Origin and Propagation of Epileptic Spasms Delineated on Electrocorticography

Summary

Purpose

Ictal electrographic changes were analyzed on intracranial electrocorticography (ECoG) in children with medically refractory epileptic spasms to assess the dynamic changes of ictal discharges associated with spasms and their relation to interictal epileptiform activity and neuroimaging findings.

Methods

We studied a consecutive series of 15 children (age 0.4 to 13 years; nine girls) with clusters of epileptic spasms recorded on prolonged intracranial subdural ECoG recordings, which were being performed for subsequent cortical resection, and in total, 62 spasms were analyzed by using quantitative methods.

Results

Spasms were associated with either a “leading” spike followed by fast-wave bursts (type I: 42 events analyzed quantitatively) or fast-wave bursts without a “leading” spike (type II: 20 events analyzed quantitatively). Twenty-three of the 42 type I spasms but none of the 20 type II spasms were preceded by a focal seizure. A “leading” spike had a focal origin in all 42 type I spasms and involved the pre- or postcentral gyrus within 0.1 s in 37 of these spasms. A leading spike was associated with interictal spike activity >1/min in 40 of 42 type I spasms and originated within 2 cm from a positron emission tomography glucose hypometabolic region in all but two type I spasms. Failure to resect the cortex showing a leading spike was associated with poor surgical outcome (p = 0.01; Fisher’s exact probability test). Fast-wave bursts associated with spasms involved neocortical regions extensively at least in two lobes within 1.28 s in all 62 spasms and involved the pre- or postcentral gyrus in 53 of 62 spasms.

Conclusions

Epileptic spasms may be triggered by a focal neocortical impulse in a subset of patients, and a leading spike, if present, might be used as a marker of the trigger zone for epileptic spasms. Rapidly emerging widespread fast-wave bursts might explain the clinical semiology of epileptic spasms.

Epileptic spasms are characterized by clusters of short contractions typically involving the axial muscles and proximal limb segments (1). Because the spasms begin mostly in infancy between months 3 and 12, they were previously called “infantile spasms.” This seizure semiology is referred to as “epileptic spasms” in the present study, because spasms occasionally persist or may even have their onset in older children (2–4). West syndrome is characterized by the triad of epileptic spasms, hypsarrhythmic EEG pattern, and arrest in psychomotor development, but not all children with epileptic spasms show hypsarrhythmia on interictal EEG or developmental delays (1,5,6). In addition, subsets of patients with epileptic spasms have other types of epileptic seizures, such as partial seizures (1,2,7–10).

In patients with epilepsy, intracranial digital electro-corticography (ECoG) monitoring is currently considered the gold standard to assess neuronal activity, including ictal discharges, with a temporal resolution of ≥5 ms and a spatial resolution of 1 cm (11,12). Ictal ECoG patterns associated with epileptic spasms have been briefly described as “diffuse fast wave activity” in a single patient (4), but more detailed descriptions or systematic studies of ictal ECoG changes during spasms are not available. It remains unclear how focal features on ECoG recordings before or during epileptic spasms can be used to tailor cortical resection for alleviating spasms. In the present study, ictal ECoG changes associated with epileptic spasms were assessed on three-dimensional reconstructed magnetic resonance imaging (MRI) in children who underwent long-term ECoG monitoring for subsequent cortical resection, and the quantitatively processed ictal ECoG findings were compared with interictal ECoG findings, neuroimaging abnormalities, and surgical outcomes.

METHODS

Patients

A total of 24 children with epileptic spasms underwent cortical resection for treatment of uncontrolled seizures in Children’s Hospital of Michigan between May 2001 and October 2004. Nine of the 24 subjects were excluded from the present study, either because cortical resection was performed immediately after the intraoperative ECoG recording without extraoperative digital ECoG recording, or because no clusters of epileptic spasms were captured during extraoperative digital ECoG recording. The subjects (N = 15; age 5 months to 13 years; nine girls and six boys) underwent scalp video-EEG monitoring, MRI, 2-deoxy-2-[¹⁸F]fluoro-D-glucose (FDG) positron emission tomography (PET), and prolonged intracranial ECoG monitoring with subdural electrodes as part of their presurgical evaluation. On MRI, seven children showed multiple cortical tubers; two children had subcortical heterotopias; one child each had a solitary cortical tuber, a brain tumor, a calcified brain lesion, and focal polymicrogyria. In the remaining two children, MRI was normal, but PET scan showed cortical regions with glucose hypometabolism in the presumed epileptic hemisphere. After resective surgery, 10 children became seizure free, and eight of them have remained seizure free for >12 months (Table 1). Two children each had >90% and 50–90% reduction of seizure frequency, and the remaining child had <50% reduction of seizure frequency. The follow-up period ranged from 3 to 45 months (mean follow-up, 25 months). The study was approved by the Institutional Review Board at Wayne State University, and written informed consent was obtained from the parents or guardians of all subjects.

