Does the effectiveness of the ketogenic diet in different epilepsies yield insights into its mechanisms?

Summary

The ketogenic diet (KD) has been used successfully in a variety of epilepsy syndromes. This includes syndromes with multiple etiologies, including Lennox-Gastaut syndrome and infantile spasms; developmental syndromes of unknown etiology, such as Landau-Kleffner syndrome; and idiopathic epilepsies, such as myoclonic-astatic (Doose) epilepsy. It also includes syndromes where genetics play a major role, such as Dravet syndrome, tuberous sclerosis, and Rett syndrome. Study of the KD in humans and animals harboring various genetic mutations may yield insights into the diet’s mechanisms. Comparison of the diet’s effectiveness with other treatments in specific syndromes may be another useful tool for mechanistic studies. The diet’s utility in epilepsy syndromes of various etiologies and in some neurodegenerative disorders suggests it may have multiple mechanisms of action.

The ketogenic diet (KD) has been used successfully in the treatment of different types of epilepsy for nearly 90 years. Given its diverse clinical utility, one question is whether the experience in different forms of epilepsy offers insight into the diet’s potential mechanisms of action. Despite meta-analyses demonstrating the diet’s effectiveness, one recent survey of practicing pediatric epileptologists showed the diet was not perceived as particularly useful in a variety of clinical scenarios, including Lennox-Gastaut syndrome (LGS; Lefevre & Aronson, 2000; Wheless et al., 2005; Henderson et al., 2006). In contrast, expert-derived algorithms have included the KD fairly early in the treatment course for certain epilepsy syndromes (Sankar et al., 2005). One of the largest studies of the KD did not demonstrate greater efficacy in any particular seizure type (i.e., myoclonic/drops vs. tonic-clonic), although epilepsy syndromes were not delineated specifically (Freeman et al., 1998). Although patients with partial-onset epilepsy might not be as likely to achieve complete seizure freedom, there is no overall difference in the diet’s effectiveness between patients with generalized epilepsy and those with partial-onset seizures (Maydell et al., 2002; Than et al., 2005). After delineating the diet’s effectiveness in various epilepsy syndromes (with selected citations), the utility of studying syndromes with known genetic mutations will be discussed. The diet’s application in different types of epilepsy suggests it might have multiple mechanisms of action.

The ketogenic diet in specific epilepsy syndromes

The KD has been shown to be useful in a variety of epilepsy syndromes. It is very effective in decreasing the incidence of atonic or myoclonic seizures in children with LGS, even within the first 48 hours of its application (Freeman & Vining, 1999). Most of the data on children with LGS (the epilepsy syndrome in which the aforementioned survey showed its greatest utility, ranking as the 5^th or 6^th option) typically are embedded in other studies of the diet, where type of seizure (rather than epilepsy syndrome) was the variable examined (Schwartz et al., 1989; Freeman et al., 1998). It can be presumed that many patients with atonic and myoclonic seizures in these series had LGS, although some may have had myoclonic-astatic (Doose) epilepsy (MAE). A more recent report demonstrated a significant rate of seizure freedom (30%) after 12 months in patients with LGS (Kang et al., 2005). Many types of pathology can lead to this clinical syndrome, so it is difficult to draw any mechanistic clues from this experience. The KD also has been shown to be useful in case series of acquired epileptic aphasia (Landau-Kleffner syndrome), another epilepsy syndrome with developmental pathology (Bergqvist et al., 1999; Kang et al., 2005).

The KD’s mechanism might be derived from comparisons with other treatments for epilepsy (Hartman et al., 2007). Clinical experience with the KD in infantile spasms (IS) has been outlined elsewhere in this monograph, but its utility offers the opportunity to consider whether the diet’s mechanism is similar to medicines currently used for this indication. As with LGS, many different types of pathology are associated with IS and a number of different anticonvulsant pathways may be useful in treating IS. These include the CRF-ACTH-cortisol axis or a GABA-related mechanism (the latter suggested by the utility of vigabatrin for infants with tuberous sclerosis and IS) (Lux et al., 2004; Mackay et al., 2004). Another potential mechanism is the neurosteroid pathway, given the utility of ACTH (which promotes both cortisol and neurosteroid production in the adrenal cortex) and ganaxolone (a synthetic neurosteroid currently undergoing a Phase II clinical trial) in some children with IS (Rogawski & Reddy, 2002). The development of animal models of infantile spasms may provide a way to test these hypotheses (Velísek et al., 2007; Lee et al., 2008).

The KD’s effectiveness in genetic syndromes also offers the opportunity for gaining mechanistic insights. Dravet syndrome (also known as severe myoclonic epilepsy of infancy) is one example of an epilepsy syndrome with a variety of genetic mutations in a voltage-gated sodium channel (Scheffer et al., 2001). One case series demonstrated the KD resulted in a greater than 75% improvement in seizure frequency in half the patients, while another case series showed a greater than 50% improvement in seizure frequency in over 60% of patients (Caraballo et al., 2005; Kang et al., 2005). Unfortunately, the genotype of these patients is unknown, so it is difficult to draw definitive pharmacogenomic conclusions about the diet’s mechanism of action in these patients.

