The management of epilepsy has become a multidisciplinary effort that is patient centred and holistic in its ethos. It is important not to forget the importance of allied health professionals and social support in the treatment of epilepsy. Epilepsy is not just a physical disease, the social implications are extremely important in a patient’s quality of life and the role of the wider health care team including psychological support and epilepsy nurse specialists are well established in patient care. Simple social measures such as providing free bus passes for patients unable to drive can have a profound impact on quality of life. The National Service Framework for Long-Term (neurological) medical conditions along with the National Institute for Health and Clinical Excellence (NICE) set out an overview for the standard of care in the United Kingdom for management of epilepsy ^{2, 6-8} .

This review will mainly focus on the pharmacological and surgical management of epilepsy but the reader is directed to the following articles for a deeper discussion of these important areas of quality of life and non-medical treatment particularly in the paediatric and elderly populations ^7-11 .

The natural history of epilepsy varies between individuals and epilepsy syndromes. It is difficult to generalize the course of the disease and accurately predict prognosis on a generic level ⁴ . In terms of pharmacology, one of the most important questions in symptomatic treatment of epilepsy is “when should I start treatment”? There are different views as to the correct answer but evidence suggests that after a single unprovoked seizure (and therefore before a definitive diagnosis of epilepsy) the recurrence rate without pharmacological intervention is around 25% ¹² . A large randomized controlled trial showed that early treatment after a single seizure affects short but not longer-term chances of being seizure free ¹³ . However, the risk of those with a second or third unprovoked seizure suffering from a further ictal episode is around 75%, therefore prophylactic treatment is generally justified in this group ^{2, 14} . Pharmacological management is a complex area that has variable efficacy and must be tailored to the individual patient.

Overall, antiepileptic drug management is effective in controlling seizures in around 60-70% of individuals although this is often achieved through a prolonged course of trial and error “popgun” pharmacy ¹⁵ . The choice of first line agent is a complex but important one. The initial selection requires the clinician to consider the individual patient and their wishes. The need for therapeutic monitoring, side effect profile, teratogenicity and even cost are all important considerations but the final aim is usually to achieve a quick, effective seizure control with a minimum of adverse effects ¹⁶ . A large randomized controlled trial (the SANAD study) by Marson et al investigated the drug of choice for newly diagnosed epilepsy of either the generalized or partial seizure type. This study has received much attention and despite its high power and good follow-up it been the subject of extensive critiques, mainly for the lack of syndromic classification and its lack of blinding. The findings, if accepted, suggest sodium valproate as first line for generalized seizures and lamotrigine for partial seizures ^{17, 18} . Notably there have been three new antiepileptics released since this study began (levetiracetam, zonisamide and pregabalin) and these are obviously not considered ¹⁹ . Final choices of pharmacologic agent are generally made individually and there is no universally accepted algorithm for therapeutic management, one example of first line drug selection is provided by French et al ² .

When considering pharmacological actions a popular first line agent for use in idiopathic generalized adult epilepsy is sodium valproate (Fig.1). The benefits of this drug were serendipitously discovered in 1963 when valproic acid was used as a solvent to dissolve various compounds that were being investigated for antiepileptogenic properties in electrical seizure models of mice ²⁰ . The exact mode of action has never been fully elucidated but what is known is that it seems to inhibit the action of succinic semialdehyde dehydrogenase ergo reducing succinic acid concentration and thus removing inhibition of L-glutamic acid decarboxylase (GAD). The GAD is then free to convert L-glutamic acid to the inhibitory neurotransmitter gamma amino butyric acid (GABA) ²¹ . More recent work has shown that it could also act via an increase in neuropeptide Y in the thalamus and temporal lobe and ergo reducing epileptiform oscillations ²² . The concept of antiepileptic agents simply increasing inhibitory factors and ergo reducing paroxysmal epileptiform discharges is almost certainly a gross simplification. Although sodium valproate is useful in a wide variety of epilepsy types including tonic-clonic, myoclonic and absence seizures, the main limitations are its teratogenicity and the wide side effect profile. Dose related effects include tremor, thrombocytopenia and weight gain while idiosyncratic reactions include hair loss and impaired liver function ²³ .

Figure 1

Figure 1. The structure of sodium valproate

Lamotrigine is a common initial monotherapy for partial seizures and this is supported by NICE guidance ⁶ (Fig.2). Its discovery in the early 1980’s came from a search for folate antagonists as this property was proposed to underlie the efficacy of the available agents. Although lamotrigine was found to be a very weak inhibitor of dihydrofolate reducatase it was efficacious in animal seizure models ²⁴ . The mode of action seems to be related to stabilization of neuronal membranes and blocking of excitatory glutamate release by sodium channel blockade although this remains controversial. Side effects are noticeably less than many antiepileptic agents, common problems include headaches and nausea. Rarer but more serious effects include cutaneous effects along the spectrum from a simple rash through to erythema multiforme and Steven-Johnson syndrome ²³ .

