Address correspondence to Alexandre N. Datta, Division of Pediatric Neurology and Developmental Medicine, University Children's Hospital, Basel, Switzerland. E-mail: email@example.com
Benign epilepsy with centrotemporal spikes (BECTS) is associated with mild cognitive deficits, especially language impairment. This study aimed to clarify whether children with BECTS with left- or right-hemispheric, or bilateral focus have specific neuropsychological language deficits when compared to healthy controls, whether these deficits correlate functionally with language network organization (typical vs. atypical), and whether cofactors such as duration, handedness, and medication have a relevant impact on language reorganization processes.
Twenty-seven patients and 19 healthy controls were examined with several neuropsychological tests (German version of the Wechsler Intelligence Scale for Children [WISC-IV], Regensburger verbal fluency test [RWT], Corsiblock forward and backward and Hand-Dominanz-Test [HDT]) and with two language paradigms on functional magnetic resonance imaging (fMRI): silent reading of word-pairs and silent generation of simple sentences.
Although neuropsychological test results only differed by trend between BECTS patients and controls, language laterality indices (LIs) in fMRI were significantly lower in patients than in controls. In particular, the anterior language network with Broca's area and the supplementary motor area (SMA) revealed the lowest LIs and showed the most bilateral or right hemispheric activations in the sentence generation task. Medication and duration of epilepsy did not have any significant effect on language reorganization and patients' performances.
Language reorganization in BECTS patients takes place in bilateral or right hemispheric language networks, with a strong focus in anterior language regions. These functional changes can be interpreted as important compensatory strategies of the central nervous system (CNS) to stabilize cognitive, especially language performance.
The impact of epilepsy on cerebral plasticity is a subject of ongoing debate. Atypical language networks in children with epilepsy have already been described in mixed refractory epilepsy cohorts in Wada test (intracarotid amobarbital procedure [IAP]) studies (Rasmussen & Milner, 1977) and several functional magnetic resonance imaging (fMRI) investigations (Anderson et al., 2006; Yuan et al., 2006). The natural development of language in childhood shows a trend to an increasingly left hemispheric adult language network, as documented in a longitudinal fMRI study by Szaflarski et al. (2006) and in a cross-sectional study by Everts et al. (2009). However, natural language development can be affected in children with epilepsy. Evidence from patients with symptomatic and cryptogenic left temporal lobe epilepsies has shown that normal left hemispheric specialization for language may be obviously hampered (Datta et al., 2009).
To disentangle the effect of pure epilepsy from lesional epilepsy on cerebral reorganization, we decided to focus this study on plasticity in children with benign epilepsy with centrotemporal spikes (BECTS), the most common idiopathic focal epilepsy in childhood and most probably a transient brain maturation pathology (Doose & Baier, 1989). Characteristically in idiopathic focal childhood epilepsies, the child's intelligence is normal and no other neurologic deficits are present (Giordani et al., 2006; Ay et al., 2009). Nevertheless, idiopathic epilepsy can affect the neurocognitive integrity of these children (Deonna et al., 2005).
Neuropsychological studies agree on the fact that children with BECTS maintain a normal global intellectual capacity, although neuropsychological deficits may be present. These deficits are dependent on the location of the epileptic discharges (Bedoin et al., 2011). In addition to deficits in nonverbal functions, dysfunctions particularly related to language have been reported (Overvliet et al., 2011). These include delayed reading, numeracy, and spelling (Pinton et al., 2006); impaired reading performance (Clarke et al., 2007; Ebus et al., 2011); delayed language development with mixed phonologic and lexicosyntactic problems (Monjauze et al., 2005); deficits in verbal fluency, verbal reelaboration, semantic knowledge, and lexical comprehension (Riva et al., 2007; Piccinelli et al., 2008); as well as oromotor deficits (Deonna et al., 1993). In addition, language lateralization evaluated with neuropsychological tests has been shown to be atypical in children with BECTS. In this context, Bulgheroni et al. (2008) showed an atypical hemispheric speech lateralization in these children while they were performing a dichotic listening task.
