Current antiepileptic drug treatment

Review

Current drug treatment of epilepsy in adults

Dougall McCorry, David Chadwick, and Anthony Marson

The choice of antiepileptic drugs

(AEDs) is rapidly increasing. This
Generalised-onset seizures Partial-onset seizures
review looks at the evidence that
guides the decision of which AED to Absence Myoclonus Tonic-atonic Primary Simple partial Complex partial
tonic-clonic
start as monotherapy and aims to aid
the choice of treatment if monotherapy Secondary tonic-clonic
fails. Unfortunately, the evidence Restricted-spectrum AEDs
supporting the prescribing of new
Ethosuximide Carbemazepine
drugs is sparse, because most Phenytoin
randomised controlled trials answer Vigabatrin
questions focused on regulatory Gabapentin
Tiagabine
requirements rather than on clinical
Oxcarbazepine
use. Ultimately, the choice of one AED
will be determined by an individual
risk-benefit assessment in which the Broad-spectrum AEDs for all seizure types
most effective drug for an individual
Valproic acid
patient is chosen, and one that would
Lamotrigine*
have the lowest risk of significant harm. Topiramate
It is the risk of chronic toxic effects and Levetiracetam*
issues of teratogenicity for women that Zonisamide
Phenobarbital
may affect the choice of drug therapy Benzodiazepines
to the greatest degree. In the future
there is a need to improve the quality of
clinical data on efficacy and harmful Figure 1. The spectrum of the effectiveness of AEDs against different seizure types. *Evidence for
effects of AEDs. effectiveness against generalised seizures is largely limited to anecdotal evidence.

Lancet Neurol 2004; 3: 729–35 heterogeneous. Most people who develop epilepsy during
their life have a relatively short-lasting susceptibility to
The past 15 years has seen the registration of many new seizures and enter remission shortly after starting treatment
drugs (vigabatrin, gabapentin, lamotrigine, topiramate, on small doses of AEDs.2,3 However, 20–30% of people who
tiagabine, oxcarbazepine, levetiracetam, and zonisamide) develop epilepsy will have a chronic epilepsy that responds
for the treatment of epilepsy. This has presented a greatly incompletely to AED therapy.
increased choice for patients and their doctors, which in One of the key features in the management of epilepsy
turn has focused attention on the paucity of information has been the differentiation between seizures that are focal
and evidence that can currently support everyday clinical in onset and those which seem to be generalised from the
practice. In this article we attempt to review the currently start. On the basis of clinical experience, there is a strong
available evidence on drug treatment for epilepsy. belief that different AEDs may be effective against different
Epilepsy is the most common serious neurological seizure types and to some degree different epilepsy
disorder, with a prevalence of 0·5–1% in the general syndromes (figure 1). Thus, drugs that have been shown to
population. On average the incidence of epilepsy is be effective against focal seizures may be relatively
approximately 50 per 100 000 population, but is highest at ineffective against generalised seizures. Despite the clinical
the extremes of life. A recent prescription cost analysis
indicated that the cost of antiepileptic drugs (AEDs) to the
UK National Health Service in 2002 (prescribed for all
All authors are at The Walton Centre for Neurology and
indications) was £142 million of which £99 million was
Neurosurgery, Liverpool, UK.
spent on new AEDs.1
Correspondence: Prof David Chadwick, The Walton Centre for
Epilepsy itself is a varied disorder with many causes Neurology and Neurosurgery, Fazakerley, Liverpool, L9 7LJ, UK.
ranging from genetic causes through to acquired brain Tel +44 (0)151 529 5461; fax +44 (0)151 529 5465;
damage and insults. Disease outcomes are also email [email protected]

Neurology Vol 3 December 2004 http://neurology.thelancet.com 729

For personal use. Only reproduce with permission from Elsevier Ltd
 Review Current antiepileptic drug treatment

