- Top of page
- Materials and Methods
- Supporting Information
Purpose: Ninety percent of patients with tuberous sclerosis complex (TSC) have epilepsy. Identification of epileptogenic areas can be difficult and studies are needed to characterize the epileptogenic network in more detail.
Methods: Five children with TSC and focal epilepsy were studied using simultaneous EEG and functional MRI recordings. Tubers were marked by a neuroradiologist on the anatomical MRI. Spike-associated BOLD (blood oxygenation level-dependent) responses were superimposed with lesions.
Results: Thirteen different types of interictal epileptiform discharges (IED) were analyzed with 12 showing a BOLD response, all involving more than one tuber.
Five studies had tubers with activations exclusively within the lesion, three studies had lesional activations extending to perilesional areas, and two studies had activations involving exclusively perilesional areas of at least one tuber. Deactivations exclusively within a tuber were found in six studies, lesional deactivations extending to perilesional areas were found in four studies, and tubers with exclusively perilesional deactivations were found in five studies.
A BOLD response was found in at least one tuber in the lobe of IED generation and presumed seizure onset (according to telemetry) in all patients. In four patients, the same tubers were involved following different IED localizations. The observed changes were always multifocal, sometimes involving tubers distant from the IED field.
Discussion: These findings suggest extended epileptogenic networks in patients with TSC, which exceed networks described in PET and SPECT studies. It was possible to identify specific interictally active tubers. EEG-fMRI provides a noninvasive method to select tubers and areas at their borders for further presurgical investigations.
Tuberous Sclerosis (TS) is a multisystem disorder that is inherited as an autosomal dominant trait and occurs in 1 in 6,000–10,000 live births (Osborne et al., 1991). Mutations in the TSC1 and TSC2 genes lead to abnormal tissue growth and differentiation affecting the brain, eyes, heart, kidneys, and skin (Crino et al., 2006). Findings in the central nervous system consist of cortical tubers, subependymal nodules, and subependymal giant-cell astrocytomas (Christophe et al., 2000). Epilepsy is found in about 90% of all TSC patients and tubers are believed to be epileptogenic. Seizures start in early childhood in most of the cases; complex partial seizures and infantile spasms are the most common seizure types (Guerreiro et al., 1998). Medically refractory seizures are closely linked with mental retardation and behavioral abnormalities (Jozwiak et al., 1998). Despite new anticonvulsants, the proportion of children with medication-resistant epilepsy remains high. In these cases, resection of epileptogenic tubers is the only potential treatment for their seizures (Fujita et al., 1997). In a great number of cases, however, it is difficult to define a target for surgical intervention. Even with new functional neuroimaging techniques and intracranial recordings, it remains challenging to distinguish epileptogenic from nonepileptogenic tubers (Chugani et al., 1998). On one hand, seizures start independently from different tubers in many patients (Bauman et al., 2005). In other patients, more than one tuber is involved in the epileptogenesis (Harvey et al., 2004). On the other hand, a widespread epileptogenicity that is not limited to the structural abnormalities seen on MRI may be responsible for difficulties in finding a candidate region for surgery. Functional neuroimaging studies provided evidence that epileptogenicity is not restricted to cortical tubers but can also affect functionally associated areas (Perreson et al., 1998; Asano et al., 2000). Intracranial EEG recordings revealed that, in particular, the tissue around and at the border of the tuber might be highly epileptogenic (Otsubo et al., 2005).
Simultaneous EEG-fMRI is a noninvasive tool to evaluate epileptogenic networks in the brain. This method allows identification of areas with Blood Oxygenation Level Dependent (BOLD) signal changes correlated with the interictal epileptic discharges (IED). Positive BOLD responses as well as negative BOLD responses therefore can delineate the irritative zone (Gotman et al., 2006). Improvement in data acquisition, using high-field scanners (3T), short recording times, and good artifact correction made it possible to obtain results with a sufficient quality using this technique in sedated children (Jacobs et al., 2007).
In the present study, we evaluated for the first time children with TSC using EEG-fMRI. We hypothesize that the underlying epileptogenic networks in patients with TSC is more wide spread than the tubers delineated on MRI, and that this network can be identified through the BOLD responses at the time of IEDs.
