Address correspondence to Elina Liukkonen, Epilepsy Unit, Helsinki University Central Hospital, PoB 280, 00029 HUS, Finland. E-mail: email@example.com
Purpose: To prospectively evaluate the efficacy of drug treatment and long-term cognitive outcome in children with encephalopathy with status epilepticus during sleep (ESESS).
Methods: Thirty-two children were diagnosed and prospectively followed up for at least 3 years at our unit between 1991 and 2007. Twenty-seven children were included in the prospective treatment study with valproate (VPA) and 17 with VPA combined with ethosuximide (ESM). Treatment response of disappearance of electrical status epilepticus during sleep (SES) was documented with overnight EEG recordings. Neuropsychological follow up for at least 5 years was available in 18 patients.
Results: Six children had atypical rolandic (AR) epilepsy, nine Landau-Kleffner syndrome (LKS), and 17 symptomatic epilepsy. Before ESESS, 20 children were cognitively normal. Prospective treatment with VPA and ESM was effective in 3 of the 17 children (18%) treated. Abolition of SES with drug treatment was observed in 16 patients. In all, 10 children (31%), 4 with AR (67%), 3 with LKS (33%), and 3 with symptomatic etiology (19%), including 9 with treatment response regained the pre-ESESS cognitive level. Unfavorable cognitive outcome was predicted by younger age at ESESS diagnosis, lower IQ at the time of the diagnosis, and no response to drug treatment when compared with those with favorable cognitive outcome. Eight of the 16 nonresponders underwent epilepsy surgery.
Discussion: Treatment response with VPA combined with ESM was observed more often than with other drug combinations. Most children with ESESS experienced permanent cognitive impairment. Cognitive outcome depends on treatment response on electroencephalography (EEG) and seizures, and on underlying etiology.
There are no prospective follow-up studies on outcome in ESESS. We prospectively investigated the treatment response and evaluated the long-term cognitive outcome of 32 children with ESESS.
In 1994 through 2000, 539 overnight video-electroencephalography (VEEG) studies were recorded at the Epilepsy Unit of Helsinki University Central Hospital. ESESS was confirmed in 27 of the 114 recordings made in children with suspected ESESS (24%). Twenty-five of these children, and two patients diagnosed earlier by ambulatory overnight EEG recording, participated in the prospective treatment and follow-up study. In addition, five patients diagnosed by ambulatory recording in 1991 and 1992 with individual treatments were included in the follow-up study.
Digital video-EEG was recorded using Telefactor equipment (Grass-Telefactor, West Warwick, RI, U.S.A.). Until 1996, an eight-channel ambulatory device (Oxford Instruments, Abingdon, Oxfordshire, U.K.), or an analog split-screen, 16-channel VEEG device (Glonner, Electronic GmbH, Munich, Germany) were also used. The international 10/20 electrode placement system was used. SES was defined as bilateral synchronous or asynchronous, or unilateral spike-and-wave discharge comprising >85% of non-REM sleep on an overnight EEG. Interruptions of 10 seconds or more were regarded as free of discharge episodes. Overnight EEG was performed (1) at diagnosis of SES, (2) at 2–4 months after change in medication for EEG response evaluation, (3) to ascertain sustained treatment response for at least 1 year with no predefined interval of the recordings and, (4) if relapse was suspected clinically.
Cognitive symptoms of ESESS were defined as deterioration from previous neuropsychological evaluation, stagnation in learning, or loss of previous skills reported by parents and/or caretakers. Expressive and/or receptive language problems and oral motor problems were documented by a speech therapist or neuropsychologist. Motor dysfunctions included ataxia, apraxia, or appearance or worsening of a unilateral deficit. Behavioral symptoms included hyperactivity, attention problems, aggressive bursts, or regression in daily life skills.
