Results of a Nationwide Veterans Administration Cooperative Study Comparing the Efficacy and Toxicity of Carbamazepine, Phenobarbital, Phenytoin, and Primidone

Epilepsin, 28(Suppl. 3):SSO-S58, 1987 Raven Press, Ltd., New York 0 International League Against Epilepsy

Dennis B. Smith, "Richard H. Mattson, ?Joyce A. Cramer, $Joseph F. Collins, §Robert A. Novelly, [[BruceCraft, and the Weterans Administration Epilepsy Cooperative Study Group

Good Samaritan Hospital & Medical Center, Portland Oregon; *Yale University School of Medicine, New Haven, Connecticut; $Veterans Administration Cooperative Studies Program, Perry Point, Maryland; *f§Veterans Administration Hospital, West Haven, Connecticut; and llNeurology Service, Veterans Administration Medical Center, Augusta, Georgia, U.S.A.

Summary: In 1985 a 5-year multicenter Veterans Admin- sociated with significantly lower incidences of intolerable istration Cooperative Study was completed that com- side effects than were primidone or phenobarbital. A be- pared the efficacy and toxicity of phenobarbital, car- havioral toxicity battery was performed whenever pos- bamazepine, phenytoin, and primidone in a double-blind sible prior to administration of any antiepileptic drug and prospective study design. A total of 622 patients, either at 1, 3, 6, and 12 months after initiation of monotherapy. previously untreated or undertreated, were entered into Significant differences in performance on all subtests of the study. Strict exclusion criteria limited confounding the battery were found between patients with epilepsy factors such as drug or alcohol abuse. Results showed and a control group matched by age, sex, and education. that each of the four drugs used as monotherapy were When the differential effects of all four drugs on behav- similarly effective in the treatment of generalized tonic- ioral toxicity were compared, few statistically significant clonic seizures, but carbamazepine was significantly differences emerged. However, carbamazepine consis- more effective in the treatment of complex partial sei- tently produced fewer adverse effects on tests of atten- zures as measured by 100% control. When the results for tionkoncentration and motor performance than did the all four drugs were combined, the data showed that ap- other three antiepileptic drugs. Key Words: Epilepsy- proximately 80% of the patients were adequately man- Seizures-Therapy -Anticonvulsants- Controlled clin- aged on monotherapy. Differences in toxicity were the ical trials-Drug-induced abnormalities-carbarnaze- most significant factor that discriminated between these pine -Phenobarbital-Phenytoin- Primidone. four drugs. Both carbamazepine and phenytoin were as-

In 1985 a nationwide multicenter study spon- generalized seizures (Mattson et al., 1985). The sored by the Veterans Administration (V.A.) was study was initiated in order to overcome short- completed. The study was a 5-year prospective, comings of previous studies of the comparative ef- double-blind trial comparing the efficacy and tox- ficacy and toxicity of antiepileptic drugs. icity of carbamazepine, phenobarbital, phenytoin, Coatsworth (1971) reviewed the literature from and primidone in adults with partial or secondarily 1920 through 1971 reporting on studies of the clinical efficacy of marketed antiepileptic drugs and found that a total of 250 case reports on the efficacy Address correspondence and reprint requests to Dr. D. B. and toxicity of these agents had been published. In Smith at Good Samaritan Hospital & Medical Center, 1015 N.W. 22nd Avenue, Suite 400, Portland, OR 97210, U.S.A. addition, results of 110 clinical trials were pub- 1 Includes Antonio V. Delgado-Escueta, Thomas R. Browne, lished during this same reporting period. Of the Peter D. Williamson, David M. Treirnan, James 0. McNamara, clinical trials, only three involved a single- or Charlotte B. McCutchen, Richard W. Homan, Wayne E. Crill, Michael F. Lubozynski, N. Paul Rosenthal, and Assa Mayers- double-blind protocol, 25 included an electroen- dorf. cephalogram (EEG) profile, five included a psycho-

