workshop emphasizes the unmet state of knowledge regarding drug action and disposition in the pediatric population. The need for further research is clearly identified. Modern molecular biology and genetics, when applied to pharmacologic studies in infants and children, will markedly advance our understanding of drug actions and enable us to practice rational therapeutics.
DIFFERENCES BETWEEN CHILDREN AND ADULTS
Presented by Ralph Kauffman, M.D.
Director of Medical Research, Children's Mercy Hospital, Kansas City, Missouri
The rules by which practitioners have extrapolated adult drug doses to infants and children have not always been based on biologic or pharmacologic logic. Moreover, they have tended to be overly simplistic, in that they assume a linear relationship between children and adults and do not incorporate the complex, nonlinear changes in growth and development that take place during childhood.
Indeed, it is the dynamic process of growth, differentiation, and maturation that sets children apart from adults. In addition to growth in physical size, dramatic changes in body proportions, body composition, physiology, neurologic maturation, and psychosocial development take place during infancy and childhood. These changes are of equal if not greater importance than growth in physical size in terms of the child's development, response to disease, and response to therapeutic modalities.
Growth and development are particularly rapid during the first 2 years of life. Body weight typically doubles by 6 months of age and triples by the first birthday. Body length increases by 50 percent during the first year. Body surface area doubles during the first year. During this metamorphosis, major organ systems grow and mature. In the first 18 months of life the child becomes ambulatory, develops socialization skills, and learns verbal language. The other period of childhood when important changes take place is puberty, when accelerated growth and sexual maturation occur. In addition, body proportions continue to change significantly from childhood to adulthood.
The respective proportions of body weight contributed by fat, protein, and water change during infancy and childhood. At birth, total body water constitutes approximately 80 percent of body weight. By 5 months of age, however, total body water accounts for only 60 percent of body weight and then remains relatively constant. However, there is a progressive decrease in extracellular water throughout childhood and into young adulthood. As the proportion of body water decreases, the percentage of body weight contributed by fat doubles by 4 to 5 months of age. In addition, protein mass increases during the second year of life as the child becomes ambulatory.
Major organ systems change in relative size as well as function during childhood. For example, the weights of liver and kidneys—the major organs responsible for drug elimination—relative to body mass are several-fold greater in the preschool-age child than in young adults. Likewise, in preschool-age children, the ratio of body surface area to body weight, which reflects the relative size of the skin as an organ, is approximately 2.5 times that in adults.
A number of quantitative and qualitative differences in the anatomy and physiology of the infant and developing child can affect the absorption of various drugs and other xenobiotic compounds:
As with the skin area, the absorptive surface area of the gut is relatively greater in the infant than in the adult.
Gastric pH is 7 to 8 at birth and then falls to less than 3 during the first 1 to 2 hours of life. However, relative hypochlorhydria persists during the first 1 to 2 months of life. Premature infants have impaired gastric acid production until after 32 weeks of gestational age. This allows orally administered acid-labile drugs that may not be absorbed intact in the mature individual to have greater bioavailability in premature and very young infants.
The infant has decreased pancreatic exocrine function, which matures by the first year. This decreased function may affect the absorption of prodrug esters of fatty acids.
The immature also exhibits permeability to large molecular species, including intact proteins that are not absorbed by the mature gut.
Gastric emptying and gut transit time may be prolonged in premature and ill newborns. On the other hand, healthy infants may have transit times shorter than those for adults.
Infants have decreased first-pass metabolism of drugs, with increased levels of uptake until the metabolic pathways resident in the gut mucosa and liver mature. This may lead to greater bioavailability of drugs that undergo significant first-pass metabolism.
Protein binding of drugs can affect drug pharmacodynamics, toxicity, distribution, and elimination. Maturational differences in protein binding are primarily an issue during infancy:
Plasma protein binding of drugs typically is decreased in the newborn and young infant. This is most pronounced in the premature infant.
Decreased binding is attributed to decreased levels of circulating protein (albumin and α-1-glycoprotein) and also to a decreased affinity for certain ligands. In some cases, decreased binding may also be attributed to competition for binding sites with other drugs or highly bound endogenous substances. The ultimate effect on binding of a specific drug depends on the relative binding affinities and molar concentrations of the competing ligands.
Binding is also pH dependent and may be decreased in the presence of acidosis.
Developmental differences in drug binding may occur with tissue binding as well as with plasma proteins.*
Glomerular filtration (regulation of renal blood flow) rates and tubular transport mechanisms all are physiologically decreased at birth but increase rapidly during the first month of life and reach maturity before the end of the first year. Maturational changes in renal function have profound implications for administration of drugs that are primarily excreted by the kidney.
Puberty is another period in development that involves major physiologic changes, including (1) large growth spurts mediated by surges in human growth hormone and other growth factors, (2) increase in luteinizing hormone and follicle-stimulating hormone concentrations with secondary sexual maturation associated with the effects of estrogen or progesterone and testosterone, and (3) gender-specific changes in body composition, with females acquiring a greater proportion of body fat and males having larger lean muscle mass.
Children differ from adults not only in anatomic and physiologic ways but also in the types of diseases from which they suffer and in their manifestations of those diseases that they do have in common with adults. This can determine the types of therapies developed for children as well as the design of studies that test new therapeutic agents. In addition, some diseases occur only in children and therefore can only be studied by studies with children. For example, newborn respiratory distress syndrome is unique to the immature newborn. Clinical trials of surfactant for the prevention or treatment of the syndrome were conducted first and only with newborns, with no studies performed with subjects in other age groups. Other examples are bronchopulmonary dysplasia and retinopathy of prematurity, which are unique to premature infants. Consequently, new approaches to the amelioration of these conditions must necessarily be studied almost exclusively in studies with infants.
