Intended for healthcare professionals

Clinical Review State of the Art

Drug treatment of epilepsy in adults

BMJ 2014; 348 doi: (Published 28 February 2014) Cite this as: BMJ 2014;348:g254
  1. Dieter Schmidt, head, Epilepsy Research Group1,
  2. Steven C Schachter, chief academic officer, Consortia for Improving Medicine with Innovation and Technology 2
  1. 1Epilepsy Research Group, Goethestr. 5, 14163 Berlin, Germany
  2. 2Departments of Neurology, Beth Israel Deaconess Medical Center, Massachusetts General Hospital and Harvard Medical School, Boston, MA, USA
  1. Correspondence to: D Schmidt dbschmidt{at}


Epilepsy is a serious, potentially life shortening brain disorder, the symptoms of which can be successfully treated in most patients with one or more antiepileptic drug. About two in three adults with new onset epilepsy will achieve lasting seizure remission on or off these drugs, although around half will experience mild to moderately severe adverse effects. Patients with epilepsy, especially the 20-30% whose seizures are not fully controlled with available drugs (drug resistant epilepsy), have a significantly increased risk of death, as well as psychiatric and somatic comorbidities, and adverse effects from antiepileptic drugs. Newer drugs have brought more treatment options, and some such as levetiracetam cause fewer drug interactions and less hypersensitivity than older ones. However, they do not reduce the prevalence of drug resistant epilepsy or prevent the development of epilepsy in patients at high risk, such as those with a traumatic brain injury. The development of antiepileptic drugs urgently needs to be revitalized so that we can discover more effective antiseizure drugs for the treatment of drug resistant epilepsy, including catastrophic forms. Antiepileptogenic agents to prevent epilepsy before the first seizure in at risk patients and disease modifying agents to control ongoing severe epilepsy associated with progressive underlying disease are also needed.


Summary points

  • Roughly 70-80% of adults with new onset epilepsy will become seizure free with current antiepileptic drugs, although around half will experience adverse effects

  • About 20-30% continue to have drug resistant epilepsy with seizures, adverse effects, increased mortality, and substantial psychiatric and somatic comorbidities

  • Newer antiepileptic drugs have brought more treatment options and increased ease of use but do not reduce the frequency of drug resistant epilepsy or prevent epilepsy in those at risk

  • There is an urgent need to revitalize the development of antiepileptic drugs to discover more effective drugs for the treatment of drug resistant epilepsy

  • Antiepileptogenic compounds that prevent epilepsy before the first seizure in at risk patients are needed, as well as disease modifying drugs to control ongoing severe epilepsy and its comorbidities

Epilepsy is a heterogeneous and serious brain disorder with multifactorial origins and manifestations. It comprises many seizure types and epilepsy syndromes,1 some of which are life shortening.2 Although 70-80% of patients with new onset epilepsy have complete seizure control with current antiepileptic drugs,3 4 unmet treatment needs remain. About half of patients report at least one adverse effect during treatment with first line antiepileptic drugs.5 6 Drug resistant epilepsy occurs in 20-30% of patients newly diagnosed with epilepsy, depending on the definition used.7 Long term observations have shown that 14% of patients with new onset childhood epilepsy who remain in remission for many years will develop refractory epilepsy while still being treated with antiepileptic drugs.4 Treatment is empirical and often based on trial and error. Seizures are widely recognized as the clinical hallmark of epilepsy, but epileptogenesis—the disease process by which epilepsy develops after brain insults or as a result of gene mutations—begins before the first seizure and probably continues after the onset of seizures.8 9 Although current antiepileptic drugs achieve symptomatic seizure relief, which is why they are more appropriately called antiseizure drugs, they do not prevent or reverse the pathological process that underlies human epilepsy or other clinical manifestations of epilepsy, such as the comorbidity of epilepsy. They therefore do not prevent the development of epilepsy, even in patients at high risk (for example, after brain injury or craniotomy),10 and nor do they exert disease modifying effects that prevent or reverse drug resistant epilepsy. Also, they do not prevent or eliminate the substantial behavioral, cognitive, and somatic comorbidities seen in many patients with epilepsy.11

These life limiting currently unmet needs provide a roadmap for the development of more effective antiseizure drugs, as well as for disease modifying and antiepileptogenic drugs.9 12 13 In this review, we critically assess the current drug treatment of epilepsy in adults and briefly examine prospects for tackling current unmet needs. Drug treatment in children with epilepsy is covered elsewhere.14 Non-drug based treatments for epilepsy, such as epilepsy surgery, diets, and brain stimulation, are beyond the scope of this review.


Worldwide about 65 million people have epilepsy,15 making it the most common neurological disorder after stroke and a major burden for public health systems.16 17 The prevalence of epilepsy varies by population. In developed countries, the annual incidence of epilepsy is nearly 50 per 100 000 population and prevalence is around 700 per 100 000.18 In low and middle income countries, such estimates are generally higher. For example, in Ethiopia, a developing country, the prevalence of epilepsy is as high as 29.5/1000 (95% confidence interval 20.5/1000 to 40.9).15 17

Sources and selection criteria

References for this review were identified through searches of publications listed by PubMed and ScienceDirect from 1 January 1980 to 1 September 2013. We used the search terms “epilepsy”, “treatment”, “antiepileptic drugs”, “efficacy”, “effectiveness”, “antiepileptogenesis”, “antiepileptogenic drugs”, “disease modification”, “adverse effects”, “antiepileptic drugs discovery”, “antiepileptic drugs preclinical development”, “antiepileptic drugs clinical development”, and “humans”. References were also identified from relevant review articles and through searches of the authors’ files. Only articles published in English were reviewed. We excluded articles published in non-peer reviewed journals. The final reference list was based on relevance to the topics covered in the review. We included publications published between 1983 and 2013, including meta-analyses. Publications of evidence classes I-IV were included because of the limited evidence base on the drug treatment of epilepsy.19