5'50 V.A. COOPERATIVE STUDY RESULTS S5 1

TABLE 1. Studies comparing major antiepileptic drugs for partial and generalized tonic-clonic seizures (1966- 1979) Comparison of Relative efficacy Comments Study Goal drug toxicity of seizure control (duration; patients) White et al., 1966 Show PRM equals PHT All drugs equal All drugs equal 2 weeks; 20 or PB institutionalized patients Millichap and Aymat, Compare PRM with PHT more toxic than Both drugs equal 8 months; 20 outpatient I968 PHT PRM children; no drug levels Cereghino et al., 1974 Show CBZ equals PHT CBZ more toxic than All drugs equal 3 weeks; 45 or PB PHT or PB (equal institutionalized behavior for all drugs) patients Troupin et al., 1975 Show CBZ equals PHT All drugs equal Both drugs equal 4 weeks; 20 outpatients Cereghino et al., 1975 Best combination of Combinations including PHT + PB 3 weeks; 41 two or three drugs CBZ more toxic than combination is best institutionalized (PHT, PB, CBZ) PB and PHT patients Rodin et al., 1976 Best drug (PRM or CBZ more toxic than All combinations equal 3 months; 45 CBZ) to add to PHT PRM; CBZ fewer outpatients; if PB fails behavioral deficits behavioral testing; no than PRM drug levels Kosteljanetz et al., 1979 Show CBZ equals PHT All drugs equal, but Both drugs equal 6 months; 19 PHT more difficult to outpatients; triple- adjust dose blind

CBZ, carbamazepine; PB, phenobarbital; PHT, phenytoin; PRM, primidone. logical evaluation, and only three included any kind tial, and generalized tonic clonic seizures.” (Anti- of statistical analysis. From 1972 to 1979, results of epileptic drugs, 2nd ed. Raven Press, 1982.) an additional seven clinical trials were reported, These conflicting recommendations emphasize but because of a variety of shortcomings (failure to the problems that existed in interpreting the com- use a double-blind randomized protocol, failure to parative literature on antiepileptic drugs. It was compare all four drugs and to monitor serum con- against that background that the V.A. Cooperative centration of the drugs, inclusion of institutional- Study was conceived. The study was designed to ized refractory patients and patients with mixed overcome shortcomings in previous trials by (1) seizure types, and insufficient sample size for sta- studying newly diagnosed patients with epilepsy tistical analysis), these studies failed to reveal any not currently receiving antiepileptic drug therapy, clear differences in the clinical effectiveness or rel- (2) utilizing a homogeneous, noninstitutionalized ative toxicities of carbamazepine, phenobarbital, population that would be representative of a typical phenytoin, and primidone (Table 1). outpatient seizure population, (3) documenting eti- It was thus apparent that despite a wealth of liter- ology by EEG and imaging studies, (4) quantifying ature dating back nearly 50 years, no convincing the frequency and severity of seizures, (5) moni- scientific evidence had been developed that could toring serum concentrations of the antiepileptic be used to justify the selection of any single anti- drugs, (6) comparing all four drugs simultaneously epileptic drug for any specific seizure type (with the by a randomized, double-blind study design, (7) exception of generalized absence seizures) in using each drug singly as monotherapy until proven adults. Despite the inconclusiveness of these unsatisfactory, and (8) using a sufficiently large studies, there was no dearth of dogmatic state- population base to allow meaningful statistical anal- ments recommending one drug over another, as ex- ysis on completion. emplified by the following quotations from three These objectives were achieved, and the results commonly used textbooks in neurology. of that study for the first time provide a scientific For the treatment of major and focal epilepsy, “it is rationale for recommending specific antiepileptic usual to begin in an adult with phenobarbi- drugs for specific seizure types. In this paper, some tone. , . .” (Brain’s Diseases of the nervous of the results of that study are summarized. system, 8th ed. Oxford University Press, 1977.) “The treatment of grand ma1 seizures in adults METHODS always begins with phenytoin. . . .” (Modern practical neurology. Raven Press, 1977.) Enrolled in the study were 622 patients newly “Carbamazepine is a major antiepileptic drug for diagnosed as having simple or complex seizures or the treatment of complex partial, elementary par- primarily or secondarily generalized tonic-clonic

