The Influence of Sulthiame on EEG in Children with Benign Childhood Epilepsy with Centrotemporal Spikes (BECTS)


  • Further members of the Sulthiame Study Group: (A) Investigators: D. Škarpa, N Barišić, and M. Jurin, University Children's Hospital, Zagreb; V. Sander, Tallinn Children's Hospital; C. Benninger and K, Schrader, University Children's Hospital, Heidelberg; H. Siemes and K.H. Spohr, Kinderklinik Pulsstraβe, Berlin; R. Knapp, Städt Klinikum, Braunschweig; P. Temin and M. Nikanorova, Institute of Paediatrics and Paediatric Surgery, Moscow; J.P. Ernst and P. Burkard, Epilepsiezentrum, Kehl-Kork; B. Kruse and B. Wilken, University Children's Hospital, Göttingen; M. Wolff, University Children's Hospital, Tübingen; M. Laub and G. Kluger, Behandlungszentrum, Vogtareuth; D. Hobusch and K. Popp, University Children's Hospital, Rostock; R. Korinthenberg, University Children's Hospital, Freiburg; M. Brozmanová and P. Sýkora, University Children's Hospital, Bratislava; B. Püst and D. Will, Kinderkrankenhaus Wilhelmstift, Hamburg; U. Stephani and B. Neubauer, University Children's Hospital, Kiel; O. Hasselmann, A. Conradi, and M. Meusers, Gemeinschaftskrankenhaus, Herdecke; M. Haupt, Kinderklinik Klinikum, Erfurt; M. Büsse and H.M/ Straβburg, University Children's Hospital, Würzburg; M. D'Hooghe, Algemeen Ziekenhuis St Jan, Brugge; G. Groβ-Selbeck and M. Kuhlenkampf, Krankenhaus, Düsseldorf-Gerresheim; J. Saracz, Heim Pál Gyermek Kórház, Budapest; G. Wohlrab, Kinderspital, Zürich; M. Brünger, Pfalzinstitut, Klingenmünster; M. Schächtele, Städt. Kinderklinik, Karlsruhe; H. Todt, University Children's Hospital, Dresden; K. Kellermann, K. Rheingans, Kinderkrankenhaus der Stadt, Köln; (B) Serum Level Observer, A. Retzow, Reinfeld; (C) Biometrician, A. Völp, Psy Consult Scientific Services, Frankfurt; (D) Steering Committee, H.M. Weinmann, Starnberg; R. Kruse, Kehl-Kork; C. Lipinski, Rehabilitionsklinik, Neckargemünd.

Address correspondence and reprint requests to Dr. T. Bast at Univ.-Kinderklinik, Abt. für Pädiatrische Neurologie, Im Neuenheimer Feld 150, D-69120 Heidelberg, Germany. E-mail:


Summary:  Purpose: To evaluate the effects of sulthiame (Ospolot; STM) monotherapy compared with placebo on the EEG in children with benign childhood epilepsy with centrotemporal spikes (BECTS).

Methods: Sixty-six patients (aged 3–11 years) entered a 6-month double-blind trial and were randomized to either STM (n = 31) or placebo (n = 35). Clinical data and general results have been reported elsewhere (1). One-hundred seventy-nine sleep EEGs were recorded at screening and after 4 weeks, 3 months, and 6 months. EEGs were analyzed by a blind reviewer using a standard protocol for each EEG. This standard protocol collected data on general changes, specific epileptiform, and nonspecific focal and generalized changes. A classification system was defined depending on rating of pathologic EEG changes. Because of the higher number of treatment-failure events (i.e., seizures) in the placebo group, there was an increasing imbalance between the two groups regarding the number of recorded sleep EEGs over time (STM, 104; placebo, 74). A Wilcoxon–Mann–Whitney U test was used to describe differences in the grade of pathology during individual follow-up between the two groups.

Results: The sleep-EEG was found to be normalized in 21 patients treated with STM (12/21 transient) and in five patients treated with placebo (4/5 transient). In the STM group, the EEG showed a marked improvement during intraindividual course when comparing the classification of follow-up EEGs at each time point with the screening EEG. Comparable improvements were not observed in the placebo group (exact two-tailed p value at 4 weeks, p < 0.0001; at 3 months, p = 0.0010; and at 6 months, p < 0.0001).

