Jacqueline A. French MD*
Marc A. Dichter MD**
* Professor, Department of Neurology, University of Pennsylvania; and Associate Director, Pennsylvania Epilepsy Center, Department of Neurology, Hospital of the University of Pennsylvania, Philadelphia, Pennsylvania
** Department of Neurology, University of Pennsylvania, Philadelphia, Pennsylvania
This chapter is one of three that address the clinical development of antiepileptic drugs (AEDs). Whereas this one addresses the “medical perspective,” the other two address the pharmaceutical and regulatory perspectives. The very fact that these different viewpoints are included is an indication that drug development must satisfy many differing goals. The goal of regulatory authorities is to obtain rigorous scientific evidence that a new AED is safe and effective. The goal of clinicians is to obtain the kind of clinically relevant information about a new AED that will lead to appropriate selection of treatment for their patients, including information about how to use a drug to its maximal advantage, and data that can determine accurate risk assessment. The first priority of the pharmaceutical industry is to obtain drug approval from regulatory bodies, which allows the drug to be marketed and sold. However, the needs of clinicians also must be addressed, if there is to be any demand for the drug once it is available. Clinicians frequently feel that their needs may be subservient to regulatory needs.
Unfortunately, clinical trials leave many clinical questions unanswered. This chapter reviews the ways that information about new AEDs is obtained during the development process. Emphasis is placed on how to interpret regulatory trials in the context of clinical use, and also on sources of information beyond regulatory trials.
HISTORY OF ANTIEPILEPTIC DRUG DEVELOPMENT
Until the middle of the nineteenth century, people with epilepsy were treated with a variety of “home remedies” and superstitious practices, the effectiveness of which is impossible to determine in the modern age. In the 1850s, bromide was introduced as the first true AED, although it was originally proposed as a treatment for epilepsy for the wrong reasons. For the next half-century bromide remained the only drug available, until just after the turn of the twentieth century, when phenobarbital was accidentally discovered to be effective in suppressing seizures. In the late 1930s, the first “rational” plan to discover AEDs was conceived by Tracy Putnam and Houston Merritt, when they developed a model of experimental epilepsy (the maximal electroshock test) and then tested compounds that had chemical structures similar to the barbiturates to find a compound that would suppress seizures while not producing sedation. They discovered phenytoin with this methodology, and within just a few years of their discovery, phenytoin was marketed for the treatment of seizures. Over the next 30 years, a small but steady stream of new antiseizure drugs was discovered and marketed and new animal screening tests were developed to complement the maximal electroshock test. These drugs included ethosuximide, primidone, mesantoin, and carbamazepine. In 1978, sodium valproate, the last of what are now referred to as the old AEDs orconventional AEDs, was approved for use in the United States, after considerable experience in Europe and considerable pressure on the U.S. Food and Drug Administration (FDA). For the most part, clinical trials of these compounds were not done in the rigorous fashion used today. Monotherapy was not considered a separate issue for approval. Wording in FDA approval statements was broad and inclusive. For example, carbamazepine was approved for “Partial seizures with complex symptomatology (psychomotor, temporal lobe), generalized tonic-clonic seizures and mixed seizure patterns” (1).
In the ensuing 15 years, an intensive AED screening program was developed at the National Institute of Neurological Disorders and Stroke and thousands of new chemical entities were tested in a variety of animal models to look for
promising new AEDs. Some clinical trials resulted from these screening tests, but most failed because of either intolerable side effects or lack of demonstrable efficacy.
The “modern era” of new AEDs dawned in 1993 with the approval of the first new AED in 15 years, felbamate. This was followed quickly by a number of new drugs, both in the United States and in Europe, including gabapentin, lamotrigine, topiramate, tiagabine, oxcarbazepine, levetiracetam, and zonisamide. Each of these drugs was first tested as adjunctive therapy in adults with intractable partial seizures, although several were tested in monotherapy trial designs or in pediatric populations almost simultaneously (see pp. 52-53). A number of additional drugs, including some with potentially novel mechanisms of action, were started in clinical trials that did not succeed. Despite the introduction of eight new AEDs, approximately 25% to 35% of adults with partial seizures are not able to become seizure free. Thus, there is a continuing need to develop new drugs that will be effective in the subgroup of patients whose disease remains intractable.
