Efficacy of ketogenic diet in severe refractory status epilepticus initiating fever induced refractory epileptic encephalopathy in school age children (FIRES)


Address correspondence to Rima Nabbout, Department of neuropediatrics, Centre de reference epilepsies rares, Hôpital Necker Enfants malades, 149 rue de Sèvres, 75015 Paris, France. E-mail: rimanabbout@yahoo.com and rima.nabbout@nck.aphp.fr


Purpose: Fever induced refractory epileptic encephalopathy in school age children (FIRES) is a devastating condition initiated by prolonged perisylvian refractory status epilepticus (SE) triggered by fever of unknown cause. SE may last more than 1 month, and this condition may evolve into pharmacoresistant epilepsy associated with severe cognitive impairment. We aimed to report the effect of ketogenic diet (KD) in this condition.

Methods: Over the last 12 years we collected data of nine patients with FIRES who received a 4:1 ratio of fat to combined protein and carbohydrate KD. They presented with SE refractory to conventional antiepileptic treatment.

Results: In seven patients, KD was efficacious within 2–4 days (mean 2 days) following the onset of ketonuria and 4–6 days (mean 4.8 days) following the onset of the diet. In one responder, early disruption of the diet was followed by relapse of intractable SE, and the patient died. Epilepsy affected the other six responders within a few months.

Discussion: KD may be an alternative therapy for refractory SE in FIRES and might be proposed in other types of refractory SE in childhood.

Pharmacoresistant status epilepticus (SE) is both a diagnostic and a therapeutic challenge. In some previously healthy children aged from 4–11years whose intractable SE follows fever, all attempts to identify intracranial infection may fail (Mikaeloff et al., 2006), including postmortem neuropathology (Baxter et al., 2003). SE may last more than 1 month, and this condition may either end in death (Baxter et al., 2003) or evolve into pharmacoresistant epilepsy associated with severe cognitive impairment (Mikaeloff et al., 2006). This condition was first called “Devastating epileptic encephalopathy in school age children” (DESC) (Mikaeloff et al., 2006), and later the more appropriate term of “Fever induced refractory epileptic encephalopathy in school age children” (FIRES) (Van Baalen et al., 2009, 2010).

Patients with recent worsening of seizure frequency with impact on brain function were shown to be excellent candidates for the ketogenic diet (KD) (Villeneuve et al., 2009). This series comprised two patients with FIRES who responded promptly to KD. We, therefore, applied this treatment to seven additional patients with FIRES for whom conventional intravenous SE therapy had failed.


We gathered patients who received KD for pharmacoresistant SE as an innovating means of treatment each time we were faced to or asked for advice for a new patient exhibiting the features of FIRES. Nine patients were followed over a period of 12 years in five centers. All patients previously had normal health and psychomotor development. Characteristics of the patients are given in Table 1.

Table 1.   Clinical and EEG characteristics of the series with response to AEDs and to the ketogenic diet
Patient/gender/age onset SE (months)Fever/etiologyLag from fever to SE (days)Lag from SE onset to KD (days)Ictal semiologyIctal EEGAEDs/steroids/pentothal before KDLag to ketonuria/to efficacy daysLag to new seizures months
  1. SE, Status epilepticus; P, patient; F, Female; M, Male; L, left; R, right; AED, antiepileptic drug; VPA, sodium valroate; CZP, clonazepam; PB, phenobarbital; PHT, phenytoin; VGB, vigabatrin; CBZ, carbamazepine; LTG, lamotrigine; TPM, topiramate; LVT, levetiracetam.

  2. aKetosis was not reached.

  3. bUse of ketocal.

  4. cNA: Not applicable; patients did not receive this medication.

  5. dDied 4 days after KD interruption.

1/F/62−/Small pox vaccination430Peribuccal cloniaBitemporal spike wavesVPA, CZP, PB, PTHVGB,CBZ/Partial clinical/No effectNo KU/NoaNS
2/F/98b+/Pharyngitis355Generalized, L eyelid and mouth cloniaR temporal and rolandic polyspikesCZP, PHT, PB, VPAVGB, LTG/No/Transient effect2/44
3/M/81+/Flu415Hypertonia, eyes deviationFrontal and temporal polyspikesCZP, VPA, VGB/NAc/NAc3/NoNS
4/M/77b+/Pharyngitis517Generalized, left hemiface cloniaAlternating migrating spikesCZP, PHT, PB/No/ NAc4/6d
5/M/54+/Gastroenteritis46Generalized, flush, mouth cloniaR frontal and temporal polyspikesVPA, CBZ, CZP/NAc/NAc3/54
6/F/86−/Pharyngitis150Eye staring, R arm automatisms, R facial cloniaDisorganized background continuous τδ activityVPA, TPM, PHT, PB/No/No effect3/46
7/F/62+/Mumps vaccination54Generalized, eyes deviation L/R, alternating hemibody cloniaMigrating focusCZP, PHT, PB, VGB, CBZ/No/No effect3/51
8/F/95b+/Unknown, myalgias68Alternating hemiclonia (L and R) palpebral cloniaBitemporal and R rolandic spikesCZP, PHT, PB, CBZ, LVT/Transitory/NAc2/63
9/M/72b−/Pharyngitis325Abdominal pain then clonic generalizedMigrating focus/mainly bitemporalCZP, PHT, PB, CBZ/No/ NAc2/46

