Chapter 60 The Ketogenic Diet
The incidence of epilepsy in children and adolescents ranges from 50 to 100 per 100,000 [Hauser and Hesdorffer, 1980; Hauser, 1994]. Antiepileptic drugs (AEDs) are the primary treatment modality and provide good seizure control in most children. However, more than 25 percent of children with epilepsy have either intractable seizures or suffer treatment-limiting adverse medication effects [Pellock, 1993, 1995, 1996; Heller et al., 1993; Pellock and Pippenger, 1993; Patsalos and Duncan, 1993]. Only a limited number of these children benefit from surgical therapy.
Uncontrolled seizures pose a variety of risks to children, including higher rates of mortality, accidents, and injuries, a greater incidence of cognitive and psychiatric impairment, poorer self-esteem, higher levels of anxiety and depression, and social stigmatization or isolation [Fisher et al., 1996]. Thus, effective treatment to control seizures is fundamental to improving overall outcome in childhood epilepsy. The ketogenic diet has proven to be an effective treatment for many children with epilepsy [Wheless, 1995a, b; Lefever and Aronson, 2000; Levy and Cooper, 2003]. In this chapter, the history of the development of the diet, current understanding of the biochemistry of ketone body formation and its relation to the anticonvulsant effect of the diet, considerations related to patient selection, and diet efficacy, complications, advantages, and disadvantages are reviewed.
History
For centuries, fasting has been used to treat many diseases, including seizures; it was even used in biblical times [Bible, The King James Version, 1982; Livingston, 1972].
Contemporary accounts of fasting (Table 60-1) include a child who did not respond to the conventional treatment of the day [Hendricks, 1995; Freeman et al., 1994], which was a combination of bromides and phenobarbital [Lennox and Lennox, 1960]. This child’s seizures were managed with prayer and fasting, producing dramatic improvement in seizure control during the period of starvation. However, when the period of starvation was terminated, the seizures returned [Conklin, 1922; Freeman et al., 1994; Bridge, 1949]. The child’s uncle, a pediatrician, enlisted the aid of John Howland at Johns Hopkins Hospital to understand how starvation and fasting controlled seizures and how one could maintain the beneficial effects of fasting [Lennox and Lennox, 1960]. Concurrently, Wilder at the Mayo Clinic suggested that a diet high in fat and low in carbohydrates could maintain ketosis and its accompanying acidosis longer than fasting [Freeman et al., 1994; Schwartz et al., 1989b]. Wilder was also the first to coin the term ketogenic diet. (See Wheless [2004] for a review of the ketogenic diet history.) The beneficial effects of this diet were initially recorded by investigators from Johns Hopkins University, the Mayo Clinic, and Harvard University. Use of the ketogenic diet was included in almost every comprehensive textbook on epilepsy in childhood that appeared between 1941 and 1980 [Penfield and Erickson, 1941; Lennox, 1941; Bridge, 1949; Livingston, 1954, 1963, 1972; Lennox and Lennox, 1960; Keith, 1963; Livingston et al., 1977; Withrow, 1980; Bower, 1980].
Until 1938, the ketogenic diet was one of the few available therapies for epilepsy, but it fell into disfavor when researchers turned their attention to the development of new AEDs [Livingston and Pauli, 1975]. After the introduction of sodium valproate, it was believed that this branched-chain fatty acid would treat those children previously placed on the ketogenic diet and that the diet could no longer be justified [Aicardi, 1994]. Use of the ketogenic diet decreased greatly until it received national media attention [NBC Dateline, 1994; Hendricks, 1995; Schneider and Wagner, 1995; Charlie Foundation to Help Cure Pediatric Epilepsy, 1994a; Freeman et al., 1990, 1994, 2000]. After an almost two-decade hiatus, renewed public and scientific interest has led to a better understanding and increased worldwide use of the ketogenic diet [Kleinman, 2004; Stafstrom and Rho, 2004; Kossoff et al., 2009b; Kossoff and McGrogan, 2005a].
