The ketogenic diet: an appropriate first-line therapy?

There are 28 anticonvulsants currently being used worldwide for the treatment of epilepsy, many of them introduced in the past 15 years. Older anticonvulsants, such as phenytoin, carbamazepine, phenobarbital and valproate, are often used first-line, based on anecdotal experience. However, according to the American Epilepsy Society and the American Academy of Neurology, only four of the newer medications have sufficient evidence to warrant their use as monotherapy (lamotrigine, oxcarbazepine, gabapentin and topiramate) [1]. Despite this, most child neurologists will use an anticonvulsant first if the clinical situation warrants its use, for example, a child with epilepsy and migraine may be treated with topiramate first, another with Lennox–Gastaut syndrome may start with rufinamide, and someone with benign rolandic epilepsy could conceivably be treated originally with levetiracetam.

The ketogenic diet (KD) is a high-fat, low-carbohydrate, nonpharmacologic treatment for epilepsy that has been in continuous use since 1921 [2]. For the past 89 years, the KD has been used almost universally for intractable epilepsy after multiple anticonvulsants have been tried unsuccessfully, but at what point does this occur? According to the 2009 International Ketogenic Diet Group Consensus Statement, the KD should be strongly considered after no more than two anticonvulsants have been tried, not the five to ten that, sadly, are often tried before the child is referred to a KD center [3]. However, all treatments for epilepsy are more effective in published clinical trials when used earlier in the course of epilepsy. What would it take to justify the use of the KD first?

The logical answer would be the presence of absolute or relative indications for its use: situations in which the KD is scientifically proven to be highly effective or, even more ideally, of greater efficacy than anticonvulsants. These indications for the KD exist and were the first and perhaps most important table in the 2009 Consensus Statement [3]. Two of these conditions for which the KD can be the only appropriate treatment, glucose transporter 1 deficiency and pyruvate dehydrogenase deficiency, are predominantly metabolic conditions notable for generally infrequent seizures. Glucose transporter 1 deficiency has recently been expanded in its clinical phenotype, including absence epilepsy, but still remains a somewhat rare disease [4].

Certain epilepsy conditions, including tuberous sclerosis complex, Rett syndrome, severe myoclonic epilepsy of infancy (Dravet syndrome) and mitochondrial disorders, also respond very well to the KD, but no studies have shown superiority over anticonvulsant medications [3]. The use of the KD earlier in these conditions rather than after many anticonvulsants have been used is probably warranted, but obviously not as a first-line treatment. Myoclonic–astatic epilepsy (Doose syndrome) is an epilepsy syndrome in which young school-age children, most commonly boys, present with the sudden onset of myoclonic and atonic seizures despite normal intelligence and a generally normal background EEG [5]. Several retrospective studies of myoclonic–astatic epilepsy from centers in Japan, Argentina and the USA have described 40–50% seizure-free response rates, which are superior to anticonvulsants [5–7]. In fact, a 2007 study by the group at the Children’s Hospital of Philadelphia (PA, USA) found that the KD was the thirteenth treatment typically prescribed for myoclonic–astatic epilepsy, but by far the most effective, leading the authors to propose that the KD “should be considered earlier in treatment” [5].

Perhaps the most convincing evidence exists for infantile spasms, a severe epilepsy syndrome affecting infants, typically aged 4–8 months, with daily tonic seizures and often developmental regression. Treatments are limited and fraught with financial implications and serious adverse effects; however, they are effective, especially corticosteroids (adrenocorticotropin hormone and oral high-dose prednisolone), demonstrating approximately 60–70% spasm-free rates, and vigabatrin, with 40–60% spasm-free rates [8]. The KD has been reported as effective in reducing spasms in large single-center retrospective studies from the USA and Korea, with evidence suggesting earlier treatment is most advantageous [9,10]. We reported our experience in using the KD as a first-line treatment for infantile spasms in 2008, typically offered to infants who presented to medical care within 2 weeks and without metabolic or nutritional concerns [11]. When used, we counsel parents that the KD will be discontinued, and corticosteroids initiated, if not completely effective in eliminating spasms within 2 weeks. To date, the KD has been successful in 11 out of 19 infants (58%), often within 5 days. Interestingly, the most common reasons families choose not to try the KD first when offered at our institution are a desire to continue breastfeeding and a reluctance to have the infant admitted for the several-day KD initiation.

