Chapter 563. Epilepsy Treatment: Ketogenic Diet and Neurostimulation

Most children with epilepsy in childhood achieve seizure freedom with anticonvulsant medication, and many outgrow their epilepsy and can be seizure-free and medication-free in adulthood. However, up to one third of patients can have intractable epilepsy, usually defined as continuing to have debilitating seizures despite the use of 2 or 3, appropriately chosen anticonvulsant medications at high therapeutic doses. For some children who have focal epilepsy, surgical resection of the responsible focus offers an alternative approach, and possibly cure of epilepsy. For others who do not have a surgical option, dietary manipulation and new techniques of neurostimulation offer approaches that may improve or even achieve seizure control. For those children whose seizure control improves, a reduction in medication dose or number can improve quality of life by reducing adverse effects.

The Ketogenic Diet (KD)

The ketogenic diet simulates the biochemical changes of fasting by providing the majority of calories in the diet from fat. The dietary fat is metabolized via acetyl coenzyme A to ketone bodies (acetoacetate 3-hydroxybutyrate and acetone) and fatty acids.1 In the context of reduced glucose availability, the body utilizes ketone bodies and free fatty acids for energy production. On a ketogenic diet, approximately 90% of calories consumed are from fat in contrast to a standard Western diet in which approximately 30% of calories are from fat. A ketogenic diet is high in fat, low in carbohydrate (10 to 15 gm/day), and provides adequate protein for health and growth (~ 1 gm/kg/day in children); typically it involves a 3:1 to 4:1 ratio of fat (grams) to carbohydrate and protein (grams).2In order to maintain ketosis, the components of each meal, snack, or beverage must preserve this ratio. The carbohydrate content of medications must be considered and minimized, usually by eliminating sugary liquid or chewable tablets in favor of standard tablet formulations. The diet can be provided using formula preparations for young infants and for patients using feeding tubes. The successful use of the ketogenic diet requires medical supervision, and the input ofa specifically trained, knowledgeable nutritionist for diet design, adjustment, and surveillance of growth and nutritional status. Parent education for the practicalities of weighing and measuring food components, managing food preferences and feeding behaviors, and adjusting for intercurrent illnesses, holidays, and variable energy demands, is essential . An intangible additional component of successful ketogenic diet programs is the ability to enthusiastically support families over the long term, who are engaged in a difficult, life-altering, but potentially rewarding process.

The ketogenic diet fell out of favor in most centers with the discovery of phenytoin and remained unpopular as newer drugs came to market. However, it has become clear that though safer than the older anticonvulsant drugs, the new drugs are not necessarily more effective. This has led to a resurgence of interest in diet treatment. In patients with epilepsy who fail to be controlled with a first anticonvulsant, the use of an alternative second anticonvulsant drug controlled epilepsy in 9%, and a third or multiple drugs controlled only 4%.3 In contrast, observational studies of the KD indicated efficacy for many children with highly refractory epilepsies — the populations reported typically had 100s of seizures per week and were treated with multiple AEDS at diet initiation.4-7 A multicenter study done more recently has shown that similar efficacy can be achieved.5 Studies of patients on the KD report up to 27% of patients having Å  90% reduction in seizures (7% seizure free) at 12 months, and 43% have at least a 50% reduction in seizures.6 A recent randomized, controlled, though not blinded study comparing immediate diet initiation to a control group who were not started on the diet for 3 months showed 38% with a > 50% seizure reduction at 3 months on the diet compared to 6% of controls.8 Seven percent had > 90% seizure control in the study group versus none of the controls. These results support the efficacy of the diet but suggest that the likelihood of success may not be as high as reported in earlier studies.

