The Necessity and Challenges of Clinical Research Involving Children
At a Congressional briefing … [i]t was my husband Joe and our daughter Becca who spoke. Becca was 16 and becoming a vibrant young lady looking forward to her life. The only obstacle that stood in her way was a malignant brain tumor … [Becca] stood at the podium next to her Dad and spoke only a few sentences. She believed it was important for her to support cancer research. It gave her hope.
Maureen Lilly, parent, 2000
As discussed in Chapter 1, children and families have benefited greatly from advances in biomedical science, medical care, and public health achieved during recent decades. For medical problems for which these advances have not resulted in prevention, cure, or improved health, the promise of future progress still gives many children and families hope in dark times.
This chapter discusses why research with adults cannot simply be generalized or extrapolated to infants, children, and adolescents and, thus, why research involving children is essential if children are to share fully in the benefits derived from advances in medical science. Most obviously, some conditions—such as prematurity and many of its sequelae—occur only in children. Similarly, certain genetic conditions such as phenylketonuria (PKU) will, if untreated, lead to severe disability or death in childhood. The diagnosis, prevention, and treatment of these conditions cannot be adequately investigated without studying children. Other conditions such as influenza and certain cancers and forms of arthritis occur in both adults
and children, but their pathophysiology, severity, course, and response to treatment may differ for infants, children, and adolescents. Treatments that are safe and effective for adults may be dangerous or ineffective for children. Many of the examples cited in this report involve drugs, but clinically significant differences between children and adults extend to other areas. Radiation therapy can, for example, disrupt normal tissue development in children.
Clinical research involving infants, children, and adolescents is, in some important respects, more challenging than research involving adults. As reviewed in this chapter, these challenges include the relatively small numbers of children with serious medical problems, the need for developmentally appropriate outcome measures for children of different ages, the complexities of parental involvement and family decision making, and the adaptations required in research procedures and settings to accommodate children’s physical, cognitive, and emotional development. Understanding and complying with the special ethical and regulatory protections for children constitutes another challenge. These various challenges underscore the need for those reviewing research protocols that include children to have adequate expertise in different areas of child health and research.
The next sections of this chapter provide additional historical context; discuss definitions used for the periods of infancy, childhood, and adolescence in different research arenas; and expand on the rationale and complex challenges of pediatric research. The last sections describe government policies adopted in recent years to encourage research involving children while protecting them from research risks.
The ideal laboratory species for accumulating data on human functions and reactions is human, and that “animal of necessity” has been widely utilized in research in the biomedical sciences in the second half of the twentieth century.
Faden and Beauchamp, 1986, p. 152
If adult humans have been the “animal of necessity” in clinical research, then children have often been “therapeutic orphans,” as characterized over 35 years ago by clinical pharmacologist Harry Shirkey (1968, p. 119). To a considerable degree, children retain this disadvantaged status, despite the recent creation of policy incentives for clinical research involving children.
As described in Chapter 1, a survey of the 1991 edition of the Physician’s Desk Reference found that approximately 80 percent of the listed medications had labels that provided no prescribing information for
children (Gilman and Gal, 1992, cited in AAP, 1995). Based on data from 1991 to 1997 involving new molecular entities with potential pediatric uses, a Food and Drug Administration (FDA) report found that 62 percent lacked labeling information for pediatric use at the time that they were initially approved for marketing (Steinbrook, 2002). The committee did not locate similar information about medical devices and biological agents.1
When drug labels lack pediatric prescribing information, physicians can still legally prescribe drugs for children on an “off-label” basis—and they do. According to Choonara and Conroy (2002), European studies suggest that at least one-third of hospitalized children and up to 90 percent of neonates in intensive care receive such prescriptions. A study in the United States, which used 1994 data on outpatient prescriptions, reported more than 1,600,000 off-label prescriptions of nebulized albuterol for children under age 12 years, nearly 350,000 prescriptions of the anti-depression drug fluoxetine (Prozac) for children under age 16 years, and more than 200,000 off-label prescriptions of methylphenidate (Ritalin, which is used to treat attention deficit disorder) for children under age 6 years (Pina, 1997; see also Turner et al., 1999 and Conroy et al., 2000). Altogether, the 10 identified drugs were prescribed more than five million times for children in age groups for which the drug label either had a disclaimer or lacked information for children. Since 1994, the FDA has reviewed supporting studies and approved pediatric labeling for several of these drugs (e.g., Prozac for ages 7 to 17 years, Ritalin for ages 6 to 12 years, leval-buterol down to age 7 years) (FDA, 2004).
The American Academy of Pediatrics has argued that the shortage of pediatric research creates an ethical dilemma for physicians, who “must frequently either not treat children with potentially beneficial medications or treat them with medications based on adult studies or anecdotal empirical experience in children” (AAP, 1995, p. 286). Many children undoubtedly benefit when physicians follow the second course. On occasion, however, some children will experience harm, either because the dose used was ineffective or because it was toxic. Even those children who receive some
benefit may not receive optimal treatment because their physicians lack validated prescribing information.
Although most of the concern about expanding research involving children has focused on differences between adults and children as they relate to drugs, it is also important to consider differences in other therapeutic arenas. For example, intraocular lens replacement after cataract surgery has been a standard of care for many years in adults, and in young children it can not only treat a loss of vision but also improve visual development. Use of the procedure with children, however, presents unique developmental issues (see, e.g., Dahan, 2000; Ahmadieh and Javadi, 2001; Good, 2001; and Pandey et al., 2001). For example, children’s eyes show lower scleral rigidity, greater elasticity of the anterior capsule, and higher vitreous pressure. Also, the refractive state of children’s eyes changes as children grow. These characteristics make intraocular lens replacement in young children surgically and developmentally different from similar procedures for adults. In addition, the sizes of the replacement lenses developed for adults are not appropriate for young children (the mean axial length of a newborn’s eye is 17 mm, whereas that of an adult is 23 to 24 mm). Surgical treatment and device placement for children present special developmental issues and requirements for pediatric studies in many other areas, including cardiology and orthopedics.
For research sponsors and investigators, the challenges of research involving children are compounded by normal developmental variability. For many conditions and interventions, separate studies are required for infants, young children, and adolescents. Several pediatric formulations of medications may ultimately be required (e.g., for acetaminophen, two strengths of chewable tablets, a low-strength “swallowable” tablet, a syrup, and drops in a different concentration for infants).
Furthermore, compared to adults, children generally represent a smaller market for commercial sponsors of research. The commercial value of various preventive, diagnostic, and therapeutic options for children, especially for rare diseases, may not be enough to offset the costs of developing them. Even for relatively common childhood conditions, the numbers of potential research participants may be small and thus require more study sites and additional costs for coordination. Development costs may also be increased because more time is often required per patient to complete study procedures and because more expensive, specialized laboratory studies may be required for small-volume biological samples. The widespread off-label prescription and use of drugs for children tend to further diminish the incentives to finance pediatric research on drugs that are already approved for use by adults. In addition, companies may be unfamiliar with the clinical, ethical, and regulatory requirements for pediatric studies, and they may
be concerned about financial or public relations consequences of adverse experiences in research involving children.
As recounted in Chapter 1, reactions against abusive or questionable research practices involving both adults and children have led to an evolving set of policies and practices to protect all human participants in research, with additional protections for children and other vulnerable populations. The adoption of special protections for child research participants and the growing awareness of researchers’ ethical obligations have curbed what are now regarded as unethical and harmful research practices. Notwithstanding these benefits, some of these protections have also made some research involving children more administratively burdensome in certain respects than research involving adults. Debate continues about what constitutes an appropriate balance between scientific priorities and protection of child participants in research.
Later sections of this chapter describe further the rationale and complex requirements for valid, safe pediatric research. The next section discusses the different definitions of infants, children, and adolescent used in different research contexts.
DEFINITIONS: INFANTS, CHILDREN, AND ADOLESCENTS
From birth into adulthood, children change and develop physically, cognitively, socially, and emotionally. Although these changes are fairly predictable, parents, clinicians, and others who interact regularly with children recognize that children of the same chronological age may develop at different rates. Practical and legal considerations, however, often dictate the organization of services and the definition of legal rights and other policies based primarily or entirely on chronological age.
Legal issues aside, age breakdowns are “a basis for thinking about study design in pediatric patients” (ICH, 2000b, p. 7). The age range specified for particular clinical studies will depend on the research question, for example, whether it involves a condition specific to infants or whether previous research shows relevant age-related physiological differences. The choice of age ranges may also be shaped by policy considerations, research organization missions, convenience, and other factors. Thus, in guidance on drug testing, FDA may reasonably define adolescence more narrowly than other agencies based on considerations of developmental physiology related to how drugs work in the body. A broader adolescent age range, however, is reasonable for the Centers for Disease Control and Prevention (CDC) and other agencies interested in the contribution of adolescent behavior, including risk taking, to problems such as sexually transmitted disease and motor vehicle-related deaths and injuries.
