Treatment of epilepsy – towards precision

Improving tolerability

Idiosyncratic drug reactions to AEDs are serious but uncommon. One of the most serious complications of carbamazepine (CBZ) use is Stevens–Johnson syndrome, a life-threatening condition that has a mortality rate of around 50% and that develops in around 21 cases per 100,000 patient exposed to CBZ⁸. This idiosyncratic reaction has been shown to be related in some populations to specific HLA markers⁹: HLA-B*1502 is an immunological marker found mainly in patients of Southeast Asian descent. Its presence is associated in this population with a fourfold increase in the risk of CBZ-associated rash and a 70-fold increase in the risk of Stevens–Johnson syndrome. Such a specific stratification of risk has influenced practice in some patient groups¹⁰ and such work has given us to the ability to target CBZ and phenytoin use in specific populations of Southeast Asian descent only to those patients in whom it is safe¹⁰, thereby avoiding increasing risk of harm in patients with newly diagnosed or refractory epilepsy. Such associations are not as firmly correlated for other drugs or in other racial subtypes. The presence of the HLA-A*301 haplotype in those of Northern European descent increases the risk of any form of CBZ-related skin reaction from 5% to 26%¹¹. Calls have been made to promote the cost-effectiveness of screening for this haplotype before CBZ is prescribed in these populations¹².

While such work has helped predict the rarer and serious drug side effects, large ongoing studies of AED use in newly diagnosed epilepsy incorporate genetic testing to try to find genetic associations with less serious but more common side effects of AED use, such as mood disorder, tremor, weight gain, or cognitive change. The mathematical and technical challenges therein are still significant, not least of which will be the clear delineation and definition of these adverse effects and the need to recruit exceptionally large cohorts.

Improving effectiveness

Currently, the choice of treatment in patients with newly diagnosed epilepsy depends on some fairly basic clinical characterisation. The SANAD (Standard and New Antiepileptic Drug) studies from 2007^7,13 showed the benefits of this, and the two treatment arms were determined by the then-contemporary clinical classification of partial or generalised epilepsy. The choice of treatments was different in each arm in this randomised open study and helped provide a context for drug preference in newly diagnosed epilepsy. In focal epilepsy, the optimal seizure responses were seen with lamotrigine and CBZ, and the former demonstrated somewhat improved tolerability.

The choice of AED for genetic generalised epilepsy has always been more limited since the traditional sodium channel blockers can be associated with an increase in seizure frequency and severity. In the generalised arm of the SANAD study¹³, sodium valproate, rather than topiramate and lamotrigine, was most effective at inducing seizure freedom. However, the deleterious effect of this drug in pregnancy has led to a continued search for safe and efficacious treatment which does not present pregnancy-related problems. The SANAD2 study looks to compare levetiracetam and sodium valproate in patients with newly diagnosed genetic generalised epilepsy and is due to report in 2019.

Given the crude clinical factors that currently influence initial treatment choice, it should be time for principles of precision medicine to come to the fore in our decision making in many stages of epilepsy care. Thankfully this day seems to be closer¹⁴. Genetic studies are becoming increasingly common in patients with epilepsy.

The discovery of a specifically treatable substrate is rare but is potentially life-saving and life-changing, most commonly noted when there is a sodium channel mutation (SCN1A) causing Dravet syndrome or a genetic Glut-1 transporter deficiency (SCL2A1). Such a discovery will firmly place stiripentol¹⁵ or a ketogenic diet¹⁶, respectively, early in the treatment pathway.

The discovery of specific genetic mutations has helped us to repurpose drugs with specific actions which may have been used in entirely unrelated conditions¹⁷. Delineation of a specific channelopathy has allowed us to predict efficacy of quinidine for its antiepileptic effect in patients with epilepsy associated with KCNT1 mutations¹⁸. It may be hoped that further analysis of genetic variations in those with Dravet syndrome will highlight those with serotonergic changes most likely to benefit from new drugs such as the amphetamine derivative fenfluramine, justifying the associated cardiovascular risk¹⁹.

The suggested promise of screening for polymorphisms in p-glycoprotein as a marker or predictor of drug resistance has not been fulfilled in the longer term²⁰, but we should retain hopes of more substantial breakthrough in coming years.

Even where there is no specific targeted treatment currently available, the ability to counsel patients fully or even just provide more insight to affected families will have a markedly positive effect in improving the journey for patients and families²¹.

In adult clinics, genetic analysis in patients with epilepsy is becoming increasingly important, most especially in those patients with associated learning difficulties or progressive encephalopathy or in those with refractory generalised epilepsy. Lindy et al.²² reported a positive yield from next-generation sequencing or copy number variation analysis in 15% of patients in a panel of tests looking at up to 70 genes.

Even in early epilepsy, before treatment has been followed up in the long term, genetic testing may shed light on the underlying pathophysiology; around 9% of children with complex febrile convulsions have genetic changes in SCN1A testing²³.

More widespread testing in adult patients can enhance diagnostic yield further. Whole exome sequencing provides a specific diagnosis in 12.5% in patients with non-lesional focal epilepsy with a positive family history²⁴.

Although any patient with refractory epilepsy may benefit from genetic screening, such testing will be of most importance in patients with early-onset seizures (less than 3 years of age), a family history of seizures, associated neurological deficit, or learning disability²⁵. In short, clinical features suggesting widespread neuronal dysfunction are more likely to have an underlying genetic abnormality uncovered by today’s testing.