Calorie-restricted ketogenic diet increases thresholds to all patterns of pentylenetetrazol-induced seizures: Critical importance of electroclinical assessment


Address correspondence and reprint requests to A. Nehlig, Ph.D., INSERM 666, Faculty of Medicine, 11 Rue Humann, 67085 Strasbourg Cedex, France. E-mail:


Purpose: Thresholds to pentylenetetrazol (PTZ) seizures were usually based only on clinical symptoms. Our purpose was to use electroclinical patterns to assess the efficacy of a ketogenic and/or calorie-restricted diet on PTZ-induced seizures.

Methods: Forty 50-day-old rats were divided in four weight-matched groups and fed controlled diets: normocalorie carbohydrate (NC), hypocalorie carbohydrate (HC), normocalorie ketogenic (NK), and hypocalorie ketogenic (HK). After 21 days, blood glucose and β-hydroxybutyrate levels were determined and seizures were induced by continuous infusion of PTZ. The clinical and EEG thresholds to each seizure pattern were compared between the different groups.

Results: The electroclinical course of PTZ-induced seizures was similar in all groups. The HK group exhibited higher thresholds than the other ones for most clinical features: absence (p = 0.003), first overt myoclonia (p = 0.028), clonic seizure (p = 0.006), and for EEG features: first spike (p = 0.036), first spike-and-wave discharge (p = 0.014), subcontinuous spike-and-wave discharges (p = 0.005). NK, HC, and NC groups were not significantly different from each other. Blood glucose and β-hydroxybutyrate levels were not correlated with electroclinical seizure thresholds. After the clonic seizure, despite stopping PTZ infusion, a tonic seizure occurred in some animals, without significant difference regarding the diet.

Conclusion: This approach permitted a precise study of the electroclinical course of PTZ-induced seizures. In addition to the usually studied first overt myoclonia, we clearly demonstrated the efficiency of a calorie restricted KD in elevating thresholds to most electroclinical seizure patterns. We confirmed the lack of efficiency of the KD to reduce seizure severity once the seizure has started.

The ketogenic diet (KD) has been successfully used for almost a century as an effective anticonvulsant therapy for intractable epilepsy (Swink et al., 1997; Vining et al., 1998; Hartman and Vining, 2007). It is effective in a variety of seizure types and syndromes in the pediatric population (Kossoff et al., 2004; Hartman and Vining, 2007) but has also been successfully used in adolescents (Mady et al., 2003) and adults (Sirven et al., 1999).

The KD is a high-fat, low-protein, and low-carbohydrate diet with a usual ratio of 4:1 of fats to carbohydrates + proteins (by weight). Seizure control appears to be more effective with high lipid/carbohydrate + protein ratios both in clinical (Freeman et al., 2000) and experimental settings (Bough et al., 2000b). In rodents, maximal anti-seizure efficacy is achieved at about 1–2 weeks after initiation of the KD (Appleton and DeVivo, 1974; Bough and Eagles, 1999; Rho et al., 1999; Bough et al., 2006) and the same delay for maximal efficacy was reported in humans (Freeman et al., 2000). In humans, the outcome of treatment with the KD seems to be unrelated to age, sex, seizure type, or frequency and at least 50% of the patients treated with the KD undergo 50% or more reduction of seizure number (Schwartz et al., 1989a; Freeman et al., 1998; Vining et al., 1998). In animals, the effects of the KD are more modest. Increases of only 15–20% in seizure thresholds have been reported in mice and rats (Appleton and DeVivo, 1974; Bough et al., 1999b; Rho et al., 1999). KDs are effective in both young (Uhlemann and Neims, 1972; Otani et al., 1984; Bough et al., 1999b) and adult rodents (Appleton and DeVivo, 1974; Muller-Schwarze et al., 1999).

Recently, it was also reported that calorie restriction in a normal carbohydrate diet is able to increase seizure threshold (Bough et al., 2003). A 50% calorie restriction appears as effective as the KD with a significantly lower ketonemia (Eagles et al., 2003). These data are in line with clinical reports, both during the initiation period of the KD and beyond if seizures cluster (Freeman and Vining, 1999). In fact, classically, calorie restriction is an integral part of the KD in humans (Bough et al., 2000a, 2000b; Greene et al., 2001, 2003).