ASSESSMENT OF EFFICACY
As noted previously, many new AEDs have been approved for use in the last several years (2). For the most part, approval has derived from large, well controlled, adjunctive, multicenter trials. These trials, most of similar design, have been an excellent way of determining efficacy to the satisfaction of regulatory bodies. Efficacy can be defined as the ability to reduce seizures in the context of a clinical trial. What these trials are not able to demonstrate is the effectiveness of these drugs. Effectiveness is defined as the value of an AED in the environment of use, or, in other words, its ability to benefit patients in clinical practice. Although efficacy and effectiveness are linked, they are not always the same. For example, a drug could demonstrate excellent efficacy, but only when administered in a five-times-per-day schedule. Because most people could not comply with such a schedule, breakthrough seizures would occur as a result of noncompliance, and the drug would not be very effective. Other examples could be given.
Demonstration of Efficacy
In 1962, the Kefauver-Harris amendments to the Federal Food, Drug and Cosmetic Act were passed. This ushered in the modern age of drug testing by requiring that pharmaceuticals be proved efficacious before marketing (3). At die same time, federal agencies were given the authority to decide whether safety and efficacy had been satisfactorily demonstrated. It was after this legislation that clinical trials, as they are now performed, began to emerge.
Two “adequate and well controlled clinical trials,” so- called pivotal trials, that demonstrate efficacy must be performed for FDA approval. Although different types of control, including historical control, can be used, placebo control is favored (4). The presence of a placebo control group in most trials to determine efficacy has a significant effect on trial design and subject selection, which is discussed later.
Clinical trials performed for regulatory purposes have several characteristics that make it difficult to extrapolate results to patients in clinical practice. Some of these are discussed in the following sections.
Epilepsy comprises a diverse group of syndromes, each with a unique clinical presentation and frequently different genetics, etiology, and possibly underlying biochemical defect (5). Most regulatory trials are performed in patients with partial epilepsy, demonstrating complex partial seizures, with or without secondary generalization. Subjects also may experience simple partial seizures, but they may or may not be acceptable as a single seizure type for trial enrollment. Partial epilepsy is advantageous because it represents the most common type of epilepsy in adults (6). Also, patients with refractory partial epilepsy are seen at large epilepsy centers, where most efficacy trials are performed. The partial seizure population also is seen as the patients with the greatest unmet need because approximately one-third have uncontrolled seizures despite medical therapy. As a result, many AEDs have been approved for treatment of partial seizures and comparatively few for other types, such as absence, myoclonus, and infantile spasms. Trials in other populations are becoming more common, however. For example, lamotrigine has received approval for use in seizures associated with Lennox-Gastaut syndrome, based on positive clinical trials (7). Topiramate was shown to reduce seizures in a trial in Lennox-Gastaut syndrome (8), as well as in a novel clinical trial of patients with primary generalized tonic-clonic convulsions (9). Large-scale randomized trials of the new drugs have not yet been undertaken in patients with absence seizures or myoclonus.
Traditionally, the patients who are enrolled in pivotal trials have very frequent seizures, a long duration of epilepsy, and have failed numerous AEDs before they are recruited. Enrollment criteria usually require a minimum of three to four complex partial or secondarily generalized seizures per month. These criteria ensure that there will be enough measurable seizures during a 3-month trial to obtain a statistically significant result. Commonly, patients experience many more seizures per month than are required by inclusion criteria. Demographic data from several recent pivotal
trials are included in Table 4.1. The use of such patients may not demonstrate a new compounds true effectiveness, and it has been hypothesized that our best conventional AEDs might fail such a difficult test (10). Even if the experimental drug can be proved effective in this challenging population, it is unclear whether the results can or should be generalized to die remainder of patients with epilepsy, who, by and large, do not have intractable disease, and have fewer seizures. The effort to conduct trials in patients with less intractable or even new-onset disease has attracted increasing support.
