Article Text

Download PDFPDF

Research paper
KOMET: an unblinded, randomised, two parallel-group, stratified trial comparing the effectiveness of levetiracetam with controlled-release carbamazepine and extended-release sodium valproate as monotherapy in patients with newly diagnosed epilepsy
  1. Eugen Trinka1,
  2. Anthony G Marson2,
  3. Wim Van Paesschen3,
  4. Reetta Kälviäinen4,
  5. Jacqueline Marovac5,
  6. Benjamin Duncan6,
  7. Sonja Buyle5,
  8. Yngve Hallström7,
  9. Petr Hon8,
  10. Gian Carlo Muscas9,
  11. Mark Newton10,
  12. Heinz-Joachim Meencke11,
  13. Philip E Smith12,
  14. Bernd Pohlmann-Eden13,
  15. for the KOMET Study Group
  1. 1Department of Neurology, Paracelsus Medical University, Christian Doppler Klinik, Salzburg, Austria
  2. 2Department of Molecular and Clinical Pharmacology, University of Liverpool, Liverpool, UK
  3. 3Department of Neurology, University Hospital Gasthuisberg, Katholieke Universiteit Leuven, Leuven, Belgium
  4. 4Department of Neurology, Kuopio Epilepsy Center, Kuopio University Hospital and University of Eastern Finland, Kuopio, Finland
  5. 5UCB Pharma, Brussels, Belgium
  6. 6UCB Pharma, Raleigh, North Carolina, USA
  7. 7Neuro Center, St Goran's Hospital, Stockholm, Sweden
  8. 8Department of Neurology, University Hospital Ostrava, Ostrava, Czech Republic
  9. 9Department of Neuroscience, 3rd Neurology Division and Epilepsy Center, Careggi Hospital, Florence, Italy
  10. 10Epilepsy Research Centre Heidelberg West, Melbourne, Australia
  11. 11Department of Epileptology, Hospital Königin Elisabeth Herzberge, Berlin, Germany
  12. 12Wales Epilepsy Research Network, University Hospital of Wales, Cardiff, UK
  13. 13Epilepsy Program, QE II Health Centre, Dalhousie University, Halifax, Canada
  1. Correspondence to Dr Eugen Trinka, Department of Neurology, Paracelsus Medical University, Christian Doppler Klinik, Ignaz Harrer Strasse 79, A-5020 Salzburg, Austria; e.trinka{at}salk.at

Abstract

Objective To compare the effectiveness of levetiracetam (LEV) with extended-release sodium valproate (VPA-ER) and controlled-release carbamazepine (CBZ-CR) as monotherapy in patients with newly diagnosed epilepsy.

Methods This unblinded, randomised, 52-week superiority trial (NCT00175903) recruited patients (≥16 years of age) with ≥2 unprovoked seizures in the previous 2 years and ≥1 in the previous 6 months. The physician chose VPA or CBZ as preferred standard treatment; each patient was randomised to standard treatment or LEV. The primary outcome was time to treatment withdrawal (LEV vs standard antiepileptic drugs (AEDs)). Analyses also compared LEV with VPA-ER, and LEV with CBZ-CR.

Findings 1688 patients (mean age 41 years; 44% female) were randomised to LEV (n=841) or standard AEDs (n=847). Time to treatment withdrawal was not significantly different between LEV and standard AEDs: HR (95% CI) 0.90 (0.74 to 1.08). Time to treatment withdrawal (HR (95% CI)) was 1.02 (0.74 to 1.41) for LEV/VPA-ER and 0.84 (0.66 to 1.07) for LEV/CBZ-CR. Time to first seizure (HR, 95% CI) was significantly longer for standard AEDs, 1.20 (1.03 to 1.39), being 1.19 (0.93 to 1.54) for LEV/VPA-ER and 1.20 (0.99 to 1.46) for LEV/CBZ-CR. Estimated 12-month seizure freedom rates from randomisation: 58.7% LEV versus 64.5% VPA-ER; 50.5% LEV versus 56.7% CBZ-CR. Similar proportions of patients within each stratum reported at least one adverse event: 66.1% LEV versus 62.0% VPA-ER; 73.4% LEV versus 72.5% CBZ-CR.