beliefs involved, there seems little evidence from clinical which the most effective drug for an individual patient is
trials to support the belief (see below). Increasingly, there is chosen, and one that would have the lowest risk of
a move to better define the potential interactions between a significant harm. AEDs are associated with both dose-related
drug and patient-specific factors, such as the type of epilepsy and idiosyncratic effects, but it is the risk of chronic toxic
or seizure, as well as to define the influence of drugs on long- effects and issues of teratogenicity for women that may affect
term prognosis. There are several important questions. First, the choice of drug therapy to the greatest degree.
do individual drugs differ in their efficacy in suppressing
seizures? Second, do individual drugs exacerbate certain Choice of a first AED
seizure types? And finally, do any of the drugs used to treat Reliable evidence about benefit will come from large-scale
epilepsy do more than simply suppress seizures: are they randomised controlled trials (RCTs) in which drugs have
antiepileptogenic and can they modify the natural history of been compared head-to-head, and in which outcomes that
epilepsy? have clinical use have been measured. Primary outcomes
The ideal drug for someone with epilepsy would not currently recommended by the International League Against
only suppress seizures but also reduce the susceptibility to Epilepsy (ILAE)12 include time to the following events after
seizures in the future (both antiseizure and randomisation: treatment failure, remission from seizures
antiepileptogenic). Epileptogenesis is a process in which for 12 months, and first seizure. Treatment failure is a
structural and functional changes occur after a brain insult composite outcome that indicates both efficacy and
that can lead to epilepsy, but epileptogenicity may also tolerability as a drug may fail because of continued seizures,
describe some processes that contribute to the progression side-effects, or a combination of both.
that is observed in some types of epilepsy. There are many It is worth considering whether AED monotherapy
animal data that support epileptogenesis as a process. This trials should be designed to detect a difference or
evidence is most clearly seen in the kindling model of equivalence. In 1998, the Commission on Antiepileptic
epilepsy and in various chronic animal models of epilepsy Drugs of the ILAE concluded that such trials should be
in which spontaneous seizures may develop after a latent designed to detect equivalence.12 In other words, trials
period following an acute brain insult such as induced should be designed to generate confidence intervals around
status epilepticus, neonatal hypoxia, or traumatic brain efficacy estimates that are narrow enough to exclude the
injury.4 Despite these models, there is a lack of any firm existence of important differences. This of course needs an
evidence that the drugs used to treat epilepsy in human a priori definition of the smallest important clinical
beings are antiepileptogenic.5 Temkin6 reviewed many difference, but at the time of writing, there is no consensus
clinical trials in which treatment was compared with no definition of this smallest clinical difference for the
treatment in individuals with significant risk for the primary outcomes in epilepsy monotherapy trials, and
development of later epilepsy. Whereas the drugs that were reported trials have used differing definitions. A further
tested consistently suppressed seizures in the short term, challenge for investigators is that most trials designed to
they did not seem to influence the long-term risk of detect equivalence will require the recruitment of
seizures. These data are supported by a first-seizure study,7 substantially more patients than those designed to detect a
which showed that treatment reduced seizure recurrence difference. For trials comparing new and standard AEDs,
after a first unprovoked seizure but did not effect long- the ILAE Commission suggested that a new drug could be
term remission rates. Thus, while a search for genuinely considered for use as a first-line agent if it is shown to have
antiepileptogenic drugs continues, current decision equivalent efficacy to, but is better tolerated than, the
making about the choice of drug treatment needs to be standard drug. A new drug showing equivalent efficacy and
based on the relative ability of drugs to suppress seizures in tolerability to a standard drug would be considered for use
the short term. as a second-line treatment.
We should of course seek to avoid prescription of drugs RCTs will also provide information about side-effects,
that might exacerbate seizures in a susceptible patient. This primarily those that are relatively common and occur soon
is a difficult area of study in a chronic variable paroxysmal after starting treatment, for example nausea or drowsiness.
condition like epilepsy, but has been reviewed by Perucca RCTs are not, however, the most appropriate methodology
and colleagues.8 A systematic approach to review did show to assess the risk of other perhaps more important side-
evidence of some specific drug effects in the exacerbation of effects including teratogenicity, rare but serious reactions
seizures. There are many reports of carbamazepine (eg, Stevens Johnson syndrome), and late effects (eg,
exacerbating typical and atypical absence seizures as well as osteoporosis). Evidence about the risk of these side-effects
myoclonic seizures.9–11 There is weak evidence that will come from observational studies, primarily case-control
ethosuximide could exacerbate tonic-clonic seizures in and cohort studies. Assessment of the risk to benefit ratio of
children,8 and somewhat more consistent evidence that AEDs therefore requires appraisal of evidence from RCTs
vigabatrin can precipitate myoclonus and absence seizures.8 and observational studies. Below, we will discuss evidence
It is perhaps here that we have the most satisfactory evidence from RCTs and systematic reviews of RCTs followed by
to prefer specific AEDs, such as valproic acid, for the evidence from observational studies. The latter is restricted
treatment of adults with generalised epilepsies. to data on teratogenicity, the risk of which plays a major part
Ultimately, the choice of an individual AED will be in the choice of drug for women of childbearing age—a third
determined by an individual risk-benefit assessment in of people with epilepsy.

730 Neurology Vol 3 December 2004 http://neurology.thelancet.com

For personal use. Only reproduce with permission from Elsevier Ltd
 Current antiepileptic drug treatment
Review

Table 1. Results of systematic reviews comparing standard-AED monotherapy

Drug comparison Number of Number of Drug for which event more Outcome: hazard ratio (95% CI)
trials patients likely (hazard ratio >1)
Treatment failure 12 month remission First seizure
Focal seizures
Carbamazepine vs valproic acid 5 813 Valproic acid 1·00 (0·79–1·26) 0·82 (0·67–1·00) 1·22 (1·04–1·44)
Carbamazepine vs phenobarbital 4 519 Phenobarbital 1·60 (1·18–2·17) 1·03 (0·72–1·49) 0·71 (0·55–0·91)
Phenytoin vs valproic acid 4 186 Phenytoin 1·23 (0·77–1·98) 1·02 (0·68–1·54) 0·81 (0·59–1·11)
Phenytoin vs phenobarbital Unknown Unknown Phenobarbital 1·97 (1·09–1·97) 0·98 (0·69–1·39) 0·78 (0·61–1·04)
Generalised seizures
Carbamazepine vs valproic acid 4 382 Valproic acid 0·98 (0·61–1·29) 0·96 (0·75–1·24) 0·86 (0·68–1·09)
Carbamazepine vs phenobarbital 3 155 Phenobarbital 1·78 (0·87–3·62) 0·61 (0·36–1·03) 1·50 (0·95–2·35)
Phenytoin vs valproic acid 5 341 Phenytoin 0·98 (0·60–1·58) 1·06 (0·71–1·57) 1·03 (0·77–1·39)
Phenytoin vs phenobarbital Unknown Unknown Phenobarbital 4·32 (1·77–10·6) 0·77 (0·46–1·28) 1·12 (0·70–1·78)
Note that 95% CIs that include 1 indicate no statistically significant difference between the outcome measure of the compared AEDs.