The first months of life are a period of particularly rapid development and change. Thus, in its section on growth and development, the Nelson Textbook of Pediatrics breaks the discussion of the first and second years of life into sections on ages 0 to 2 months, 2 to 6 months, 6 to 12 months, 12 to 18 months, and 18 to 24 months (Behrman et al., 2004). Thereafter, the textbook uses larger age ranges.
The discussion later in this chapter on pediatric pharmacokinetics underscores the significance for drug studies of developmental differences and changes during the earliest period of life. The importance of these differences and changes is reflected in FDA’s guidance on pediatric drug testing, which categorizes those under age 2 years as infants (FDA, 1994a; 21 CFR 201.57(f)(9)(i)) and also distinguishes the category of neonates (infants less than 28 days of age).2 The drug testing guidelines of the International Conference on Harmonisation (ICH) identify two additional categories—preterm infants and early neonates (infants less than 7 days of age) (ICH, 2000b; FDA, 2000b). The guidelines emphasize the unique and highly variable pathophysiology of preterm infants and their different responses to medications by stage of gestation. The guidelines stress the importance of involving neonatologists and neonatal pharmacologists in the development of research protocols for studies with neonates.
In health services research, epidemiologic studies, and public policy discussions, infants are typically defined as children under age 1 year, and early neonates and neonates are defined as described above for the FDA. These age categories, which are based on National Center for Health Statistics (NCHS) guidelines for vital statistics reporting (Kowaleski, 1997), reflect in part the interest of public health officials and epidemiologists in data to guide policies and programs to reduce infant mortality. Approximately half of all deaths among individuals under age 20 years involve infants, and the majority of these occur soon after birth (Arias and Smith, 2003).
The term child is often used broadly—as it is in the title and, frequently, the text of this report—to refer to individuals ages 0 to 19 years (i.e., under age 20 years) or to cover all persons below the age at which a person can provide legal consent to medical treatment (usually age 18 years in the
United States). The definitions of child in federal regulations on human research protections are cast in legal terms and do not cite an age or age range.3
In guidance encouraging testing of drugs in studies involving children for the purpose of establishing safe dosing levels, the FDA categorizes those ages 2 to 11 years (“up to 12 years”) as children (FDA, 1994a; 21 CFR 201.57(f)(9)(i)). The ICH guidelines cited above also use this age range. The guidelines stress the importance of investigating effects of medications on growth and development in this age group. The Nelson Textbook of Pediatrics uses a slightly broader age range—ages 2 to 12 years—and distinguishes early childhood (ages 2 to 5 years) and middle childhood (ages 6 to 12 years) (Behrman et al., 2004).
The National Institutes of Health (NIH) policy statement on the inclusion of children in research is expansive, defining a child as “an individual under the age of 21 years” for purposes of that policy (NIH, 1998a, unpaged).4 The policy does not differentiate among infants, older children, and adolescents and does not present a rationale for the age range selected. The policy does, however, require research proposals to describe the rationale for including or excluding particular age groups. The NIH statement notes that its policy applies notwithstanding the different age range used by FDA. It also notes that the definition differs from the regulations of the Department of Health and Human Services governing children’s participation in federally conducted, supported, or regulated research. Under NIH policy, an 18-year-old would be an adult for consent purposes (under state law) but a child for study inclusion purposes.
The term adolescent seems particularly variable in its definition, depending on the medical, public health, or psychosocial context in which it is used. NCHS observes that adolescence is “generally regarded as the period of life from puberty to maturity; [but] the meaning of ‘puberty’ and ‘maturity’ are often debated by health professionals” (NCHS, 2000, p. 19). Ado-
The regulations state that children are “persons who have not attained the legal age for consent to treatments or procedures involved in the research, under the applicable law of the jurisdiction in which the research will be conducted” (45 CFR 46.402(a); 21 CFR 50.3(o)). Chapter 5 and Appendix B discuss these regulations and state policies that allow adolescents, under certain circumstances, to make decisions about health care in their own right.
The policy statement was developed in response to language in House and Senate Appropriations Committee reports for fiscal year 1996 that noted the need for the more widespread inclusion of children in research (NIH, 1998a).
lescence is clearly a period of physical and psychosocial maturation and vulnerability related to hormonal changes, changes in appearance, and transition toward adult roles and responsibilities. The metabolic and other effects of hormonal changes associated with puberty may alter disease processes (e.g., in patients with diabetes) or contribute to the onset of medical problems (e.g., depression or polycystic ovary syndrome), with corresponding implications for disease prevention, diagnosis, and management (see, e.g., Janner et al., 1994; Travers et al., 1995; Angold et al., 1999; NRC/ IOM, 1999; Schultz et al., 1999; Driscoll, 2003; and Sarnblad et al., 2003). Changes associated with puberty are an important consideration in much drug research, and the combination of physical, emotional, and social changes makes adolescence a particularly challenging period for psychosocial research.
In its regulations on pediatric drug testing, FDA uses a narrow definition of adolescence—ages 12 to 15 years (“up to 16 years”) (FDA, 1994a; 21 CFR 201.57(f)(9)(i)). The ICH guidelines, however, refer to adolescents as those aged 12 to 16 or 18 years (with the observation that the upper limit “varies among regions”) (ICH, 2000b, p.10). When adolescents are included in studies that also include adults, the guidelines suggest that it may appropriate “to consider studying adolescent patients … in centers knowledgeable and skilled in the care of this special population” (ICH, 2000b, p. 10). Recently, in draft guidance on the testing of medical devices, FDA proposed a broader age for adolescents—12 to 21 years—citing “the impact that a device could have on a growing adolescent as well as the effect growth could have on the device” as a rationale for the upper age limit (FDA, 2003b, p. 3). The agency noted that other factors—including weight, physiological development, and neuromuscular coordination—may be more relevant than chronological age for the assessment of device safety and effectiveness.
The Nelson Textbook of Pediatrics describes three periods of adolescence: early (ages 10 to 13 years), middle (ages 14 to 16 years), and late (ages 17 to 20 years and beyond) (Behrman et al., 2004).5 The text observes that the first visible signs of puberty usually occur between the ages of 8 and 13 years and thus the period of early adolescence overlaps with the period of middle and late childhood. In a general statement on the age limits for pediatrics practice, the American Academy of Pediatrics (AAP)
noted that the responsibility of pediatrics may continue through age 21 years and even beyond under special circumstances (AAP, 1988).
Clearly, complete consensus does not exist about the age ranges that define infancy, childhood and adolescence. Definitions may reasonably vary depending upon the type of research being conducted. Given the practical challenges of recruiting adequate numbers of children for clinical research, as discussed later in this chapter, investigators may opt for as wide an age range as can be justified given the particular intervention or condition being investigated.
THE RATIONALE FOR PEDIATRIC DRUG RESEARCH
I was a brand-new fellow in pediatric hematology-oncology, in July 1963 … Long-term survivors of childhood leukemia were so rare that I cannot recall a single conversation on the topic during my fellowship … It has been a privilege to participate in studies leading to the dramatic increase in the proportion of long-term survivors.
Joseph V. Simone, 2003, p. 627-628
Much of the progress in pediatric oncology that Simone’s statement acknowledges stems from a concerted effort to identify promising chemotherapeutic agents for childhood cancers, set priorities for the testing of these agents, and then design and conduct clinical trials through a national cooperative network of investigators and research institutions. This effort has been necessary because childhood cancers sometimes differ biologically from adult cancers and because other physiologic differences between children and adults may affect the ways in which chemotherapeutic agents work in the body. Agents that are effective with adults are not always effective with children.
In general, several features distinguish pharmacotherapy in children from that in adults and explain why medicines must be studied in research with children to ensure their safe and effective use (IOM, 2000b; see also, Kearns and Winter, 2003, and Reed and Gal, 2004). These features include
requirements for age-appropriate formulations that allow the accurate, safe, and palatable administration of medicines to children of a wide range of weights and with a wide range of developmental characteristics;
age- and development-dependent changes in how medicines are distributed in and eliminated from the body (pharmacokinetics);
age- and development-dependent changes in the response to medicines (pharmacodynamics);
age- and development-dependent changes in the adverse effects of medicines, both short and long term; and
unique pediatric diseases that require development of unique pediatric medications.
