The Ketogenic Diet in Pediatric Epilepsy

Jong M. Rho, M.D.

Key Points
--Ketogenic diet helps control seizures in up to two-thirds of children refractory to anticonvulsant drugs
--Children with partial seizures may not respond as well to the diet
--The role of ketone bodies as markers of seizure control is imprecisely defined
--Their role as direct anticonvulsant agents is also unknown
--Tests for transporter defects and enzyme deficiencies help screen diet candidates
--Important clinical questions about the ketogenic diet remain unanswered


Most practicing child neurologists possess a basic understanding of the ketogenic diet---the highfat, low-protein, and low-carbohydrate diet for treating children with epilepsy who do not respond to or cannot tolerate drugs. This diet mimics the biochemical changes associated with starvation and induces, among other changes, production of ketone bodies (mainly beta hydroxybutyrate, and to lesser extent, acetoacetate and acetone), which has been implicated in the mechanisms of seizure control.

Before the historic first announcement of results with the ketogenic diet by the Mayo Clinic [Wilder 1921], bromides and Phenobarbital were the only effective options for antiepileptic therapy. Fasting as a method for controlling epilepsy had been reported sporadically long before this time. However, it was only with such reports in the clinical journals did other
medical centers begin adopting the ketogenic diet as an effective treatment for intractable cases of epilepsy.

Still, the diet remained generally under-utilized, being utilized mainly at institutions such as the Mayo Clinic and Johns Hopkins, until the 1990s when a national television program aired a report on a child whose epilepsy was cured by the ketogenic diet. Later, the Charlie Foundation (named after the young patient) was formed and a television movie recounting Charlie’s success with the ketogenic diet was produced. It also prompted a flood of inquiries to pediatric neurologists and epileptologists about this treatment option.

Even today, however, after a long history of clinical use and the recent surge of professional interest and research, many questions remain unanswered about the ketogenic diet. The purpose of this review is to update neurologists by highlighting key issues and controversies pertaining to this important diet-based therapy for epilepsy. In particular, the review will focus on the potential role of ketone bodies in energy metabolism and in seizure control, the indications and contraindications for the ketogenic diet in pediatrics, and other practical clinical questions involving efficacy, patient selection, monitoring, and side effects.


Diet Overview

The classic ketogenic diet starts with a period of fasting and relative dehydration aimed at achieving ketosis, which is reflected by elevated ketone levels in the urine. There are several variations on this theme (e.g., the medium-chain triglyceride or MCT diet), all based on the principle that one can “force” the organism to use fatty acids and ketones as the main sources of energy, while decreasing the utilization of carbohydrates [Freeman et al, 2000]. This important diet initiation is usually accomplished in a hospital or specialized outpatient setting. An example of the dietary regimen after this introductory phase is listed as follows:

--Restrict total calories to 75% of the recommended daily allowance, with 90% of the calories coming from fat
--Restrict protein to 1 gm/kg
--Provide a very small amount of carbohydrate
--Restrict fluid intake to 60-70 cc/kg/d
--Supplement this diet with vitamins and minerals

The target for the overall ratio by weight of ketogenic foods (i.e., fats) to anti-ketogenic foods (carbohydrates and proteins) is usually 4:1 or 3:1. Maintaining such a high ketogenic potential usually entails a diet of fatty oods, creams, and special oils. A dietitian or a nurse with special
nutrition training is required to tailor ketogenic diets for children and to help families adhere to exacting strict regimen. Also, before being placed on the diet, the child with intractable epilepsy should always be evaluated within the context of a childhood epilepsy center. Only a thorough evaluation in such a setting will ensure that surgical or pharmacological treatments with potentially greater success rates are not overlooked.

Most treatment centers leave the child on the diet for 2 to 6 months to assess potential efficacy. If the patient is seizure-free, as with anticonvulsant therapy, the therapy continues indefinitely for at least a two-year period. With such success, the patient can be weaned from the diet. There is clinical evidence to suggest that early intervention with the diet may even alter the processes of epileptogenesis and render a permanent cure. However, if a patient is taken off the diet (thus breaking ketosis) and seizures recur, it may be difficult to regain control after reinitiation of the diet due to a persistent of high insulin levels in the face of lowered glucagons. Therefore caution is recommended in prematurely removing patients who have been successfully controlled.