Efficacy
As presented in Table 60-2, initial reports began to appear in the 1920s and 1930s that documented the efficacy of the ketogenic diet [Geyelin, 1921; Wilder, 1921; Peterman, 1925; Helmholz, 1927; McQuarrie and Keith, 1927; Lennox and Cobb, 1928; Barborka, 1928a, 1929; Helmholz and Keith, 1930, 1932; Pulford, 1932; Fischer, 1935; Wilder and Pollack, 1935; Helmholz and Goldstein, 1937; Wilkins, 1937]. Over the next 60–70 years, many more clinical reports appeared (Table 60-3) [Keith, 1942; Prasad et al., 1996]. About one-third to one-half of children appeared to have had an excellent response to the diet, defined by cessation or marked reduction in seizure activity. About 10 percent of children become seizure-free after initiation of the ketogenic diet, and typically stop the diet after 2 years. Of those children who become seizure-free, a minority have recurrence after the diet is stopped [Martinez et al., 2007]. Risk factors that increase the likelihood of recurrence are epileptiform abnormalities on electroencephalograms (EEGs) obtained within 12 months of diet discontinuation, a focal abnormality on magnetic resonance imaging (MRI), lower initial seizure frequency, and tuberous sclerosis complex [Martinez et al., 2007]. When the diet has been discontinued, it has usually been due to lack of efficacy. It appears that younger children are more likely to have a more favorable response than older children [Freeman et al., 1998a].
Recent studies attest to the efficacy and safety of the ketogenic diet in infants [Nordli et al., 2001; Kossoff et al., 2002; Klepper et al., 2002; Kang et al., 2005; Eun et al., 2006]. Nordli et al. retrospectively reviewed their experience with 32 infants who had been treated with the ketogenic diet, 17 of whom had infantile spasms [Nordli et al., 2001]. Most infants (71 percent) were able to maintain robust ketosis. The overall effectiveness of the diet in infants was similar to that reported for older children; 19.4 percent became seizure-free, and an additional 35.5 percent had a greater than 50 percent reduction in seizure frequency. The diet was particularly effective for children with infantile spasms. Kossoff et al. retrospectively reviewed their experience with the ketogenic diet as a treatment for infantile spasms during a 4-year period [Kossoff et al., 2002]. Twenty-three children, aged 5 months to 2 years, 39 percent of whom had symptomatic infantile spasms and 70 percent of whom had hypsarrhythmia, were started on the ketogenic diet. The children had been previously exposed to an average of 3.3 AEDs (range 0–7), and 74 percent had previously failed or relapsed after adrenocorticotropic hormone (ACTH) or steroids. The average time on the diet was 1.6 years, with 56 percent remaining on the diet at 12 months, 46 percent of whom were over 90 percent improved (three were seizure-free), and 100 percent were over 50 percent improved. Factors that predicted greater than 90 percent improvement at 12 months were age less than 1 year (p = 0.02) and exposure to ≤3 AEDs (p = 0.03). Improvement in development was related to seizure control. The same authors retrospectively reviewed their experience with the ketogenic diet as initial monotherapy for infantile spasms [Kossoff et al., 2008e]. At the end of 1 month, 8 of 13 (62 percent) infants were seizure-free (by parental report); however, the EEG normalization did not occur for 2–5 months. Time to spasm freedom was 6.5 days, suggesting that, if the ketogenic diet is used as initial treatment, a 2-week trial period is sufficient to judge efficacy for infantile spasms. In 2010, the same authors reported their 14-year experience with 104 consecutive infants prospectively started on the ketogenic diet [Hong et al., 2010]. Thirty-seven percent became spasm-free for at least a 6-month period, within a median 2.4 months of starting the ketogenic diet. A normal EEG was eventually obtained in 18 percent, and 17 percent demonstrated resolution of hypsarrhythmia. Eun et al. reviewed their experience treating 43 children with intractable infantile spasms (mean duration 13.4 months) with the ketogenic diet. Retrospective analysis revealed that the diet resulted in 53.5 percent of the patients being seizure-free; 81 percent had a greater than 50 percent reduction in seizure frequency after a mean duration of 9 months [Eun et al., 2006]. Complete seizure control was obtained within 4 weeks in 78.3 percent of the children who obtained seizure-freedom. The improvements in seizure control correlated with improvements in EEG findings. As part of the Korean multicentered experience, Kang et al. retrospectively reviewed efficacy in 49 infants treated with the ketogenic diet [Kang et al., 2005]. Only 1 of 14 children with severe myoclonic epilepsy of infancy was seizure-free at 12 months, but 56 percent had a greater than 50 percent seizure reduction. Carraballo et al. reported similar results, with 10 percent seizure-free, and 55 percent achieving a greater than 50 percent seizure reduction at 12 months [Carraballo et al., 2005]. In a small study (n = 3) of patients with a diagnosis of early infantile epileptic encephalothopy, none achieved seizure control [Kang et al., 2005].
