EDITORIAL
Clinical Trial Design for Antiepileptic Drugs James J. Cereghino, MD In this issue of the Annals, Sachdeo and colleagues report on the efficacy of felbamate for partial onset seizures. They utilized a vahd active treatment control design, which was designed to show and did show a difference in efficacy 11).These results may be considered pivotal in justifying labeling for a monotherapy indication. Epilepsy poses unique problems in demonstrating efficacy to the Food and Drug Administration (FDA). Drugs are available to control seizures in many people. Uncontrolled seizures may result in status epilepticus. At what point is it ethical to use an unmarketed drug rather than a marketed drug? Can a true placebo control ever be used? Based on these considerations, add-on trials (both crossover and parallel) with therapy-resistant patients have become the mainstay of controlled clinical trials for antiepileptic drugs. Placebo control trials have been performed but usually in patients with nonconvulsive seizures (eg., absence epilepsy) or in hospitalized or institutionalized patients under 24-hour observation in protocols with strict criteria for termination of the trial based on safety considerations. Such trials are difficult to perform and have been infrequently done. Similarly, newly diagnosed untreated patients have seldom been used in trials, except for absence epilepsy patients. A major criticism of add-on trials in patients with uncontrolled seizures is that these patients cannot be considered truly representative of the universe of all epilepsy patients and that an efficacious drug may in fact be missed. Add-on trials present difficulties in the areas of drug interactions, increased toxicity, and decreased seizure control. Interactions between the new drug and the concomitant antiepileptic drug or drugs may alter blood levels of any drugs in either direction. Systematic manipulations of the drug dosages have been incorporated successfully into trial designs but add a level of complexity to both trial implementation and trail interpretation. Add-on trials are likely to overestimate toxicity and underestimate maximum tolerated dosage of the new drug, particularly if the toxic effects of the drugs are similar and additive. The magnitude of a new drug effect on seizure control is likely to be less in an add-on trial than in a monotherapy trial. In May 1988, the Epilepsy Branch of the National Institute of Neurological Disorders and Stroke (NINDS) convened a workshop on design issues in controlled clinical trials of investigational antiepileptic drugs. The members represented a variety of disciplines (adult neurology, child neurology, epidemi-
ology, biostatistics) and segments of the epilepsy research community (academia, the pharmaceutical industry, the FDA, NINDS). Workshop goals were to examine critically current trial practices and propose alternative approaches. Proceedings were not published. The workshop yielded a rich vein of clinical trial design ideas. Three alternative trial designs, which appear feasible and ethical, were proposed and are currently being evaluated: (1) valid active control treatment design, (2) concentration control design, and ( 3 ) inpatient seizure evaluation design. A valid active treatment control design uses a known active drug at a specified low concentration (or dosage) and specifies early exit criteria. Sachdeo and colleagues 111 have used valproate at the recommended starting dose as the active control in the expectation that the dosage would not maintain therapeutic control in this drug-refractory patient population. Two patients did complete the 112-day study at this dosage; this could represent true drug efficacy or natural variation in seizure frequency. A major source of controversy in the epilepsy research community has focused on placebo control versus active treatment control studies. In the United States, the FDA rejects efficacy claims based on a finding of no difference between an experimental compound and a marketed antiepileptic drug 12f. The reasons that such claims are not well founded have been repeatedly stated 13, 41. Interestingly, European regulatory authorities tend to take a pragmatic approach and are likely to accept comparative studies against an active control in support of efficacy. A quote from the minutes of a fairly recent bilateral discussion between the European Commission and the FDA on the subject of drug regulations says: “As regards trial design, it is noted that placebo-controlled studies are not always required, indeed, may even be questioned in Europe” { S ] . A concentration control design is based on the assumption that it is reasonable to expect a stronger correlation between blood drug level and response than between drug dosage and response, i.e., the quantity of drug actually available in the blood should be more informative than the amount ingested. Thus clinical trials comparing several plasma concentrations could be more powerful than trials comparing different dosages. This would not be feasible for a drug for which there is no correlation between blood level and efficacy. The demonstration of efficacy would be between the high and low concentration groups. This would approximate the actual clinical situation in which patients are started Editorial: Cereghino: Trial for Antiepileptic Drugs 393
at a low dosage (concentration) and increased based on clinical response. Patients who exit the trial for seizures in the low-dosage group would ethically be expected to continue in a follow-up protocol where they would be given an opportunity to achieve a higher concentration. The inpatient seizure evaluation design can be done at epilepsy centers in which antiepileptic drugs are partially or completely discontinued while patients are in the hospital to characterize their seizures and/or determine whether surgery is a feasible treatment [4]. At the conclusion of this inpatient seizure evaluation, the patient can participate in a controlled trial while remaining in the hospital. These trials could be monotherapy and placebo control. The duration can be limited to a specified number of seizures or a specified number of days, whichever comes first. At least one inpatient seizure evaluation trial has been completed and submitted for publication and another has recently been initiated. The results from the inpatient seizure evaluation design, if positive, will effectively permit investigators to do additional pivotal monotherapy studies much earlier than in the past. The epilepsy community has long been interested in showing efficacy and safety of new antiepileptic compounds as rapidly as possible while at the same time exposing as few patients as possible to an ineffective drug. Concurrent with the new era of identifying putative anticonvulsants in the laboratory based on the concept of “rational design of drugs” is a new era of more efficient trial design in the clinic designed to bring safe, effective drugs to market more rapidly. The article by Sachdeo and colleagues {l]in this issue of the Annals is one of the harbingers of the new era of more efficient clinical design.