TABLE 4.1. DEMOGRAPHICS FOR RECENT, RANDOMIZED, PLACEBO-CONTROLLED ADJUNCTIVE TRIALS OF NEW ANTIEPILEPTIC DRUGS
Another reason why clinical trials do not necessarily mirror clinical practice is that in trials, patients are titrated to fixed doses, which may either be excessive for that individual, leading to unnecessary toxicity, or too low, leading to suboptimal seizure control. In fact, some of the newer AEDs are being used at doses higher than those that were used in clinical trials, whereas others are used at lower doses. Trial results, as determined by 50% responder rates or percentage seizure reduction rates in recent clinical trials, have been significantly colored by the doses selected for use in the trial. Drugs that are tested at lower doses appear well tolerated, but less effective, whereas drugs tested at higher doses may appear highly effective, but with more toxicity. This appearance is highlighted by the way in which outcomes are statistically calculated in most trials. Patients who are placed on a higher dose than they can tolerate may drop out before the trial is completed. If this occurs, they still may be counted as “responders” if they experienced seizure reduction before dropout, even though the observation period can be short. Table 4.2 shows the highest dose studied in pivotal trials for recently approved AEDs, and the responder rate and the dropout rates at that dose. Also included is the dose now typically considered an average dose for that AED, when used as add-on therapy. In some cases, the dropout rate was higher than the responder rate.
The most confounding aspect of pivotal trials is that they do not allow patients to reach their own “ideal” target dose. All practitioners have experience with patients who are not seizure free when a new AED is initiated, but who eventually can be titrated to a dose that renders them seizure free. Not uncommonly, these patients need to be down-titrated on their concomitant AED medication for higher doses of the new AED to be tolerated. This type of manipulation is not typically accommodated during a clinical trial. Therefore, it is a common belief that clinical trials underrepresent the efficacy of many drugs, even in a population that is difficult to control.
TABLE 4.2. COMPARISON OF DOSE STUDIED IN ADJUNCTIVE, RANDOMIZED, CONTROLLED TRIALS COMPARED WITH CLINICAL PRACTICE, AND ITS EFFECT ON EFFICACY AND DROPOUT RATES
Another variable that clearly colors the results of clinical trials, and the perception of drugs when they are approved, is the selection of titration rates. There are several examples
of drugs that may have been titrated too rapidly in clinical trials, leading to a relatively higher dropout rate. One example is lamotrigine. Studies have clearly demonstrated that the incidence of rash is tied to titration rate, as well as background medication (11,12). A very slow titration rate is necessary when lamotrigine is added to valproic acid (12, 13), either alone or in combination with enzyme-inducing AEDs. Unfortunately, this information was not as clear when clinical trials were being performed. The trial that may have been most affected by this was the trial in patients with Lennox-Gastaut syndrome (7). Many patients in this trial were receiving valproic acid, and in retrospect the titration rate was too rapid, leading to a 1% incidence of serious rash. This, in part, led to a black box (serious) warning in FDA labeling when lamotrigine is used in children.
In some instances, rapid titration leads to decreased tolerability because of cognitive or other side effects. This was probably the case for topiramate and zonisamide, both of which are better tolerated when titrated more slowly (14, 15).
The titration rate for pivotal trials may be selected early in drug development, when less information is available. In many cases, the titration rate used in these trials is placed in FDA labeling. Clinicians should be aware that this may not represent the most useful titration rate in all patients. Often, slower titration leads to better tolerability, particularly in patients who are sensitive to drug side effects. On the other hand, a faster titration rate may be safe and effective, and might be considered when a rapid effect is desirable. Unfortunately, rigorously obtained data often are not available to guide the practitioner on the relative risk of rapid titration. One exception to this rule is levetiracetam. A blinded study compared initiation at 1, 2, or 4 g (16). Results indicated that the dropout rates were similar, although side effects were greater in the group started at the highest dose. Of the newer AEDs, only lamotrigine cannot be titrated rapidly owing to safety concerns (13). The others can be rapidly titrated, but an increase in side effects will most likely result.