Epilepsia, Vol. 28, Suppl. 3, 1987 S52 D.B. SMiTH ET AL. seizures. All patients were previously untreated or Newly Diagnosed Epilepsy Patient undertreated. A double-blind prospective study design was used. Patients were followed for as long as Evaluated as Primary or Secondarily 6 years (average, 3 years) at 10 V.A. medical Generalized Tonic-Clonic centers (Augusta, GA; Boston, MA; Dallas, TX; Durham, NC; Los Angeles, CA; Minneapolis, MN; Complex or Elementary San Diego, CA; Seattle, WA; Sepulveda, CA; and Partial Seizures West Haven, CT, U.S.A.). The study population included men and women (age range, 18 to 70 I Randomize to Studv Druq I years). Nonveterans were also entered into the Does Patient Ha& study. Exclusion criteria were strict (Table 2). Inadequate Seizure Control? I Study design No Details of the study design have been described Yes r elsewhere (Delgado-Escueta et al., 1983; Mattson Does Patient Have et al., 1983; Cramer et al., 1983). Figure I schemati- Unacceptable Side Effects? cally summarizes the principles of the study design. failure,cross Yes Patients were randomly assigned to receive therapy to other single No with carbamazepine, phenobarbital, phenytoin, or drug if partial primidone. Inadequate seizure control with the ini- seizures or to drug failure two drugs if cross to tially assigned drug was counted as a drug failure tonic-clonic another drug and the patient was randomly assigned to a second study drug. (Patients with generalized tonic-clonic Continue on Study Drug seizures were randomly assigned to two study FIG. 1. Flow chart of study design drugs.) If drug treatment resulted in unacceptable toxicity, this was also counted as a drug failure, and the patient received a second single study drug by tion (i.e., the length of time that the patient con- random assignment. Patient compliance was moni- tinued to take the randomly assigned drug). Addi- tored throughout the study by measurement of an- tional assessments included rating scales of seizure tiepileptic drug levels and by pill counts at each frequency, systemic toxicity, and neurotoxicity that follow-up visit. were designed specifically for this study (Cramer et Details of drug administration and statistical con- al., 1983). These scales were designed to reflect the siderations are given by Mattson et a]. (1985). effects of seizures and drug toxicity on the quality of life of the patient. A composite score was com- Evaluations pleted by the evaluating physician at every clinic A primary end point for evaluating the efficacy visit. A score of 0-20 points corresponded with a and toxicity of the study drugs was patient reten- good clinical response, a score of 20-35 repre- TABLE 2. Exchions sented a satisfactory but suboptimal response, and a score of 35-49 represented a fair to poor but not 1. Under age 18 years clearly unacceptable outcome; a score of 50 or 2. Diagnosis of epilepsy not confirmed 3. Classification of specific seizure type not possible more represented an unacceptable outcome re- 4. Etiology of seizure is neoplastic, progressive degenerative, quiring a change in therapy (Cramer et al., 1983). metabolic, demyelinating, or active infection (a patient with a previously resected tumor that is not recurrent is Behavioral toxicity battery acceptable) 5. Patient has generalized seizures but not tonic-clonic (e.g., All patients had extensive neuropsychological only absence) evaluation before administration of any antiepi- 6. Patient is abuser of alcohol (24drinks per day) leptic drug and at 1, 3, 6, and 12 months after initia- 7. Patient is abuser of other drugs (e.g., narcotics) 8. Patient has progressive neurologic disorder tion of monotherapy with one of the study drugs. 9. Patient has serious medical disorder (unstable or requiring Seventy-five control subjects representing all 10 intervention or both) or active infection study sites were selected to match the epilepsy pa- 10. Patient is psychotic, grossly organic, or severely depressed 11. Patient’s intelligence quotient (WAIS) full-scale is less than tient group closely in terms of age, education, sex, 85 and absence of significant medical problems. This 12. Patient is a willful noncomplier group of controls was administered the same bat- 13. Repeated seizures may occur under a variety of circumstances related to acute medical disorders such as tery of neuropsychological tests on two occasions uremia, hypoglycemia, etc., and are not considered 30 days apart. epilepsy for the purpose of this study The tests selected were designed to reflect neuro-