Focal epilepsies Lamotrigine has been compared with carbamazepine in

There have been several RCTs that have compared the effects four published trials that recruited patients with focal
of standard AEDs such as carbamazepine, phenytoin, seizures.21–24 In three of these trials, doses were escalated
valproic acid, and phenobarbital, and their results have been according to clinical need,21–23 and all found that
summarised in meta-analyses in a series of Cochrane carbamazepine was significantly more likely to be
systematic reviews.6,13–16 Results from reviews providing data withdrawn, mainly because of side-effects. However, none of
on the subgroup of patients with focal-onset seizures are these trials found a significant difference in terms of seizure
summarised in table 1. control. Two found no difference in the proportion of
A meta-analysis of trials comparing carbamazepine and patients who were seizure free in the final 24 weeks of a
valproic acid found no difference for time to treatment 48 week trial,21,22 or the final 16 weeks of a 24 week trial;23
failure (hazard ratio [HR] 1·00, 95% CI 0·79–1·26), and a however, these outcomes lacked both statistical power and
significant advantage for carbamazepine for time to clinical usefulness. Two trials found no difference in time to
12 months of remission (HR 0·82, 95% CI 0·67–1·00) and first seizures after randomisation. However, estimates were
time to first seizure (HR 1·22, 95% CI 1·04–1·44). A meta- in favour of carbamazepine and confidence intervals were
analysis that compared carbamazepine with phenobarbital wide,21,22 and although lamotrigine seems better tolerated
found no significant advantage for either drug. The than carbamazepine, we have insufficient evidence about
estimates indicated that phenobarbital was more likely to seizure control to inform a choice between these two drugs
be withdrawn, presumably due to side-effects, but fared for patients with focal seizures.
better for time to first seizure. Similarly, a meta-analysis Gabapentin and carbamazepine have been compared in
comparing phenytoin found no significant differences, one monotherapy trial recruiting patients with focal
although the estimates suggested that phenytoin was more seizures.25 Patients were randomly assigned to receive 600 mg
likely to be withdrawn than carbamazepine, but fared carbamazepine daily, or one of three doses of gabapentin
better for time to first seizure. A meta-analysis comparing daily (300 mg, 900 mg, or 1800 mg). The primary outcome
phenobarbital and phenytoin found that phenobarbital was was time to exit, reasons for which included a single tonic-
significantly more likely to be withdrawn, but that there clonic seizure, three complex partial seizures, and status
was no significant difference for 12 months of remission or epilepticus. No significant difference was found between the
time to first seizure. We therefore have limited data on the groups taking carbamazepine and 900 mg or 1800 mg
comparison between standard AEDs. Where estimates gabapentin daily. Significant differences were found between
indicate no difference, it is important to understand that doses of gabapentin and between the 300 mg doses of
the confidence intervals are wide, hence important gabapentin and carbamazepine. The protocol for this trial
differences have not been excluded and equivalence cannot deviates substantially from everyday clinical practice, and
be inferred. Meta-analyses that compare carbamazepine the results therefore do little to inform clinical practice.
with valproic acid do, however, support guidelines that Topiramate has been compared with standard AEDs in
recommend carbamazepine as first-line treatment for one trial in which clinicians chose either carbamazepine or
patients with focal seizures.17,18 valproic acid as the preferred standard AED.26 The subgroup
In the past decade, many new AEDs have been developed for whom carbamazepine was chosen as the standard drug
and marketed, including gabapentin, lamotrigine, had focal seizures and were randomly assigned to
levetiracetam, oxcarbazepine, topiramate, and zonisamide. receive 600 mg carbamazepine daily, 100 mg topiramate
All have shown efficacy as add-on treatments for patients daily, or 200 mg topiramate daily. For the analyses, data for
with drug-resistant focal seizures when compared with the 100 mg and 200 mg topiramate groups were pooled. No
placebo.19,20 Fewer head-to-head trials have been undertaken difference was found between topiramate and
in which standard and newer AEDs have been compared. carbamazepine for time to treatment withdrawal, time to

Neurology Vol 3 December 2004 http://neurology.thelancet.com 731

For personal use. Only reproduce with permission from Elsevier Ltd
 Review Current antiepileptic drug treatment