Requirements for Age-Appropriate Formulations
For orally administered drugs, children of different ages often need or prefer formulations that differ from those used with adults. These formulations may include liquids, chewable tablets, rapidly dissolving tablets, and more palatable flavors. A critical impetus for the Federal Food, Drug, and Cosmetic Act of 1938, (the basis for the modern FDA) was a tragedy involving a liquid sulfonamide that cost the lives of over 100 adults and children (Ballentine, 1981; Wax, 1995; Hilts, 2003). When the sulfonamides—the first truly effective antimicrobials—were developed in the 1930s, they came in pill form, which was not suitable for very young children. For a liquid formulation, the manufacturer’s chief chemist tried several solvents before devising a so-called elixir of sulfanilamide dissolved in diethylene glycol, a pleasant-tasting but toxic substance. The mixture, which also included flavoring agents, was evaluated for taste and fragrance but not safety. At least 34 children and 71 adults died of kidney failure after taking this elixir (Wax, 1995).6
Instances of harm to children from other unsafe drug formulations continue to occur. For example, in 1982, FDA warned that 16 premature infants had experienced “gasping syndrome” (resulting from metabolic acidosis leading to respiratory distress and other severe effects) and then died after being given intravenous medicines that contained excessive amounts of benzyl alcohol as a preservative (CDC, 1982). As recently as 2001, the warnings about this syndrome were added to labeling information for an intravenous drug—with the same preservative—that has been used on an off-label basis to treat cardiac arrhythmias in infants (de Vane, 2001). In developing countries, pediatric deaths have been associated with local formulations of acetaminophen containing diethylene glycol (Hanif et al., 1995; O’Brien et al., 1998).
Evaluations of pediatric formulations need to consider not only the safety of excipients (more or less inert substances used as vehicles or media for administering or diluting medications) but also the concentrations of the medicines being administered. For example, many adult intravenous preparations have concentrations of medicines such that the appropriate dose for a small infant is too small (e.g., less than 0.01 milliliter) to be accurately measured in a clinical syringe. Such medicines may also have low water solubility; when a nurse or doctor tries to dilute the medicine in intravenous solutions, the medicine precipitates out and may clog the intravenous line. If they reach the body, the precipitated medications may lodge in the lung and cause serious, even fatal, harm.
Not all formulations of drugs specifically for children come from pharmaceutical companies. For example, parents may be advised to give a half or a quarter of an adult tablet or to crush tablets and place them in applesauce. These strategies can produce both dosing errors and, possibly, drug instability because of uncertainty about the stability of the medicine in foods.
Professional, extemporaneous formulations (e.g., liquids based on crushed tablets) for pediatric medications may be helpful when they are produced by pharmacies with appropriate expertise. However, the validation of the stability and the absorption of the medicine from such formulations (i.e., its bioavailability) is often inadequate, particularly when compared with the standards of “good manufacturing practices” and “good clinical practices” defined by the FDA for formulation development and evaluation by pharmaceutical companies (Nahata, 1992, 1999; ICH, 1996; see also, FDA, 1997 and 21 CFR 210 and 211). In addition, many pharmacists lack the expertise to produce adequate and safe extemporaneous pediatric formulations, so their availability is limited. As discussed later in this chapter, one goal of the Best Pharmaceuticals for Children Act of 2002 is to encourage pharmaceutical companies to develop safe, effective formulations of drugs for children.
Development Affects Drug Distribution in and Excretion from the Body: Pharmacokinetics
Pharmacokinetic studies investigate the way in which medicines are absorbed (including when they are given orally, topically on the skin, or rectally), the way in which they are distributed among organs in the body, and the relationship between the dose and the concentration of a medicine in the blood. (Drug concentrations are typically measured in blood.) These studies are critical in pediatric care because they provide the basis for different formulations of drugs for children that avoid the dangers of either toxicity or underdosing based on extrapolation from studies conducted
with adults. A complete description of developmental changes in pharmacokinetics cannot be included here, but a few examples demonstrate the need for proper pediatric pharmacokinetic studies before clinical trials or general clinical use of medicines with children.
One important reason for pediatric drugs studies is that the absorption of medicines after oral administration (the most common route) varies with age and differs among specific medicines. For example, the anticonvulsant phenytoin is poorly absorbed in newborns; thus, although it might theoretically be a good medicine for the treatment of seizures in newborns, unreliable absorption limits the drug’s use for this purpose. A factor that can affect drug absorption is gastrointestinal transit time. The more rapid transit time of young infants can lead to poor absorption of many sustainedrelease formulations developed for adults.
Absorption of drugs through the skin also varies with age. Compared with adults, newborns and young children have a larger skin surface per pound of weight, and other characteristics of infants (e.g., skin thickness, blood) also affect absorption of topically applied substances such as steroid creams. The result can be a much larger dose per pound in a small child with the consequent risk of adverse effects. Examples of such adverse effects include growth inhibition from topical corticosteroids and central nervous toxicity from hexachlorophene, which is used to clean the skin of newborns.
Age and level of development are related to several pharmaco-kinetically relevant variations in the relative sizes of organs, the ways in which medicines attach to proteins in the blood plasma, and the physiologic processes that exclude chemicals from sites such as the central nervous system (e.g., processes that involve what is often called the blood-brain barrier). For example, if sulfonamides are given to premature infants, the drugs will interfere with the safe removal of bilirubin (a by-product of blood metabolism) from the blood by plasma proteins. If permitted to accumulate in the bloodstream, bilirubin can penetrate the infant’s immature blood-brain barrier, which can, in turn, lead to kernicterus (yellow staining of the brain) and brain damage. The objective of pharmacokinetic studies with infants, children, and adolescents is to identify such toxic effects before rather than after a drug is used in pediatric clinical care.
Medicines are excreted from the body by metabolism in the liver and other organs. They are also filtered and excreted by the kidney. Increasingly, scientists are learning that metabolism is mediated by specific enzymes and that each of these enzymes has a different metabolic pattern depending on the child’s stage of development. A specific medicine may be metabolized by one or several enzymes, each with a different pattern of action related to developmental stage. In addition, a child’s stage of development will affect drug excretion and the half-life of the drug (the time it
takes for half the drug to be excreted from the body, which is one of the determinants for choosing a dosing schedule, e.g., once a day versus three times a day).
For example, in newborns, the enzyme glucuronyl transferase, which metabolizes the drug chloramphenicol, has a very low level of activity. When chloramphenicol was first used in newborns in the 1950s, the dose was extrapolated from adult doses without knowledge of the drug’s metabolic pathway. At the time, it was also not possible to determine concentrations of drugs in serum because of a lack of technology to measure drug concentrations in the small quantities of blood that can be safely extracted from neonates, infants, and small children. With the doses selected by individual pediatricians, the drug reached much higher concentrations in infants than in adults, resulting in death from the “gray baby syndrome” (see, e.g., Weiss et al., 1960). The technologies available today allow medicines like chloramphenicol to be more fully evaluated before they are used with pediatric patients.
The rates of maturation of drug clearance pathways after the neonatal period are highly variable and relate to the specific metabolic enzymes or kidney mechanisms that are responsible for drug excretion. (Other variables such as disease state and drug-drug interactions can also influence drug clearance.) Research shows that although neonates excrete many medicines much more slowly than adults, prepubertal children often excrete drugs much more rapidly than adults. Thus, although overdosing based on the extrapolation of doses for adults to doses for neonates is likely, underdosing of children is common.
During puberty, drug clearance and half-life move toward adult levels (Goodman et al., 2001). The changes during this period depend on the specific pathways of clearance of a given medicine (i.e., which enzymes are involved in its metabolism). This process has been studied best for cytochrome P-450 1A2 (CYP1A2), an enzyme that metabolizes caffeine and theophylline (Lambert et al., 1983, 1986; Le Guennec and Billon, 1987). The enzyme activity decreases from childhood levels at earlier pubertal stages in girls than in boys (and, thus, at an earlier age, because girls begin undergoing pubertal stages at younger ages than boys). Once adolescents have reached Tanner stage 4 (of the five stages of sexual maturity identified by Tanner ), drug metabolism and clearance closely resemble adult values.
Recent studies have resulted in changes in a number of recommendations or warnings about the use (or nonuse) of specific medications by children (Meadows, 2003). These changes emphasize the importance of pharmacokinetic studies of drugs that are expected to be beneficial for children of various ages.
Development Affects Response to Medications: Pharmacodynamics
In addition to developmental changes in drug metabolizing enzymes, developmental changes can also affect the drug receptors that mediate how medicines act in the body, that is, their pharmacodynamics. This area has not been studied as systematically as drug metabolism.
As examples of pharmacodynamic differences between adults and children, phenobarbital and antihistamines may produce sedation in adults but excitation and hyperactivity in children. This paradoxical response is thought to be related to central nervous system receptors that have not matured. Other drugs may simply not work in children of certain ages because a receptor that permits the activity of the drug is not present or has not yet developed. Major future challenges for pediatric drug development will be to develop methods to (1) assess receptor development, (2) create clinical tools to assess the relationship between how the body processes a drug and what response the drug triggers in adults, and then (3) determine whether the same relationship holds for children of various ages.
FDA and ICH guidelines provide for minimizing the burden of full clinical trials with children if a combination of clinical information about the similarity of the disease in children and adults, pharmacokinetic and pharmacodynamic measurements, and safety and other studies provide sufficient confidence that efficacy can be extrapolated from adults or older children and that an appropriate dose can be defined (FDA, 2000b). To judge the acceptability of such extrapolations, more data are needed on changes in drug receptors by stage of development and associated pharmacodynamic outcomes.