A few early studies evaluated use of the ketogenic diet in adults (Table 60-4). Some revealed improved seizure control [Barborka, 1928b, 1930], but subsequent reports concluded that the diet is not particularly beneficial in adolescents and adults with epilepsy [Notkin, 1934; Lennox, 1941; Bridge, 1949; Livingston, 1954, 1963, 1972; Withrow, 1980; Mosek et al., 2009]. Reasons cited for this were poor dietary compliance, types of seizures seen in these age groups, and the developmental differences in the ability of the brain to use ketone bodies. As a result, past studies suggested that the optimum age of response to the diet may be in young children before the onset of adolescence. Recent studies have challenged the belief that older patients could not maintain therapeutic ketosis or comply with the rigors of the diet, and that the ketogenic diet was less efficacious in this age group. Sirven et al. [1999] reviewed the tolerability, efficacy, and adverse events in 11 adult patients prospectively begun on the classic ketogenic diet for refractory symptomatic epilepsy. All were on stable medication, and had not achieved seizure control with two or more medications; four had prior surgery. At eight months of follow-up, 1 patient was seizure-free, 5 patients had a greater than 50 percent decrease in seizure frequency, 1 patient had a less than 50 percent seizure decrease, and 4 patients discontinued the diet. Common adverse events included gastrointestinal complaints (100 percent), menstrual irregularities in all women, and increased serum cholesterol and cholesterol high-density lipoprotein (HDL) ratios. Subjective improvements in thinking and mood, without a decrease in AEDs, were reported in 7 patients, 1 without improved seizure control. Mady et al. reviewed their experience with 45 adolescents who had been on the ketogenic diet for an average duration of 1.2 years [Mady et al., 2003]. They found no evidence to support the belief that the diet was not efficacious and was too restrictive in this age group. Adolescents with multiple seizure types did best, and those with simple and complex partial seizures had the poorest response. The retention rate for motivated adolescents on the diet was not significantly different than reports in younger children.
Recent studies have critically re-examined the benefits of the diet in selected groups of patients (see Table 60-3). Schwartz et al. [1989a, 1989b] reported on the results of metabolic profiles and seizure control in 59 epileptic children receiving a normal diet or one of three forms of the ketogenic diet:
Patients fasted for 18 hours and then were placed on one of the diets and monitored for 6 weeks. All three diets produced a significant increase in total ketone body (aceto-acetate, β-hydroxybutyrate) levels that was most marked using the classical ketogenic diet [Freeman et al., 2000]. Ketone body concentrations reached a maximum in the afternoon and were frequently lower in the morning. Urinary ketones were measured and reflected changes noted in the serum. All three diets improved seizure control and none was superior to the others during the 6 weeks of observation. The first randomized trial of the classical and medium-chain triglyceride ketogenic diets was performed about 20 years later, and confirmed comparable efficacy over a longer follow-up period (1 year) [Neal et al., 2009]. Overall, these investigators found that 81 percent of patients had greater than a 45 percent reduction in seizures. Drowsiness, which occurred in 25 percent of patients during initiation of the diet, usually resolved. The medium-chain triglyceride diet was considered less palatable and associated with more side effects, including diarrhea and vomiting.
Kinsman et al reviewed 59 patients with severe intractable epilepsy; 80 percent had multiple seizure types; and 88 percent were on multiple AEDs [Kinsman et al., 1992]. Improved seizure control occurred in 67 percent of patients; 64 percent were able to reduce the amount of AEDs they were taking; 36 percent became more alert; and 23 percent had better behavior. Comorbid neurologic conditions in this group of patients with refractory epilepsy included mental retardation (84 percent), microcephaly (15 percent), and cerebral palsy (45 percent). Seizure type did not predict success with the diet. Additionally, 75 percent of the improved patients continued the diet for at least 18 months, confirming the efficacy and palatability of the diet and willingness to continue administration of the diet on the part of patients and their families. This study confirmed the earlier work by Livingston, demonstrating that 52 percent of patients had complete control and an additional 27 percent had improved control [Livingston, 1972].
The first large prospective evaluation of the ketogenic diet was conducted in 150 consecutive children aged 1–16 years (mean age 5.3 years) [Freeman et al., 1998a]. The children were followed for a minimum of 1 year, had previously been on an average of 6.24 medications, and were on a mean of 1.97 medications at the diet’s initiation. Seventy percent of children had an intelligence or developmental quotient of <69. The children averaged 410 seizures per month before the ketogenic diet. At 6 months, 71 percent remained on the diet and 32 percent had a greater than 90 percent decrease in seizures. At 1 year, 55 percent remained on the diet, 7 percent were seizure-free, and 27 percent had a greater than 90 percent decrease in seizure frequency. There was no statistically significant difference in seizure control based on age, sex, or seizure type, although none of the patients had only partial seizures. Most of those discontinuing the diet did so because it was either insufficiently effective or too restrictive.