Idiopathic generalised epilepsies

Results from the Cochrane reviews of
Two or more spontaneous standard AEDs for patients with
seizures generalised seizures are also
summarised in table 1. Trials included
in these reviews recruited patients with
generalised tonic-clonic seizures with
Focal-onset Generalised- or Unclassifiable or without other types of generalised
seizures onset seizures seizures seizure. Most trials predate the ILAE
classification of epileptic syndromes;32
hence, in a large proportion of trials
patients were classified according to
seizure types but not syndromes. In
First-line AED Woman of child- Not woman of addition, significant numbers of
Carbamazepine bearing years child-bearing patients are likely to have been
years
misclassified. For example, on closer
Alternatives First-line AED
Lamotrigine Lamotrigine First-line AED inspection of trials comparing
Phenytoin Valproic acid carbamazepine and valproic acid, a
Topiramate Alternatives significant proportion of patients over
Valproic acid Valproic acid Alternatives the age of 25 years who presented with
Topiramate Lamotrigine seizures were classified as having
Topiramate
generalised seizures, which is extremely
unlikely. It is more likely that these
Figure 2. A suggested treatment algorithm.
patients had focal seizures and had
been misclassified. Meta-analysis of
first seizure, or the proportion of patients seizure free for the trials shows no significant differences between standard
last 6 months of follow-up. Although this trial found no drugs for time to treatment withdrawal, time to first seizure,
difference for these outcomes, confidence intervals are wide or time to 12 month remission. Existing data from RCTs do
and the results do not confirm equivalence; hence, this trial not inform a choice among standard AEDs for patients with
falls short of providing data that informs a choice between an idiopathic generalised epilepsy. Guidelines recommend
carbamazepine and topiramate. valproic acid as a treatment of first choice for patients with
Initial monotherapy trials of oxcarbazepine compared it this type of epilepsy, supported by data from observational
with carbamazepine.27,28 These trials found no difference studies that indicate a good response, particularly for
between these two drugs, but lacked power to show juvenile myoclonic epilepsy,33,34 as well as reports that
equivalence. In addition, results of the largest trial were indicate worsening of myoclonus and absence seizures for
confounded by the exclusion of 70 of the 235 randomly patients taking phenytoin or carbamazepine.9–11
assigned patients from the efficacy analyses.27 There are few studies of the new AEDs in patients with
More recently, oxcarbazepine has been compared with an idiopathic generalised epilepsy. Lamotrigine is
phenytoin in two monotherapy RCTs29,30 and valproic acid recommended in guidelines as an alternative to valproic
in one monotherapy RCT.29 All three trials were of similar acid,17 but there is no RCT evidence to support this use.
design, and lasted 56 weeks. Primary efficacy outcomes were Similarly, there is no RCT evidence to support the use of
the proportion of patients who were seizure free, and the levetiracetam and no high quality studies to support the use
primary tolerability outcome was time to withdrawal due to of zonisamide. Topiramate has been compared with valproic
side-effects. None of these trials found a difference between acid in one trial, which has already been described above.26
oxcarbazepine and its comparators in the proportion of No difference was found for the primary outcomes, although
patients who were seizure free. Although no difference was the results did not confirm equivalence, and fall short of
found when compared with valproic acid for time to informing a choice between topiramate and valproic acid.
treatment withdrawal due to side-effects,31 when compared
with phenytoin, oxcarbazepine fared significantly better for Unclassified epilepsy
time to treatment withdrawal due to side-effects in both There is no evidence from RCTs to inform the management
adults30 and children.29 Although no difference was found of unclassified epilepsy. Such patients are excluded from
for the primary efficacy outcomes, these trials also lacked many trials. Guidelines recommend the use of broad-
the power to exclude the possible existence of an important spectrum drugs such as valproic acid and lamotrigine
difference, and do not prove equivalence. (figure 2).
In summary, for patients with focal seizures, we have
evidence to support a choice of carbamazepine from the Long-term adverse effects of AEDs
standard AEDs. We have no reliable evidence that any of the AEDs have been associated with a wide range of chronic
newer drugs are either superior or inferior to carbamazepine adverse effects that are often identified many years after a
or other standard drugs. drug’s introduction into clinical practice. Of the new AEDs,

732 Neurology Vol 3 December 2004 http://neurology.thelancet.com

For personal use. Only reproduce with permission from Elsevier Ltd
 Current antiepileptic drug treatment
Review