Intermediary Metabolism and Ketone Bodies

Elevated levels of ketone bodies have been strongly associated with seizure control and seizure freedom, and all practicing neurologists employ them as biochemical markers of treatment. However, the ketosis produced by the ketogenic diet may not be the main factor in controlling epileptic seizures in children [Schwartzkroin 1999]. Nevertheless, the clinical goal has historically been to achieve high urine ketone levels, and the importance of this time-honored practice can only be appreciated through an understanding of intermediary metabolism.

When the glycolytic pathway is deprived of glucose, as during starvation or the ketogenic diet, free fatty acids are mobilized as substrates for mitochondrial oxidation (Figure 1). In addition, certain amino acids may be converted to ketoacids that can provide other substrates (e.g., alanine to pyruvate) for Krebs cycle activity. The hepatic microsomal system can also convert fatty acids to dicarboxylic acids (via omega oxidation). These dicarboxylic acids require carnitineesterification for urinary secretion [Sankar & Sotero de Menezes, 1999].

Free fatty acids are not readily available to the neuron itself. However, fatty acids can undergo aseries of conversions and translocations to produce acetate substrates for ketone body production. These ketone bodies are carried across the blood-brain barrier (by a fasting-inducible transporter called the monocarboxylic acid transporter) and into the neuron where they are available as an energy substrate for cerebral metabolism.

Thus, one major physiologic role for ketone bodies is to provide an alternative energy substrate for brain and muscle under conditions of fasting or a high-fat diet. In a classic study of fasting obese volunteers, for example, glucose utilization accounted for only 29% of the brain’s oxygen
consumption while ketones extraction accounted for 52% [Cahill 1966]. Playing another major physiological role, ketone bodies act as the principle source of energy during early postnatal development. Furhter, they are the substrates for the carbon skeleton of lipids that comprise the cell membranes of growing brains and organs. Thus, ketones are involved in both the energy supply and lipid biosynthesis of the embryonic central nervous system (CNS).

But do ketone bodies exert a direct antiepileptic effect? Can they modulate neuronal excitability? Several clinical studies have now shown that diet-induced ketosis (especially at very high concentrations) seems to correlates with the level of seizure control. (The most recent studies will be discussed later.) Also, abrupt loss of seizure control has long been known to occur within hours after ketosis is broken [Huttenlocher, 1976].

Thus the compelling question remains: are ketones directly responsible for anticonvulsant activity? Or are they just an epiphenomenon of some other diet-induced physiological change? These questions have been explored in varied experimental settings.
&Mac183; ß One recent animal study, for example, showed that ketone bodies do not directly alter the excitatory or inhibitory hippocampal synaptic transmission. [Thio 2000] Neither beta-hydroxybutyrate nor acetoacetate affected whole cell currents evoked by glutamate, kainite, or gamma aminobutyric acid (GABA) in cultured hippocampal neurons. The ketone bodies also failed to prevent spontaneous epileptiform activity in the hippocampal-enterorhinal cortex slide seizure model.
&Mac183; Results from our laboratory in cultured mouse neocortical neurons were similar, with no effects of the ketone bodies on the classic neuronal targets of anticonvulsants. Investigators should also be aware that beta-hydroxybutyrate is a stereoisomer, with the D-isomer being the biologically relevant species. The non-physiologic L-isomer possesses anticonvulsant activity both in vivo and in vitro, and is due to the presence of a contaminant, dibenzylamine.
&Mac183; Similarities in the chemical structures of beta-hydroxybutyrate and GABA have led to speculation about GABAergic inhibition induced by the ketogenic diet. Results from studies are conflicting, with one showing no changes in whole brain GABA [Al-Mudallal 1996] and another demonstrating that ketonescan increase GABA in synaptosomes. [Erecinska 1996].
&Mac183; Finally, magnetic resonance spectrophotometric techniques have shown elevated levels of cerebral ketones in patients who are successfully controlled by the ketogenic diet [Pan et al., 1999].

Overall, the experimental evidence supporting a direct link between ketone and seizures is far from convincing. Indeed, as with the underlying causes of the seizures themselves, the ameliorating actions of the ketogenic diet may be multiple, with a host of diet-influenced metabolic changes acting in concert to decrease membrane excitability.

But even as research continues, the clinical connection between peripheral ketone levels and seizure control still impels clinicians to confront more practical questions. For example, what assay method should be employed to monitor diet efficacy? Urine dipsticks are commonly used for this purpose but these measure acetoacetate, the less prominent ketone body. Which ketone body actually correlates best with seizure control is unknown. If beta-hydroxybutyrate ketone is actually the preferred marker, a new reflectance meter (Keto-Site™, GDS Diagnostics) will assay the D-isomer from a small drop of blood. But then, what is the “therapeutic concentration”
for either of these ketones? And what does the peripheral level predict about the brain level?