When to start treatment

The Multicentre Study of Early Epilepsy and Single Seizures trial shows that starting antiepileptic drugs after a first seizure reduces the risk of a second seizure compared with no treatment or delayed treatment.20 Immediate treatment increased the time to second seizure (hazard ratio 1.3, 95% confidence interval 1.1 to 1.6) and first occurrence of a tonic-clonic seizure (1.5, 1.2 to 1.8). It also reduced the time to achieve two year remission of seizures (P=0.023).20 Table 1 details factors that place patients at high risk for recurrence. It is justifiable to recommend treatment after the first seizure in patients at higher risk of recurrence because such patients have a slightly better long term outcome with early versus delayed treatment (table 2). Accordingly, the new practical clinical definition of epilepsy proposed by the International League Against Epilepsy (ILAE) includes certain patients after their first seizure. Such patients are those with a probability of further seizures “similar to the general recurrence risk (≥60%) after two unprovoked seizures, occurring over the next 10 years,” or, as in the previous definition, those patients who have had two unprovoked seizures more than 24 hours apart.21 One important consequence of the revised definition of epilepsy is that all clinicians who encounter patients after a first seizure need to be familiar with the varied clinical presentations of seizures, especially those that are non-convulsive, as well as the appropriate investigations to determine the underlying cause.22

Table 1

Prognostic index from the Multicentre Study of Early Epilepsy and Single Seizures trial* 20

View this table:
Table 2

Estimates of seizure recurrence risk from the Multicentre Study of Early Epilepsy and Single Seizures trial*

View this table:

For patients diagnosed as having epilepsy,22 treatment with an antiepileptic drug is usually recommended, especially if further seizures might cause serious morbidity or mortality. Underlying this standard recommendation is a 73% risk of seizure recurrence (95% confidence interval 59% to 87%) within four years of two unprovoked seizures.23 The risk of a third seizure is nearly twice as high in patients whose seizures have a known cause as in those with idiopathic or cryptogenic seizures,23 as defined by the ILAE.24 Nevertheless, no randomized controlled trials have been performed in unselected patients who have had two or more seizures, and the size of the treatment effect of antiepileptic drugs is currently unknown.25 If the diagnosis of epilepsy is uncertain, it may be best not to start antiepileptic drugs but to undertake further evaluations, such as electroencephalography monitoring, or adopt a watch and wait approach.26

Selecting the first antiepileptic drug

Ideally, antiepileptic drugs should fully control seizures, be well tolerated with no long term safety problems (such as teratogenicity, hypersensitivity reactions, or organ toxicity), and be easy for clinicians to prescribe and patients to take (once or twice daily, no drug interactions, and no need for serum monitoring).26 The introduction of more than 15 antiepileptic drugs since the 1980s has provided more choice but has made it more difficult, even for epilepsy specialists, to select the optimum drug for individual patients because each drug has its advantages and limitations (figs 1-4 ).


Fig 1 Characteristics of widely used first generation antiepileptic drugs for the treatment of epilepsy9 27 28 29


Fig 2 Characteristics of widely used second generation antiepileptic drugs for the treatment of epilepsy9 27 28 29


Fig 3 Characteristics of widely used third generation antiepileptic drugs for the treatment of epilepsy9 27 28 29


Fig 4 Mechanisms of action of antiepileptic drugs, which act by diverse mechanisms, mainly involving modulation of voltage activated ion channels, potentiation of GABA, and inhibition of glutamate.27 30 Approved antiepileptic drugs have effects on inhibitory (left hand side) and excitatory (right hand side) nerve terminals. The antiepileptic efficacy in trials of most of these drugs as initial add-on does not differ greatly, indicating that seemingly similar antiseizure activity can be obtained by mechanisms aimed at diverse targets. However, putative mechanisms of action were determined only after discovering the antiseizure effects; mechanism driven drug discovery has been largely ignored.9 Abbreviations: AMPA, α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid; GABA, γ-aminobutyric acid; GAT-1, sodium dependent and chloride dependent GABA transporter 1; SV2A, synaptic vesicle glycoprotein 2A. Modified, with permission, from Nature Reviews Neurology.28

Effectiveness in new onset epilepsy

Most patients with newly diagnosed epilepsy have a constant course that can be predicted early on.31 32 About 50% of patients with new onset focal or generalized seizures, as internationally defined,24 become seizure free while taking the first appropriately selected and dosed first line antiepileptic drug (assuming that patients have access to healthcare resources; box 1).4 7 33 34

Box 1 Disease burden and treatment gap

The global burden from epilepsy as measured by disability adjusted life years increased by 30% between 1990 and 2010.35 In 2010, the disease burden from epilepsy was higher than that for Alzheimer’s disease and other dementias, multiple sclerosis, and Parkinson’s disease combined.35

The often substantial comorbidity of epilepsy includes injury and drowning, depression and anxiety associated with high suicide rates, and mortality three times the rate expected in the general population, including sudden unexplained death in epilepsy (SUDEP).36 Estimates indicate that most people with epilepsy in developing countries live in rural and remote areas and have no easy access to skilled medical care.37 38 The difference between the number of people with active epilepsy and the number who are being appropriately treated in a given population at a given point in time is known as the treatment gap.37

Epilepsy imposes a large economic burden on patients and their families, particularly in rural and remote regions and the developing world.37 Throughout the world, epilepsy imposes an additional hidden burden associated with stigmatization and discrimination against patients and their families in the community, workplace, school, and home. Social isolation, emotional distress, dependence on family, poor employment opportunities, and personal injury add to the suffering of people with epilepsy.38 Because of the seriousness of the disorder and its psychosocial dimensions, it is worrying that epilepsy is often suboptimally diagnosed and managed, even in developed countries, and especially among certain socioeconomic groups.11 39

How to close the treatment gap

The Global Campaign Against Epilepsy, which is jointly sponsored by the World Health Organization, ILAE, and International Bureau for Epilepsy, advocates using phenobarbital to close the high treatment gap in low income countries.40 The suggested first step is for all patients with epilepsy to be given phenobarbital, which will control seizures in most of them. In resource poor countries, phenobarbital can cost as little as $5 (£3; €3.7) to $10 a year. Phenobarbital has an extremely low potential for misuse.41 Its use in developed countries has been limited by a comparative trial that showed that phenobarbital and primidone (which is metabolized to phenobarbital) were less well tolerated than phenytoin or carbamazepine.42 This finding is less relevant for resource poor countries when the only choice is between phenobarbital or no treatment at all. The side effects of phenobarbital—mainly sedation, possible mild cognitive impairment, and depression—can be minimized by using the lowest possible effective dose.41 Thus, phenobarbital is the current drug of choice for large scale, community based programs, particularly in rural and remote areas of developing countries.41 Despite the availability of phenobarbital for more than 90 years and its modest cost, the treatment gap for epilepsy still exceeds 90% in many developing countries.41

The current evidence base of comparative efficacy among first line antiepileptic drugs is limited to a surprisingly few class I trials (box 2).