The prospective treatment study with VPA and ESM started in 1994. After the diagnosis of ESESS, high-dose VPA with trough blood level of at least 600 μm/L (86.5 μg/ml) was introduced. If no EEG response was observed, ESM was added in children with bilateral synchronous SES and/or atypical absence seizures (AAs). If no EEG response was achieved with ESM combined with VPA, or if ESM was not indicated based on the above mentioned criteria, subsequent treatment was chosen between benzodiazepines, prednisone, and other AEDs excluding sodium blocking agents, without a predefined sequence. Treatment decision was defined by the clinical situation, such as seizure type and frequency or adverse effects. Benzodiazepines were often the next choice, given in moderate doses, such as 0.3–0.5 mg/kg per day of clobazam. Weaning of a previous drug was done as quickly as possible depending on the drug combination, usually within a week or two. No more than three AEDs (the third in that case was a benzodiazepine) were prescribed concomitantly. Patients who had not obtained EEG response after trials of four or more drugs, including steroids, were considered to be drug resistant.
Neuropsychological testing was done at the diagnosis of ESESS and prospectively once a year for 3 years (one of the evaluations possibly without IQ assessment), and subsequently with 1–4 year intervals. In the six patients diagnosed before the prospective study, the evaluations were less regular. According to age and developmental level, test batteries of Bayley (1993) and Griffiths (Alin-Åkerberg & Nordberg, 1991) giving a developmental quotient DQ, and IQ tests giving verbal intelligence quotient (VIQ) and performance intelligence quotient (PIQ), WPPSI-R (Wechsler, 1995), WISC-R (Wechsler, 1984), or WISC III (Wechsler, 1999) were used. Cognitive levels were classified as: normal, IQ >85; nearly normal, IQ 70–85; mild intellectual impairment, IQ 50–69; moderate intellectual impairment, IQ 30–49; and severe intellectual impairment, IQ below 30. Neuropsychological test results before symptoms of ESESS were available for 13 children (Table 1). In the others, estimation of previous cognitive level was based on developmental milestones and/or evaluations by speech or occupational therapists.
Perinatal vascular (unil. atrophy and thal. injury, PVL)
Response to drug treatment was evaluated and defined as:
1EEG response: complete disappearance of the SES discharges in at least two overnight EEGs. For this, no spreading of the discharge within a hemisphere or between hemispheres was allowed. Focal continuous low amplitude (<200 μV) spiking was accepted as response.
2Seizure response: reduction of any epileptic seizure type by at least 90% within 3 months after starting new drug treatment.
3Cognitive response: outcome was defined as good (favorable) if pre-ESESS cognitive level was regained, moderate if IQ decrement of one level was observed, and poor in all other cases. Moderate and poor outcomes were defined as unfavorable.
4Fisher exact test was used for statistical calculations of discrete variables and two-tailed t-test for continuous variables. P-value <0.050 was considered statistically significant. For the follow-up evaluations not necessary on clinical grounds, approval of the hospital ethics committee and written parental consent were obtained.
All 32 included patients were treated at our epilepsy unit between 1991 and 2007 with clinical follow-up for at least 3 years since the ESESS diagnosis. Seizure types, previous neurologic abnormalities, cognitive level, as well as etiologic and imaging data are presented in Table 1. Of the 15 children without known etiology (idiopathic), six were classified having atypical rolandic epilepsy (AR; Dalla Bernardina et al., 2002), eight as typical LKS, and one child with naming and comprehension problems without prominent auditory agnosia (Patient 9) as variant LKS. Among the 17 children with symptomatic etiology, 12 had signs of vascular pre-/perinatal insults and 4 had cortical dysgenesis. Eleven children had congenital hemiplegia and 5 had shunted hydrocephaly. Eight children had unilateral thalamic injury on magnetic resonance imaging (MRI). Three patients had unilateral SES.
Mean age at epilepsy onset was 3.1 years (range 0.4–7.6 years), see Fig. 1. Four children with LKS (44%) had no clinical seizures. Epilepsy onset was earlier in the symptomatic group than in the idiopathic group (p = 0.024). The presenting symptoms of ESESS are shown in Table 2. Clinical symptoms emerged at mean age of 4.4 years (range 2.3–6.6 years). The symptom onset in the symptomatic group was later than in the idiopathic group (p = 0.025). ESESS was diagnosed at the mean age of 5.4 years (range 2.7–9.2 years). Mean delay from symptom onset to diagnosis of ESESS was 1.0 years (range 0.1–4.8 years). No statistically significant differences were observed between the groups.