Epilepsia, Vol. 28, Suppi. 3, 1987 V.A. COOPERATIVE STUDY RESULTS s53

psychological functioning in four major areas: (a) TABLE 3. Percent of patients by drug group free of all general intellect, (b) attention/concentration/mental seizures flexibility, (c) motor/motor manipulation, and (d) Seizure type emotionaUmood states. Intellectual ability was Generalized measured by the Wechsler Adult Intelligence Scale Tonic-clonic PTL (WAIS), a well-standardized measure of general Drug group (12 mos) (18 mos) cognitive function based on 11 verbal and non- Phenobarbital 58 33 verbal subtest activities. The behavioral test bat- Phenytoin 48 34 tery consisted of the following tests: Digit Symbol, Primidone 63 26 Carbamazepine 55 65* Digit Span, Critical Flicker Discrimination, Dis- All 55 42 criminative Reaction Time, Word Finding, Index Finger Tapping, Lafayette Pegboard, and Color * p < 0.05. Naming. Mood and emotional states were measured by the profile of mood states (POMS) (Lorr tonic-clonic seizures for 12 months was not signifi- et al., 1971). A complete description of this test cantly different among the study drugs: 55% of pa- battery and the procedure for its administration has tients were seizure free in the carbamazepine been previously detailed (Smith et al., 1986; No- group, 58% in the phenobarbital group, 48% in the velly et al., 1986). phenytoin group, and 63% in the primidone group. In contrast, control of partial seizures was signifi- RESULTS cantly better with carbamazepine than with the Seizure control other three study drugs after 18 months of follow- Of the several measures employed in this study up (p < 0.05) (Table 3). Partial seizures were con- for assessment of the comparative efficacy of the trolled in 65% of the patients receiving carbamaze- study drugs in seizure control, the most straightfor- pine as opposed to 33% taking phenobarbital, 34% ward and simple analysis was examination of the taking phenytoin, and 26% taking primidone. percentage of patients seizure free after 1 year. Another method of comparing the efficacy of Combined results for all four drugs showed that the these four drugs was to examine the number of pa- percentage of patients with fully controlled seizures tients on each drug remaining active in the study remaining in the study for 1 year after drug levels over time. A decreasing retention of patients over reached the therapeutic range was significantly time represents an inability of the drug to manage higher for patients whose primary seizure type was seizures, whether because of poor seizure control, tonic-clonic seizures than for those whose primary unacceptable toxicity, or both, Figure 2 shows the seizure type was either simple or complex partial number of patients successfully managed with each seizures (56% versus 39% of patients seizure-free). drug during 36 months of follow-up. This analysis For both groups, seizure control was similar at 6 combines data for all seizure types (53% of all the months and 1 year, with most seizure break- patients had more than one seizure type). Retention throughs occurring within the first 1 to 3 months rates were significantly better among patients re- (Homan et al., in press). Complete control of ceiving carbamazepine or phenytoin than among

-phenobarbital --- phenytoin .-.-. primidone ...... -- carbamazepine

-a&-=-. c C FIG. 2. Number of patients, by treatment group, remaining active in the study for 36 c -._.-._._._._._._._.-.-.-.-. months. (Reprinted by permission of The 5 40- New England Journal of Medicine.) $2 2

2o t 01 I I 1 I I I I I I I I 1 3 6 9 12 15 18 21 24 27 30 33 36 Months

Epileppsia, Vol. 28, Suppl. 3, 1987 s54 D. B. SMITH ET AL.

TABLE 4. Reasons for drug failure in the four treatment groups Treatment group CBZ PB PHT PRM All patients Reason (n = 101) (n = 101) (n = 110) (n = 109) (n = 421) Toxicity alone 12 19 18 36 85 Toxicity plus seizures 30 33 29 35 127 Seizures alone 3 4 1 3 11 Total failures 45 56 48 74 233

~~ CBZ, carbamazepine; PB, phenobarbital; PHT, phenytoin; PRM, primidone.

those receiving primidone (p < 0.001). Phenobar- drug (Fig. 3). Sixty percent of patients were ade- bital also had a better retention rate than primidone quately managed by monotherapy with the first (p < 0.05). Retention rates were better for carbam- drug they received by random assignment. Of those azepine or phenytoin than for phenobarbital, but who had treatment failure with the first drug and this difference did not reach statistical significance. who were switched to a second drug (either be- These results, however, do not indicate the cause of toxicity, lack of seizure control, or both), reasons for a patient failing to remain on therapy 55% were successfully treated with the second ran- with a single anticonvulsant (see Table 4). Very few domly assigned drug also used as monotherapy. In patients failed to remain on their original antiepi- this selected patient group, 20% could not be ade- leptic drug because of a lack of seizure control quately managed on monotherapy. Of this 20%, alone; most failures occurred because of both tox- 49% could be adequately managed by treatment icity and lack of seizure control. with two drugs in combination. Patients receiving primidone experienced the highest incidence of toxicity. Most of these patients Behavioral effects experienced unacceptable toxicity within the first 1 Of the 622 patients entered into the study, 618 to 3 months of therapy. The side effects included received the full battery of neuropsychological tests nausea, vomiting, dizziness, ataxia, and somno- at the initiation of the study. All 75 control subjects lence, and caused early discontinuation of the drug. completed the full battery of tests. Table 5 com- Patients tolerating the initiation of therapy with pri- pares the demographic characteristics of the con- midone, however, experienced no more toxicity trol subjects and the patients with epilepsy. with chronic therapy than did patients taking any of The two groups match closely for age, sex, and the other three antiepileptic drugs. education. Educational level was coded from 1 through 9, with the highest level of education Monotherapy (Ph.D.) represented by a 1 and the lowest level of Analysis of the percentage of patients remaining education (grade school or less) represented by a 9. in the study on monotherapy for 1 year showed The general intellectual ability, as measured by the clearly that the majority of patients remained ade- WAIS score, represents a normal range of function quately managed with monotherapy regardless of for patients with epilepsy, and the wide variability

