Extrapolation of Safety and Efficacy Data to Children
Extrapolation to children of safety and efficacy data generated for adults requires careful attention to potentially important differences between these two populations. The safety of drugs, for instance, needs to be supported by appropriate research with the targeted age group. Some medications that are completely safe for adults may produce toxic effects in children. An assessment of pharmacokinetic-pharmacodynamic relationships, however, by use of a surrogate and comparison of those results with those for adults, may suffice as a basis for approval of the drug for use in the pediatric population or help determine the doses to be used in clinical trial.
Unlike drugs, the majority of which are for oral administration, the majority of therapeutic biologics are for parenteral use only. Thus, many of the formulation issues for biologics to be used in the pediatric population are similar to the formulation issues for parenteral drugs to be used in the adult population. Because of their distinctive properties, the use of biologics results in unique safety concerns that require different types of monitoring, such as for adventitious agents that occur as a result of treatment with the biologic, or for reactogenicity. There is some concern about therapeutic biologics because little is known about the long-term effects of treatment with such agents, especially their effects on developing children. Many questions about the evaluation of biologics in the pediatric population need to be addressed; no simple approach is available.
The requirements for the safety and efficacy of medical devices are different from those for the safety and efficacy of drugs and biologics. Not only can
the size of a device pose engineering problems, but hormonal influences and the activity levels of patients also need to be factored into the design of devices. Obtaining approval for use of a device in the pediatric population is also not an easy undertaking; ample evidence indicating that it has been properly designed for children and that the safety and efficacy are demonstrated rather than presumed must be made available.
Special considerations must be exercised when extrapolating safety and efficacy data in adults to children in the areas of drugs, biologics, and medical devices. The summaries of the presentations that follow explore such issues, including when and how safety and efficacy data can be used for children.
SPECIAL CONSIDERATIONS FOR EVALUATING MEDICAL DEVICES IN INFANTS AND CHILDREN
Presented by David W. Fugal, Jr., M.D., M.P.H.
Director, Center for Devices and Radiological Health, U.S. Food and Drug Administration
Nearly 8,000 manufacturers have medical devices registered with the U.S. Food and Drug Administration (FDA). Many of these are specifically designed for children. Examples include heart valves, fetal bladder stents, medical equipment for the neonate, and hydrocephalus shunts. There are also life-supporting and life-sustaining products such as ventilators, rate-responsive pacemakers, and hemodiaylsis machines. These devices often cannot just be devices adapted from adult use. Even something as simple as the change needed in the size of a catheter lumen changes the flow characteristics and may critically affect product performance. Children grow and implants such as heart valves and hip prostheses need to be designed to accommodate growth or be designed for replacement. Changes in the type and level of activity occur rapidly from infancy to adulthood; some devices need to be developed with consideration given to the hormonal and body size changes that rapidly occur at adolescence.
To gain approval for a device for children, there must be evidence that it is properly designed for children, as well as a demonstration of safety and effectiveness. In contrast to drugs, for which formulation changes and improved palatability might be sufficient to develop a pediatric formulation, developers of pediatric devices must be focused on the performance and controls needed to adapt a device for a pediatric patient. Recently, the FDA centers responsible for drugs and biologics reviewed their approved products to identify those with inadequate clinical information to support pediatric labeling and published a list calling for pediatric studies. Contemplating a similar task for devices is daunting. Even aside from the fact the nearly half of all newly marketed devices are exempt from premarket applications, 4,500 new devices are approved
or cleared for marketing in the United States each year. A device-by-device approach is not likely to work; more promising are the incentives to develop new devices for children.
The Center for Devices and Radiological Health (CDRH) gives priority to applications for novel products and products which address unmet medical needs and special populations. CDRH encourages consultation on device trials protocols in children under investigational device exemptions (IDEs—the experimental device application comparable to the IND for drugs and biologics). One of the most important issues to discuss is the evidence that needs to be developed for marketing clearance or approval.