It is very frustrating to many clinicians that by the time AEDs are approved, there rarely has been a clear “therapeutic range” established. However, there is a good reason for this. For many AEDs, particularly those that are hepatically metabolized, serum levels are very variable and cannot be predicted by dose (17). As noted previously, patients are randomized to a dose, and not a serum level. The dose that they are randomized to may be too high or too low for them. Also, as noted, overly high or low doses may have been selected for study. In addition, in this very refractory population, even those who achieve a high serum level might not respond (18). Therefore, level-response relationships rarely are established in a randomized, placebo-controlled, dose-response add-on trial (19). In general, the only information that can be obtained is a range of plasma levels achieved in the trials. Optimal serum levels usually are determined by open-label postmarketing investigations in which patients can be titrated to their optimal dose.
Clinical trials are of short duration. Typically, seizure reduction counts are based on 3 to 4 months of observation. This may be too short to see evidence of the development of tolerance. On the other hand, some drugs may demonstrate an increase in efficacy or a decrease in side effects over time. Increasing efficacy with extended use has been postulated for valproic acid and the vagus nerve stimulator (20). Again, such effects are missed over the typical drug study duration.
Another issue that is difficult to address in controlled clinical trials is the potential for becoming seizure free over the long-term. The goal of epilepsy therapy is a “cure,” as manifested by elimination of seizures over years and even decades. Many patients report that introduction of a new AED provides short-term seizure relief, but seizures ultimately recur. This can be devastating, both medically and emotionally. Controlled clinical trials simply are not able to determine how many patients will remain free of seizures. It therefore is critically important to follow patients after randomized, controlled trials have been completed, to determine long-term efficacy. Unfortunately, this is rarely done in a rigorous manner.
Pivotal trials are performed in the adult population first. For most studies, upper age cutoffs are imposed at 65 to 70 years of age. Lower age cutoffs have been lowered gradually. Whereas the lower cutoff used to be 18 years of age, more recent trials have studied patients as young as 12 years of age (21). Although even younger patients eventually will be tested in randomized, controlled trials leading to a pediatric indication, the elderly rarely are studied in a rigorous manner, despite the high incidence of epilepsy in this age group, with approximately 25% of new cases of epilepsy occurring in elderly people (22). Moreover, the etiologies and pathophysiologic processes are thought to be different in this group, potentially leading to a differentiation in drug response. Also, there are differences in pharmacokinetics and ability to tolerate specific side effects, such as fatigue and cognitive dysfunction, that may distinguish elderly individuals from their younger counterparts (23). Therefore, it is not clear that results from pivotal trials should be generalized to the elderly. One blinded study explored the response to carbamazepine and lamotrigine in a newly diagnosed
elderly population, and found that lamotrigine was better tolerated (24). More such studies are needed, and indeed several are ongoing.
Monotherapy versus Polytherapy
Pivotal trials for determining safety and efficacy of new AEDs almost always are designed as adjunct studies. In other words, the drug under investigation is added to any AEDs the subject may already be taking to control his or her seizures. This design allows a placebo control group, which otherwise would be very difficult for such a serious disease. In a recent article, Robert Temple, of the FDA, advocated an add-on study design for “trials…in which omitting standard therapy would generally be unacceptable. Such studies are not directly informative about a drug as monotherapy, but they do provide interpretable evidence of effectiveness in a well-defined setting” (25). In fact, many patients with refractory epilepsy are treated with polytherapy, and it is critically important to determine the effectiveness, safety, and side effect profile in this setting. However, eventually, as less severely ill patients, or even patients with newly diagnosed epilepsy, receive new AEDs, it also is important to obtain information about drug activity in the monotherapy setting. The best way to obtain this information is still controversial, both in the United States and Europe.