60% managed

Failure with drug 1

FIG. 3. Percent of patients successfully managed on monotherapy with first or second drug randomly assigned (combined Failure with monotherapy data for all drug groups).

(20% no seizures in 1 yr) V.A. COOPERATIVE STUDY RESULTS s5.5

57 - 56 - Controls at 1 month 55 - .z ...... All patients pretreatment 54 - All patients at 1 month 53 - ./ '? ----- 52 - -.-.I ./ \. ,_ .__c ./.-.-._ ." 1. 51 - \. ./ 50 50 \-.- \ 49 - '.-,-.-. .* -_-. 48 - .. -. '. 6 47- '. 2 46 - b- 45 - 73 44 - =.=.'" ...... _\*.. .-2 43 - - %. 42 - *..\ -0 41 - 5 40 - % 5 39 - x, 38 - 37 - 36 I I I I I I I I I Finger Finger Lafayette Color Digit Flicker POMS-T POMS-A POMS-F Tapping Tapping Pegs Naming Symbol Dom. Total UonDom.

shown in the study population is no more than The control group differed significantly from the would be expected in such a large sample. WAIS epilepsy group on all but three behavioral measures scores were not obtained for the control subjects, at trial I and on all but two measures at trial 11. A but the two groups were comparable with regard to most important observation was that a practice ef- educational level, and the correlation between edu- fect appeared to occur for selected performances in cation and I.Q. is high (Smith et al., 1986). the control group. Virtually no practice effect was The mean performances of both the control seen between trial I and trial I1 in the epilepsy group and the patients with seizures for each of the group. These patients tended to show only slight behavioral measures are compared in Fig. 4. The improvement at trial I1 for Pegboard, Digit Symbol, results are shown for both the initial (trial I) and and Word Fluency tests. 30-day (trial 11) performances. For the epilepsy The total behavioral toxicity score was derived subjects, trial I performances were obtained under by combining the scores of the individual subtests a relatively drug-free condition, whereas at trial 11, transformed into weighted ordinal units of change all epilepsy patients had been receiving mono- based upon published norms. The higher the score therapy with one of the four anticonvulsant medica- on the behavioral toxicity battery, the more deterio- tions for approximately 30 days. ration from baseline performance is indicated. A negative score indicates an improvement in perfor- TABLE 5. Putient charucferistics mance compared with baseline predrug test scores. Characteristic Control Epilepsy When the effect of each antiepileptic drug on the total behavioral toxicity battery was examined at Number 75 618 Age (years, mean) 38.6 (-+13.1) 40.4 (t15.3) both 1 and 3 months, significant differences be- Range (years) 17-72 18-82 tween the drugs were apparent (Fig. 5). None of the Sex (males) 72% 85% groups showed an improvement in performance at Education (coded) 4.1 4.9 Education range 1-8 1-8 either 1 or 3 months when comparisons were made Verbal I.Q. 100.5 (? 14.5) with their predrug test scores. Carbamazepine pa- Range 55- I42 tients showed the least deterioration, achieving a Performance I.Q. 99.2 (t13.6) Range 52-134 total battery score of 16 at 1 month and 14 at 3 Full-scale I.Q. 100.0 (t13.5) months. Patients with primidone also showed rela- Range 52- 135 tively little deterioration at 1 and 3 months, having Adapted with permission from Smith et al., 1986. a total score of 14 at 1 month and 18 at 3 months. In Numbers in parentheses are standard deviation. contrast, patients taking phenobarbital or phe-

Epilepsia, Vol. 28, Suppl. 3, 1987 S56 D.B. SMITH ET AL.

