62ournal of Neurology, Neurosurgery, and Psychiatry 1994;57:682-687 PAPERS A multicentre comparative trial of sodium valproate and carbamazepine in adult onset epilepsy A Richens, D L W Davidson, N E F Cartlidge, D J Easter on behalf of the Adult EPITEG Collaborative Group Abstract The long term efficacy and safety of sodium valproate and carbamazepine in adult outpatients with newly diagnosed primary generalised or partial and secon- darily generalised seizures were com- pared in a randomised, open, multi- centre study at 22 neurology outpatient clinics. Patients were randomised to oral sodium valproate (Epilim EC enteric coated 200 mg tablets twice daily, n = 149) or oral carbamazepine (100 mg twice daily increasing to 200 mg twice daily in week 2, n = 151) and followed up for three years. If clinically necessary, dosages were regularly increased until seizures were controlled or toxicity developed. Sodium valproate and carbamazepine controlled both primary generalised and partial seizures equally effectively over- all. Significantly more patients on sodium valproate than carbamazepine (126/140 (90%) v 105/141 (75%), p = 0.001) remained on randomised treatment for at least six months. Skin rashes occurred significantly more often in carba- mazepine recipients than in sodium val- proate recipients (11.2% v 1-7%, p < 0.05) and carbamazepine was associated with a higher withdrawal rate because of adverse events (15% v 5% on sodium val- proate) in the first six months of treat- ment. There was no difference between the drugs in the rate of withdrawal because of poor seizure control at any stage, regardless of seizure type. At the end of the three year trial period, over 70% of the available patients were still on randomised treatment or had recently stopped treatment after achieving full seizure control. Sodium valproate and carbamazepine were both associated with a high degree of overall seizure control regardless of seizure type and both have good long term tolerability in adult patients with newly diagnosed epilepsy. Recommendations are made for a higher initial dosage regime for sodium val- proate in partial seizures. (7 Neurol Neurosurg Psychiatry 1994;57:682-687) After a period when anticonvulsant polyther- apy had been widely used in the treatment of epilepsy, it became apparent that a single drug was capable of controlling as many as 75% of adult outpatients with epilepsy.' Several studies had indicated that seizure control was affected by seizure type. Carbamazepine was reported to be the treatment of choice for partial seizures and sodium valproate for absences.23 Tumbull et al compared phenytoin and sodium valproate monotherapy in 140 pre- viously untreated adults for at least two years.4 Both drugs were found to be effective in the control of primary generalised and partial seizures, with a better response in patients with generalised attacks. Callaghan, et al com- pared carbamazepine, phenytoin, and sodium valproate in 181 previously untreated patients, and concluded that there was little difference in efficacy between the three drugs for either primary generalised or partial seizures.' We undertook this multicentre prospective trial to compare the long term efficacy and tolerability of sodium valproate and carba- mazepine in untreated newly diagnosed adults with primary generalised or partial seizures with or without generalisation presenting to neurological clinics. The relation between dose and seizure control was also investigated. Patients and methods Three hundred patients with newly diagnosed idiopathic generalised tonic clonic seizures or partial seizures with or without generalisation were recruited for this three year study at 22 outpatient clinics in the United Kingdom. To be eligible, patients (either sex) had to be over 16 years of age and to have had at least two generalised tonic-clonic or partial seizures with or without generalisation in the previous six months. Patients with both pri- mary generalised tonic-clonic and other gen- eralised seizures (for example, myoclonic or typical absence seizures) were included, but the efficacy analyses were based solely on the tonic-clonic seizure data. Patients with an accompanying renal, hepatic, or CNS disorder, abnormal liver function tests, low platelet count, or other Department of Pharmacology and Therapeutics, University of Wales College of Medicine, Heath Park, Cardiff CF4 4XN, UK A Richens Dundee Royal Infirmary, Dundee DD1 9ND, UK D L W Davidson The Royal Victoria Infirmary, Queen Victoria Road, Newcastle upon Tyne NE1 4LP, UK N E F Cartlidge Medical Affairs Department, Sanofi Winthrop Ltd, Onslow Street, Guildford GUI 4YS, UK D J Easter Correspondence to: Dr Easter Received 5 July 1993 and in final revised form 4 October 1993. Accepted 12 October 1993 682 on October 20, 2020 by guest. Protected by copyright. http://jnnp.bmj.com/ J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.57.6.682 on 1 June 1994. Downloaded from