Several earlier studies measured the threshold to pentylenetetrazol (PTZ)-induced seizures in rats fed a KD (Bough et al., 2000a, 2000b; Nylen et al., 2005) or subjected to calorie restriction (Bough et al., 1999b; Eagles et al., 2003) but these studies were only based on the observation of clinical seizure symptoms. In the present work, we used either a KD and/or calorie-restricted diet in rats receiving a PTZ infusion in the tail vein but we performed threshold measurements under EEG recording. Indeed, progressively increasing doses of PTZ lead first to spike-and-wave discharges that are followed by myoclonic, clonic, and tonic–clonic seizures by increasing the dosage of the convulsant (Andre et al., 1998). Our purpose was to control whether the efficacy of the KD and/or calorie restriction was similar on the different types of seizures induced by PTZ and whether clinical signs of seizure activity can be considered as reliable and sufficient indexes to assess seizure activity.


Animals and diets

Forty male Wistar rats, weighing 154–206 g were randomly divided into four weight-matched groups (10 rats per group), with different diets based on carbohydrate, fat, and calorie intakes. Animals were put on their respective diet at postnatal day 50 (P50). The first animal group was given 0.3 kcal/g per day—considered as 100% of daily calorie requirement (Rogers, 1979)—of the usual laboratory certified UAR A04C carbohydrate rodent diet (UAR, Villemoisson-sur-Orge, France) and named normocalorie carbohydrate group (NC). The same chow with 15% calorie restriction was given to a second group, named hypocalorie carbohydrate group (HC). Two additional groups were fed a ketogenic diet (KD). We used a pharmaceutical product, Ketocal (Nutricia, North America, Gaithersburg, MD, U.S.A.) characterized by a 4/1 lipid/carbohydrate + protein calorie ratio, with 100% long chain triglycerides, balanced in vitamins and oligoelements. Using the latter diet, we defined two other groups, a normocalorie ketogenic group (NK) receiving the normal 0.3 kcal/g/day caloric intake for rats and a hypocalorie ketogenic group (HK) with 15% caloric restriction of the KD. Animals were kept in separate cages to control for their individual calorie intake. The carbohydrate, protein, and lipid composition of both diets is indicated in Table 1. After an overnight fast, appropriate amounts of each diet were provided every morning. Water was given ad libitum. Animals were kept in individual cages while they received the different types and amounts of food. Food was provided every day on the basis of the animals' weight and maintained constant for the first three to four days, even if the animals lost weight. This procedure was applied to allow specific adaptation to the KD.

Table 1.  Composition of the carbohydrate and ketogenic diet
Calorie intake (cal/kg/day)Normocalorie carbohydrate diet (NC) 300Normocalorie ketogenic diet (KC) 300
Composition (g/kg animal body weight/day)Percentage of calorie intakeComposition (g/kg animal body weight/day)Percentage of calorie intake
  1. The hypocalorie HC and HK diets are similar to the NC and NK diets, respectively, with 15% calorie restriction, i.e., 255 calories/kg/day.

  2. In this table, only the energetic components were indicated since they contribute to the calorie intake. Both diets were balanced in minerals in vitamins. Water was available ad libitum.

Lipids  3.1 9.330   90.0 

Animals were kept in uncrowded breeding facilities at 22°C under a 12 h/12 h normal light/dark cycle (lights on at 07:00 h). All animal experimentation was performed in accordance with the rules of the European Community Council Directive of November 24, 1986 (86/609/EEC), and the French Department of Agriculture (License No. 67–97). The experimental protocol was agreed by the ethical Animal Research Committee Board of University Louis Pasteur (CREMEAS, #AL/02/05/03/07) and all efforts were made to minimize animal suffering.