All of the prior reports of efficacy of the ketogenic diet described open-label trials. Only one randomized, blinded crossover study has been performed. This study evaluated the efficacy of ketogenic diet for atonic-myclonic or drop seizures associated with Lennox–Gastaut syndrome [Freeman and Vining, 1999; Freeman, 2009; Freeman et al., 2009b]. Twenty children, experiencing at least 15 atonic seizures per day, were fasted for 36 hours, and then received the ketogenic diet for 1 week, in conjunction with a solution of glucose or saccharin. They then crossed over to the other study arm for an additional week. Physicians and parents were blinded to the solution composition and level of ketosis. Fasting and the ketogenic diet resulted in a decrease in the numbers of atonic seizures, which was only partially reversed with the addition of the glucose solution. The improvement in seizure control seen even in the glucose arm may explain the effectiveness of the less restrictive diets (Atkins-like and low glycemic).
A single randomized controlled trial of the ketogenic diet has been performed in children aged 2–16 years [Neal et al., 2008a]. Children with various seizure types or epilepsy syndromes experiencing daily seizures were recruited. They were randomly assigned to receive the ketogenic diet immediately (n = 73), or after a 3-month delay, with no other changes in their treatment (n = 72, control group). Children were also randomly assigned to the classic or the MCT ketogenic diet. After 3 months, the percentage of patients with baseline numbers of seizures despite treatment was significantly lower (p <0.0001) in the diet group (62 percent) than in the controls (136.9 percent). The children in the diet group had a mean of 13.3 seizures per day at baseline, and after 3 months of treatment, 1 was seizure-free; 28 (38 percent) had a greater than 50 percent reduction in seizures and 5 (7 percent) had a greater than 90 percent reduction. Efficacy was the same for symptomatic generalized or symptomatic focal epilepsies. The most common adverse events were constipation, vomiting, lack of energy, and hunger.
The same authors reported on the 1-year follow-up of those children randomized to the classic or MCT versions of the ketogenic diet [Neal et al., 2009]. Of the 125 children who started the diets, data from 64 were available at 6 months, and from 47 at 12 months. At 12 months, there was no significant difference between groups in number achieving greater than 50 percent (17.8 vs. 22.2 percent) or 90 percent (9.6 vs. 9.7 percent) seizure reduction. There was no significant difference in tolerability of the diet, except increased reports of lack of energy after 3 months, and vomiting after 12 months, both in the classical group.
In recent years, two alternative diet regimens that are less restrictive and perhaps more palatable have emerged: the modified Atkins diet [Kossoff and Dorward, 2008a] and a low glycemic index treatment [Pfeifer et al., 2008] (Figure 60-1). No prospective, comparative trials have been performed with the three treatments to evaluate relative efficacy and side-effect profiles. The Atkins diet is less restrictive than the ketogenic diet. The modified Atkins diet is similar in fat composition to a 1:1 ketogenic ratio diet, with approximately 65 percent of the calories from fat sources; it can be started as an outpatient, with guidance from a nutritionist. Three out of 6 patients studied on this diet had a greater than 50 percent seizure reduction; 2 were seizure-free [Kossoff et al., 2003]. Subsequent studies in children, adolescents, and adults reveal that 45 percent have a 50–90 percent seizure reduction, and slightly over 25 percent have a greater than 90 percent seizure reduction (Table 60-5) [Kossoff et al., 2003, 2006, 2007, 2008b, 2008d; Kang et al., 2007b; Carrette et al., 2008; Ito et al., 2008; Weber et al., 2009; Porta et al., 2009; Kumada et al., 2010].