The evidence needed to establish the safety and effectiveness of devices is somewhat different from the evidence needed to establish the safety and effectiveness of drugs and biologics. Device evidence is graded appropriate to the level of risk, ranging from nonclinical product performance standards to rigorous clinical trials. Although there is more flexibility in the evidence needed for approval, there is more room for disagreement. A provision in the Food and Drug Administration Modernization Act (PL 105–115) establishes procedures to resolve disputes and provisions for meetings with FDA officials to arrive at agreed upon criteria, evidence and study end points that in the absence of major scientific developments will be adequate to establish the safety and efficacy of the product.
Another mechanism to meet special needs is the humanitarian device exemption (HDE). This mechanism requires that fewer than 4,000 patients could use the device each year. No evidence of device effectiveness is required for an HDE, although informed consent and Institutional Review Board (IRB) oversight is necessary. Unlike orphan drug approval, HDE approval does not confer marketing exclusivity, and, in general, the HDE is more analogous to a drug treatment-IND-with-cost recovery than an orphan drug approval.
Finally, CDRH has a radiological health program that affects children. The standards set for radiation-emitting products range from clinical to nonmedical consumer devices. One of the programs for assessing medical x-ray exposure is the Nationwide Evaluation of X-ray Trends (NEXT) program, which sets standards and collects data from a testing program that recently measured the exposure to children from pediatric chest x-rays.
In summary, while developing devices for children presents many of the same challenges as developing drugs and biologics for children, and the tools differ, the need and potential remain great.
DEFINING SURROGATE ENDPOINTS AND BIOMARKERS FOR DRUG ACTION IN TRIALS WITH PEDIATRIC SUBJECTS
Presented by Robert Temple, M.D.
Associate Director for Medical Policy, Center for Drug Evaluation and Research, U.S. Food and Drug Administration
It is not clear whether the use of biomarkers and surrogates in the pediatric population poses problems different from the potential problems resulting from their use in adults. The U.S. Congress, in writing FDAMA, specifically incorporated into the law a provision that FDA had introduced in 1992, called accelerated approval, which allowed the use of surrogate markers that were less than well established but that nonetheless had a reasonable likelihood of predicting clinical benefit. FDA defines a surrogate endpoint as "a laboratory measurement or physical sign that is used in therapeutic trials as a substitute for a clinically meaningful endpoint that is for a direct measure of how a patient feels, functions or survives and is expected to predict the effect of the therapy" (57 Federal Register 13234–13242, 1992).
There is a tension between the advantages and disadvantages of the use of surrogates. As a general matter, reliance on surrogates can be faster, cheaper, and more efficient than use of clinical endpoints, but surrogates do not measure the endpoint of real interest. An additional drawback is that reliance on a surrogate results in a much smaller amount of controlled safety data than would be obtained from a trial with a clinically relevant endpoint. In addition, many risks do not manifest their effects immediately in children, for whom the problems raised by abnormalities in childhood might not show up until adulthood; short-term trials with surrogates give no opportunity to discover such effects.
This idea of relying on a surrogate endpoint is not new, but past practices have sometimes been reconsidered. In the past, FDA approved drugs that lowered ventricular premature beat (VPB) rates and were believed to lead to symptomatic improvement. But concerns arose over the proarrhythmic effects of these agents, and, notably, after a heart attack antiarrhythmic drugs were found to be lethal in some settings; thus these drugs are not being approved even for symptomatic treatments without evidence that they do not decrease survival. The FDA also approves drugs that lower blood pressure or lipids without pre-marketing evidence of other effects. FDA sometimes approves drugs because they shrink tumors, especially in patients with refractory illness, even though the consequences for the patient's health or survival are unproven.
Even when FDA bases initial approval on surrogate effects, subsequent outcome studies are of interest and importance. Moreover, the labeling for the drug reflects what is known. For example, if FDA approves a drug that lowers cho-
lesterol levels it is approved for lowering cholesterol levels, not for improving survival. If a company wants to have the drug approved for improving survival, it must prove that it does. Thus, FDA usually distinguishes between approval of a drug on the basis of a surrogate effects and giving it a further claim on the basis of a surrogate effects.