Adverse Effects of Medicines by Developmental Stage
The very fact that children are growing and developing places them at risk of adverse effects that are not observed in adults. For example, in addition to other side effects that occur in adults, corticosteroids can alter physical growth. To cite another example, tetracyclines stain developing teeth but not teeth that are fully developed.
Not just medicines used for long periods but also those used for relatively short periods with children may have long-term consequences that are not evident for years. Although study designs for short-term clinical studies need to take potential developmental toxicity into account, long-term follow-up and surveillance are important. Such surveillance is important, for example, in tracking developmental and other outcomes for children treated for cancer with chemotherapy agents. Collaborative groups such as the Children’s Oncology Group (COG) have played an important role in monitoring children for long-term outcomes, and their strategies
may serve as models for those studying the developmental consequences of experimental interventions for other acute or chronic illnesses. Novel epidemiologic methods will likely be needed for long-term follow-up of otherwise healthy children treated with various medicines on a short-term basis.
Medicines for Unique Pediatric Diseases
When a medicine is being developed for both adults and children, at least some data are usually available from studies with adults before studies with children start. Even for medicines with unique pediatric indications, studies with children often follow phase 1 trials that provide an initial assessment of the drug’s tolerability and pharmacokinetics in adults. As molecularly targeted therapies are developed with increasing frequency, however, it is likely that more new drugs will be developed specifically for the treatment of conditions that occur only in children. If an agent is not appropriate for initial testing with healthy adult volunteers but findings from findings from prior laboratory and animal studies are favorable, then initial testing is likely to be performed with children who have such a condition. The agent may never be administered to adults. Thus, it is inappropriate to assume or require that drugs must invariably be tested in adults before pediatric clinical trials are initiated.
IMPLICATIONS FOR PEDIATRIC INVESTIGATION
The preceding sections of this chapter emphasize the need for expanded efforts to develop appropriate and safe formulations of medicines for children. The following section outlines a safe, logical and efficient process for doing so.
Before drug studies are initiated with children, several steps should be completed to minimize the number of children required for research protocols and to maximize the quality of the data collected. Whenever possible, the first step should be the completion and evaluation of phase 1 studies with adults to investigate the tolerability, bioavailability and pharmacokinetics of the drug of interest and provide the basis for designing phase 1 and phase 2 pharmacokinetic studies involving children. When feasible, data from these studies or laboratory techniques developed during their conduct may be used to minimize the numbers and volumes of blood samples necessary for pediatric studies. For drugs such as new anticancer agents for which it is desirable to expose as few children as possible to doses too low to be therapeutic, data from phase 1 trials with adults may be used to guide the selection of a pediatric starting dose and a dose escalation scheme for early-phase testing of the drug.
In some cases, the second step should involve additional preclinical
toxicology studies using animals to assess potential developmental toxicity before trials with children are started. The third step should be for investigators to assess the need and specific requirements for special formulations (e.g., liquids or chewable tablets) that are suitable for children at different ages. Taken together, the data from adult and animal studies will guide the development of a rational series of clinical studies that involve children.
For many drugs, the first pediatric study is a single-dose pharmacokinetic study that involves the age groups that are likely to use the drug. This type of study is designed to discover major age-related differences in pharmacokinetics in these groups and to generate preliminary data for the design of subsequent multidose and efficacy studies. In the United States, the majority of single-dose pediatric pharmacokinetic studies involve children with an illness likely to be treated by the drug under study. Depending on what is known in advance based on trials with adults, preclinical studies, and other information, some of these early-phase trials—unlike typical phase 1 trials involving adults—may arguably be viewed as having the prospect of providing a direct benefit to the child participants. The assessment of such research may also consider whether other treatment alternatives have been exhausted and whether the probable outcome without the experimental intervention has been assessed. As described in Chapters 3 and 4, depending on the findings from the specific assessment of potential harms and potential benefits, early-phase studies involving children may or may not be approvable under federal regulations.
After the completion of phase 1 (safety and pharmacokinetic) trials, the next step is the development of appropriate phase 2 (safety and efficacy) studies. As discussed further below, a major challenge for drug manufacturers and pediatric investigators is designing these studies to incorporate appropriate outcome criteria. For all phase 2 studies involving children, trials should be designed to minimize risks and the number of study participants (while ensuring statistically meaningful results).
Broad phase 3 trials with children may or may not be appropriate for a particular drug, depending on the agent and the condition that it is intended to treat. For some “orphan” indications (i.e., rare conditions), a phase 3 or randomized trial may not be practical. Similarly, even if there is a plan for such a trial, the timing may be significantly delayed beyond the drug’s approval by the FDA for adult use. This situation applies to many anticancer agents. In such cases, safety and efficacy assessments should continue beyond marketing approval and may include postmarket surveillance or targeted cohort studies.
One of the greatest ethical challenges is assessing the optimal timing for the initiation of pediatric pharmacokinetic and efficacy trials. Most medicines entering phase 1 trials with adults are never approved because of safety concerns or a lack of efficacy (Kaitin, 2003). For this reason, one
could argue that research with children should await FDA approval based on studies with to avoid children’s exposure to the excessive risks of early-phase trials. As noted above, some drugs are for conditions that do not exist in adults and may be inappropriate or impossible to test in adults. Moreover, particularly for seriously ill children who have exhausted standard treatment options and who may not survive until the FDA approves an investigational drug, approval of a trial may be warranted based on initial favorable findings from preclinical studies and trials with adults. Thus, judgment is required in balancing conflicting concerns. Assessment of each proposed trial should take into account the availability of other therapies, the severity of the disease in question, the assessment of adverse event profiles for adults and other relevant data, and the availability of a suitable pediatric formulation for testing.
In sum, the overall goal for pediatric drug development should be to evaluate the safety and efficacy of drugs for children as soon as possible after adequate data for adults are generated to guide the design of safe and efficient pediatric studies. When data for adults are not available or when waiting for such data may do more harm than good, particular care must be taken to justify proceeding with pediatric trials. Likewise, if a drug will be marketed for adults without a plan in place to evaluate it with children, the rationale should be explained (e.g., no equivalent condition in children, safety concerns based on the results of studies with adult, formulation problems, or a lack of pediatric clinical end points to judge efficacy). Such an explicit rationale will help pediatricians understand the possible implications for “off-label” use of the drug.
CHALLENGES OF DESIGNING AND CONDUCTING PEDIATRIC STUDIES
Clearly, determining how children’s development affects drugs in the body is both a major rationale and a significant challenge for pediatric research. In addition, researchers committed to clinical research involving infants, children, and adolescents face a number of other challenges beyond those typically encountered in research involving adults. These challenges range from practical and methodological problems to regulatory requirements and ethical concerns. Those conducting initial and continuing reviews of pediatric research should have sufficient expertise—personally or through consultation—to assess the appropriateness of the investigator’s strategies for managing challenges relevant to their research topic. This section discusses these challenges.
Defining and Measuring Outcomes and Other Variables
Among the challenges facing pediatric research is defining appropriate outcome measures. This challenge has basic two dimensions. One is the determination of what outcomes are meaningful for children of different ages in the context of a specific study—and then developing reliable and valid ways of measuring these outcomes. The other is the determination of normative data for purposes of comparison, for example, for comparisons between healthy children of different ages and children known or suspected of having a medical problem.
In addition, because children are developing physiologically, anatomically, cognitively psychologically, and socially, possible developmental effects of medications and other interventions may need attention. Assessment of such effects may require lengthy follow-up studies that track children for years, even decades (see further discussion below).
Identifying Age-Appropriate Clinical Outcomes
Among meaningful outcomes, death is a relatively straightforward outcome across age groups, although the accurate identification of the cause of death is a continuing concern, particularly in studies relying on death certificates. For example, for deaths among children, health officials have made particular efforts to distinguish cases of child abuse from instances of sudden infant death syndrome. As research has led to improved therapies and prolonged survival for children with many fatal childhood illnesses (e.g., acute lymphoblastic leukemia, severe prematurity, and cystic fibrosis), death has become—as hoped—an uncommon or long-delayed outcome of these illnesses. As a result, other outcome measures become more significant in further clinical studies.
Survival after diagnosis and disease-free survival are also meaningful outcomes across age groups, but considerable caution may be needed in interpreting survival statistics and changes in survival across time. In particular, changes in diagnostic technologies or the frequency of their use can push the time of diagnosis earlier for many individuals; this can make survival statistics look better independent of any real improvement. Such changes are particularly relevant to genetic disorders that are now being diagnosed prenatally or at birth, before clinical signs develop.