Box 2 Preferred first line antiepileptic drugs for new onset and refractory epilepsy in adults2643

All drugs that are regarded as first line for new onset cases are also considered for patients with refractory epilepsy because they differ from one another in their pharmacological profile.

New onset partial epilepsies
  • Carbamazepine

  • Gabapentin

  • Lamotrigine

  • Levetiracetam

  • Oxcarbazepine

  • Topiramate

  • Valproate

New onset idiopathic generalized epilepsies
  • Lamotrigine

  • Topiramate

  • Valproate

Refractory partial epilepsy
  • Lacosamide

  • Pregabalin

  • Zonisamide

  • Perampanel

  • Clobazam

Refractory idiopathic generalized epilepsies
  • Clobazam

  • Levetiracetam

Class I evidence for the comparative efficacy and effectiveness of drugs for new onset epilepsy is limited.19 The SANAD trial, a large randomized unblinded pragmatic study of antiepileptic drug monotherapy in new onset epilepsy, showed similar efficacy of carbamazepine, lamotrigine, and oxcarbazepine, but a lower comparative efficacy of gabapentin and topiramate, for treating focal seizures.44 Time to 12 month remission was the primary efficacy parameter. Compared with carbamazepine, the hazard ratios (95% confidence interval) were 0.72 (0.58 to 0.89; P<0.05) for gabapentin, 0.81 (0.66 to 1.00; P<0.05) for topiramate, 1.01 (0.83 to 1.22; P>0.05) for lamotrigine, and 0.92 (0.73 to 1.18; P>0.05) for oxcarbazepine. A hazard ratio greater than one indicates that 12 month remission occurs more rapidly on that drug than with carbamazepine.44

In another unblinded randomized study, levetiracetam monotherapy was as effective as controlled release carbamazepine for focal seizures or extended release valproic acid/valproate for generalized seizures in patients with new onset epilepsy.45 The hazard ratio for time to treatment withdrawal was 1.02 (0.74 to 1.41) for levetiracetam versus extended release valproic acid and 0.84 (0.66 to 1.07) for levetiracetam versus controlled release carbamazepine.45

For treatment of refractory partial epilepsy, taking into account baseline risk, random effects meta-analysis was used to derive pooled estimates of odds ratios and number needed to treat or number needed to harm (NNT/NNH).46

Sixty two placebo controlled trials (12 902 patients) and eight head to head randomized controlled trials (1370 patients) were included. Pooled odds ratios for responder and withdrawal rates (versus placebo) were 3.00 (95% confidence interval 2.63 to 3.41) and 1.48 (1.30 to 1.68), respectively. Indirect comparisons of responder rate based on relative measurements of treatment effect favored topiramate (1.52, 1.06 to 2.20) over all other antiepileptic drugs, whereas gabapentin (0.67, 0.46 to 0.97) and lacosamide (0.66, 0.48 to 0.92) were less efficacious, without significant heterogeneity. When analyses were based on absolute estimates (NNTs), topiramate and levetiracetam were more efficacious, with gabapentin and tiagabine being less efficacious. Withdrawal rates were higher with oxcarbazepine (1.60, 1.12 to 2.29) and topiramate (1.68, 1.07 to 2.63), and lower with gabapentin (0.65, 0.42 to 1.00) and levetiracetam (0.62, 0.43 to 0.89). However, differences were too small to make any conclusions about which new drugs had superior effectiveness. The choice of drug for refractory partial epilepsy should therefore be guided more by other aspects, such as patient characteristics and pharmacoeconomics, than only by evidence from randomized trials.46

Figure 5 lists dosages and effective plasma concentrations of antiepileptic drugs for the treatment of epilepsy in adults. The incidence of many adverse events can be reduced by slow titration and avoiding high dosages.


Fig 5 Dosages and effective plasma concentrations of antiepileptic drugs for the treatment of epilepsy in adults26 47

Tolerability and safety of drugs in new onset epilepsy

Given the similar efficacy of many first line antiepileptic drugs in new onset epilepsy, comparative tolerability and safety become important considerations when selecting treatment.

Figure 6 provides an overview of the main tolerability and safety considerations for currently available antiepileptic drugs. The evidence base for the comparative tolerability of individual drugs given as monotherapy is limited to short term randomized controlled trials, which typically show a similar proportion of patients with adverse effects when comparing newer drugs such as levetiracetam and zonisamide with carbamazepine.33 48 In the SANAD trial, about 50% of patients reported at least one adverse effect from carbamazepine or valproic acid/valproate as well as from newer drugs, such as lamotrigine, gabapentin, oxcarbazepine, and topiramate.5 6 There is thus no compelling evidence that these recently approved drugs are better tolerated than older ones.5 6 49 50 With regard to safety, valproic acid/valproate seems to be the most teratogenic antiepileptic drug on the market,51 52 and newer drugs, such as gabapentin and levetiracetam, cause fewer or no dermatological hypersensitivity reactions and do not induce or inhibit hepatic enzyme function.


Fig 6 Overview of adverse effects of individual antiepileptic drugs.9 26 29

Pharmacogenomics may be helpful in selecting specific antiepileptic drugs.53 People of Asian descent who take carbamazepine, lamotrigine, or phenytoin and carry the HLA-B*15:02 allele have a significantly increased risk of developing Stevens-Johnson syndrome or toxic epidermal necrolysis.54 In drug specific analysis, the carrier rate of HLA-B*15:02 was significantly higher in patients with carbamazepine related Stevens-Johnson syndrome or toxic epidermal necrolysis than in carbamazepine tolerant controls (92.3% v 11.9%, P<0.005; odds ratio 89.2, 19.2 to 413.8). This was also true in patients with phenytoin related Stevens-Johnson syndrome or toxic epidermal necrolysis compared with phenytoin tolerant controls (46.7% v 20.0%, P=0.045; 3.50, 1.10 to 11.18.55 Screening is therefore recommended before starting these drugs in patients with Han Chinese and South East Asian ancestry.56

With older antiepileptic drugs, drug interactions can greatly lower the efficacy of other drugs, including other antiepileptic drugs when taken in combination; this is not a problem with newer non-enzyme inducing agents, such as gabapentin, lamotrigine, and levetiracetam (fig 7).