0 1 month 0 3 months

FIG. 5. Total behavior toxicity battery scores by drug group at 1 and 3 months. The higher the score, the worse the performance.

Carbamazepine Phenobarbital Phenytoin Primidone nytoin showed significant deterioration in the total Digit Span, Critical Flicker Fusion, Word Finding, behavioral toxicity score at both 1 and 3 months (p and Discriminative Reaction Time. Overall, the < 0.001 at 1 month); scores for the phenobarbital, group taking carbamazepine showed the highest phenytoin, and primidone groups all showed deteri- number of “best” scores on subtests of the POMS oration in total scores at 3 months, while patients and on tests measuring attentionkoncentration. taking carbamazepine showed no such deteriora- Scores for the individual subtests showed no signif- tion at 3 months (p < 0.002 at 3 months). Drug icant pattern of effect over time among the four an- levels for all groups were within the well-accepted ticonvulsants. therapeutic range. The total behavioral toxicity score for phenobarbital was 25 and 20 at 1 and 3 months, respectively, while for phenytoin the DISCUSSION scores were 21 and 23. Differences between the four drugs for scores on In 1978, when planning for the V.A. Cooperative individual subtests of the behavioral toxicity bat- Study was completed, valproate had been approved tery occasionally attained statistical significance, only for use in generalized absence seizures, and but no consistent pattern emerged. For example, on phenobarbital, phenytoin, primidone, and carbam- the Word Finding subtest, patients in the carbamaz- azepine were still the most commonly used antiepi- epine group performed significantly better than pa- leptic drugs in the treatment of adults with epilepsy. tients taking phenytoin or phenobarbital (p < 0.008) All these drugs possess considerable efficacy in the at 3 months, and better than phenytoin patients treatment of generalized tonic-clonic seizures (p < 0.06) at 12 months. Scores for carbamazepine (White et al., 1966; Ives, 1951; Cereghino et al., patients were not significantly better than those for 1974), but selection of one of these drugs over an- patients taking any of the other drugs at either 1 or other as the initial therapy for any individual pa- 6 months. Analysis of covariance using pretreat- tient was influenced almost entirely by personal ment score as covariant was used to determine statistical significance in all cases. Although the patients taking carbamazepine had better scores on TABLE 6. Total number of best and worst scores an the Word Finding subtest at I and 6 months than subtests of the behavioral toxicity battery at I, 3, 6, and 12 months of therapy did the patients taking the other three drugs, this ~~ ~ ~~ difference did not reach statistical significance. Attention/ Results were further analyzed by counting the POMS Motor concentration number of times that a patient treatment group Best scores scored better than two or more other treatment Carbamazepine 11 4 10 groups with a confidence level of > 0.05. Similarly, Phenytoin 2 1 2 the number of worst scores for each subtest at each Primidone 9 3 1 rating period was determined using the same cri- Phenobarbital 2 I 0 teria. Worst scores Table 6 tabulates the numbers of best and worst Carbamazepine 0 0 1 scores (combined results from tests at I, 3, 6, and Phenytoin 13 4 0 Primidone 2 5 6 12 months) on subtests of the POMS, motor func- Phenobarbital 9 0 5 tion, and attention/concentration. Subtests of ~ ~~ motor function included Index Finger Tapping, La- “Best” and “worst” indicate scores that are significantly different from scores attained by two or more of the other groups fayette Pegboard, and Color Naming. Subtests of for the same subtest at the same rating period. attention/concentration included Digit Symbol, POMS, profile of mood states.