To support accelerated marketing approval of new drugs and biologics aimed at serious pediatric diseases, FDA has in recent years accepted evidence from phase 3 efficacy trials that use “surrogate” physiological measures if they are “reasonably likely, based on epidemiologic, therapeutic, pathophysiologic, or other evidence, to predict clinical benefit” (21 CFR 314.510). This acceptance may be accompanied by requirements for addi-
tional postapproval follow-up studies to demonstrate long-term clinical benefit and safety in the target population. Many convenient physiological measures (e.g., blood pressure, tumor shrinkage, and the levels of different substances in the blood) do not fulfill the FDA criteria for this kind of “surrogate” outcome measure. Although the measures track changes relevant to the administered treatment, these changes may not correspond to physical or mental outcomes that directly improve the child’s life. For example, a drug may shrink a tumor without affecting survival, function, or quality of life.
Pediatric investigators and FDA staff may find it difficult to agree on the best surrogate for a particular disease entity. An acceptable measurement must not only be relevant to the child’s long-term well-being but also be easy to use with children and reproducible when used at multiple sites. Further, sufficient normative data should be available to permit statistical comparisons and analyses.
An example of a surrogate physiologic marker is forced expiratory volume at 1 second (FEV1). For many lung disorders, FEV1 is used as a measure of obstruction (blockage) of airway passages. Until the 1990s, no FDA-approved therapy had been developed for the treatment of patients with cystic fibrosis. Two new therapies were developed in the 1990s, a drug to breakdown bronchial mucus (dornase alfa [Pulmozyme]) and an inhaled antibiotic (tobramycin solution for inhalation [TOBI]). Through a series of meetings, the FDA and the cystic fibrosis community decided that FEV1 was the optimal surrogate (in conjunction with the frequency of pulmonary infections) for this disorder because of the published data relating FEV1 to the survival rate (Kerem et al., 1992; Corey et al., 1997; Liou et al., 2001) and the hospitalization rate (Emerson et al., 2002). As a result, FEV1 was a key factor in FDA approval of these therapies and has become the primary outcome measure for most subsequent clinical trials (Fuchs et al., 1994; Ramsey and Boat, 1994; Ramsey et al., 1999).
Like most surrogates in pediatric research, FEV1 is far from perfect. First, as new therapies continue to improve survival and function (CFF, 2002), FEV1 becomes a less sensitive marker of change and of early disease. Second, measurement of FEV1 requires a voluntary maneuver that most children under 6 years of age cannot reproducibly perform, which means it cannot serve as a surrogate outcome measure for this group. As a result, both Pulmozyme and TOBI were initially approved only for children ages 6 years and older. Newer techniques are being developed to measure lung function in infants and toddlers (Gappa et al., 2001; Castile et al., 2000). FEV1 exemplifies the challenges of developing a surrogate outcome measure for children, especially young children.
Assessment of outcomes related to physical or cognitive functioning may be more complicated in pediatric studies than in adult studies, given
normal developmental differences. The inability to walk or talk is characteristic of infants but is a serious problem for a 3-year-old. In evaluating protocols, institutional review boards (IRBs) will need to consider whether the proposed outcome measures are developmentally appropriate.
In addition to survival and functioning, outcomes related to the prevention of pain, nausea, vomiting, and other symptoms may be crucially important to children and families, depending on the child’s medical condition and its treatment. Among symptoms, instruments to assess pain are the furthest advanced. For infants and young children, symptom assessment strategies commonly used with adults, for example, numerical rating scales with 10 representing severe pain and 0 no pain, need to be replaced by other more developmentally appropriate methods. Such methods in use include face scales (representing a continuum for grimaces to big smiles) and observation of infant behaviors such as crying and body movement (see, e.g., Wong and Baker, 1988; Broome et al., 1990; Krechel and Bildner, 1995; Hockenberry-Eaton et al., 1999; Gallo, 2003; Manworren and Hynan, 2003; and Naar-King et al., 2004). If child-appropriate instruments do not exist for a symptom or group of children, the development and validation of such instruments can take years.
Quality-of-life measures have become important indicators of clinical benefit and therefore important in measuring the efficacy of new treatments. Again, instruments developed for adults are usually not suited for children. Instrument development for pediatric uses is complicated and slowed by the need to create forms appropriate for children at different stages of development. Work continues to develop, test, and refine generic (for all conditions) and disease-specific instruments for both healthy and ill children of different ages.7 For example, the PedsQL 4.0 instrument has both self-report forms and parent-proxy report forms for children ages 5 to 7, 8 to 12, and 13 to 18 years and a parent-proxy form only for children ages 2 to 4 years (Varni, 2003). This instrument has disease-specific modules for asthma, arthritis, cancer, cardiac disease, and diabetes. Other pediatric instruments are being developed or have been developed for some of the same conditions as well as for other conditions, including cystic fibrosis, cerebral palsy, atopic dermatitis, and obesity (see, e.g., Henry et al., 1997; Gee et al., 2000; Quittner et al., 2000; and Bullinger et al., 2002).
Development of new treatments for children requires an assessment of both the beneficial effects and the safety of the therapy. Critical to such an analysis are comparative data about what constitutes normal values for a wide range of physiological variables. Such comparative data must be age appropriate and, frequently, disease specific. Many laboratory norms are based on easily accessible data for healthy adults. For pediatric studies, efforts must be made to ensure that laboratories processing clinical samples have age-appropriate norms. The collection of data for the development of such norms is often difficult because routine laboratory studies (e.g., chemistry profiles) are rarely performed for healthy children. The lack of normative data becomes an even greater issue for children with rare conditions. For example, to assess the potential toxicity of new drugs for premature infants or children with AIDS, it is essential to have baseline “normal-for-the-population” laboratory parameters, such as white blood cell counts and liver function test results. These children may already have abnormal white blood cell counts, which makes it more difficult to monitor the effects of drugs that may have bone marrow suppression as a toxic side effect.
An advantage of disease-specific clinical trials cooperatives or networks is their greater ability to collect age- and disease-specific normative data across multiple trials. Baseline data collected before administration of a study drug(s) are most useful for this purpose. Investigators should be encouraged to publish these data, and journals should also be encouraged to accept such publications.
Administering Interventions and Measurements
When they evaluate a pediatric protocol for potential harms and benefits and for appropriate efforts to minimize risks to child participants, IRBs should consider the qualifications and expertise of investigators and others to conduct the research, taking into account the procedures and age groups described in the protocol. They should likewise assess the appropriateness of the facilities and settings for the proposed research.
Particularly in studies involving infants and young children, the administration of interventions and measurements may lead to complications not encountered with adults. Some complications are physiological. For example, it is more difficult to draw blood from the small veins of infants or toddlers than from those of older children and adults. Furthermore, the smaller volume of blood in children sets limits on how much blood can safely be drawn. Fortunately, newer analytical methods permit accurate assays using much smaller volumes of blood than was possible in the past.
Other challenges in administering pediatric studies are behavioral. Young children may not understand instructions. If they do, they may still be too immature to cooperate consistently, for example, by staying still when asked. Certain procedures that depend on verbal feedback from the research participant (e.g., more complex or complete assessments of pain, hearing, or other sensations or sensory functions) may not be possible with very young children. Older children and adolescents may present different challenges, for example, rebellion against adult authority, including that of their parents and the investigators. Peer pressure may also be a constraint for studies that require participants to be “different” (e.g., to take medications during school hours or to miss after-school activities to go to a clinic for an assessment). Depending on the study and the ages of the children involved, providing adequate time for training and preparation of children before the initiation of study protocols may be useful. Children may be more receptive to procedures requiring cooperation if they have a good understanding in advance of what is expected of them.
Whether undertaken for therapeutic or research purposes, tests and treatments for children often require or benefit from modifications in procedures, equipment, staffing, patient and family communication, and other dimensions of care that are tailored to their special developmental needs. Thus, it may be less stressful for all involved when persons drawing blood or spinal fluid from a child have been trained to work with infants and children and their families and when they have routine access to small-gauge needles, appropriate topical anesthetics, and even colorful bandages. Indwelling catheters may be used in pediatric studies to avoid multiple needle sticks. Children should always be given the option to receive a topical anesthetic to reduce needle-stick pain.
Laboratory personnel accustomed to analyzing blood and other biological samples from children may be more accepting of small samples than personnel who mainly work with adult specimens. Furthermore, depending on expectations about possible adverse events, the safety of research protocols may be increased by the availability of personnel and facilities prepared for pediatric emergencies.
Children with chronic disorders have frequently had previous experiences (both good and bad) with medical procedures. Careful discussion of the child’s perceptions and fears of any procedures before the initiation of research protocols may help correct misunderstandings and allay the child’s fears. The involvement of child-life specialists and child psychologists may reduce the stress on children and families during clinical studies as well as during usual clinical care.
Undertaking Research When Study Populations Are Small
Because most children today are healthy, children suffering from serious conditions such as cancer or heart disease are relatively few in number compared with the number of adults who have such conditions. As one indicator of children’s good health, children aged 0 to 19 years accounted for 29 percent of the total U.S. population of 272.7 million in 1999 but only 2 percent of all deaths—about 55,000 deaths for children compared with more than a half million deaths for adults aged 20 to 64 years and 1.8 million for those age 65 years and over (NCHS, 2000).