Measurement of thresholds to pentylenetetrazol-induced seizures

Rats were maintained on their respective diet for at least three weeks. During the second week, epidural EEG electrodes were implanted bilaterally over the frontoparietal cortex. For the surgery, the anaesthesia procedure associated i.p. injections of 37 mg/kg ketamine (Imalgene 1000, Merial, Lyon, France) and 5.5 mg/kg xylazine (Rompun, Bayer, Leverkusen, Germany). To prevent infection, strict aseptic conditions were applied during surgery. All animals were given a preventive intramuscular injection of benzylpenicilline (Extencilline, 60 000 units) after surgery. PTZ seizure threshold determination was performed between the 21st and 25th day of diet on a similar percentage of animals of each group on every testing day. All experiments were performed between 10:00 a.m. and 4:00 p.m. A freshly prepared 20 mg/ml PTZ solution was infused at a rate of 6 ml/h through the tail vein using a 24-gauge catheter. To introduce the intravenous catheter, animals were immobilized in adapted plexiglas boxes. After catheter insertion, animals put into recording cages in which they were freely moving. EEG was recorded by means of a computer-assisted device (Deltamed, France) using bipolar anteroposterior left and right derivations. EEG baseline was recorded for at least 10 min.

Thresholds to clinical seizure events were determined, considering behavioral and motor activity, and EEG paroxysmal patterns. PTZ infusion was stopped as soon as a sustained typical clonic seizure started in order to avoid the occurrence of a tonic seizure that often led to death. At the end of the experiment, the animals received a lethal dose of pentobarbital.

Measurement of blood glucose and β-hydroxybutyrate

Before animal sacrifice, two drops of blood were taken from the tail vein, and used for the measurement of blood glucose and β-hydroxybutyrate levels, respectively. We used the Accu-Chek glucometer (Roche Diagnostics, Meylan, France) for glucose measurement and the β-ketone strip test on the Medisense Optium Xceed reader for β-hydroxybutyrate. Prior to use, these two easy and fast techniques were compared with data obtained from the same samples with classical spectrophotometric assessment of plasma levels of glucose and β-hydroxybutyrate after decapitation of anesthetized animals and immediate analysis on blood drops taken from the blood freely flowing from the decapitated body. Both strip techniques were found reliable.

Statistical analysis

Threshold cumulative doses of PTZ were calculated from the latency to each manifestation, the concentration of the PTZ solution, and the infusion rate. Thresholds were expressed in milligrams of PTZ infused by kilograms of animal body weight. Data from the different groups underwent first an ANOVA test for multiple comparisons followed, if allowed, by Fisher PLSD test. In the ketogenic groups, a regression analysis was performed between seizure thresholds and the blood β-hydroxybutyrate level. In all groups, a similar regression analysis between seizure thresholds and the blood glucose level was performed.


Animals and diet

Diets were well tolerated by all animals, which exhibited normal behavior. Within the first week, a loss of weight was noticed in the two hypocalorie groups. During this period of weight loss, calorie intake was calculated on the basis of the initial weight. Within three to four days, animals went back to their initial weight. Calorie intake was then again calculated on the daily weight basis.

At the time of the experiment, there was no significant weight difference between the NC and HC groups. The rats from the ketogenic groups were significantly lighter than those from carbohydrate groups, although they all received equivalent calorie intake and all ate daily the totality of the food provided. The weight of NK rats was significantly reduced by 24% compared to the NC and HC rats while the weight of HK rats was even more reduced (39% less than the carbohydrate-fed groups). The weight of the HK rats was also significantly lower than that of the NK rats (Fig. 1).

Figure 1.

Effects of the different diets on body weight. Boxes represent mean values, the first and third quartile, minimal and maximal values. Double asterisk denote p < 0.01, statistically significant difference from NC and HC rats. ##p < 0.01, statistically significant difference between HK and NK rats.

Glucose and β-hydroxybutyrate blood levels

In the NC and HC groups, calorie restriction (15%) did not induce significant differences in glucose or β-hydroxybutyrate blood levels. Carbohydrate-diet-fed animals exhibited very low β-hydroxybutyrate blood levels (< 150 μM). Animals fed the KD exhibited significantly higher blood β-hydroxybutyrate levels and lower glucose levels compared to carbohydrate-fed animals. Moreover, the β-hydroxybutyrate level was higher in the HK group (3190 μM) than in the NK group (2146 μM; p < 0.001). Blood glucose level was also significantly lower in HK (4.5 mM) than in NK rats (5.8 mM; p < 0.001) (Fig. 2).

Figure 2.