Conclusions LEV monotherapy was not superior to standard AEDs for the global outcome, namely time to treatment withdrawal, in patients with newly diagnosed focal or generalised seizures.

  • Levetiracetam
  • valproic acid
  • carbamazepine
  • epilepsy
  • monotherapy
  • neurophysiol
  • clinical
  • neuroepidemiology
  • surgery
  • MRI

Statistics from Altmetric.com

Introduction

Levetiracetam (LEV) is a newer antiepileptic drug (AED) with a unique mode of action1 ,2 and a broad spectrum of efficacy against focal3–5 and generalised6 ,7 seizures. In a double-blind monotherapy trial in newly diagnosed patients, LEV demonstrated similar seizure freedom rates compared with controlled-release carbamazepine (CBZ-CR) after 6 and 12 months.8

Double-blind randomised controlled trials remain the gold standard for assessing efficacy and tolerability for regulatory purposes. However, available regulatory trials in epilepsy often fail to inform treatment decisions made by clinicians and their patients. Such trials have short duration, fixed dosing schedules and restricted patient population, resulting in poor external validity.9 ,10 Treatment decisions for epilepsy should be informed by long-term trials that measure outcomes with real clinical utility (of effectiveness rather than efficacy), and recruit a broad population representative of everyday practice. Blinding long-term epilepsy trials can limit their external validity, as important groups, such as women of childbearing age are under-represented, while maintaining the blind is impractical due to problems, such as drug interactions. Also, the blind would need to be broken for patients who discontinue study medication in order to inform future treatment choices. We urgently need long-term treatment outcome data in epilepsy; these will largely come from unblinded randomised trials, accepting the trade-off between internal and external validity.

The Standard and New Antiepileptic Drugs (SANAD) trial was a long-term (up to 6 years) unblinded study that identified lamotrigine as a first-line treatment for patients with focal-onset seizures,11 and valproate (VPA) as a first-line treatment for patients with generalised-onset seizures.12 SANAD results have been used in the development of the latest German treatment guidelines for neurology,13 and have triggered an update of the UK National Institute for Health and Clinical Excellence epilepsy guideline.14 However, SANAD could not include LEV, because it was not licensed at the time when SANAD was designed. In KOMET (Keppra vs Older Monotherapy in Epilepsy Trial), we have compared the effectiveness of LEV and extended-release VPA (VPA-ER) and CBZ-CR.

Methods

Patients

Patients were recruited to this study at specialist clinics at the discretion of the physician. Patients aged ≥16 years were included if they had two or more unprovoked seizures in the previous 2 years with at least one during the previous 6 months, classified according to the International League Against Epilepsy.15 Neuroimaging and electroencephalography were not compulsory prior to entry, if clinical information was sufficient for a diagnosis of epilepsy.

Patients were excluded if they had been treated with LEV, VPA or CBZ for any indication or treated for epilepsy with any other AED in the last 6 months. Acute seizure treatment was allowed with a maximum of 2 weeks' duration providing it had been stopped at least 1 week before screening. All participants provided written informed consent before entering the study. The study was approved by the local ethics committees for every study centre, and was conducted in accordance with the International Conference on Harmonisation Good Clinical Practice guidelines and the Declaration of Helsinki. Ongoing medical review of all data was conducted by the sponsor.

Study design

KOMET (N01175; NCT00175903) was a multicentre, unblinded, randomised, 52-week, controlled superiority trial with a two-parallel-group design (figure 1).

Figure 1

Study design. AED, antiepileptic drug; CBZ, carbamazepine; CBZ-CR, controlled-release carbamazepine; LEV, levetiracetam; SV, safety visit; V, visit; VPA, sodium valproate; VPA-ER, extended-release sodium valproate; W, week.

At screening, the clinician decided whether VPA or CBZ would be the standard first-line treatment. Central randomisation, stratified by best recommended treatment, was done. Within the VPA stratum, patients were randomised (1:1) to treatment with LEV (UCB Pharma, Belgium) or VPA-ER (Sanofi-Aventis, France). Within the CBZ stratum, patients were randomised (1:1) to treatment with LEV or CBZ-CR (Novartis, Switzerland). Treatment allocation was concealed by use of an Interactive Voice Response System via telephone to manage the randomisation process.