Fig 7 Summary of drug interaction properties of common antiepileptic drugs.56 57 58 59 *Inducers of cytochrome P 450 enzyme system. †Need to monitor serum concentrations

The evidence on the potential adverse effects of long term enzyme induction with antiepileptic drugs has been recently reviewed.60 Clinical problems can occur as a result of pharmacokinetic interactions altering the serum concentration and, possibly, the efficacy or the adverse effects of concurrently taken antiepileptic drugs and other drugs when the inducer is introduced or withdrawn.60 Enzyme induction will continue for as long as the patient takes the inducer and will affect future drugs that are prescribed. The enzyme inducing effects of antiepileptic drugs therefore have implications for the general health of people with epilepsy.

Whether enzyme inducing antiepileptic drugs should still be used as first line treatment for newly diagnosed epilepsy when many non-inducing, equally effective, alternatives are available is a point for discussion and further research. Several clinical scenarios require patients to be switched to a non-enzyme inducing antiepileptic drug. Examples include patients who need pharmacotherapy for cancer or those with other life threatening diseases treated with drugs that are inducible by concurrent antiepileptic drug treatment.

People with epilepsy who are established on enzyme inducing antiepileptic drugs should be screened regularly for associated long term problems, such as osteoporosis and sexual dysfunction. The patient’s other care providers should be advised about the potential for harmful pharmacokinetic interactions. Switching patients to non-enzyme inducing drugs to avoid these interactions should be done with caution, particularly if seizures are not fully controlled. For seizure-free patients, the risks and benefits of switching need to be carefully weighed given the paucity of data on comparative likelihood of seizure control. In all such situations, the benefits and risks of both courses of action should be discussed with patients and their families.60

Finally, patients may be using over-the-counter dietary supplements or herbal preparations, some of which may interact with antiepileptic drugs—for example, Gingko biloba or St John’s wort can interact with hepatically metabolized antiepileptic drugs.61

Optimizing drug regimens

Iatrogenic overtreatment is a leading cause of poor tolerability to antiepileptic drugs. This can occur by unnecessarily exceeding the recommended dosage for a particular drug (fig 5) or through pharmacokinetic or pharmacodynamic effects of other, including inappropriately prescribed, drugs.62 Adverse effects and the patient’s perceived risk of adverse effects or safety risks may compromise adherence to the prescribed dose. Poor adherence, in turn, may lower treatment efficacy, with potentially fatal results,63 and paradoxically cause heightened or prolonged adverse effects by not allowing tolerance of adverse effects to develop.64

Target plasma concentrations are available for several antiepileptic drugs (fig 5) but are less useful for optimizing dosages and dosing schedules than for monitoring the patient’s clinical course and adherence to therapy.26 Except for phenytoin, for which monitoring is strongly recommended, particularly at concentrations above 20 mg/L because of the non-linear saturation dose kinetics, monitoring of plasma concentrations of other drugs is needed only to confirm suspected non-adherence or to evaluate unexplained toxicity or uncontrolled seizures in individual cases.26 65 Even so, although therapeutic drug monitoring may improve the benefit to risk ratio of treatment, there are many practical limitations,65 including latency in the occurrence of adverse effects or seizures and constraints in when the blood can be sampled, owing to travel time to phlebotomy services. In addition, further work is needed to clarify the role of drug monitoring in improving seizure control during pregnancy and identifying serum drug concentrations that may be considered safe for fetal exposure.66

Drug resistant epilepsy

Drug resistant epilepsy is one of the most important unmet needs in the daily management of epilepsy,11 and it provides a challenge to our understanding of the mechanisms underlying drug resistance and how it can be overcome or avoided (fig 8).


Fig 8 Possible determinants of antiepileptic drug resistance in human and experimental epilepsies.67 Modified, with permission, from Nature Reviews9

Any patient in whom at least two trials of adequately selected and dosed antiepileptic drugs have not brought sustained remission fulfils the ILAE criteria for drug resistant epilepsy.68 Many other definitions exist for different purposes.7 26 69 Epilepsy may also be considered drug resistant if treatment does not stop seizures for 12 months, for whatever reason. By this wide definition, which is based on an influential hospital based observational study,7 and which is increasingly being used in the US, 36% of newly treated patients have drug resistant seizures.7 However, if the definition of frequent and severe seizures despite optimal treatment is used, with alternative treatments such as surgery being included, only 5-10% of newly diagnosed patients are estimated to have drug resistant seizures.70 A diagnosis of absolute drug resistance may require failure of at least six antiepileptic drugs, because about 17% of patients become seizure free when additional antiepileptic drugs are given, even when two to five drugs have previously failed to control seizures.32 71 72 These data suggest that there is no room for complacency among physicians treating patients who have had persistent seizures over many years despite taking multiple antiepileptic drugs.

The mechanisms underlying drug resistant epilepsy are still not fully understood (fig 9).8 73 74 Current theories include the transporter hypothesis, the target hypothesis, the network hypothesis, the gene variant hypothesis, and the intrinsic severity hypothesis.8 75 However, none of these hypotheses can convincingly explain how drug resistance arises in human epilepsy,67 and a new synthesis or breakthrough in understanding is needed. Interestingly, a history of depression and a high frequency of seizures before treatment onset have been associated with drug resistance.76 77 These and other observations suggest that common neurobiological factors may underlie disease severity, psychiatric comorbidity, and drug resistant epilepsy, although more work is clearly needed.67


Fig 9 Examples of novel targets that are particularly interesting for development of antiseizure or antiepileptogenic drugs. Modified, with permission, from Nature Reviews.9 GABAA=γ-aminobutyric acid type A; NKCC1=bumetanide sensitive sodium-(potassium)-chloride cotransporter 1; MHC=major histocompatibility complex; NRF2=nuclear factor erythroid 2 related factor 2; NRSF=neurone restrictive silencer factor; TGFβ=transforming growth factor β; VLA4=very late antigen 4 (α4β1 integrin)