The only existing fMRI study in children with BECTS confirmed atypical language networks compared to healthy controls, especially for the anterior language network (Lillywhite et al., 2009).
To get a more comprehensive understanding of language development, the process of reorganization, and the influence of epileptic discharges, a combinatory approach including neuropsychological as well as functional neuroimaging methods is required. In the present study we aimed to verify not only whether BECTS causes deficits in language performances, but also whether these deficits are correlated with the functional reorganization of language networks in contrast to healthy controls. We further aimed to determine the predictability of left-handedness for an atypical language network. In addition, we intended the following: (1) to clarify the impact of duration of epilepsy on reorganization, (2) to investigate the effect of left hemispheric discharges on left hemispheric functions, and (3) to study a possible crowding-out effect for right hemispheric functions. Crowding-out implies in this case the takeover of the right temporal region of the brain by the hierarchically higher placed language network, at the expense of typical right hemispheric functions.
Twenty-seven children with a diagnosis of BECTS were recruited over a period of 2.5 years from the ambulatory clinics for Child Neurology at the children's hospitals in Basel, Aarau, Bellinzona, Berne, Lucerne, and St. Gallen.
Inclusion criteria: Girls (n = 13) and boys (n = 14) between 7.4 and 13.1 years of age (mean 9.9, standard deviation [SD] 1.5) were included with or without antiepileptic treatment (n = 15 with, n = 12 without treatment). The decision for antiepileptic monotherapy with sultiam (n = 14) or a combination of two drugs (n = 1 with sultiam and levetiracetam) was not only based on the number of seizures but also on a short time interval of only a few weeks between the seizures. Three patients had a high frequency of epileptic discharges, which motivated the application of antiepileptic treatment.
With the exception of one boy with left-handedness and one with ambidexterity, participants were right-handed. All patients had included in their clinical diagnostic work-up standard wake and sleep electroencephalography (EEG) or sleep deprivation EEG. To specify the EEG focus (left, right, or bilateral (fronto)-centrotemporal), we chose the most active focus during somnolence and the first non–rapid eye movement (NREM) sleep state N1. We classified patients with very predominant epileptic focus (>80% of all discharges) as epilepsy with unilateral focus; patients were classified as bilateral in the case that both foci were active (<80% for each focus, e.g., 60% and 40%). We chose the focus that appeared early in their epileptic disease (at least in the EEG studies of the first year of their disease). For the small number of BECTS patients, those with preferentially anterior (frontocentral) and preferentially posterior (centrotemporal) rolandic foci were all assigned to the same group of either right, left, or bilateral foci according to the definitions described earlier. Epileptogenic discharges ranged from spikes to sharp waves but did not vary in waveform morphology within one patient's EEG recordings over several years. Additional slow waves in combination with spikes or sharp waves were uncommon.
Exclusion criteria were the following: (1) any other epilepsy than BECTS; and/or (2) any parenchymal pathology which could affect the diagnosis of BECTS, for example, a pathologic magnetic resonance imaging (MRI); (3) other accompanying neurologic disorders such as cerebral palsy, brain tumor or neurometabolic diseases, and mental retardation. Comorbidity of attention deficit/hyperactivity disorder (ADHD) was registered, but was not necessarily regarded as an exclusion criterion, due to its common comorbidity in the population of BECTS (Table 1): three patients had a diagnosis of ADHD, but none of them received specific medication.
Table 1. Patients' characteristics (age, sex, age at onset of epilepsy, number of seizures, side of the EEG focus, antiepileptic drug(s))
Nineteen healthy girls (n = 6) and boys (n = 13), age range between 8.6 and 13.3 years (mean10.9, SD 1.6) participated as a control group. Two boys were left-handed. Children with any epileptiform discharges on EEG or any personal or family history of epilepsy or developmental delay, especially in speech, neurologic, or psychiatric disorders were excluded.
