1. Top of page
  2. Abstract
  3. Method
  4. Results ()
  5. Discussion
  6. References

Aim  We observed a dramatic response to the ketogenic diet in several patients with highly refractory epilepsy whose seizure frequency had recently worsened. This study aimed to identify whether this characteristic was a useful indication for the ketogenic diet.

Method  From the 70 patients who received the ketogenic diet during a 3-year period at our institution, we retrospectively selected patients with focal epilepsy. There were 22 children, 13 females and nine males, aged from 5 months to 18 years 6 months (mean 6y 9mo, SD 5y 11mo). Fifteen had symptomatic and seven had cryptogenic focal epilepsy. Seizure frequency 1 week before initiating the ketogenic diet was compared with that at 1 month and at the last visit on the diet.

Results  Eleven patients were responders (defined as reduction of seizures by more than 50%) at 1 month. Responders were higher (p=0.046) in the group with a recent worsening of seizures than in those with stable seizure frequency. Seven patients were still seizure-free at 6 months on the diet. Tolerability was excellent in 10 patients. Five patients stopped the diet because of early side effects.

Interpretation  The ketogenic diet may be a valuable therapeutic option for children with pharmacoresistant focal epilepsy, particularly those with a recent deterioration of seizure control and neurological status. Because of its rapid effect, the ketogenic diet may be a useful support to intravenous emergency drugs in such a situation.

The ketogenic diet has, for nearly a century, been repeatedly advocated in the treatment of childhood drug-resistant epilepsy.1,2 Many investigators have confirmed the benefits that this treatment can offer to young patients with epilepsy,3–6 including infants.7,8

However, the best indications for using the diet have not been clearly defined since it was shown to be helpful in treating a large range of different seizure types and epilepsy syndromes. Special interest has been given to various conditions associated with epilepsy including myoclonic astatic epilepsy,5,9 infantile spasms,7 Dravet syndrome,10 atypical absences,11 acquired epileptic aphasia,12 tuberous sclerosis,13 and Rett syndrome.14,15 One study found slightly better results in generalized as opposed to focal epilepsy although this was not significant.16 Lack of complex focal seizures was reported as a predictive factor for early, complete, and sustained response.3 However, studies that addressed the effects of the ketogenic diet according to the type of seizures or epilepsy failed to identify any significant difference.17,18 Animal studies have concentrated more on models for generalized convulsions19,20 than on focal seizures,21 although in both instances ketosis proved to be efficient. These models failed to generate any hypothesis regarding possible mechanisms of antiepileptic and anticonvulsive properties, despite the increase in neurogenesis after kainic-acid-induced seizures in mice.22 Therefore, the choice of the best candidate for a ketogenic diet remains to be determined.

We recently observed a dramatic response to the ketogenc diet in several patients with highly refractory focal epilepsy, in the setting of recent worsening, mainly in the form of nearly continuous seizures or status epilepticus. In order to ascertain whether this characteristic could be of interest as an indication for the ketogenic diet, we selected from our ketogenic-diet database a subgroup that was both exhaustive and as homogeneous as possible regarding the type of epilepsy. It consisted of children with focal epilepsy.


  1. Top of page
  2. Abstract
  3. Method
  4. Results ()
  5. Discussion
  6. References

Ketogenic diet procedure

On initiating the ketogenic diet, we administered the 4:1 ratio of lipid to non-lipid advised by Freeman and Vining, with the minimum of 1g/kg proteins, a mean 65ml/kg/day liquid, and vitamin supplementations.23 The ketogenic diet was started over a 3-day period, after 24 hours fasting, during a 5-day admission to hospital. The last day was dedicated to the validation of parental education. Tube feeding was used for patients with status epilepticus or very frequent seizures affecting consciousness. Since this study was done before the availability of a commercial ketogenic diet (KetoCal, SHS International Ltd, Fulda, Germany), all patients had the classic diet used for meals administered by tube feeding. Clinical and biological follow-up comprised monitoring of glucose blood levels and ketosis every 4 to 6 hours. Whenever glucose levels fell under 3mmol/l, intensive clinical monitoring was performed in order to detect signs of hypoglycaemia. In patients with very high seizure frequency, including those with status epilepticus, a low glucose level (<2.5mmol/l) was an indication for glucose administration.