The accelerated approval rule promulgated in 1992 and incorporated into law in 1997 (U.S. Food and Drug Administration Modernization Act §112, 1997) is aimed at drugs that treat serious illnesses for which no good therapy is available; in such cases FDA is prepared to rely on less well-established surrogates. There is currently no regulatory definition of a "well-established surrogate." The definition of "the strength of evidence" needed for a surrogate to support accelerated approval is that the surrogate must be "reasonably likely, based on epidemiologic, pathophysiologic, and other kinds of evidence, to predict outcome" (57 Federal Register 58942–58960, 1992). Presumably, a well-established surrogate has stronger support than is sufficient to be "reasonably likely" to predict benefits. Furthermore, it should be appreciated that in clinical trials there are endpoints that are not surrogates because they represent a real clinical benefit, but they are also not the final goal of therapy. These have been called intermediate endpoints. For example, increased exercise tolerance in patients with heart failure or decreased symptoms of hyperglycemia in patients with diabetes are valuable clinical benefits, but in both groups of patients a decrease in the rate of morbidity is hoped for as the final endpoint.
The main problem with surrogate endpoints is that they may not predict the desired outcomes. There are two possible reasons for this. First, the relationship between the surrogate and clinical endpoint may not be the causal relationship that is presumed. For example, improved bone mineral density does not always lead to a lower fracture rate. Second, drugs have more than one effect, and the surrogate endpoint measures only the one that is thought to be desirable. The drug could have other effects that are adverse. An outcome trial gives the "sum" of all these effects, but a trial with a surrogate endpoint does not. Examples include diuretics, which lower blood pressure but which also lower potassium levels, probably leading to arrhythmics, and antiarrhythmics, which decrease ventricular heartbeat rates but which probably increase the incidence of ventricular tachycardia and, perhaps, asystole. The net effect (beneficial or adverse) may depend on the populations and the duration of treatment. There may be some cumulative adverse effects that will not be observed in a short period of time.
In a number of situations it would be desirable to know the long-term effects of treatments, both in children and in adults. This is the case for such conditions as chronic lung disease, asthma, various inflammatory diseases, attention deficit disorder, and depression and psychosis. In many of these cases, the drugs that are being used have significant cardiovascular and other effects.
With these concerns in mind, there are certain situations that tend to support reliance on a surrogate endpoint, including consistent epidemiology, existence
of appropriate animal models, and adequate understanding of disease pathogenesis. There are also practical reasons for relying on surrogate endpoints, such as a lack of existing therapy for a serious disease or the difficulty in conducting an outcome study, particularly when the effect would be very delayed or when the event rate is low.
Even if one can conclude that a drug is likely to behave in children like it does in adults, there could still be a concern as to whether the concentration-response relationship is the same. If the drug is not toxic and the drug is generally given on the plateau part of the dose-response curve, dosing can initially rely on the adult dose-response curve, modified for size, with a requirement to collect resulting safety information. If a drug is potentially toxic, an effectiveness trial is needed to establish an appropriate dose. Assessment of pharmacokinetic and pharmacodynamic relationships by use of a surrogate and then comparison of those with the relationships for adults might suffice as a basis for approval or might at least help determine the doses to be used in a clinical trial.
PREDICTION OF LONG-TERM EFFECTS ON POSTNATAL BRAIN DEVELOPMENT
Presented by Mark Batshaw, M.D.
Chair, Department of Pediatrics, George Washington University Medical Center, and Director, Children's Research Institute, Children's National Medical Center
Brain development occurs rapidly from early embryogenesis to beyond birth. Some development is completed before birth, such as proliferation, neurogenesis, migration, and some synaptic generation. Other aspects occur after birth; it is these areas that can be affected by drugs given to children.
The organization of the brain starts at about the fifth month of gestation and progresses through about the sixth year of childhood. This involves the outgrowth of neurons and of the dendritic and axonal spines, the development of synapses, and the selective elimination of neural processes, as well as the proliferation of the glial network. In early childhood, when synapses are forming, the use of drugs may have very subtle effects.
Commonly occurring disorders involving neuropsychologic and neurophysiologic abnormalities can serve as surrogates for what could happen to a typically developing child who is exposed to drugs. For example, disorders of myelination, which starts at birth and continues for the first 2 years of life, can be studied to determine the possible effects of certain drugs on myelination. In addition, animal models, such as the trisomy 16 mouse, or the "quaking mouse," can be used as models of abnormal neuronal organization and myelination.