TABLE 1.

Patient data

Patient Number	Age (yr)/Gender	Age of Seizure Onset	MRI	Ictal ECoG Pattern Associated with Spasms	Cortical Resection	Surgical Outcome (Follow-up [months])
1	0.4/Boy	2 day	Solitary tuber in Rt precentral G	Type I	Rt pre- and postcentral G	Seizure-free [15]
2	1/Boy	8 months	Solitary tumor in Lt postcentral G	Type I	Lt post- and precentral G	Seizure-free [3]
3	1/Girl	3 months	Multiple tubers	Type I	Rt T-P	Seizure-free [33]
4	1/Girl	3 months	Multiple tubers	Type I	Rt F	Seizure-free [45]
5	2/Boy	3 months	Multiple tubers	Type I	Lt T-P	50-90% Reduction [36]
6	3/Girl	2 months	Multiple tubers	Type I and Type II	Rt T-P-O-F	Seizure-free [30]
7	4/Girl	4 months	Heterotopia in Lt F and Rt T	Type I and Type II	Lt F	<50% Reduction [5]
8	5/Boy	10 months	Multiple tubers	Type I	Lt T-P	Seizure-free [32]
9	7/Girl	6 months	Multiple tubers	Type I	Lt F-T-O	Seizure-free [4]
10	7/Boy	5 months	Multiple tubers	Type I	Lt T-P-O	>90% Reduction [29]
11	7/Girl	6 years	Normal	Type I and Type II	Rt T-P-F	>90% Reduction [21]
12	7/Boy	22 months	Calcification in Lt T	Type I	Lt T-P-F	Seizure-free [40]
13	10/Girl	4 months	Heterotopia in Rt F and Rt P	Type I and Type II	Rt F-P	50-90% Reduction [33]
14	11/Girl	3 months	Polymicrogyria in Lt F	Type I and Type II	Lt F-P-T-O	Seizure-free [27]
15	13/Girl	9 years	Normal	Type II	Lt T-P-F	Seizure-free [15]

Lt, left; Rt, right; G, gyrus; F, frontal; P, parietal; T, temporal; O, occipital; Type I, a leading spike followed by fast-wave bursts; Type II, fast-wave bursts without a leading spike.

Subdural electrode placement

For extraoperative video-ECoG recording, platinum grid electrodes (10 mm intercontact distance) were surgically implanted. The total number of electrode contacts in each subject ranged from 64 to 128. The placement of intracranial electrodes was guided by the results of scalp video-EEG recording, MRI, and interictal glucose metabolism on PET. The primary sensory-motor cortex also was covered with electrodes for subsequent functional mapping in all 15 cases. A single dual-contact electrode strip was placed in the interhemispheric space to record the ECoG of homotopic medial frontal-parietal regions bilaterally in four patients (patients 1, 7, 11, and 13). All electrode plates were stitched to adjacent plates or the edge of dura mater or both, to avoid movement of subdural electrodes after placement. In addition, intraoperative pictures were taken with a digital camera before dural closure, to enhance spatial accuracy of electrode display on three-dimensional brain surface reconstructed from MRI (13).

Extraoperative video-ECoG recording

Extraoperative video-ECoG recordings were obtained by using a 128-channel Stellate HARMONIE digital system (sampling rate, 200 Hz; Stellate Inc., Montreal, Quebec, Canada) for 2–5 days. Antiepileptic medications (AEDs) were discontinued or reduced during ECoG monitoring until a sufficient number of habitual seizures were captured. Ictal ECoG recordings were visually reviewed by two of the three electroencephalographers (E.A., A.S., and J.S.). Clinical manifestations were assessed by using synchronized digital videos with 30 frames/s, and surface electromyogram (EMG) recordings from the left and right deltoid muscles were added as necessary. According to the various types of ictal scalp EEG findings reported in previous studies (1,3,6,14,15), ECoG data for each patient were assessed specifically with regard to the three electro-graphic patterns: (a) the presence of “leading” spike activity (including sharp-wave activity) consistently preceding widespread fast-wave bursts associated with spasms, (b) the presence of fast-wave bursts associated with spasms, and (c) the presence of focal seizures.

A focal seizure was defined as a sustained, focal, and rhythmic ECoG change not explained by level of arousal and clearly distinguished from background ECoG and interictal activity (16,17). Brief bursts of spikes and periodic spikes at a frequency of <2 Hz were not considered to be focal seizures for this analysis (17–19).