Transcript
62ournal ofNeurology, Neurosurgery, and Psychiatry 1994;57:682-687
PAPERS
A multicentre comparative trial of sodiumvalproate and carbamazepine in adult onsetepilepsy
A Richens, D LW Davidson, N E F Cartlidge, D J Easter on behalf of the Adult EPITEGCollaborative Group
AbstractThe long term efficacy and safety ofsodium valproate and carbamazepine inadult outpatients with newly diagnosedprimary generalised or partial and secon-darily generalised seizures were com-pared in a randomised, open, multi-centre study at 22 neurology outpatientclinics. Patients were randomised to oralsodium valproate (Epilim EC entericcoated 200 mg tablets twice daily, n =149) or oral carbamazepine (100 mg twicedaily increasing to 200 mg twice daily inweek 2, n = 151) and followed up for threeyears. If clinically necessary, dosageswere regularly increased until seizureswere controlled or toxicity developed.Sodium valproate and carbamazepinecontrolled both primary generalised andpartial seizures equally effectively over-all. Significantly more patients onsodium valproate than carbamazepine(126/140 (90%) v 105/141 (75%), p = 0.001)remained on randomised treatment for atleast six months. Skin rashes occurredsignificantly more often in carba-mazepine recipients than in sodium val-proate recipients (11.2% v 1-7%, p < 0.05)and carbamazepine was associated with ahigher withdrawal rate because ofadverse events (15% v 5% on sodium val-proate) in the first six months of treat-ment. There was no difference betweenthe drugs in the rate of withdrawalbecause of poor seizure control at anystage, regardless of seizure type. At theend of the three year trial period, over70% of the available patients were still onrandomised treatment or had recentlystopped treatment after achieving fullseizure control. Sodium valproate andcarbamazepine were both associated witha high degree of overall seizure controlregardless of seizure type and both havegood long term tolerability in adultpatients with newly diagnosed epilepsy.Recommendations are made for a higherinitial dosage regime for sodium val-proate in partial seizures.
(7 Neurol Neurosurg Psychiatry 1994;57:682-687)
After a period when anticonvulsant polyther-apy had been widely used in the treatment ofepilepsy, it became apparent that a single drugwas capable of controlling as many as 75% ofadult outpatients with epilepsy.' Several studieshad indicated that seizure control was affectedby seizure type. Carbamazepine was reportedto be the treatment of choice for partialseizures and sodium valproate for absences.23Tumbull et al compared phenytoin andsodium valproate monotherapy in 140 pre-viously untreated adults for at least two years.4Both drugs were found to be effective in thecontrol of primary generalised and partialseizures, with a better response in patientswith generalised attacks. Callaghan, et al com-pared carbamazepine, phenytoin, and sodiumvalproate in 181 previously untreatedpatients, and concluded that there was littledifference in efficacy between the three drugsfor either primary generalised or partialseizures.'We undertook this multicentre prospective
trial to compare the long term efficacy andtolerability of sodium valproate and carba-mazepine in untreated newly diagnosed adultswith primary generalised or partial seizureswith or without generalisation presenting toneurological clinics. The relation betweendose and seizure control was also investigated.
Patients and methodsThree hundred patients with newly diagnosedidiopathic generalised tonic clonic seizures orpartial seizures with or without generalisationwere recruited for this three year study at 22outpatient clinics in the United Kingdom.To be eligible, patients (either sex) had to
be over 16 years of age and to have had atleast two generalised tonic-clonic or partialseizures with or without generalisation in theprevious six months. Patients with both pri-mary generalised tonic-clonic and other gen-eralised seizures (for example, myoclonic ortypical absence seizures) were included, butthe efficacy analyses were based solely on thetonic-clonic seizure data.
Patients with an accompanying renal,hepatic, or CNS disorder, abnormal liverfunction tests, low platelet count, or other
Department ofPharmacology andTherapeutics,University ofWalesCollege ofMedicine,Heath Park, CardiffCF4 4XN, UKA RichensDundee RoyalInfirmary, DundeeDD1 9ND, UKD LW DavidsonThe Royal VictoriaInfirmary, QueenVictoria Road,Newcastle upon TyneNE1 4LP, UKN E F CartlidgeMedical AffairsDepartment, SanofiWinthrop Ltd, OnslowStreet, GuildfordGUI 4YS, UKD J EasterCorrespondence to:Dr EasterReceived 5 July 1993and in final revised form4 October 1993.Accepted 12 October 1993
682 on O
ctober 20, 2020 by guest. Protected by copyright.