Figure 2

Figure 2. The structure of lamotrigine

Absence seizures seem to differ distinctly in their neurophysiological aetiology from other seizure disorders. There is good electophysiological data to show that the pathologic process is driven by abnormal neuronal oscillation in thalmocortical circuits. This circuit is thought to be physiologically important in normal alertness and wakefulness. Abnormalities in absence seizures may be mediated by glutamic overexcitation that becomes phase-locked with inhibition derived from GABAergic neuronal connections ^25-27 . This is thought to be the primary reason why many antiepileptic drugs that increase GABA cause an increase in spike-wave propogation and ergo may precipitate seizures of the absence type. The usual first line medication in these cases is ethosuximide, a member of the succinimide class of theraputic agents(Fig.4). The mode of action of this antiepileptic is, yet again, rather controversial. It is efficacious in pentylenetetrazol (PTZ) induced animal convulsion models and classically it has been viewed as a blocker of T-type calcium channels in the thalmocortical circuit. However, some of the experiments that demonstrated this have been criticised for their single neuron methodology and lack of external validity ²⁸ . Safety concerns are limited and mainly centre on idiosyncratic hypersensitivity reactions and the potential to precipitate tonic-clonic seizures in some patients ²³ .

Figure 3

Figure 3. Schematic showing the normal thalmocortical relay circuit. Thalamic relay neurones are labelled (TR), thalamic reticular nucleus neurones (NRT) and cortical pyramidal neurones (cortex). Glutaminergic neocortical layer VI pyramidal neurones project onto the glutaminergic thalamic relay neurones, setting up a positive feedback loop. Inhibition of this loop is provided by GABAergic reticular nucleus neurones. Cortical pyramidal cells also project onto the reticular nucleus neurons that then connect to thalamic relay neurones. Inherited ion channelopathies in this circuit are thought to underlie circuit hyperexcitability and epileptiform rhythmic spike-wave discharges .

Figure 4

Figure 4. The structure of ethosuximide

There are numerous other epilepsy syndromes and over 20 antiepileptic agents licensed in the United Kingdom. Many of the inherited childhood epilepsy syndromes have very particular phenotypes and seem to encompass a wide range of underlying pathology. Likewise, treatment for these syndromes differs significantly. It is beyond the scope of this article to cover the details of all antiepileptic drugs, ergo the reader is referred to a review article by Duncan et al for a summary of the other licensed antiepileptic agents. This article provides details of the agent’s putative modes of action, elimination, year of introduction, dosing and main safety issues ²⁹ .

The classic aim of epilepsy surgery was a complete removal of the epileptogenic focus without neurological deficit ³⁰ . This goal fails to consider more complex psychosocial and quality of life issues inherent in most patient encounters. There have been various outcome measures applied to surgical management of epilepsy syndromes. Classically seizure frequency and overall mortality were the markers used, but over time the importance of morbidity and patient quality of life have become more recognised. The efficacy of surgery, like antiepileptic drugs, must take these complex factors into account ³¹ .

There are two main divisions of epilepsy surgery; that taken with curative intent and that performed as a palliative procedure.

The selection criteria vary between centres but patients are generally considered for surgical intervention when they fulfil the core criteria of:

• Focal seizures with a lesion identifiable on imaging (in most cases) that is amenable to surgical intervention

• Supportive electrophysiological data, often in the form of an scalp or invasive electroencephalogram (EEG)

• Refractory to medical therapy or intolerable side effects

• No contraindications to surgery

The commonest and best-studied forms of epilepsy surgery are anteromesial or localized neocortical resection for mesial temporal sclerosis (MTS). This accounts for up to 75% of adult epilepsy surgery ^{32, 33} . The focal removal of a unilateral sclerotic hippocampal lesion with an anteromesial resection in well-selected patients has a seizure free success rate of around 60-70%. The results for neocortical resections are slightly less encouraging with a seizure free rate of around 50% but with an overall improvement rate of 85%. Surgical morbidity and mortality is generally small in this group ³⁴ . The results at one year seem to predict the long-term outcome accurately ³⁵ . The commonest permanent deficit is a disruption of Meyer’s loop causing a superior homonymous quadrantanopic visual field defect that excludes the patient from being eligible to hold a United Kingdom driving license. Improved localization of lesion and eloquent cortex has occurred with the introduction of intracranial electroencephalography with depth electrodes and judicious use of functional magnetic resonance imaging to localize language and other important cortical regions ^36-38 . The general trend with surgical refinement and experience is towards improvement and it is hoped that advances such as in vivo white matter bundle identification with diffusion-weighted tractography the risk of inadvertent field defects will be further reduced. The use of surgery in this class of patients is widely accepted and on a population level this intervention appears beneficial, certainly there is no appropriate medical therapeutic alternative.