James J. Cereghino, M D Epilepsy Branch NINDS, NIH Betbesda, M D
References 1. Sachdeo R, Kramer LD, Rosenberg A, Sachdeo S. Felbamate monotherapy: controlled trial in partial onset seizures. Ann Neurol 1992;32:386-392 2. Leber PD. Hazards of inference: the active control investigation. Epilepsia 1989;3O(suppl l):S57-S63 3. Makuch RW, Pledger G, Hall DB, et al. Statistical issues in drug research and development. In: Peace KE, ed. Active control equivalence studies. New York: Marcel Dekker, 1990:225-262 4. Pledger GW, Kramer LD. Clinical trials of investigational antiepileptic drugs: monotherapy design. Epilepsia 1991;32:7 16-72 1 5. Rees PJ. Monitoring clinical studies in the European Community: comparison of clinical research in the U.S. and Europe. J Clin Res Pharmacoepidemiol 1991;6:1-10
Correction In the article “Recurrent Status Epilepticus in Children,” published in the June issue of the Annals (Shinnap s, MaytalJ, Krasnoff L, Moshe SL, 1992;31:598604),an error was made by the Annals in Table 3. The corrected table is printed below. Annals apologizes for this error. Table 3. Recurrent Status Epilepticus (SE) and Neurological Status Normal n
Idiopathic Remote symptomatic Febrile Acute symptomatic Progressive encephalopathy Total
394 Annals of Neurology Vol 32 No 3 September 1992
n
2 2 Episodes
Abnormal 2 2 Episodes
n
24 24 1 18 0 0 29 24 0 18 15 1 6 0 0
0 18 5 3
0 8 1 1
6
4
63 2
32
14
95
Clinical Trial Design for Antiepileptic Drugs James J. Cereghino, MD In this issue of the Annals, Sachdeo and colleagues report on the efficacy of felbamate for partial onset seizures. They utilized a vahd active treatment control design, which was designed to show and did show a difference in efficacy 11).These results may be considered pivotal in justifying labeling for a monotherapy indication. Epilepsy poses unique problems in demonstrating efficacy to the Food and Drug Administration (FDA). Drugs are available to control seizures in many people. Uncontrolled seizures may result in status epilepticus. At what point is it ethical to use an unmarketed drug rather than a marketed drug? Can a true placebo control ever be used? Based on these considerations, add-on trials (both crossover and parallel) with therapy-resistant patients have become the mainstay of controlled clinical trials for antiepileptic drugs. Placebo control trials have been performed but usually in patients with nonconvulsive seizures (eg., absence epilepsy) or in hospitalized or institutionalized patients under 24-hour observation in protocols with strict criteria for termination of the trial based on safety considerations. Such trials are difficult to perform and have been infrequently done. Similarly, newly diagnosed untreated patients have seldom been used in trials, except for absence epilepsy patients. A major criticism of add-on trials in patients with uncontrolled seizures is that these patients cannot be considered truly representative of the universe of all epilepsy patients and that an efficacious drug may in fact be missed. Add-on trials present difficulties in the areas of drug interactions, increased toxicity, and decreased seizure control. Interactions between the new drug and the concomitant antiepileptic drug or drugs may alter blood levels of any drugs in either direction. Systematic manipulations of the drug dosages have been incorporated successfully into trial designs but add a level of complexity to both trial implementation and trail interpretation. Add-on trials are likely to overestimate toxicity and underestimate maximum tolerated dosage of the new drug, particularly if the toxic effects of the drugs are similar and additive. The magnitude of a new drug effect on seizure control is likely to be less in an add-on trial than in a monotherapy trial. In May 1988, the Epilepsy Branch of the National Institute of Neurological Disorders and Stroke (NINDS) convened a workshop on design issues in controlled clinical trials of investigational antiepileptic drugs. The members represented a variety of disciplines (adult neurology, child neurology, epidemi-
ology, biostatistics) and segments of the epilepsy research community (academia, the pharmaceutical industry, the FDA, NINDS). Workshop goals were to examine critically current trial practices and propose alternative approaches. Proceedings were not published. The workshop yielded a rich vein of clinical trial design ideas. Three alternative trial designs, which appear feasible and ethical, were proposed and are currently being evaluated: (1) valid active control treatment design, (2) concentration control design, and ( 3 ) inpatient seizure evaluation design. A valid active treatment control design uses a known active drug at a specified low concentration (or dosage) and specifies early exit criteria. Sachdeo and colleagues 111 have used valproate at the recommended starting dose as the active control in the expectation that the dosage would not maintain therapeutic control in this