The preferred method of demonstrating efficacy in monotherapy would be to compare placebo with active drug. Concerns about patient safety preclude this approach. Several trial designs have been devised that attempt to address safety concerns, but at the same time demonstrate effectiveness of a new drug as monotherapy (26,27). These designs use a “pseudoplacebo” in place of a true placebo. Patients in the pseudoplacebo arm receive some treatment to prevent catastrophic seizures or severe worsening, but not enough to prevent the complex partial seizures that are being evaluated in the study. Previously performed trials of this sort have used low-dose valproic acid or a low dose of the study drug as pseudoplacebo. Two commonly used trial designs use a pseudoplacebo arm. In one, the surgical withdrawal design, the subjects are undergoing medication withdrawal for the purpose of presurgical evaluation. When most or all medication has been eliminated, the experimental AED or placebo/pseudoplacebo is added in a randomized, blinded fashion. The trial ends after subjects have experienced a prespecified number of seizures (“failures”) or have gone a certain time (usually 7 to 10 days) without that number of seizures having occurred (“completers”). Analysis is based on how many patients complete in the pseudoplacebo arm compared with the active arm.. These trials now are considered by many experts to be too short to generalize results to the outpatient setting.. A second design is performed in outpatients. Again, patients are randomized to pseudoplacebo or active drug, added on to baseline therapy. Therapeutic failure is determined on the basis of escape criteria, such as doubling of seizure frequency or increase in seizure severity. If more patients receiving pseudoplacebo reach escape criteria compared with those receiving active drug, the drug is determined to be effective in monotherapy. These protocols now are commonly used when a monotherapy indication is sought (28, 29, 30, 31).
This trial design, although it provides clear information for regulators, is of little value to clinicians because it does not address improvement when patients convert to monotherapy, but rather compares the patients converting to the study drug with a population that clearly will worsen as a result of being randomized to pseudoplacebo. This provides evidence only that the study drug is better than nothing.
Monotherapy studies also may be performed in newly diagnosed patients. In Europe, studies typically use an active control design, in which a new drug is compared with a standard (32,33). Some studies compare high and low doses. For example, one study in newly diagnosed patients demonstrated that 900 and 1,800 mg of gabapentin were more effective than 300 mg in preventing seizures requiring exit from the study (33). These types of dose-controlled studies in the newly diagnosed are somewhat dangerous. Most newly diagnosed patients who will become seizure free on a given drug appear to do so on relatively low levels of that drug (24). Thus, comparing a low dose with a high dose of a new drug in this population may demonstrate reasonably similar efficacy in both groups, leading to the erroneous conclusion that the drug is not substantially better than placebo, when, in fact, both doses of the study drug may be effective.
Monotherapy trial designs are controversial. Issues have been raised about the ethics of randomizing patients to treatment arms that are known to be suboptimal (pseudoplacebo), and also about the artificial nature of some of the designs, making it difficult to translate results to real-life situations (34, 35, 36, 37). There possibly is more dissatisfaction with regulatory trials leading to monotherapy indications than with any other type of industry-sponsored trials. In this arena, only the active-controlled trials performed in Europe provide information useful to clinicians in treating patients. Discussion of the optimal way to perform these trials is ongoing. It is hoped that better solutions will be found in the future, which will provide useful regulatory and clinical data.
In a more controversial vein, it can be argued that all of the new (and old) AEDs that have been approved as adjunctive therapies also work in monotherapy, and probably at approximately the same efficacy as drugs that have been more rigorously studied in monotherapy. Thus, one can ask, especially given the ethical issues discussed earlier and the technical difficulties of carrying out monotherapy trials, whether these kinds of trials need to be performed at all. An alternative would be to have AEDs approved for use in partial seizures or primary generalized seizures of a
given type without referring to adjunctive or monotherapy. This is especially important because adjunctive implies that a given drug will not work by itself, but only as a “helper” to some other form of therapy, and this clearly is not the case for AEDs. The FDA's current position on this issue is that they can approve a drug only for an indication in which there is rigorous proof of its efficacy, and because AEDs can be ethically tested initially only in adjunctive, add-on trials, that is the only indication they can receive from the FDA. This is a major area in which there is a conflict between the regulatory concerns and those of clinicians.