In addition, certain malignancies occur only in children, such as neuroblastoma and Wilms' rumor. Congenital heart defects are generally diagnosed and treated during infancy and childhood and exhibit a range of pathophysiologies and hemodynamics that are uncommon in adults.
Furthermore, certain infectious diseases, such as rubella, rubeola, mumps, pertussis, and Haemophilus influenzae infection occur primarily in childhood. Therefore, studies that document the efficacies of vaccines for the prevention of these infections must be carried out with healthy infants and young children. Data derived from studies with adults cannot be extrapolated to children. For example, in contrast to adults and older children, children under age 2 years did not develop protective immunity to H. influenzae from the first commercially available vaccine, because of differences in their immune responses. Subse-
quently, the conjugated vaccine, which confers immunity to children less than 12 months of age, was developed after appropriate evaluation in this age group.
In developing drugs for pediatric populations, investigators must recognize that some diseases that are superficially similar between children and adults may have very different pathophysiologies in children. For example, human immunodeficiency virus (HIV) infection and AIDS may occur at any age. However, it was realized within a few years after the start of the AIDS pandemic that the clinical course of the infection in infants differs in several important aspects from that in adults. It became apparent that the Centers for Disease Control and Prevention criteria for AIDS, developed out of experience with the adult population, did not apply to infants and young children. In addition, progression of the infection in infants tended to be more rapid than in adults. Before current antiviral therapies became available, survival times for children were shorter than those for adults. Studies focused on the pediatric age group led to advances in understanding of the pediatric manifestations of the disease, and these advances have led to major shifts in therapeutic approaches for infants and children with HIV infection.
As another example, hypertension occurs in infants and children as well as adults, although the dominant causes of hypertension are different between the two groups. Whereas most cases of hypertension in adults are classified as primary or essential, most cases of hypertension in children are secondary and are most commonly associated with renal disease. Essential hypertension is rarely identified until late childhood or adolescence. This has profound implications for the design of studies that test the effects of antihypertensive drugs in children and the choice of new antihypertensive drugs to be tested in studies with children.
Finally, children may be more or less vulnerable than adults to the adverse effects of drugs. This possibility must always be considered when designing a drug trial or developing a new drug for pediatric use. There have been numerous examples of greater or unpredictable vulnerability of children to adverse effects, including the following:
An entire generation of children suffered enamel dysplasia from exposure to tetracycline antibiotics during critical periods of formation of dentition, an adverse effect that could never have been anticipated on the basis of data from clinical trials conducted with adult populations.
Verapamil was given to infants to convert supraventricular tachycardia on the basis of experiences with adults. After a series of infant deaths associated with its use, the different response in infants was recognized and verapamil is no longer used for this indication in infants.
Desflurane, an inhalation anesthetic, provides rapid, smooth, and safe anesthesia induction in adult patients and appeared to be an ideal anesthetic for children. However, in pediatric patients it causes an unacceptable incidence of laryngospasm, breath holding, and hypersecretion when used as an induction
agent. This life-threatening adverse event could not have been predicted from the results of trials with adults.
Conversely, children can be less vulnerable to adverse drug effects than adults. For example, infants are less susceptible to renal toxicity from aminoglycoside antibiotics, possibly due, in part, to a reduced ability to concentrate the drugs intracellularly in renal tubular and epithelial cells. In addition, hepatotoxicity from the general anesthetic halothane is relatively rare in children, whereas it is not uncommon in adults. As another example, the risk of isoniazid-induced hepatitis is negligible in children, whereas the reported incidence is 23 per 1,000 patients in middle-age adults.
Because children differ from adults in so many important aspects and are dependent on adults for their welfare, certain ethical and logistical issues must also be considered when designing and conducting research that involves children. The anatomic, physiologic, and psychologic dynamics of growth and development must be accommodated at various developmental stages. This frequently requires age-appropriate modification of approaches typically used in studies with adult subjects, particularly when obtaining informed consent and conducting risk-benefit analyses.
In terms of informed consent, children cannot independently consent to participate in research; thus, surrogate consent or permission must be provided by a parent or guardian. When gaining surrogate consent or authorization to participate in research, it is important to ensure that the child's interests are protected. It is essential to give the child with sufficient cognitive ability the opportunity to assent to participation.
When evaluating risks and potential benefits, investigators must consider the child's age, developmental status, and condition. For example, children at various ages have unique fears, anxieties, and perceptions of body threat that must be considered. To the young child, being in a strange environment and separated from parents may be a greater threat than the procedures directly related to the study. For the adolescent, perceived loss of privacy associated with, for example, urine collection may be a source of embarrassment that is far more threatening than venipuncture. It is therefore important to take whatever steps are feasible to minimize fear, anxiety, and discomfort associated with a study. In addition, consideration of parents' perceptions is important in terms of their views of the possible benefits versus risks and the amount of discomfort or inconvenience that they perceive their child will suffer by participation in the study.
Regarding the logistics of conducting studies with children, several issues deserve consideration. For example, if multiple blood samplings are required, there may be a need to use indwelling cannulae and to time research blood sampling so that it coincides with clinical sampling when possible. Acute blood loss should not exceed 3 percent of total blood volume, which can be quite limiting in younger children and infants. This may necessitate adapting assays to smaller
clinical sample sizes. When developing outcome measures, assessment tools might have to be adapted so that they are applicable to younger age groups. For example, in conducting analgesic studies the same pain assessment methods used for a 25-year-old individual cannot be used for a 5-year-old child.