SES disappeared in 12 patients (38%, Patients 1–3, 7, 9–12, 16, 21, 23, and 25) within 4 months of drug change. In further three responders (Patients 4, 8, and 24), the clinical condition and seizure frequency improved considerably within 3 months after drug change, but due to limited availability of overnight EEG during the summer holiday season, EEG response was first documented at 2–3 months after clinical improvement. In one child partial EEG response (centroparietal continuous discharge), seizure freedom, and improved clinical condition were observed at 7 months and full response 1.5 years after drug change (Patient 22). EEG response to drug treatment was thus observed in a total of 16 patients (50%), for treatment details see Table 3. EEG was normalized in two children (Patients 3 and 4), occasional focal spikes were observed in 11 (Patients 2, 7–12, 16, and 23–25) and focal continuous low-amplitude spiking in 3 children (Patients 1, 21, and 22). Mean age at EEG response was 8.1 years (range 6.2–12.2 years). Within the symptomatic ESESS group, two of the six EEG responders (Patients 22 and 25), and one of the 11 nonresponders (Patient 27) had unilateral SES. The difference was not statistically significant. The duration of SES, estimated by the interval from observed onset of ESESS symptoms to EEG response was mean 3.5 years (range 1.2–6.9 years) in the 16 treatment responders. Sustained treatment response was documented in all responders by overnight EEG recordings with mean follow-up of 57 months (range 15 months to 10 years). The first follow-up EEG was done 2–13 months after the observed EEG response.
Table 3. Effective medication and duration of ESES in the EEG responders (N = 16)
Sixteen children (50%) were resistant to drug treatment. Fourteen children were treated with 4 to 12 drugs (mean 6 drugs), whereas two children received only 3 drugs due to compliance problems. Five of them (Patients 5, 20, 30, 31, and 32, four of whom had symptomatic etiology) still had SES when last recorded at mean age of 10.1 years (range 7.9–14.1 years) with mean SES duration of 6.5 years (range 4.7–10.1 years). In three children with idiopathic etiology, SES resolved ultimately spontaneously without any recent drug change at 10.0, 10.6, and 13.8 years, respectively (Patients 6, 14, and 15). SES duration in these three children was 6.2–7.8 years. Eight children—one with LKS, three with polymicrogyria, and four with vascular etiology underwent epilepsy surgery—callosotomy (five: Patients 17–19 and 27, and 28), hemispherotomy (Patient 26), parietooccipital resection (Patient 29), and multiple subpial transection (Patient 13) at mean age of 7.3 years (range 4.0–9.8 years). Before surgery, all seven children with symptomatic etiology experienced marked cognitive deterioration and six had frequent daily AAs. The child with LKS had been without speech for 5.6 years. The mean duration of ESESS before surgery was 3.4 years (range 1–5.6 years).
Prospective VPA treatment was started at the diagnosis at mean age of 5.7 years (range 3.0–8.9 years) in 27 children. There were no treatment responses to VPA monotherapy with mean trough blood level of 696 micromoles/L (100 micrograms/ml). Subsequently, 21 children were treated with the combination of VPA and ESM. Four early dropouts from ESM treatment occurred (three adverse effects and one dosing problem). Seventeen children were treated with ESM (5 AR, 2 LKS, and 10 symptomatic) with mean ESM dose of 28 mg/kg per day (range 18–40 mg/kg per day). EEG response was achieved in three children (18%), see Table 3. In three more children, ESM contributed to the EEG response in other drug combinations (Patients 1, 2, and 4), see Table 3. Consequently, ESM was effective in 6 of 18 of the patients (33%).
Clobazam, clonazepam, lamotrigine, and vigabatrin were used in different combinations in >30% of the patients. Topiramate, nitrazepam, sulthiame, gabapentin, and vigabatrin were given to 10–30% and clomethiazole, diazepam, acetazolamide to <10% of the patients. Of the 21 patients who received benzodiazepines, clobazam in combination with VPA resulted in EEG response in three children. Prednisone with an initial dose of 2 mg/kg per day was used in 14 patients (44%), 2 of 6 of the atypical rolandic, 6 of 9 of the LKS, and 6 of 17 of the symptomatic group. Prednisone was introduced within 1 year of the ESESS diagnosis in nine children. Mean duration of the 2 mg/kg per day prednisone dosage was 5 weeks, and mean tapering period to dosage of 1 mg/kg every other day was 6 weeks. Mean duration of prednisone treatment was 52 weeks (range 9–174 weeks). Treatment response was achieved within 2 months in two children with LKS (Patients 9 and 11), in one of them repeatedly with adrenocorticotropic hormone (ACTH) for SES relapse (Table 3). Consequently, prednisone was effective in the treatment of SES in 2 of 14 children (14%). Treatment lag from the ESESS diagnosis was 0.2 and 1.9 years in the two responders and 0–9 years (mean 1.6 years) in the 12 nonresponders. Five of these children subsequently responded to other treatments while seven were resistant to all drug treatments.