Starting doses (LEV 500 mg/day, VPA-ER 500 mg/day, CBZ-CR 200 mg/day, administered twice daily as equal doses) were up-titrated over 2 weeks to the initial target doses (LEV 1000 mg/day, VPA-ER 1000 mg/day, CBZ-CR 600 mg/day). If a seizure occurred, doses could be increased according to the clinician's judgement to a maximum of LEV 3000 mg/day, VPA-ER 2000 mg/day and CBZ-CR 1600 mg/day. Patients who did not tolerate higher doses could revert to lower doses, but the dose could not fall below the initial target dose.

Study visits were scheduled at weeks 0, 6, 12, 26 and 52 (evaluation visit). Patients recorded the number and type of seizures and any adverse events (AEs) using daily record cards. Their intensity (mild, moderate or severe), as judged by the investigator, was recorded. Serious AEs were defined as those which resulted in death, were life-threatening, required prolonged hospitalisation, resulted in persistent or significant disability or incapacity, or were congenital anomalies. A health-related quality-of-life questionnaire (Patient Quality of Life Inventory in Epilepsy-31 (QOLIE-31-P)) and a health status questionnaire (EQ-5D) were completed at weeks 0, 6 (EQ-5D only), 12, 26 and 52 by all patients in countries where a local language version was available.

Outcome measures

The primary outcome measure was time to withdrawal from study medication (treatment withdrawal) calculated from randomisation to the day after the last intake of study medication for the overall comparison of LEV with standard AEDs (VPA-ER and CBZ-CR). Patients could withdraw consent at any time, or could be withdrawn according to the investigator's clinical judgement. Secondary outcome measures were time to first seizure calculated from randomisation, and treatment withdrawal and seizure freedom rates at 6 and 12 months.

Exploratory outcome measures were changes from baseline to the last assessment for all QOLIE-31-P and EQ-5D scores. Tolerability was evaluated by recording treatment-emergent AEs, their intensity (mild, moderate or severe) and seriousness. Severe AEs were those affecting the patient's ability to work normally or to carry out usual activities, and those of definite clinical consequence. Serious AEs were those classed as life threatening, resulting in death, requiring hospitalisation or causing a persistent or significant disability/incapacity.

Statistical analysis

Sample size calculations were based on the primary outcome (time to treatment withdrawal) for the overall comparison of LEV with standard AEDs across the combined VPA and CBZ strata. The calculation was based on the log rank test of survival in two groups followed for a fixed time (12 months) and assuming a constant HR. The assumed 12-month retention rate for the standard AED group was 60%. To detect a clinically relevant absolute difference in retention rate between LEV and standard AEDs of 7% (eg, 60% vs 67%) with a power of 90% using a two-sided test (α=0.05), 982 patients per treatment group were needed (total 1964).

Supportive analyses of the primary endpoint were also conducted for each individual stratum (VPA and CBZ). Secondary endpoints were analysed in a similar fashion.

The primary analysis was by intention-to-treat (ITT) and included all randomised patients. In order to assess the effectiveness of study treatment in particular seizure types, and to assess a potential interaction between treatment and seizure type, a post hoc subgroup analysis was performed using data from patients who had a history of focal or generalised seizures only (excluding unclassified, unknown or mixed seizure types).

The safety population (tolerability analyses) consisted of all patients who received one or more doses of study medication, including those with unknown intake. Patients who were randomised but not treated, and those who did not give informed consent, were excluded from the safety population.

Kaplan–Meier survival curves were plotted for time to treatment withdrawal and time to first seizure. For the time to treatment withdrawal analysis, all treated patients who withdrew from the study prior to day 365 were considered as having the event. Treated patients who completed the study or withdrew after day 365 were censored at day 365 or at completion of the study (for those who completed just prior to day 365). Untreated (but randomised) patients were censored at day 1 (1 day after randomisation). For the time to first seizure analysis, patients with no reported seizure during the 1-year treatment and observation period were censored at the date of last intake of study medication, date of early termination, date of week 52 visit or day 364, whichever was earliest. Time to treatment withdrawal and time to first seizure were calculated from randomisation.