Finally, age-appropriate dose formulations must be considered. Palatable liquid formulations are preferable for orally administered medications for infants and young children. If solid dosage formulations must be administered, they may have to be administered with an age-appropriate food, and when doing so, the age of the child must be considered (e.g., infants cannot take the dosage with solid food).
In summary, growth and development add complexity to the already complex world of drug metabolism and pharmacodynamics. This process is neither simple nor linear. It should be obvious, then, that the results from studies with adults typically cannot be extrapolated to children. The younger the child, the more this is true. Studies of drugs for use in children must be designed to take into effect the complex developmental changes that can affect drug biodisposition and pharmacodynamics. Developmental effects may not always be predictable. As the gaps in knowledge are gradually and painstakingly filled in, the development of new therapeutic agents for children will be enhanced, drugs will be given to children with greater precision and safety, and in the end, infants and children will be the ultimate beneficiaries.
DIFFERENTIAL RESPONSES OF CHILDREN TO DRUGS
Presented by John T. Wilson, M.D.
Professor and Chief of Clinical Pharmacology, Department of Pediatrics, Louisiana State University Health Sciences Center, Shreveport
Past negative experiences with the differential responses of children to drugs—such as the deaths that occurred with elixir of sulfanilamide, tooth staining after tetracycline treatment, kernicterus after sulfisoxazole treatment, and cardiovascular collapse after chloramphenicol treatment—highlight the long-standing and sometimes severe effects that can occur when drugs are not properly studied before their use in children. Insufficient information about pediatric dosing prompted investigations of the pharmacokinetics and pharmacodynamics of drugs in children. This work with children and immature-animals showed a maturational pattern for metabolic clearance of many drugs. The publication of the 1994 FDA Pediatric Rule and enactment of Section III of the Food and Drug Administration Modernization Act in 1997 allowed (1) extrapolations of information to children only when efficacy and drug action were similar in adults, and (2) a 6-month extension of patent exclusivity when a sponsor performed studies with children at the request of FDA. These actions ex-
posed the need for the identification of differential ding responses in children and their underlying mechanisms and illustrated the need for clinical research in this area if extrapolations of information from studies with adults were to be made with validity. What was the extent and magnitude of a differential response, and how much harm had occurred from not knowing the proper pediatric dose? Answers to these and other questions proved difficult from a search with conventional literature index systems. To obtain this information in a structured format amenable to literature searches, the following classification system was developed and tested with existing data.
A hierarchical classification system has, for all its categories, a maturational base that can be inserted at any level to limit or expand the question being asked with regard to differential drug responses in children. The lower the insertion, the more focused the question. Some items in this base are absorption and disposition, pharmacodynamics, childhood diseases (action on or confounding by), and vulnerability of the affected locus (i.e., a window of maximum risk). The following are categories of the classification system:
Response. Drug response is divisible into efficacy (achieving the desired response to the drug) or safety; either or both components can be decreased, increased, or considered appropriate for children relative to adults. Safety issues tend to focus on side effects, idiosyncratic reactions, or toxicity in children.
Differentiation. An efficacy or safety response is differentiated in children by either intensity (quantitative) or type (qualitative). An intensity or type of reaction that is different in children compared with that in adults may occur and is customarily referenced against drug dose or concentration in plasma.
Characteristics. Additional sets of characteristics identify and describe differential drug responses in children: whether the response is overt or covert, immediate or delayed, acute or chronic, and temporary or permanent. A further description for each includes incidence, magnitude, confounding, and treatability.
Orientation. Lastly, responses are described with regard to orientation of the question to cause or outcome. Orientations include drug, clinical, or chemical event; outcome (or a surrogate marker for the desired clinical outcome); and direct or indirect cause. This classification system allows one to operationalize the relevant response differences as they relate to human maturation and the pharmacology of a drug. Applications of the classification system to known differential drug responses in children illustrate its utility.
Questions about differential efficacy are qualified according to intensity and confounded as they apply to antipyretics as a drug class. Febrile children show nonlinear pharmacodynamics (the initial temperature sets the extent of fall in temperature), a lag period between time of maximum level in plasma and response, and slope and cyclic functions embedded in the temperature response to pyrogen and drug (Brown et al., 1993, 1998; Wilson et al., 1982). This has not been shown for adults.
Another drug class to which efficacy, intensity, and magnitude have been applied is the antihypertensive class of drugs (angiotensin-coverting enzyme inhibitors and nifedipine). The efficacy, intensity, and acute action of a drug applied to a disease can be shown with theophylline. The effect of the ding diminishes with increasing severity of asthma. In addition, a delayed efficacy response is seen when hypothyroid children of different ages are given thyroid replacement. In contrast to delayed effects, however, a chronic characteristic is noted for many psychotropic drugs given to children with behavioral disorders.