http://jnnp.bmj.com
/J N
eurol Neurosurg P
sychiatry: first published as 10.1136/jnnp.57.6.682 on 1 June 1994. Dow
A multicentre comparative trial ofsodium valproate and carbamazepine in adult onset epilepsy
blood dyscrasia on initial assessment wereexcluded. Patients with absences ormyoclonic jerks alone were excluded, as werefemale patients who were pregnant, lactating,or planning to become pregnant during thestudy. No concomitant anticonvulsants werepermitted during the study; any patient whorequired additional antiepileptic medicationduring the study was withdrawn. All patientsgave written or verbal informed consentbefore participating in the study, which wasconducted in accordance with the Declarationof Helsinki (1964) as amended in Tokyo(1975) and which was approved by ethicscommittees local to each clinic.On entry, patients were stratified by age,
sex, seizure type, and centre and randomisedby telephone to either sodium valproate (n =149) or carbamazepine (n = 151) by a com-puterised minimisation programme based inthe Medical Department at Sanofi WinthropLtd.The recommended initial dose for sodium
valproate (Epilim enteric coated tablets) was200 mg orally twice daily; if clinically neces-sary, the dose could be increased by 200 mgtwice daily at weekly intervals until seizureswere controlled or until toxic symptomsappeared, up to a maximum daily dose of2400 mg. The recommended initial dosage ofcarbamazepine was 100 mg orally twice daily,increasing in the second week to 200 mg twicedaily. Further increases of 200 mg/day at fort-nightly intervals were permitted as required tocontrol seizures or until signs of toxicityappeared, up to a maximum of 1600 mg/day.Patients in whom treatment was discontinuedbecause of adverse events or whose seizurecontrol was not satisfactory were regarded astreatment failures, were switched to thealternative drug, and were followed up forthe rest of the three year study. A detailedanalysis of their progress after failure onrandomised treatment is outside the scope ofthis report.
EFFICACY AND SAFETY ASSESSMENTSPatients were assessed at one month, then atthree-monthly intervals for the first two yearsof the trial and at about six-monthly intervalsthereafter (maximum duration of follow upseven years). An EEG taken at entry was usedto aid classification of seizure type.6 Seizurecontrol was assessed from a daily seizure diarycompleted by the patient. Serum drug con-centrations were determined after threemonths' and 12 months' treatment, and atother times at the discretion of the investiga-tor to monitor patient compliance. Routinehaematology and liver function tests were car-ried out at entry, at three months, and then atthe discretion of the investigator. To monitordrug safety, at each visit patients were askedwhether they had been affected by treatmentin any way, and any adverse events reportedwere recorded. Events were graded as severe ifthey warranted discontinuation of treatment.Treatment was continued if events weregraded moderate or mild. All concomitantmedication was recorded at each visit.
STATISTICAL ANALYSISThree hundred patients were recruited to thisstudy, which was designed to detect an overalldifference in seizure control of 15%-20%between treatments with a = 0-05 and ,3 =0-80. All statistical tests were two tailed at the5% significance level.
Seizure data were analysed on an intentionto treat basis including all randomisedpatients with a diagnosis of epilepsy whoreceived study treatment and who had ade-quate follow up data on treatment (at leastone visit). Patients who failed to take theirmedication regularly or who showed poorcompliance with the study medication wereincluded in all efficacy analyses as if they hadremained on treatment.
For the remission analysis, the seizure datarecorded at each visit was used to determinethe time to achieve six, 12, and 24 months offreedom from seizures for each patient overthe 36 month study period. Cumulativeremission distributions were computed foreach treatment group and expressed as unad-justed Kaplan-Meier plots, censored forsevere adverse events, deaths, and dropouts.
Efficacy was also assessed by a retentionanalysis comparing the number of patientswho remained in the study on randomisedtreatment. The Kaplan-Meier method wasused to compute cumulative distributions foreach treatment group, censoring for deathsand dropouts only.The effects of all relevant prognostic factors
(for example, treatment, seizure type, pre-treatment seizure frequency, age, sex, etc) onthe remission and retention analyses wereanalysed by Cox regression models.7
Safety was assessed in all patients who wereon randomised treatment, or who hadchanged from one treatment to the other.Fisher's exact test was used to compareadverse event incidence between treatmentsfor the randomised treatment period only.The incidence of treatment failure due to
severe adverse events or poor seizure controlon each treatment was also compared withFisher's exact test.
