The impact of microbiota and ketogenic diet interventions in the management of drug-resistant epilepsy

1 INTRODUCTION

Epilepsy is one of the most common neurological disorders, affecting more than 50 million people worldwide and 6 million people in Europe.^{1, 2} Epilepsy groups heterogeneous disorders that affect people of all ages and whose common feature is epileptic seizures.^2-5 Uncontrolled seizures can lead to developmental delay, cognitive deficits, memory and learning difficulties, and they can also cause sudden unexpected death (SUDEP).⁶

Current treatments include anti-seizure medicines (ASM), medical diets, epilepsy surgery and stimulation devices. ASM can control seizures in about two-thirds of people with epilepsy. However, about one-third of patients suffer from drug-resistant epilepsy (DRE), that is, epilepsy that is not controlled by two or more correctly selected and dosed ASM.^7-10 The mechanisms underlying drug treatment failure are multifactorial,⁸ and a re-evaluation of the concept of DRE, including a more comprehensive definition of the different subtypes, is currently underway.¹¹

The causes of epilepsy include structural, genetic, infectious, metabolic and immune factors.^12-16 However, the cause of epilepsy remains unknown in a significant proportion of patients. Epileptogenesis may result from a combination of various factors, including lifestyle and environmental causes, such as the use of toxic substances (such as alcohol and tobacco) and certain medications during pregnancy. Meningoencephalitis can also cause inflammation of the brain and increase the risk of epilepsy. The gut-brain axis has also been implicated in this respect.¹⁷

Changes in the gut microbiota can affect brain function and lead to inflammation in the gut and brain, which may contribute to epileptogenesis (Figure 1). The gut microbiota produces a variety of compounds that can affect brain function, such as neurotransmitters and short-chain fatty acids (SCFAs).^18-20 Some of these compounds have been shown to have anticonvulsant effects in rodents.^21-23 However, in patients, the mechanisms by which the microbiota might cause or contribute to epileptogenesis are completely unclear.

Here, we provide a timely mini-review on the impact of microbiota and diet in patients with DRE. We highlight open questions in the field with the aim of suggesting future directions. Gaining more knowledge in this area is likely to pave the way for new and more effective therapeutic approaches, such as in vivo modulation of the microbiota to ameliorate the burdens in DRE.

2 ROLE OF THE MICROBIOTA IN DRUG-RESISTANT EPILEPSY: MICROBIAL INTERVENTIONS TO TREAT DRUG-RESISTANT EPILEPSY

Evidence for a relevant role of the gut microbiota in human epilepsy comes from few and correlative studies. Descriptive studies with relatively limited sample size have suggested that gut commensals differ between individuals with drug-sensitive (DSE) and drug-resistant (DRE) forms of epilepsy (summarized in Table 1). Some of those studies showed an enrichment of various gut microbes in patients with DRE in comparison with drug-sensitive patients and healthy controls, such as Clostridium XVIII, Dorea, Coprobacillus, Ruminococcus, Akkermansia, Neisseria, and Coprococcus²⁴ or Actinobacteria, Verrucomicrobia, Nitrospirae, Blautia, Bifidobacterium, Subdoligranulum, Dialister and Anaerostipes.^25-28

In fact, various microbe-based or microbiota-altering interventions have shown promising efficacy in reducing seizure frequency in patients with DRE. Fecal microbiota transplantation (FMT), initially used to treat symptoms of Crohn's disease (CD), had a positive effect on the frequency of seizures in a young woman who had also suffered from epilepsy for 17 years. Prior to the procedure, the patient underwent ASM treatment to maintain seizure control.³¹ Interestingly, FMT not only alleviated the symptoms associated with Crohn's disease, but also the patient remained seizure free for over two years without ASM. This suggests that the improvement in epilepsy symptoms may have been due to the effect of FMT on the patient's gut microbiota, which in turn may have altered the interaction of the microbe with central nervous system function, suggesting the potential of FMT as a treatment for epilepsy and also in reducing systemic inflammation.³¹ In fact, clinical trials aimed at understanding the efficacy and safety of FMT in the treatment of epilepsy began several years ago, but data are not yet available (ClinicalTrials.gov ID NCT02889627).

Other studies have confirmed that probiotics or microbiota-modifying treatments can decrease seizure frequency. In a pilot study, the use of a mixture of probiotics (eight bacterial subspecies of Lactobacillus, Bacteroides and Streptococcus) reduced seizure frequency and improved the quality of life in patients with DRE.²⁹ Unfortunately, in this study the gut microbiota of the treated individuals was not analyzed.