At least 90% reduction in seizure frequency was observed in 12 children (43% of those with clinical seizures), at mean age of 7.7 years (range 4.6–11.1 years). All obtained EEG response as well. Seizure response was observed within 3 months of drug change simultaneously with the EEG response in five children. In three children, seizure response was observed with previous medication 1.1–1.5 years before the EEG response. In four patients with symptomatic etiology, seizure reduction was observed in association with taper of oxcarbazepine and introduction of another AED 0.5–1.4 years before EEG response (Patients 22–25). In three children of the idiopathic group (Patients 1, 2, and 4), atypical absences and epileptic negative myoclonus disappeared when ESM was introduced.
Twenty-eight children (88%) had three or more evaluations of cognitive level. Mean neuropsychological follow-up was 5.4 years (range 0.3–13.4 years) since ESESS diagnosis. In 18 patients (56%), follow-up was 5 years or longer. One severely intellectually impaired child with only one testing and DQ of 14 was excluded from the analyses. The cognitive course of the eight surgically treated children is reported until surgery. Postsurgical follow-up will be reported in a separate paper (Peltola et al., unpublished data). None of the surgically treated patients regained their pre-ESESS cognitive level.
On latest neuropsychological follow-up evaluation at a mean age of 11.1 years (range 4.5–18.1 years), favorable cognitive outcome was observed in 10 children (32%), see Table 4. Seven children (23%) had moderate outcome with cognitive deterioration of one level, such as normal to nearly normal, and 14 children (45%) had poor outcome. In all, 21 children (68%) had unfavorable outcome.
Table 4. Cognitive outcome at latest IQ evaluation
GroupPre-ESESS cognitive level
Pre-ESESS cognitive level based on standardised testing (N = 13) or developmental data (N = 19).
In surgically treated children (N = 8) presurgical evaluation. ii, intellectual impairment.
Atypical rolandic (N = 6)
LKS (N = 9)
Symptomatic (N = 16)
Nine of the 10 children with favorable cognitive outcome were EEG responders. The only child with no deterioration of cognitive level without EEG response was a mildly intellectually impaired boy with behavioral symptoms of ESESS treated with valproate, ethosuximide, vigabatrin, and lamotrigine (Patient 31).
Factors associated with cognitive outcome
In Fig. 2, the cognitive course of the children according to EEG response with drug treatment is illustrated by the better of the two IQ subsets, (VIQ or PIQ). There was a statistically significant difference in mean IQ at the time of the ESESS diagnosis in the drug responders compared with the nonresponders, mean IQ 92 and 65, respectively (p = 0.002). In Fig. 3 respective IQ scores are presented in relation to EEG response achieved with drug treatment.
EEG response with drug treatment was achieved by nine children with good cognitive outcome (90%), by four children with moderate outcome (57%), and by three with poor outcome (21%). There was a significantly significant increase in proportion of EEG responders within the favorable outcome group compared to the unfavorable outcome group (p = 0.006). There was no statistically significant association of age at the EEG response with cognitive outcome.
When age at epilepsy onset, symptom onset, ESESS diagnosis, and duration of SES were analyzed, statistically significant difference was observed only in the age of ESESS diagnosis, which was younger in the unfavorable group than in the favorable group (p = 0.048). IQ at the time of ESESS diagnosis was higher in those with favorable outcome (mean 90, range 51–111) compared with those with unfavorable outcome (mean 72, range 40–105, p = 0.016).
Four children with AR (67%), three with LKS (33%), and three with symptomatic ESESS (19%) had favorable cognitive outcome. All except one child with symptomatic etiology and mild intellectual impairment also showed EEG response. None of the children with dysgenetic etiology had favorable cognitive outcome. Two of the 3 children in whom the EEG response was characterized by focal continuous spiking and 7 of the 13 children with normal EEG or occasional spikes had favorable cognitive outcome. Seven of the 12 seizure responders had favorable cognitive outcome.