The KD is effective treatment for most seizure types, though it is least likely to be effective in complex partial seizures.9,10 In some children (17% in one study),10 a virtually immediate and complete response is seen within 2 weeks, and is usually maintained. Others may respond more slowly, but some reduction in seizures is expected within 2 weeks.11 In those children with less than a 50% reduction in seizures, the dietary restriction for the child and burden of producing ketogenic meals is such that the diet is withdrawn. Generally, there is a complete cessation or reduction in seizures within 2 weeks in children that will benefit from the KD.11 If there is less than a 50% reduction In seizures, the dietary restriction for the child and burden of producing ketogenic meals is such that the diet is withdrawn. In addition to Lennox-Gastaut syndrome which was highly represented in the initial studies, small series have described efficacy in children with highly refractory epilepsies such as myoclonic astatic epilepsy (Doose syndrome),12 severe myoclonic epilepsy of infancy (Dravet syndrome),13 infantile spasms,14,15 and tuberous sclerosis.16 The diet can be effective across age groups from infancy to adulthood,17 though it is rarely used in adults, largely due to poor palatability. However, the increase in use and acceptability of other high-fat diets, such as the Atkins diet for weight reduction, suggests that the KD may have a greater potential role in adults with epilepsy now than at any time in the past.

The KD is not a “natural” diet, and although usually tolerated very well has a significant list of potential adverse effects.Prior to starting the KD, patients are screened for metabolic disorders, especially disorders of fatty acid oxidation, and some mitochondrial disorders for whom diet initiation could be associated with a serious risk of deterioration. Acute complications of diet initiation include hypoglycemia (less likely with prediet metabolic screening), acute acidosis (as ketones, which are more acidic than glucose build up in the blood), and associated vomiting, and suppression of thirst leading to dehydration which exacerbates acidosis and vomiting. Management of these acute complications is the primary reason for initiation of the diet in a hospital setting. The patient is usually fit for discharge on the third or fourth day. The most common long-term complication is constipation; the most common serious complication is nephrolithiasis (6.7%).2,18 Growth failure is frequent, especially if the diet is used for many years.19-21 Other complications include increased bruising (seldom frank coagulopathy), increased infection rate, pancreatitis, and reduced GI motility.

The mechanisms by which the ketogenic diet reduces seizure frequency is unknown. In clinical situations, loss of ketosis is often associated with seizure breakthrough in children on the diet. However, there is conflicting animal data regarding the possibility of a direct anticonvulsant effect of ketone bodies, and they have not been shown to alter excitatory neurotransmission to date.22 Other hypotheses for which there is some experimental support, include a possible anticonvulsant effect of elevated levels of polyunsaturated fatty acids (PUFAs), or an indirect effect of PUFAs on mitochondrial energy metabolism via activation of “uncoupling” proteins, inhibition of glycolysis or an effect on potassium-sensitive ATP channels due to reduced glucose levels. There are also effects on neurotransmitters, particularly γ-aminobutyric acid (GABA) and norepinephrine which could act to decrease cortical excitability.23 Given the variety of seizure types and epilepsy syndromes that respond to the ketogenic diet, and the variability in response characteristics, it seems likely that different mechanisms, or combinations of mechanisms may operate in different clinical scenarios.

Modified Atkins Diet

This diet was developed as a temporizing measure for patients awaiting admission for KD initiation at Johns Hopkins Hospital. Patients were instructed to reduce carbohydrate intake to 10 to 15 grams per day, and to increase dietary fat in replacement. They were permitted free protein intake. The diet approximates a 2:1 ketogenic diet (KD) ratio. Many patients become ketotic and ~ 50% had improvement in their seizures.24 A further prospective open-label study of 30 adults reported 33% with a > 50% reduction in seizures at 6 months.25 The diet has been offered on an outpatient basis. This diet may be an attractive option for adults who are familiar with the Atkins diet. It differs from the Atkins diet in maintaining the more rigorous “induction” phase of the standard Atkins diet indefinitely, rather than relaxing the restrictions as is otherwise allowed. Medical supervision is required to monitor for and prevent acidosis, nephrolithiasis, and growth failure, all of which may occur in ketotic patients.