Selection and evaluation of patients

During a 3-year period at our centre, Hôpital Saint Vincent de Paul, we administered the ketogenic diet to 70 children with drug-resistant epilepsy (treatment failure after at least three antiepileptic drugs had been given alone or in combination). In order to identify a frequent and homogeneous study group, we excluded patients with generalized epilepsy syndromes, including infantile spasms, Lennox−Gastaut syndrome, epileptic encephalopathy with continuous spike waves in slow sleep, myoclonic−astatic epilepsy, pharmacoresistant absence epilepsy, and other symptomatic generalized epilepsies. We therefore selected for this retrospective study the 22 children diagnosed with symptomatic or cryptogenic focal epilepsy on the basis of history, clinical examination, and imaging, and who had at least one focal seizure recorded on an electroencephalogram (EEG).

We distinguished patients who had experienced recent seizure worsening from those with a stable epileptic condition. Recent worsening was defined as a more than 100% increase in seizure frequency during the previous month. This resulted in very high frequency seizures or status epilepticus, requiring intravenous antiepileptic drugs. At this point, when the patients had received intravenous drugs including benzodiazepines, it was no longer possible to distinguish high-frequency seizures from status epilepticus because the level of consciousness was altered in both instances. For this reason, we decided not to make any distinction and, for the purpose of the study, to label the whole group as ‘status epilepticus’. The cognitive or motor impact of such seizure worsening was evaluated on clinical grounds.

Patients were evaluated during the week before they started the ketogenic diet, 1 month after introduction of the diet and at the last visit on the diet. The modification of seizure frequency after introduction of the ketogenic diet fell into the following categories: freedom from seizures; reduction of seizures by more than 50% (both categories being defined as ‘responders’); and seizure reduction of 50% or less (defined as ‘non-responders’). For responders, the ketogenic diet was maintained after 1 month if well tolerated.


The study group comprised nine males and 13 females aged from 5 months to 18 years 6 months (mean 6y 9mo, SD 5y 11mo). Fifteen participants presented with symptomatic epilepsy (focal cortical dysplasia in five patients, Rasmussen encephalitis in four, sequelae of grey mater encephalitis in two, and one instance each of polymicrogyria, microcephaly with diffuse cortical atrophy, hypomelanosis of Ito syndrome, and Sturge−Weber syndrome). The seven others had cryptogenic epilepsy (including one with autosomal dominant nocturnal frontal lobe epilepsy). The age at onset of epilepsy ranged from 1 day to 12 years (median 2y 7mo). The characteristics of the epilepsy are given in Table I. On the basis of clinical and EEG reports, all patients had simple focal seizures, with complex focal seizures reported in 14, and secondary generalized seizures in four patients. All patients had daily seizures except one who suffered from clusters of seizures for 2 to 3 days every other week.

Table I.   Patients’ clinical data
PatientSexAge at epilepsy onsetAge at start of ketogenic dietAetiologySeizures /d before dietFunctional impact
  1. aResponders.

 1aMale6mo18y 6moIto syndromeStatus epilepticusCognitive deterioration, hemiplegia
 2aFemale6y10y 6moRasmussen diseaseVery frequent seizuresCognitive deterioration, hemiplegia
 3aFemale1.5mo2y 6moFocal cortical dysplasiaVery frequent seizuresHemiplegia
 4aFemale8y 2mo8y 4moGrey-matter encephalitisStatus epilepticusCognitive deterioration, bilateral hemiplegia
 5Male6mo1ySturge–Weber syndromeStatus epilepticusNone
 6aFemale8y 2mo16y 8moRasmussen diseaseVery frequent seizuresHemiplegia
 7aFemale8mo3y 5moCryptogenic epilepsyStatus epilepticusHemiplegia
 8Female8d4y 3moCryptogenic epilepsyVery frequent seizuresNone
 9aFemale7y10yCryptogenic epilepsyStatus epilepticusHemiplegia
10Male12y16yRasmussen diseaseVery frequent seizuresNone
11aFemale5y7y 9moCryptogenic epilepsy3Cognitive deterioration
12aFemale4mo1yFocal cortical dysplasia10None
13Male5mo1y 1moFocal cortical dysplasia30None
14Male1d5moFocal cortical dysplasia20None
15Male1y3y 10moFocal cortical dysplasia25None
17Female3y12y 9moRasmussen disease8None
18aFemale3mo4y 8moGrey-matter encephalitis5None
19Female2mo3y 6moMicrocephaly1.3None
20Female1y 6mo3y 6moCryptogenic epilepsy10None
21Male10mo2y 3moCryptogenic epilepsy15None
22Male5y16y 3moAutosomal dominant nocturnal frontal lobe epilepsy10None

Ten patients (including seven with symptomatic epilepsy) had experienced recent seizure worsening with or without secondary generalization, based on clinical and EEG assessment. In eight of them, seizure worsening was associated with negative functional impact consisting of unilateral (six patients) or bilateral (one patient) motor and/or cognitive deterioration (three patients). Patients were receiving a mean of 2.7 antiepileptic drugs (Table II). After seizure worsening, nine of the 10 patients had received i.v. clonazepam and phenytoin, six i.v. phenobarbital, and one pentobarbital without success. This last patient had pentobarbital for 48 hours with a suppression burst pattern on EEG. Seizures recurred on withdrawing pentobarbital.