vigabatrin therapy is associated with irreversible, typically in those with idiopathic generalised epilepsy. In these
symptomless, bilateral loss of concentric visual fields in 40% patients, lamotrigine may be effective, but some women may
of chronically exposed patients.35–37 Twice yearly visual-field require valproic acid for adequate seizure control,34 which is
assessments are recommended in patients continuing with an important consideration in pregnancy. The potential
therapy. With other agents available, vigabatrin should be place for use of topiramate and levetiracetam in these
used as a drug of last resort in adults with partial epilepsy. patients is uncertain.
The efficacy of vigabatrin for infantile spasms,38 particularly
when associated with tuberous sclerosis,39 has been shown. Unsuccessful monotherapy
The drug still has an important role in paediatric practice, Before considering alternative medication, consideration
although visual-field loss paradoxically is difficult to reliably should be given as to whether the optimum dose has been
assess in children. reached. Seizure response and adverse effects should be used
Recent concerns have focused on bone density and to guide therapy, rather than serum drug concentrations, for
osteoporosis. Concern began with the discovery of high rates all drugs except phenytoin. In the case of phenytoin, the
of osteomalacia in institutionalised patients with epilepsy checking of serum concentrations may be helpful, although
who were taking AEDs long-term.40 There is, however, no the therapeutic range is only a rough guide; many patients
evidence of higher fracture rates in patients not in obtain seizure freedom with concentrations below this
institutions when compared with the general population, range, whereas others tolerate and require concentrations
once fractures occurring during seizures are accounted for.41 above this range. For patients who continue to have seizures
Most,42–44 but not all,45,46 cross-sectional studies have found a despite an adequate dose of an AED, an explanation of non-
reduction in bone-mineral density in patients with epilepsy compliance should be considered. Thought must be given as
who are taking AEDs. Because of their design, results of these to whether the diagnosis of epilepsy is correct, whether the
studies are confounded and it is not possible to reliably epilepsy syndrome has been correctly classified, and whether
untangle any potential effects of AEDs on bone health from the appropriate AED for the epilepsy syndrome has been
other factors associated with epilepsy, such as reduced chosen.
physical activity, diet, or comorbid illness. Whether there are The longer patients continue to have seizures after their
differences in the effect on bone health of the newer drugs initial diagnosis, the lower the probability of subsequent
when compared with older agents is unclear. Consensus remission.2 Whereas about 70% of patients diagnosed with
guidelines argue against screening with bone densitometry epilepsy can expect long-term remission,2 only 16% of patients
in patients whose only risk factor for osteoporotic fracture is who find the first drug to be ineffective in suppressing seizure
epilepsy.47 activity (as opposed to poor tolerability) can subsequently
expect to become free from seizures.51 Therefore, in patients
Teratogenicity who have seizures despite a good trial of their first AED, a
The risk of teratogenic effects has an important influence on cautious prognosis can be given. Patients who fail a second
the choice of AED for women of childbearing age. AED have a small chance of obtaining seizure freedom.52
Observational studies have consistently shown an increased Guidelines recommend that after efficacy failure of two
risk of teratogenic effects associated with the standard drugs. appropriate AEDs, surgical options should be considered.17
However, studies have tended to be small and have used
various designs, both prospective and retrospective, with few Alternative monotherapy or combination therapy?
being population based. As a result, estimates have varied, After monotherapy failure the physician has a choice: to add
although there seems to be a two to threefold increase in the on a new therapy or to replace the current drug. The
rate of major fetal malformations compared with a potential advantages of switching include the limiting of
background rate of 2–3%.48 Of the standard AEDs, valproic toxicity and the separate assessment of drugs. By contrast,
acid is of increasing concern because it is associated with a polytherapy could offer theoretical advantages of more rapid
higher risk of congenital abnormalities, including neural- seizure control and higher seizure-freedom rates through the
tube defects.48 There is also growing concern about exploitation of pharmacodynamic interactions. The first trial
developmental delay in children exposed to valproic acid in that specifically tried to compare add-on with alternative
utero.49 In view of these concerns, recent guidelines therapy has recently been published.53 After monotherapy
recommend particular caution when prescribing valproic failure, patients were randomly allocated to adjunctive
acid to women of childbearing age.17 Fewer data are available therapy or alternative monotherapy, although the choice of
on the teratogenic effects of newer AEDs. To provide drug and dosing strategy were chosen by the physician. In
information about the teratogenic effects of new and the assigned treatment groups no significant difference was
standard AEDs, several pregnancy registers have been found between time to assigned treatment failure or for the
established. Lamotrigine is the new AED for which there is cumulative number of patients remaining seizure free for
the largest amount of data, and interim results of the UK 12 months, with remission rates in both groups at about
pregnancy register suggest that the risk of a major congenital 15%. Recruitment was difficult and therefore the trial lacked
abnormality associated with lamotrigine is low and similar the power to exclude a clinically important difference.
to that of carbamazepine.50 Uncertainty of whether alternative monotherapy or
Although use of valproic acid may be readily avoided in combination therapy is the best therapeutic option therefore
women with focal epilepsies, to do so may be more difficult continues. If seizure remission is achieved with combination

Neurology Vol 3 December 2004 http://neurology.thelancet.com 733

For personal use. Only reproduce with permission from Elsevier Ltd
 Review Current antiepileptic drug treatment

Table 2. Results of the systematic reviews of new drugs as add-on therapy in partial-onset epilepsy

AED Number Daily dose Effect estimate (95% CI) for a 50% reduction Relative risk (95% CI) of treatment
of trials range (mg/day) in seizures compared with placebo* withdrawal compared with placebo*

Gabapentin 5 600–1800 OR 1·93 (1·37–2·7) 1·05 (0·68–1·61)

Tiagabine 3 16–56 RR 3·16 (1·97–5·07) 1·80 (1·2–2·7)
Topiramate 9 200–1000 RR 3·32 (2·52–4·39) 2·06 (1·38–3·08)
Lamotrigine 11 200–500 OR 2·71 (1·87–3·61) 1·10 (0·81–1·50)
Levetiracetam 4 1000–3000 OR 3·78 (2·62–5·44) 1·21 (0·88–1·66)
Oxcarbazepine 2 600–2400 OR 2·51 (1·88–3·33) 1·72 (1·35–2·18)
Zonisamide 3 400 OR 2·46 (1·61–3·74) 1·46 (1·02–2·62)
*All results are calculated for all doses. Note that 95% CIs that include 1 indicate no statistically significant difference. OR=odds ratio; RR=risk ratio.

therapy, AED withdrawal does carry a risk of return of outlined in table 3. For valproic acid, topiramate,
seizures;54 remaining on combination therapy carries the gabapentin, and levetiracetam, a dominant mechanism of
modest chance that monotherapy with the second agent action has not been convincingly shown. Although the
could give seizure freedom. Patients must be counselled appropriate drugs for syndromic diagnosis should be used,
about the potential benefits and risks, and should be allowed rational therapy suggests avoiding combination therapy with
to control the decision-making process. two AEDs that have the same primary modes of action.