Quantitative analysis of interictal spike frequency recorded on ECoG

Quantitative analysis of interictal spike frequency was performed on ECoG, by using Stellate HARMONIE software, as previously described and validated (18,19). In short, three distinct 10-min ECoG segments during quiet wakefulness were selected from the video-ECoG data. In cases in which frequent spikes (>30 spikes/min in an electrode) were seen and spike distribution visually appeared consistent, three distinct 5-min segments instead of 10-min segments were selected (19). Subsequently, the averaged spike frequency for each intracranial electrode was obtained from the three ECoG segments. In cases in which extremely frequent interictal spikes (a total number of spikes >5,000 per segment) were seen and spike distribution visually appeared consistent among three segments, a spike frequency for a single ECoG segment instead of three distinct ECoG segments was used for subsequent analyses (19).

Quantitative analysis of ictal spike activity on ECoG recording

Quantitative ictal ECoG analyses were applied to at least three epileptic spasms with the fewest artifacts. To identify the entire epileptogenic region, additional seizures with all distinct ictal-onset zones also were quantitatively analyzed.

Epileptic spasms

Voltage mapping was applied to any spike activity consistently preceding spasms, by using Stellate HARMONIE software and the methods similar to those described in previous studies (20,21). Initially, the highest peak of spike activity was visually identified in each channel, and the spike activity with its highest peak occurring earliest among all channels was defined as a leading spike (22). To assess the dynamic changes of spike activity associated with spasms, sequential change of the “spike voltage” was subsequently measured every 5 ms from 20 ms before until 100 ms after the highest peak of leading spike. To assess the absolute magnitude of spike activity, “spike magnitude” (defined as the area under spike activity between 10 ms before and after the highest peak of leading spike) also was calculated for each channel. To eliminate the effect of pulse artifacts on voltage mapping (20), a low-frequency filter of 4.0 Hz was applied to the ECoG recording. Both spike voltage and spike magnitude for each channel were displayed on the surface of the 3D-reconstructed MRI volume, as described later (Fig. 1 and Supplementary Video).

FIG. 1.

A 7-year-old girl with intractable epileptic spasms associated with tuberous sclerosis complex. A: Ictal electrocorticogram (ECoG) showed a focal periodic spike-and-wave activity in the left temporooccipital region. The focal periodic spike-and-wave discharges were followed by focal fast-wave bursts in the same region. The focal seizure discharges gradually propagated to the left subtemporal region with minimal clinical correlation, and the focal seizure discharges subsided 5 s before the initiation of cluster of spasms. Each spasm was associated with a single giant spike followed by widespread fast-wave bursts. A low-frequency filter of 4.0 Hz and a high-frequency filter of 100 Hz were applied. B: The magnitude of averaged focal periodic spike-and-wave activity was highest in the left temporooccipital region. C: The magnitude of leading spike activity associated with a spasm was also highest in the left temporooccipital region. According to the ictal voltage mapping (see Supplementary Video), the leading spike originated from the left temporooccipital region and propagated to the left parietal, temporal, and frontal regions within 100 ms. D: The subtraction ictal ECoG magnitude coregistered to MRI revealed increased fast-wave magnitude (24–100 Hz) associated with a spasm in the left temporooccipital, parietal, subtemporal, and sensory-motor cortex.

Focal seizures

To increase the signal-to-noise ratio, voltage mapping was applied to the averaged spike activity associated with focal seizures consisting of periodic spike–wave activity. Recordings from all channels within ±200-ms window were averaged ≤10 times by using the peak of leading spike, as long as the voltage field visually appeared stable (21). Spike voltage as well as spike magnitude was measured as described earlier, by using the highest peak of leading averaged spike.

Quantitative analysis of rhythmic discharges on ictal ECoG recording

For patients showing continuous rhythmic activity associated with spasms or focal seizures, the “subtraction ictal ECoG magnitude” was calculated, by using the Stellate HARMONIE software. The subtraction ictal ECoG magnitude was obtained for each channel by subtracting the “averaged interictal ECoG magnitude” from the “ictal ECoG magnitude” of rhythmic discharges and represents the ictal event-specific increase of ECoG magnitude compared with the interictal baseline. Finally, the subtraction ictal ECoG magnitude was coregistered to three-dimensional MRI, as described later. The method of quantifying the ictal as well as interictal ECoG magnitude was reported in our previous study (23).

To represent the gradient of ECoG magnitude at visible ictal onset and subsequent spatial changes of ictal ECoG magnitude, an epoch of interest was placed at the very beginning of visually recognized fast-wave bursts associated with spasms or rhythmic discharges associated with focal seizures (23). To increase the signal-to-noise ratio and maintain the temporal resolution, epoch length was determined so that each epoch contained ≥10 ictal waves of primary interest, according to the ictal discharge frequency at the epoch. The frequency of ictal discharge was estimated by using the statistics cursor included in the Stellate HARMONIE software, and epoch length was set to 5.12, 2.56, and 1.28 s, when the ictal discharge frequency of primary interest at the first epoch ranged from 2 to 4 Hz, 4 to 8 Hz, and >8 Hz, respectively. Further to increase the signal-to-noise ratio for seizures starting gradually with very low amplitude fast-wave activity >8 Hz and slowly propagating to other lobes (24), epoch length was set up to 2.56 s (25). To obtain reference-free topographic maps of spectral measures, all signals were remontaged to an average reference (26).