Conclusions: STM had marked effects on the EEG in BECTS, which led to normalization in the majority of the patients. Most of those whose EEGs were not normalized showed improvement in the grade of EEG pathology. Normalization persisted in >50% of patients during the investigation. Spontaneous normalization in the placebo group reflects the wide spectrum of individual courses, which must be considered when analyzing drug effects on EEG in BECTS.

Recently we reported a 6-month randomized, double-blind, multicenter trial that compared sulthiame (STM; Ospolot) with placebo as primary agent in the treatment of benign childhood epilepsy with centrotemporal spikes (BECTS) (1). The trial demonstrated that STM is safe and effective in the prevention of seizures in children with BECTS.

Besides the clinical effectiveness of STM, there have been reports concerning positive effects of the drug on the EEG in BECTS (2). Some reports point to the role of interictal EEG changes for developing neuropsychological deficiencies in BECTS (3,4). As the disease is associated with a wide variability of intraindividual EEG courses, including spontaneous normalization (5), a blinded, placebo-controlled investigation is necessary for analyzing drug effects on the EEG in BECTS. Besides the global judgment of a complete normalization of the EEG, more detailed effects of drugs should be analyzed. To address this question, it was necessary to develop a tool for intraindividual comparison of EEGs suitable for a multicenter setting. Thus a classification system was constructed for this study.


Children with a diagnosis of BECTS [according to International League Against Epilepsy (ILAE) classification](6) and two or more seizures during the past 6 months were admitted to the study. Patients were required to be aged between 3 and 10 years (weight between 10 and 50 kg). Patients with severe organic diseases; relevant renal, thyroid, or hepatic dysfunction; acute porphyria; a history of mental illness; relevant hypersensitivity; or somatic signs of puberty were excluded, as was any patient with an antiepileptic drug (AED) pretreatment after the sixth month of life (exception: acute intervention of <1 week). Written informed consent was obtained from parents or guardians of each patient before admission to the study. The 66 patients (40 boys, 26 girls) between ages 3.1 and 10.7 years (mean, 8.3 years) were admitted to the study in 26 centers in Europe. The 31 study participants were randomized to STM treatment (5 mg/kg/day), and 35, to placebo. The primary efficacy variable was the rate of treatment failure events (TFEs) per group. Besides safety-related criteria, any first following seizure was defined as a TFE. Of the 66 children with two or more seizures during the past 6 months, only 10 of 35 patients of the placebo group completed the study without TFE, compared with 25 of the 31 patients treated with STM. The trial was stopped after a preplanned, adaptive interim analysis because STM was found to be superior in preventing TFEs. In the STM group, four of 31 patients were removed from treatment because of seizures (TFEs), and two were withdrawn for administrative reasons unrelated to treatment efficacy or tolerability. In the placebo group, 21 of 35 patients had a seizure, and removal from treatment was initiated by the parents in two patients (both end points were evaluated as TFEs); two additional patients were withdrawn for administrative reasons (Fig. 1). Further details are described elsewhere (1).

Figure 1.

Fraction of patients without treatment failure events (Kaplan-Meier plot) [0].

According to the study protocol, EEGs (10–20 electrode positions), both awake and during sleep, were recorded at screening and after 4 weeks, 3 months, and 6 months (total number, 363). In this report we focus on only the sleep EEG (n = 179) to rate at the same vigilance stage. Twenty-six EEGs were recorded during sleep stage 1, and 153 EEGs, during deeper stages of sleep (at least stage 2) (Table 1). All sleep EEGs were included for further analyses. Because of the higher number of TFEs in the placebo group, there was an increasing imbalance in the number of recorded EEGs in both groups (STM, 104; placebo, 75). The distribution of sleep EEGs at screening, 4 weeks, 3 months, and 6 months is shown in Table 1.

Table 1.  Number and classification of sleep EEGs
 Number of sleep EEG
 Screening4 wk3 mo6 mo
 St. 1St. 2TotalSt. 1St. 2TotalSt. 1St. 2TotalSt. 1St. 2Total
  1. St. 1, Sleep stage 1; St. 2, Sleep stage 2 or deeper; Grade, see Table 2; STM, sulthiame.