Epilepsia, Vol. 28, Suppl. 3, 1987 V.A. COOPERATIVE STUDY RESULTS s5 7

biases and by fragmentary and sometimes con- An interesting finding was that the pretreatment flicting information of relative efficacy from pub- intellectual performance for the epilepsy group was lished clinical trials (Tassinari and Roger, 1975). generally normal. The distribution of I.Q. scores The V.A. Cooperative Study has for the first time for the combined epilepsy sample was very similar provided some comparative data that can be used to that expected in the normal population on the to justify the selection of the antiepileptic drug that basis of Wechsler’s (1955, 1981) normative data. is most likely to be the most successful overall This finding is partly due to the strict exclusion cri- choice. teria, requiring a full-scale I.Q. of > 85. Nonethe- Mattson et al. (1985) have pointed out that the less, although the epilepsy group had a normal level results of this study “indicated that carbamazepine of intellectual performance, significant differences and phenytoin are most likely to be successful were consistently observed between patients with overall when used as the initial antiepileptic drug in epilepsy and the control group. This finding is par- adults with partial seizures or secondarily general- ticularly significant because the results of all pre- ized tonic-clonic seizures or both.” It was clear vious studies have been confounded to some extent that the choice between these two drugs required by drug effect (Dodrill and Troupin, 1974; Gesch- consideration of other factors. wind et al., 1980; Giordani et al., 1985; Klgve and Toxicity rather than seizure control proved to be Matthews, 1969; Thompson and Trimble, 1982). the most significant factor differentiating between The differential behavioral effects of the four an- the four study drugs. While the highest rate of tox- ticonvulsant drugs were clearly evident. Again, the icity was seen in the patients taking primidone, this importance of these findings is heightened when toxicity occurred almost exclusively within the first one considers the study design through which they 1 to 2 months of therapy. It is clear that if patients were obtained, i.e., not only was each drug com- tolerate the initiation of therapy with primidone, pared with each other drug used as monotherapy, they may do equally as well as patients taking any but, more importantly, results were compared of the other three drugs. Results from the behav- against a predrug baseline, each patient serving as ioral toxicity battery suggest that if patients tolerate his own control. the initial side effects of primidone, they perform Even though the results on individual subtests of somewhat better on some subtests of the POMS the behavioral toxicity battery were not dramati- and perform similar to patients taking phenytoin or cally revealing, the total battery score did show carbamazepine on subtests of motor function and that carbamazepine had fewer behavioral effects of attentionkoncentration. than did the other three drugs. None of the drug- Despite the fact that a number of previous treated groups showed any improvement in scores studies have pointed out the advantages with related to practice effects, in contrast to the control monotherapy, the results of this study are the first group. These results suggest that all four drugs from a double-blind prospective trial to confirm the studied have some adverse effect on motor skills, effectiveness of initiating therapy with a single cognition, and mood. Nonetheless, the best perfor- drug. In this study, approximately 80% of the pa- mances on subtests of the behavioral toxicity bat- tients were successfully managed for l year on tery were attained by patients taking carbamaze- monotherapy regardless of drug, i.e., they had ade- pine, and this difference was present at 1, 3, 6, and quate seizure control and lacked intolerable tox- 12 months of therapy. icity. These data show that while some patients On the basis of the results of this long-term pro- may require polytherapy for successful control, the spective comparison of phenobarbital, primidone, majority can be successfully managed with mono- carbamazepine, and phenytoin, the most successful therapy. This study also indicates that most pa- drugs when used as initial antiepileptic treatment in tients who are going to fail to respond to a single adults appear to be carbamazepine and phenytoin. antiepileptic drug will do so within the first 3 All four drugs demonstrated equal efficacy in pa- months of therapy. tients with secondarily generalized tonic-clonic One of the unique features of this study was the seizures, but carbamazepine was significantly more opportunity to prospectively examine the neurobe- effective than either primidone or phenobarbital in havioral effects of each of these four drugs and to controlling partial seizures. The behavioral effects compare the drug-treated state with a relatively of carbamazepine as measured by the neurotoxicity drug-naive state in the same patient. In addition, battery used in this study were less than any of the for the first time, patients with epilepsy who were other three drugs. This finding, coupled with the drug naive or undertreated could be compared with relatively equal antiepileptic efficacy of all four a control group matched for age, sex, and educa- drugs, indicates that carbamazepine may be the tion. preferred drug for initiation of therapy in some Epilepsia, Vol. 28, Suppl. 3, 1987 S58 D. B. SMITH ETAL.