ResultsSTUDY POPULATIONA total of 140 of 149 patients randomised tosodium valproate and 141 of 151 patients ran-domised to carbamazepine could be evaluatedon an intention to treat basis. The distributionof age, sex, weight, EEG, medical history,intellectual status, seizure type, and seizurefrequency was similar in each treatmentgroup, supporting the preliminary demo-graphic findings of this study reported pre-viously.8 The mean age of sodium valproateand carbamazepine recipients was 33 yearsand 34 years respectively. The seizure type atentry was classified on clinical grounds as"primary generalised" in 69 patients in eachtreatment group and as "partial with or with-out generalisation" in 71 sodium valproaterecipients and in 72 carbamazepine recipients.The EEGs done at or near entry were
683 on O
ctober 20, 2020 by guest. Protected by copyright.
http://jnnp.bmj.com
/J N
eurol Neurosurg P
sychiatry: first published as 10.1136/jnnp.57.6.682 on 1 June 1994. Dow
reported by the local hospital services as nor-mal in 32% of patients, with 32% showingfocal features and 11% generalised spike waveactivity. Subsequently, an independent assess-ment of all available EEG records (by Dr CBinnie, Maudsley Hospital, London) indi-cated that a small proportion (about 20%) ofthe patients with "primary generalised"seizures had focal features that could haveindicated a partial origin to the seizures, butin the absence of clinical evidence of focalonset, the original investigator's assessment ofseizure type at entry was used for all efficacyanalyses. The reasons for exclusion of the 19patients from the efficacy analyses were nodata after randomisation (seven); incompletedata on treatment (seven); violation of entrycriteria (not treated) (two); not epilepsy(two); and died in train incident (presumedsuicide) (one).
TREATMENT FAILURESAdverse events led to withdrawal of treatmentin 7/140 (5%) sodium valproate recipientsand in 22/141 (15%) carbamazepine recipi-ents in the first six months of treatment. Poorseizure control resulted in treatment with-drawal in 4/140 (3%) of sodium valproaterecipients and 3/141 (2%) of carbamazepinerecipients over this period; all these patientshad frequent partial seizures at entry.Significantly more sodium valproate recipi-ents (90%) remained in the study at sixmonths compared with carbamazepine recipi-ents (75%, p = 0-001).
Over the period from six months to threeyears, treatment failed because of poor seizurecontrol in 15/126 (12%) of sodium valproaterecipients and in 12/105 (12%) of carba-mazepine recipients, and because of severeadverse events in 5% of sodium valproaterecipients and 3% of carbamazepine recipi-ents, respectively. Poor seizure control com-bined with adverse events at low drug doseswas reported for two (2%) additional patientson sodium valproate.
EFFICACY ANALYSISAbout half of the patients who were able totolerate the trial drugs remained completelyfree of seizures for the first six months ontreatment (fig 1); most of the remainderachieved at least six months' remission before
100sn
*° 8070
O 60._ 50
40I'D 30r-
a. 2010
- Sodium valproate--- Carbamazepine
0.0
0 E
to ~ 4
0 6 12 18 24 30 36
Months on treatment
Figure 1 Remission analysis: percentage of all patientsfree ofseizures for 6, 12, and 24 months.
Figure 2 Remission analysis: percentage ofpatientsfree ofseizures for 12 months: patients with primary generalisedsezzures.
the end of the three year trial period. By theend of three years, 80% of the patients avail-able for assessment had achieved 12 months'remission and 60% had achieved 24 months'remission. A similar proportion of patients ineach treatment group had achieved six, 12,and 24 months' remission by three years.
In patients with primary generalisedseizures who tolerated the trial drugs, therewas no difference between valproate and car-bamazepine in the 12 month remission rates(fig 2). In partial seizures, similar overall 12month remission rates were achieved for eachtreatment (sodium valproate 72%, carba-mazepine 76%) by the end of the three yeartrial period (fig 3).The full Cox model analysis, incorporating
all the prognostic covariates showed that thedifferences in remission rates between thedrugs were minimal, with a marginal and sta-tistically insignificant advantage for carba-mazepine compared with sodium valproate inthe six month (risk ratio (RR) 0-92, 95% con-fidence interval (95% CI) 0-69-1 24), 12month (RR = 0-89, 95% CI 0-65-1-21) and24 month (RR= 0-83, 95% CI 0-59-1d19)remission data. In strong contrast to this, thenumber of seizures during the six monthsbefore treatment had a significant effect on sixmonth remission (RR = 0'52, 95% CI0-28-0-97) and a similar effect on 12 month(RR= 0-52, 95% CI 0-27-1 01) and 24month remission (RR= 0.49, 95% CI0-23-1 06), indicating a considerably higherrisk of failure on either drug for patients withfour or more seizures before entry.