Interestingly, different types of mutations in the same gene are involved in some patients with GEFS+ (reviewed in Helbig et al., 2008). Comparisons of the diet’s efficacy in patients with similar clinical syndromes but different genetic mutations (e.g., Dravet syndrome with either SCN1A or GABRG2 mutations) offer a unique tool that has not been used thus far. Multicenter trials of the KD probably would be necessary, based on the small numbers of patients seen in individual clinics with each disorder. Use of the KD in animals harboring mutations seen in these syndromes (e.g., transgenic mice) might provide important mechanistic information (Dutton & Escayg, this Supplement).

Rett syndrome is another example of an epilepsy syndrome where a variety of genetic mutations in a single gene product account for a significant number of cases. In Rett syndrome, a defect in carbohydrate metabolism was proposed based on elevated pyruvate (blood, CSF) and lactate in some patients (Haas et al., 1986). This prompted the use of the KD in a number of patients, with good results in terms of seizure control, social interaction, and stereotypical behaviors (Haas et al., 1986). An opportunity for translational work exists in a transgenic mouse model of Rett syndrome that has decreased levels of ATP, glutamine, and glutamate, measured with magnetic resonance spectroscopy (Saywell et al., 2006). Given other rodent data showing an increase in the brain’s energy state and specifically, glutamine and glutamate levels during the KD (Appleton & DeVivo, 1974; Bough, et al., 2006), it would be interesting to see if the diet corrected these deficiencies.

Tuberous sclerosis is a genetic syndrome where the KD has been useful (Kossoff et al., 2005). It may be one of the conditions in which seizures recur after the diet has been discontinued after two years of seizure freedom (Martinez et al., 2007). Mutations in one of two proteins, hamartin and tuberin, are seen in many patients with tuberous sclerosis. Tuberin is phosphorylated by AMP-activated kinase (AMPK) in response to certain stresses, such as energy starvation (Inoki et al., 2003). The KD might represent one form of cellular stress that involves the AMPK pathway, as preclinical data (cited previously) show a change in the brain’s energy state during the KD. Interestingly, inhibition of one component of this pathway with rapamycin prevents the development of epilepsy in mice harboring a TSC1 mutation (Zeng et al., 2008). Whether the KD has a similar mechanism in this condition remains to be seen.

Not all clinical experience with the KD has been positive. Lafora body syndrome, one of the neurodegenerative progressive myoclonus epilepsy syndromes that involves defects in two proteins critical to glycogen synthesis, laforin (EPM2A) and malin (EPM2B), would appear a priori to be a particularly useful application of the KD (Cardinali et al., 2006). The rationale of this study was that the KD decreases glycogen synthesis and therefore, it may decrease Lafora body accumulation and progression of disease. Only five patients were studied in this series, but there was both clinical and neurophysiological evidence of continued deterioration in four of the five patients. The authors suggested that earlier implementation of the diet or studying a limited genotype of the condition might have yielded different results.

Genetic mutations in identified epilepsy syndromes are one means of dissecting the KD’s mechanism, but the study of syndromes treated with certain medications or adjunctive treatments may provide a different and potentially complementary tool. MAE is one example of a clinical syndrome where comparisons have been made between the KD and medications (Oguni et al., 2002). In this series, the KD was the most effective treatment studied, eliminating myoclonic and astatic seizures in fifteen of 26 patients. Drugs with GABAergic actions (e.g., benzodiazepines and valproate) were not as successful as the diet in this series; ACTH and ethosuximide showed intermediate results. These data suggest the diet may not be acting in a GABA- or glucocorticoid-related manner in MAE. Further comparisons of the diet to medications in other epilepsy syndromes might be useful for suggesting relevant neurotransmitter or biochemical pathways.

Conclusion

The KD’s utility in a variety of different epilepsy syndromes and other diseases (discussed elsewhere) suggests that like many other neurotherapeutic agents, it may have multiple mechanisms of action, each of which may be more relevant to a specific disease state. For example, the diet induces the liver to generate ketone bodies, which probably serve as an alternative fuel source for children with certain blocks in metabolism such as GLUT-1 deficiency. Ketone bodies might serve a similar function in certain neurodegenerative disorders or in cancer (Reger et al., 2004; Seyfried, this Supplement). Inconsistencies in studies attempting to correlate seizure protection with levels of ketone bodies suggest something other than ketone bodies may be involved in the KD’s beneficial effects in epilepsy (Huttenlocher 1976; Ross et al., 1985; Thio et al., 2000). The diet’s utility in genetic syndromes and comparisons with medications with known mechanisms are two strategies that offer the possibility of further study of its anticonvulsant (and possibly disease-modifying) effects. Its utility in a variety of different epilepsies suggests that it might have multiple mechanisms of action.