ASSESSMENT OF PHARMACOKINETICS
The behavior of a drug, once it enters the body, depends on it properties of absorption, metabolism, distribution, and elimination. Many AEDs have complex pharmacokinetic properties. Understanding the unique pharmacokinetic characteristics of AEDs is critical to their appropriate use. Most pharmacokinetic characteristics should be thoroughly explored at the time of AED approval. Usually, the first step in obtaining such understanding derives from preclinical investigations. Recently, it has become possible to predict some hepatic metabolic effects in vitro. Liver enzyme tests can indicate toxicity and effect on hepatic metabolism, including inhibition of cytochrome P450 enzymes. Liver tests also may help to establish the route of metabolism (e.g., oxidation or glucuronidation) (38). Further information is obtained from early phase 1 and phase 2 trials, which usually include trials specifically aimed at elucidating interactions between AEDs. At this stage, interactions with commonly used medications such as oral contraceptives and warfarin also may be explored (39). By the time large, phase 3, double-blind, placebo-controlled trials begin, most common interactions will have been thoroughly studied. Yet, when AEDs are approved, some pharmacokinetic data still may be lacking. For example, there may be limited data on the behavior of drugs in certain populations, such as the elderly, children, and patients with chronic liver or kidney dysfunction. Pharmacokinetic properties in pregnancy also may be unknown. More important, clinical trials focus on population pharmacokinetics and less on the behavior of a drug in a given individual under clinical conditions. For example, it is very common to down-titrate and ultimately discontinue background AEDs once a new AED is introduced. Because many of the older AEDs either induce or inhibit new AED metabolism (40), changes in serum level of the new AED can be expected. The clinician who is treating such a patient is caught in a dilemma: At what point should the new AED dose be adjusted during downtitration of the background drug, to maintain equivalent serum levels and seizure control? Such questions rarely are addressed in regulatory trials.
ASSESSMENT OF SAFETY AND TOLERABILITY
It can be considered an axiom that no drug is absolutely safe or without side effects. Consequently, AEDs, which often must be taken for years and in relatively high doses, commonly have issues that relate to both tolerability and longterm safety. Both of these are assessed during the development and clinical evaluation of new AEDs, but the ability of short-term studies on limited populations of highly selected individuals to detect all the issues related to either tolerability or safety is severely limited. For the purposes of this discussion,tolerability is related to the development of side effects from a drug that are unpleasant and may limit the drug's usefulness, whereas safety refers to serious or life-threatening side effects.