In our patients, three clinically different groups emerged: ESESS with AR, LKS, and symptomatic etiology. The first group comprised six children without clinically significant structural changes on MRI (Boxerman et al., 2007), initial clinical course consistent with benign focal epilepsy of childhood (Commission of ILAE, 1989; Engel, 2001), normal or nearly normal early development, and an atypical course evolving to ESESS. In eight children the clinical picture was consistent with classical LKS (Landau & Kleffner, 1957; Beaumanoir, 1992) with previously normal or slightly delayed speech development followed by typical auditory and expressive problems.
The symptomatic ESESS group of 17 patients included 4 children with dysgenetic etiology. Two of them had unilateral extensive polymicrogyria and mild congenital hemiplegia, a condition that has previously been described in association with SES, both with self-limiting benign course and cognitive deterioration (Guerrini et al., 1998; Caraballo et al., 1999; Ohtsuka et al., 2002). One child (Patient 16) had an EEG and seizure response (frequent daily AAs) with combination of VPA and ESM. The other child (Patient 19) did not respond to any medication and underwent anterior callosotomy at age 8.2 years with total disappearance of extremely frequent AAs. Both children experienced significant loss of their cognitive abilities. One of our patients had sequelae of prenatal cytomegalovirus (CMV) infection (Patient 17).
We selected VPA and ESM for first-line AEDs based on previous reports (Marescaux et al., 1990; Yasuhara et al.,1991; Paquier et al., 1992; Roulet Perez et al., 1993). However, no prospective studies were available. The use of ESM with the defined criteria was regarded reasonable because of its efficacy in atypical absence seizures. In our study, ESM combined with VPA or other drugs was effective in one-third of the children, supporting the similar observation of Inutsuka et al. (2006). Contrary to results of previous studies (Marescaux et al., 1990; Sinclair & Snyder, 2005; Kramer et al., 2009), we did not find corticosteroid treatment useful, even if given within 1 year of the diagnosis. The efficacy of benzodiazepines was disappointing as well. LEV was not effective in the four children treated by us, but it was recently reported to be effective in 40–50% of treated patients (Aeby et al., 2005; Kramer et al., 2009). However, our observation that various combinations can be effective warrants active drug trials in individual patients.
There are only few data on the temporal relationship of drug treatment with the disappearance of SES. Inutsuka et al. (2006) reported treatment response even if remission was observed 1–3 years after treatment onset, and some remissions were observed at 13 years or later. In our study, 14 of 16 EEG responses were observed at 1–6 months after drug change. However, the two symptomatic children with delayed response documentation (Patients 22 and 24) and another idiopathic drug responder (Patient 3), all with favorable outcome, were 10–12 years old at the EEG response. Moreover, two children achieved abolition of SES not associated with recent drug changes at 10 years. Both children had mild intellectual impairment at follow-up (Patients 6 and 15). These five children, comprising 15% of our study series may also be regarded as spontaneous recoveries, although disappearance of SES is generally reported to occur somewhat later, in teens (Tassinari et al., 2005). To assess the real effect of treatment, the need for frequent EEG recordings is obvious. We found that seizure response coincided with or preceded EEG response in all drug responders, suggesting that decreasing seizures could be regarded as a positive predictor of drug efficacy in ESESS. Maybe cessation of frequent seizures is as important as abolition of SES for favorable cognitive prognosis, as illustrated by the two late EEG responders in our series.
The observation that the children with subsequent EEG response to drug treatment were cognitively better at the time of ESESS diagnosis compared to those with no drug response suggests that children who experienced marked cognitive problems by the time of the diagnosis may have more severe epileptic encephalopathy to begin with. This is in line with the observation of no significant differences when age at epilepsy onset and ESESS symptom onset were compared between those with favorable and unfavorable cognitive outcome, but those with unfavorable outcome were younger at SES diagnosis.
We observed that cognitive deterioration continued in all three patient groups at least for 1 year after the diagnosis of ESESS, although treatment was started immediately in all except three LKS children with late referral of >2 years after LKS diagnosis. This finding suggests that the initially introduced drugs were not effective. This could, however, also mean that the abolition of SES and cognitive improvement was spontaneous. However, the close temporal association between later changes in drug treatment and EEG response in most drug responders is an argument against spontaneous recovery.