Low Glycemic Index Diet Treatment

This diet was developed at the Massachusetts General Hospital in Boston, in response to the experimental hypothesis that overall reduction in glucose levels or the regulation of glucose levels within narrow parameters may be associated with the efficacy of the KD.Total carbohydrate intake is gradually reduced to 40 to 60 grams/day and “allowed” carbohydrates are chosen for their low glycemic index < 50 (glycemic index describes the tendency of foods to increase blood glucose as compared to giving glucose itself which has a glycemic index of 100). The proportion of calories from fat is increased, though not as much as in the KD or modified Atkins diet, and approximates a 1:1 ketogenic ratio. Patients are not generally ketotic. This diet is initiated in the outpatient setting. A pilot study reported efficacy similar to the KD in initial follow-up.26 Longer-term studies and comparison with a traditional KD will be informative but are not yet available.

In the context of the treatment of epilepsy, neurostimulation describes the chronic intermittent electrical or magnetic stimulation of sites in the central or peripheral nervous system with the goal of raising the seizure threshold and reducing the frequency of seizures. These relatively new techniques are used adjunctively in patients with intractable epilepsy. Electrical vagus nerve stimulation (VNS) devices are in clinical use, and stimulation of the anterior thalamus, subthalamic nucleus, mesial temporal lobe, responsive stimulation of seizure foci, and trancranial magnetic stimulation are being investigated. Devices providing electrical stimulation must be implanted with the attendant risks of anesthesia and infection. Transcranial magnetic stimulation is delivered externally at approximately 2 monthly intervals. Overall efficacy of neurostimulation strategies has been modest to date, with very few patients achieving freedom from seizures by this means. However, unlike anticonvulsant drugs, neurostimulation is rarely associated with cognitive or behavioral adverse effects. Once implanted, compliance with therapy is essentially guaranteed. In addition, the use of neurostimulation may be associated with improvements in various aspects of quality of life. These features, along with the possibility of reducing the dose or number of anticonvulsant medications, make neurostimulation an attractive option for some patients, adult and pediatric, with intractable epilepsy. Treatment with neurostimulation is expensive. However, decreased health care utilization and decreased medication-related costs following VNS implantation have been described, which may result in cost savings over the life of the treatment.27,28

Vagus Nerve Stimulation (VNS)

The VNS device was approved for use in adults and children older than 12 years of age with refractory focal-onset seizures by the United States Food and Drug Association (FDA) in 1997. It is the only approved device at this time. More than 35,000 individuals with epilepsy have been implanted with the device, of whom approximately 30% were younger than 18 years at implantation.29 Stimulation of the vagus nerve results in synchronization or desynchronization of cortical electrical activity depending on the stimulus parameters used.30 A desynchronized state is associated with reduced seizure frequency and duration in multiple animal models and in humans.30,31 The left vagus nerve is chosen because it carries fewer cardioactive fibers than the right, and those innervate the ventricles rather than the atria with less likelihood of altering the heart rate. The stimulator battery is usually implanted subcutaneously or in a subpectoral pocket in the infraclavicular area, and the lead to the vagus nerve is threaded subcutaneously into the neck where it is placed around the nerve. This is usually an outpatient procedure. Treatment may be initiated perioperatively or shortly thereafter, and the “dose” may be adjusted by altering the battery output, stimulus duration, or frequency at 2- to 4-week intervals until the optimal settings are determined. Typically the stimulator provides a repeated stimulus of 0.5 to 3 milliamps, lasting up to 30 seconds every 1 to 5 minutes. On-demand stimulation, usually of slightly higher power, can be delivered by the patient/caretaker by passing a handheld magnet over the battery location. This may abort or decrease the duration of a seizure and reduce the duration of the postictal state in some patients.