Influence of the different diets on blood glucose and β−hydroxybutyrate levels. Boxes represent mean values, the first and third quartile, minimal and maximal values. Double asterisk denote p < 0.01, statistically significant difference from NC and HC rats. ##p < 0.01, statistically significant difference between HK and NK rats.

Electroclinical description of PTZ-induced seizures

Thresholds were finally determined in 37 of the 40 animals, because of the failure of the catheterization procedure in 1 HK and 2 NC rats.

In all groups, PTZ-induced seizures were very stereotyped considering both clinical and EEG manifestations. For each feature, mean PTZ cumulative dose is specified below only for NC animals. We clinically observed successively: (1) the loss of normal exploration behavior (13.9 mg/kg PTZ); (2) brief episodes of stop-and-stare behavior sometimes with vibrissae and facial twitching, considered as clinical absences (22.1 mg/kg); (3) subcontinuous absences and mild jerks (first jerk: 33.9 mg/kg); (4) first overt myoclonia with slight elevation of forelimbs (37.9 mg/kg); (5a) clonic seizure of the limbs (the PTZ infusion was then stopped: 39.9 mg/kg), followed by a tonic axial seizure in some cases (5b); (6) loss of postural tonus and erratic jerks during several minutes in all cases.

On the EEG we could see: (a) isolated spikes (first spike: 6.5 mg/kg PTZ); (b) bursts of bilateral spike-and-wave discharges, occurring on a normal desynchronized EEG background (23.1 mg/kg); (c) subcontinuous spike-and-wave discharges (at least 50% of each 20 s recording period: 28.3 mg/kg); (d) recruiting spike-and-wave activity of higher amplitude (exactly matching with clonic seizure, PTZ infusion stopped); (e) polyspike activity, followed by major EMG artifacts; (f) major depression of EEG activity with very high amplitude isolated slow spike-and-waves (Fig. 3).

Figure 3.

Typical recording of the evolution of changes in seizure EEG patterns induced by the continuous i.v. infusion of PTZ. Carbohydrate, ketogenic and/or calorie-restricted-fed animals exhibited the same seizure course. (A) Early stages: isolated spike (a) and first spike-and-wave discharge (SWD) (b); followed by (B) subcontinuous recurrent bursts of SWDs. The PTZ threshold dose for this feature was determined at the onset of the first SWD followed by recurrent ones lasting for at least 50% of the recording time (c). (C) On a background of subcontinuous SWDs, onset of recruiting high amplitude spike-and-wave activity (d) followed by polyspike discharges (e). (D) Very high amplitude isolated slow spike-and-waves (f) and brief bursts of polyspikes on a depressed EEG background.

No clinical motor manifestation was seen with isolated spikes (a), and no clear EEG background modification was seen during the period when rats lost their normal exploratory behavior (1). In particular no correlation was shown between the latency to these two criteria. The first spike-and-wave burst (b) sometimes preceded but most often matched the first clinical absence (2). Each clinical absence characterized by sudden cessation of behavior with staring had an EEG correlate expressed as a spike-and-wave discharge, as described in humans. Likewise, the clonic seizure (5a) occurred concurrently with the recruiting spike-and-wave discharge (d), and the tonic seizure (5b) with polyspike discharges (e). Spike-and-wave discharges recorded together with clinical absence or myoclonia exhibited the same EEG pattern.

Seizure thresholds and diets

The EEG events induced by PTZ did not differ according to the diet or calorie intake. Namely, the amplitude and frequency of baseline EEG before and between seizures were similar. Likewise, the duration, amplitude, and frequency of individual spike-and-wave discharges were identical in the four groups.

The result of multivariate ANOVA allowed the study of differences between the diet groups for the previously defined features of PTZ-induced seizures (p = 0.035). The threshold to the first spike was significantly lower in the NC group than in the HK, NK, and HC groups (p = 0.036). For other clinical and EEG thresholds, the HK group was significantly different from the three other groups and the NK, HC, and NC groups were not significantly different from each other.

Considering clinical thresholds (Fig. 4), the HK group had significantly higher thresholds than the NK, HC, and NC groups for the clinical absence (p = 0.003), the first overt myoclonia (p = 0.028), and the clonic seizure (p = 0.006). Considering EEG thresholds (Fig. 5), the HK group experienced higher thresholds than the three other groups for the first spike-and-wave discharge (p = 0.014) and subcontinuous spike-and-wave discharges (p = 0.005). The same trend was seen for other clinical and EEG thresholds, but without significant difference.