A multicentre comparative trial ofsodium valproate and carbamazepine in adult onset epilepsy
In these primary analyses, patients whowere intolerant of the trial drugs were cen-
sored from the analysis from the time of with-drawal. Such patients could, however, beregarded clinically as treatment failures. In a
subsidiary analysis retaining these patients as
treatment failures, essentially similar resultswere obtained to those found in the primaryanalysis: Overall, patients on sodium val-proate achieved higher six month (79% v
72%, RR = 1-21, 95% CI 0-91-1-61) and 12month (69% v 64%, RR= 1 14, 95% CI0-84-1 55) remission rates but similar 24month remission rates (52% v 51 %, RR =1-01, 95% CI 0.71-1.43) to patients on car-
bamazepine: Patients with primary gener-alised seizures achieved higher 12 monthremission rates on sodium valproate (76%)than on carbamazepine (62% RR = 1-41,95% CI 0-91-2-18), whereas patients withpartial seizures had similar overall remissionrates (62% v 66%, RR= 0-84, 95% CI0Q54-1430) on either drug. None of these dif-ferences was statistically significant--by Coxmodel analysis.Cox model analysis of the proportion of
patients remaining on treatment (retentionanalysis) showed that patients with primarygeneralised seizures who received sodiumvalproate were significantly more likely toremain on randomised treatment than were
carbamazepine recipients (RR = 0-34, CI0-16-0.72), with a non-significant advantagefor carbamazepine in partial seizures (RR =
1-60, 95% CI 0-84-3 05). There was a signifi-cant benefit (RR = 0O41, 95% CI 0.17-0.96)for sodium valproate over carbamazepine inyounger patients (under 25 years old) and inpatients who had had less than four seizures inthe six months before the study (RR = 0.35,95% CI 0.16-0.80).
Using the methodology given by Kay,9 a
competing risks analysis was done by separat-ing the Cox model into two components-adverse events and poor seizure control (thecombined category was ignored because ofsmall numbers). For poor seizure controlthere was no difference between treatments(RR = 1-09, 95% CI 0-56-2.11). For adverseevents however, there was a significant treat-ment effect in favour of valproate (RR = 0-48,95% CI 0 25-0.93).
MEAN DRUG DOSE LEVELSPatients on sodium valproate reached a mean
dose of 924 mg/day at month 24, those withpartial seizures requiring higher doses (mean1066 mg/day) than those with primary gener-alised seizures (mean 821 mg/day). Patientson carbamazepine reached a mean dose of516 mg/day at two years and there was no dif-ference between patients with primary gener-alised seizures (mean 520 mg/day) and thosewith partial seizures (mean 511 mg/day).
ADVERSE EVENT ANALYSISThe adverse event analysis (which includespatients changed from one drug to the other) isbased on reports from 174 patients treatedwith sodium valproate (165 for more than
three months) and from 178 patients treatedwith carbamazepine (147 for more than threemonths). An interim report on part of thesedata has been presented previously.'0The overall prevalence of adverse events
was similar in the two groups, 86 of 174(49.4%) sodium valproate recipients and 87of 178 (48 9%) carbamazepine recipientsreporting events at some time in the three yeartrial period.The table shows the adverse events that
were reported by four or more patients.Rashes occurred significantly more often oncarbamazepine (11 2% of patients) than onsodium valproate (1 7%, p < 0 05) and forcarbamazepine, were predominantly reportedwithin the first three months of treatment.Other drug related adverse events with carba-mazepine seemed to be dizziness (6&7% v2 9%), headaches (6-1% v 3A4%), and ataxia(2.2% v 0%) (all non-significant).
Significantly more patients on sodium val-proate reported weight gain (12- 1% v 1 1 %, p< 0.05), usually after at least three months ontreatment. Other drug related adverse eventswith valproate seemed to be tremor (5.2% v1.7%), alopecia (2.9% v 06%), and appetiteincrease (2-3% v 0%) (all non-significant).
Treatment was withdrawn because ofadverse events in 9*3% of patients on val-proate (no predominant event) and 17-7% ofpatients on carbamazepine (predominantlyrashes with or without other events). In thefirst six months on treatment, rash alone or incombination accounted for 13 (59%) of the22 (15.6%) withdrawals on carbamazepine.There was no clear pattern of adverse eventsfor the seven patients (5%) withdrawn fromvalproate treatment.
In the subsequent 30 months on treatment,six patients (4.3%) on valproate and threepatients (2-2%) on carbamazepine were with-drawn after adverse events. Late events onvalproate were (a) rash and oedema, (b)alopecia, (c) thrombocytopenia, (d) tremor,
Table Adverse events reported byfour or more patients atany time during treatment
*p = 0-05; **p = 0-01 between treatments, Fisher's exact test.Data from 174 patients on valproate (165 for at least threemonths) and 178 patients on carbamazepine (147 for at leastthree months).