Recent progress in our understanding of mechanisms involved in ictogenesis and epileptogenesis now permits a shift towards target based validation studies in animal models of refractory epilepsy or epileptogenesis. Systems biology approaches are a promising source for targets. Such approaches take advantage of newer high throughput technologies to profile large numbers and types of molecules by using functional genomics, transcriptomics, epigenomics, proteomics, and metabolomics, enabling identification of causal pathways from the myriad of competing hypotheses, and thus assisting in defining candidate targets.78 Molecular profiling of epileptic brain tissues from animal models and humans also holds promise to identify new ictogenic and epileptogenic drug targets, and it might be possible to discover a final common pathway of genes consistently induced at human epileptic foci.78 This is supported by the recent identification of several promising pathways and potential drug targets, with particularly interesting examples illustrated in fig 10. More extensive discussion of individual targets is available elsewhere.9

No class I evidence has shown superior efficacy for any particular antiepileptic drug with market authorization for treating drug resistant epilepsy.11 In addition, there is no evidence that modern antiepileptic drugs have substantially lowered the proportion of patients with drug resistance.11 New add-on antiepileptic drugs are only moderately more effective than placebo. In a recent meta-analysis of 54 randomized controlled add-on trials in 11 106 patients with refractory epilepsy, the benefit in efficacy between adding a new antiepileptic drug and adding placebo was only 6% for freedom from seizures and 21% for a 50% reduction in seizure frequency.79 This suggests that better strategies for finding more effective antiseizure drugs are needed for refractory epilepsy.

Failure of the first drug to induce sustained seizure remission and drug resistant epilepsy

There are two options for patients who continue to have seizures despite taking the first antiepileptic drug: an alternative monotherapy (substitution) or combination therapy (add-on), which usually means adding a second drug to the current monotherapy.26 80 Randomized trials have not provided evidence of which strategy is best.81 82 Although substitution is preferable for patients with serious idiosyncratic side effects from the first drug, many physicians prefer add-on treatment with small increments in dose, mainly because it avoids the possibility of breakthrough seizures after discontinuation of the baseline drug.26 In addition, add-treatment has become easier to implement and maintain with modern non-enzyme inducing drugs.60

For patients whose clinical course meets the study specific definition of drug resistant epilepsy,68 relatively short term randomized controlled trials show that the chance of freedom from seizures declines with successive drug regimens, most markedly from the first to the third antiepileptic drug, especially in patients with localization related epilepsies.32 In one representative observational study from an epilepsy clinic, seizure-free rates decreased from 61.8% for the first antiepileptic drug to 41.7% after one drug proved ineffective.71 In patients who had no response to the first drug, the proportion who subsequently became seizure free was much smaller (11%) when treatment failed because of lack of efficacy rather than intolerable side effects (41%) or an idiosyncratic reaction (55%).7 Encouragingly, a longitudinal observational study encompassing almost 40 years of follow-up found that nearly four of five patients whose seizures were not initially controlled after two trials of suitable antiepileptic drugs eventually entered remission for at least one year, and half had at least a five year remission.83 Idiopathic or cryptogenic causes were the only significant predictor of entering remission in this study.

Treatment of special patient groups

One of the standards of good clinical care is to tailor the treatment of epilepsy on the basis of the patient’s individual needs.26


People with epilepsy, and particularly women, have a higher risk of bone fracture than the general population.84 This increased risk is secondary to epilepsy (that is, breaking a bone during a seizure) and use of antiepileptic drugs, especially enzyme inducing ones.60 85 These drugs independently increased the risk of fracture in the Women’s Health Initiative study,86 as well as in a Danish population based case-control study and a Korean study.85 87 The Women’s Health Initiative determined the associations between the use of antiepileptic drugs and falls, fractures, and bone mineral density over an average of 7.7 years of follow-up in women aged 50-79 years in a longitudinal cohort analysis. After adjustment for covariates, use of antiepileptic drugs was positively associated with total fractures (hazard ratio 1.44, 1.30 to 1.61), all site specific fractures including the hip (1.51, 1.05 to 2.17), clinical vertebral fractures (1.60, 1.20 to 2.12), lower arm or wrist fractures (1.40, 1.11 to 1.76), other clinical fractures (1.46, 1.29 to 1.65), and two or more falls (1.62, 1.50 to 1.74), and was not associated with baseline bone mineral density or changes in bone mineral density (P≥0.064 for all sites). Use of more than one antiepileptic drug and use of enzyme inducing antiepileptic drugs were significantly associated with total fractures (1.55, 1.15 to 2.09 and 1.36, 1.09 to 1.69, respectively). The Women’s Health Initiative concluded that in clinical practice, postmenopausal women who use antiepileptic drugs should be considered at increased risk for fracture and attention to fall prevention may be particularly important in these women. Antiepileptic drugs, especially enzyme inducing ones, have been shown to decrease bone mineral density and alter bone metabolism. Induction of cytochrome P can accelerate the metabolism of vitamin D to polar inactive metabolites.60 The use of risedronate plus calcium and vitamin D has been shown to prevent the occurrence of new fractures in male patients with a high risk of fractures.88 Further studies are needed to clarify the mechanisms by which enzyme inducing antiepileptic drugs have these effects on bone and whether newer non-enzyme inducing drugs have advantages over enzyme inducing ones.

Pregnant women and neonates

Although two of three women with epilepsy who become pregnant remain seizure free throughout pregnancy, antiepileptic drug dosages may need to be adjusted, particularly when seizures occur in the first trimester. Women prescribed lamotrigine and possibly levetiracetam, topiramate, and oxcarbazepine may also need dose adjustment to compensate for the increased clearance of these drugs during pregnancy and to reduce the risk of breakthrough seizures.89 90 91

Offspring of women with epilepsy who took an antiepileptic drug during pregnancy seem to have an increased risk of being small for gestational age and having a one minute Apgar score of less than 7.91 Many antiepileptic drugs are associated with major congenital malformations, and prescribers should routinely consult an updated package insert or patient information leaflet for the latest recommendations of regulatory agencies. For example, the use of valproic acid monotherapy in the first trimester is associated with significantly increased risks for six of the 14 malformations under consideration. The adjusted odds ratios were as follows: spina bifida, 12.7 (7.7 to 20.7); atrial septal defect, 2.5 (1.4 to 4.4); cleft palate, 5.2 (2.8 to 9.9); hypospadias, 4.8 (2.9 to 8.1); polydactyly, 2.2 (1.0 to 4.5); and craniosynostosis, 6.8 (1.8 to 18.8).92 93