New AEDs as add-on treatments Conclusion

Table 2 summarises the results of intention-to-treat analyses The evidence to support the prescribing of new AEDs is
from systematic reviews in patients with drug-refractory sparse, indicating that most RCTs answer questions focused
localisation-related seizures. Although all these drugs show on regulatory requirements rather than clinical use. A large
efficacy, caution should be used in comparing outcomes as pragmatic RCT (the Standard and New Antiepileptic Drugs
populations differed and doses used varied. [SANAD] trial), which is comparing new and standard
AEDs, will report in the next 2 years and should partially
AED combinations after monotherapy failure address these concerns. In asbence of good evidence for the
There is no evidence in human beings to suggest that a difference of efficacy of different AEDs, our practice will be
particular combination of AEDs is more efficacious than guided by concerns about potential unwanted adverse
another. A rational approach is commonly advocated; if two effects. However, even here, poor methodology of studies
AEDs are used together, their combined effect will be greater hampers decision making. These issues particularly affect
if the known mechanisms of action are different. women of childbearing age.
Unfortunately, mechanisms of action for AEDs are only In the future there is a need to improve the quality of
partially known and are not universally agreed; subtle but clinical data on the efficacy and harm of AEDs. There is also
significant differences may mean that combining drugs with a developing agenda in the area of pharmacogenetics which,
similar mechanisms may be effective. Avoidance of once methodological issues are addressed, may help clarify
combinations with pharmacokinetic interactions simplifies choices from the availability of new AEDs.
management. The putative modes of action of AEDs are
Authors’ contributions
All authors contributed equally to this review.
Table 3. Summary of principal modes of action of AEDs
Conflict of interest
Drug GABA- Blockade of Blockade of Unknown DMcC, DC, and AM have received hospitality and/or sponsorship to
mediated voltage-gated voltage-gated attend conferences from Sanofi-Synthélabo, GlaxoSmithKline,
inhibition Na channels Ca channels Novartis, Pfizer, Jansen Cilag, and UCB Pharma. DC’s university
department has also received research support from these companies.
Phenobarbital Yes ·· ·· ··
Phenytoin ·· Yes ·· ·· Role of the funding source
Carbamazepine ·· Yes ·· ·· No funding was received for this review.
Valproic acid Yes Yes Yes ··
Ethosuximide ·· ·· Yes ··
Benzodiazepines Yes ·· ·· ·· Search strategy and selection criteria
Vigabatrin Yes ·· ·· ·· The Cochrane Library (2004 issue 1) and the Cochrane
Lamotrigine ·· Yes ·· ·· Epilepsy Group controlled trials register were searched. The
Gabapentin Yes ?Yes ?Yes ·· latter was compiled from searches of MEDLINE (1966–2004)
Levetiracetam ·· ·· ·· Yes as well as journal handsearching, which was undertaken by the
Topiramate ?Yes Yes Yes ·· Cochrane Epilepsy Group, and the Cochrane Collaboration as
Oxcarbazepine ·· Yes ·· ·· a whole. When found, results of systematic reviews were
Tiagabine Yes ·· ·· ·· selected for inclusion. Where no systematic reviews were
found, randomised controlled trials were selected.

734 Neurology Vol 3 December 2004 http://neurology.thelancet.com

For personal use. Only reproduce with permission from Elsevier Ltd
 Current antiepileptic drug treatment
Review