Time to treatment withdrawal and time to first seizure were analysed using a Cox's proportional hazards regression model. The treatment effect (HR) was described using two-sided 95% CIs. In this model, a HR of <1 favoured LEV, while a HR of >1 favoured standard AEDs. This superiority trial was to be considered positive if the null hypothesis of no difference between LEV and standard AEDs was rejected in favour of LEV. The primary analysis (time to treatment withdrawal, LEV vs standard AEDs) was stratified according to the standard recommended treatment at baseline. Results for each stratum are also reported for the subgroups with only focal or only generalised seizures.

No adjustment for multiple comparisons was required since only one primary analysis was planned and conducted. For the time to treatment withdrawal and time to first seizure analyses, no further adjustments for dropouts or missing data were required, as these patients were censored accordingly.

Results

Patients

This study was carried out in a community setting between February 2005 and October 2007 in 269 centres across 23 European countries and Australia.

One thousand seven hundred and one patients were screened and 1698 were randomised (figure 2). Ten patients were excluded from the ITT population (six with no documented informed consent; four because of non-compliance with International Conference on Harmonisation Good Clinical Practice). Patients in the VPA stratum (n=696) were randomised to LEV (n=349) or VPA-ER (n=347) and those in the CBZ stratum (n=992) were randomised to LEV (n=492) or CBZ-CR (n=500).

Figure 2

Patient disposition. CBZ, carbamazepine; CBZ-CR, controlled-release carbamazepine; CRF, case report form; GCP, good clinical practice; ITT, intent to treat; LEV, levetiracetam; VPA, sodium valproate; VPA-ER, extended-release sodium valproate.

The LEV and standard AED treatment groups were similar with respect to baseline demographics and epilepsy characteristics (table 1). Within the VPA stratum, 65.8% (458/696) had only primary generalised seizures (table 2). Within the CBZ stratum 86.5% (858/992) had only focal seizures.

Table 1

Baseline demographics and epilepsy characteristics (intent-to-treat population)

Table 2

Time to treatment withdrawal and time to first seizure by VPA and CBZ strata for subgroups of patients with focal or generalised seizures only (intent-to-treat population excluding unclassified/unknown seizure types)

Overall, 1266 patients (75.0%) in the ITT population were still taking their randomised drug at 12 months with a similar proportion in the LEV and standard AED groups (figure 2). Treatment was discontinued because of AEs in 181 patients (10.7%) and because of lack of efficacy in 60 (3.6%). Seventy-four patients (4.4%) withdrew consent for personal reasons not related to AEs or efficacy, 37 (2.2%) discontinued for other reasons and 70 (4.1%) were lost to follow-up.

Median (range) daily doses (including the up-titration period) were: LEV 987 mg/day (250–2807), VPA-ER 987 mg/day (500–2263), CBZ-CR 588 mg/day (180–1422).

LEV versus standard AEDs

The time to treatment withdrawal was longer in patients treated with LEV compared with standard AEDs, but the difference was not significant (HR 0.90, 95% CI 0.74 to 1.08) (table 3; figure 3A). Time to first seizure was significantly longer for patients in the standard AEDs group compared with the LEV group (HR 1.20, 95% CI 1.03 to 1.39) (table 3; figure 4A).

Table 3

Time to treatment withdrawal, treatment withdrawal rates, time to first seizure and seizure freedom rates for LEV and standard AEDs (intent-to-treat population)

Figure 3

Kaplan–Meier survival curves for the time to treatment withdrawal (intent-to-treat population) for: (A) LEV vs standard AEDs; (B) LEV vs VPA-ER in the VPA stratum; (C) LEV vs CBZ-CR in the CBZ stratum. AED, antiepileptic drug; CBZ, carbamazepine; CBZ-CR, controlled-release carbamazepine; LEV, levetiracetam; VPA, sodium valproate; VPA-ER, extended-release sodium valproate.