Neuropsychologic measures can be used to study the subtle abnormalities that drugs may produce. Such measures might be neuroanatomical, such as neuroimaging by magnetic resonance imaging (MRI) or computed tomography (CT). Some are neurophysiological, such as positron emission tomography (PET), single-photon emission computed tomography (SPECT), and functional MRI. Cognitive testing and neurobehavioral measures can also be used, although caution is in order with full-scale intelligence quotation (IQ) scores because they can mask subtle neuropsychologic patterns of abnormality.
Functional neuroimaging can also be used. This has recently been done with learning disability, especially dyslexia, by using PET and functional MRI scanning. In this case, glucose utilization or blood flow in various brain regions can be observed. Studies have shown decreased activation, that is, less glucose utilization or blood flow, in the left temporal parietal cortex, and the superior temporal cortex during rhyming-directed phonologic awareness. In dyslexia, persons have difficulty understanding phonemes, an understanding that occurs primarily in this temporal parietal cortex region. When one performs functional neuroimaging with dyslexic children, the occipital region lights up normally, but the temporal parietal region does not. One might be able to, in a similar way, look at functional neuroimaging in children before and after they receive a drug that could have adverse effects.
In terms of IQ testing and neuropsychology, it is difficult to predict long-term outcome in younger children. One of the main reasons for this is that the best predictor of intelligence is language. In the first 2 years of life, language is not well developed. The most commonly used test for children under 2 years of age is the Bailey Scales of Infant Development II, which provides global developmental status. The Wechsler intelligence scales are the most commonly used intelligence tests for children ages 3 to 16 years. These tests have the ability to assess verbal and nonverbal skills, working memory, and processing skills. These tests also provide a way of looking for abnormalities in cognition caused by drugs.
Some neuropsychologic tests can help pinpoint abnormalities in a specific area of the brain. For example, the Stroop test looks for cognitive flexibility. Slow performance is associated with left hemisphere dysfunction and traumatic brain injury. The Wisconsin card sort test assesses frontal and prefrontal functioning. These tests reveal much more subtle abnormalities than would be found by simply performing IQ studies. Investigators developing therapies such as radiation to the brain or drugs that are going to affect biogenic amines or excitotoxins (e.g., glutamate and glycine) should consider the value of these tests for determination of long-term adverse effects. As an example, animal models could provide an indication that a particular drug may affect the cerebellum. Tests that assess cerebellar function in humans might be in order.
Adaptive skill assessment, that is, the flexibility of the person in terms of performing daily living skills, can also be done to determine if a medication is
having an effect on the ability of the patient to care for himself and herself. These tests identify a pattern of functioning different from the neuropsychological pattern. In addition, a number of child behavior checklists look at behavioral functioning and child competency.
EVALUATING BIOLOGIC THERAPEUTICS IN PEDIATRIC CLINICAL TRIALS
Presented by Karen Weiss, M.D.
Director, Division of Clinical Trial Design and Analysis, Center for Biologics Evaluation and Research, U.S. Food and Drug Administration
The definition of a biologic, as set forth in the Public Health Service Act, is "any virus, serum, therapeutic serum, toxin, antitoxin, vaccine, blood, blood component or derivative, allergenic product or analogous product, or arsphenamine or its derivative, or any other trivalent organic arsenic compound applicable to the prevention, treatment or cure of diseases or injuries in man."* That definition does not fully express the diversity of therapeutic products currently regulated by the Center for Biologics Evaluation and Research (CBER). Examples of therapeutic biologics will be discussed below, as will some differences between drugs and biologics. It is worth noting that Section III of FDAMA addresses exclusivity as an incentive for conducting needed pediatric studies. However, that exclusivity applies only to products approved under Section 505 of the Food, Drugs, and Cosmetics (FD&C) Act. Since most biologics are licensed under the Public Health Service Act, they are excluded from the exclusivity provisions of FDAMA.
Some of the more "traditional" therapeutic biologics fall into the broad category of cytokines, including interferons and interleukins and the category of hematopoietic growth factors, including erythropoietins and the colony-stimulating factors. Another class of therapeutic biologics is the monoclonal antibodies. The first licensed monoclonal antibody was Orthoclone (OKT3), for the treatment of renal allograft rejection. More recently, CBER has licensed monoclonal antibodies for the prevention of renal allograft rejection, for the treatment of Crohn's disease, and for the treatment of certain cancers.