 No sleep EEG  2  2  3  4
 Total number of sleep EEG32629326296182431922
 Normal EEG01131720110112810
 Unspecific pathologic grade 0011011123112
 Specific focal pathologic grade 121113077347066
 Specific focal pathologic grade 2033011000033
 Specific focal pathologic grade 311011000123011
 No sleep EEG  3  2  2  2
 Total number of sleep EEG626323202311112178
 Normal EEG000044044011
 Unspecific pathologic grade 0000000011011
 Specific focal pathologic grade 1314172911123134
 Specific focal pathologic grade 2257033033011
 Specific focal pathologic grade 3178145011011

At the end of the clinical trial, all EEGs were analyzed by one author (T.B.) visually within a 5-week period. The results were documented by using a standard protocol for each EEG. The EEG rater was blind with regard to the identity of the investigational products. EEGs from the same patient were analyzed independently (i.e., no comparison was made to a previous or subsequent EEG). According to the standard protocol, the following data were collected:

  • • General: vigilance and background activity (awake EEG), sleep stages, generalized epileptiform activity and technical data.
  • • Unspecific foci: Type of focus (delta or theta slowing, pathologic alpha or beta, alpha reduction, asymmetry), side and localization. In case of several unspecific foci, each focus was described separately.
  • • Specific foci: Type (typical rolandic morphology, others like polyspikes), side, localization, appearance (single, groups, chains), time fraction of epileptic activity, percentage of sharp waves showing ipsilateral and contralateral propagation.

All of these items were determined for each single focus in case of multifocal activity.

  • • To determine the extent of EEG pathology, a rating system was developed by the authors, which included a count of the amount of spiking, propagation, and generalization of any specific focus. Each focus was rated separately. If there were three specific foci, and additional 3 rating points were given (Table 2). All rating points were summed for all foci (Table 2). EEGs were classified as (a) normal; (b) nonspecifically pathologic (grade 0): no specific focus but any kind of other pathology like nonspecific foci, abnormal background activity, or generalized spike–wave complexes; or (c) specifically pathologic (grades 1–3): specifically pathologic EEG with at least one specific focus (grade 1, <5 rating points; grade 2, 5–7 rating points; grade 3, >7 rating points) (Table 2).
  • • The results presented later apply to the trial's intention-to-treat data set, which included all randomized patients. Comparability of the study groups was evaluated by testing the numbers of specific foci as well as the EEG grades at baseline for between-group differences. The changes versus baseline in these three measures were compared between the treatment groups to assess the treatment effect on the sleep EEG. All treatment group comparisons were performed with exact Wilcoxon–Mann–Whitney U tests, with SPSS Version 9 statistical software on a PC system running under MS Windows 95. All p values reported are two-sided and are intended to be descriptive.
Table 2.  Rating of specific focus pathology
of spiking
per time
  1. In case of three specific foci, 3 additional rating points were given.

No focus0  No0
<25%1  <25%1
>70%8  >75%8


At screening, no statistical difference could be found between the two groups regarding recording time, sleep stages, and the classification of pathology (t test: p = 0.82). Screening sleep EEG was normal in one patient (STM group); however, this patient was included because a previous EEG had shown typical centrotemporal sharp waves, and he remained in the study accordingly. In another patient, the sleep EEG at screening was only nonspecifically pathologic, but the awake EEG, recorded on a different day, demonstrated a specific centrotemporal focus. In five patients, no sleep EEG at screening was available, whereas the awake EEG showed specific pathology.

Eighty foci were identified in 59 patients at screening; 41 patients had one focus, 15 patients had two, and three patients had three independent foci. Sharp-wave foci were detected in the following regions: centrotemporal (n = 47), temporal (n = 17), temporoparietal (n = 6), centroparietal (n = 4), precentral (n = 4), and occipital (n = 2).

Intraindividual comparisons during follow-up showed that the number of patients with EEG normalization in the STM group increased over time. Normalization of specifically pathologic sleep EEGs also was observed in the placebo group, but in a much smaller fraction of patients (Table 3). The overall EEG classification at each time point is displayed in Fig. 2.

Table 3.  Normalization of individual sleep EEGs over time
(n = 31)
(n = 35)
  • STM, sulthiame.

  • a

     Because of terminal failure event.

  • b

     Previous awake EEG showed a typical rolandic focus.

No normalization616
Transient normalization124
Constant normalization91
No follow-up EEGa111
Screening normalb10
No sleep EEG at screening23
Figure 2.