Materials and Methods
- Top of page
- Materials and Methods
- Supporting Information
Five patients with focal epilepsy and diagnosis of TSC (one female, four males, mean age 5.2 ± 5.1) were recruited in the Department of Neuropediatrics of the University of Kiel, Germany and the Northern German Epilepsy Centre, Raisdorf. In all patients, the diagnosis of TSC was established following the current clinical diagnostic criteria for TSC (Roach et al., 1998). Clinical work-up included video-EEG monitoring, echocardiogram, renal ultrasound, ophthalmologic, and dermatologic examination. In addition, in all except one patient, the diagnosis was confirmed by molecular genetics studies (mutations in the TSC1 and TSC2 genes) at the time of the study.
All patients had medically refractory epilepsy; in all but one the epilepsy started before the second year of life. At the time of examination, all but one patient received two medications to control their seizures, but remained with daily seizures. Only patient 1 was seizure free with sulthiame monotherapy for two months at the time of the examination, but started to have seizures again in the clinical follow-up. A clear lateralization of the seizure onset could be seen in only two patients in long-term video EEG monitoring (patients 4 and 5). Seizures were classified as different types if they showed distinguishable EEG onset in regard to topography, pattern and clinical pattern. More than one seizure pattern could be clearly identified in three patients: patient 3, with 3 seizure types and patients 1 and 2, with 2 seizure types. On routine sleep EEG, multifocal interictal epileptiform discharges were seen in all patients. The clinical details and MRI findings are shown in Table 1.
Table 1. Clinical information
|Patient||Age||Epilepsy onset||Seizure type||Seizure onset on EEG||AED||Genetic diagnosis||Neurologic involvement||Cardiac||Renal||Dermatologic||MRI findings|
|1||1||6 m||CPS||F R P L||SUL||TSC 2||Mild developmental delay||1||—||Hypomelanotic nodules||20||9||1|
|2||4||2 y||CPS Atypical absences Hypermotor seizures (nocturnal)||F L T R||LTG, SUL||TSC 1||Developmental delay||1||—||Hypomelanotic nodules||13||3||—|
|3||2||6 m||CPS Atypical absences||F bi CP R O R||OXC, VGB||TSC 2||Mild developmental delay||—||—||Hypomelanotic nodules, facial angiofibroma||19||6||1|
|4||14||6 y||CPS Hypermotor seizures (nocturnal)||F R||LTG, OXC||n.a.||Behavioral disorder||—||—||Hypomelanotic nodules, facial angiofibroma||5||1||—|
|5||5||4 m||Infantile spasms, CPS, Myoclonic jerks||FP R||VPA, VGB||TSC 2||Severe developmental delay, autistic disorder, left hemipareses||1||1||Hypomelanotic nodules, facial angiofibroma, harmar toma||18||6||—|
Informed consent was obtained from the legal guardians of all patients. The study was carried out according to the Declaration of Helsinki and was approved by the Ethics Committee of the University of Kiel, Schleswig, Holstein.
EEG-fMRI data were only acquired in children who had a clinical indication for a brain MRI, either for presurgical evaluation or as a follow-up examination. The anatomical MRI was acquired in the same scanning session and no additional sedation to that applied for the clinical MRI was needed.
All patients required sedation, as their age and mental impairment prevented them from remaining quiet in the scanner. All children had been sedated for EEG sleep studies before and their sedative medication for the EEG-fMRI session was chosen according to the tolerance of the individual patient. Thus, patients were sedated with either chloral hydrate (Chloralhydrat; average dose approx. 50 mg/kg) or chlorprotixen (Truxal; average dose approx. 20 mg/kg). During the scan, a pediatrician was present and vital parameters were monitored using MRI compatible machine.
The EEG was continuously recorded inside the MRI scanner (3-Tesla Philips Achieva, 8-channel SENSE head coil, Philips Medical Systems, Best, The Netherlands) from 30 scalp sites (10–20 system plus FC1, FC2, CP1, CP2, FC5, FC6, CP5, CP6, TP9, TP10) with a reference located between Fz and Cz. Sintered Ag/AgCl ring electrodes were attached using a “EasyCap” (Falk-Minow Services, Herrsching-Breitbrunn, Germany), which is part of the MR-compatible EEG recording system “BrainAmp-MR” (Brainproducts Co., Munich, Germany). Electrode impedance was kept below 7 kOhm. Two additional electrodes were placed on the infraorbital ridge of the right eye for recordings of the vertical EOG and on the left perivertebral region for electrocardiogram (ECG) recording. Data were transmitted from the amplifier (5 kHz sampling rate, 250 Hz low-pass, and 0.03 Hz high-pass filters) via an optic fiber cable to a computer located outside the scanner room.