Because of fMRI scanning, children with fixed dental braces had to be excluded as well.
A comprehensive neuropsychological examination was administered to each child by a neuropsychologist using the German version of the Wechsler Intelligence Scale for Children (WISC-IV; Petermann & Petermann, 2008). This test battery assesses various aspects of age-adequate cognitive functions: language comprehension, logical reasoning, processing speed, and verbal short-term span. In addition, language production was measured using two different word fluency tasks of the Regensburger verbal fluency test (RWT), semantic and phonematic word fluency. Children were asked to name as many animals (semantic) or words beginning with the initial letter “s” (phonologic) in 1 min (Aschenbrenner et al., 2000). To measure the visuospatial memory and visuospatial short-term span Corsiblock forward and backward was used (Orsini et al., 1987). Finally, a test for handedness was applied (Hand-Dominanz-Test, HDT, Steingrüber, 1971). Statistical analyses of performances on the neuropsychological tests were based upon age- and gender-standardized data and compared between groups using unpaired t-tests. Data was analyzed using standard statistical software (SPSS, version 20, IBM, New York, NY, U.S.A.). A significance threshold of p = 0.05 was set for all analyses.
fMRI as a method offering high sensitivity to local changes in cerebral blood oxygenation was used to detect differences in brain activation between children with epilepsy and healthy controls while performing different, but mainly speech related tasks. Children were familiarized with all tasks outside the scanner and were trained to perform the tasks silently.
All tasks were programmed using E-Prime Software (version 1.1.3; Psychology Software Tools, Sharpsburg, PA, U.S.A.) and generated by a PC. Stimuli were back-projected via an MRI-compatible projector onto a screen that could be viewed through a mirror attached above the scanner's head-coil. To significantly decrease the intensity of the noise of the machine, children were given earplugs.
Language activation task
To measure language production, the silent generation of simple sentences (subject-verb-complement) was applied: A noun was presented on the screen for 5,000 msec and children were asked to form a simple sentence (e.g., “street”: “I'm walking on the street”). One block of activation contained 5 words and lasted for 25,000 msec followed by a control condition of 25,000 msec when children had to look at a fixation cross.
The silent reading of word-pairs: Children were asked to read two words presented on the screen silently. Each of the five word-pairs of one activation block was presented for 5,000 msec. Control condition blocks lasted for 25,000 msec.
Both tasks consisted of five blocks of control condition and four blocks of activation.
Imaging was performed on a 3T MRI system (Magnetom VERIO; Siemens Healthcare, Erlangen, Germany) with a standard head coil. The imaging protocol included a sagittal T1-weighted three-dimensional (3D) high-resolution data set acquired with a magnetization prepared rapid gradient echo (MPRAGE) sequence (inversion time (TI) 1,000 msec) providing an isotropic spatial resolution of 1 × 1 × 1 mm3 and echo planar (EPI) sequences with a voxel size of 3 × 3 × 3 mm3, 38 slices, a field of view (FOV) of 228 mm, a slice thickness of 3 mm, a repetition time (TR) of 2,500 msec, a echo time (TE) of 28 msec, and a matrix size of 76 × 76. For the task Silent generation of simple sentences 94 volumes and a scan time of 235 s and for the task silent reading of word-pairs 96 volumes were acquired with a scan time of 240 s. Slices were positioned parallel to the anterior commissure-poster commissure (AC-PC) line.
Data processing and analysis
For data analyses the statistical parametric mapping software package (SPM5) implemented in Matlab (version 6.5.1, 2003; Mathworks, Inc., Natick, MA, U.S.A.) was used. All sessions were subjected to standard preprocessing procedures (Ashburner & Friston, 1997) including reslicing and realignment with second mode movement correction, co-registration to the structural T1-weighted image, normalization to the standard Montreal Neurological Institute (MNI) Template, and smoothing with a 9-mm isotopic 3D Gaussian filter. To extract the contrast of interest and the model design, the smoothed images were subjected to a first-level analysis where vectors of stimulus onsets were entered. Each condition (active and passive) consisted of 12 scans. As additional covariates the movement parameters extracted in the realignment were included to remove residual variance. Performed tasks with moving parameters above 4 mm of translation and 2 degrees of rotation were excluded. To contrast the activation, the active condition substracting the control condition was specified. These contrast images were submitted to a two-sample t-test to extract group effects with a threshold of p = 0.001 and a cluster size of 10 voxels.