Other than epilepsy syndromes, there are several clinical scenarios for a child with epilepsy that suggests a neurologist should consider the KD early. Recent evidence from France suggests that a worsening seizure frequency within the past month correlates better with success than relative stability [12]. Many neurologists have interpreted these results as predictive of the beneficial role of the KD in status epilepticus. Another reported condition is that of a child with severe epilepsy who is formula-fed only [13,14]. Most of these children are infants but there are also many with gastrostomy tubes for nutrition. There are three KD formulas in existence, including Nutricia’s (MD, USA) KetoCal^®, Solace Nutrition’s (MD, USA) KetoVolve^® and Ketonia™ in South Korea. Not only have two studies suggested that formulas may have superior efficacy compared with solid food KDs [13,14], but they probably have a lower incidence of dyslipidemia, due to lower cholesterol content [15].

It seems logical that the KD should be used sooner in all of these epilepsy syndromes and situations. However, there are several unfortunate practical aspects inherent to the KD that makes it perhaps a less attractive first-line option. In order to start the KD first, the KD center must be willing to begin the KD urgently (that day) and quickly mobilize their KD neurologist, dietitian, kitchen and educational team to teach families. Insurance must approve the use of the KD, which can already be problematic, even for children with intractable epilepsy, and requires days of coordinating. When used, ketogenic formulas have to be ordered and shipped quickly to be immediately available in the child’s home.

As well as these logistical problems, there are more subjective factors that may play a role in deciding whether to use the KD first. Parents have to allow the KD time to work, sometimes days as in the case of infantile spasms, but occasionally longer. The KD is absolutely a lifestyle change for families, even for those children on formulas. Evidence for this came this year in a study surveying parents of children who had discontinued the KD years prior, many of whom had intractable epilepsy and became seizure-free with the KD [16]. When asked, surprisingly only 48% would have tried the KD before medications years ago had they been given the choice back then by their treating neurologist [16]. These results confirm our suspicion that for most families medications are simpler and easier than dietary treatments and, even if successful, the KD can be a hardship. Would adolescents or adults agree to disrupt their daily life in a public way with a dietary change, or would they decide to try a more private, easier tablet twice daily? The failure to recruit patients for a dietary study of chronic migraine in adolescents, enrolling only eight patients over a 4-year period, suggests diets are difficult decisions for teens [17]. Lastly, would pediatricians and emergency department physicians refer children to KD centers for dietary treatment or would they start a medication that day? Child neurologists have to be strong advocates and have confidence in the efficacy of the KD to use it first-line. In my personal experience using the KD for new-onset infantile spasms, one major factor in whether a family decides to implement the KD is which particular child neurologist consults and then counsels the parents.

Despite all of these problems in using the KD first, there may be several good solutions. The modified Atkins diet and low glycemic index treatments are recent, ‘alternative’ KDs that have been demonstrated in clinical trials to be safe and effective for children and adults with intractable epilepsy [18,19]. Advantages include an out-patient initiation, which requires an hour of instruction typically, a lack of protein, calorie or fluid restriction, which appeals to older children and adolescents, the presence of copious recipes and information on the internet and in books, and perhaps a reduced need for dietitian support and direct involvement [20]. These less restrictive diets could potentially be more appealing to both a neurologist and family considering diets versus medications. Another potential solution may be to increase education about the KD to dietitians and child neurologists and, even more importantly, to the pediatricians and emergency department physicians who are on the ‘front-line’ when epilepsy begins. Lastly, creating an ‘easier’ KD, especially for those receiving formula-only diets, would be helpful. For infantile spasms specifically, if an internet-located, standardized and age-/weight-based, formula program was widely available, more dietitians could quickly create and ship the KD to families.

In my opinion, there is currently sufficient evidence to warrant the use of the KD as a first-line therapy in very select situations. First, select epilepsies (e.g., infantile spasms and myoclonic–astatic epilepsy) and conditions (e.g., infants and those with gastrostomy tubes). Second, select families, including those who are motivated, compliant and willing to give the KD time to work. Lastly, select child neurologists and dietitians knowledgeable in the KD who are confident in its efficacy and able to implement it within 1–2 days. I think this concept of KD initial ‘monotherapy’ will continue to grow in interest, especially as a result of the increasing use of the modified Atkins diet, simplified ketogenic formulas and worldwide awareness of the value of dietary treatments.

Acknowledgements

The author gratefully acknowledges the content review by Gerry Harris.

Financial & competing interests disclosure

Eric H Kossoff is on the Scientific Advisory Board of Atkins Nutritionals, Inc. He has research support for a clinical study from Nutricia, Inc., related to their product KetoCal^®. The author has no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

No writing assistance was utilized in the production of this manuscript.