When safety is the major consideration, categories 2 to 4 mentioned above assist in focusing on the question. That of high intensity, incidence, and of a delayed nature is illustrated by a skin rash that followed 5 to 7 days after children received a loading dose of phenytoin. It was found in approximately 54 percent of children and 5 to 10 percent of adults. A threshold in plasma was found and metabolite analysis allowed postulation of in situ (skin) metabolism that led to a rash in those children poor for formation of p-hydroxyphenytoin (Wilson et al., 1976, 1978). Anticonvulsant drug safety with regard to categories of type, overt actions, and drug class is shown by a paradoxical reaction to phenobarbital, an increase in absence seizures with carbamazepine, and worsening of cognition in children treated with valproic acid. A change on category 3 to temporary classifies pediatric safety concerns for benzodiazepines. Paradoxical reactions to diazepam, aggression with clonazepam, and disinhibition with midazolam are examples.
In sum, efficacy and safety components of study design should be strengthened by a structured classification system that yields data for analysis of differential drug responses in children. Although fundamental pharmacology studies with immature-animal models are useful, when it comes to determining qualitative or even quantitative differences in drug responses in children, it is essential that children be studied and that data be categorized in a way that facilitates their access in multiple ways when one asks questions about differential ding responses in children. It is hoped that indexing systems will incorporate this new classification so that much needed pediatric drug data can be filed, retrieved, and collated in a comprehensive manner.
MOLECULAR BASIS OF DRUG METABOLISM
Presented by Thierry Cresteil, Ph.D.
Institut Gustave Roussy, Villejuif, France
The ontogeny of drug-metabolizing enzymes in humans could explain several differences in the consequences of drug exposure in adult and pediatric populations. Obstetricians must be mindful of the teratogenic potential of drugs, pollutants, or toxic compounds, and pediatricians need to be attentive to the ef-
fects of maturation in modifying the pharmacokinetics and clearance of a drug or in impairing the elimination of endogenous molecules from the body.
North American and European studies have demonstrated that the average number of drugs ingested during pregnancy is 10.3, with the range being from 3 to 29. As a consequence, most infants have drugs and metabolites stored in their tissues in the microgram or milligram range after labor and parturition. A survey conducted in an intensive care unit revealed exposure to an average of 3.4 drugs per infant in neonatal practice, and exposure was generally inversely related to birth weight. Further exposure can occur through breast-feeding.
The knowledge of the biotransformation pathways that a drug undergoes in the human liver and the capacity of the fetal and neonatal liver to catalyze these reactions will allow one to predict the metabolic fate and the potential risk of drug toxicity at every stage of development.
In the human liver, a multienzyme system is responsible for the biotransformation of hydrophobic molecules. In a first step, hydrophobic molecules are converted by phase I enzymes and are further conjugated to cellular acceptors by phase II enzymes. The balance between phase I and II enzymes controls the accumulation of activated molecules in the cell and thus the formation of adducts to protein or DNA. Because of the wide variety of substrates, phase I and phase II enzymes exist as isoforms, with a partial and overlapping substrate specificity, which could be investigated in vitro with human liver microsomes.
Cytochromes P450 (P450 or CYP) are the major phase I enzymes and are mostly expressed in the liver but are expressed to a lesser extent in the intestine, kidneys, and lungs. In the adult human liver, more than 15 isoforms of P450 are expressed, but there is wide variability for a given isoform. These variations result from different genetic backgrounds and from heterogeneous chemical exposures among individuals. Some substrates are specific to a single isoform: for example, debrisoquine is hydroxylated only by CYP2D6. A majority of compounds, however, could undergo metabolism by several routes, each catalyzed by different enzymes, leading to the formation of metabolites with variable therapeutic or toxic capacities.
Lastly, the expression of these enzymes could be modulated by the administration of molecules capable of stimulating their synthesis. In addition to drugs like barbiturates or steroids, molecules present in the diet (e.g., alcohol or flavonoids) or in the environment (e.g., polycyclic aromatic hydrocarbons) could act as inducers of P450.
Ontogeny of Monooxygenase System
Yaffe and colleagues first described the monooxygenase activities in the human fetal liver in 1970. A few years later extensive studies by Pelkonen and coworkers (1973) demonstrated that the human fetal liver contained an appreciable amount of P450 and its associated electron transfer chain components and was able to actively carry out a variety of reactions. However, certain activities remained extremely low and suggested that P450s could develop independently in the liver (Pelkonen et al., 1973). By preparing antibodies against human P450s, it was determined by immunoblotting that CYP2C was absent from the fetal liver, whereas CYP3A was actively synthesized, a finding that emphasized the role of CYP3A during the fetal period (Cresteil et al., 1985).
Microsomes prepared from the fetal liver were active not only against exogenous drugs but also against endogenous compounds like lipids and a wide variety of steroids. In 1982 it was demonstrated that if the level of 6β-hydroxylation of testosterone was lower in the fetal liver than in the adult liver, the 16α-hydroxylation of dehydroepiandrosterone (DHEA) was several times higher in fetal liver preparations (Cresteil et al., 1982). Therefore, it was concluded that the major function of P450 in the fetal liver is to eliminate endogenous hydrophobic molecules. To confirm this information based on protein determinations, reverse transcription-polymerase chain reaction (RT-PCR) has been used to detect the presence of ribonucleic acids (RNAs) encoding P450 isoforms. In fetuses of 11 to 24 weeks of gestational age, no RNA coding for CYP1A, CYP2A, CYP2B, or CYP2E has been detected, whereas RNA encoding CYP3A was the major isoform found in the liver (Hakkola et al., 1994).