Clearly, many questions remain about the physiological relevance and the practical utility of monitoring ketone bodies in the ketogenic diet.


Special Indications and Contraindications

Several specific inborn errors of metabolism can upset mitochondrial function and lead to dysfunctional glycolysis. Children with these special conditions may be strong candidates for the ketogenic diet, which will provide an important alternative energy source capable of crossing the blood-brain barrier (BBB) and sustaining cerebral energy metabolism.

The best example of such a condition is the family of glucose transporter defects (e.g., GLUT-1 deficiency) where glucose cannot penetrate the BBB [DeVivo 1991]. Two other conditions with less clear indications for special diet are pyruvate dehydrogenase complex deficiency where acetyl CoA production is blocked [Wexler 1997] and glycolysis-upsetting phosphofrucokinase deficiency [Swoboda 1997]. The benefit of the ketogenic diet is even less certain in mitochondrial cytopathies due to Complex I deficiency, which presents in infancy with hypoketogenic hypoglycemia and hepatomegaly [Sankar & Sotero de Menezes, 1999].

_______________________________________________________________________
Insight
“In difficult-to-control seizures in infants, especially myoclonic seizures, it’s important to consider an early look at glucose in the spinal fluid even if the child is afebrile. With glucose transporter defects, the ketogenic diet is the treatment of choice.”
Dr. Riviello
_______________________________________________________________________

Most inborn errors of metabolism involving mitochondrial transport of fatty acid oxidation are absolute contraindications for the ketogenic diet. These include, for example, deficiencies in carnitine (primary or secondary), carnitine palmitoyltransferase I or II, and translocase. The most common fatty acid disorder to be vigilant for is the medium-chain acyl dehydrogenase deficiency (MCAD). Other such deficiencies include those of long-chain acyl dehydrogenase, short-chain acetyl CoA dehydrogenase, long-chain 3-hydroxyacyl-CoA, and medium-chain 3-hydroxyacyl-CoA.

Clues to an inborn error of metabolism include developmental delay, hypotonia, exercise intolerance, and easy fatigability. In children with these presenting symptoms, several tests can determine if the child is suitable for the ketogenic diet. The recommended biochemical screening tests (in addition to the routine laboratory studies such as liver function tests, complete blood count, etc.) are for urine organic acids, serum amino acids, and serum lactate and pyruvate. As implied in Figure 1, findings of highly elevated dicarboxylic acids in the urine signal a problem with the normal pathway of intermediary metabolism (either mitochondrial cytopathy or a fatty acid oxidation defect) and this warrants further investigation.


Clinical Question

The ketogenic diet is highly effective in some children, but efficacy rates have varied depending on the study. Results from large prospective multicenter trials using either the classic Hopkins diet or the modified medium chain triglyceride (MCT) oil-diet are listed in Table 1. In general, more recent studies have reported lower rates of seizure control, probably due to better tracking of drop-outs (i.e., intention-to-treat analysis) and longer follow-up periods. Overall, about one-third of children come close to seizure freedom on the ketogenic diet, one third have reductions in seizure frequency, and one third do not respond. In recent prospective, multicenter studies, only 10% actually become seizure-free [Vining 1998; Freeman 1998].

Despite generally high efficacy rates in these children who are unresponsive to drugs, many questions about the ketogenic diet require further study. Determining which seizure types respond best to the diet, for example, has been a subject of debate for decades. The early controversy centered on cryptogenic versus idiopathic efficacy [Keith 1963, Livingston 1972]. And more recently, despite some reports of efficacy in both partial and generalized seizures [Schwartz 1989, Freeman 1998] many patient type- for example, those with partial seizures arising from temporal lobe pathology-still appear relatively resistant to the diet’s effects. In fact, patients with partial seizures have been excluded from most studies assessing the clinical efficacy of the ketogenic diet.

Other remaining points of controversy include the benefits of the classic diet versus the modified MCT oil diet, the potential of vagal nerve stimulation as a therapeutic alternative in these drug refractory patients, the long-term developmental effects of restricted protein and calories, and the effect of age on efficacy. On this last point, note that the diet has historically been considered more effective in infants and children because ketone extraction from periphery to brain is more efficient in the developing brain. The clinical data with the ketogenic diet in the adults is sparse, with approximately half the patients responding with greater than 50 % seizure reduction [Sirven et al., 1999].