After the placement of consecutive epochs, an amplitude spectrum (x-axis unit: Hz; y-axis unit: μV/Hz) was created for each epoch and each channel, by using a Fast Fourier Transformation. The ictal-discharge frequency was determined for each ictal event, based on the primary peak of the amplitude spectral curve for the first epoch in the most-consistent-onset electrode. The software subsequently calculated the ictal ECoG magnitude (unit, μV) at each epoch within preset frequency bands, which were defined as the summation of all frequency components under the amplitude spectral curve within the given frequency band. The frequency bands were preset as follows: 2–4, 4–8, 8–12, 12–16, 16–24, 24–32, 32–64, and 64–100 Hz. The ictal ECoG magnitude of 32–64 Hz was calculated without a 56- to 64-Hz component, if visual inspection revealed a 60-Hz artifact peak on the amplitude spectral curve on all electrodes (23). Ictal ECoG magnitudes for multiple frequency bands (up to four consecutive bands) were calculated, if a seizure event was found, where multiple rhythmic discharges at different preset frequency bands were visually recognized simultaneously at the epoch of interest and its ictal ECoG magnitude exceeded 2 SDs from the mean of its averaged interictal ECoG magnitude (23).

The subtraction ictal–interictal ECoG magnitude within specific frequency band(s) was calculated for each channel by subtracting the mean of the averaged interictal ECoG magnitude from the ictal ECoG magnitude and displayed on the surface of the 3D-reconstructed MRI, as described later (Fig. 1). The averaged interictal ECoG magnitude consisted of the ECoG magnitude within the same preset frequency band(s), which was averaged among 40 epochs clearly representing the interictal state selected from the sections ≥1 h before the ictal onset and ≥1 h after the seizure offset.

Imaging acquisition protocol

FDG-PET studies were performed as previously described (27,28), and the scalp EEG was monitored throughout PET examinations to verify that all scans were interictal. MRI including a T₁-weighted spoiled gradient echo (SPGR) image as well as fluid-attenuated inversion recovery image also were obtained (27,28). Planar x-ray images (lateral and anteroposterior images) were acquired with the subdural electrodes in place for determining the location of the electrodes on the brain surface. Three metallic fiducial markers were placed at anatomically well-defined locations on the patient’s head for coregistration of the x-ray with the MRI, as previously described (27–29).

Image analysis

To identify regions with cortical decreases in glucose metabolism, an objective program (27) that delineates cortical areas with >10% relative hypometabolism compared with the homotopic region (28) was used for patients without tuberous sclerosis complex (patients 2, 7, 11–13, and 15). Areas with decreased glucose metabolism were then marked in blue in the PET image volume. For patients with tuberous sclerosis complex (patients 1, 3–6, 8, 9, and 10), this approach was not practical because of multiple bilateral hypometabolic regions. Therefore cortical areas showing abrupt decreases of glucose metabolism compared with the surrounding cortex (supplementary figure) were considered as glucose hypometabolic areas representing cortical tubers and overlying dysplastic cortices in patients with tuberous sclerosis complex (30). Cortical areas were classified into three categories as follows: (a) glucose hypometabolic regions, (b) normometabolic regions (>2 cm away from the hypometabolic region), and (c) the border between hypometabolic and normal cortex (within 2 cm from the hypometabolic region but not completely within the hypometabolic region). Neuroimaging abnormalities were defined by MRI alone in patient 14, who had a previous epilepsy surgery and underwent no repeated FDG-PET scan after the surgery.

FDG-PET and MRI SPGR image volumes were coregistered, as previously described (27,28,31). To reconstruct surface views corresponding to the planar x-ray image, three virtual markers were defined in the SPGR MR image volume at the same position as in the planar x-ray image, as previously described (27–29). As a result, a surface view was created which corresponds to the planar x-ray image position and where the location of electrodes was directly defined on the brain surface.

Delineation of ECoG data on 3D-reconstructed MRI

Electrode positions (x- and y-axis values) on the planar x-ray coordinate were measured for every electrode by using Microsoft PowerPoint (Microsoft Corporation, Redmond, WA, U.S.A.) and were registered into the Stellate HARMONIE system as well as the SurGe Interpolation Software (Web site: http://mujweb.cz/www/SurGe/surgemain.htm ). The topographic map derived from the x-ray images was used to display the interictal spike frequency as well as the sequential change of spike voltage and subtraction ictal ECoG magnitude coregistered to 3D MRI.