The low glycemic index treatment [Pfeifer et al., 2005; Pfeifer et al., 2008] includes approximately 20–30 percent of calories from protein and 60–70 percent from fat. Total carbohydrates are gradually decreased to 40–60 g/day (about 10 percent of calories), using foods with a low glycemic index [Foster-Powell et al., 2002]. A retrospective review of the efficacy of the low glycemic treatment demonstrated 50 percent of patients (n = 20) with refractory epilepsy experiencing a greater than 90 percent in seizure reduction [Pfeifer et al., 2005]. Of these 20 patients, 9 were transitioned from the ketogenic diet and 11 were initiated on the low glycemic treatment. Those with a greater than 90 percent seizure reduction had an average blood glucose level of 72.6 mg/dL. In a second retrospective review, 76 children on the low glycemic index diet – 42, 50, 54, 64, and 66 percent, respectively – achieved a greater than 50 percent seizure reduction from baseline seizure frequency at follow-up intervals of 1, 3, 6, 9, and 12 months [Muzykewicz et al., 2009]. Efficacy did not differ between partial onset and generalized seizure types. Increased efficacy was correlated with lower serum glucose levels at the 1- and 12-month follow-up visits. Side effects were minimal, with only three patients reporting transient lethargy. Children were placed on the low glycemic treatment for several reasons:
Mechanism of Action
While clinical reports support the efficacy of the ketogenic diet, several theories have emerged to explain the diet’s mechanism of action [Schwartzkroin, 1999]. Four major areas have been investigated: cerebral acidosis; water balance; the direct effect of ketones or lipids; and alteration in brain energy substrates. The importance of ketone body formation was recognized early in the search for the mechanism of action of the ketogenic diet [McQuarrie and Keith, 1927]. Initially, acidosis and partial dehydration were considered likely contributors to its success [Lennox, 1928; Bridge and Lob, 1931]. Even recently, the diet’s effects on water and electrolyte metabolism were thought responsible for its efficacy [Millichap et al., 1964]. Animal studies show no change in intracellular pH of the cerebral cortex on the ketogenic diet [Al-Mudallal et al., 1996; Harik et al., 1997]. Preliminary studies using MR spectroscopy have demonstrated that children on the diet do not develop alteration of brain water and electrolyte distribution [Seymour et al., 1999] or cerebral acidosis [Marks et al., 1997; Seymour et al., 1999]. This suggests that, although ketoacidosis occurs in these children, changes in brain pH may not be a critical factor. Animal models and human evidence suggest a prominent role of ketonemia [Uhlemann and Neims, 1972; Appleton and DeVivo, 1973, 1974; DeVivo et al., 1975; Al-Mudallal et al., 1995; Stafstrom, 1999a]. One study considered elevated plasma lipids a predictor of seizure control in children [Dekaban, 1966]. A subsequent analysis of blood metabolites in nine children on the ketogenic diet revealed that a rise in total serum arachidonate correlated with improved seizure control [Fraser et al., 2003]. Despite its long history of use and proven value, the exact mechanism of action of the diet remains unknown [Nordli and DeVivo, 1997].
Oxidation of Fatty Acids: Ketogenesis
Ketonemia is essential for the ketogenic diet to work. Ketonemia occurs as a result of fatty-acid oxidation during fasting or while on the ketogenic diet [Hawkins and Biebuyck, 1980; DeVivo, 1980; Cahill, 1982; Mayes, 1996a, 1996b]. Fatty-acid biosynthesis (lipogenesis) takes place in the cytosol, whereas fatty-acid oxidation occurs in mitochondria and generates adenosine triphosphate (ATP) [Roe and Coates, 1995]. The oxidizable substrate may come from dietary sources (ketogenic diet) or from mobilization of peripheral adipose stores (starvation) (Figure 60-2). The brain does not directly use fatty acids but readily oxidizes ketone bodies [Hawkins and Biebuyck, 1979, 1980]. Increased fatty-acid oxidation leading to ketone body formation by the liver is characteristic of starvation or the ketogenic diet (Figure 60-3) [Mayes, 1996a]. Glucose, present in small concentrations, is necessary to facilitate ketone body metabolism. This is referred to as the permissive effect of glucose. Acetoacetate, a ketone body constituent, continually undergoes spontaneous decarboxylation to yield acetone that is volatilized in the lungs and gives the breath its characteristic odor.
The liver is the only organ capable of synthesizing significant quantities of ketone bodies that are released into the blood (see Figure 60-3). The liver is equipped with an active enzymatic mechanism for the production of acetoacetate. Once formed, acetoacetate cannot be significantly metabolized back to fatty acids in the liver because it lacks the enzyme 3-oxoacid-CoA transferase. This accounts for the net production of ketone bodies by the liver. Ketone bodies are then transported to and oxidized in extrahepatic tissues proportionately to their concentration in the blood. Oxidation and brain influx rates of ketone bodies are proportional to their blood concentration of approximately 12 mmol/L. At this level, the oxidative machinery and uptake mechanisms of the cell are saturated [Mayes, 1996a].