Figure 4.

Clinical threshold doses of PTZ (mg/kg) in young adult rats fed a selected diet during 21 days. Values represent means ± S.D. °°p<0.01, statistically significant difference between HK and the three other groups (NC, NK and NK animals).

Figure 5.

EEG threshold doses of PTZ (mg/kg) in young adult rats fed 21 days a selected diet.Values represent means ± S.D. Single asterisk p < 0.05, double asterisk p < 0.01, statistically significant differences from NC animals. °°p < 0.01, statistically significant difference between HK and the three other groups (NC, NK, and NK animals).

In all groups, none of the studied thresholds to PTZ-induced seizure events was correlated with blood glucose levels, as shown by the regression analysis (0.624 < r < 0.254). Likewise there was no significant relation between any threshold and blood levels of β-hydroxybutyrate in the ketogenic groups (0.586 < r < 0.021) (data not shown).

Despite the interruption of PTZ infusion when clonic seizures began, a tonic seizure occurred in some animals, with no significant difference regarding the diet (NC = 3/9; HC = 2/10; NK = 3/10; HK = 4/9).


Diets, weight, glucose and ketosis

Our results demonstrate the usability of a human KD formula, Ketocal, as an efficient, easy to provide, and palatable KD for rats. In a previous work from our group, we tested another KD consisting of a home-made mixed solution of lipid and amino acids used for human parental nutrition. This KD was given to the animals by gavage and compromised their health status that led us to test a palatable KD, well balanced and accepted by the rats (unpublished data). In opposition, the vanilla aromatized KD we used this time was very easily accepted; animals ate it spontaneously from the first day they were exposed to it, were in good health, and exhibited normal behavior. The daily preparation and the stability of this KD permitted free access to food for the rats over the whole day, thus avoiding intermittent fasting induced by sequential feeding, as previously reported (Likhodii, 2001).

Previous authors showed that the efficiency of KD in elevating thresholds to PTZ-induced seizures depends on a high lipid/carbohydrate + protein ratio (Bough et al., 2000b). Medium-chain triglycerides are known to induce higher ketosis than long-chain triglycerides without better clinical efficacy in the management of seizures (Huttenlocher et al., 1971; Schwartz et al., 1989b; Thavendiranathan et al., 2000). Although human and rat cerebral metabolism is not strictly similar, we chose to use a 4/1 ratio KD with long-chain triglycerides because it is the commonly used diet in clinical human practice (Hartman and Vining, 2007; Hee Seo et al., 2007).

Despite equivalent calorie intake, KD-fed animals exhibited a lower weight than the carbohydrate diet-fed ones. Elevated levels of ketone bodies in the brain are known to induce lower body weight even with normal calorie intake. The relationship between brain levels of β-hydroxybutyrate and reduced body weight has been explained by an increase in metabolism and heat production induced by ketosis (Bough and Eagles, 2001; Bough et al., 2000b, 2006; Davis et al., 1981; Sakaguchi et al., 1988). In the present study, as in previous ones, the body weight of rats subjected to the normocalorie KD diet was decreased by 24% compared to NC/HC rats. Furthermore, the weight of HK rats was further reduced by 13% compared to that of NK rats. These data confirm that a normocalorie and 15% hypocalorie KD with a relatively low 4/1 ratio is leading to reduced body weight. Likewise, a KD with increasing ketogenic ratios led to increased reduction in body weight gain (Bough et al., 2000b). However, in the present study, the KD did not impair body growth since a daily increase in body weight was noticed for each animal in each group after the 4th day of feeding. This is in line with data from previous studies reporting that only calorie-restricted diet by 40% or more can inhibit body growth in rats (Klurfeld et al., 1989).

Thresholds were expressed as the cumulative dose of PTZ per animal body weight, as commonly used (Bough et al., 1999b). At the time of threshold determination, it was not possible to have age, time on diet, and weight-matched animals. We chose to favor age and time on diet as they are known to influence seizure threshold. According to previous data, we determined seizure threshold after at least 21 days of exposure to the KD, to ensure steady maximal protection against seizures that is known to be delayed from the plateau level of β-hydroxybutyrate that is reached within five days (Appleton and DeVivo, 1974; Bough and Eagles, 1999; Bough et al., 2000b).