drug-refractory patient population. Two patients did complete the 112-day study at this dosage; this could represent true drug efficacy or natural variation in seizure frequency. A major source of controversy in the epilepsy research community has focused on placebo control versus active treatment control studies. In the United States, the FDA rejects efficacy claims based on a finding of no difference between an experimental compound and a marketed antiepileptic drug 12f. The reasons that such claims are not well founded have been repeatedly stated 13, 41. Interestingly, European regulatory authorities tend to take a pragmatic approach and are likely to accept comparative studies against an active control in support of efficacy. A quote from the minutes of a fairly recent bilateral discussion between the European Commission and the FDA on the subject of drug regulations says: “As regards trial design, it is noted that placebo-controlled studies are not always required, indeed, may even be questioned in Europe” { S ] . A concentration control design is based on the assumption that it is reasonable to expect a stronger correlation between blood drug level and response than between drug dosage and response, i.e., the quantity of drug actually available in the blood should be more informative than the amount ingested. Thus clinical trials comparing several plasma concentrations could be more powerful than trials comparing different dosages. This would not be feasible for a drug for which there is no correlation between blood level and efficacy. The demonstration of efficacy would be between the high and low concentration groups. This would approximate the actual clinical situation in which patients are started Editorial: Cereghino: Trial for Antiepileptic Drugs 393
at a low dosage (concentration) and increased based on clinical response. Patients who exit the trial for seizures in the low-dosage group would ethically be expected to continue in a follow-up protocol where they would be given an opportunity to achieve a higher concentration. The inpatient seizure evaluation design can be done at epilepsy centers in which antiepileptic drugs are partially or completely discontinued while patients are in the hospital to characterize their seizures and/or determine whether surgery is a feasible treatment [4]. At the conclusion of this inpatient seizure evaluation, the patient can participate in a controlled trial while remaining in the hospital. These trials could be monotherapy and placebo control. The duration can be limited to a specified number of seizures or a specified number of days, whichever comes first. At least one inpatient seizure evaluation trial has been completed and submitted for publication and another has recently been initiated. The results from the inpatient seizure evaluation design, if positive, will effectively permit investigators to do additional pivotal monotherapy studies much earlier than in the past. The epilepsy community has long been interested in showing efficacy and safety of new antiepileptic compounds as rapidly as possible while at the same time exposing as few patients as possible to an ineffective drug. Concurrent with the new era of identifying putative anticonvulsants in the laboratory based on the concept of “rational design of drugs” is a new era of more efficient trial design in the clinic designed to bring safe, effective drugs to market more rapidly. The article by Sachdeo and colleagues {l]in this issue of the Annals is one of the harbingers of the new era of more efficient clinical design.
James J. Cereghino, M D Epilepsy Branch NINDS, NIH Betbesda, M D
References 1. Sachdeo R, Kramer LD, Rosenberg A, Sachdeo S. Felbamate monotherapy: controlled trial in partial onset seizures. Ann Neurol 1992;32:386-392 2. Leber PD. Hazards of inference: the active control investigation. Epilepsia 1989;3O(suppl l):S57-S63 3. Makuch RW, Pledger G, Hall DB, et al. Statistical issues in drug research and development. In: Peace KE, ed. Active control equivalence studies. New York: Marcel Dekker, 1990:225-262 4. Pledger GW, Kramer LD. Clinical trials of investigational antiepileptic drugs: monotherapy design. Epilepsia 1991;32:7 16-72 1 5. Rees PJ. Monitoring clinical studies in the European Community: comparison of clinical research in the U.S. and Europe. J Clin Res Pharmacoepidemiol 1991;6:1-10
Correction In the article “Recurrent Status Epilepticus in Children,” published in the June issue of the Annals (Shinnap s, MaytalJ, Krasnoff L, Moshe SL, 1992;31:598604),an error was made by the Annals in Table 3. The corrected table is printed below. Annals apologizes for this error. Table 3. Recurrent Status Epilepticus (SE) and Neurological Status Normal n
Idiopathic Remote symptomatic Febrile Acute symptomatic Progressive encephalopathy Total
394 Annals of Neurology Vol 32 No 3 September 1992
n
2 2 Episodes
Abnormal 2 2 Episodes
n
24 24 1 18 0 0 29 24 0 18 15 1 6 0 0
0 18 5 3
0 8 1 1
6
4
63 2
32
14
95