Comparison of EEG pathology over time Wilcoxon-U-Test: p= exact 2-tailed p-value when comparing individual changes in both groups.

When analyzing individual courses regarding the grade of pathology, there was a clear tendency for improvement in the STM group in cases without complete normalization. The individual EEG changes after 4 weeks are displayed in Fig. 3 for the STM group and in Fig. 4 for the placebo group.

Figure 3.

Individual EEG changes over first 4 weeks of therapy: STM patients.

Figure 4.

Individual EEG changes over first 4 weeks: Placebo patients.

When analyzing individual changes over time by comparing the grade of pathology, the U test underlines the different courses between the two groups (two-sided p values: 4 weeks, p < 0.0001; 3 months, p = 0.0010; and 6 months, p < 0.0001).


STM has been demonstrated to be effective in BECTS regarding seizure control (1). We report the first placebo-controlled EEG investigation into the benefit of STM on EEG pathology. It has not been clear whether the interictal EEG changes in BECTS have an influence on neuropsychological findings and behavior. If one alleges the hypothesis of adverse effects caused by interictal changes (3,4), a drug treatment should not only control seizures, but should preferably also improve the EEG. Besides seizure control, only two drugs, clonazepam (CZP) and STM, are reported to have positive effects on the EEG in BECTS (7,8). In a trial comparing CZP, valproate (VPA), and carbamazepine (CBZ) in BECTS (7), global EEG normalization after 4 weeks was observed in 15 of 20 patients treated with CZP and in one of 10 patients receiving VPA. In contrast to CZP, no EEG normalization could be found in the 10 patients treated with CBZ in that study. In addition, CBZ may have the disadvantage of worsening EEG characteristics and precipitating of epilepsy with continuous spikes and waves during slow sleep (CSWS) in a considerable number of patients (3,4,9–13), although this influence is controversial (14). Early observations showed positive effects of STM treatment on the interictal EEG in BECTS (2,11,15). Of 58 patients treated with STM, 51 showed initial normalization, which turned out to have been transient in 19 of them (2).

In this study we attempted to replace the global assessments of normalization, improvement, no changes, or deterioration when comparing two EEGs by a more specific classification of EEG pathology to detect individual changes. In our understanding, the grade of EEG pathology in BECTS is determined mainly by the number of specific foci, the amount of spiking, and the extent of contralateral propagation and generalization (Table 2).

Amount of spiking, contralateral regional propagation and complete generalization were rated exponentially because a higher grade of pathology should be assigned to EEGS with (a) one focus that is apparent 80% of the time, (b) one focus with a temporal proportion of 50% and all of the sharp waves generalizing, or (c) three independent foci, each with 25% of the whole EEG activity. All these examples were classified as grade 3 EEGs.

In any case with three foci, three additional rating points were added, leading to at least grade 2 EEG, even if there was a little spiking, propagation, or generalization. Three independent foci were found in three patients (two STM and one placebo). Although it is well known that sharp-wave activity in BECTS is activated during deep sleep, we did not distinguish between light or deep sleep, as the number of comparable EEGs would have been decreased substantially. According to the study design, individual follow-up EEGs were available only when the patient remained in the study. This led to a marked imbalance in the number of recorded EEGs over time for the two study groups, because many more patients treated with placebo had TFEs (Fig. 1). As displayed in Table 1, there also was a tendency toward normalization or improvement in the placebo group. The increasing time interval from the previous seizure may promote a normalization of EEG changes. One can speculate that patients with a higher grade of EEG pathology may relapse in a higher number. However, patients lost to follow-up because of a seizure were previously classified to have grade 1 screening EEGs in most cases in our study (Fig. 4). If one assumes a correlation between EEG pathology and seizure relapse, there should be a bias toward an increasing number of less pathologic or normal EEGs in the placebo group compared with STM, because of the higher number of seizures in this group. In contrast to this assumption, we found a stronger tendency for normalization in the STM group (Fig. 2). Because there was a substantial treatment-group difference in favor of STM at week 4, when most of the placebo patients were still in the trial, it is not very likely that the EEG data were biased in favor of STM by patients dropping out because of a terminal event. The observed (transient or constant) EEG normalization in five patients of the placebo group demonstrates the variability of EEG pathology in the natural course of BECTS (5).

In summary, this study demonstrates that STM improves pathologic EEG changes during individual follow-up in a considerable proportion of patients with BECTS.