685
on October 20, 2020 by guest. P
rotected by copyright.http://jnnp.bm
j.com/
J Neurol N
eurosurg Psychiatry: first published as 10.1136/jnnp.57.6.682 on 1 June 1994. D
extrapyramidal symptoms, raised liverenzymes, and heart failure, (e) somnolenceand fatigue, and (f) fatigue. Late events withcarbamazepine were (a) menstrual disorderand fatigue, (b) psoriatric rash, and (c) rash(unspecified).The number of pregnancies on treatment
reported during the study was higher amongcarbamazepine recipients (3.4%) than amongsodium valproate recipients (0 6%). Includingthose patients who had stopped their treat-ment before or shortly after becoming preg-nant, eight patients randomised to sodiumvalproate and 13 to carbamazepine becamepregnant during the three year trial period.One patient on valproate elected for termina-tion of pregnancy (no abnormalities reported)and one had a seizure during delivery. Therewere two adverse events reported for patientsrandomised to carbamazepine: one case offoetal malformation (pyloric stenosis) and oneof premature birth. All other patients had nor-mal pregnancies and deliveries.
Five deaths were reported during the trial,none of which was considered by the investi-gator to be drug related. Two deaths were inthe valproate group (man aged 58 from car-diovascular disease and a man aged 31 in atrain incident (presumed suicide) shortly afterrandomisation (no data on treatment)). Twodeaths were in patients on carbamazepine (awoman aged 66 from a heart attack and a managed 58 from cancer of the bronchus). Thefifth patient (a woman aged 34) died of anastrocytoma nine months after withdrawalfrom the trial (on carbamazepine).No clinically significant changes in haema-
tology or biochemistry were noted on eithertreatment. White cell counts seemed to bedepressed (by 8%) on carbamazepine treat-ment, but the wide scatter in pretreatmentlevels limited any detailed interpretation ofthese data. Mean platelet counts wereunchanged during treatment with either val-proate or carbamazepine. One patient with anabnormally low platelet count (below150 x 109) before treatment, received severalmonths of treatment with valproate at dosesup to 2000 mg/day before falling plateletcounts prompted drug withdrawal (with fullrecovery).
Alanine aminotransferase (ALT, SGPT)and aspartate aminotransferase (AST,SGOT) activities showed minor and insignifi-cant changes during treatment with eitherdrug. Alkaline phosphatase activity wasextremely variable, particularly in youngerpatients with residual maturational effects,but absolute change, percentage change, tran-sition, and frequency analyses indicated thatcarbamazepine treatment was associated withincreased alkaline phosphatase (mean change0-3 months + 15% on carbamazepine v + 1%on valproate).
DiscussionSodium valproate and carbamazepine are wellestablished as safe and effective drugs for thetreatment of all types of epilepsy (except thatcarbamazepine is not recommended in certain
types of generalised seizures"). Until recentlythere has been a widely held view, however,(mainly from anecdotal evidence) that val-proate was best for primary generalisedseizures, and carbamazepine for partialseizures. Our main objective in this trial wasto compare directly these two drugs whenused in the routine treatment of newly diag-nosed adult onset epilepsy, both generalisedand partial, in busy hospital neurology clinics.The analysis of treatment failures showed
that in the first six months of treatment 22 outof 141 patients on carbamazepine treatment(15-6%) were withdrawn because of severeadverse events (mainly rashes in the first fewweeks of treatment). Thereafter, there waslittle difference in failure rates between thetreatments, whether for adverse events, poorseizure control, or both. At the end of thethree year trial period, over 70% of the avail-able patients were still on randomised treat-ment or had recently stopped treatment afterachieving full seizure control.The remission analyses showed that both
sodium valproate and carbamazepineachieved a high degree of seizure control, 80%of patients available for assessment having hadat least 12 months' freedom from seizures,and 60% at least two years' freedom over thethree year trial period.
Although there were no differences in over-all control, patients with partial seizures onsodium valproate seemed to take longer toachieve remission. This was most likely due tothe low initial regime for sodium valproateand the protocol recommendation that thedose be increased only if needed to controlseizures. Patients randomised to sodium val-proate started at 200 mg twice daily (currentdata sheet recommendation 600 mg/day).Analysis of mean daily doses for valproateindicated a target dose of 821 mg/day for pri-mary generalised seizures, but a higher dose(1066 mg/day) to achieve similar control withpartial seizures. For carbamazepine, there waslittle difference in mean daily dose for primarygeneralised seizures (520 mg/day) and for par-tial seizures (511 mg/day).