To cite another example, the American Cancer Society has estimated that approximately 1.3 million new cases of cancer would be diagnosed in 2003, of which an estimated 9,000 would involve children ages 0 to 14 years (ACS, 2003). For any specific cancer in children, the numbers are much smaller. For example, each year approximately 300 children are diagnosed with retinoblastoma (a tumor of the retina), 1,100 are diagnosed with astrocytoma (a brain tumor), and 2,400 are diagnosed with acute lymphoblastic leukemia (based on data from 1977 to 1995) (Ries et al, 1999). Moreover, as treatments for childhood cancer have improved and rates of cancer-free survival have increased, researchers find fewer children with relapsing disease who are potentially available for the study of new chemotherapeutic agents.
Exceptions exist to the “small-numbers” phenomenon. Each year approximately 75,000 babies are born very prematurely at 31 weeks gestation or earlier, and more than 350,000 babies have low or very low weights at birth (Martin et al., 2002). (For various reasons, however, many of these newborns will not be appropriate to include in research.) Asthma is a growing problem among children, and the increase in childhood obesity has become a major concern. A number of less serious conditions, such as otitis media, acne, and mild allergies, are quite common.
Although the numbers of children affected by some of these illnesses are greater than the numbers affected by rare genetic disorders, not all children with a disorder or condition will be eligible or able to participate in clinical trials. For any one research location, the numbers available for a study are usually quite low.
For many pediatric studies, the relative scarcity of potential study participants means that it often takes considerable effort and some creativity to enroll and retain sufficient numbers of children who meet the criteria for study participation. It may also mean that studies must extend for quite long periods just to secure enough participants. One recently published article on the prevention of fungal infections in children and adults with chronic granulomatous disease reported that it took 10 years to enroll 39 participants, most of whom were children at the time that the study started (Gallin et al, 2003).
Without a large enough number of participants, studies may not be able to generate statistically reliable estimates of differences between a study drug and a control treatment or placebo. A review of randomized controlled trials published in the Archives of Diseases in Childhood from 1982 to 1996 reported that about half of the studies recruited less than 40 children, with medians of 80 children for multicenter trials and 36 children for single-center studies (Campbell et al., 1998; see also Pattishall, 1990 and, generally, Freiman et al., 1986 and Moher et al., 1994). The authors comment that these small studies “have inadequate power to detect small or moderate treatment effects and result in a significant chance of reporting false-negative results” (Campbell et al., 1998, p. 196).
Children’s developmental differences create further complications that often require separate subanalyses or studies with infants, older children, and adolescents to assess the safety and efficacy of an intervention. The production of reliable estimates of effects for each subgroup usually increases the total number of research participants required for a study. In some cases, depending on the condition and the question being investigated and various technical considerations, special study designs can allow the use of smaller numbers of participants. For example, in so-called crossover studies, research participants act as their own control group, typically by first receiving and being evaluated on the experimental intervention and then receiving and being evaluated on a placebo or a standard treatment. In the study by Gallin and colleagues (2003) cited above, participants were randomly assigned to receive an experimental drug or a placebo at enrollment but were then switched annually to the alternative.
As discussed further in Chapter 4, one justification for the use of placebo-controlled trials is that they may allow smaller sample sizes than may be required for studies that compare the efficacy of two standard treatments or the efficacy of an experimental and a standard intervention. Such designs may raise ethical questions, and they require particularly careful design, monitoring, and analysis. Nonetheless, they have many advantages when their use is ethically and scientifically appropriate.
Large, multisite trials have become increasingly important for studies with adults as researchers seek to understand differences in medical conditions and treatment effects among population subgroups and to demonstrate reliably treatment effects that are modest but still important. Even more often than with adults, research involving children requires multisite trials and fairly long periods of participant enrollment to generate the minimum required numbers of study participants. The relative success of pediatric oncology researchers, as described below, is based in part on the large number of institutions participating in COG, which makes state-of-the-art care more accessible to children and families. In recent years other disease-specific clinical trials (Goss et al., 2002) are beginning to follow the COG model to increase participant enrollment and improve trial efficiency.
In reviewing all protocols but especially protocols involving children, IRBs should be particularly attentive to potential problems in achieving adequate numbers and classes of study participants and to the appropriate strategies for managing such problems. Such strategies may include modifications of classic clinical trial designs and innovative approaches to recruiting children and then retaining them throughout the course of the study. At the same time, reviewers of research protocol should be alert to the appropriateness of strategies for recruiting children, including payments to children or parents. This topic is discussed further in Chapter 6.
Designing Long-Term Studies
Importance of Long-Term Studies in Pediatric Populations
As mentioned earlier in this chapter, long-term studies are a particular challenge and a particular need for many serious pediatric medical conditions. Assessing the possible developmental effects of medical treatments or interventions may require extremely lengthy follow-up, well beyond what the immediate study outcomes appear to mandate. For example, the adverse sequelae associated with the use of cranial radiation to prevent central nervous system spread of leukemia in children did not become evident until many years following the introduction of this therapeutic approach (Cousens et al., 1988; Roman and Sperduto, 1995). Despite the success of radiation in the short term, the eventual recognition of late effects, including impaired intellectual function, profound neuroendocrine abnormalities, and second central nervous system malignancies, ultimately resulted in efforts to eliminate cranial radiation from among the options for the treatment of leukemia.
Many other conditions may require long-term follow-up in order to understand the physical, psychosocial, or economic consequences of the condition and its treatment. For example, very-low-birth-weight infants must be monitored at least until they reach school age to assess in detail the sequelae of prematurity and its treatment (Fazzi et al., 1997). Likewise, if one wants to test certain kinds of preventive interventions, then the study must follow study participants for at least as long as it is expected to take the target condition to develop naturally without treatment.
Studies of the effects of certain drugs in children may also require long-term follow-up that is not necessary for adults. Because of developmental changes in hepatic and renal function, children may experience positive or negative changes in response as they receive certain drugs over a long period. For children receiving drug therapy for a chronic condition, the need for long-term follow-up for both safety and efficacy is therefore obvious.
Implementing Long-Term Studies with Children
The performance of long-term studies with children involves a number of logistical and ethical challenges. The investigator, for example, must have a clinical trial infrastructure that permits tracking and periodic evaluation of trial participants over many years. It is obvious that families may move, but so may researchers. Within research institutions, where both physicians and other study staff tend to change positions over time, the study infrastructure must have an “institutional memory” to manage ongoing data collection and interaction with research participants despite staff turnover. In addition to being a logistical challenge, long-term studies are usually extremely expensive. Study sponsors are rarely willing to provide funding for long-term follow-up, and few institutions have the wherewithal to support such studies independently.
Long-term studies with children may also raise ethical issues that are not relevant for studies with adults. The most obvious of these involves informed consent. Although federal regulations provide for investigators to seek children’s assent when appropriate (as discussed in Chapter 5) and also allow their dissent under certain circumstances, it is almost always the parents who give legal permission for a child to participate in a study. For long-term studies, parents and children may be approached periodically for continued permission or assent, particularly if the nature of the research changes or when important developmental milestones are reached.
If a longitudinal study that started in childhood extends into adulthood, continuing participants will eventually become legally competent to consent to participation in their own right. For research that requires an individual’s continued contribution (e.g., through periodic interviews or procedures), these participants can either consent or decline to continue participation when they become adults. In general, continued research use of data already collected would be permitted for the purposes covered by earlier permission and assent. Questions have, however, been raised about when new permission or consent must be sought for new research uses of stored biological specimens and about what to do with such specimens and related analyses when an individual withdraws from a study (see, e.g., NBAC, 1999, Weir, 2000; Botkin, 2001).8 These are important questions with significant ethical and scientific implications, the analysis of which goes beyond the tasks for this study.
Working with Families
Family-centered care is increasingly being appreciated as one of the goals for the clinical services provided to children (see, e.g., Shelton et al., 1987; Johnson et al., 1992; Gerteis et al., 1993; MCHB, 2003; IOM, 2003b). It emphasizes care that respects and involves both the child and the family and that understands and accommodates their strengths, their coping strategies, and their cultural, religious, and other values. Family-centered care also promotes shared decision making.
Although family members are sometimes involved in discussions about the participation of adults in research, parents or other legally designated persons are almost always involved in discussions and decisions about children’s participation in research. Discussions and decisions may also involve grandparents and other family members. As explained in Chapter 3, if research involves more than minimal risk but is not expected to benefit a child participant, federal rules usually require that both parents give permission for participation. In addition, federal rules require that, when appropriate, children must “assent” to participation in research.
The multiplication of participants in discussions and decisions about a child’s involvement in research can place considerable demands on the interpersonal skills of the investigator and others involved in discussions about research participation. It may also increase the time needed for explanations and decision making and the potential for disagreement among those involved. When research involves acutely ill or injured children, investigators will usually face parents who are confronting complex and difficult decisions while under extraordinary stress. Chapter 5 examines some of the complexities of establishing the conditions for informed family decision making about children’s participation in research.