Exploration of Neonatal Period
To understand the ontogeny of P450 isoforms, the liver content of individual P450 proteins and RNA and the capacities of these proteins to metabolize endobiotic and xenobiotic compounds were investigated by using postmortem liver samples from newborns and children between the ages of 1 hour and 10 years who had died from various pathologies, such as infection, hypotrophy, malformation, respiratory distress, or sudden death. To eliminate the possible interaction of pathologies with the regulation of drug-metabolizing enzymes, the level of expression was carefully examined. In most cases the cause of death did not play a crucial role in the regulation of expression of the monooxygenase system. Similarly, infants receiving drugs known for their inductive capacity were excluded from the study. The total P450 content was shown to remain stable from the first trimester of gestation to 1 year of age and account for about 30 percent of the adult level (Treluyer et al., 1997).
CYP2C proteins were confirmed to be definitively absent from the fetal liver but to rise during the first week after birth, regardless of gestational age at
birth. After 1 week the level of CYP2C remains fairly stable up to age 1 year but does not exceed 30 percent of the adult level.
Two enzymatic activities that depend on CYP2C—the hydroxylation of tolbutamide, and the demethylation of diazepam—were measured in these samples and were found to parallel the evolution and rise in the level of the protein content after birth. The coordinated increases in these enzymatic activities suggest that the two proteins are coregulated. This is also evidenced indirectly by the examination of the RNA contents of developing livers. RNAs coding for CYP2C8, CYP2C9, and CYP2C18 were individually estimated by RT-PCR: RNA for CYP2C9 was the predominant RNA synthesized in the liver but the levels of RNAs for CYP2C isoforms increased during the first week following birth. It is likely that the activation of CYP2C genes occurs during the first week after birth, promoting the synthesis of CYP2C proteins and related activities.
This early rise in the levels of CYP proteins in the liver was also confirmed with in vivo data. Urine samples from infants given diazepam for sedative purposes were collected and analyzed. The production of metabolites was very low in infants aged 1 to 2 days, was notably higher after 1 week of age, and then remained stable up to age 5 years (Treluyer et al., 1997).
The major P450 subfamily expressed in the human liver is the CYP3A subfamily, which comprises three isoforms: CYP3A4, CYP3A5, and CYP3A7. Although highly homologous in terms of protein sequence (CYP3A5 is 87 percent homologous to CYP3A7 and 84 percent homologous to CYP3A4), these proteins display different substrate specificities and different patterns of expression. Thus, CYP3A7 is mostly expressed in the fetal liver, whereas CYP3A4 is the major P450 isoform present in the adult liver, accounting for 40 percent of the total P450 content (Cresteil et al., 1985). In subsequent studies, CYP3A4, CYP3A5, and CYP3A7 were recognized by antiserum against CYP3A4: the amount of reacting material was similar in all samples, whatever the age of the child, indicating that the overall level of the three isoforms was nearly constant from the first third of gestation to adulthood (Lacroix et al., 1997).
To discriminate between CYP3A4 and CYP3A7 two approaches have been used
By using an oligonucleotide specific for CYP3A4, the amount of RNA was estimated. As expected, the CYP3A4 RNA content was low in fetuses and increased after birth to reach a plateau during the first week after birth.
Specific substrates were designed to probe the two proteins. For that purpose, CYP3A4, CYP3A5, and CYP3A7 have been cloned and transfected into a human-derived cell line. Expressed CYP3A7 is able to actively convert DHEA into its 16-hydroxy metabolite, whereas CYP3A4 retains a very low level of activity. The 6-hydroxylation of testosterone, which is extensively catalyzed by CYP3A4 and which is only moderately catalyzed by CYP3A7, is used to probe the CYP3A4 activity (Mäenpää et al., 1993). With these two sub-
strates the relative evolution of both proteins was estimated during the perinatal period. As expected, the activity of CYP3A4 was shown to rise during the first week after birth, whereas CYP3A7 exhibits a high level of activity in fetuses, is maximal during the first week after birth, and thereafter declines to reach an extremely low level in adults (Lacroix et al., 1997).
These studies confirm that recombinant P450 proteins retain the same catalytic properties as the native protein present in liver microsomes and suggest the suitability of this model for the prediction of biotransformation reactions. This model was further used to evaluate the hydroxylation of cortisol by liver microsomes, since this reaction of 6β-hydroxylation has been proposed as an index that can be used to probe CYP3A activity. Only the CYP3A4 protein and, to a lesser extent, the CYP3A5 protein are able to actively hydroxylate cortisol, whereas CYP3A7 exhibits a very low level of activity. This result contradicts a recent report in which the 6-hydroxy cortisol/cortisol ratio was high at birth and progressively declined at the moment when CYP3A4 is supposed to replace CYP3A7. To date there is no explanation for this discrepancy.
Another point to consider is the polymorphic expression of CYP3A5. Twenty to 25 percent of the adult population is positive for CYP3A5, The proportion of the fetal and neonatal populations who are positive is the same, indicating that the fraction of the population that expresses CYP3A5 is the same at all ages. As for CYP3A4, the surge in the CYP3A5 level occurs during the first week after birth, and the amount of protein detected in liver microsomes remains nearly constant during the first year of life.
For CYP1A2, the protein develops very late during the postnatal period: the first rise in the protein level is observed only during the 3 months after birth, progressively increasing into adulthood. The evolution of the protein is correlated with the increase in the associated enzymatic activities, such as the deal-kylation of methoxy-and ethoxyresorufin (Sonnier and Cresteil, 1998).
This late evolution of the CYP1A2 protein was confirmed with several other substrates. For instance, caffeine could be demethylated by CYP1A2 and hydroxylated by 3A: when caffeine was incubated in vitro with liver microsomes, the C8 hydroxylation was active in fetal preparations, whereas the demethylation by CYP1A2 remained negligible before a rise in its level during the first trimester (Cazenave et al., 1994).