These results indicate the need for a higherstarting dose for sodium valproate in partialseizures than that used in this study. The pre-sent data sheet recommendation is 600mg/day in divided doses rising as necessary to1000-2000 mg/day. The results presentedhere suggest an increase to 800-1000 mg/daysoon after the start of treatment for patientswith partial seizures regardless of response toensure full early seizure control for both pri-mary generalised and partial seizures inadults.The results presented on efficacy are simi-
lar to those reported by Reynolds and his col-leagues'2 who compared phenobarbitone,phenytoin, carbamazepine, and sodium val-proate in adults with newly diagnosedepilepsy. In their study, all four drugsachieved similarly high degrees of seizure con-trol, and there were no significant differencesin efficacy between the drugs regardless ofseizure type.
686
on October 20, 2020 by guest. P
rotected by copyright.http://jnnp.bm
j.com/
J Neurol N
eurosurg Psychiatry: first published as 10.1136/jnnp.57.6.682 on 1 June 1994. D
A multicentre comparative trial ofsodium valproate and carbamazepine in adult onset epilepsy
Comparable efficacy of sodium valproateand carbamazepine in both primary gener-alised and partial seizures has also been shownin paediatric patients. Reynolds and col-leagues'3 compared phenytoin, carba-mazepine, and sodium valproate in childrenwith newly diagnosed epilepsy and showedthat all drugs achieved similarly high degreesof seizure control both overall and by seizuretype and that there were no significant differ-ences in efficacy between the drugs regardlessof seizure type. In a paediatric version of thepresent trial, Verity et al'4 showed that val-proate and carbamazepine were equally effec-tive in achieving high levels of seizure controlin both primary generalised and partialseizures. In that trial, patients on valproateseemed to achieve better long term remissionthan those on carbamazepine, regardless ofseizure type, although the differences were notstatistically significant.Our results differ from those of Mattson et
al'5 who reported that whereas there was nodifference with secondary generalised tonic-clonic seizures, carbamazepine provided sig-nificantly better seizure control than sodiumvalproate in adult patients with predominantlycomplex partial seizures. Their study was,however, carried out in a different andrestricted patient group that was not represen-tative of the range of patients with epilepsypresenting to neurology clinics; the patientswere predominantly male war veterans, 31%of whom had trauma related epilepsy, with amean age of 47 years. The study also had aconsiderably higher dropout rate (27% in thefirst 12 months v 5% in the present study).Furthermore, whereas over 98% of ourpatients had newly diagnosed epilepsy (noneon treatment), in the study by Mattson et al,'550% of the patients had already been on med-ication and 26% were still on medicationshortly before entry.The adverse event profiles for sodium val-
proate and carbamazepine reported here aresimilar to those reported by these investigatorsand others,'6 although the actual incidence ofspecific adverse effects seems to be verydependent on the treatment policy used (forexample, minimal effective dose (as usedhere) v target serum drug concentrations).
Treatment was withdrawn because ofadverse effects in 18% of the patients on car-bamazepine (50% due to early rashes) and9% of the patients on sodium valproate (nodominant event). There was a considerablyhigher number of pregnancies in patients oncarbamazepine, but the data are not availableto establish whether all of these were unin-tended or due to reduced efficacy of contra-ceptive medication after hepatic enzymeinduction.
In conclusion, we have shown in this trial inadult patients with newly diagnosed seizures,that sodium valproate and carbamazepinewere of similar overall efficacy whether forprimary generalised seizures or partialseizures. Apart from a high incidence ofsevere rashes in the first few weeks of treat-ment with carbamazepine, both drugs showed
equivalent long term tolerability. The resultspresented here and elsewhere would supportan initial dosage regime for sodium valproateof 600 mg/day for all seizure types, increasingto 800 or 1000 mg/day within the first monthin patients with partial seizures to ensure thatfull seizure control is achieved as soon as pos-sible after start of treatment.We gratefully acknowledge the advice, support, and statisti-
cal expertise provided by Dr A L Johnson (MRC BiostatisticsUnit, Cambridge) and Mr M James (Sanofi Winthrop Ltd) inthe data analysis and to Mr P Simmonds (Sanofi WinthropLtd) for the data presentation. Our thanks go also to Dr CBinnie of the Maudisey Hospital, London for his independentassessment of the pretreatment EEG records.