Table II.   Antiepileptic drugs used at the onset of the ketogenic diet (number of patients)
Clobazam 8
Phenobarbital 3
Topiramate 3
Valproate 3
Stiripentol 2
Lamotrigine 1
Oxcarbazepine 1
Pentobarbital 1
Levetiracetam 1

Statistical analysis

We compared the population characteristics of responders and non-responders (aetiology, age of onset of epilepsy, age when the ketogenic diet was started, time lag between onset of epilepsy and starting the diet, and recent worsening vs stable seizures frequency). In order to find out whether this recent worsening was correlated to the response to the diet or whether there were other factors underlying this positive correlation, we compared the age at onset of epilepsy, the age when the ketogenic diet was started, and the time lag from epilepsy onset to starting the diet, in the group of patients with recent worsening versus the group with stable epilepsy. The analysis was retrospective. We used the Mann−Whitney U test to compare quantitative variables (age of epilepsy onset, time lag before diet started) and Fisher’s Exact Test for qualitative variables (recent worsening vs stable epilepsy).

This study was performed in accordance with French ethical guidelines.

  1. Top of page
  2. Abstract
  3. Method
  4. Results ()
  5. Discussion
  6. References
Table III.   Results of the ketogenic diet
PatientSexSeizure reduction (%) at 1mo of dietTime to response (d)Time to seizure relapse (mo)Diet duration (mo)
  1. aResponders. NA, not applicable.


The age when the ketogenic diet was introduced ranged from 5 months to 18 years 6 months (median 6y 6mo). The time from onset of epilepsy to introduction of the diet ranged from 2 months to 18 years (median 4y 4mo) and the time from status epilepticus to ketogenic diet introduction ranged from 48 hours to 2 weeks.

Efficacy at 1 month (Table III)

Ten patients were responders at 1 month (numbers 1−4, 6, 7, 9, 11, 12, 18). Eight of them were seizure-free; two others experienced more than 90% seizure reduction (12, 18). The median time to seizure improvement was 3 days. Eight out of the 10 responders had responded during the first week on the diet.

The ketogenic diet was stopped during the first month in seven patients: in two (10, 15) because of inefficacy after 10 days and 3 weeks respectively, and in the five others (3, 8, 9, 20, 21) because of adverse events (severe vomiting and fatigue). Although they tolerated the ketogenic diet poorly, these five patients were responders at 1 week, including two who became seizure-free, with seizure control maintained at 1 month (3, 9). The three others relapsed one week after stopping the diet.

There was no correlation between the response at 1 month and the age at onset of epilepsy, the age at which the diet was started, or the duration of the epilepsy before the ketogenic diet was started (data not shown). Comparison of patients with recently increased seizure frequency (group 1) with those who exhibited no recent worsening (group 2) showed that the number of responders was significantly higher for the former (7 out of 10 vs 3 out of 12; p=0.046). Moreover, all seizure-free responders belonged to group 1, but only one to group 2. These two groups showed no difference in terms of age when the diet was introduced or of time from epilepsy onset to starting the ketogenic diet. In the group with recent deterioration of seizure control, 7 out of 10 patients (1−4, 6, 7, 9) experienced a negative impact on motor or cognitive function (isolated or associated; see Table I). All seven patients dramatically improved 1 month after starting the ketogenic diet. Response to the diet was not restricted to seizure control but was associated with cognitive improvement and with disappearance of the recent motor deficit.

Long-term efficacy (Table III)

The duration of the diet ranged from 5 days to 12 months (mean 5.3mo, median 4mo).

Ten patients were maintained on the diet for more than 1 month (1, 4, 7, 11−13, 16, 18, 19, 22; Table I). Eight of them were responders at 1 month, seven remained seizure-free, and one had only occasional seizures. Response to the diet was maintained for at least 6 months in seven of the 10 responders and for at least 1 year in two patients. In two patients with cryptogenic epilepsy, the ketogenic diet was stopped after 6 and 11 months of seizure freedom (7, 11), but the epilepsy subsequently relapsed after 2 and 9 months respectively. The reintroduction of the diet controlled seizures again for one of them.