Figure 4

Kaplan–Meier survival curves for the time to first seizure (intent-to-treat population) for: (A) LEV vs standard AEDs; (B) LEV vs VPA-ER in the VPA stratum; (C) LEV vs CBZ-CR in the CBZ stratum. AED, antiepileptic drug; CBZ, carbamazepine; CBZ-CR, controlled-release carbamazepine; LEV, levetiracetam; VPA, sodium valproate; VPA-ER, extended-release sodium valproate.

LEV versus VPA-ER

Time to treatment withdrawal was similar for LEV and VPA-ER (HR 1.02, 95% CI 0.74 to 1.41) (figure 3B). For the comparisons by seizure type, no significant differences were found, but there were trends favouring VPA-ER in patients with primary generalised seizures (HR 1.16, 95% CI 0.79 to 1.71), and LEV in those with focal seizures (HR 0.73, 95% CI 0.37 to 1.44) (table 2).

Estimated treatment withdrawal rates at 6 and 12 months were similar for LEV and VPA-ER, both for all patients and those with generalised seizures only (table 4). The estimated overall withdrawal rates (95% CI) at 12 months were 22.0% (18.0–26.7) with LEV and 21.6% (17.7–26.4) with VPA-ER.

Table 4

Treatment withdrawal and seizure freedom rates (Kaplan–Meier estimates) for the VPA stratum for all patients and those with generalised seizures only, and for the CBZ stratum for all patients and those with focal seizures only (intent-to-treat population excluding unclassified/unknown seizure types)

Time to first seizure favoured VPA-ER over LEV (HR 1.19, 95% CI 0.93 to 1.54), but this was not statistically significant (figure 4B). A similar effect was seen when analysis included patients with generalised seizures only and those with focal seizures only (table 2).

Estimated seizure freedom rates at 6 and 12 months were higher with VPA-ER than LEV, both for all patients and in those with generalised seizures only (table 4). The estimated overall seizure freedom rate (95% CI) at 12 months was 58.7% (53.1–63.9) for all patients in the VPA stratum (excluding those with unclassified/unknown seizure types) treated with LEV and 64.5% (58.9–69.5) for those treated with VPA-ER.

LEV versus CBZ-CR

Time to treatment withdrawal was longer for LEV than for CBZ-CR, but the difference was not statistically significant (HR 0.84, 95% CI 0.66 to 1.07) (figure 3C). Results in patients with focal seizures only were similar to those in all patients (HR 0.84, 95% CI 0.65 to 1.09) (table 2). In the group of patients with generalised seizures only, the HR favoured LEV (HR 0.49, 95% CI 0.16 to 1.49), but the number of patients was very small (table 2).

Estimated treatment withdrawal rates at 6 and 12 months were higher with CBZ-CR than LEV, both for all patients and those with focal seizures only (table 4). The estimated overall withdrawal rate (95% CI) at 12 months was 25.2% (21.6 to 29.3) for LEV and 28.8% (25.1 to 33.0) for CBZ-CR.

Time to first seizure was longer in the CBZ-CR group compared with the LEV group, but the difference was not statistically significant (HR 1.20, 95% CI 0.99 to 1.46) (figure 4C). Results in patients with focal or generalised seizures only were similar to those in all patients (table 2), with a significant difference between treatment groups in those with focal seizures only in favour of CBZ-CR (HR 1.24, 95% CI 1.01 to 1.52).

The estimated seizure freedom rates at 6 and 12 months were higher with CBZ-CR than LEV, both for all patients and those with focal seizures only (table 4). Overall, the estimated seizure freedom rate (95% CI) at 12 months was 50.5% (45.8 to 55.1) for all patients in the CBZ stratum (excluding those with unclassified/unknown seizure types) treated with LEV and 56.7% (51.8 to 61.2) with CBZ-CR.

Quality of life

There were no clear differences between LEV and standard AEDs in the impact on health-related quality of life as measured by the QOLIE-31-P, or health status as measured by the EQ-5D (data not shown). No clear trends over time were noted, apart from an improvement in the seizure worry subscale score in all treatment groups.