Relation between interictal ECoG, ictal ECoG, and neuroimaging data

To determine the relation between the presumed generators of epileptic spasms and focal irritability ranked by interictal spike activity, the distance between the electrodes showing the maximal ECoG abnormalities associated with spasm events and the maximal interictal spike frequency was individually measured. To determine the relation between the presumed generators of epileptic spasms and cortical neuroimaging abnormalities, the distance between the electrode showing the maximal ECoG abnormalities associated with spasm events and glucose hypometabolic regions on FDG PET, as described earlier, also was measured. To determine the clinical significance of objectively measured ictal ECoG parameters, the association between the ECoG parameters on the nonresected cortex and the surgical outcomes was tested by the Fisher’s exact probability test, by using eight patients who were seizure free for ≥1 year and five patients not seizure free after resective surgery. Two patients who were seizure free <1 year (see Table 1) were excluded from the statistical analysis.

RESULTS

Scalp video-EEG findings

The background activity consisted of posterior dominant sinusoidal activity that reacted to eye opening and closure, and the frequency of posterior background rhythm was within normal limits (32,33) except for patient 7, who showed slow posterior background rhythm (Supplementary Table 2). All patients showed intermittent or consistent focal background disorganization in the presumed epileptic hemisphere. None of the patients showed hypsarrhythmia on interictal EEG. All patients showed multifocal interictal epileptiform activity more frequently in the presumed epileptic hemisphere, with the exception of patient 2, who showed unilateral interictal epileptiform activity.

Epileptic spasms were captured during scalp video-EEG recording in 12 of the 15 patients (Supplementary Table 2). Ictal EEG onset consisted of spike activity followed by fast-wave bursts superimposed on delta wave activity in five patients (patients 1, 5, 7, 12, and 14), spike activity followed by delta wave activity without clear-cut fast-wave components in three patients (patients 3, 4, and 6), fast-wave bursts superimposed on delta wave activity without a consistent preceding spike activity in three patients (patients 11, 13, and 15), and spike activity followed by fast-wave bursts without clear-cut delta wave activity in patient 10. Focal seizures were noted in 10 of 15 patients, and ictal EEG onset was localized to the epileptic hemisphere in all 10 cases (Supplementary Table 2).

Two types of ictal ECoG changes associated with spasms

Each spasm was associated with either a leading spike followed by fast-wave bursts (type I; Figs. 1–3) or fast-wave bursts without a leading spike (type II; Fig. 4). In total, 18 distinct sites of type I spasms were noted in 14 patients, and 42 type I spasms with the fewest ECoG artifacts were quantitatively analyzed. A leading spike associated with a spasm had a focal origin in all 42 type I spasms and involved the pre- and/or postcentral gyri within 0.1 s in 37 of 42 type I spasms (Supplementary Video). Conversely, a total of 11 distinct sites of type II spasms were noted in six patients, and 20 type II spasms with the least artifacts were quantitatively analyzed (Supplementary Table 3). Thirty-six of the 42 type I spasms were noted in patients with cortical lesions on MRI (patients 1–6, 8–10, 12, and 14), whereas 16 of the 20 type II spasms were noted in patients without cortical lesions on MRI (patients 7, 11, 13, 15). The Fisher’s exact probability test showed that type I spasms were associated with patients with cortical lesions on MRI, and type II spasms were associated with patients without cortical lesions on MRI (p < 0.001).

FIG. 3.

A 5-month-old boy with intractable epileptic spasms and a cortical tuber in the right precentral gyrus. A: Frequent interictal epileptiform activity was noted in the right medial precentral gyrus. Red line, The location of central sulcus. B: Ictal electrocorticogram (ECoG) showed a focal leading spike activity in the right medial frontal region with the maximal amplitude at electrode 4 on the precentral gyrus. After the spike activity, regional fast-wave bursts emerged from the right medial and lateral frontal region as well as the right parietal region. It should be noted that the fast-wave discharges minimally involved the left medial frontal region, which was correlated with milder spasm-related movement of the right-sided limbs compared with the left. A low-frequency filter of 4.0 Hz and a high-frequency filter of 100 Hz were applied. C: The magnitude of leading spike activity associated with a spasm was highest in the right medial precentral gyrus. D: The subtraction ictal ECoG magnitude coregistered to MRI revealed increase of fast-wave magnitude (16–64 Hz) associated with a spasm in the right medial and lateral frontal regions, as well as the right parietal region.

FIG. 4.