Calculation of laterality indices
The laterality index (LI) for different regions of interest (ROIs) was calculated. ROIs were created using the software Masks for Region of Interest Analysis (MARINA; Walter et al., 2003). For the task silent generation of simple sentences the following ROIs were used: hemispheric, supplementary motor area (Brodmann Area BA 6), Wernicke's area on the superior temporal gyrus (BA 22), and Broca's area on the frontal inferior part (44/45). ROIs of the task silent reading of word-pairs were: hemispheric, Broca's area on the frontal inferior part (BA 44/45), Wernicke's area on the superior temporal gyrus (BA 22), angular gyrus (BA 39), and fusiform gyrus (BA 37). The LIs of these ROIs were calculated with the LI-Toolbox of Wilke & Lizba (2007). LIs were calculated with the following formula:
For left hemispheric language areas, a typical activation is a value of >0.2, a bilateral (atypical) activation is between −0.2 and +0.2, and a right hemispheric (atypical) activation is below the value of −0.2 and thus located in the right hemisphere (Hertz-Pannier et al., 1997; Springer et al., 1999). To determine the correlations between treatment, age of onset, duration of epilepsy, number of seizures and the LIs, Spearman's-rho was used.
Results of the neuropsychological examination are displayed in Table 2 with mean values and SDs of patients and healthy controls. To compare behavioral performances of patients and controls, an unpaired t-test was applied. With a significance threshold of p = 0.05 no group analysis reached significance. With the exception of the Corsiblock forward and backward, patients showed a poorer but not significantly different performance overall when compared to controls. SDs were also not significantly higher for patients than for controls. The lack of statistical difference in neuropsychological test performances was not only measurable for classically left-hemispheric language tasks, but also for working memory, perceptual reasoning, visuospatial short-term span as well as for cognitive processing speed (Table 2). Patients' performances varied from poor to excellent, whereas the control group was more homogeneous.
Table 2. Neuropsychological differences (WISC IV, Corsiblock and word fluency test [RWT]) between patients and controls: No significant differences have been found
WISC index values
Hand Dominance Test
Subgroup analyses for left hemispheric functions with LIs in left hemispheric BECTS or in right hemispheric BECTS, as well as analyses of right hemispheric functions with left hemispheric or right hemispheric BECTS, could not be performed due to small group sizes.
Comparison of laterality indices in fMRI
For the sentence generation task, 20 patients and 16 controls, and for the reading task, 17 patients and 11 controls, fulfilled the fMRI movement criteria. LIs were calculated for each ROI. Because normal distribution was not given (most likely because of small sample size) the Mann-Whitney U-test was applied. For the sentence generation task, patients showed significantly lower LIs in the inferior frontal gyrus (IFG; z=−2.682; p=0.007), the supplementary motor area (SMA; z=−2.133; p =0.033), and hemispherically in total (H) (z=−2.006; p=0.046), when contrasted to healthy controls. The LI of the superior temporal gyrus (STG) did not reach significance but showed a strong trend (p=0.08). The contrast image between patients minus controls showed remaining activation for patients mainly in the right middle temporal gyrus, the left fusiform gyrus, and the right inferior prefrontal gyrus (Fig. 1). This indicates an atypical bilateral activation of language areas in BECTS, whereas the control group showed more left-sided activations (Table 3). To describe the patient group in more detail, LIs for each group of epileptic focus-side are displayed separately in Table 3, although no significant results were found. Contrary to the silent generation of simple sentences task, the reading task did not reveal any significant differences between BECTS patients and controls. The analyzed ROIs were the following: hemispheric (H), inferior frontal gyrus (IFG), superior temporal gyrus (STG), angular gyrus (ANG), and fusiform gyrus (FUS). All described activations are shown in Figs. 1 and 2.