Three non-responders were maintained on the diet because of improved quality of life (16, 19, 22). One seizure-free responder stopped the diet for personal reasons and was lost to follow-up (2). Two responders stopped the diet during the first month because of significant adverse events (3, 9) but continued to benefit in terms of seizure control for 2 and 12 months respectively.


Tolerability was relatively good. No adverse events were reported in 10 patients. Five patients had to stop the diet because of early side effects. For four patients this consisted of severe vomiting (two patients were tube-fed), which was associated with severe asthenia in three of them. Severe vomiting did not resolve on decreasing the strength of the diet (from 4:1 to 3:1 lipid to non-lipid ratio), nor on fractioning the meals. The fifth patient experienced severe anorexia.

Minor side effects were reported in seven other patients. Four had non-symptomatic hypoglycaemia during the first 3 days of the ketogenic diet requiring oral glucose supplementation. Three patients reported drowsiness that resolved spontaneously in two of them and after adjustment of the dose of antiepileptic drugs in one.


  1. Top of page
  2. Abstract
  3. Method
  4. Results ()
  5. Discussion
  6. References

This study shows that children with pharmacoresistant focal epilepsy who have experienced a recent severe worsening of seizure frequency with negative functional impact are likely to benefit from the ketogenic diet. This benefit most often occurs within the first week of the diet. The diet also permits cognitive and motor functions to return to previous levels.

We could find no evidence from the literature that a recent deterioration of seizure control in focal epilepsy is a particularly good indication for the ketogenic diet. Although one study reported a better response in generalized than focal epilepsies,16 others failed to identify any significant difference in response according to the type of epilepsy.17,18 We therefore decided to exclude this factor and to concentrate on a single type of epilepsy, selecting patients with focal epilepsy because they represent one major group of pharmacoresistant epilepsy.

This study is retrospective with a relatively small sample but the population of patients was fairly homogeneous, including patients with Rasmussen syndrome who had reached a stable condition before the recent worsening of seizures and returned to their previous condition after starting the ketogenic diet. We can therefore hypothesize that the worsening was mainly related to the seizure increase.

There are theoretical reasons to predict that recent seizure worsening is a good indication that a child may benefit from the ketogenic diet. Seizures increase cerebral glucose consumption, as shown by ictal positron emission tomography (PET).24,25 Increased seizure frequency accelerates this process, provided sufficient glucose supply is offered to the brain. The ketogenic diet reduces the consumption of glucose by the brain and provides ketone bodies as an alternative fuel.26 The delivery rate of ketone bodies to the brain is limited by the inability of the brain to generate reserves of this fuel.27 Therefore, the sudden increase of energy needs is likely to remain unmet, thus preventing seizure recurrence. Indeed, Noh et al. identified a protective effect of ketosis on the hippocampus in an animal model.21 This might explain the better response in patients with status epilepticus and nearly continuous seizure activity than in patients with stable sporadic seizures.

The ketogenic diet in our series was particularly useful for patients not responding to intravenous administration of benzodiazepines, phenytoin, or barbiturates. We recently applied the ketogenic diet to the management of status epilepticus in children suffering from devastating epileptic encephalopathy with dramatic efficacy in half the cases.28

Although our data are preliminary, they might suggest that a recent increase in seizure frequency, developing into very frequent seizures or status epilepticus resistant to standard emergency drugs, should be treated with the ketogenic diet, particularly when associated with neurological regression because of seizure worsening. Our results suggest the need for a prospective randomized trial in order to validate the place of the ketogenic diet in therapeutic guidelines.

It takes several hours for drugs to prove inefficacy. Although useless and dangerous escalation may then become tempting,29,30 doses should be reduced to prevent adverse effects. The issue is then the time lag to efficacy (ketosis) and tolerability. Intravenous fluid, if necessary in this context, could be free of glucose. A recently available liquid formulation of the ketogenic diet (KetoCal) is easily delivered through a gastric tube and offers a possible means for such emergency therapy. Median time lag to seizure improvement in our study was 3 days. Gradual initiation of the ketogenic diet without fasting results in better tolerability and allows ketosis to occur within 2 or 3 days,6 but glucose fasting generates ketosis within less than 15 hours27 and in one study the response was quicker for fasted children (within 5d vs 14d).31The question will then arise of the appropriate duration of the diet in this indication.


  1. Top of page
  2. Abstract
  3. Method
  4. Results ()
  5. Discussion
  6. References