A recent guideline for treatment of women with epilepsy also suggested that intrauterine exposure to valproate monotherapy reduces cognitive outcomes for offspring, as has also been suggested for phenytoin and phenobarbital.94 If clinically possible, antiepileptic drugs known to be associated with congenital malformations, including valproate, as well as combinations of antiepileptic drugs, should be avoided during pregnancy, especially during the first trimester. Similarly, valproate, phenytoin, phenobarbital, and antiepileptic drug polytherapy should be avoided throughout pregnancy if clinically possible to prevent unfavorable cognitive outcomes in offspring.94

The risk of major congenital malformations seems to be influenced not only by the specific antiepileptic drug but also by dose and other variables.95 96 The lowest malformation rate was seen in the International Registry of Antiepileptic Drugs and Pregnancy (EURAP) with less than 300 mg per day of lamotrigine and less than 400 mg per day of carbamazepine compared with valproic acid and phenobarbital at all studied doses, and with carbamazepine at doses greater than 400 mg per day.95 Folate supplementation (≥0.4 mg folic acid/day) is recommended during pregnancy because this lowers the risk of cognitive teratogenicity in babies born to women with epilepsy.96 Primidone and levetiracetam pass into breast milk in amounts that may be clinically important, unlike valproate, phenobarbital, phenytoin, and carbamazepine.96

Older people

The change in pharmacokinetics and higher sensitivity to adverse events of many antiepileptic drugs associated with aging usually require more cautious selection of drugs and dosing in older people. Lower glomerular filtration rates should prompt reduced doses of renally excreted drugs. Changes in body fat, albumin, and cytochrome P450 also occur, and oxcarbazepine related hyponatremia may be more common.26 97 In addition, concomitant diseases, such as hypertension, are common in this age group and often require medication, increasing the possibility of drug interactions with antiepileptic drugs. Therefore, monotherapy with a well tolerated antiepileptic drug that is not associated with drug interactions, such as gabapentin and lamotrigine,98 low dose topiramate,99 and levetiracetam (no class I evidence available), is preferable. Providers should be aware that adherence to antiepileptic drug regimens may be more difficult in older people with cognitive decline.

Patients with comorbidities

Many disorders are more common in people with epilepsy than in the general population, including cardiac, gastrointestinal, and respiratory disorders; stroke; dementia; and migraine.100 Alzheimer’s disease and migraine are not only more common in patients with epilepsy but are also risk factors for the development of seizures, suggesting a bidirectional association and shared disease mechanisms.100

The lifetime community based prevalence of depression, suicidal ideation, and generalized anxiety disorder is twice as high in patients with epilepsy than in the general population.101 Depression and anxiety substantially affect quality of life and are associated with an increased suicide rate.102 The psychiatric comorbidities of epilepsy may also manifest as psychogenic non-epileptic seizures or panic attacks.103 Psychiatric comorbidities are associated with a worse response to the treatment of the epilepsy, whether by drugs or surgery. Comorbid mood and anxiety disorders have also been associated with more adverse effects when taking antiepileptic drugs.77

Before starting antidepressants in patients with epilepsy, it is important to look for possible iatrogenic causes of depression. Antiepileptic drugs such as phenobarbital, vigabatrin, topiramate, tiagabine, levetiracetam, and clobazam can induce depressive symptoms in patients with epilepsy. Several second generation (carbamazepine and valproate) and third generation (lamotrigine and pregabalin) drugs are associated with mood stabilizing properties, so discontinuation of one of these could precipitate depression.77 103 Patients with both epileptic seizures and psychogenic non-epileptic seizures may benefit from reducing high doses of antiepileptic drugs or the number of drugs given, if possible.103

Treatments for psychiatric disorders in patients with epilepsy are severely lacking. Current clinical experience suggests that carbamazepine, valproate, and lamotrigine cannot counteract established depression in patients with epilepsy. Although pregabalin is approved for both epilepsy and generalized anxiety disorder, it is has not been comprehensively studied as treatment for patients with epilepsy and comorbid psychiatric disorders.

The ability of antidepressants to counteract depression in patients with epilepsy has not been properly studied.77 103 Only two double blind controlled trials have been reported. One small study showed that high dose amitriptyline was superior to placebo against major depressive episodes.104 Reassuringly, the other trial found that sertraline did not increase seizure frequency or severity.105 Selective serotonin reuptake inhibitors (SSRIs) and selective noradrenaline reuptake inhibitors (SNRIs) have been assessed in patients with intractable epilepsy in open label trials, and fewer seizures were seen during treatment with the SSRIs fluoxetine or citalopram.103 The only exception among antidepressant drugs was bupropion, which caused more seizures in patients with epilepsy.103 Taken together, these studies suggest that SSRIs and SNRIs may reduce seizures and depressive symptoms in patients with epilepsy and depression, although further controlled trials are needed. Work is also needed to evaluate anecdotal observations that SSRIs and SNRIs increase the number of seizures in patients with slow hepatic metabolism or when taken in overdose.103 Until then, SSRIs with minor pharmacokinetic interactions, such as escitalopram and citalopram, should be considered as first line drugs, followed by sertraline. Fluoxetine and paroxetine interfere with cytochrome P450, so their use may require antiepileptic drug dosages to be adjusted.10

Patients with status epilepticus and prolonged acute convulsive seizures

Tonic-clonic status epilepticus is associated with serious morbidity and mortality, and treatment depends on seizure stage (fig 10).106 Unfortunately, this is a therapeutic area in which there are few randomized trials, and their absence has impeded definitive assessment of alternative therapeutic options, particularly in treatment of stage 2 and stage 3 seizures. The regulatory agencies have not licensed drugs for status epilepticus because of the lack of randomized studies.106


Fig 10 Doses and routes of administration of drugs used to treat different stages of tonic-clonic status epilepticus. From the consensus document of the workshop of European epileptologists106