Glucose is the principal substrate for brain metabolism. Under certain conditions (e.g., fasting, ketogenic diet), the human brain uses ketone bodies for fuel [Owen et al., 1967]; the movement of ketone bodies into the brain is dependent on a monocarboxylic transport system [DeVivo, 1980]. Acetoacetate and β-hydroxybutyrate (BHB), the two constituents of ketone bodies, are metabolized primarily in the mitochondrial compartment. In the brain, the main pathway for the conversion of acetoacetate to acetoacetyl-CoA involves succinyl-CoA (Figure 60-4). Acetoacetyl-CoA is split to acetyl-CoA and oxidized in the tricarboxylic acid cycle. The enzymes that break down BHB and acetoacetate into acetyl CoA are regulated developmentally, with maximal expression early in life [Hawkins et al., 1971; Dahlquist et al., 1972; Nehlig, 1999]. This is consistent with the higher utilization of ketones by the brain in children compared to adults [Kraus et al., 1974]. Ketone bodies not only serve as a source of energy, but also contribute to the synthesis of the neurotransmitters, glutamate and gamma-aminobutyric acid (GABA), cerebral metabolic pathways normally dependent on glucose metabolism [Cremer, 1971; Sokoloff, 1973; DeLorey and Olsen, 1994; Rodwell, 1996; Cooper et al., 1996; Yudkoff et al., 2007; Lund et al., 2009]. Furthermore, fatty-acid oxidation increases brain ATP concentration [Mayes, 1996a, 1996b]. Elevation of brain ATP concentration has been verified in an animal model of the ketogenic diet [DeVivo et al., 1978; Nakazawa et al., 1983], suggesting that the ketogenic diet improves cerebral energetics. Increased alpha-ketoglutarate on the diet may also increase activity of the GABAA shunt [DeVivo, 1980; McGeer and McGeer, 1989; Peng et al., 1993]. Although whole-brain GABA levels are not changed in an animal model of chronic ketosis [Al-Mudallal et al., 1996; Harik et al., 1997], local changes in GABA concentration may occur [Rho et al., 1999b; Yudkoff et al., 2001a]. Two children studied with a new technology, two-dimensional double-quantum MR spectroscopy, showed low initial GABA levels that increased over time on the ketogenic diet [Wang et al., 2003]. The addition of acetoacetate or BHB to rat synaptosomes increased the rate of GABA formation by 35 and 43 percent, respectively [Erecinska et al., 1996]. The addition of β-hydroxybutyrate to cultured rat astrocytes suppresses GABA-transaminase in time- and dose-dependent manners [Suzuki et al., 2009]. This suppression of astrocytic GABA degradation may be another mechanism for increased GABA concentration. At the same time, the conversion of glutamate to aspartate, which has excitatory properties in the brain, is reduced [Erecinska et al., 1996; Yudkoff et al., 1997; Yudkoff et al., 2001b]. Thus, improved cerebral energetics, along with decreased excitatory (glutamatergic) and increased inhibitory (GABAergic) neurotransmission, may contribute to the efficacy of the ketogenic diet [Peyron et al., 1994; Greene et al., 2003].
Clinical Studies
Ketonemia is necessary but not sufficient for ketogenic diet-induced seizure control [Vining, 1999b; Stafstrom and Spencer, 2000]. Urine ketones are the only readily available inexpensive approach to ketone assessment, and typically the desired range is 80–160 mmol/L. Seizure control correlates significantly (p = 0.03) with serum BHB levels greater than 4 mmol/L, although urine ketones of 160 mmol/L can be found when blood BHB levels exceed 2 mmol/L [Gilbert et al., 2000]. Although serum ketones, particularly BHB, are believed to reflect the immediate state of ketosis more accurately, it is not certain if it is the presence of these ketone bodies that produces the antiseizure effect. Brown et al. [1998] measured serum BHB levels during routine clinic visits and during illness in 12 children. Eight experienced an increase in seizures and decrease in BHB during illness, and 4 had no seizure increase and no lowering of BHB levels. This suggests that BHB levels are important for the antiseizure effect. Subsequently, Freeman et al. [1998] evaluated serum BHB levels of 35 children 3 months after diet initiation. There was a significant correlation between higher levels of BHB and seizure control. The mean serum BHB level in patients with greater than 90 percent seizure control was over 6 mM.