Tolerability

In the VPA stratum, similar numbers of patients treated with LEV and VPA-ER reported at least one treatment-emergent AE, and similar numbers of drug-related AEs were reported (table 5). Most treatment-emergent AEs reported (table 6) were mild or moderate. Serious AEs were reported by 39 patients (11.3%) treated with LEV and 20 (5.8%) with VPA-ER. The most common serious AEs were convulsion (LEV seven patients, VPA-ER three patients) and epilepsy (LEV three patients; VPA-ER four patients). Discontinuation of treatment due to AEs was similar in patients treated with LEV (6.1%) and VPA-ER (4.7%).

Table 5

Summary of treatment-emergent AEs (safety population)

Table 6

Incidence of treatment-emergent AEs ≥5% in any of the treatment groups (safety population)

In the CBZ stratum, similar numbers of patients treated with LEV and CBZ-CR reported at least one treatment-emergent AE (table 5). More drug-related AEs were reported in the CBZ-CR group. Most treatment-emergent AEs were mild or moderate (table 6). Serious AEs were reported by 67 patients (13.7%) treated with LEV and 41 (8.2%) with CBZ-CR. The most common serious AEs were convulsion (LEV nine patients; CBZ-CR four patients) and ‘grand mal convulsion’ (generalised tonic-clonic seizure) (LEV four patients; CBZ-CR one patient). More patients treated with CBZ-CR discontinued treatment due to AEs than those treated with LEV (18.8% vs 9.8%).

Eight deaths were reported; five in patients treated with LEV: brain neoplasm (two); suspected cardiac arrhythmia; head injury sustained in road traffic accident; radiation injury (one each), two in the CBZ-CR group (subarachnoid haemorrhage; acute myocardial infarction), and one in the VPA-ER group (sudden unexpected death in epilepsy). None of the deaths were considered to be related to study medication.

Discussion

The choice of first AED is probably the most important treatment decision that a patient with epilepsy will make. Such treatment decisions should be informed by pragmatic, longer-term, head-to-head trials, accepting the trade-off between external and internal validity if unblinded. Monotherapy studies in patients with newly diagnosed epilepsy present a number of methodological and statistical challenges.

In KOMET, we have compared LEV with the standard treatments for generalised (VPA) and focal (CBZ) seizures. Dosing regimens were flexible and modified-release (ER or CR) formulations of standard treatments were used, given reports that they are better tolerated than immediate-release formulations.16–20

The results across strata for our primary outcome and time to treatment withdrawal, yielded a HR (95% CI) of 0.90 (0.74 to 1.08) suggesting that LEV is not superior to standard AEDs. Overall results across strata for time to first seizure, the main efficacy outcome, suggest an advantage for standard treatments compared with LEV (HR 1.20, 95% CI 1.03 to 1.39).

VPA is widely regarded as the treatment of choice for generalised epilepsies or difficult-to-classify seizures.12 In SANAD, VPA was better tolerated than topiramate and more efficacious than lamotrigine. For time to treatment failure, VPA was significantly better than topiramate (HR 1.57, 95% CI 1.19 to 2.08), with no significant difference between VPA and lamotrigine (HR 1.25, 95% CI 0.94 to 1.68). For the subgroup of patients with idiopathic generalised epilepsy, VPA was significantly better than either lamotrigine (HR 1.55, 95% CI 1.07 to 2.24) or topiramate (HR 1.89, 95% CI 1.32 to 2.70).12

In our trial, for the stratum comparing LEV and VPA-ER, the HR (95% CI) for time to treatment withdrawal was 1.02 (0.74 to 1.41), suggesting similarity, but the CI does not exclude the possibility that LEV is 26% better or 41% worse (HR Scale). This stratum included predominantly (∼70%) patients with generalised seizures, although some had focal seizures and arguably were not offered the best standard treatment. A subgroup analysis of only the patients classified with generalised seizures gives an HR for treatment withdrawal of 1.16 (95% CI 0.79 to 1.71) suggesting a non-significant advantage for VPA-ER; however, the differences in the proportion withdrawing from treatment at 6 and 12 months is small with narrow 95% CIs; 0.9% at 6 months (CI −6.2 to 8.1) and 3.0% at 12 months (CI −4.7 to 10.7).