All patients had the following clinical anatomical acquisitions: T2 transversal, coronal, sagittal (FOV = 200 mm, Matrix 400 × 400, 36 slices, 3-mm slice thickness, TR = 4,352 ms, TE = 100 ms, flip angle 90°), T2 FLAIR coronal (FOV = 208 mm, Matrix 208 × 208, 50 slices, 3-mm slice thickness, TR = 12,000 ms, TE = 160 ms, °, Sense factor 2). In addition, a 3D-T1-weighted anatomical acquisition (FOV = 224 mm, Matrix 224 × 224, 150 slices, 1-mm slice thickness, TR = 8.4 ms, TE = 3.6 ms, Sense factor 1.3) was performed and later coregistered with the functional images. At the end of each clinical examination, continuous BOLD fMRI data were acquired for 20 min (T2*-weighted single-shot EPI sequence, FOV = 200 mm, 30 slices, Matrix 64 × 64, 3.5-mm slice thickness, TR = 2,250 ms, TE = 45 ms, flip angle = 90°, 540 dynamics).
EEGs were filtered offline using Brain Vision Analyzer (Brain Products) and BESA (Brain Electrical Source Analysis, Megis Software GmbH, Gräfelfing, Germany). Gradient artifacts were corrected using an averaged artifact subtraction method (Allen et al., 2000). This was followed by pulse artifact removal with an Independent Component Analysis (Srivastava et al., 2005) and spatial filters based on artifact and brain signal topography (Siniatchkin et al., 2007a).
Two experienced neurophysiologists reviewed the filtered EEG to mark the IEDs and spike-wave-bursts. These were classified into distinct IED types according to spatial distribution and morphology for each patient if more than one type was present. Semiautomatic detection of the manually selected IEDs was then performed offline in BESA using a spatiotemporal pattern search and all marked IEDs were visually reviewed. No IED types were excluded.
The fMRI images were motion corrected and smoothed (Gaussian kernel, FWHM = 8 mm) using SPM 2 software (Statistical parametric mapping, http://www.fil.ion.ucl.ac.uk/spm/). fMRIstat software was used for further statistical analysis (Worsley et al., 2002). Each IED that was noted in the EEG was included in the analysis as an event, with IEDs differentiated into different event types based on their spatial distribution. One general linear model was constructed in which each IED type was entered as a separate contrast of the same analysis. In the case of events occurring in close temporal proximity, the model HRFs were assumed to summate linearly. Thus, a close succession of IEDs would be modeled as a larger and longer BOLD change than an individual IED. We performed four separate analyses each using a different model HRF. Models consisted of a single GAMMA function with a FWHM of 5.4 s, with peaks 3, 5, 7, or 9 s after the event, resulting in four separate statistical maps. A single combined statistical map was then created by taking the highest absolute value of each voxel. This allowed some variation in the latency of the BOLD response (both between patients and within regions of each patient) while retaining information about its expected shape (Bagshaw et al., 2004). Significant responses were defined as 10 or more contiguous voxels with t > 3.1 (p = 0.05, corrected for four analyses). Significant positive and negative BOLD responses were visualized using Anatomist (http://brainvisa.info/).
Evaluation considering the lesion and EEG localization
An experienced neuroradiologist (AR) reviewed all anatomical MRIs and highlighted the contours of all tubers, subependymal nodules, and giant cell astrocytomas on the T1 images. The fMRI results were coregistered to and displayed on these images. We correlated the different lesions with the BOLD responses in each patient, looking at lesional as well as perilesional BOLD changes. A BOLD change was considered perilesional if it extended past the border highlighted by the neuroradiologist. As a second parameter, the concordance between BOLD responses and the EEG focus was examined. A BOLD response was considered to be within the EEG focus if it was in the brain area considered to generate such IED topography on the surface EEG, as determined by BESA through averaged voltage maps of the IEDs.
- Top of page
- Materials and Methods
- Supporting Information
In this study, we applied simultaneous EEG-fMRI for the study of epileptogenic networks in a pediatric series of TSC patients. In all patients, multiple, but not all, tubers were involved following the interictal discharges. Sometimes they were located distant to the IED topography, possibly over tissue that was normal on the MRI, suggesting that extended brain areas might be involved during interictal spiking in patients with TSC. In all cases, BOLD responses were limited to parts of the lesions, but extension beyond tuber boundaries and even exclusive involvement of perilesional areas was noted. Different IED types in one patient often lead to involvement of the same tubers.