Other important curative surgical procedures include removal of space occupying lesions that are epileptogenic or for focal cortical dysplasias. It is difficult to interpret outcomes in these groups due to the diverse range of pathologies and expected prognoses depending on the underlying cause. A recent retrospective study analysed 120 cases of focal cortical dysplasia (including MTS) and found that histological subtype and age of onset was not prognostic ³⁹ . It seems that patients with dysembryoplastic neuroepithelial tumours (DNETs), cystic lesions and vascular malformations have at least as high refraction rates as in MTS ⁴⁰ .

In patients with epilepsy that is beyond medical or surgical cure the option of palliation is profoundly important. Radical resections may benefit patients with refractory epilepsy from vascular defects, acquired dysfunction, malformations or developmental abnormalities beyond the remit of lesionectomy ^{41, 42} . These operations can be quite radical and may extend to complete hemispherectomy in children. In the complex multiseizure Lennox-Gastaut syndrome corpus callostomy is sometimes used with moderate success although no clinical trials have assessed this efficacy of this technique ⁴³ . One interesting method of ameliorating the frequency of seizures is to create a series of cortical incisions below the pia mater with the aim of inhibiting paroxysmal cortical depolarizations. The efficacy of this technique is still being questioned but advocates report good rates of success ⁴⁴ . The apparent benefits of palliative surgery are often only counted in terms of reduced seizure frequency. Clinical decisions and eligibility criteria in these areas are often difficult with minimal evidence to guide best practice ⁴⁰ .

As already alluded to, many of the anti-epileptic medications on license have come about, at least in part, due to serendipitous circumstances. The vogue in the scientific community is to rely less on fortune and to try and create medicines based on scientific reasoning. In this respect there have been several partial successes. Vigabatrin was the first ‘designer’ anticonvulsant agent and despite the severe risks of visual field defects and potential to lead to psychiatric disturbances it still has a role in West’s syndrome, a rare and particularly severe form of infantile epilepsy ⁴⁵ . Other examples of newer drugs with varied efficacy include gabapentin, pregabalin, tigabine, topiramate, levetiracetam and zonisamide. Despite the apparent plethora of new drugs it is notable that the new agents seem to be more ‘fine-tuning’ than ‘radical new cure’. This may be related to the relative inaccuracy of current animal seizure models and their inability to identify new therapeutic mechanisms, instead producing ‘me too’ drugs ⁴⁶ . The extent to which this is true is debatable, and with better understanding of epileptogenesis there may be scope to develop newer, more accurate models on which to test novel compounds. The alternative of primary in vivo experimentation is generally considered ethically suspect and technically impractical. With these existing barriers and the current scientific framework we are left to join patients in hoping for epiphanic pharmaceutical success, be it by molecular targeting or animal models ⁴⁶ .

The concept of individually designed and personalised medicines is central to the field of pharmacogenomics. The individual balance of pharmacodynamics and pharmacokinetics markedly affects the efficacy and side-effect profile of drug on each individual. Current dosing strategies and treatment schemes are based on Gaussian concepts of population-based responses. There exists a developing science by which drug treatment could conceivably be more accurately tailored to an individual. Key areas are drug absorption that may be more accurately determined by individual typing of transporters and counter-regulatory enzymatic eliminatory function. Examples include the current research into the expression of P-glycoprotein (pgp) that may determine the initial bioavailability of antiepileptic agents ^{47, 48} . Drug distribution is a second important part of pharmacogenomics, and knowledge of the extent to which a drug is apparently distributed in individuals should improve dosing accuracy. There have been multiple studies into the ABCB1 gene polymorphism and some have shown clinically significant effects on cerebrospinal fluid levels of antiepileptics ⁴⁹ . The extent to which the may be important in refractory epilepsy is still not known but further study is certainly warranted. There are many other areas in which pharmacogenetics may improve epilepsy treatment. These range from dosing improvement with better understanding of excretory functions to side effect predicition and event prediction of teratogenicity risk ⁴⁷ .

Over the past two decades stereotactic neurosurgical targeting has transformed operative practice and significantly improved patient outcomes. These benefits have come largely from the surgical enhanced accuracy provided by the technique. Analogous new developments include the application of magnetoencephalography and diffusion-weighted tractography to epilepsy surgery. These modalities promise to improve localisation of lesions and improve the resolution of resection margins with the goal of complete lesion removal and greater retention of eloquent cortex ⁵⁰ . The clinical applications of these technique are still being explored but progress must be made carefully as there are already cautionary tales in the literature where over-reliance on the imaging without other support have lead to detrimental outcomes ^{51, 52} . The role of white matter tractography is now being investigated specifically in epilepsy surgery ⁵³ . Likewise, transcranial magnetic stimulation and deep brain stimulation in epilepsy are being increasingly studied. They remain in their infancy but the potential for these techniques are significant and results so far are generally positive especially in refractory cases ^{54, 55} . Other invasive treatments with potentially revolutionary benefits include targeted direct application of stem cells, pharmaceuticals and even viral particulates to epileptogenic foci. Only further time and study will reveal how successful these strategies will be ²⁹ .