AEDs are designed to work on the brain and limit the activity of neurons that are hyperactive during seizures. However, it is imperative that these drugs not interfere with the normal function of the brain. It actually is remarkable that any such compounds can be developed at all, rather than that they all may confer unpleasant side effects in some individuals. Initially, in phase 1 studies, AEDs are tested in a very small number of people who have not been chronically exposed to drugs, and these would likely be those most sensitive to side effects of the drug. Titration rates, dosage, and duration of exposure all affect the likelihood of intolerable side effects. Once the drug gets past this small group of volunteers, it usually is tested on several hundred to a few thousand subjects before it is released. These are patients with intractable epilepsy who typically are on multiple drugs, and who are used to tolerating a variety of side effects while trying to get their seizures under control. In addition, they are highly selected for the clinical trials and often are afforded special medical attention during the clinical trial. Thus, there are reasons why they would be more susceptible to side effects (multiple drugs) as well as less susceptible (experience, enhanced level of care). Typically, at higher doses, and especially during add-on trials, many new AEDs cause some degree of sedation (which may be called fatigue or lethargy). Sometimes this may decrease as a patient continues to take the medication, a process of growing tolerance. It is hoped that a similar tolerance will not occur to the antiseizure effect of the drug. Other important central nervous system issues of tolerability relate to decreases in cognitive function and changes in affective behavior (e.g., irritability, anxiety). Both of these may be hard to quantitate, especially when specific tests usually are not performed. Other, more easily assessed side effects commonly seen include dizziness, ataxia, diplopia, gastrointestinal symptoms, and headache. It is common for these to be reported during controlled, double-blind studies, and in some cases there is little difference between those on active drug and those on placebo (which is one of the more important aspects of placebo control). In general, if
dropout rates are not too high, tolerability issues neither cause the premature termination of a trial nor prevent a successful outcome of a trial and registration of a drug. However, once a drug is approved, a number of additional considerations develop. First, a much larger number of less carefully selected patients are given the drug, and new problems may arise that were not appreciated in the smaller exposure. Second, the drug may be used in patients who have had less experience with AEDs and therefore may be less tolerant. Third, patients stay on the drug for longer periods and new problems with tolerability may be noted, such as excessive weight gain or loss, or hair loss. Fourth, patients may be on other types of drugs that were excluded from the clinical trials (e.g., antibiotics, oral contraceptives, antihypertensives, antidepressants), and tolerability issues may develop based on unanticipated pharmacokinetic or pharmacodynamic interactions.
The most problematic tolerability issue in interpreting clinical trials is distinguishing phamacodynamic side effects from those produced by the drug in isolation. Most of the data we have about AEDs derive from adjunctive trials. When AEDs are combined, side effects often are magnified beyond those seen with either drug alone. Often, these pharmacodynamic side effects may be seen with one drug combination, but not others. The magnitude of this problem was highlighted by several monotherapy studies in outpatients. As noted previously, these studies often involve adding a new drug, then removing all background drugs. In such a trial performed with lamotrigine, dizziness occurred in 20% of patients when lamotrigine was used as adjunctive therapy, but in only 7% when background AEDs were withdrawn (28).
Thus, tolerability seen in small clinical trials may not be identical to that seen when a drug becomes available to the larger population of patients with epilepsy, or when the drug is used as monotherapy.
Assessment of Safety
All new chemical entities undergo laboratory (animal) investigation for safety before they are available for testing in humans. These tests include both short-term assessments in multiple species and longer-term assessment for chronic toxicity, especially carcinogenesis (41). In addition, drugs are tested in two species of animals for possible teratogenesis, although it is not clear that for this purpose animal testing can be readily extrapolated to humans. Safety issues then are very carefully monitored during clinical trials and any adverse event, even if apparently unrelated to drug administration, is recorded and reported.
Some potentially serious safety issues are not easily routinely monitored. For example, adverse effects on hormonal status in women or on behavioral states in general may be missed either if they are subtle or take a relatively long exposure to be manifest. Thus, it would not be expected that a propensity to polycystic ovaries or a problem with fertility would be noted in a typical 12- or 24-week exposure to a new AED. Similarly, depression is relatively common in patients with epilepsy, and can be multifactorial, so it is possible that subtle but significant changes in affect could be missed during a brief trial. In general, one sees what one is looking for or, in some cases, things that cannot be ignored. Thus, if trials are not set up to examine specific parameters, safety issues that relate to these conditions could be overlooked. This clearly is not an issue for AEDs alone, but because these drugs are used for prolonged periods in patients who already have a number of associated problems, some types of safety issues may be more problematic.
Perhaps one of the most serious issues in safety assessment relates to the fact that evaluations during brief clinical trials clearly are unable to ascertain serious adverse events that occur only uncommonly or rarely. The recent example of the emergence of very serious safety issues after felbamate was approved is a graphic example of this. In a number of individuals, aplastic anemia or hepatic failure developed, both of which were not seen, or even hinted at, during the clinical trials (42). The combination of felbamate being the first of the new AEDs released and an aggressive and effective marketing campaign resulted in felbamate being taken by more than 100,000 people in the United States within 1 year of approval. These two very serious, and in some cases, fatal, toxic side effects caught the medical community by surprise, but unfortunately emphasize the fact that rare idiosyncratic side effects of any new drug will be detected only after widespread use. Adequate postmarketing surveillance is the only method by which such events will be detected and long-term safety ascertained.