In the report by Veggiotti et al. (1999) on follow-up of cryptogenic and symptomatic ESESS, about one-half of the patients without previous severe developmental delay regained their previous cognitive level. In the present study, the overall cognitive outcome was similar, although a smaller proportion of children with symptomatic etiology regained their previous level. Unilaterality of SES does not seem to affect treatment response or cognitive outcome, which is in line with the observation by Kramer et al. (2009). In those two studies, SES duration of >18–24 months was associated with a drop of the cognitive level both in the cryptogenic and symptomatic group, which was not observed in our study. The definition of the EEG response in the present study was, however, different compared with that of previous studies. As illustrated in Fig. 3, the tendency of improvement of IQ observed only after EEG response defined by very strict criteria suggests that reduction of SES may not be sufficient for cognitive improvement. In our study, residual continuous focal spiking did not adversely affect long-term IQ outcome, but effects on specific neuropsychological tasks were not studied.
The atypical rolandic group had the most favorable course, with normal or nearly normal cognitive outcome in four of six children. This occurred despite delayed diagnosis (Fig. 1), long duration of EEG abnormality (Table 3), and marked cognitive deterioration during SES (Fig. 3), but required an ultimate cessation of SES (Fig. 3). In one child (Patient 3) the ultimate outcome could be explained by spontaneous recovery at age 12.2 years. All others were 8 years or younger at the time of response, and the association of the EEG response with drug changes was evident.
In the LKS group of nine children, those with normal or only slightly impaired nonverbal cognitive abilities at the time of the diagnosis responded more readily to treatment and had favorable ultimate outcome also as concerns auditory agnosia and expressive language. The three nonresponders with severe language impairment also had low nonverbal IQ at the end of follow-up (Fig. 2). In our study, the duration of SES was not correlated with cognitive outcome, in contrast to two previous studies (Giovanardi Rossi et al., 1999; Robinson et al., 2001).
Only three children with symptomatic etiology (19%) regained their previous cognitive level. In this subgroup, ESESS frequently presented with motor symptoms, such as worsening of unilateral deficit or even loss of independent walk, which may not be an easily recognizable sign of epileptic encephalopathy. Cognitive stagnation or deterioration in neurologically impaired children with epilepsy may easily be overlooked if regular developmental evaluations are not done.
Our results on the importance of EEG and seizure response on cognitive outcome in ESESS suggest that intensive treatment efforts are worthwhile, although spontaneous recovery may be possible. Abolition of SES is a prerequisite for normal cognitive development. Complete cognitive recovery seems rare, especially in patients with symptomatic etiology. Cognitive recovery was observed in the drug responders of the idiopathic group, and further deterioration was halted in the symptomatic group, whereas in the drug nonresponders it continued to slowly deteriorate (Figs. 2 and 3). The large proportion of children with symptomatic etiology with marked cognitive deterioration underscores the importance of early consideration of surgery.
Only one-third of the children with ESESS have favorable long-term cognitive outcome without permanent deterioration of cognitive level. Abolition of SES and termination of seizures are mandatory for cognitive recovery. Etiology and the underlying epileptic condition seem to have major impact, as severe permanent cognitive deterioration is rare in children with atypical rolandic epilepsy with ESESS, whereas patients with LKS and symptomatic ESESS have considerably worse long-term outcome. Children with early onset epilepsy and congenital hemiplegia, hydrocephalus, or thalamic injury may have increased risk of ESESS, and their development should be monitored carefully. In the present study, duration of SES did not have an impact on the outcome. In our experience, the combination of ESM with VPA may be a reasonable choice for first drug treatment until more prospective data on other treatments and new AEDs are available. Treatment response should be monitored by careful follow-up of clinical, EEG, and cognitive aspects. Surgical treatment options should be considered early in drug-resistant symptomatic ESESS.
This study has been supported Arvo and Lea Ylppö Foundation, Rinnekoti Foundation and Pediatric Department of the Helsinki University Central Hospital.
Disclosure: Liukkonen E has served as a paid consultant for UCB-Pharma. The remaining authors have no conflicts of interest. 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.