The diversity of family contexts also creates challenges for investigators. Family decision making about health care in general and about research specifically is not done in a vacuum. Rather, it is affected by social and cultural factors. These factors have an important influence on families’ perceptions of health, illness, and treatment and on their views of the role of patients and family members in decision making. Thus, in discussions about children’s participation in research, investigators are also called upon to be culturally sensitive and competent and to negotiate between families’ beliefs and the tenets of biomedicine.
Some research involving children raises highly sensitive issues for their families and, perhaps, their communities. Studies of such important issues as suicide, drug use, sexual behavior, family dysfunction, and other topics are sensitive enough when they involve adults. When studies on these topics involve children, the challenges for investigators grow. Parents may be reluctant to permit a child’s participation in sensitive research. As discussed
in Chapter 5, IRBs may approve federal requirements for waiver of parent permission in certain situations, and this committee encourages IRBs to consider carefully protocols that propose such waivers. In general, however, parental involvement and permission are advisable.
Meeting Ethical and Regulatory Standards for Pediatric Research
As discussed in Chapters 1 and 3, recent decades have seen the development and refinement of ethical principles to guide research with vulnerable populations, including children. These principles have been translated into government regulation, which, in turn, have directed the development of institutional structures and procedures to implement the principles and regulations.
The regulations restrict the range of research that can be undertaken with children, particularly studies that involve more than minimal risks for healthy children or children who have no prospect of directly benefiting from participation in the research. For example, traditional phase 1 clinical trials that use healthy volunteers to test the safety of a new medication face higher hurdles to approval if investigators propose to include children. Placebo-controlled trials of a drug’s efficacy likewise face more scrutiny when children are involved if the trial does not hold the prospect of benefiting the children who receive the placebo but does expose them to more than a minimal amount of risk. Chapter 4 discusses in detail the criteria for approving research that includes children.
Notwithstanding the ethical rationale for the special regulations that protect child participants in research, the regulations add to the administrative burdens on investigators in preparing, justifying, and implementing research protocols. Investigators may find themselves confronting unexpected questions and criticisms from IRBs, changing procedures for obtaining parents’ and children’s agreement to participate in research, redesigning protocols, and delaying the enrollment of participants. When investigators have not been adequately educated about the special regulatory protections for children, they may be at a particular disadvantage in this process.
The extended time required to gain IRB approval for pediatric studies, even though it is necessary for participant safety, may frustrate both the investigators and the families of children with life-threatening disorders. In multicenter trials, some study sites are never able to enroll participants because the site does not receive IRB approval until after study enrollment is already completed at other sites. This not infrequent scenario leads to significant disappointment among researchers and potential research participants at the site. Chapter 8 discusses further these and other issues with multicenter trials.
Training and Retention of Clinical Investigators
Not only are child participants in short supply for much pediatric research, but so are properly trained and experienced pediatric investigators. Pediatricians who pursue careers in clinical research must receive specialized, post-residency training. Until recently, most subspecialty pediatric fellowship training was directed toward laboratory science rather than human-based research. To conduct the kinds of pediatric research described in this report, it is important to attract recently trained pediatricians into fellowship programs that are oriented toward clinical research and that have strong curricula that cover study design, biostatistics, the ethical conduct of trials, and similar essentials of sound clinical investigation. To ensure that new and established investigators alike are sufficiently knowledgeable about the protection of human research participants, NIH requires education on this topic for all investigators submitting new NIH grant or contract applications or receiving non-competing awards for research involving human participants (NIH, 2001).
NIH offers institutional grants, such as the K-30 clinical research curriculum awards, that are specifically intended to stimulate academic institutions to expand or improve the training of clinical investigators. Recently, NIH announced a new career development program aimed specifically at multidisciplinary clinical research (NIH, 2003b). As examples of the core components of a multidisciplinary curriculum, the program announcement listed clinical research methodology, epidemiology, biostatistics, informatics, ethical issues, safety of research participants, regulatory requirements, team leadership and management, grant writing, and interactions with industry.
In 1996, the Federation of Pediatric Organizations revised and reaffirmed its 1990 Statement on Pediatric Fellowship Training (FOPO, 1991), which is once again being revised. The statement stresses that fellowship training should prepare trainees to be competent in clinical care, education, and research. Such training should occur at sites that have sufficient faculty who are committed to scholarship and research excellence and who, collectively, have appropriate expertise in hypothesis-driven investigations. In addition to providing direct research experience, programs may also provide that trainees serve in some capacity on or with an IRB.
Once clinical faculty are hired, their retention depends, in part, on adequate mentoring by successful clinical investigators, sufficient protected time to conduct research, and the rapid critique of grant proposals by faculty members, including the chair and the mentor(s). Clinical investigators should be more successful if they work in an environment with a supportive research infrastructure that includes appropriate staff (including research nurses and grant administrators), adequate access to computers
and data sets (if needed), and capable administration of institutional and government policies on the ethical conduct of human research.
DATA ON THE EXTENT OF CHILDREN’S PARTICIPATION IN CLINICAL RESEARCH
Data on the extent of children’s participation in research are limited for most conditions. Most data located by the committee involve children with cancer. Cancer and heart disease are the leading disease-related causes of death for children, but they account for a small percentage of all childhood deaths. About half of childhood deaths occur in infancy (primarily as a result of congenital anomalies, short gestation, or complications related to pregnancy), and almost a third are the result of intentional or unintentional injuries (NCHS, 2000). As discussed elsewhere in this report, clinical trials of emergency care with both adult and pediatric populations face particular difficulties related to informed consent.
The National Cancer Institute (NCI) reports that in 1998 and 1999 some 50 percent of child cancer patients (ages 0 to 14 years) participated in clinical trials undertaken by NCI cooperative groups compared to 20 percent of patients ages 15 to 19 years and less than 3 percent of patients age 20 years or over (NCI, 2002b; data from Sateren et al., 2002; see also, Shochat et al., 2001). An earlier study based on data obtained from 1991 to 1994 reported that 70 percent of child cancer patients (ages 0 to 19 years) were enrolled in trials (Tejeda et al., 1996). A recent study calculated that about 70 percent of child cancer patients under age 15 years were registered by pediatric oncology trial groups compared to only 24 percent of patients ages 15 to 19 years (Liu et al., 2003).
The committee found no comparable data on the percentages of children with other conditions enrolled in clinical trials. A simple search of the NIH clinical trials website (http://www.clinicaltrials.gov/) generated a list of 163 leukemia studies for children from birth to age 17 years, 14 studies for cystic fibrosis, 12 studies for prematurity, and 29 studies for diabetes (all types). The search also yielded 146 studies for the terms infant and infancy, 238 for the term neonatal, and 159 for the terms adolescent and adolescence. The list of trials includes studies that do not test therapies. Closer examination of the listed studies suggests that some do not actually involve children. The number of studies is not a direct indicator of the total enrollment in clinical trials because trials vary greatly in size.
Some studies suggest that recruitment of children from minority populations may be particularly difficult. For example, in a longitudinal study of the risk factors for the development of cardiovascular disease during childhood, the recruitment of sufficient numbers of minority children took 2
years whereas researchers needed only 1 year to recruit the target number of nonminority children (Grunbaum et al., 1996). An analysis of enrollment in pediatric cancer treatment trials suggested, however, that minority children were proportionately represented (Bleyer et al., 1997). Some other studies report similar results (NINR, 1993; Rosella, 1994; Villarruel, 1999), whereas some show underrepresentation (Bonner and Miles, 1997; Peterson and Sterling, 1999). A recent report from the General Accounting Office suggested that FDA needed to improve its monitoring of the inclusion of minority children in research regulated by the agency (GAO, 2003).
The high referral and enrollment rates for child cancer patients in clinical trials likely reflects a number of factors, including the imminent threat posed by many types of cancer and the success of researchers in increasing survival rates for important childhood cancers, such as leukemia. Other factors appear to include the large number of institutions participating in cooperative trials, the relatively large number of therapies being tested, and the concentration of cancer treatment for children in institutions participating in cooperative trials. Care at these institutions is widely regarded by pediatricians and families as “state of the art.” The committee found no other serious condition with a clinical trials cooperative group as large as COG, which includes nearly 240 member institutions in the United States, Canada, and other countries. According to its website, COG typically has approximately 100 trials open to enrollment at any time, with approximately 5,000 participants enrolled and some 35,000 being actively monitored (COG, no date).
In contrast to the number of institutions participating in COG, the National Institute of Child Health and Human Development funds 13 centers in the Pediatric Pharmacology Research Network (PPRU, no date) as well as 16 centers in the Neonatal Research Network (NICHD, 2003). As noted above, prematurity, problems associated with childbirth, and low birth weight are the leading causes of death among children.