Further evidence showing that the microbiota may play a role in seizures comes also from a study of six patients with DRE which were treated with antibiotics.³⁰ While on antibiotics, five patients were seizure-free and the sixth reported a significant reduction in the seizure frequency. Here, the antibiotics were used to treat different infections, such as pneumonia otitis or pertussis and some have known inhibitory activity on cytochrome p450, enzymes which increases serum concentrations of some ASMs.^{30, 32} The patients were exposed to different ASM treatments, and one of them was also treated with a ketogenic diet. The positive effect of the antibiotics on seizure frequency disappeared in all the patients within 2 weeks after stopping the antibiotic treatment.³⁰

In conclusion, human studies of microbiota modulation by FMT, probiotics or antibiotics, although few and with small numbers of patients, show improvement in terms of reduction of seizure frequency, suggesting the microbiota as a possible target for therapeutic strategies in the management of DRE.

3 KETOGENIC DIETS IN THE MANAGEMENT OF DRE

Alternative treatments for DRE include non-pharmacological approaches (e.g., vagal nerve stimulation) and dietary treatments such as ketogenic diets (KDs),^33-35 which have been shown to improve conditions in not only metabolic diseases but also several neurological disorders. Restricting carbohydrate intake forces the metabolism to replace the preferred energy source, glucose, with fatty acids, which are oxidized in the liver to form ketone bodies. The latter can be used by cells that cannot metabolize fatty acids, such as the brain,^{36, 37} to generate energy.

Since ancient times, prolonged starvation has been widely used to treat patients with epilepsy.³⁸ In 1921, Dr. Wilder introduced the hypothesis that fasting induces a beneficial effect related to ketonemia in patients with epilepsy and proposed that KD would have had similar effects, due to its property of inducing a metabolic state of ketosis in the host, but without the malnutrition effects induced by starvation.³⁹ KD is a form of dietary intervention based on a high-fat intake accompanied by moderate protein, and very low or no carbohydrate. The classical ketogenic diet has been extendedly used in DRE management. Modifications of the composition in terms of fats, carbohydrates and proteins have led to less restrictive forms of the diet, such as the modified Atkins diet (MADKD), the low-glycemic index diet (LGID), and the middle-chain triglyceride ketogenic diet (MCTKD), all of them bonded with the term of (non-classical) ketogenic diets (KDs) (Table 3).

In this chapter, we will focus on what is currently known about the effect of KDs on the microbiota in DRE (Table 2). While there are now a considerable number of studies investigating the effects of classical KD on gut microbiota changes and seizure frequency, to the best of our knowledge, we have not found studies demonstrating the effects of modified ketogenic diets on the gut microbiota in patients with DRE. However, in line with the aims of our review to highlight the impact of KD on the management of DRE and to emphasize the need for personalized dietary plans according to patients' needs (e.g., food intolerances, cultural factors or food preferences), we consider it valuable to share information on the success of ketogenic therapies other than the classic ketogenic diet, even in the absence of results related to gut microbiota composition. By doing so, we also want to emphasize the need for and importance of conducting microbial studies in DRE patients following different forms of ketogenic diets. Alternatives to the classical KD, which offer the same neurological benefits, could improve dietary compliance and facilitate comparative studies on variations in gut microbiota composition.

3.2 Modifications of the classical KD

As discussed, classical KD has shown good results in the management of DRE, although adherence to the diet has also been reported as one of the main challenges due to its strongly restrictive carbohydrate allowance. More flexible diets have been established as alternative therapeutic approaches to classical KD for patients with DRE, such as the modified Atkins diet, the low-glycemic index diet and the medium-chain triglyceride ketogenic diet (Table 3).

TABLE 3. Main differences among the ketogenic diets used in DRE.

	Classical KD	MAD	LGID	MCTD
Composition
Ratio fat: protein + CH	4:1	1:1	1:0.6	2:1
Restrictions	High. Food must be carefully weighted	Moderate. No calories restriction or fluids. No need to weigh the food CH: 10-15 g/day, up to 30 g/day when seizures are improving	Moderate. No need to weigh the food CH: 40-60 g/day, and only CH with glycemic index <50	Moderate. Admits more variety of food. No need to weigh the food
References	Kossoff et al.48; Sharma and Jain49; Verrotti et al.50; Yoon et al.51

• Note: fat; carbohydrate and protein.
• Abbreviations: CH, carbohydrates; KD, ketogenic diet; LCT, long-chain triglycerides.LGID, low-glycemic index diet; MAD, modified Atkins diet; MCT, middle-chain triglycerides; MCTD, medium-chain triglyceride diet.