Another issue in the assessment of safety relates to the use of AEDs in the pediatric population. These patients are undergoing profound changes in the brain during their normal development, and it is important that drugs that work on the brain not interfere with these developmental processes. It might take years before adverse effects on these developmental events manifest, and it also would likely take careful, detailed assessments of both cognitive and behavioral performance to detect any problems (43). Unfortunately, an adequate analysis has not been performed on any AED, either “old” or “new.” Although gross problems have not been identified, the issue of more subtle changes in general has not been addressed. It would be extremely difficult, and quite impractical, to conduct adequate long-term developmental safety trials during the registration process for drugs that will be used in the pediatric population. On the other hand, it is hard to argue with the importance of such issues. Perhaps, good postmarketing trials could address these issues.
One other major safety issue relates to potential teratogenicity of new AEDs. Most people with epilepsy are young, and half are women. Thus, many people on chronic AED therapy are women who will likely want to become
pregnant and have children despite their epilepsy. Data on older AEDs suggest that using one drug in relatively moderate doses does not markedly increase the risk for major fetal malformations over that of offspring of women with epilepsy who are not taking AEDs, in that the risk is still on the order of 4% to 6% (44). Less information is available for the newer AEDs. As mentioned, the validity of preclinical testing is questionable. Fortunately, pregnancy registries have been established in many countries throughout the world to collect data about possible adverse outcomes of pregnancies exposed to AEDs. Over time, these should provide good data about the relative risks of each of the available AEDs.
ALTERNATIVES TO REGULATORY TRIALS
If regulatory trials cannot provide sufficient information to inform the treating physician, what are the alternatives? There have been many attempts to “fill the knowledge gap,” each with its own set of problems.
One solution has been to perform a meta-analysis of several trials in an attempt to arrive at comparative efficacy and safety data. Recently, such an analysis was done for many of the new drugs. This study revealed that efficacy rates often paralleled tolerability. In other words, the higher the responder rate, the more tolerability problems existed (2,45). Although this approach has some utility, it suffers from the same problems as do the regulatory trials from which it derives, namely, fixed doses and titration schedules and a refractory patient sample.
What of other solutions? Many studies have been performed that purport to mimic clinical practice. Typically, these trials follow a cohort of patients over a long duration in an open fashion. Variables that typically are assessed include discontinuation rates, adverse events, and percentage seizure freedom (46,47). These trials can definitely provide very important efficacy and safety information, particularly in populations that are not routinely studied in randomized, prospective trials. However, there are drawbacks to this approach, including the absence of a control group, and potential bias with regard to choice of patients that go on each drug.
What would be most useful would be a randomized “use” trial, in which patients are randomized to one of the new drugs, titrated to their ideal dose, and followed over a long period of time. Such trials were common in the 1980s and provided a wealth of data that still is relevant today (48, 49). Some of the European active-controlled monotherapy trials in newly diagnosed patients are performed in this fashion (32,33). A large-scale use trial was initiated recently in Europe. This type of trial should provide clinicians with important information on the effectiveness of the new AEDs.
Controlled clinical trials are essential in proving that drugs are safe and effective for use. However, they cannot fully inform the clinician in the use of new AEDs. The typical patient in practice is not treated in the same fashion as a subject in a clinical trial. The need for scientific rigor and a controlled environment in clinical investigation is crucial, but at the same time often introduces a certain rigidity in population selection, titration, dose selection, and manipulation of background drugs that reduces the generalizability of the results. The gaps in knowledge need to be filled by additional studies, which try to address clinical questions in as rigorous a way as possible.