It is also accepted that the classic ketogenic diet produces the most ketone bodies [Schwartz et al., 1989a]. Despite the fact that ketone bodies partially replace glucose for cerebral metabolism [DeVivo, 1980; Haymond et al., 1983], cerebral glucose levels are unaltered [DeVivo et al., 1978; Al-Mudallal et al., 1995]. Ketones from the circulation are transported across the blood–brain barrier by facilitated diffusion, using a monocarboxylate transport system [Moore et al., 1976; Pellerin et al., 1998]. The efficacy of the diet in childhood and the apparent slightly lower efficacy in older children and adults may be due to maturational changes in this transport system [Persson et al., 1972; Kraus et al., 1974; Dodson et al., 1976; Dahlquist and Persson, 1976; DeVivo, 1980; Williamson, 1985]. A child’s ability to extract ketones from the blood into the brain is 4–5 times greater than that seen in adults [DeVivo, 1983]. However, even adult animals placed on the ketogenic diet can upregulate brain monocarboxylate transporter levels [Leino et al., 2001]. MR spectroscopy has documented the fact that both fasting and intravenous BHB infusion in humans increase brain BHB levels [Pan et al., 2000, 2001]. However, proton MR spectroscopy (1H-MRS) performed in occipital gray matter in five children on the ketogenic diet demonstrated a single resonance identified as acetone, and no detectable BHB or acetoacetate [Seymour et al., 1999]. Subsequent experiments in rats and mice showed that acetone is an anticonvulsant and that chronic administration may enhance its action [Likhodii et al., 2002, 2003; Rho et al., 2002]. MR spectroscopy has also been used to study changes in cerebral energetics induced by the ketogenic diet. Alteration of tricarboxylic acid cycle activity by ketosis, resulting in an increased adenosine triphosphate to adenosine diphosphate (ATP:ADP) ratio or greater cerebral energy, has been hypothesized to have an anticonvulsant effect [Masino and Geiger, 2008]. This hypothesis is supported by recent experiments in patients with Lennox–Gastaut syndrome that used MR spectroscopy (31P) to document improvement in cerebral energy metabolism on the ketogenic diet [Pan et al., 1999].
Additionally, during chronic ketosis, adaptive mechanisms are active that increase the cerebral extraction of ketone bodies [Gjedde and Crone, 1975]. These mechanisms may be why ketosis develops promptly within several days after initiation of the diet, but the anticonvulsant effect may be delayed for 1–2 weeks [Appleton and Devivo, 1973, 1974]. This observation suggests that ketosis per se is insufficient to explain the anticonvulsant effect. Once ketone bodies are extracted, it is postulated that there is a secondary biochemical change or a cascade of biochemical effects that has some form of anticonvulsant effect [Prasad et al., 1996].
The importance of ketosis was demonstrated by Huttenlocher in 1976. He studied two children with a prior history of myoclonic seizures, who were seizure-free on the ketogenic diet. After a 50-minute infusion of glucose, the first patient’s serum ketones decreased 67 percent, with no change in serum pH, and a seizure occurred. The second child’s EEG changed from normal to diffuse polyspike and slow-wave complexes accompanied by myoclonic jerks after 90 minutes of intravenous glucose. Another study, involving nine children with intractable atypical absence seizures treated with the MCT diet, noted a significant decrease in the mean number of epileptiform discharges in treated patients [Ross et al., 1985]. A rise in serum ketones was the only prominent early biochemical change. DeVivo et al. also reported two children suffering from seizures resulting from a glucose transporter type I defect who were treated with the ketogenic diet and achieved complete control of seizures and improvement in neurologic development [DeVivo et al., 1991]. These observations and others suggest a pivotal role for ketone bodies in providing an alternative energy source and in achieving a still unknown role in seizure control [DeVivo, 1983; Schwartz et al., 1989b; Huttenlocher, 1976; Livingston, 1954; Aicardi, 1992; DeVivo et al., 1978; Withrow, 1980; Millichap et al., 1964; Lamers et al., 1995].
Experimental studies
At the present time, there are several animal models used to study the effects of the ketogenic diet. (For review, see Stafstrom [1999a]; Bough et al. [2002]; Stafstrom and Bough [2003]; Bough and Stafstrom [2004]; Hartman et al. [2007]; Hartman et al. [2008]; Raffo et al. [2008].) In general, these animal models demonstrate that the ketogenic diet provides protection for partial-onset seizures with secondary generalization and generalized myoclonic, tonic, and tonic-clonic seizures. The mechanism appears to be a lowering of neuronal excitability and raising of seizure threshold [Thavendiranathan et al., 2003]. However, some studies have failed to show a protective effect of the ketogenic diet in models where animals were acutely challenged with a convulsant stimulus [Bough et al., 2000a, 2002; Thavendiranathan et al., 2000; Samala et al., 2008]. Differential efficacy in animal models undoubtedly relates to regional differences in protein phosphorylation and the concentration of GABA and other neurotransmitters, and differences between provoked and spontaneous seizures [Szot et al., 2001; Ziegler et al., 2002; Martillotti et al., 2006].
