Both fasting and consumption of low-carbohydrate diets produce changes in energy metabolism that involve a shift to lower circulating levels of glucose, the most prevalent sugar in the general circulation, and an elevation in ketone bodies (acetone, the only true ketone, and acetoacetic and β-hydroxybutyric acids), which are among the breakdown products of fat catabolism. The advantage of the KD over fasting is that it permits maintenance of an adequate energy supply so the conditions of low glucosemia and high ketonemia can be maintained indefinitely. Importantly, not all organs and tissues use ketone bodies. For example, muscle can use them when the levels first rise, but then it switches to triglycerides. The central nervous system (CNS) is the primary consumer of ketone bodies when glucose levels are low (Cahill et al., 1966). The CNS, however, cannot support its metabolism on ketone bodies alone (Sokoloff, 1973) and is dependent upon the maintenance of glucose levels near the lower end of the normal physiological range.
For clinicians, the primary goal in the use of KDs is finding dietary conditions that are most beneficial therapeutically. These conditions include antiepileptic effectiveness, minimal side effects and compliance, the last of which often depends upon palatability. Most of the diets used clinically have attempted to address all of these important issues.
Clinical efforts at refining dietary therapies have generally held to the thesis that an important key to seizure control is achievement and maintenance of an elevated level of ketone bodies, as monitored either from the urine or from the blood (Mike, 1965). Interest in the levels of ketonemia (or ketonuria) is based upon the rationale of shifting the metabolism away from carbohydrates and toward fats, and from the demonstration that seizure control increased with increased levels of ketone bodies (Huttenlocher, 1976) (Fig. 1). While urinary levels often differ greatly from blood levels, both can serve as useful indices and permit identification of noncompliance.
Two main KD types have been used. The classical long-chain KD, based on animal fat, has a ketogenic ratio (fat: carbohydrate + protein) of 4:1 or 3:1. The medium-chain triglyceride (MCT) diet, which allows more carbohydrate (yet achieves a similar degree of ketosis as the classic KD), has a ketogenic ratio of <2.5:1. Administration of either diet is usually restricted with respect to the number of calories provided and menus are designed to provide variety while maintaining prescribed levels of ketosis and seizure control (e.g., Freeman et al., 2007). Often, if there is a positive response to the diet but incomplete seizure control, better control can be obtained by a further reduction in total calories. Recent clinical focus has shifted toward the use of foods with a low glycemic index (LGI) (Pfeifer & Thiele, 2005). The LGI diet permits greater variety than do the classical and MCT KDs and reflects a change in perception with respect to the relative merits of managing levels of ketone bodies versus glucose levels in blood or CNS.
For researchers using experimental animals, the area of primary interest is in identifying the mechanisms by which manipulation of diets affords seizure protection. An underlying concern is how well the experimental animals model any of the human epilepsies. Research directed toward understanding mechanisms of ictogenesis and how these might be linked to diet has focused on the variables that change with manipulation of diet. These variables include roles for the levels of blood (and CNS) ketone bodies, the separate roles of acetone, acetoacetic acid and β-hydroxybutyric acid, levels of blood glucose, body weight and adiposity, levels of sympathetic hormones in blood (epinephrine) and CNS (norepinephrine) and the number and identity of their receptors (Szot et al., 2001), and aspects of brain energetics (catabolic pathways, energy-linked ion channels (KATP/ADP) (Ma et al., 2007) or ion transporters (Na+/K+ ATPases) and voltage-regulated ion channels of various kinds.
In contrast to the findings of Huttenlocher (1976) in humans with epilepsy, in an animal model there is little dependence of PTZ seizure threshold upon ketonemia among rats of a single age cohort (Bough et al., 2000a) (Fig. 2). Interestingly, rats fed KDs show elevated thresholds as demonstrated by slow infusion of PTZ, but the same rats show more severe seizures when subjected to maximum electroshock (MES) (Bough et al., 2000b). Normal (nonepileptic) mice fed a KD appear not to be protected from PTZ-induced seizures but they are protected from maximal seizures (MES) (Uhlemann & Neims, 1972). Given these differences in protection afforded nonepileptic strains of mice and rats consuming KDs, it is reasonable to be concerned about the transferability of such studies to humans.
In rodents, KDs are marginally effective in elevating seizure threshold when fed ad libitum, but their effectiveness is increased considerably when they are modestly restricted in calories. Among the first experimental demonstrations that calorie restriction, alone, provided seizure protection was that of Seyfried's group (Greene et al., 2001). They found that seizures induced by handling in EL mice were reduced when standard rodent chow diets were provided in restricted quantities and that seizure protection appeared better correlated with blood glucose levels than with levels of β-hydroxybutyric acid, the blood ketone body most commonly measured. An important feature of this demonstration was that the mice were epileptic, suggesting that they are closer to humans with epilepsy than are normal animals provoked to have seizures by either chemical or electrical insults. Models of the latter type, while less directly related to the clinical condition, have added support for the role of calorie restriction and the importance of lowered levels of blood glucose rather than the elevation of ketone bodies (Eagles et al., 2003). That work showed that normal Sprague–Dawley rats fed highly restricted rodent chow diets enjoyed seizure protection indistinguishable from that of rats fed a more modestly restricted ketogenic diet, despite having blood levels of β-hydroxybutyric acid that were indistinguishable from those of rats fed rodent chow ad libitum (Fig. 3).
The first evidence showing neuronal effects in vivo of consumption of varied diets was that of Bough et al. (2003), who used field potential studies to show a requirement for increased stimulus intensity to elicit postsynaptic responses when rats were fed either ketogenic or calorie-restricted rodent chow diets. A current focus of experimental work is on brain bioenergetics. Animals use carbohydrates exclusively for brain metabolism when they are fed ordinary diets ad libitum, but a portion of brain metabolism is shifted to fat breakdown products when carbohydrates (or total calories) are restricted. The question of interest is whether the CNS is afforded more energy (i.e., ATP) when an animal consumes normal diets, a calorie-restricted diet, or a KD. There is evidence to suggest that, perhaps paradoxically, brain bioenergetics are improved when animals are fed a calorie-restricted KD (Sato et al., 1995).