Other CYP proteins are expressed during the perinatal period. Thus, CYP4A which supports the hydroxylation of fatty acids, is present in the fetal liver and its levels increase immediately after birth to its final levels during the first days after birth, probably in relation to the hydrolysis of triglycerides and the release of fatty acids that occur during this period (Treluyer et al., 1996). This release of fatty acids could also play a role in the regulation of CYP2E1. The protein is absent from fetal samples and surges during the first hours following birth. It is well known that ketone bodies generated during fatty acid
degradation act by stabilizing the CYP2E1 protein. It is likely that the high level of plasma ketone bodies during the first days after birth (which, in fact, correspond to those after a 3- to 4-day fasting for adults) is responsible for the sudden surge in CYP2E1 in the livers of newborns (Vieira et al., 1996, 1998). Finally, CYP2D6 and CYP2A6 levels rise during the first days or week following birth (Jacqz-Aigrain et al., 1993; Treluyer et al., 1991).
These data allow the classification of P450 isoforms into three groups according to their patterns of expression:
the fetal group that includes CYP3A7 and CYP4A, which act on endogenous substrates (steroids and fatty acids) and which are implicated in the elimination of those substances from the body;
an early neonatal group composed of a majority of P450s that develop quickly during the hours or days after birth; and
the late neonatal protein CYP 1A2.
The data presented here emphasize the role of phase I enzymes during the early neonatal period in the human liver and raise questions about the susceptibility of children to hazards in the food chain, drug intake, or chemical exposure. Enzymes expressed in the human liver early in life could activate or inactivate chemicals. Moreover, the relative intensities of the different pathways that drugs undergo could vary in relation to age. For example, the metabolism of imipramine leads to the formation of several derivatives: in adults, the major metabolite formed is desmethylimipramine, which is essentially formed by CYP1A2, whereas the hydroxylation at position 2 is supported by CYP2D6.
According to age, the metabolism of imipramine demonstrates the relative evolution of P450: in fetuses there is no CYP2D6 and no CYP1A2. The biotransformation of imipramine remains very low (Cresteil, 1999). Immediately after birth the levels of CYP2D6 increase and the level of formation of 2-hydroxy imipramine is significantly increased. Later, levels of CYP1A2 surge and the formation of the desmethyl derivative predominates. This demonstrates that the evolution of the different pathways that drugs undergo is related to the evolution of the P450 responsible for the reaction, which could differ between children and adults. This makes it hazardous to extrapolate data for adults to children. To conclude, the xenobiotic metabolizing system is well developed in the livers of human newborns and neonates. The levels of most phase I and II enzymes rise during the first weeks after birth, regardless of gestational age at birth. The capacity of the human liver to eliminate xenobiotic compounds during the neonatal period is effective and the intensity of biotransformation depends primarily on the level of maturation of phase I enzymes.
To anticipate the metabolic fate of chemicals in the developing liver, it is possible to use proteins expressed in eukaryotic cell expression system. This has been done with steroids as a model and could be extended to other drugs. For example, an antiprotease molecule is extensively metabolized in vitro by adult human liver microsomes into four derivatives. As shown with recombinant human P450 the metabolism seems to be mostly dependent on CYP3A4 and, to a lesser extent, on CYP2C9 and CYP3A5, whereas CYP3A7 has no or little activity. This could result in a low level of biotransformation in the fetal and early neonatal liver. A consequence could be that in the case of a pregnant woman treated with antiprotease, biotransformation in the fetal liver should be very limited.
With knowledge of the biotransformation pathway of a drug and the ontogenic profiles of CYP proteins, it becomes possible to predict the metabolism and to potentially estimate the risk of drug exposure during the perinatal period. However, these biochemical data can give only an estimation and require experimental confirmation before definitive conclusions can be reached about the therapeutic or toxicological effects of chemicals in developing beings.
GENE EXPRESSION AND ONTOGENY OF DRUG METABOLISM
J. Steven Leeder, Pharm. D., Ph.D.
Associate Professor of Pediatrics and Pharmacology, Children's Mercy Hospital and Clinics
At present, there is significant interest in understanding how pharmacogenetics and pharmacogenomics may improve the knowledge of interindividual variability in the clinical responses to therapeutic agents. In the context of pediatric pharmacotherapy, genetic and environmental determinants of variability are superimposed on a changing background of developmental and maturational processes that add further complexity to the optimal use of medications. For example, the weight of a newborn will double by 5 months of age and triple by 1 year of age, whereas caloric expenditures increase three-to fourfold over this same period of time. Thus, it is not unlikely that other functions, such as drug biotransformation, will undergo profound changes in this period of rapid growth and development. Current knowledge related to the expression of drug-metabolizing enzymes in infants and children during the first year of life is reviewed below with a focus on the cytochromes P450 (CYPs).
For many years, it was considered dogma that drug biotransformation capability was limited at best in the fetus and newborn but increased over the first year of life to levels in toddlers and young children that generally exceeded the adult capacity. In fact, in several situations examination of clinical pharmacokinetic data has revealed discernible patterns of drug clearance that can be attrib-
uted to developmental differences in ding biotransformation. As knowledge of mammalian drug biotransformation processes has increased over the past few years, it has become apparent not only that there are developmental differences in expression among drug-metabolizing enzyme families (CYPs, glucuronosyltransferases, etc.) but that individual drug-metabolizing enzymes may have unique developmental profiles (Cresteil, 1998) that influence the therapeutic response, desired or undesired, to a given agent. Specific examples are discussed below.