3.2.1 Modified Atkins diet

The modified Atkins diet (MAD) is a variation of the Atkins diet that is used to treat epilepsy. The classic Atkins diet is a high-fat, low-carbohydrate diet developed by Dr. Robert Atkins in the 1970s, initially intended for weight loss. Due to the restricted amount of carbohydrates (20 grams per day) the Atkins diet was reported to induce ketosis and therefore the production of ketone bodies,⁵² which has encouraged its use in DRE. Indeed, three out of six patients aged between seven and 52 years old, showed a >50% reduction in their seizure frequency following the introduction of the Atkins diet. Several studies then moved on to the MAD, which limits carbohydrate consumption to 10 g per day for children and 15 g for adults, and unlike the classical KD, fats account for around 65% of calories and there is no restriction on protein (which accounts for around 30% of calories).^{49, 53, 54} In contrast to the classical KD, where most target patients are children, MAD has been tested in adult cohorts with promising results. In one clinical trial, almost half of the patients had at least 50% reduction in seizures after one and 3 months on MAD, with improvement starting as early as the first 2 weeks of treatment, and after three and 6 months, two patients were seizure-free.⁵⁴ In another study, 66% of patients had >30% reduction in seizure frequency after 6 months of MAD,⁵⁵ and in an observational study, 36% of patients with DRE had a ≥50% reduction in seizure frequency and 16% were seizure-free after 3 months of MAD. Approximately one quarter of patients had <50% seizure reduction, no improvement or worsening seizures after 3 months of treatment.⁵⁶

A recent clinical trial involving 160 patients between the ages of 10 and 55 showed that MAD combined with drug treatment was more effective in reducing seizure frequency than ASM therapy alone. Around 26% of the patients receiving MAD and ASM together showed a 50% reduction in seizure frequency and improved quality of life within 6 months after starting the diet.⁵⁷ Similarly, a randomized controlled trial showed that the number of children with a >50% reduction in seizure frequency after the introduction of MAD was higher than the number of children in the group receiving ASM alone after 12 weeks of treatment.⁵⁸

Several studies have highlighted the benefits of the MAD in patients with DRE, showing an overall improvement of >50% of seizure frequency in approximately 50% of the patients.^{48, 53-55, 59, 60} Interestingly, there were no significant differences between MAD and KD when comparing children with 90% seizure frequency reduction after 6 months of diet, but KD was more effective in cases with >50% seizure reduction.⁶¹

In general, these results suggest that MAD is a good alternative to KD, as patients appear to be more compliant with this diet compared to KD, with no adverse effects in both the short and long term.

3.2.2 Low-glycemic index diet

The low-glycemic index diet is used as a more flexible alternative to the classical KD, as patients are allowed to consume a higher amount of carbohydrates (approximately 40–60 g) daily,⁵¹ which are monitored and should have a low-glycemic index (<50 glycemic index). Calories from fat should account for about 60% of the total intake.⁵¹

A study in children aged between 1.5 and 17 years showed that more than 70% of patients (of a total of 42) had a ≥50% reduction in seizures after just 2 weeks on a low-glycemic index diet.⁶² Kim and colleagues⁶³ showed that more than half of the children with DRE (36 patients in total) on low-glycemic index diet had ≥50% reduction in seizure frequency after 3 months of treatment, and 6% were seizure-free for 1 year, with very low incidence of adverse events. Similarly, a retrospective study in Italy⁶⁴ showed a ≥50% reduction in half of the patients (n = 15) in a cohort of children and young adults (11.3–22 years old). This diet has also been used in other conditions, such as the Angelman syndrome, which is characterized by drug-resistant seizures. Here, patients on this diet showed a reduction in seizures or were even seizure-free, with very mild to low side effects.⁶⁵

Although this diet is more palatable than classical KD and offers good clinical outcomes for patients with DRE, studies of low-glycemic index diets have mainly been conducted in children or young adults, so the effect of the diet in the adult population is not well established.

3.2.3 Medium-chain triglyceride ketogenic diet

The fats present in the classical KD are mainly of long-chain triglycerides (LCT). Alternatives are medium-chain triglycerides (MCT), which can be used to produce ketones more easily than long-chain triglycerides, thus reducing the total fat required and allowing more carbohydrate and protein to be included in the diet.^{50, 66} The medium-chain triglyceride ketogenic diet was first described in the 1970s and provided up to 60% of the calories in the form of MCTs.⁶⁷ Due to tolerability problems,^{68, 69} the diet was modified to include 30% of calories from the MCT and 30% from LCT.⁷⁰

The effects of the MCT ketogenic diet have been studied in patients with DRE. Indeed, after 3 months on the new diet, children showed a significant reduction in seizure frequency compared with the baseline. More than 60% of the patients (n = 16) showed a 50% reduction in seizure frequency and almost 30% were seizure-free.⁷¹ In addition, supplementation with the MCT oil in six adult patients with DRE showed a reduction in the seizures rate⁷² and a comparison between MCT and KD in 94 children with DRE showed that MCT was as effective as KD in reducing the number of seizures.⁶⁶ Therefore, good management of the MCT diet could provide a good substitute for those patients who cannot maintain KD or other dietary alternatives.⁷³

Taken together, this evidence suggests that modifications of the classical KD could provide similar success rates in reducing seizure frequency⁷⁴ and this opens the possibility of using less restrictive diets to treat DRE patients. However, the reality is not as simple and unanswered questions remain. In a randomized controlled study, in pediatric patients⁷⁵ the MAD and the low-glycemic diet were not as effective as classical KD, whereas LGD offered fewer adverse effects than MAD and KD with a good balance of seizure reduction.