References 19 Marson AG, Kadir ZA, Chadwick DW. New 36 Luchetti A, Amadi A, Gobbi G. Visual field defects
1 National Institute of Clinical Excellence. Newer antiepileptic drugs: a systematic review of their associated with vigabratin monotherapy in children.
drugs for adults with epilepsy, full guidance (TA76). efficacy and tolerability. BMJ 1996; 313: 1169–74. J Neurol Neurosurg Psychiatry 2000; 69: 566.
March 2004. www.nice.org.uk/pdf/ 20 Marson AG, Hutton JL, Leach JP, et al. 37 Sills GJ, Butler E, Forrest G, Ratnaraj N, Patsalos PN,
TA076fullguidance.pdf (accessed Nov 2, 2004). Levetiracetam, oxcarbazepine, remacemide and Brodie MJ. Vigabatrin, but not gabapentin or
2 Annegers JF, Hauser WA, Elveback LR. Remission of zonisamide for drug resistant localization-related topiramate, produces concentration-related effects
seizures and relapse in patients with epilepsy. epilepsy: a systematic review. Epilepsy Res 2001; on enzymes and intermediates of the gaba shunt in
Epilepsia 1979; 20: 729–37. 46: 259–70. rat brain and retina. Epilepsia 2003; 44: 886–92.
3 Medical Research Council Antiepileptic Drug 21 Brodie MJ, Richens A, Yuen AW. Double-blind 38 Elterman RD, Shields WD, Mansfield KA,
Withdrawal Study Group. Randomised study of comparison of lamotrigine and carbamazepine in Nakagawa J, US Infantile Spasms Vigabatrin Study
antiepileptic drug withdrawal in patients in newly diagnosed epilepsy. UK Lamotrigine/ Group. Randomized trial of vigabatrin in patients
remission. Lancet 1991; 337: 1175–80. Carbamazepine Monotherapy Trial Group. Lancet with infantile spasms [comment]. Neurology 2001;
4 Walker MC, White HS, Sander JW. Disease 1995; 345: 476–79. 57: 1416–21.
modification in partial epilepsy. Brain 2002; 22 Brodie MJ, Overstall PW, Giorgi L. Multicentre, 39 Chiron C, Dumas C, Jambaque I, Mumford J,
125: 1937–50. double-blind, randomised comparison between Dulac O. Randomized trial comparing vigabatrin
5 Berg AT, Shinnar S. Do seizures beget seizures? An lamotrigine and carbamazepine in elderly patients and hydrocortisone in infantile spasms due to
assessment of the clinical evidence in humans. with newly diagnosed epilepsy. The UK Lamotrigine tuberous sclerosis. Epilepsy Res 1997; 26: 389–95.
J Clin Neurophysiol 1997; 14: 102–10. Elderly Study Group. Epilepsy Res 1999; 37: 81–87. 40 Dent CE, Richens A, Rowe DJ, Stamp TC.
6 Temkin NR. Antiepileptogenesis and seizure 23 Nieto-Barrera M, Brozmanova M, Capovilla G, et al. Osteomalacia with long-term anticonvulsant therapy
prevention trials with antiepileptic drugs: meta- A comparison of monotherapy with lamotrigine or in epilepsy. BMJ 1970; 4: 69–72.
analysis of controlled trials. Epilepsia 2001; carbamazepine in patients with newly diagnosed 41 Vestergaard P, Tigaran S, Rejnmark L, Tigaran C,
42: 515–24. partial epilepsy. Epilepsy Res 2001; 46: 145–55. Dam M, Mosekilde L. Fracture risk is increased in
7 Musicco M, Beghi E, Solari A, Viani F. Treatment of 24 Reunanen M, Dam M, Yuen AW. A randomised epilepsy. Acta Neurol Scand 1999; 99: 269–75.
first tonic-clonic seizure does not improve the open multicentre comparative trial of lamotrigine 42 Farhat G, Yamout B, Mikati MA, Demirjian S,
prognosis of epilepsy. First Seizure Trial Group and carbamazepine as monotherapy in patients with Sawaya R, El-Hajj Fuleihan G. Effect of antiepileptic
(FIRST Group). Neurology 1997; 49: 991–98. newly diagnosed or recurrent epilepsy. Epilepsy Res drugs on bone density in ambulatory patients.
8 Perucca E, Gram L, Avanzini G, Dulac O. 1996; 23: 149–55. Neurology 2002; 58: 1348–53.
Antiepileptic drugs as a cause of worsening seizures. 25 Chadwick DW, Anhut H, Greiner MJ, et al. A 43 Pack AM, Olarte LS, Morrell MJ, Flaster E, Resor SR,
Epilepsia 1998; 39: 5–17. double-blind trial of gabapentin monotherapy in Shane E. Bone mineral density in an outpatient
9 Snead OC 3rd, Hosey LC. Exacerbation of seizures in patients for newly-diagnosed partial seizures. population receiving enzyme-inducing antiepileptic
children by carbamazepine. N Engl J Med 1985; Neurology 1998; 51: 1282–88. drugs. Epilepsy Behav 2003; 4: 169–74.
313: 916–21. 26 Privitera MD, Brodie MJ, Mattson RH, et al. 44 Sato Y, Kondo I, Ishida S, et al. Decreased bone mass
10 Shields WD, Saslow E. Myoclonic, atonic, and Topiramate, carbamazepine and valproate and increased bone turnover with valproate therapy
absence seizures following institution of monotherapy: double-blind comparison in newly in adults with epilepsy. Neurology 2001; 57: 445–49.
carbamazepine therapy in children. Neurology 1983; diagnosed epilepsy. Acta Neurol Scand 2003;
107: 165–75. 45 Akin R, Okutan V, Sarici U, Altunbas A, Gokcay E.
33: 1487–89. Evaluation of bone mineral density in children
11 Liporace JD, Sperling MR, Dichter MA. Absence 27 Dam M, Ekberg R, Loyning Y, Waltimo O,
receiving antiepileptic drugs. Pediatr Neurol 1998;