Table 3. Mean values of laterality indices (LI) of patients with left, right of bilateral epileptic focus versus controls and of the entire patient group for all regions of interest (ROI) and for the whole hemisphere for the tasks “silent reading of words pairs” and “silent generation of simple sentences”
Silent reading of word-pairs
Silent generation of simple sentences
Bold: almost significant difference (p = 0.052) in the superior temporal gyrus between patients and controls with lower LIs for patients during “silent reading of word-pairs.”
Bold italics: Significant differences (p < 0.05) between BECTS patients and controls for the specific ROI (hemispheric, SMA and frontal inferior) during “silent generation of simple sentences.”
The activations of treated (n = 15) and nontreated (n = 12) patients:
Bivariate, one-sided correlations between treated patients and the LI of silent generation of simple sentences were all not significant (H: r = −0.053; p=0.412; IFG: r = −0.139; p=0.285; STG: r = −0.035; p=0.441; SMA: r = −0.035; p =0.441). The correlation of the task silent reading of word-pairs with the LIs also failed to produce significant results (H: r = −0.337; p=0.093; IFG: r = 0.217; p=0.202; ANG: r = 0.313; p=0.111; FUS: r = −0.217; p=0.202), with the exception of a trend for LI STG with r = −0.409, p=0.052.
The group of BECTS with left hemispheric discharges compared to right or bilateral discharges did not reveal divergent Lis, although inference statistics could not be performed due to small group sizes. Nevertheless, independent of the side of discharge, they all showed an atypical bilateral network for sentence generation and reading (Table 3).
Duration of epilepsy did not show any significant negative correlation with LIs (silent generation of simple sentences and silent reading of word-pairs). Neither the age of onset (silent generation of simple sentences and silent reading of word-pairs) nor the patients' age correlated significantly with LIs (silent generation of simple sentences and silent reading of word-pairs).
In idiopathic focal epilepsies with a lack of any lesion by definition, it may be suggested that the initial typical developmental course of language representation progressively deviates from the normal trajectory (dysmaturation), resulting in atypical patterns. Therefore, benign idiopathic epilepsies such as BECTS are an ideal model to study the direct impact of epilepsy on cerebral reorganization.
In our cohort of BECTS patients, we found a significantly lower language LI compared to our healthy control group (Table 3). The anterior language network showed the lowest LIs for the sentence-generation fMRI task, whereas the posterior language network did not show significantly lower values compared to controls. During sentence generation, which is one of the most lateralizing fMRI language tasks (Hertz-Pannier et al., 1997, 2002), it is mainly the network of productive language that is activated, including semantic and syntactic language aspects. In our study, sentences of rather simple words had to be built. This did not demand too much from the patients, since they were allowed to adapt the difficulty of the generated sentences to their personal capabilities. Therefore, the atypically lateralized and extended language network seen in these patients in fMRI is not a sign of the complex nature of the task. Because these changes in the language network can be seen in simple language tasks such as sentence generation, they can be interpreted as an effect of language reorganization.
In contrast, during the reading task which can be rated as a less strongly lateralizing language task, no significant differences were found in any region of interest (ROI).
Of interest, neuropsychological testing of our subjects did not reveal any significant differences between BECTS patients and controls, neither for language production and fluency tasks (RWT) nor for language comprehension (VC: vocabulary, similarities and comprehension subtests). Descriptive analysis of the data showed, however, that all mean values of patients, with the exception of spatial working memory and short-term memory, were lower and that SDs of the WISC index values were higher, in contrast to healthy controls. The fact that neuropsychological test results only showed a tendency toward difference from controls can be interpreted as a capability of BECTS patients to compensate in a “latent” manner for minor productive or receptive language deficits by recruiting a wider language network. Therefore, their performances turn out not to be impaired in daily life but rather in more sophisticated and demanding test situations.