The CYP3A subfamily consists of three members in humans: CYP3A4, CYP3A5, and CYP3A7. The term CYP3A refers to these isoforms in a collective sense since, historically, it has been difficult to differentiate one from the others on the basis of immunochemical or catalytic properties. Nevertheless, the human CYP3A subfamily is one of the most important drug-metabolizing families in humans. CYP3A4 is the major isoform expressed in adult liver (Schuetz et al., 1994) and intestine (Kolars et al., 1994) and is known to metabolize more than 50 different drugs of diverse chemical structure. CYP3A7 is predominantly expressed in fetal liver (Komori et al., 1990), whereas CYP3A5 is the major CYP3A isoform expressed in human kidney (Schuetz et al., 1992); in contrast, CYP3A5 is expressed in only 25 percent of liver samples (Schuetz et al., 1994). In vitro studies indicate that CYP3A7-dependent DHEA 16α-hydroxylase activity is very high in fetal liver and shows maximal activity in the early neonatal period with a progressive decline thereafter. In contrast, CYP3A4 activity as measured by testosterone 6β-hydroxylation is essentially absent from the fetal liver but increases during the first week of post natal life (Lacroix et al., 1997).
Acquisition of CYP3A activity in vivo is largely inferred from studies of cortisol 6β-hydroxylation in newborns. Nakamura and colleagues (1998a) observed that the ratio of 6β-hydroxycortisol to free cortisol in spot urine samples obtained within 24 hours of birth was higher in term newborns (16.6 ± 1.9; n = 39) than in premature newborns (5.3 ± 0.9, n = 42; p < .001). Significant positive correlations were observed between the 6β-hydroxycortisol-to-free cortisol ratio and gestational age as well as, to a lesser extent, between the ratio and birth weight, suggesting that the level of CYP3A activity is higher in more mature infants. However, the ratio in term infants declined over the first 3 to 5 days after birth to levels comparable to those in premature infants and to levels similar to those observed in adults (Nakamura et al., 1998a). Subsequent work has revealed that the ratio of 6β-hydroxycortisol to free cortisol in the urine of marare infants on the day of birth is independent of that in the urine of their mothers and, presumably, therefore, is independent of maternal CYP3A activity (Nakamura et al., 1999). Although it appears that CYP3A4, CYP3A5, and CYP3A7 are all capable of hydroxylating cortisol in the 6β position, CYP3A4 appears to be 5.7- and 11.4-fold more active than CYP3A5 and CYP3A7, respectively (T. Cresteil, Institute Gustave Roussy, Villejuif, France, personal communication, May 28, 1999).
A literal interpretation of the in vivo data presented by Nakamura and colleagues (1998a, 1999) therefore implies that CYP3A4 activity is compromised in premature newborns, whereas in term newborns, CYP3A4 activity is highest on the day of birth, with a gradual decline during the postnatal period. The data for term newborns are in marked contrast to the in vitro data obtained with human fetal hepatic microsomes cited above (Lacroix et al., 1997). In this regard, concern has been raised that the urinary 6β-hydroxycortisol-to-free cortisol ratio in not an adequate representation of hepatic CYP3A activity (Watkins, 1994), and thus, developmental changes in renal CYP3A5 activity could conceivably account for the observed maturational profile of cortisol 6β-hydroxylation, a possibility that has not been addressed directly either in vitro or in vivo. On the other hand, pharmacokinetic studies of other drugs that can serve as markers of CYP3A activity, such as midazolam, indicate that CYP3A activity in newborns is indeed reduced compared with that in older infants.
When administered intravenously, midazolam clearance reflects the CYP3A activity in the liver (Kinirons et al., 1999). On the basis of data from a population study of intravenous midazolam pharmacokinetics in critically ill neonates (Burtin et al., 1994), it is apparent that although considerable interindividual variability in midazolam clearance exists in this patient population, clearance (and thus hepatic CYP3A activity) is markedly lower in neonates less than 39 weeks of gestation (1.2 ml/kg/min) and greater than 39 weeks of gestation (1.8 ml/kg/min) relative to the clearance of 9.1 ± 3.3 ml/kg/min observed in infants greater than 3 months of age (Payne et al., 1989). The finding of low concentrations of 1 '-hydroxymidazolam in neonates (Burtin et al., 1994) provides further confirmation that functional CYP3A activity is limited in the newborn period.
The midazolam data cited above suggest that CYP3A activity increases approximately five fold over the first 3 months of life (Payne et al., 1989). Carbamazepine (CBZ) represents an additional therapeutic entity that can be used to follow the maturation of CYP3A function in children since the formation of its major metabolite, carbamazepine 10,11-epoxide (CBZ-E), is largely a CYP3A4-mediated process (Kerr et al., 1994). Data from therapeutic drug monitoring databases and pharmacokinetic studies indicate that the rate of CBZ clearance is greater in children than in adults (Pynnönen et al., 1977; Riva et al., 1985) necessitating the administration of higher doses (on a milligram-per-kilogram basis) to children to achieve and maintain therapeutic concentrations. Korinthenberg and colleagues (1994) demonstrated that the ratio of CBZ-E to CBZ in plasma decreases over a period of time spanning the first year of life to 15 years of age. Although factors other than CYP3A4 activity (i.e., microsomal epoxide hydrolase activity) may also influence this ratio, these data are consistent with increased CYP3A4 activity in early childhood, with a gradual decline to levels approximating those in adults occurring around adolescence.