For time to first seizure in the VPA stratum, results (HR 1.19, 95% CI 0.93 to 1.54) are similar to the overall results, suggesting an advantage for VPA-ER. Time to first seizure can be influenced by initial titration rates and drug pharmacokinetics, and ideally a longer-term measure of seizure control is required. We were able to assess the proportion of patients remaining seizure-free at 12 months. For the VPA stratum, 58.7% were seizure-free on LEV and 64.5% on VPA-ER. The estimated absolute difference was −5.7% (95% CI −13.4% to 1.9%) suggesting a small advantage for VPA-ER.

Women of childbearing age represented a significant proportion of the epilepsy population. Increased human teratogenicity21–23 and risk of impaired cognitive function in children exposed in utero to VPA24 ,25 are perhaps the most important adverse effects of VPA. It is therefore imperative that we identify effective treatments that are safer in pregnancy. Our study is the first to point towards a comparable time to treatment withdrawal for a newer AED compared with VPA-ER in patients with predominantly generalised seizures. Recent preliminary data suggest a low risk of teratogenicity with LEV (UK and UCB AED pregnancy registries)26 ,27 and a lower risk of delayed development following in utero exposure to LEV compared with VPA.28 ,29 Based on these results, LEV might be a viable first-line treatment for patients with generalised seizures, particularly for women of childbearing age.

The choice of CBZ as a standard treatment for patients with focal seizures was underscored in the 2006 International League Against Epilepsy treatment guidelines.30 SANAD identified lamotrigine as a preferred option for patients with focal seizures11 but these results were not available when KOMET was planned. In the CBZ stratum, our results for time to treatment withdrawal suggest a non-significant advantage for LEV compared with CBZ-CR (HR 0.84, 95% CI 0.66 to 1.07). Results were similar when restricted to patients with focal seizures (HR 0.84, 95% CI 0.65 to 1.09). For time to first seizure we found a non-significant advantage for CBZ-CR (HR 1.20, 95% CI 0.99 to 1.46), and a significant advantage for CBZ-CR when the analysis was restricted to patients with focal seizures (HR 1.24, 95% CI 1.01 to 1.52). Estimates of the proportion of patients remaining seizure-free at 12 months find only a small advantage for CBZ-CR: 50.5% on LEV and 56.7% on CBZ-CR were seizure-free at 12 months (absolute difference −6.1% (95% CI −12.8% to 0.5%)). The analysis restricted to patients with focal seizures finds a similar absolute difference (−8.1, 95% CI −15.3 to 1.0).

The results for time to treatment withdrawal and time to first seizure suggest a trade-off between efficacy and tolerability, with LEV being better tolerated but CBZ-CR being slightly more effective. Better tolerability of LEV is supported by the lower incidence of treatment withdrawal due to side effects (9.8% vs 18.8%), although longer-term outcome data are required to confirm the latter. It should be noted that the initial target dose of CBZ-CR (600 mg/day) may have been unnecessarily high, resulting in a higher discontinuation rate due to AEs. In addition, a higher incidence of serious AEs was observed with LEV than CBZ-CR (13.7% vs 8.2%).

Interpretation of our findings should take into account study limitations. Possible sources of bias include selection of patients at the discretion of the physician, and unblinded treatment. Additionally, the choice of best recommended treatment was not standardised in accordance with expert recommendations. KOMET was designed to be relevant to routine clinical practice and, thus, it was not mandatory for clinicians to confirm the diagnosis using brain imaging or electroencephalograph before selecting treatment with VPA or CBZ. Finally, although patients in KOMET were followed up over a 12-month treatment period, there is still a need for data collected over a longer duration.

We have found LEV to be non-superior to both VPA-ER and CBZ-CR for the global outcome and time to treatment withdrawal. KOMET provides data for clinicians and patients considering LEV as a first-line treatment. Guidelines and policy decisions will be further informed by studies collecting longer-term seizure data.