The only fMRI study that has studied language networks in BECTS patients' was published by Lillywhite et al. (2009). They showed that children with BECTS had more atypical language networks compared to healthy controls. This indirect sign of cerebral reorganization was noticed especially for the anterior language network (Lillywhite et al., 2009). They postulated a high-level language difficulty in BECTS patients, since their subjects showed a significantly different performance in the language production test but not in the simple sentence repetition test when compared to controls. This difference could not be confirmed by our study and may be ascribed to the applied cognitive tasks which were most likely not challenging enough.
Our study has several limitations. First, our goal to disentangle the impact of epilepsy on patients' cognitive right and left hemispheric performances of the different subgroups (left, right, and bilateral focal discharges) failed due to the number of patients being too small. Second, left-handedness is a highly predictive parameter for atypical language network in childhood temporal lobe epilepsy (Yuan et al., 2006; Datta et al., 2009), but also for adult epileptic patients (Pujol et al., 1999) and nonepileptic patients with congenital cerebral palsy (Staudt et al., 2002). In our cohort of patients with BECTS, the impact of left-handedness in language laterality could not be calculated statistically because there was only one sinistral patient. Third, the range of individual duration of epilepsy in our cohort of patients between 7.4 and 13.1 years of age was probably not wide enough to reveal any significant correlations. Fourth, we compared pharmacologically treated to nontreated patients in our study, who did not differ significantly in their LI or in their neuropsychological results. This could be an indication of the heterogeneity of our patient cohort concerning verbal and nonverbal abilities and does not necessarily mean that antiepileptic treatment does not influence the course of the disease and the neuropsychological outcome. Our patients were all at different stages of their disease at the moment of fMRI and neuropsychological testing. In addition, the indication to treat a patient was normally based on the number and frequency of seizures and less on the spike intensity. High spike intensity was clearly not always a sign of high seizure frequency in BECTS, and spike intensity also characteristically changed from one EEG to another in some patients. Bilateral foci also often independently altered their predominance through the course of the disease. High spike intensity was exceptionally a reason to treat patients with clear neuropsychological deficits with the intention to minimize the potential negative effect on circumscribed right- or left-hemispheric functional areas. These arguments highlight the heterogeneity of this rather small cohort of 28 patients and the impossibility of drawing any conclusions from this comparison between treated and untreated patients. For this reason a comparison between left-/right-hemispheric and bilateral focus and the impact on left- and right-hemispheric functions was impossible.
To conclude, our study findings indicate that BECTS patients reorganize their language in more bilateral or right-hemispheric language networks, especially for the anterior language areas, without significantly impacting their cognitive, especially language performances.
We believe that longitudinal fMRI and neuropsychological studies on patients with BECTS will be the best approach to distinguish the influence of nonlesional focal epilepsies such as BECTS on functional brain regions and their development.
We would like to thank all the families and children who took an active part in our study. We also thank Prof. Klaus Scheffler and Dr. Markus Klarhöfer from the Division of Radiological Physics, University Hospital Basel and Karsten Specht from the Department of Biological and Medical Psychology, University of Bergen, Norway for teaching in SPM. We thank Dr. M. Stöcklin for assisting with statistics and Mrs Janet Maccora for correcting the English text. We thank Dr. Patricia Dill and Dr. Elisabeth Böhringer from our team as well as Dr. Oswald Hasselmann from St.Gallen for their help in recruiting patients. We thank the Basel University Children's Hospital Matching Funds and the Liga gegen Epilepsie Schweiz for supporting A.N. Datta and for the financial aid for the study. We thank the Freiwillige Akademische Gesellschaft Basel for the financial support of N.Oser. We thank Desitin Schweiz for sponsoring the book vouchers for the children participating in the study.
All authors have full disclosure of any conflicts of interest. We confirm that we have read the Journal's guidelines for ethical publication. We affirm that our report is consistent with the Journal's guidelines.