4 CONCLUSIONS AND FUTURE PERSPECTIVES

As we discussed in the previous chapters, there is increasing evidence that the gut microbiota is an important player in drug-resistant epilepsy and an essential mediator of dietary effects on the host organism (summarized in Tables 1 and 2). Indeed, diet is one of the major forces shaping the microbiota in terms of composition and metabolism, contributing to host homeostasis and disease susceptibility. Nutrition is currently a complementary and alternative approach in the treatment of neurological disorders, including epilepsy, with beneficial effects particularly in drug-resistant patients. In this review, we aim to highlight the importance of the diet and its interactions with the host microbiota as a new and promising approach to the management of DRE. We have pointed out here that some DRE patients treated with KD respond efficiently to dietary intervention, with >50% reduction in seizures, whereas another proportion of patients do not respond, with <50% reduction in seizures frequency even after several months of KD treatment. The reasons for this variable response to KD among patients remain completely unclear. One of the major challenges in clinical studies is translating findings from small cohorts into effective approaches for more patients with DRE (and other neurological disorders). Because of the heterogeneity of DRE and the interpersonal differences in the composition of the patients' microbiome, and thus possible differences in dietary interaction effects, larger cohort studies are needed. Positive effects after dietary intervention depend on the metabolic and microbiome profile of the individual, highlighting the importance of personalized dietary treatments.⁷⁶

In line with this, detailed analyses of the impact of potential confounding factors (e.g., sex, age, ethnicity, ASM use, environmental and lifestyle factors, including diet) on the microbiota composition/function in DRE patients and its impact on the modulation of seizure frequency upon KD introduction are still almost completely lacking.⁷⁷ Therefore, analyzing the microbiota composition and function, such as the metabolic profile, of each patient before and after dietary intervention will pave the way to more personalized treatments and even predict the success of the intervention, similar to what has been shown in other diseases.⁷⁸

The classical ketogenic diet is perceived as a restrictive diet, and its consumption could also cause some adverse effects, such as nausea, vomiting, diarrhea, constipation and weight loss, which could contribute to lower acceptance or adherence to the diet. Modifications of classical KD, termed as ketogenic diets, have been developed to reduce the side effects and improve acceptance and compliance to the diet. However, there is a paucity of data on the effect of ketogenic diets on the composition of the patient's microbiota. It is also unclear whether ketogenic diets would improve seizure frequency and quality of life in those patients who do not respond to classical KD, as there is very little data on this topic. To address these unanswered questions, it is necessary to understand the effect of the diet on the host microbiota and on the host central nervous system. Analysis of the gut microbial composition together with bacterial metabolic functions would help to identify molecular pathways or metabolites that could be used in the clinic. It has been suggested that the gut microbiota can influence the phenotype and function of mucosal and systemic immune cell responses and central nervous system function through the dissemination of bacterial products or metabolites, and this is strongly dependent on the substrates available through dietary intake.⁷⁹ Identification of the specific bacterial products and metabolites in each dietary state in different patient populations is currently needed to deepen our understanding of the pathogenesis of diseases such as DRE and to develop new and more effective therapeutic strategies. Interestingly, only bacteria such as Akkermansia muciniphila and Parabacteroides merdae mediated seizure protection in germ-free (GF) KD-fed or antibiotic-treated specific-pathogen-free (SPF) KD- or control diet-fed mice.²³ The authors suggested that KD may be responsible for changes in the expression of neuroactive molecules, such as α-glutamyl-amino acids, α-aminobutyric acid (GABA) and glutamate, which may contribute to seizure protection. Despite these promising results, information on similar pathways in humans is lacking. Addressing this question through metagenomics and metatranscriptomics of the microbiota before and after KD treatment, together with metabolomics on host samples (such as serum), would provide comparative results across human populations, and ultimately lead the way to new targets for treating DRE and other neurological diseases through personalized dietary or pre- and probiotic interventions (Figure 2).

CONFLICT OF INTEREST STATEMENT

The authors declare no commercial or financial conflict of interest.