In the first stage (early status epilepticus), buccal midazolam has become an important out-of-hospital treatment option. A randomized controlled trial showed that buccal midazolam achieved seizure cessation in 8 min compared with 15 min for rectal diazepam (P<0.01). The rate of respiratory depression did not differ between groups.107 In UK community practice, rectal diazepam and unlicensed buccal midazolam are the two treatment options used for acute epileptic seizures. In practice, outside the US rectal diazepam is rarely used, with unlicensed buccal midazolam being widely recommended and prescribed by physicians. More recently a licensed preparation of buccal midazolam has become available.108 In a double blind study of children and adults with convulsions that had lasted for more than five minutes, and who were still seizing when paramedics arrived, midazolam given by intramuscular autoinjector had equal efficacy to intravenous lorazepam, with comparable safety. The primary efficacy outcome in this study was absence of seizures on arrival at the emergency department, without emergency medical system rescue therapy. 109 Patients treated with intramuscular midazolam were more likely to have stopped seizing on arrival at the emergency department and were less likely to be admitted to the hospital or an intensive care unit.109

In the second stage (established status epilepticus), preferred treatment choices include intravenous valproate, levetiracetam, and lacosamide among the newer antiepileptic drugs, as well as the older agents fosphenytoin, phenytoin, and phenobarbital (fig 8). In the third stage (refractory status epilepticus), midazolam, thiopentone, and propofol are available choices (fig 10). Further treatments such as various anesthetics and non-pharmacological treatments may be considered as well, including immunotherapy for cryptogenic refractory status epilepticus.110 None of the drugs for the second or third stage has been studied in sufficiently powered randomized controlled trials, and multicenter randomized controlled comparative trials are needed.111 Treatment in individual cases should include consideration of any underlying causes of status epilepticus.111

Stopping antiepileptic drugs

Patients who become seizure free and remain so for a prolonged time often wish to discontinue treatment. The decision to discontinue antiepileptic drugs should be based on the patient’s risks of seizure recurrence after discontinuation, which overall is twice as high in the two years after discontinuing drugs compared with continuing to take them (box 3). Other studies suggest that the risk of seizure recurrence when patients stop taking antiepileptic drugs is as high as 34% (27% to 43%), with a wide range of 12-66%.112 Adults seem to have a higher risk of recurrence than children (39% v 31%).113 The revised ILAE definition of epilepsy states that “epilepsy is considered to be resolved for individuals who either had an age dependent epilepsy syndrome but are now past the applicable age or those who have remained seizure free for the last 10 years and off anti-seizure medicines for at least the last five years.”21

Box 3 Stopping antiepileptic drugs in patients in remission

High risk profile for seizure recurrence off antiepileptic drugs106
  • Being 16 years or older

  • Taking more than one antiepileptic drug

  • Having seizures after starting drug treatment

  • History of generalized tonic-clonic seizures

  • History of myoclonic seizures

  • Having an abnormal electroencephalogram in previous year

When it may be safe to discontinue114115
  • Freedom from seizures for more than two years implies a 60% chance of persistent remission in certain epilepsy syndromes

  • Favorable factors:

    • Control easily achieved on a low dose of one drug

    • No previous unsuccessful attempts at withdrawal

    • Normal neurological examination and electroencephalogram

    • Primary generalized epilepsy except juvenile myoclonic epilepsy

    • Benign syndromes

Considerations for counseling patients include driving, pregnancy, work, and family. Other considerations are that a recurrent seizure may be embarrassing and stigmatizing for the patient. It could also result in loss of a driver’s license or, rarely, accidental or seizure related death. Furthermore, restarting antiepileptic drugs after seizure recurrence does not guarantee immediate and sustained resumption of seizure control.116 However, the impact of ongoing drug related side effects and drug interactions may argue in favor of discontinuing treatment. It may be advisable to offer discontinuation using a slow taper schedule in suitable patients after a thorough and documented discussion of the pros and cons.

This recommendation also applies to stopping antiepileptic drugs for patients in seizure remission after epilepsy surgery.117 118 Epilepsy surgery improves the prognosis of surgical candidates, with rates of freedom from seizures of 50-80%, depending on the cause of epilepsy, type and site of surgery, age group, and duration of follow-up.118 Unfortunately, in many series, outcomes are given without reference to whether patients are seizure free on or off antiepileptic drugs, the latter often being referred to as an indicator of cure.119 It has long been recommended that antiepileptic drugs are continued for at least one or two years after surgery, largely on the basis of antiepileptic drug withdrawal policies in non-surgical cohorts. Furthermore, it was suggested that after successful epilepsy surgery, duotherapy is preferable to monotherapy to maintain seizure control.120 This raises the question of whether it is justifiable to discontinue antiepileptic drugs after surgical remission in all patients.

After temporal lobe surgery, the average proportion of adults who were cured (at least five years seizure free and off drugs) was only 25%.119 A systematic review found that cure was more common in children (27%) than in adults (19%).121 This was confirmed in a recent Swedish study—twice as many children were seizure free and had stopped antiepileptic drugs than adults 10 years after surgery.122

Why are postoperative cure rates so much lower than the overall surgical seizure freedom rates reported in the literature? Firstly, the follow-up periods of observation may have been too short. It takes time to achieve complete discontinuation of antiepileptic drugs and at least an additional five years of remission off these drugs to establish cure.117 Most published studies had shorter postoperative follow-up intervals. Secondly, postoperative seizure freedom rates are not stable but decline over time,123 and the extent of this decline probably depends on underlying disease.124 Median long term (>5 years) seizure freedom rates ranged from only 27% to 66%,121 122 closer to the reported cure rates of 19-45% in a recent review.117

How can seizures relapse after antiepileptic drug withdrawal in patients who have had complete resection of the presumed cause of epilepsy? One possible explanation is that most epilepsies do not develop from alterations in a single localized target; rather, they arise from complex alterations that result in a wide epileptic network in the brains of individual patients.125 Variability of network properties and the extent of these networks may explain why even complete resection does not guarantee cure, because only part of the potentially epileptogenic network may have been removed.126 The development of antiepileptogenic drugs in the future may improve cure rates for medical treatment, and the discovery of biomarkers to assess the extent of the epileptogenic network in an individual patient may offer a chance to improve surgical cure rates.

There is no proof that antiepileptic drug withdrawal itself negatively affects long term seizure outcomes in patients who have become seizure free under drug treatment or after epilepsy surgery. Discontinuation of drugs merely unveils the natural course of the epileptic disorder in medically treated patients and unmasks true postoperative outcome. Given the available evidence, the risk of relapse is probably determined more by the clinical characteristics of the epilepsy syndrome or failure of the surgical procedure to eliminate relevant epileptogenic brain networks than by antiepileptic drug withdrawal and its timing.