A 13-year-old girl with nonlesional intractable epileptic spasms. A: Glucose hypometabolic regions [blue-coded areas; >10% decrease of fluorodeoxyglucose (FDG) uptake compared with the homotopic region] involved the left temporoparietal-frontal regions. B: Very frequent interictal epileptiform activity was noted in the left parietal, frontal, and temporal regions, independently. C: Ictal electrocorticogram (ECoG) associated with spasm 2 showed widespread fast-wave bursts over the left hemisphere without a leading spike activity. A low-frequency filter of 8.0 Hz and a high-frequency filter of 100 Hz were applied. D: The subtraction ictal ECoG magnitude coregistered to MRI for this spasm showed increase of fast-wave magnitude (32–100 Hz) extensively in the left parietal, temporal, and frontal regions, including the postcentral gyrus. E: Another spasm event (Spasm 3) showed increase of fast-wave magnitude (32–100 Hz) in the left parietal, frontal, and temporal regions, but the distribution of increased fast-wave magnitude is different from that for Spasm 2. F: The distribution of increased fast-wave magnitude for Spasm 5 is different from Spasms 2 or 3.

Spasm-related fast-wave bursts on ECoG

The frequency of fast-wave bursts on ECoG reached the gamma range, exceeding 32 Hz in all 62 spasms. Very fast gamma bursts >64 Hz were noted in 36 of 62 spasms, and beta bursts ranging from 16 to 32 Hz were intermixed with gamma bursts in 28 of 62 spasms. The duration of fast-wave bursts associated with a spasm ranged from 1 to 5 s. Fast-wave bursts associated with a spasm involved the neocortical regions extensively in at least two lobes within 1.28 s in all 62 spasms and involved the pre- and/or postcentral gyri in 53 of 62 spasms (Figs. 1–5). A distance >2 cm was found between the electrodes showing the maximal magnitudes associated with leading spike and fast-wave bursts in 31 of 42 type I spasms, and they were located in different lobes in 10 of 42 type I spasms (noted in patients 2, 3, 5, 6, 8, and 15).

FIG. 5.

A 4-year-old girl with intractable epileptic spasms associated with subcortical heterotopia. A: Glucose hypometabolic regions (blue-coded areas; >10% decrease of fluorodeoxyglucose uptake compared with the homotopic region) involved the left frontal region as well as the pre- and postcentral gyri. Red line, The location of central sulcus; white broken line, the resection margin (frontal lobectomy). B: Frequent interictal epileptiform activity was noted in the left precentral gyrus and frontal region. C: Ictal electrocorticogram (ECoG) findings for Spasm 1 consisted of a preceding focal seizure characterized by periodic spike-and-wave discharges in the left frontal region, followed by a regional leading spike activity in the left frontal region, and subsequently followed by widespread fast-wave bursts associated with a spasm. D: Ictal ECoG findings for Spasm 2 also consisted of a preceding focal seizure characterized by periodic spike-and-wave discharges in the left frontal region, followed by a regional leading spike activity in the left frontal region, and subsequently followed by widespread fast-wave bursts associated with a spasm. E: Ictal ECoG findings for Spasm 4 consisted of a focal leading spike activity in the precentral gyrus followed by widespread fast-wave bursts associated with a spasm. It should be noted that the origin of leading spike for Spasm 4 was not resected because of the proximity to the motor cortex. F: Ictal ECoG findings for Spasm 5 consisted of widespread fast-wave bursts without a leading spike.

Temporal relation between fast-wave bursts and clinical semiology

Evolution of fast-wave bursts was correlated with clinical onset of spasms. According to video assessment of seizure semiology, spasms with fast-wave bursts involving the pre- and/or postcentral gyri were associated with more prominent movements of limbs, whereas those without involvement of the pre- or postcentral gyrus showed more subtle movements (1), such as slight head nodding or widening of the eyes. Fast-wave bursts associated with spasms were noted in the other hemisphere within 1.28 s in all four patients (1, 7, 11, and 13) who underwent placement of dual-contact interhemispheric strip electrodes. Of these four patients, patients 1 and 7 showed asymmetrical spasms with more prominent movement of the contralateral limbs, a finding that was correlated with higher fast-wave magnitude in the presumed epileptic hemisphere compared with the homotopic area (Fig. 3).

Focal seizures recorded on ECoG

During extraoperative ECoG recording, at least one focal seizure was noted on ECoG for each patient, and a total of 28 focal seizure-onset zones were noted in the 15 patients (Figs. 1 and 2). Ten sites of focal seizures preceded spasms, six followed spasms, and 12 were unrelated to spasms (Supplementary Table 3).

FIG. 2.

A 1-year-old boy with intractable epileptic spasms associated with a tumor in the left postcentral gyrus. A: Ictal electrocorticogram (ECoG) showed a focal fast-wave discharge confined to the postcentral gyrus (electrode B5), which was followed by a single giant spike and subsequently followed by widespread fast-wave bursts in the left parietal, temporal, and frontal regions. Widespread fast-wave bursts were associated with a clinical spasm. A low-frequency filter of 4.0 Hz and a high-frequency filter of 100 Hz were applied. B: The subtraction ictal ECoG magnitude coregistered to MRI revealed an increase of fast-wave magnitude (16–32 Hz) in the left postcentral gyrus. White broken line, The resection margin. C: The magnitude of leading spike activity associated with a spasm also was highest in the left postcentral gyrus. D: The subtraction ictal ECoG magnitude coregistered to MRI showed increase of fast-wave magnitude (16–64 Hz) associated with a spasm in the left temporal, parietal, and frontal regions.