A limited number of animal studies have investigated the effect of age at diet onset versus efficacy. The diet provides a greater level of seizure protection to younger animals [Bough et al., 1999c; Rho et al., 1999a], but older animals also demonstrate an elevated seizure threshold [Appleton and DeVivo, 1974; Hori et al., 1997; Bough and Eagles, 1999b]. While it would seem intuitive that increasing ketonemia would correlate with efficacy of the ketogenic diet, experimental studies have not revealed a positive correlation between them [Bough et al., 1999a, 2000b; Likhodii et al., 2000]. However, rats that developed higher levels of ketosis also showed higher thresholds for seizure induction [Bough and Eagles, 1999b]. Caloric restriction can significantly influence seizure threshold and augment the effects of the ketogenic diet on seizure control [Bough et al., 2000a, 2002, 2003; Greene et al., 2001; Eagles et al., 2003; Mantis et al., 2004; Cheng et al., 2004]. A role for enhanced cerebral energetics in efficacy of the ketogenic diet was first proposed by DeVivo et al. [1978]. Bough et al. [2006] used gene expression profiling to confirm a coordinated upregulation of transcripts for genes encoding proteins involved in energy metabolism in adolescent rat hippocampus, including those specific to mitochondria, after maintenance on a ketogenic diet [Bough et al., 2006]. These data support the hypothesis that a ketogenic diet enhances brain metabolism. Several animal models provide evidence that the ketogenic diet can prevent epileptogenesis [Muller-Schwarze et al., 1999; Stafstrom et al., 1999b; Rho et al., 1999c, 2000; Su et al., 2000; Todorova et al., 2000; Bough et al., 2003; Kossoff and Rho, 2009a; Kossoff et al., 2009c] and mediate neuroprotection [Zieglar et al., 2003; Sullivan et al., 2004; Noh et al., 2005; Gasior et al., 2006; Maalouf et al., 2007; Kim and Rho, 2008; Maalouf et al., 2009]. These studies also support the clinical observation that children can be gradually weaned off the ketogenic diet, resume a normal diet, and not experience loss of seizure control.
Selection of Candidates for the Diet
Despite the fact that several thousand patients have been treated, there currently are no consensus criteria as to which patients are candidates for the diet [Maria et al., 1997] (see Tables 60-1 to 60-4). Many seizure types appear to respond to the ketogenic diet. In general, several groups of children are potential candidates for treatment. These include the following children:
Little is known about whether the ketogenic diet should be considered in patients who are being evaluated for epilepsy surgery. Forty-seven children with intractable epilepsy and surgically remediable focal malformations of cortical development were treated with the ketogenic diet [Jung et al., 2008b]. Those seizure-free at 3 months continued on the diet, while the rest underwent surgery. The seizure-free rate at 2 years was similar between the two groups, leading the authors to suggest that, in potentially surgically remediable epilepsy, a 3-month trial of the ketogenic diet may be an option. However, in those patients in whom a malignancy or vascular malformation is detected, surgery is preferable. The ketogenic diet may be the preferred initial therapy in children with seizures and specific metabolic defects or seizures associated with specific neurologic syndromes (see Box 60-1 and Figure 60-5) [DeVivo et al., 1973, 1991; Haas et al., 1986; Wijburg et al., 1992; Shafrir and Prensky, 1995; Melvin et al., 1996; Wexler et al., 1997; Bergqvist et al., 1999b; Weber et al., 2001; Klepper et al., 2002, 2004; Liebhaber et al., 2003; Busch et al., 2005; Kang et al., 2005, 2007a; Kossoff et al., 2005b; Lee et al, 2008; Brockmann, 2009, Joshi et al., 2009; Mantis et al., 2009 ? mouse model Rett?s syndrome; Nylen et al., 2009 ? mouse model SSADH; Seo et al., 2010]. These specific metabolic disorders are reviewed in other sections of this book.
For some patients, the ketogenic diet is contraindicated (Box 60-2). Children with acute intermittent porphyria should not be treated with the ketogenic diet because carbohydrate restriction can be harmful in this condition. The ketogenic diet may also worsen some mitochondrial diseases, pyruvate carboxylase deficiency, or organic acidurias [Demeritte et al., 1996b; Nordli, 2000; Kleinman, 2004; Kossoff et al., 2009b]. When indicated, metabolic screen, including urine amino and organic acids, serum amino acids, lactate, pyruvate, and carnitine profile, should be performed before starting the ketogenic diet [Wheless, 2001]. In addition, the combination of the use of the ketogenic diet with topiramate or zonisamide appears to be associated with an increased risk of acidosis.