adults with either generalized tonic-clonic sei- vs phenytoin. A controlled clinical trial in focal motor and zures, partial seizures, or both. However, in view generalized epilepsy. Arch Neurol I979;36:22-4. Lorr M, McNair DM, Droppleman LF. Manual: Projle ofmood of the variability of individual response, other states. San Diego: Educational and Industrial Testing Ser- factors may need to be considered when an antiepi- vice, 1971. leptic drug is selected. Mattson RH, Cramer JA, Delgado-Escueta AV, et al. A design for the prospective evaluation of the efficacy and toxicity of antiepileptic drugs in adults. Neurology (NY) 1983;33(suppl 1):14-25. REFERENCES Mattson RH, Crarner JA, Collins JF, et al. Comparison of carbamazepine, phenobarbital, phenytoin, and primidone in Cereghino JJ, Brock JT, VanMeter JC, Penry JK, Smith LD, partial and secondarily generalized tonic-clonic seizures. N Engl J Med 1985:313:145-51. White BG. Carbamazepine for epilepsy. A controlled prospective evaluation. Neurology 1974;24:401- 10. Milichap JG, Aymat F. Controlled evaluation of primidone and Cereghino JJ, Brock JT, VanMeter JC, Penry JK, Smith LD, diphenylhydantoin sodium. Comparative anticonvulsant effi- White BG. The efficacy of carbamazepine combinations in cacy and toxicity in children. JAMA 1968;203:738-9. epilepsy. Clin Pharmacol Ther ‘1975;18:733-41. Novelly RA, Schumartz MM, Mattson RH, Crarner JA. Behav- Coatsworth JJ. Studies in the clinical efficacy of markefed anti- ioral toxicity associated with antiepileptic drugs: Concepts epileptic drugs. NINDS Monograph No. 12. Washington, and methods of assessment. Epilepsia 1986;27:331-40. D.C.: U.S. Government Printing Office, 1971. Rodin EA, Rim CS, Kitano H, Lewis R, Rennick PM. A com- Cramer JA, Smith DB, Mattson RH, et al. A method of quantifi- parison of the effectiveness of primidone versus carbamaze- cation for the evaluation of antiepileptic drug therapy. Neu- pine in epileptic outpatients. J Nerv Ment Dis 1976;163:41-6. rology (NY)1983;33(suppl 1):26-37. Smith DB, Craft BR, Collins J, Mattson RH, Cramer JA, and the Delgado-Escueta AV, Mattson RH, Smith DB, Cramer JA, V.A. Cooperative Study Group 118. Behavioral character- Collins JF. Principles in designing clinical trials for antiepi- istics of epilepsy patients compared with normal controls. leptic drugs. Neurology (NY) 1983;33(suppl 1):8-13. Epilepsiu 1986;27:760-8. Dodrill CB, Troupin AS. Effects of repeated administration of a Tassinari CA. Roger J. Prognosis and therapy of complex partial comprehensive neuropsychological battery among chronic seizures with barbiturates, hydantoins, and other drugs. Adv epileptics. J Nerv Ment Dis 1974;161: 185-90. Neurol I9753 1 :201- 19. Geschwind N, Shader RI, Bear D, North B, Levin K, Chetham Thompson PJ, Trirnble MR. Anticonvulsant drugs and cognitive D. Behavioral changes with temporal lobe epilepsy: Assess- functions. Epilepsiu 1982;23:531-44. ment and treatment. J Clin Psychiatr 1980;41:89. Giordani B, Berent S, Sackellares CJ, et al. Intelligence test per- Troupin AS, Green JR, Halpern LM. Carbamazepine (Tegretol) formance of patients with partial and generalized seizures. as an anticonvulsant. A controlled double-blind comparison Epilepsia 1 985;26:37-42. of diphenylhydantoin (Dilantin). Actu Neurol Scand Homan RW, Miller B, and the V.A. Cooperative Study Group. 1975;75(suppl):13-26. Causes of treatment failure with antiepileptic drugs vary over Wechsler D. Wechsler adult intelligence scale. New York: The time. Neurology (in press). Psychological Corp., 1955. Ives ER. Comparison of efficacy of various drugs in treatment of Wechsler D. Wechsler adult intelligence scale-revised. New epilepsy. JAMA 1951;147:1332-5. York: The Psychological Corp., 1981. Kl@e H, Matthews CGL. Neuropsychological evaluation of the White PT, Plott D, Norton J. Relative anticonvulsant potency of epileptic patient. Wisc Med J 1969;68:296-301. primidone: A double-blind comparison. Arch Neurol Kosteljanetz M, Christiansen J, Dam AM, et al. Carbamazepine 1966;14:31-5.

Epilepsiu, Vol. 28. Suppl. 3, 1987