Focal seizures preceding spasms

In total, 24 focal seizures preceding spasms with fewest ECoG artifacts were quantitatively analyzed. All focal seizures preceding spasms were associated with type I but not type II spasms. The duration of preceding focal seizures ranged from 10 s to 1 min. At least a 1-s interval was noted between the termination of such preceding focal seizures and the onset of spasms (Fig. 1), except for a single event in which no interval was found between the focal seizure and a spasm (Fig. 2). Clinical manifestations related to focal seizures preceding spasms included arousal, behavioral arrest, and staring off.

Focal seizures after spasms

All focal seizures after spasms consisted of periodic focal spike–wave discharges evolving from spasm-associated fast-wave bursts, building up in amplitude and propagating to the adjacent cortices, and in total, seven focal seizures after spasms with the least ECoG artifacts were quantitatively analyzed. The duration of such periodic focal spike–wave discharges ranged from 10 to 30 s. Clonic body movements were seen in a single event in patient 13, in whom periodic spike–wave discharges involved the precentral gyrus. Automatisms of the mouth and upper limbs were seen in a single event in patient 15, accompanied by periodic spike–wave discharges involving the medial temporal structures. Clinical manifestations for the other types of focal seizures after spasms were minimal.

Focal seizures unrelated to spasms

In total, 21 focal seizures unrelated to spasms from 12 distinct onset zones were quantitatively analyzed. The duration of such focal seizures ranged from 10 s to 2 min. Facial grimacing was seen in a single event in patient 1, associated with focal periodic spike–wave discharges involving the precentral gyrus. Crawling movements were seen in three events in patient 2, during which focal fast-wave bursts involved the postcentral gyrus. Clinical manifestations for the other types of focal seizures unrelated to spasms included arousal, behavioral arrest, and staring off.

Relation between ictal and interictal ECoG data

The presumed origin of epileptic spasms determined by quantitative methods was associated with higher interictal spike frequency in the majority of spasm events (Figs. 3 and 5). The electrode showing the maximal spike magnitude associated with a leading spike showed frequent interictal spikes (>10/min) in 27 of 42 type I spasms, occasional spikes (1–10/min) in 13 of 42 type I spasms, and rare spikes (<1/min) in two of 42 type I spasms. The electrode showing the maximal fast-wave magnitude associated with a spasm showed frequent interictal spikes in 30 of 62 spasms, occasional spikes in 25 of 62 spasms, and rare spikes in seven of 62 spasms. The electrode showing the maximal focal seizure magnitude showed frequent interictal spikes in 35 of 52 focal seizures, occasional spikes in eight of 52 focal seizures, and rare spikes in nine of 52 focal seizures.

Conversely, the electrode showing the maximal spike frequency predicted either the regions showing the maximal spike magnitude associated with a leading spike, the maximal fast-wave magnitude associated with a spasm, or the maximal focal seizure magnitude in nine of 15 patients. In four of the remaining six patients, the electrode showing the maximal spike frequency was not identical to but located within 2 cm from one of the previously mentioned regions.

Relation between ictal ECoG and PET data

The presumed origin of epileptic spasms determined by quantitative methods was associated with focal neuroimaging abnormalities in the majority of spasms (Fig. 5). The electrode showing the maximal spike magnitude associated with a spasm was located within glucose hypometabolic regions in 20 of 40 type I spasms, at the border between hypometabolic and normal cortex in 18 of 40 type I spasms, and in normometabolic regions in two of 40 type I spasms. The electrode showing the maximal fast-wave magnitude associated with a spasm was located within glucose hypometabolic regions in 25 of 59 spasms, at the border between hypometabolic and normal cortex in 17 of 59 spasms, and in normometabolic regions in 17 of 59 spasms. The electrode showing the maximal focal seizure magnitude was located within glucose hypometabolic regions in 30 of 49 focal seizures, at the border between hypometabolic and normal cortex in 17 of 49 focal seizures, and in normometabolic regions in two of 49 focal seizures.

Relation between ictal ECoG data and surgical outcomes

Seven of the eight patients in whom the cortex showing the maximal spike magnitude associated with leading spikes was completely removed became seizure free. Conversely, none of the four patients in whom the cortex showing the maximal spike magnitude associated with spasms was not removed because of the proximity to eloquent cortices became seizure free (Fig. 5). Failure to resect the origin of leading spikes was associated with poor surgical outcome (p = 0.01).

Six of the seven patients in whom the cortex showing the maximal fast-wave magnitude associated with spasms was completely removed became seizure free. Conversely, four of the eight patients in whom the cortex showing the maximal fast-wave magnitude was not removed became seizure free. The present study failed to demonstrate an association between the incomplete resection of the cortex showing the most prominent fast-wave bursts related to spasms and poor surgical outcome (p = 0.2).