Emerging treatments

Novel approaches to the development of new drugs are emerging.9 These offer hope of finding more effective antiseizure drugs to treat ongoing drug resistant epilepsy, antiepileptogenic agents to prevent symptomatic or genetic epilepsy before the first seizure, and disease modifying agents to mitigate established epilepsy. Our understanding of the mechanisms mediating the development of epilepsy, the causes of drug resistance, and the emerging role of pharmacogenetics for drug discovery have grown substantially over the past decade.9 127 Finally, new strategies are being explored, such as joint endeavors between academia and industry, identification and application of tools for new target driven and systems biology based approaches, and comparative preclinical proof-of-concept studies and innovative clinical trials designs.9

Barriers to the development of new drugs for drug resistant epilepsy

Reliance on established animal models that were used to bring previous antiepileptic drugs to the market as the preferred method to test experimental compounds as well as clinically inadequate trial designs in humans are roadblocks in the development of more effective antiepileptic drugs for drug resistant epilepsy.11 Novel preclinical and clinical approaches for the discovery and development of drugs with more effective antiseizure activity have recently been suggested.9 28 128 Potential targets for future drug discovery and development have been proposed (fig 10).

The currently accepted minimum measure for efficacy in randomized controlled trials is a statistical difference between the placebo arm and the treatment arm in the proportion of patients showing at least a 50% reduction in seizure frequency versus the baseline period.11 129 130 131 This bar is disappointingly low from a clinical perspective because 50% seizure reduction has not been shown to be beneficial for overall health or quality of life of patients,132 and nor does it satisfy the requirements for a driver’s license.133 This policy has led to the approval of several new antiepileptic drugs without demonstrated superiority over older ones, and which have entered the market at higher prices. A concern with placebo controlled trials is the increasingly unpredictable and unexpectedly high placebo response rates, which have been held responsible, at least in part, for the failure of new antiepileptic drugs to show efficacy in placebo controlled add-on trials.128 134 Another concern is that placebo use seems to be associated with an increased rate of sudden unexplained death in clinical trials.125

Clinical features such as a history of epilepsy surgery or lifetime exposure to seven or more antiepileptic drugs are associated with a low placebo response,135 136 which may maximize the treatment effect of the experimental antiepileptic drug versus placebo. However, limiting clinical trials to patients with these clinical features may restrict the generalizability of the findings. If variations of placebo mechanisms are left uncontrolled, it will be more difficult to document any specific effects of a drug. Novel clinical trial designs for the development of antiepileptic drugs that de-emphasize the use of placebo controls have recently been proposed.9 131 A further concern is that current trial designs do not take into account the heterogeneity of the causes and severity of disease in trial participants with drug resistant epilepsy. Although clinical features such as lifetime exposure to an increasing number of antiepileptic drugs seem to be associated with a decreased likelihood of eventual remission in patients with new onset epilepsy,7 32 71 83 current trial designs do not stratify patients on the basis of the severity of disease as measured by the total number of antiepileptic drugs they have taken, for example. This needs more attention and, if confirmed, may render a comparison of efficacy results between trials with individual antiepileptic drugs more difficult.


Most patients will achieve lasting remission of seizures on generally well tolerated antiseizure drug treatment, and the availability of many new antiepileptic drugs over the past three decades has brought more treatment options. Yet about 20-30% of patients continue to experience seizures despite all available drug options, and even more are at high risk of neuropsychiatric comorbidities.

New drugs with fewer side effects and better efficacy than the currently available ones are urgently needed. Antiepileptogenic and disease modifying agents are also needed. Because many large drug companies have stopped innovating in this therapeutic area, it is becoming increasingly important for foundations and government agencies to fund the discovery of new antiepileptic drugs, and to do so at a level commensurate with the substantial prevalence and costs of drug resistant epilepsy.

Areas for future research

To identify:

  • Molecular targets that may lead to the discovery of novel drugs to treat drug resistant epilepsy

  • The cellular mechanisms that trigger epileptogenesis after brain insults

  • The cellular mechanisms that lead to seizure remission without relapse on or off antiepileptic drugs (cure)

  • Molecular targets that may lead to the discovery of novel drugs to prevent epilepsy before the first seizure

  • Molecular targets that may lead to the discovery of novel drugs to prevent psychiatric and cognitive comorbidity

Additional educational resources—guidelines

  • American Academy of Neurology and American Epilepsy Society. Efficacy and tolerability of the new antiepileptic drugs I: treatment of new onset epilepsy. 2004.

  • American Academy of Neurology and American Epilepsy Society. Efficacy and tolerability of the new antiepileptic drugs II: treatment of refractory epilepsy. 2004.

  • American Academy of Neurology and American Epilepsy Society. Management issues for women with epilepsy—focus on pregnancy: teratogenesis and perinatal outcomes. 2009.

  • American Academy of Neurology and American Epilepsy Society. Management issues for women with epilepsy—focus on pregnancy: obstetrical complications and change in seizure frequency. 2009.

  • National Institute for Health and Care Excellence. The epilepsies: the diagnosis and management of the epilepsies in adults and children in primary and secondary care. 2012.

  • Guidelines for Management of Epilepsy in India (GEMIND).

  • Glauser T, Ben-Menachem E, Bourgeois B, Cnaan A, Guerreiro C, Kälviäinen R, et al; ILAE Subcommission on AED Guidelines. Updated ILAE evidence review of antiepileptic drug efficacy and effectiveness as initial monotherapy for epileptic seizures and syndromes. Epilepsia 2013;54:551-63


Cite this as: BMJ 2014;348:g2546


  • Contributors: DS wrote an early version of most sections of the manuscript and revised the manuscript. SCS edited early and revised versions of the manuscript, contributed as author to sections of the manuscript, and is guarantor.

  • Competing interests: We have read and understood the BMJ Group policy on declaration of interests and declare the following interests: DS has received hospitality and consulting fees in the past two years from Eisai, Sun, UCB, and Viropharma. None of the companies has had any input to the manuscript. SCS: none declared.

  • Provenance and peer review: Commissioned; externally peer reviewed.


View Abstract