Earlier studies raised the question of calculating the PTZ dose in accordance to weight. Since the weight of KD-fed animals is usually lower than their carbohydrate-fed counterparts, the use of weight might artificially increase thresholds in animals with lower weight (Bough and Eagles, 1999; Bough et al., 1999b; Nylen et al., 2005; Auvin et al., 2006). Therefore, we repeated our statistical analysis considering latencies to each PTZ-induced clinical and EEG seizure stage with no consideration of weight. In these conditions, the latency to the clonic seizure was still significantly higher in HK compared to NK, NC, and HC animals (Fig. 6).

Figure 6.

Latency to the clonic seizure (min) not corrected for weight effects in young adult rats fed 21 days a selected diet. Values represent means ± S.D. °°p < 0.01, statistically significant difference between HK and the three other groups (NC, NK, and NK animals).

PTZ-induced seizures

Systemic administration of PTZ is commonly used in rodents to induce generalized seizures for testing antiepileptic drugs. PTZ can be administered by subcutaneous, intraperitoneal, or intravenous route (Zaczek et al., 1986; Löscher et al., 1991). Timed intravenous infusion allows a precise dose determination (Pollack and Shen, 1985) and the follow-up of the progressive evolution in clinical severity of the seizure. The severity of PTZ-induced seizures depends on the cumulative dose of PTZ injected and ranges from absences with low doses (Snead, 1992; Andre et al., 1998; McLean et al., 2004) to mild motor then tonic–clonic seizures that can evolve to status epilepticus with high doses (Mirski and Ferrendelli, 1984; el Hamdi et al., 1992; Andre et al., 1998).

In the present study, the continuous intravenous PTZ infusion and systematic independent analysis of clinical and EEG events allowed us to be very precise in the determination of thresholds, to observe the entire EEG course of seizures for each animal, and to fully describe the grading of the clinical events. To our knowledge, this study is the first one considering the occurrence of the very early first spike on the EEG preceding all clinical events. This appears of interest since the threshold to the first spike was influenced by the diet or calorie restriction. Another early event with clinical and EEG correlates that was also reliably influenced by the diet was the first isolated absence with spike-and-wave discharges followed by return to normal behavior and EEG background. These absence episodes observed in every single animal occurred several times, but could be differentiated from the secondary absence status characterized by activity arrest, with or without mild isolated jerks, and subcontinuous spike-and-wave discharges. These results emphasize the utility of concurrent use of clinical observation together with EEG recording for a more precise study of subtle PTZ-induced seizure features.

Seizure thresholds and diets

As it was already reported, the clinical correlates of PTZ-induced seizures with their characteristic progression in severity were not affected by the nature of the diet, be it carbohydrate or ketogenic in nature (Bough et al., 1999b). Although the commonly assessed thresholds are initial overt myoclonic jerks and tonic seizure without simultaneous EEG recording (Löscher and Fiedler, 1996), other clinical features such as absence-like seizures are very stereotyped and well correlated with paroxysmal EEG features.

Considering the first overt myoclonia, PTZ seizure threshold has already been shown to be more elevated in rats fed at least 20 days with a 10% calorie restricted 6.3/1 KD compared to carbohydrate-fed animals with the same calorie intake, whatever the age at diet onset, from shortly weaned pups (P22) up to adult rats (P126). The same authors showed that up to P50, the elevation of the threshold to the first overt clonus of the forelimb was significantly higher as the diet onset was earlier (Bough et al., 1999b). In the present study, we were able to reproduce this difference with an easy to provide and well-accepted human diet.