Our EEG-fMRI study demonstrated a sufficient sensitivity. All but one IED type were associated with BOLD responses. These results correspond with other studies in children (Jacobs et al., 2007; DeTiege et al., 2007). In adult EEG-fMRI studies, positive BOLD responses strongly correlate with the irritative zone, i.e., the EEG-focus or a lesion, while negative BOLD responses are more likely to be observed in distant brain areas (Kobayashi et al., 2005; Salek-Haddadi et al., 2006). In this study, we analyzed both positive and negative BOLD responses and their concordance with the EEG topography and cortical tubers. We could not find different patterns for positive and negative BOLD responses as both were found in lesional areas, areas of IED topography, and distant brain areas in similar amounts. This corresponds with our previously published study, which found positive and negative BOLD responses encompassing parts of the irritative zone equally (Jacobs et al., 2007). This relative increased prevalence of negative BOLD responses in children may be related to sleep, sedation, or age, i.e., conditions that cause changes in baseline activity (Siniatchkin et al., 2007b). There has been evidence that negative BOLD responses may result from different baseline states while still representing the same activity as positive BOLD responses (Shulman et al., 2007). Therefore, in the following part of the article, we will discuss BOLD responses without differentiation between positive and negative.
Correlation between BOLD responses and lesions
There is a need for functional neuroimaging studies in TSC patients to identify epileptogenic brain areas and describe targets for a surgical intervention. PET can sometimes be highly sensitive in distinguishing epileptogenic from nonepileptogenic tubers (Chugani et al., 1998). However, this technique is insufficient and provides inconclusive results in some patients (Sood & Chugani, 2006). Interictal SPECT studies alone show a very low sensitivity in localizing focal epileptogenic areas (Harvey & Berkovic, 1994) but a combination with ictal studies has demonstrated sensitivity as high as 95% (Véra et al., 1999). However, ictal SPECT can be misleading if an insufficient number of seizures is evaluated or if there is a rapid seizure propagation as is often seen in children. EEG-fMRI may represent an alternative to other functional neuroimaging techniques.
All of our patients had multiple tubers and presented with multifocal interictal discharges and seizure patterns, suggesting that more than one tuber could be involved in the epileptogenic process. High tuber counts were found in all but one of our patients, indicating the severity of their disease; patients with a high tuber count are likely to suffer from severe impairment and epilepsy (Goodman et al., 1997). Especially in patients with high tuber counts, the identification of epileptogenic tubers with surface EEG is very difficult due to insufficient spatial resolution. These patients may benefit from combining the temporal resolution of the EEG with the spatial resolution of a high field MRI. A recent study showed that EEG-fMRI may be useful for presurgical evaluations in epilepsy patients (Zijlmans et al., 2007).
Our study demonstrated that BOLD changes were limited to restricted areas of the tuber (often on one side or at the border of the lesion), as has been described before in patients with polymicrogyria (Kobayashi et al., 2005). The presence of BOLD responses in lesional areas may indicate an intrinsic epileptogenicity, as suggested in many other studies (Koh et al., 2002; Kobayashi et al., 2005; Weiner et al., 2006; Jacobs et al., 2007). Additionally, BOLD responses were found in perilesional areas, sometimes exclusively. This agrees with findings of intracranial recordings and MEG dipole analysis (Perreson et al., 1998; Koh et al., 2002). Most likely, the developmental disorganization of cortical layers extends past the borders of the tuber, which can be seen on MRI. This finding is important for the planning of intracranial electrode placement and all efforts should be made to include all lesional borders in the electrode coverage.
Interestingly, one patient showed a positive BOLD response in a subependymal astrocytoma. These tumors are known to cause neurological complications such as the obstruction of CSF pathways and consecutive hydrocephalus (Goh et al., 2002), but normally not regarded as part of the epileptogenic process and their removal was not associated with seizure improvement (Cuccia et al., 2003). However, these tumors derive from stem cells and consist of proliferating neuronal and glial cells (Sharma et al., 2004), therefore groups of undifferentiated or immature neurons may be part of the epileptogenic network.
Correlation between BOLD response and EEG
BOLD responses in tubers were not limited to the lobe corresponding to the topography of the interictal discharge. In 77% of the studies, a BOLD response was seen in one tuber lying in the brain area that corresponded with the IED topography. Additional tubers were found showing BOLD responses in other brain regions that were located distant from the IED. Patient 1, for example, showed bifrontal lesional negative BOLD responses following a left temporo-occipital IED. This phenomenon is not yet understood, but it is very unlikely that this observation is only a result of the limited spatial resolution of the EEG or a propagation phenomenon.