The age of patients ranged from 54–98 months (mean 74 months, or 6 years). They had no personal or familial antecedents. A nonspecific febrile illness had preceded the first seizures by 1–6 days (mean 4 days). At the onset of neurologic symptoms, six patients still had fever.

Patients mainly exhibited partial seizures involving perisylvian areas, and within <24 h they progressively lost consciousness, while seizures became repetitive and nearly continuous.

Interictal electroencephalography (EEG) mainly showed diffuse delta–theta activity with spikes mostly involved the temporal lobe on both sides alternatively, extending to rolandic areas in six patients. Ictal EEG showed low amplitude fast activity together with rhythmic spikes and polyspikes over both temporal lobes, usually alternatively, and migrating over large areas of both hemispheres in seven patients.

Cerebrospinal fluid (CSF) showed 1–9 cells/mm3 and normal protein level and electrophoresis. Extensive virologic investigations in blood, CSF, and urine remained negative in all patients. In the three last patients followed in our center, search for autoimmune antibodies (VGKC, NMDAR, AMPAR, and GABABR) were negative (patients 4, 8, and 9). Plasma levels of lactate and pyruvate were normal, and plasma amino acid and urine organic acid chromatographies showed no abnormalities.

On admission for SE, magnetic resonance imaging (MRI) showed no hypersignal of the white matter, but two patients had hypersignal of the mesial temporal gray matter on bothsides, including one previously reported (patients 1 and 5).

Patients failed to respond on many antiepileptic drugs (AEDs) for SE (Table 1) including intravenous benzodiazepines (eight patients), phenytoin (seven patients), and/or sodium thiopental (Pentothal; six patients).


All patients received a 4:1 KD and removal of glucose administration, including intravenous infusion. After 24 h fasting, the first patients were fed with standard KD locally made with “home made ingredients” until a commercial preparation became available (KetoCal [Ketocal, SHS, Liverpool, United Kingdom]—Patients 2, 4, 8, and 9). Diet was administered via gastric tube. Glycemia was monitored every 3 h for the first 3 days and then every 6 h, and patients were administered glucose via gastric tube in case the blood sugar level fell to <2.5 mmols/L. Urine ketosis was monitored daily with Labstix (Bayer Diagnostics, Puteaux, France). Because patients were treated in various centers, no attempt was made to measure ketone bodies in the plasma.


Treatment of the SE

Ketonuria was reached within 2–4 days (mean 2.8 days) for eight patients. Only Patient 1 failed to reach ketonuria, and he was receiving steroid therapy when starting the diet.

Seizures stopped in seven patients, within 2–4 days (mean 2 days) following the onset of ketonuria and 4–6 days (mean 4.8 days) following the onset of the diet. KD failed to control seizures in two patients (Patients 1 and 3). Patient 1, who was on steroid therapy, did not reach ketonuria, but for Patient 3 we could identify no risk factor for failure.

Patients recovered consciousness within 24–48 h following seizure cessation, and recovered motor functions within the following weeks.

One patient (Patient 4) experienced a dramatic course. Three days after cessation of SE with KD, following a change in the medical team in the intensive care unit (ICU), the diet was abruptly interrupted “since this indication did not correspond to evidence based medicine.” SE recurred within a few hours and the patient died 10 days later, without attempt with new KD.

Long-term course

All six responders remained on the diet for 6 months to 2 years (mean 1 year). Within a few months seizures recurred (Table 1), but these consisted of isolated seizures occurring once or twice a week. Parents and physician discussed stopping the KD after seizure recurrence because of what they considered the “severity” of the diet.