CBER also regulates cell-based therapies when the cells are manipulated (e.g., selected or expanded). One example of such a cell-based therapy is Carticel, which is autologous cartilage derived from an auricular biopsy specimen, expanded in culture and placed into defects of the femoral condyle. CBER also regulates cell-based therapeutics when cells are from xenogeneic sources (e.g., mesencephalic porcine cells under investigation for the treatment of Parkinson's
disease). A relatively new field of clinical investigation is gene therapy. Examples of investigational gene therapies include an adenoviral vector containing the cftr (cystic fibrosis transmembrane regulator) gene for cystic fibrosis and plasmids containing fibroblast growth factor for the treatment of coronary artery disease. CBER also regulates organ transplantation when the organs come from xenogeneic sources and combination products when the primary mode of action is the cellular component (e.g., hepatocytes housed in filters for the treatment of hepatic failure and pancreatic islet cells in alginate capsules for the treatment of diabetes mellitus).
Biologics differ from drugs in important ways. One is that the starting or source material for many biologics is cellular or cell substrate in origin. Such products have the potential to transmit infectious agents. The manufacturing procedures must be sufficient to remove or inactivate infectious agents. Biologic licensing regulations call for strict controls over the entire manufacturing process to ensure, to the extent possible, that the biologic product is sterile, pure, and potent.
Another difference between biologics and drugs is that many therapeutic biologics are large proteins. Administration of foreign protein, particularly chronically, may result in an immune reaction. Although a serum immunologic response (i.e., a protective antibody) is the basis for the effectiveness of vaccines, the development of an immune response after exposure to a therapeutic biologic is not desirable. An immune response may simply manifest as an elevated serum antibody titer that has no clinical sequelae. At the other extreme it may result in unwanted effects such as altered pharmacokinetics, which could render a product ineffective, or it may manifest as serious or life-threatening anaphylaxis. The likelihood that an immune reaction will occur tends to be higher with repeated administration of a foreign protein, but such a reaction could even occur after single-dose administration. Information on serum antibody levels after exposure to the biologic and clinical aspects of immune reactions is among the important types of safety data that must be generated during clinical development. The detection of serum antibody formation requires a sensitive laboratory assay; if no assay is available, the sponsor needs to develop and validate an assay. This procedure may be costly and time-consuming.
The development of immune reactions from administration of a biologic also has implications in the design of studies with animals and the ability to use animal data to support human trials. Generally, for drug products, long-term dosing (e.g., 9 to 12 months) is a necessary prerequisite to chronic dosing in humans. However, nonhuman animals, including nonhuman primates, often develop neutralizing antibodies to biologics, sometimes after only a single exposure, making repeat dosing potentially impossible and irrelevant.
Unlike drugs, the majority of which are for oral administration, the vast majority of therapeutic biologics are for parenteral use only. Thus, many of the biological product formulation issues for pediatric populations are similar to the
formulation issues for parenteral drugs: the safety and feasibility of different volumes and concentrations, the safety of various preservatives, the appropriate packaging of vials or prefilled syringes for pediatric use, and so forth.
There is no simple approach to the evaluation of a biologic product for use in the pediatric population. One needs to think about the intended action or the indication of the particular biologic and the disease or condition that it is proposed to treat. Questions that should be considered during development include the following: Is the disease unique or very common in the pediatric population? Are there alternative therapies? What are the safety and efficacy profiles of those alternative therapies that will make it imperative or important to develop a new agent? What is the relevance and suitability of the adult safety and efficacy data? Will it be possible to extrapolate efficacy data from older to younger children? The answers to these questions are among the considerations for optimizing the program development for biologics for the pediatric population.
The evaluation of the clinical efficacy of biologics is similar to that for drugs and devices. There may be unique safety concerns for biologic products because of their distinctive properties that will require different types of monitoring, such as for adventitious agents or for immune reactions. Little is known about the long-term effects (e.g., years) of treatment with therapeutic biologics, especially the effects on growing children. As more and more biologics are licensed for use for chronic treatment, it will be important to collect and examine data from postmarketing use, including studies from registries and Phase 4 clinical trials (controlled studies following market approval), to further the understanding of this important class of agents.