The efficacy of vagal nerve stimulation was demonstrated in 2 pivotal studies in adults and children greater than 12 years of age.32,33 In these studies, patients were randomized to “high”-level (greater power output) versus “low”-level stimulation. In the first study of 67 implanted patients, the mean reduction in seizure frequency was 26.5% in those treated with high-level stimulation versus 6.1% in the group treated with low power level. The second study reporting results of 196 implanted patients, randomized to the same treatment groups, showed a mean seizure reduction of 28% versus 15% in high- and low-dose groups, respectively. A study of sixty children, aged 3 to 18 years at implantation who were enrolled in the pivotal studies or an early compassionate use protocol, showed a median seizure reduction of 23% in 60 children at 3 months, 31% at 6 months, and 42% in 46 patients at 18 months. Subsequent studies of efficacy after varying duration of treatment suggest continued and possibly incremental efficacy over time, although the effect of changes in anticonvulsant therapy cannot be assessed. One registry-based, retrospective study of 269 patients who did not have medication changes over a 12-month period had improved seizure reduction over time (45% at 3 months, 58% at 12 months).35 Alteration of stimulus parameters was not associated with improvement. To date, no consistent clinical predictors of a positive outcome have been identified.

The largest pediatric experience was reported by Helmers et al. and described retrospective, uncontrolled, multicenter study of 125 pediatric patients followed for up to 1 year.36 Most patients (59) had intractable focal epilepsy, and the study included 43 patients with Lennox-Gastaut syndrome. There was no restriction on change of anticonvulsant drug or dose during the study. The that the mean reduction in seizures was 36% at 3 months and 45% at 6 months. Medication changes were allowed during the study. The proportion experiencing a 50% or greater reduction in seizure number was 53% at 3 months and 57% at 6 months. Age at implantation (less than 6 years or 6 to 12 years of age) did not affect efficacy. Assessment of change in quality of life parameters including alertness, seizure clustering, verbal communication, duration of the postictal period, school achievement, mood, memory, and ability to ambulate indicated improvement in many patients, which was independent of seizure response. This finding reflects similar observations in adult studies. Longer-term follow-up (median follow-up of 2 years) of 46 patients aged 2 to 18 years at implantation reported an incremental response over time,37 similar to reports in adults. Median seizure reduction compared to baseline was 50%, 63%, 83%, and 74% at 6, 12, 24, and 36 months, respectively.

There is very limited data available regarding the effect of VNS therapy in specific epilepsy syndromes aside from refractory focal epilepsy. A multicenter retrospective study of 50 patients with Lennox-Gastaut syndrome aged 5 to 27 years at implantation showed that 40% achieved a 50% or greater reduction in seizures at 6 months following VNS placement.38 The median decrease in the frequency of drop seizures and atypical absences was 55% and 73%, respectively, at 3 months compared to baseline, and compared to a median reduction of only 23% for partial seizures. The median reduction in drop seizures and atypical absences increased to 88% and 81%, respectively, at 6 months. No data are given for partial seizures at this time interval. Eleven of 24 patients with generalized seizures (by clinical and EEG criteria), aged 4 to 40 years, had a > 50% reduction in seizures at 3 months,39 and in 14 patients with idiopathic generalized epilepsy, 8 (57%) had > 50% reduction in seizures.40 Five were able to reduce the dose and number of their anticonvulsant medications. Patients with tuberous sclerosis often have epilepsy that is highly refractory to medications. Two small studies have suggested reasonable efficacy (approximately 50% of patients with a 50% reduction in seizure frequency) of VNS therapy in this difficult-to-manage subset of patients.41,42

Surgical complications are relatively rare in children (0% to 3.5%) and mainly include wound/deep pocket infection.43,44 Stimulus-related adverse effects of cough, dysesthesia, voice alteration/hoarseness, and hypersalivation are relatively common (up to 39%)44 but are often transient and can be reduced by alteration in stimulus parameters if persistent. Normal functions of the vagus nerve include effects on bronchial caliber, pulmonary secretions, heart rate, gastric and pancreatic secretions, peristalsis, and baro- and chemoreceptors in the great vessels. Effects of implantation and chronic stimulation of the nerve are gradually being recognized. Obstructive sleep apnea is worsened by VNS treatment in some adult patients,45 and similar findings are postulated in children but are not as yet well documented.46,47 Caution is recommended in using this treatment in children with preexisting obstructive sleep apnea. Other vulnerable groups requiring careful assessment may be those with swallowing dysfunction, with the risk of increased aspiration, and those with cardiovascular disease.