Compared to the NC diet, the HC, NK, and HK diets induced a higher threshold to the first spike, with highly significantly different values between HK and NC rats. For the EEG thresholds to the first spike-and-wave discharge and to subcontinuous spike-and-waves, as well as nearly all clinical thresholds (clinical absence, overt myoclonia, and clonic seizure) the PTZ cumulative dose was significantly very higher in the HK group than in all other groups. Threshold levels between NC, HC, and NK rats were not significantly different. The enhancement of the KD anticonvulsant effect by calorie restriction was already emphasized for the commonly assessed PTZ-induced seizure patterns, like myoclonia (Bough et al., 1999b). The present results extend this conclusion to all PTZ-induced seizure thresholds, even for subtle precocious events. Based on our data, it could be tempting to recommend calorie restricting the ketogenic diet in humans. Although in some teams, like ours for example, a 10% calorie restriction of the ketogenic diet is recommended in clinical practice, more data from different groups would be necessary to make this recommendation.

As in previous studies, no correlation could be found between seizure thresholds and blood β-hydroxybutyrate levels in the ketogenic groups (Bough et al., 1999a, 2000b). We showed this lack of correlation whatever the severity of seizure considered. The anticonvulsant activity of the KD does not only depend on the blood level of β-hydroxybutyrate and several other mechanisms of action have been hypothesized, among which the optimization of cellular metabolism and the achievement of a specific metabolic state were suggested (Bough and Rho, 2007). In the same way, lower seizure sensitivity to the carbohydrate calorie restricted fed animals has been suggested, although it does not involve ketosis (Bough et al., 1999b). However, in the present study the 15% restricted carbohydrate diet did not show any anticonvulsant activity. Calorie restriction was reported to be anticonvulsant only with the use of 35% restricted high-carbohydrate diet (Eagles et al., 2003).

The only clinical feature that was not significantly influenced by the diet was the first behavioral change and the first jerk but the same trend as for other clinical features could be seen. The lack of significance most likely originates in the larger variation in the evaluation of these features. Thus, it appears critical, when assessing PTZ seizure threshold without simultaneous EEG recording, to only consider overt myoclonia inducing forelimb lifting, excluding minor jerks.

In several previous studies concerning the use of KD in rats, two different models were used to assess protection over seizures: (1) the determination of the threshold to the first overt forelimb jerk induced by PTZ, and (2) the modification of seizure severity with the extension/flexion ratio in the maximal electroshock stimulation (MES) model. In the latter model, KD-fed rats displayed a higher threshold without reduction in seizure intensity (Hori et al., 1997). These experimental data seem to indicate a difference between the KD ability to delay or abort seizure onset and its inability to affect seizure severity once the seizure has started, sometimes even with a higher severity in KD and/or calorie-restricted rats than in rats fed the normal carbohydrate laboratory chow (Mahoney et al., 1983; Otani et al., 1984; Thavendiranathan et al., 2000; Bough et al., 2000a, 2003). In order to be able to assess the effects of the diets on both seizure threshold and severity, we compared different thresholds corresponding to different stages of seizures in a single model in order to avoid variability related to the mode of seizure induction. Even if a clear difference in thresholds could be shown especially for HK-fed animals, all rats exhibited finally the same seizure severity during the PTZ infusion. Furthermore, a similar percentage of animals exhibited tonic seizures in all groups although PTZ infusion was stopped at the onset of the clonic seizure. These data confirm the lack of reduction in seizure severity by the KD despite a protection against seizure onset in the same model.


The present approach permitted to precisely study the electroclinical course of PTZ-induced seizures. We especially differentiated isolated absences on a normal EEG background from absence status with myoclonic jerks that occurred with a higher PTZ dosage. The nature of the diet or the calorie restriction did not influence the successive seizure patterns. As previous authors, we confirmed the influence of diet/calorie restriction on the first overt myoclonia. In young adult rats, we clearly demonstrated the diet/calorie efficacy in elevating nearly all other PTZ-induced seizure thresholds, including in particular the most precocious events, i.e., the first EEG spike, and the first isolated electroclinical absence. This work confirmed the lack of efficiency of the KD to reduce seizure severity once the seizure has started. These results support the usefulness of the electroclinical assessment of PTZ-induced seizures in rats to assess the KD efficiency.


This work was supported by a grant from INSERM (U 666). The authors are grateful to SHS for providing freely Ketocal.

We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines. This protocol has received written consent from ethical Animal Research Committee Board of University Louis Pasteur in Strasbourg (CREMEAS, #AL/02/05/03/07).

Disclosure of conflicts of interest

There is no conflict of interest in the work reported in the present manuscript.