seizures and carbamazepine in adults. Epilepsia 1994; Jakobsen K. A double-blind study comparing
oxcarbazepine and carbamazepine in patients with 19: 129–31.
35: 1026–28. 46 Valimaki MJ, Tiihonen M, Laitinen K, et al. Bone
newly diagnosed, previously untreated epilepsy.
12 Commission on Antiepileptic Drugs. Considerations mineral density measured by dual-energy x-ray
Epilepsy Res 1989; 3: 70–76.
on designing clinical trials to evaluate the place of absorptiometry and novel markers of bone
new antiepileptic drugs in the treatment of newly 28 Reinikainen KJ, Keranen T, Halonen T,
Komulainen H, Riekkinen PJ. Comparison of formation and resorption in patients on antiepileptic
diagnosed and chronic patients with epilepsy. drugs. J Bone Miner Res 1994; 9: 631–37.
Epilepsia 1998; 39: 799–803. oxcarbazepine and carbamazepine: a double-blind
study. Epilepsy Res 1987; 1: 284–89. 47 Royal College of Physicians Consensus Group.
13 Marson AG, Williamson PR, Clough H, Hutton JL, Osteoporosis—clinical guidelines: summary and
Chadwick DW, Epilepsy Monotherapy Trial Group. 29 Guerreiro MM, Vigonius U, Pohlmann H, et al. A
double-blind controlled clinical trial of recommendations. London: Royal College of
Carbamazepine versus valproate monotherapy for Physicians, 1999: 1–13.
epilepsy: a meta-analysis. Epilepsia 2002; 43: 505–13. oxcarbazepine versus phenytoin in children and
adolescents with epilepsy. Epilepsy Res 1997; 48 Yerby MS. Management issues for women with
14 Tudur Smith C, Marson AG, Williamson PR. epilepsy: neural tube defects and folic acid
Carbamazepine versus phenobarbitone 27: 205–13.
30 Bill PA, Vigonius U, Pohlmann H, et al. A double- supplementation. Neurology 2003;
monotherapy for epilepsy (Cochrane Review). In: 61 (suppl 2): S23–26.
The Cochrane Library, Issue 4, 2004. Chichester, UK: blind controlled clinical trial of oxcarbazepine versus
John Wiley & Sons, Ltd. phenytoin in adults with previously untreated 49 Adab N, Jacoby A, Smith D, Chadwick D. Additional
epilepsy. Epilepsy Res 1997; 27: 195–204. educational needs in children born to mothers with
15 Taylor S, Tudur Smith C, Williamson PR, epilepsy. J Neurol Neurosurg Psychiatry 2001;
Marson AG. Phenobarbitone versus phenytoin 31 Christe W, Kramer G, Vigonius U et al. A double-
blind controlled clinical trial: oxcarbazepine versus 70: 15–21.
monotherapy for partial onset seizures and
generalized onset tonic-clonic seizures (Cochrane sodium valproate in adults with newly diagnosed 50 Morrow JI, Russell A, Irwin B, et al. The UK Epilepsy
Review). In: The Cochrane Library, Issue 4, 2004. epilepsy. Epilepsy Res 1997; 26: 451–60. and Pregnancy Register Interim results. In:
Chichester, UK: John Wiley & Sons, Ltd. 32 Commission on Classification and Terminology of Proceedings of the Association of British
the International League Against Epilepsy. Proposal Neurologists Autumn Meeting, 1–3 October 2003.
16 Tudur Smith C, Marson AG, Williamson PR. J Neurol Neurosurg Psychiatry 2004; 75: 797–806.
Phenytoin versus valproate monotherapy for partial for revised classification of epilepsies and epileptic
onset seizures and generalized onset tonic-clonic syndromes. Epilepsia 1989; 30: 389–99. 51 Kwan P, Brodie MJ. Early identification of refractory
seizures (Cochrane Review). In: The Cochrane 33 Delgado-Escueta AV, Enrile-Bacsal F. Juvenile epilepsy. N Engl J Med 2000; 342: 314–19.
Library, Issue 4, 2004. Chichester, UK: John Wiley & myoclonic epilepsy of Janz. Neurology 1984; 52 Kwan P, Brodie MJ. Epilepsy after the first drug fails:
Sons, Ltd. 34: 285–94. substitution or add-on? Seizure 2000; 9: 464–68.
17 Scottish Intercollegiate Guidelines Network. 34 Nicolson A, Appleton RE, Chadwick DW, Smith DF. 53 Beghi E, Gatti G, Tonini C, et al. Adjunctive therapy
Diagnosis and management of epilepsy in adults. A The relationship between treatment with valproate, versus alternative monotherapy in patients with
national clinical guideline. April 2003. lamotrigine, and topiramate and the prognosis of the partial epilepsy failing on a single drug: a
www.sign.ac.uk/pdf/sign70.pdf (accessed Nov 2, idiopathic generalised epilepsies. multicentre, randomised, pragmatic controlled trial.
2004). J Neurol Neurosurg Psychiatry 2004; 75: 75–79. Epilepsy Res 2003; 57: 1–13.
18 Wallace H, Shorvon SD, Hopkins A, O’Donoghue M. 35 Iannetti P, Spalice A, Perla FM, Conicella E, 54 Deckers CL, Czuczwar SJ, Hekster YA, et al.
Guidelines for the clinical management of adults Raucci U, Bizzarri B. Visual field constriction in Selection of antiepileptic drug polytherapy based on
with poorly controlled epilepsy. London: Royal children with epilepsy on vigabatrin treatment. mechanisms of action: the evidence reviewed.
College of Physicians, 1997. Pediatrics 2000; 106: 838–42. Epilepsia 2000; 41: 1364–74.

Neurology Vol 3 December 2004 http://neurology.thelancet.com 735

For personal use. Only reproduce with permission from Elsevier Ltd