DISCUSSION

Pathomechanisms of epileptic spasms

The pathomechanism of epileptic spasms has been studied by using a number of neuroimaging techniques such as interictal and ictal glucose metabolism PET scan (34–36), ictal and interictal single-photon emission computed tomography (SPECT) (37,38), and near-infrared spectrophotometry (38,39). These studies led to the hypothesis that the neocortex may trigger spasms at least in a subset of children with epileptic spasms, but none of these studies showed the dynamic changes of ictal discharges with high temporal and spatial resolution. In the present study, dynamic ECoG changes associated with epileptic spasms were quantitatively measured and delineated on a 3D-reconstructed MRI with high temporal and spatial resolution. It was demonstrated that each spasm was associated with either a focal leading spike followed by widespread fast-wave bursts (type I) or widespread fast-wave bursts without a leading spike (type II). Such spasm-related leading spikes were associated with interictal spike activity as well as focal decreased glucose metabolism on PET in the majority of spasm events, and resection of the origin of leading spike was associated with good surgical outcome. Therefore our results support the notion that cortex may trigger epileptic spasms, and a focal leading spike, if present, may be used as a marker of the trigger zone in a subset of patients with epileptic spasms.

Rapid involvement of widespread cortex by fast-wave bursts might explain the clinical manifestations of epileptic spasms, because fast-wave burst activity was correlated with the clinical onset of spasms, and involvement of the sensory-motor cortex by the fast-wave bursts appeared to correlate with the severity of contralateral limb movement. Similar observations on scalp EEG recordings have been previously reported (40,41). Brain regions showing fast-wave propagation often showed normal glucose metabolism on PET and did not need to be resected to achieve seizure-free outcome in some cases. We hypothesize that spasm-associated fast-wave bursts might be derived from a corticosubcorticocortical pathway rather than a corticocortical pathway, because fast-wave bursts even >64 Hz rapidly and extensively involved the noncontiguous neocortex simultaneously in the present study. It is still unknown what subcortical structures are involved in the propagation of epileptic spasms, but it was hypothesized that the brainstem (34,37), basal ganglia (34,37), and thalamus (42) may play a role in the propagation of spasms.

Caution should be recommended in interpretation of the results in the present study, because the number of subjects was relatively small, and some of the patients had a short postsurgical follow-up period. It also should be noted that the majority of patients in the present study had a normal posterior background rhythm instead of hypsarrhythmia. This observation indicates that our patients may represent a selected group of children with epileptic spasms (1).

Methodologic considerations

Several methodologic issues must be discussed in the present study. One of the major limitations of extraoperative ECoG recordings is sampling error. It is still uncertain whether widespread fast-wave bursts associated with type II spasms were intrinsically generated by the neocortex without a focal trigger impulse or leading spike not recorded simply because of sampling error in type II spasms. Even by maximizing the number of recording electrodes, all epileptogenic regions could not be covered by subdural electrodes in some patients (Fig. 1 and Supplementary Figure), who had multiple epileptogenic foci in different lobes (Supplementary Table 3). In those patients, we extended the resection, based on the location of structural lesion delineated on MRI and PET, as well as additional intraoperative ECoG recording. Technical limitations also include movement-related artifacts. Ictal ECoG changes associated with prominent movement artifacts were not quantitatively, but were visually analyzed in the present study so that all distinct seizure-onset zones were assessed. Medications may affect FDG-PET imaging and ictal discharge patterns on ECoG recording. It has been reported that cerebral metabolic rates for glucose are globally decreased by AEDs (43,44). However, use of asymmetry analysis eliminates global effects of these medications on PET. Therefore the effects of AEDs on FDG-PET studies are considered to be relatively small in the present study. The effect of medications on ictal-discharge patterns on EEG has not been reported previously. Because the sampling rate of digital EEG system used in the present study was 200 Hz, fast-wave components between 64 and 100 Hz (6) were clearly detected in some seizure events, but ultrafast-wave components >100 Hz (42,45) could not be recorded in the present study.

Future directions

A major long-range objective is the establishment of the optimal resection criteria in children with epileptic spasms by using multimodality quantitative analyses including intracranial ECoG and neuroimaging such as MRI and PET. Although the present study failed to demonstrate the association between spasm-related fast-wave bursts and surgical outcomes, this finding does not rule out the clinical utility of fast-wave bursts on ECoG in the decision making about resection margin, because the number of subjects was small in the present study. Because resective surgery for infantile spasms is currently performed as an effective treatment in a number of epilepsy surgery centers (46–50), including our own (51), a multicenter collaborative effort would probably facilitate the establishment of common guidelines in the surgical treatment of epileptic spasms.