Seizure semiology indicated focal onset and EEG discharges were perisylvian. Chewing movements, lateral deviation of the head to one side, modification of heart rate, or mydriasis were the most frequent manifestations of mesial temporal lobe involvement, but clonic jerks of the mouth and drooling expressed opercular extension. Seizure semiology revealed that both sides of the brain were affected, and in some patients generalization occurred.


KD is an alternative to pharmacotherapy in refractory SE in FIRES, and its efficacy has been reported in two cases of FIRES (Villeneuve et al., 2009). Following a first attempt previously reported with favorable outcome (Mikaeloff et al., 2006) and the impressive improvement for patients with partial epilepsy presenting with recent worsening and repetitive seizures including cognitive impairment (Villeneuve et al., 2009), we advised this therapeutic possibility in FIRES.

In the present series of refractory SE triggered by fever, KD was effective in seven of eight patients who reached ketonuria. In one patient (Patient 1), ketonuria could not be reached and the efficacy of KD could, therefore, not be assessed, possibly because of concomitant steroid treatment. β-Hydroxybutyrate level was not measured in the plasma, and it is, therefore, not possible to determine whether the reported level >4 mmol/L that might optimize response to KD had been reached (Gilbert et al., 2000). The observation of Patient 4 is particularly challenging, since following seizure control the patient was no longer in ketosis (negative ketone bodies on labstix), and the patient died following intractable relapse of SE.

The major contraindication to administering KD during SE is possible pyruvate carboxylase and beta oxidation deficiencies, but both contraindications of the KD are rare and usually easy to diagnose (Saudubray et al., 1999).

Although the mechanism of action of KD has long remained mysterious, some hint is presently appearing. There is growing evidence that a fundamental shift from glycolysis to intermediary metabolism induced by the KD is necessary and sufficient for clinical efficacy (Kim do & Rho, 2008). KD contributes to provide energy to the neurons, as clearly established for glucose transporter deficiency (GLUT1-deficiency) and pyruvate dehydrogenase deficiencies. KD has become the treatment of choice for these diseases (De Vivo et al., 2002; Klepper, 2008). On the other hand, in the context of glucose restriction, ketone bodies and polyunsaturated fatty acids may play various mechanistic roles, including enhancement of the mitochondrial respiratory chain (MRC), ATP production, and decrease of production of reactive oxygen species (ROS) [for review see (Kim do & Rho, 2008)]. SE triggers the mechanisms of cell death including apoptosis and necrosis. The apoptotic pathways involve the release of cytochrome C from mitochondria via activation of caspases 3 and 8 (Henshall et al., 2000, 2001). In addition, the increased generation of mitochondrial ROS mediated by either inhibition of MRC activity or decreased ATP is a critical factor of the cell death cascade that is believed to occur during epileptogenesis (Kunz et al., 2000; Kubova et al., 2001; Kudin et al., 2002). KD prevents neuronal loss in the mouse exposed to kainic acid and decreases significantly ROS in the hippocampus (Sullivan et al., 2004).

In addition to our reports on the efficacy of KD in repetitive seizures and focal SE in children (Villeneuve et al., 2009), a few case reports emphasize the possible therapeutic role of KD in nonconvulsive SE in children using the Atkins modified diet (Kumada et al., 2010) and in adults (Wusthoff et al., 2010).

Treatment of patients with SE involves in addition to the epileptologist the ICU physicians. To date, the treatment algorithms in childhood and adult SE are characterized by the lack of randomized controlled trials and remain mostly empirical. Class I evidence is available for only benzodiazepines as first-line treatment for SE (Abend & Dlugos, 2008; Trinka, 2009). Although KD seems to have some efficacy in this context, its widespread use in SE after failure of first-line intravenous AEDs could be delayed for at least three reasons: the risk of hypoglycemia on starting KD, the carbohydrate content of intravenous and oral compounds, and the unresolved question of the best timing of tube feeding in the ICU. Hypoglycemia is a concern on beginning KD (Kossoff et al., 2009), but patients with repetitive seizures and SE are well monitored in the ICU. Restricted carbohydrate administration, particularly glucose perfusion, is required and must be accepted by the ICU team. Finally, the timing of the tube feeding is also controversial, since there are no recommendations regarding this issue, although recent publications emphasized the possibility of beginning early tube feeding without major risks for patients in the ICU (López-Herce et al., 2008; Joffe et al., 2009). In this series, the use of the ready formula (KetoCal) in the last patients facilitated the tube feeding.

At this point, the presently reported observations merely provide a proof of concept that should encourage performing a prospective trial in order to validate this innovative therapeutic approach. Ketosis might be a novel strategy to safely and effectively treat refractory SE.


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. The authors declare no conflicts of interest.