Other members of the Adult EPITEG Collaborative Group,without whose participation and enthusiasm the trial could nothave been undertaken, were Dr D Bates, Royal VictoriaInfirmary, Newcastle upon Tyne; Dr HG Boddie, NorthStaffordshire Royal Infirmary, Stoke on Trent; Dr CRAClarke, St Bartholomew's Hospital, London; Dr P Cleland,Sunderland District General Hospital, Sunderland; Dr RNCorston, New Cross Hospital, Wolverhampton; Dr RE Cull,Royal Infirmary, Edinburgh; Dr WJK Cumming, WithingtonHospital, Manchester; Dr CJK Ellis, Poole General Hospital,Poole; Dr AK Gupta, Dudley Road Hospital, Birmingham;Professor MJG Harrison, Middlesex Hospital, London; Dr GHarwood, St Bartholomew's Hospital, Rochester, Kent andGreenwich District Hospital, London; Dr D Jefferson,Derbyshire Royal Infirmary, Derby; Professor C Kennard,Charing Cross Hospital, London; Dr LA Loizou, PinderfieldsGeneral Hospital, Wakefield; Dr PAH Millac, Leicester RoyalInfirmary, Leicester; Dr P Newman, Middlesbrough GeneralHospital, Middlesbrough; Dr VH Patterson, Claremont StreetHospital for Nervous Diseases, Belfast, and Dr DI Shepherd,North Manchester General Hospital, Manchester.
2 Jeavons PM, Covanis A, Gupta AK. Monotherapy withsodium valproate. In: Cranger R, Angeleri F, Penry JK,ed. Advances in Epileptology. New York: Raven Press,1980:415-8. (Epilepsy International Symposium NoXII.)
3 Jeavons PM. Choice of drug therapy in epilepsy.Practitioner 1977;219:542-56.
4 Turnbull DM, Howel D, Rawlins MD, Chadwick DW.Which drug for the adult epileptic patient: phenytoin orvalproate? BMJ 1985;290:815-9.
5 Callaghan N, Kenny RA, O'Neill B, Crowley M, GogginT. A prospective study between carbamazepine, pheny-toin and sodium valproate as monotherapy in previouslyuntreated and recently diagnosed patients with epilepsy.J Neurol Neurosurg Psychiatry 1985;48:639-44.
6 Commission on Classification and Terminology of theInternational League against Epilepsy. Proposal forrevised clinical and electroencephalographic classifica-tion of epileptic seizures. Epilepsia 1981;22:489-501.
7 Cox DR. Regression models and life tables. J R Stat SocSeries B 1972;34:187-220.
8 Cartlidge NEF. The adult EPITEG trial: a comparativemulticentre clinical trial of sodium valproate and carba-mazepine in adult onset epilepsy. Part 1: description oftrial and demographic data. In: Chadwick D, ed.Proceedings of the fourth international symposium on sodiumvalproate and epilepsy. London: Royal Society ofMedicine Services, 1989:106-13.
9 Kay R Treatment effects in competing-risk analysis inprostate cancer. Biometrics 1986;42:203-1 1.
10 Davidson, DLW. The adult EPITEG trial: a comparativemulticentre clinical trial of sodium valproate and carba-mazepine in adult onset epilepsy. Part 2: Adverse effects.In: Chadwick D, ed. Proceedings of the fourth internationalsymposium on sodium valproate and epilepsy. LondonRoyal Society of Medicine Services, 1989:114-24.
11 Chadwick D. Diagnosis of epilepsy. Lancet 1990:36:291-6.
12 Heller AJ, Chesterman P, Elwes RDC, Reynolds EH,Crawford P, Chadwick D, Johnson AL. Phenobarbitone,phenytoin, carbamazepine or sodium valproate for newlydiagnosed adult epilepsy: a randomised, comparativemonotherapy trial. J Neurol Neurosurg Psychiatry 1994;57:(in press).
13 de Silva M, McArdle B, McGowan M, Neville BGR,Johnson AL, Reynolds EH. A prospective, randomisedcomparative monotherapy clinical trial in childhoodepilepsy. In: Chadwick D, ed. Proceedings of the fourthinternational symposium on sodium valproate and epilepsy.London: Royal Society of Medicine Services, 1989:81-6.
14 Verity CM, Hosking F, Easter DJ. A multicentre compara-tive trial of sodium valproate and carbamazepine in pae-diatric epilepsy. Dev Med Child Neurol 1994 (in press).
15 Mattson RH, Cramer JA, Collins JF and the Departmentof Veterans Affairs Epilepsy Cooperative Study No 264Group. A comparison of valproate with carbamazepinefor the treatment of complex partial seizures and secon-darily generalized tonic-clonic seizures in adults. N EnglJ Med 1992;327:765-71.
15 Brodie MJ. Established anticonvulsants and treatment ofrefractory epilepsy. Lancet 1990;336:350-5.
687
on October 20, 2020 by guest. P
rotected by copyright.http://jnnp.bm
j.com/
J Neurol N
eurosurg Psychiatry: first published as 10.1136/jnnp.57.6.682 on 1 June 1994. D