The Cystic Fibrosis Foundation supports various research programs, including 14 centers in the Therapeutics Development Network. The National Institute of Allergy and Infectious Diseases (NIAID) and the National Institute for Child Health and Human Development (NICHD) support 18 centers of the Pediatric AIDS Clinical Trials Group (plus a data management center and a coordinating operations center) (NICHD, no date). A centralized national registry of clinical trials is clearly needed to better understand the number of children (particularly healthy children) participating in clinical research and to promote more cooperative clinical trial groups. An earlier IOM report (2003a) recommended the creation of such a central registry of all clinical trials.
INITIATIVES TO INCREASE INVOLVEMENT OF CHILDREN IN RESEARCH
Given the challenges described above, many investigators and sponsors of research have been reluctant to undertake research with children. Responding to concerns from pediatricians and family advocacy groups, policymakers have attempted, particularly in the last decade, to stimulate research that involves infants, children, and adolescents.
Congress has long provided funding and directives to NIH and other agencies to support research on a variety of child health problems. In 1963, it established NICHD as part of NIH. This NIH unit supports a range of biomedical, epidemiologic, and other research with the goal of ensuring “that every person is born healthy and wanted, that women suffer no harmful effects from the reproductive process, and that all children have the chance to fulfill their potential for a healthy and productive life, free of disease or disability” (NICHD, 2002, unpaged). Other government agencies also focus on children’s health problems, for example, the venerable Maternal and Child Health Bureau, first created early in the last century.
Legislators have also attempted to influence the behavior of pharmaceutical companies. In 1997, the U.S. Congress reinforced initiatives already underway at FDA by including incentives in the Food and Drug Administration Modernization Act of 1997 (P.L. 105-115) for these companies to conduct pediatric studies. For drugs for which FDA had requested pediatric studies, the legislation provided that a company could obtain 6 additional months of patent protection for a drug by conducting studies to provide dosing and safety information for children.
Roberts and colleagues (2003, p. 906) reported that “[b]etween July 1998 and April 2002, 53 drugs were granted pediatric exclusivity and 33 drug products have new labels with pediatric information.” For seven of the drugs, pediatric studies led to “major adjustments in the dosing instructions” (p. 910). To cite an example, the new information showed that doses for midazolam hydrochloride (Versed, a sedative widely used for surgical procedures) should start at lower initial levels for children with congenital heart disease and pulmonary hypertension. Between April and January 2004, pediatric use labeling became available for another 33 drugs (FDA, 2004).
The Best Pharmaceuticals for Children Act of 2002 (P.L. 107-109) renewed the exclusivity provisions. It also called for NIH to sponsor pediatric tests of certain drugs already approved but not tested or not fully tested for their effects with children. This list, published in January 2003, identi-
fied the dozen highest-priority drugs needing pediatric testing (DHHS, 2003a). Most of the drugs are no longer under patent, and therefore, the “pediatric exclusivity” incentives are largely irrelevant. The secretary of DHHS announced that $25 million in federal funds would be allocated to support research on these drugs in 2003.
Recently, Congress passed the Pediatric Research Equity Act of 2003, (P.L. 108-155). It gives FDA the authority to require pediatric studies of certain drugs and biological products. In the future, companies submitting requests for approval to market a new drug or biologic (or a new indication, formulation, dosing regimen, or route of administration for an already approved product) will be required to submit information about the safety and effectiveness of the product in relevant pediatric populations. Testing may not be required if the agency determines that it is appropriate to extrapolate data from studies with adults (usually with some supplementary pediatric pharmacokinetic or other study data). Submission of pediatric data may be deferred under certain conditions (e.g., the adult but not the pediatric studies are completed). The requirements may also be fully or partially waived under several conditions (e.g., pediatric studies are impossible or highly impractical or existing evidence suggests that the drug would be unsafe or ineffective for children). If a waiver were granted on the basis of evidence that a drug would be unsafe or ineffective, the drug label would be required to include information to that effect.
Other agencies that are not part of FDA or NIH may also have been influenced by these initiatives. For example, in its program of Centers for Education and Research on Therapeutics (CERTs), which supports research and education on effective therapeutics (i.e., drugs, medical devices, and biologics), the Agency for Health Research and Quality has designated one center to focus specifically on medical therapies for children (UNC, 2003). The CERTs program was authorized in P.L. 105-115.
Food and Drug Administration Rules and Policies
Even though the best and brightest pediatric minds have helped us establish dosages for children, we’re finding out that the dose is different than we thought in some cases. And that probably came as a surprise to most of us.
Richard Gorman, quoted by Meadows, 2003, unpaged
In 1979, FDA issued regulations on the content and format of labels for human prescription drugs (FDA, 1979a). The regulations stated that if drug companies made statements about the pediatric use of a drug for an approved indication, the statement had to be based on substantial evidence from adequate and well-controlled studies, unless FDA waived that require-
ment. As discussed in Chapter 1, from the early 1970s into the 1990s, the proportion of drugs with specific pediatric dosing information stayed at a low level (about 20 percent of the drugs listed in the Physician’s Desk Reference).
In 1994, FDA required drug manufacturers to determine whether they had data sufficient to support labeling information on pediatric use. If they did, they were then to request permission from FDA to make changes in their labels based on that data (FDA, 1994a). The rules also provided that clinical trials with children might not be required to support labeling if sufficient evidence existed that the disease and the drug’s effects allowed extrapolation from the results of trials with adults.
Four years later, in issuing new regulations, FDA observed that “[t]he response to the 1994 rule has not substantially addressed the lack of adequate pediatric use information for marketed drugs and biological products” (FDA, 1998e, p. 66632). Under the new regulations, FDA could, in some cases, require drug companies to conduct studies of new and existing drugs to determine their safety and efficacy in children. In October 2002, a federal court struck down this so-called pediatric rule, holding that FDA exceeded its statutory authority (Albert, 2002; ruling at AAPS v. FDA, 2002). In response, Congress passed the Pediatric Research Equity Act of 2003, as described earlier.
National Institutes of Health Policies
NIH has encouraged pediatric research in ways both general and specific. General strategies include support for the clinical trial cooperative groups that are intended to facilitate high-quality, multisite adult and pediatric studies and, as one consequence, reduce the number of trials with undersized study populations (i.e., “underpowered” studies in the language of statistics and research methods).
In 1998, following directives in House and Senate Appropriations Committee reports for fiscal year 1996, NIH issued specific policies and guidelines for including children as research participants (NIH, 1998a). NIH analyses suggested that 10 to 20 percent of NIH-supported studies inappropriately excluded children.
The 1998 policies focus on disorders and conditions that affect adults and that may also affect children. Children are to be included in research conducted or funded by the agency unless their exclusion is justified on scientific or ethical grounds. For some medical conditions, children’s developmental characteristics might suggest the need for a separate, child-only study. Proposals to NIH for funding must include a section on the participation of children that discusses the rationale for excluding children from a study or that provides a plan for their inclusion. Such plans must describe
the relevant expertise of investigators and the appropriateness of the facilities to be used. The plans are then reviewed as part of the NIH peer review process for proposals. The 1998 NIH guidelines also describe the federal rules for protecting children in research.
Congress encouraged, but did not require, NIH to establish pediatric research priorities (U.S. Congress, 1995; NIH, 1998a). Although NIH does not appear to have developed an overall set of priorities, some individual institutes have identified priorities for certain clinical problems or services, including kidney disease (NIDDK, 2001), HIV/AIDS (NIH, 2003b), and emergency medical services (NIH, 2003b).
Recent years have seen a number of actions to encourage research involving children and help investigators cope with the many methodological, practical, and ethical challenges of pediatric studies. One apparent result is an increase in the number of children participating in research. Another is an increase in the number of priority drugs that have labeling information for at least some pediatric age groups.
More can be done. Chapter 8 includes a recommendation for strengthening educational programs to develop pediatric investigators who are prepared to design and conduct valid, ethical clinical research. In addition, to ensure the continuation of well-designed and well-conducted pediatric research that will improve children’s health, federal policymakers should sustain and extend other aspects of the critical financial and infrastructure for this research. They should likewise support important research that is often not attractive to commercial sponsors, including long-term studies and projects to improve outcome measures relevant to infants, children, and adolescents. Because most pediatric conditions are sufficiently uncommon that statistically sound, ethical research requires multiple study sites, the federal government should continue to establish and fund discipline-specific, age-relevant research groups or consortia with the expertise and administrative infrastructure to conduct multicenter studies.
Furthermore, as research continues to become more international, it is important for governments, investigators, industry, and international organizations to cooperate to support the conduct of ethical multinational pediatric studies, increase the pace of therapeutic development for rare pediatric conditions, and move toward greater consistency in the regulatory protections for child participants in research. The next chapter, which summarizes the regulatory framework for protecting human research participants in the United States, also briefly reviews international guidelines and efforts to develop consistent or uniform—that is, harmonize—regulations across countries.