Therefore, it can be hypothesized that an IED may be associated with BOLD changes not only in the lesional areas corresponding with the discharge, but also in distant lesional areas. These additional lesions might be potential irritative zones and contribute to the spread or generation of IEDs. This might also explain why some patients showed involvement of the same tuber following different types of IEDs. In patient 5, for example, the same occipital tuber showed a lesional positive BOLD response following IEDs over the right parietal region and a negative BOLD response following independent IEDs over the right central region.
In our patients, the seizure onset zone was determined following a long-term telemetry study. Tubers lying in the seizure onset zone showed BOLD changes in all patients, mainly negative responses. This suggests that EEG-fMRI could identify the tubers that are mainly involved in the epileptogenic process. However, intracranial recording would still be needed to confirm that the seizure truly originated from the delineated tuber, as sometimes multiple tubers were localized in one lobe. Nevertheless, in cases where multiple tubers are found in an area representing the seizure onset on surface EEG, EEG-fMRI in combination with other functional studies could help determine depth electrode placement in the correct lesion.
Widespread involvement of different brain structures during interictal discharges
Despite observations in patients with polymicrogyria and heterotopia in whom a single lesion could be delineated as epileptogenic, our patients showed involvement of several tubers in each study. This may represent the multifocal character of the patient's disease. In all but patient 4, we found evidence of bilateral changes of brain activity following the IEDs. Moreover, PET and SPECT studies sometimes showed multifocal epileptic activity in several tubers, which were associated with multifocal interictal discharges on surface EEG, poor surgical outcome, and severe cognitive impairment of some patients with TSC (Rintahaka & Chugani, 1997; Chugani et al., 1998). However, changes extending the lesional areas were not observed in these studies. Our results show an even more widespread involvement of brain tissue following the interictal discharges. All of our patients showed severe neurological impairment as a result of TSC. Neurophysiological changes extending the areas of lesional changes have been described previously in patients with TSC (Jambaque et al., 1993). The severe impairment of our patients may either reflect primary structural changes extending beyond the lesions visible on MRI, or secondary functional changes due to the frequent seizures (Palmini et al., 1991, 1995; Jambaque et al., 1993;). Both factors, lesions and early seizure onset, are known to correlate with reduced intellectual function (O'Callaghan et al., 2004). The multifocality of the BOLD responses in different tubers and the occurrence of distant positive and negative BOLD responses might reflect a large involvement of brain tissue during interictal discharges. The impact of interictal discharges on development and cognition is a controversial issue (Binnie & Martson, 1992; Holmes et al., 2006), but frequent multifocal changes of neuronal activity and blood flow, as we observed, could interfere with both. EEG-fMRI studies in a larger group of patients may help better understand these networks, especially since it could provide a noninvasive method to examine patients with different severities of the disease and even asymptomatic carriers of the gene.
- Top of page
- Materials and Methods
- Supporting Information
Figure S1. This figure shows positive and negative BOLD responses in patient 2 (study 2). The strongest positive response is seen over the right temporal regions, which corresponds with the spike marked over the right temporal area T4/T8. The strongest negative response is seen in a left frontal tuber, which showed response to the interictal discharge in both studies in this patient and lies in the area of his presumed seizure onset zone. Visualization was performed as described in Fig. 1.
Figure S2. In patient 4 (study 5), a strong positive BOLD response was observed within a tuber in the right temporal lobe. The patient was scanned twice with similar results. The localization of the EEG spike correlated well with the BOLD responses inside the tuber described. The patient showed two types of seizures: typical sleep related hypermotor frontal seizures starting over the right frontal lobe and complex partial seizures with an EEG onset over the right temporal area. Both tubers with BOLD responses therefore correlate well with the area of seizure onset. Visualization was performed as described in Fig. 1.
Figure S3. The strongest negative BOLD response in study 10 of patient 5 was observed within and at the border of a right parietal tuber, which corresponded well with the map of the right parietal spike. A positive BOLD response was found within a right occipital tuber. The seizure onset in these patients was observed over the right fronto-parietal area. Visualization was performed as described in Fig. 1.
Please note: Blackwell Publishing are not responsible for the content or functionality of any supplementary materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.
Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.