The potential adverse effects of the ketogenic diet are well known (Table 2). In recent years, the clinical literature has focused on nephrolithiasis, growth retardation, and the potential for cardiac disease. Some of the acute toxic effects can be serious and careful monitoring is required.

Because many children with intractable epilepsy are on valproic acid, the special issue of potential exacerbation of drug side effects by the diet becomes another key issue. In particular, because carnitine deficiency is well documented with valproic acid use, supplementation is recommended in documented cases of deficiency (e.g., plasma free carnitine < 20 _mol/L after the first week of life or an esterifed to free ratio of > 0.4).

The ketogenic diet also increases the risk of nephrolithiasis, a risk that may increase in patients taking carbonic anhydrase inhibitors such as acetazolamide or, potentially, with newer broad-spectrum anticonvulsants that act (in part) at this same enzyme (e.g., topiramate and zonisamide). Preliminary experience indicates that children can be treated safely with such agents combined with the ketogenic diet.

In summary, although its mechanism of seizure control is imprecisely defined and severalpractical details of therapy (e.g., patient selection) require further study, the ketogenic dietremains a valuable option for therapy in the most drug-resistant cases of pediatric epilepsy. Before initiating a trial of the ketogenic diet, the clinician must ensure that the patient has had adequate trials of at least 2-3 anticonvulsants, and has been carefully considered for potential epilepsy surgery or vagus nerve stimulation. A thorough diagnostic metabolic work-up and a frank evaluation of the family’s potential to comply with the diet are also mandatory. In carefully selected patients without other options, the ketogenic diet can provide major benefits, both in terms of seizure control and quality of life.

Q&A

Q1 Why not just use the Atkins diet? Isn’t that ketogenic?

Dr. Rho: Many different diets are variations on the ketogenic theme but we
simply don’t have the clinical data to say which works best.

Dr. Morton: We also need to counsel patients who have initiated this diet on
their own. Just because the do-it-yourself Atkins book is available at Barnes and Noble and it’s talked about in a keto chat room doesn’t make it safe. I had one patient who became hypokalemic on the diet.


Q2 Is he vagal nerve stimulator replacing the ketogenic diet as an option for children who are not traditional surgical candidates?

Dr. Rho: To some degree, that was our experience at the University of Washington in Seattle. The overall use of the ketogenic diet at several major centers has fallen somewhat in the past few years.

Dr. Bourgeois: It depends on the seizure type. At Children’s Hospital in Boston we still prefer the ketogenic diet for those with Lennox-Gastaut and similar epilepsies. The VNS might be considered for those few children with partial seizures who are not surgical candidates, but overall our use of the ketogenic diet has stayed about the same.


Q3 What is the role of the family in success of the ketogenic diet?

Dr. Rho: The diet involves an exacting formulation and regimen and the family and social structure of the patient is critical to its success. If the family cannot help maintain complete compliance, ketosis cannot be achieved. Even small lapses such as not eating the whole meal (to maintain the proper ratio) or eating substances that contain sugar (whether a candy bar or even certain anticonvulsants) can undermine the diet. Family support is critical in maintaining a child on this diet.


Table 1. Clinical Efficacy of the Ketogenic Diet

Study Diet Seizure-Free Seizures_ Follow-up
By&Mac179; 90%

Livingston 1954 Classic 43% 3 months
(n= 304) (“controlled”)

Schwartz 1989 Classic 46% 3 months
(n=59) MCT 37% 3 months
Mod MCT 41% 3 months

Kinsman 1992 Classic 29% 38% 31 months
(n=58) (50-99%)

Swink 1996 Classic 22% 22% 12 months
(n=22)

Freeman 1998 Classic 7% 20% 12 months
(n= 150, intent to treat)

Vining 1998 Classic 10% 20% 12 months
(n=51, intent to treat)


MCT: Medium chain triglyceride



Table 2. Side Effects of the Ketogenic Diet

Possible long-term effects of high fats (cholesterol, triglycerides)
Growth retardation due to protein deficiency
Vitamin and mineral deficiencies
Constipation
Kidney stones
Elevated uric acid production
Impaired immune defenses (possibly related to neutrophils)
Metabolic acidosis
Liver failure

[insert slide 8 from Dr Rho’s talk; note that permission will be needed from author and publishers of ref Sankar 1999, Figure 1]
Reproduced with permission from ref [Sankar 1999]