Phenytoin is widely used for the treatment of seizure disorders in children and adults. Biotransformation of phenytoin to (S)-5-(4-hydroxyphenyl)-5-
phenylhydantoin (S-HPPH) by CYP2C9 and subsequent conjugation with glucuronic acid represent the principal metabolic pathway by which the ding is eliminated from the body. Phenytoin can also be metabolized by CYP2C19 to yield R-HPPH. Under normal conditions, 95 percent of the HPPH recovered in the urine is the CYP2C9 product S-HPPH (Fritz et al., 1987). However, as plasma phenytoin concentrations increase from 5 to 60 µM, the contribution of CYP2C19 to overall phenytoin biotransformation is estimated to increase three-fold (Bajpai et al., 1996). Nevertheless, changes in phenytoin pharmacokinetics during development provide some insight into the maturation of CYP2C9 function.
In preterm infants, the phenytoin half-life is prolonged and highly variable (75.4 ± 64.5 hours) relative to that in term infants less than 1 week postnatal age (20.7 ± 11.6 hours) or to that in term infants greater than 2 weeks of age (7.6 ± 3.5 hours) (Loughnan et al., 1977). In vitro, CYP2C9-mediated phenytoin metabolism is saturable (Bajpai et al., 1996). Bourgeois and Dodson observed that saturable phenytoin metabolism was not apparent until approximately 10 days postnatal age suggesting that the acquisition of functional CYP2C9 activity was delayed over this time period. Data derived from phenytoin dosage individualization procedures in Japanese children ages 6 months to 15 years indicate that the Michaelis-Menten parameter Km, is less than 20 µM (5 µg/ml) in the majority of patients (Chiba et al., 1980). These data indicate that CYP2C9 is primarily responsible for phenytoin elimination in this population given that phenytoin Km values determined in vitro are 5 and 70 µM for CYP2C9 and CYP2C19, respectively (Bajpai et al., 1996). The Japanese data further indicate that Vmax values clearly decline as one approaches adolescence. Although changes in phenytoin bioavailability may also contribute to the latter finding, the investigators cite data indicating that the fractional excretion of HPPH in urine does not vary over the age range studied, thus implying that the observed decrease in Vmax is a function of decreased CYP2C9 activity during childhood (Chiba et al., 1980). This fact would then account for the higher phenytoin dosage requirements (on the basis of body weight) in younger children compared with those in adults (Leff et al., 1986).
Caffeine and theophylline are two compounds that are widely used in infants and children. For both compounds, CYP1A2 is the primary route of metabolic clearance. For theophylline, clearance is considerably lower in infants at the time of birth but increases over time. Furthermore, immature infants have a very limited capacity to metabolize the drug, and the majority is excreted unchanged in the urine. CYP1A2-mediated metabolism to 1,3-dimethyluric acid becomes quantitatively more important with increasing age, a process that appears to be completed at about 5 to 6 months of age (Kraus et al., 1993). These changes in theophylline biotransformation are accompanied by increased dosage requirements over the first year of life (Nassif et al., 1981). Elimination of caffeine is also dependent upon CYP1A2 activity, and its developmental profile is similar to that of the ophylline (Aranda et al., 1979; Le Guennec and Billon, 1987). Consis-
tent with in vitro data (Cresteil, 1998), functional CYP1A2 activity is among the last of the P450 activities to be acquired by the newborn and appears to be further delayed in breast-fed infants (Le Guennec and Billon, 1987).
The ontogeny of CYP2D6 activity has not been well characterized to this point in time. Investigators have initiated a study designed to test the hypothesis that within the first year of life acquisition of CYP2D6 activity consistent with genotype is dependent upon both gestational age and postconceptional age. Dextromethorphan phenotyping (administration of 0.3 mg of dextromethorphan DM as Robitussin® Pediatric per kg of body weight after the last evening feed techniques) is conducted with infants during the first year of life at six points timed to coincide with well-baby visits to a primary care physician. Overnight urine recovered from diapers is analyzed by high-pressure liquid chromatography for levels of dextromethorphan and three metabolites. Although the data are preliminary and have not yet been subjected to peer review, data to date for 45 samples obtained 13.9 ± 2.9 (mean ± standard deviations) days after birth indicate that at this postnatal age, the CYP2D6 phenotype (ratio of dextromethorphan to dextrorphan in urine) is consistent with the corresponding genotype determined as described above. However, changes in the pattern of dextromethorphan metabolite excretion in urine suggest that maturation of CYP3A (possibly intestinal CYP3A) is delayed relative to that of hepatic CYP2D6, occurring over the first 4 months of life.
In summary, available information concerning the developmental regulation of individual CYP isoforms is inferred from pharmacokinetic studies of drugs considered to be model substrates for those particular CYP isoforms. In most cases, the data consist of serial measurements of the parent compound from which clearance is estimated and compared with values obtained for adults. Concurrent data for metabolites would be extremely valuable since the ability to characterize individual drug biotransformation pathways becomes more likely. Longitudinal phenotyping studies with healthy children and specific disease populations may help bridge the gap between preclinical in vitro drug biotransformation studies. Subsequent pharmacokinetic studies. These studies may provide important information concerning the effects of disease processes on drug disposition. Ultimately, the goal of developmental pharmacogenetic studies is to better understand the determinants of interindividual variability during childhood such that pharmacotherapeutics can truly be optimized for children of all ages.