Acknowledgments

Bernd Pohlmann-Eden, MD, PhD, former head of Epilepsy Centre Bethel, Bielefeld-Germany, now at the Epilepsy Program, QE II Health Science Centre, Dalhousie University, Halifax, Canada, was the primary investigator and coordinator of this trial. The authors would like to thank the members of the KOMET Study Group for their contribution to the study and collection of data. The KOMET Study Group comprised 269 principal investigators who recruited patients in 269 sites across 24 countries. The authors thank Pascal Edrich, MSc, Biostatistician, (UCB Pharma employee at the time of study conduct and preparation of the clinical study report) for initial statistical analysis, Michel Bourgois, PhD, (Business & Decision (Life\Sciences), Brussels, Belgium; UCB Pharma consultant) for support with additional post hoc statistical analysis, Françoise Tonner, MD, (UCB Pharma employee at the time of study conduct and preparation of the clinical study report) and Robert Chan, MD, (UCB Pharma) for critical review of the manuscript and Svetlana Dimova, MD, PhD, (UCB Pharma) for critical review and coordination of the manuscript preparation.

References

Footnotes

  • GCM is deceased.

  • Funding KOMET was funded and sponsored by UCB Pharma who was responsible for the design and conduct of the study, and collection, management, analysis and interpretation of the data. Medical writing and editorial assistance was provided by Jennifer Stewart, MSc, (QXV Communications, Macclesfield, UK), and was funded by UCB Pharma.

  • Competing interests ET has acted as a paid consultant to Eisai, Medtronics, Bial and UCB Pharma. He has received research funding from UCB Pharma, and speakers' honoraria from Bial, Cyberonics, Desitin Pharma, Eisai, Gerot and UCB Pharma. AGM has received research funding from UCB Pharma, Eisai, GlaxoSmithKline (GSK) and Pfizer, has acted as a paid consultant to UCB Pharma and Cyberonics and has received speakers' honoraria from Sanofi-Aventis and GSK, and travel grants from GSK, UCB Pharma, Janssen-Cilag, Eisai and Sanofi-Aventis. WVP has acted as a paid consultant to and has received speakers' honoraria from UCB Pharma, Pfizer, Janssen-Cilag, Valeant, Johnson & Johnson (J&J), Eisai, Sanofi-Aventis, Novartis and GSK. He has received research funding from UCB Pharma, and travel grants from UCB Pharma, Pfizer, GSK, Janssen-Cilag and Novartis. RK has acted as a board member for UCB Pharma and Eisai, and as a paid consultant to UCB Pharma, Eisai, Orion Pharma and Janssen-Cilag, and has received research grants from UCB Pharma, speakers' honoraria from UCB Pharma, Eisai, Pfizer, Orion Pharma, Sanofi-Aventis and Janssen-Cilag, and travel grants from UCB Pharma, Eisai, Pfizer, Orion Pharma, Sanofi-Aventis, Cephalon and J&J. JM and SB were full-time employees of UCB Pharma at the time when the KOMET study was conducted and the results analysed. BD is a full-time employee of UCB Pharma. YH has been a paid advisory board member to UCB Pharma and Pfizer. PH has no relevant financial relationships outside the submitted work. GCM (deceased) had received speakers' or consultancy fees and/or travel/accommodation grants for participation in medical congresses and investigators' meetings from Cyberonics, Eisai, Eli Lilly, Novartis, Pfizer, Sanofi-Aventis and UCB Pharma. MN has received a speakers' honorarium from UCB Pharma. H-JM was a paid consultant to UCB Pharma and Eisai. He has received unrestricted research grants from Janssen-Cilag, Pfizer, and UCB Pharma and speakers' honoraria from Cyberonics, Desitin Pharma, Eisai, Novartis, Janssen-Cilag, Pfizer, and UCB Pharma. PS has received an unrestricted research grant from UCB Pharma, speakers' honoraria from UCB Pharma and Eisai, and travel grants from UCB Pharma and Janssen-Cilag. B P-E was a paid consultant to Desitin Pharma, UCB Pharma, Janssen-Cilag and Pfizer. He received research and trial grants from UCB Pharma and Pfizer, and speakers' honoraria from UCB Pharma, Pfizer, Desitin Pharma, Eisai and Janssen-Cilag.

  • Ethics approval The study was approved by the local ethics committees for every study centre.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles