SEARCH

SEARCH BY CITATION

Keywords:

  • Intracerebral EEG;
  • High frequency;
  • Spikes

Abstract

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgments
  7. REFERENCES

Summary: Purpose: High-frequency activity has been recorded with intracerebral microelectrodes in epileptic patients and related to seizure genesis. Our goal was to analyze high-frequency activity recorded with electroencephalograph (EEG) macroelectrodes during the slow wave immediately following interictal spikes, given the potential importance of this presumed hyperpolarization in transforming spikes into seizures.

Methods: Depth electrode EEG recordings from 10 patients with intractable focal epilepsy were low-pass filtered at 500 Hz and sampled at 2,000 Hz. Spikes were categorized according to localization and morphology. Segments of 256 ms were selected immediately following (postspike), and 2 s before each spike (baseline). Power was estimated in subgamma (0–40 Hz), gamma (40–100 Hz), high frequency (100–200 Hz), and very high frequency (250–500 Hz) bands.

Results: Changes in power above 100 Hz were seen in 22 of 29 spike categories, consisting primarily of a widespread decrease in frequencies above 100 Hz. This decrease became spatially more restricted as frequencies increased, and coincided with the localization of largest spikes for the highest frequencies. High-frequency power decreases were prominent in the hippocampus but less common in amygdala and neocortex. High-frequency power increases were observed in the amygdala.

Conclusions: Thus high-frequency EEG activity can be recorded with macroelectrodes in humans and may provide insights on neuronal mechanisms related to human epilepsy. This activity undergoes consistent modifications after EEG spikes. We propose that the reduction in high frequencies reflects a postspike depression in neuronal activity that is more pronounced in the region of spike generation. This depression is almost always seen in hippocampus but less in amygdala.

High-frequency oscillations, ranging from 100 to 250 Hz, and termed ripples, have been described in the hippocampus and entorhinal cortex of normal rodents (1–4). Bragin et al. (5,6) have found similar oscillations in the human hippocampus in recordings obtained through microwires implanted with depth electroencephalograph (EEG) macroelectrodes in epileptic patients with intractable seizures. Although the frequency of these oscillations was lower than in animal studies (80–160 Hz), ripples were found in epileptogenic as well as normal temporal lobes (TLs) and interpreted as representing physiological activity. High-frequency ripples appear associated with memory consolidation (1), while even higher (7) frequency activities greater than 500 Hz, which have been described in the normal neocortex of animals (8–11) and humans (12–14), may be related to sensory information processing (15–17).

High-frequency oscillations in the range of 250–500 Hz have also been reported under pathological conditions such as interictal epileptic discharges. These oscillations, called fast ripples, were recorded during the interictal period in the hippocampus and entorhinal cortex from kainic acid-treated rats with chronic seizures and from epileptic patients, using microelectrodes or microwires (5,18,19). Unlike ripples, fast ripples occurred preferentially in the hippocampus ipsilateral to the seizure onset and less often from the contralateral side. Other features such as localization in different hippocampal structures, relation to sleep, and pathogenesis seem to confirm the distinction between ripples and fast ripples (5,18,20,21). It has been proposed that generators of fast ripples are important in the process of epileptogenesis (22). Electrographic seizure onset is associated with an increase in fast frequencies in the seizure focus of experimental animals, suggesting that fast frequencies could be related to seizure initiation (23–26).

The EEGs of most patients with focal epilepsy show interictal epileptic spikes. They can be followed by slow waves, and experimental studies have shown a robust hyperpolarization following spikes caused by intrinsic and synaptic mechanisms (27–29). Studies with brain stimulation in humans have demonstrated that there is an increased discharge threshold after interictal spikes in patients with TL epilepsy (30,31) and neocortical epilepsies (32). Like seizures, interictal spikes indicate epileptogenesis, but recent evidence suggests that they may have a protective role against seizures (33–35). On the other hand, postspike hyperpolarization is a phenomenon that may also be involved in the transformation of spikes into seizures (36). No firm link has been established between high-frequency oscillations and spikes or the subsequent slow wave. Understanding this relationship should advance our comprehension of the all-important postspike hyperpolarization.

The aim of this study was to address these issues by measuring the changes in power of high-frequency activity in the EEG of epileptic patients using intracerebral depth electrodes. This assumes that high frequencies can be recorded with EEG macroelectrodes, allowing the investigation of all the brain regions evaluated for clinical purposes.

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgments
  7. REFERENCES

Patient selection

Ten consecutive patients (six men, average age 34 years, range 22–49) with medically intractable epilepsy were evaluated with intracranial electrode implantation between September 2004 and May 2005. The Montreal Neurological Institute and Hospital Research Ethics Committee approved this study and informed consent was obtained from each patient.

Stereoelectroencephalographic studies were performed when comprehensive noninvasive presurgical workup yielded inconclusive data: the sites for electrode placement were individualized according to the clinical history, seizure semiology, neuroimaging, and surface electrophysiological investigations. Table 1 summarizes the imaging and EEG findings. Five patients had mesial TL epilepsy and the other five, neocortical epilepsy. This classification was determined by the localization of the SEEG seizure onset and of the MRI abnormalities, independently of spike localization. One patient (patient 5) classified in the group of mesial TL epilepsy had an unclassified neurocutaneous syndrome with multiple white matter signal hyperdensities in the MR scan, she also had right mesial temporal atrophy, and the seizure onset was in the mesial structures of the right TL. Two patients had previous left cortico-amygdalo-hippocampectomy (patients 6 and 7) and one had undergone a perilesional resection (patient 9).

Table 1. Demographic, neuroimaging, and electrophysiological data of 10 patients undergoing implantation studies
PatientAge/ genderMR neuroimagingScalp EEGImplanted electrode locationsInterictal SEEGSeizure onset SEEG
  1. M, male; F, female; T, temporal; AVM, arteriovenous malformation; Ant., anterior; Post., posterior; Inf, inferior; R, right; L, left; Bil., bilateral; A, amygdala; H, hippocampus; AC, anterior cingulate; C, central; Ci, cingulate; Fr, frontal; FC, frontocentral; FT, frontotemporal; I, insula; MFG, middle frontal gyrus; MH, middle hippocampus/parahippocampus; PM, premotor; T, temporal; TP, temporoparietal; TPo, temporal pole; OF, orbitofrontal; OP, operculum; PO, parietooccipital.

 134 FMR and interictal FDG PET, normalInterictal: RT Ictal: RTDepth (5): RA, RH, RMH, RAC, ROFRA, RH, RMHR mesial T
 249 FR mesial T atrophyInterictal: Bil. T Ictal: LTDepth (6): LA, LH, LMH, RA, RH, RMHLA, LH, LMH RA, RH, RMHDiffuse bil. mesial T and T neocortex
 345 MHypertrophic LT poleInterictal: Bil. T Ictal: LTDepth (5): LTPo, LA, LH, LMH, LOFLTPo, LA, LH, LMHL mesial T
 429 MBil. mesial T atrophy, R > LInterictal: Bil. T Ictal: LDepth (6): LA, LH, RA, RH, RMH, RILA, LH RA, RH, RMHR MH
 542 FR mesial T atrophy, R centrum semi-ovale small white matter hypersignals, L T arachnoid cystInterictal: Bil. T Ictal: Bil. TDepth (6): LA, LH, LMH, RA, RH, RMHLA, LH, LMH RA, RH, RMHR mesial T, rare L mesial T
 622 ML P and ant. T surgical cavities with gliosis, L atrium cavernous angiomaInterictal: Bil. T Ictal: Bil. TDepth (2): LH, LMHLHEEG changes after clinical seizure onset
 725 MNeurocysticercosis, L post. T gliosis and cystInterictal: LT Ictal: diffuseDepth (3): LTP, LOF, LCiLTP, LOFL TP
 836 MFocal cortical dysplasia, L OP and MFG (confirmed on pathology)Interictal: Bil. Ictal: diffuseDepth (6): LAC, LOF, LOP, LMFG, RAC, ROFLOP, LMFGL Inf Fr convexity
 936 MLarge L CP porencephalic cyst with surrounding gliosisInterictal: L F C Ictal: L FCDepth (7): LTP, LA, LOF, LOP, LAC, LPM, LPOLOF, LOP, LPOL MFG and ant. cingulate gyrus
1028 FSuspected R ant. I focal cortical dysplasia, Ictal SPECT: R Fr pole hyperperfusionInterictal: Bil. FT Ictal: ArtifactsDepth (7): LAC, LPC, LOF, RAC, RPC, ROF, RIROFR pars orbitalis

Recording methods

Stereoelectroencephalographs (SEEGs) were acquired using the Harmonie long-term monitoring system (Stellate, Montreal, Canada). The implantation method consisted of a combination of intracerebral depth and cortical surface electrodes according to the methods described by Olivier et al. (37). Electrode bundles were implanted stereotactically using an image-guidance system (SSN Neuronavigation System, Mississauga, Ontario, Canada) through percutaneous holes drilled in the skull.

Intracranial depth electrodes (electrode bundles) were manufactured on site by wrapping 3/1,000 inch (0.076 mm) stainless steel wire around a 10/1,000 inch (0.254 mm) stainless steel central core. These wires were coated with Teflon except for regions where the insulation was stripped to form electrode contacts. In total, there were nine contacts on each electrode bundle that were spaced along the length of the core wire at 5-mm intervals. The deepest contact (contact 1) was made from the tip of the core wire and had an uninsulated length of 1 mm, while more superficial contacts (contacts 2–9) were formed from stripped sections of the marginal wire that was tightly wound to make 0.5-mm long coils. The effective surface area for each of the eight superficial contacts was 0.80 mm2, and was 0.85 mm2 for the single deep contact. Impedances are measured once immediately after implantation; they usually vary between 10 and 20 kohms. Table 1 includes the localization of the electrode bundles in each patient.

The SEEG was low-pass filtered at 500 Hz and sampled at 2,000 Hz. The analysis was performed in a bipolar montage. Each channel compared two adjacent contacts of the same electrode bundle.

Interictal epileptic spike selection and categorization

SEEG recordings were visually reviewed to detect and classify the interictal spikes, which were categorized according to their localization and morphology. A category was defined as a set of electrographic events with the same spatial distribution (e.g., involving the left hippocampus and minimally the left amygdala) and the same morphology (e.g., large negative amplitude in the deepest hippocampal contact, decreasing in more lateral contacts). A maximum of 5 categories were retained per patient, keeping only the categories with the most frequent spikes. EEGs were reviewed until 60 spikes were found in each category. Only spikes with interspike intervals greater than 2 s were marked. Postspike segments of 256 ms were selected immediately following each marked spike, starting at the first inflection point following the spike peak. Baseline segments of 256 ms were selected 2 s before each marked spike for statistical comparison (Fig. 1A). When the interval between two spikes was shorter than 4 s, the baseline segment was selected in the first interspike interval of 4 or more seconds before the sample (Figs. 1B and 1C). It may be noticed that we are analyzing the EEG immediately after the spike, a time that is different from the time during which ripples and fast ripples have been mostly recorded (5,6,18,20,21).

image

Figure 1. Baseline segments selection criteria. (A) Interspike interval >4 s: b1 and b2 are the baseline segments corresponding to sp1 and sp2 respectively and are located at least 2 s before the corresponding marked spikes. (B) Interspike interval between 2 s and 4 s: the interval between sp1 and sp2 is less than 4 s; therefore, b2 is located in the interspike interval with a minimum duration of 4 s and closest to sp2 but prior to it. Consequently, in this example b2 is in the same interspike segment as b1. (C) Interspike interval between 2 s and 4 s in a burst of spikes: The spikes in the burst that were not selected (interspike interval <2 s) have been crossed-out. The baseline of sp1 (interspike interval <4 s) is located in the interspike interval (>4 s) before the burst. (D) Selection of the second-baseline segment when interspike intervals >4 s: the second-baseline segments (sb1 and sb2) are located before baseline segments b1 and b2, respectively, in the corresponding intervals. (E) Selection of the second-baseline segment when interspike intervals ∼4 s: the crossed-out second-baseline segment is not valid because it is located less than 2 s after sp1; as a result, sb2 is located in the interspike interval (>4 s) before sp1. Notice that sb1 is before b1 and sb2 is before b2, however, sb2 can also be after b1. (F) Selection of the second-baseline segment when interspike interval is between 2 s and 4 s: Similar to (E), in this case sb1, sb2, and b1 are also located in the same interspike interval. However, as the interspike interval between sp1 and sp2 is less than 4 s, b2 is located not in this interval but the same interval as sb1, sb2, and b1.

Download figure to PowerPoint

To determine whether our sampling of the baseline was stable and representative, we compared each baseline segment to a neighboring baseline segment, with the assumption that there should be no statistical difference between these two samples. For this verification, we used only 8 spike categories from 6 patients, assuming that these were representative of all the spike categories. A new baseline segment (second-baseline segment) was selected to correspond to each baseline segment. Each second-baseline segment was located before its matched baseline segment (Fig. 1D). The new baseline segments were at least 2 s away from any spike (Figs. 1E and 1F). A repeated measurement t test analysis was performed to compare baselines and second baselines.

Measurement of frequency band energy by spectral analysis

The power spectral density was estimated in each segment (baseline and postspike) in 4 frequency bands: subgamma (0–40Hz), gamma (40–100 Hz), high frequency (HF, 100–200 Hz), and very high frequency (VHF, 250–500 Hz). The power spectral density was determined using the fast Fourier transform (FFT) function in MATLAB (The Mathworks, Natick, MA, U.S.A.). A frequency resolution of 7.8 Hz was attained using a 256 points FFT epoch, given the 2,000 Hz sampling rate (FFT on 0.128 s, cosine tapering window, 50% overlapping epochs). The signal power was converted to log-scale in order to obtain a more Gaussian distribution (38).

Power changes between postspike periods and baselines were assessed for each channel in the four frequency bands using Student's t test for paired samples. The same test was used to compare baseline and second-baseline segments. The level of significance was set at p < 0.01. All power changes reported below passed this level of significance. Statistical analyses were carried out with the SPSS program (SPSS, Chicago, IL, U.S.A.).

Power changes, increases or decreases, were considered present in one electrode bundle when there were significant differences in one or more bipolar channels from that bundle, in one or more frequency bands, with the following exception: electrode bundles with differences limited to one frequency band in one channel or in two distant channels were not considered as having significant changes since such differences were sometimes found when comparing baseline and second-baseline segments (see Results below and Fig. 2A). Therefore, significant differences in power had to affect at least two neighboring channels (Fig. 2B) or more than one frequency band in one channel (Fig. 2C) to be classified as power changes in a bundle. The most common situation, with changes in multiple channels and multiple bands, is illustrated on Fig. 2D. Once the bundles with power changes were identified, the specific frequency bands showing changes in this bundle were defined by the change in at least one channel (for instance, in electrode LH of Fig. 4A, all bands were considered to show a change since each band shows a change in at least one channel). Finally, it was considered that there were power changes in one spike category when there were changes following that spike in at least one of the electrode bundles.

image

Figure 2. Evaluation of changes in electrodes. (A) electrode with significant changes in the HF band in two distant channels. The change was not considered significant for that bundle, even though it affected two channels, because they were not neighboring channels (RCi1-2 and RCi5-6). (B) bundle with significant changes in the subgamma band in two neighboring channels (LA2-3 and LA3-4). The change was considered significant for that bundle; (C) bundle with changes in two bands in the same channel. The changes were considered significant because they were present in two bands although it was only in one channel (LOP1-2); (D) bundle with changes in more than one band and more than two channels. This type of change was the most frequent. The change was considered significant because it was present in three electrodes (LMH1-2, LMH2-3, and LMH3-4) and affected the four bands. RCi: right cingular cortex; LA, left amygdala; LOP, left operculum (anterior margin of the cyst): LMH, left middle hippocampus/parahippocampus. *p < 0.01

Download figure to PowerPoint

image

Figure 4. Topographic distribution of the changes in the different bands. (A) temporal spike. Spike located mainly in the inner channels of hippocampal and parahippocampal bundles with the clearest spike in LH1-2 (patient #5). Changes in high frequencies are restricted to these two bundles while changes in the subgamma band affect also the amygdalar bundle. Changes in gamma band are present in channels with spike (LH1-2, LH2-3, LH3-4, LMH1-2, and LMH2-3). Changes in HF band are present in the same channels than gamma band but not in LH3-4 (a channel with a lower amplitude spike). Changes in VHF are restricted to the channel with the clearest spike (LH1-2); (B) neocortical spike. Spike located in the LOF bundle with the clearest spike in channel LOF7-8 (patient #7). Changes in high frequencies are limited to this bundle, while changes in the subgamma band are also present in the LC bundle. Changes in the gamma band are present in five neighboring channels including the channel with the clearest spike. Changes in the HF band are present in three channels including the channel with the clearest spike (LOF7-8) and the surrounding channels (LOF6-7 and LOF8-9). Changes in VHF band are restricted to the channel with the clearest spike. LA, left amygdala; LH, left hippocampus; LMH, left middle hippocampus/parahippocampus; LOF, left orbito-frontal; LCi, left cingulate. *p < 0.01

Download figure to PowerPoint

RESULTS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgments
  7. REFERENCES

Twenty-nine spike categories were identified: 19 in patients with mesial TL epilepsy (patients 1–5) and 10 in patients with neocortical epilepsy (patients 6–10). The most frequent difference between spike categories of the same patient was the localization. However, in four patients, spikes of the same localization but different polarity (three patients) or morphology (one patient) were identified as different categories. In another two patients, spikes with similar morphology but different spatial spread were placed in different categories. The 19 spike categories of patients with mesial TL epilepsy were located in the mesial TL; 2 of the 10 categories from patients with neocortical epilepsy were also located in the mesial TL; the other 8 were located in the neocortex. Activity in the HF and VHF bands could be identified in the SEEG; the changes were not artifactual since they were orderly and not random, as demonstrated below. We concluded that activity in these frequency bands could be recorded from EEG macroelectrodes, and thus we proceeded with examination of the postspike changes.

To assess baseline stability, we evaluated the significance of the differences between the pairs of baselines (8 spike categories) in the four frequency bands for all the channels. In total we had 1,096 p values, 16, or 1.5%, of which were significant (p < 0.01). This percentage is similar to the 1% expected by chance from the 0.01 level of significance. There was neither consistent pattern in the frequency bands showing significant differences nor in the spatial distribution of the differences. We therefore concluded that baseline samples were stable and representative.

Figure 3 shows an example of baseline and postspike segments, with different timescales and filtering. In this case, a reduction in HF and VHF is apparent following the spike. In many cases, the changes were not obvious by visual inspection but were revealed by the statistical comparison of the baseline with the postspike segments. In the postspike segments, changes in HF or VHF bands were measured in 76% of spike categories. Fig. 4 shows two examples of the most frequent pattern (4A: mesial temporal spike category; 4B: neocortical category). There were two main findings: (i) the spatial extent of the power changes showed a gradual reduction as frequencies got higher, with widespread changes in subgamma and a restricted distribution in VHF. Changes in VHF coincided with the distribution of the largest spikes, while changes in the subgamma band could be more extensive than the spike (Fig. 4); and (ii) in most spike categories, the direction of the change varied across the different frequency bands, with a clear-cut separation between the subgamma band and frequencies above 40 Hz (gamma, HF and VHF bands). In the subgamma band, the power always increased, but in 23 of 29 spike categories, the power in frequencies higher than 40 Hz decreased. An increase in all frequency bands was rare, except in the amygdala. We explain below the details of the presence of changes in the different frequency bands, and their spatial localization and direction. We will also address the relationships between these changes and the region of seizure onset.

image

Figure 3. Changes in high-frequency activity after spikes. (A) unfiltered EEG section showing one spike. The two highlighted portions represent the baseline and postspike segments; (B) the same EEG section with high-pass filtering at 80 Hz; (C) the baseline (left) and postspike (right) segments are shown expanded, high-pass filtered at 80 Hz (top) or 250 Hz (bottom). The decrease in high frequencies is apparent immediately after the spike in panel B. The expanded segments in panel C show that the decrease includes the HF (top) and VHF (bottom) bands, being maximal immediately following the spike.

Download figure to PowerPoint

Presence of power change

There were significant power changes in at least one frequency band in all 29 spike categories: 22 (76%) in high frequencies (HF and VHF), 26 (90%) in the gamma band, and 28 categories (96%) in the subgamma band. In 16 categories, the changes affected the four bands; 5 had changes in HF, gamma, and subgamma bands, but not in the VHF band; 4 showed changes restricted to the gamma and subgamma bands; 3 had changes only in the subgamma band. Finally, in one category, there was a change in all the bands except subgamma. For each bundle, the changes in a frequency band were almost always associated with changes in all the lower frequency bands: for example, a change in HF was accompanied by one in gamma and subgamma, or a change in VHF was accompanied by one in HF, gamma, and subgamma (Figs. 4–7). Only four exceptions to this rule (4 bundles) were found in the 97 analyzed bundles (although only 53 bundles were implanted, some of them were analyzed more than once for patients having more than one spike category).

image

Figure 5. Similar changes in amygdala and hippocampus in the postspike segment of a temporal lobe spike category (patient #2). Channel RP12 shows changes in all the frequency bands; channels RA1-2, RA2-3, RH1-2, RH2-3, RH3-4 show changes in HF, gamma, and subgamma bands; RH4-5, RMH2-3, and RMH3-4 show changes in gamma and subgamma bands; channels RA3-4 and RA4-5 show changes only in subgamma band. Changes in VHF, HF, and gamma are decreases and changes in subgamma band are increases. RA, right amygdala; RH, right hippocampus; RMH, right middle hippocampus/parahippocampus. *p < 0.01

Download figure to PowerPoint

image

Figure 6. Different changes in amygdala and hippocampus in the postspike segment of a temporal lobe spike category (patient 5). Channel RA12 shows an increase in all the frequency bands. The rest of the significant changes consist in a decrease in high frequencies and an increase in lower frequencies. Channel RMH1-2 shows changes in all the frequency bands; channels RH1-2, RH2-3, and RMH2-3 show changes in HF, gamma, and subgamma bands; RMH3-4 shows changes in gamma and subgamma bands; channels RA4-5, RH3-4, and RH4-5 show changes only in subgamma band. RA, right amygdala; RH, right hippocampus; RMH, right middle hippocampus/parahippocampus. *p < 0.01

Download figure to PowerPoint

image

Figure 7. Power increase in the postspike segment of a neocortical spike (patient 8). Channel LOP2-3 shows increase in HF, gamma, and subgamma bands; LOP1-2, LOP3-4, and LOP4-5 show changes in gamma and subgamma bands; channels LOP5-6, LOP6-7, LOP7-8, and LOP8-9 show changes only in subgamma band. Increases in VHF and HF in LOP1-2 are not significant. LOP, left operculum; LMFG, left middle frontal gyrus. *p < 0.01

Download figure to PowerPoint

Spatial distribution of the power changes in the different frequency bands

Since the localization of changes in high-frequency activity has been related to the epileptic focus, we compared the localization and extent of changes between the four frequency bands. They spanned a greater region in lower frequencies than in higher frequencies. From 16 categories with changes in all four frequency bands, 11 showed a gradual extension of the spatial distribution of change from VHF to subgamma (Figs. 4–7). In another 4 categories among the 16 with changes in the four bands, the extent of changes was highest in the subgamma band, but without clear differences among gamma, HF, and VHF. In the last category with changes in all four bands, the changes in the subgamma band were less extensive than those in frequencies higher than 40 Hz; nevertheless, the VHF, HF, and gamma bands still showed a gradual increase in spatial extent.

In four of the five categories with changes in all but the VHF band, there was again a gradual extension of the spatial distribution of change from HF to subgamma. In the remaining category, the extent of the changes in the gamma band was larger than that of HF, but the changes in the subgamma band were more restricted spatially. The gradation in the spatial spread was present between gamma, HF, and VHF in the one category without changes in the subgamma band.

Changes in frequencies higher than 40 Hz were always seen in bundles that showed a spike in at least one channel, and such channels were close to the spike. Changes in the subgamma band (in the 28 spike categories with changes in this band) were present in all but one of the bundles showing spikes, but were also seen in bundles that did not show spikes (Figs. 4A and 4B). For example, in spike categories of a patient with mesial TL epilepsy, the changes in frequencies higher than 40 Hz were in the amygdalar and hippocampal bundles while those in the subgamma band could include cingulate and orbitofrontal bundles as well. The higher the frequency band, the more likely the changes were restricted to the channels containing the spike. The channels with VHF changes were always the channels with the largest spikes. The channels with HF changes were either in the electrode with the highest spike or in electrodes immediately adjacent. Changes in the HF and gamma bands were seen in a progressively higher number of neighboring channels (Figs. 4–7). In only four bundles (four spike categories), changes in frequencies higher than 40 Hz were outside the spiking region.

Power decrease versus power increase

The power in the subgamma band always increased, but a decrease was the most frequent pattern in frequencies higher than 40 Hz: 23 of 29 spike categories had a decrease in frequencies higher than 40 Hz in at least one bundle, and in 20 the decrease included the HF and VHF bands. We studied the patterns of change (decrease vs. increase) in the different frequency bands in individual channels, in different channels of the same bundle, and in different bundles of the same spike category.

Within individual channels, the patterns of change showed a clear difference between frequencies higher and lower than 40 Hz. In the gamma, HF and VHF bands, they showed the same direction (decrease or increase) in 38 of the 40 channels in which they were present in these three bands. In the other two, the power in the gamma band increased and the power in the HF and VHF bands decreased. The HF and VHF bands always changed in the same direction. Changes in the subgamma band could have the same direction as that of the three high-frequency bands (bundle RA in Fig. 6 and bundle LOP in Fig. 7) or the opposite direction (Fig. 4 and 5, and bundles RH and RMH in Fig. 6).

Within the same bundle, the direction of change in each band was the same in all its channels in 71 of 72 bundles showing changes (Figs. 4–7). If there was a power decrease in one frequency band, all the channels of the bundle with changes in this band showed a decrease. If the change was an increase, all the channels of the bundle with changes in this band showed an increase.

Table 2 summarizes the direction of the changes in all the spike categories. In the hippocampus, the vast majority of spikes showed a decrease in power in HF or VHF, whereas increases and decreases were more evenly distributed in neocortex and amygdala. Fig. 5 illustrates a decrease in hippocampus and amygdala, and Fig. 6 illustrates a decrease in the hippocampus and an increase in the amygdala. Fig. 7 shows an increase in the neocortex.

Table 2. Direction of the changes in all the spike categories
 AmygdalaHippocampusNeocortex
  1. aNumber of spike categories having a spike in the amygdala, hippocampus, and neocortex.

N of spike categoriesa19218
Power decrease6 (32%)17 (81%)3 (37.5%)
Power increase5 (26%)1 (5%)1 (12.5%)
No change8 (42%) 3 (14%)4 (50%)  

In summary, the most frequent change seen in the VHF, HF, and gamma bands was a power decrease, and in the subgamma band, a power increase. The direction of change was similar across the VHF, HF, and gamma bands and across the different channels of the same bundle. Power decrease in HF and VHF was seen in the hippocampus in 81% of the 21 temporal spike categories (15 categories from patients with mesial TL and 2 from one patient with neocortical epilepsy), in the amygdala in 31% of temporal categories, and in neocortex in 37.5% of neocortical categories. Increase in HF and VHF was seen in the hippocampus in 5% of temporal spike categories, in the amygdala in 26% of temporal categories, and in neocortex in 12.5% of actual neocortical categories. The remaining 7 categories showed changes only below 100 Hz or below 40 Hz.

Relation between localization of power changes and seizure onset

This relation could only be studied in patients for whom there were spike categories in channels involved in seizure onset and in channels not involved in seizure onset, thus allowing a comparison. In patients with mesial TL epilepsy, all spike categories were recorded in contacts involved in seizure onset, and the comparison was therefore impossible. In patients with neocortical epilepsy, changes in high frequencies were seen in channels involved in the onset of seizures with good clinical-EEG correlation (three of four spike categories), in channels that showed the first ictal changes in seizures with poor clinical-EEG correlation (one of three categories), and in channels not involved in the ictal onset (one of three categories). Therefore, it was not possible to find any relationship between changes in high frequencies and the region of seizure onset.

DISCUSSION

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgments
  7. REFERENCES

In this study, we demonstrate a decrease in the power spectral density of high-frequency bands (>100 Hz) during the postspike slow wave. Importantly, this variation in high-frequency activity could be detected with EEG macro intracerebral electrodes, which, in contrast to the limited coverage of microwires (18), allowed us to evaluate and compare several recording sites in different brain regions. We were able to identify changes in high frequencies compared with the baseline following a high proportion of spikes (76% of spike categories) in patients with mesial TL and with neocortical epilepsy. However, this high-frequency activity should not be equated with the high-frequency oscillations (ripples and fast ripples) described in the hippocampus and entorhinal cortex of epileptic humans and of nonepileptic and epileptic rodents. We studied power changes in different frequency bands immediately after a spike, while ripples and fast ripples are short-lasting oscillatory events associated with EEG transients (6,21). We nevertheless analyzed separately the frequency bands corresponding to these two high-frequency oscillations (HF and VHF bands), since a considerable body of work has resulted in the separation into different pathophysiologic mechanisms of ripples and fast ripples according to frequency.

The most common pattern of postspike change is represented schematically in Fig. 8A. It consists of (i) an increase in subgamma activity that includes the region of spike generation but often extends beyond it; and (ii) a decrease in higher frequencies, with a gradual shrinking of the space involved as the frequency band gets higher. In addition, the decrease in the VHF band is limited to the regions of largest spike amplitude.

image

Figure 8. Schematic representation of the most frequent patterns of changes. The illustration shows the region of the temporal lobe but similar results were found in other regions. Each rectangle represents one electrode bundle (top, amygdala; middle, hippocampus; lower, posterior hippocampus, or parahippocampus). The changes in different frequency bands are represented by sinusoids at different frequencies (see box on figure). The lengths of the sinusoids indicate the spatial extent of the changes in the bundle. The green shaded region represents the extent of the spike in each bundle. Darker green areas indicate the locations where the spikes are most prominent. (A) Widespread increase in subgamma power represented by the long red sinusoid and spatially graded decrease in gamma, HF, and VHF, represented by the progressively shorter blue sinusoids. The VHF decrease is only present in the region where the spike is largest (dark green shading); (B) A different pattern was often seen in the temporal lobe, with a decrease in high frequencies in the hippocampus but an increase in the amygdala.

Download figure to PowerPoint

This increase in subgamma activity following an epileptic spike may result from the prolonged depression in excitability that characterizes this phase, and coincided with the slow wave seen in the EEG after the spike (39). Experimental studies have shown that the paroxysmal depolarization shift associated with an epileptiform EEG spike is followed by a decrease in excitability that may result from a combination of intrinsic and synaptic currents (27–29). Early work performed in the penicillin-induced epileptic focus has demonstrated that this inhibition is widely spread around the focus (36,40).

Similarly, we hypothesize that the decrease in gamma, HF, and VHF activity represents the suppression of short postsynaptic potentials, and of neuronal firing, as a result of the widespread inhibition. The observed spatial gradation in frequencies could reflect the most profound inhibition nearest the spike. It is not clear, however, why the shortest postsynaptic potentials should be the most resistant to inhibition. Since changes in HF and VHF always occurred in parallel, our results do not provide evidence for a separate role for these two bands in the immediate postspike period.

The frequent pattern described earlier was not only the most common, but was also almost always present in the hippocampus. The amygdala showed sometimes the same pattern, but in a significant number of cases, it showed either no change or even an increase in high frequencies following spikes (Fig. 8B). High-frequency oscillations have often been recorded in the hippocampus and entorhinal cortex (1,9,18,41,42), but they have been less often studied in the amygdala (7). The absence of a decrease in high frequencies following the spike suggests that the depression in excitability occurring in the amygdala during the postspike period is not as effective at suppressing fast postsynaptic potentials as it is in the hippocampus. If high frequencies represent a mechanism for seizure initiation, then our finding supports the view that the amygdala is characterized by high epileptogenic properties as in the in vivo kindling model (43,44) and in vitro preparations (45).

Our study also demonstrated that high-frequency changes were seen following spikes in neocortical regions, but again less often than in the hippocampus. This finding suggests that the same phenomena may be at play in all epileptogenic regions, with regional variations. Neocortical regions very rarely showed increases in high frequencies, such as those seen in the amygdala, possibly reflecting the lesser epileptogenicity of neocortex compared to amygdala, as indicated by clinical investigations (46,47) and by experimental studies made in the kindling model (43,44).

The changes in high-frequency activity were observed over relatively wide regions: commonly over several contacts and several bundles for HF activity and one or two contacts for VHF activity (contacts are 5 mm apart). We have not examined whether the waveforms are synchronous over the regions in which we see those changes, or whether they are time-locked to the spike. They are not very likely to be synchronous since high-frequency activity has been reported to be synchronous only over a very small volume (9,48–50). It is more probable that this activity changes simultaneously over large regions, as a result of the widespread hyperpolarization discussed earlier.

We could not establish a link between postspike changes and the region of seizure onset, probably because of the lack of diversity in our group. It would be necessary to study a larger and more diverse group of patients to assess this relationship.

In conclusion, we have demonstrated that EEG activity between 100 and 500 Hz can be recorded in epileptic humans with intracerebral macroelectrodes, extending the potential for future analysis of these discrete phenomena. Our findings also identify, for the first time, changes in high and very high frequencies that suggest profound and site-specific modifications in neuronal excitability following interictal epileptic spikes. These data, which are in line with experimental evidence obtained with microelectrode recordings from epileptic rodents, may shed light on the pathophysiology of epileptiform discharges by directly addressing the underlying mechanisms in the human brain.

Acknowledgments

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgments
  7. REFERENCES

Acknowledgment:  This project was supported by the Canadian Institutes of Health Research grant MOP-10189. Elena Urrestarazu was supported by a scholarship for research of the Department of Education of the Basque Government.

REFERENCES

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgments
  7. REFERENCES
  • 1
    Draguhn A, Traub RD, Bibbig A, et al. Ripple (approximately 200-Hz) oscillations in temporal structures. J Clin Neurophysiol 2000;17:36176.
  • 2
    Chrobak JJ, Lorincz A, Buzsaki G. Physiological patterns in the hippocampo-entorhinal cortex system. Hippocampus 2000;10:45765.
  • 3
    Chrobak JJ Buzsaki G. High-frequency oscillations in the output networks of the hippocampal-entorhinal axis of the freely behaving rat. J Neurosci 1996;16:305666.
  • 4
    Buhl DL, Buzsaki G. Developmental emergence of hippocampal fast-field “ripple” oscillations in the behaving rat pups. Neuroscience 2005;134:142330.
  • 5
    Bragin A, Engel J Jr., Wilson CL, et al. High-frequency oscillations in human brain. Hippocampus 1999;9:13742.
  • 6
    Staba RJ, Wilson CL, Bragin A, et al. Quantitative analysis of high-frequency oscillations (80-500 Hz) recorded in human epileptic hippocampus and entorhinal cortex. J Neurophysiol 2002;88:174352.
  • 7
    Ponomarenko AA, Korotkova TM, Haas HL. High frequency (200 Hz) oscillations and firing patterns in the basolateral amygdala and dorsal endopiriform nucleus of the behaving rat. Behav Brain Res 2003;141:1239.
  • 8
    Ikeda H, Wang Y, Okada YC. Origins of the somatic N20 and high-frequency oscillations evoked by trigeminal stimulation in the piglets. Clin Neurophysiol 2005;116:82741.
  • 9
    Grenier F, Timofeev I, Steriade M. Focal synchronization of ripples (80-200 Hz) in neocortex and their neuronal correlates. J Neurophysiol 2001;86:188498.
  • 10
    Jones MS, MacDonald KD, Choi B, et al. Intracellular correlates of fast (>200 Hz) electrical oscillations in rat somatosensory cortex. J Neurophysiol 2000;84:150518.
  • 11
    Shimazu H, Kaji R, Tsujimoto T, et al. High-frequency SEP components generated in the somatosensory cortex of the monkey. Neuroreport 2000;11:28216.
  • 12
    De Curio G, Mackert BM, Burghoff M, et al. Somatotopic source arrangement of 600 Hz oscillatory magnetic fields at the human primary somatosensory hand cortex. Neurosci Lett 1997;234:1314.
  • 13
    Inoue K, Harada T, Kaseda Y, et al. Effects of movement on somatosensory N20m fields and high-frequency oscillations. Neuroreport 2002;13:186164.
  • 14
    Curio G, Mackert BM, Burghoff M, et al. Localization of evoked neuromagnetic 600 Hz activity in the cerebral somatosensory system. Electroencephalogr Clin Neurophysiol 1994;91:4837.
  • 15
    Hashimoto I. High-frequency oscillations of somatosensory evoked potentials and fields. J Clin Neurophysiol 2000;17:30920.
  • 16
    Gobbele R, Waberski TD, Simon H, et al. Different origins of low–and high-frequency components (600 Hz) of human somatosensory evoked potentials. Clin Neurophysiol 2004;115:92737.
  • 17
    Curio G. Linking 600-Hz “spikelike” EEG/MEG wavelets (“sigma-bursts”) to cellular substrates: concepts and caveats. J Clin Neurophysiol 2000;17:37796.
  • 18
    Bragin A, Mody I, Wilson CL, et al. Local generation of fast ripples in epileptic brain. J Neurosci 2002;22:201221.
  • 19
    Bragin A, Engel J Jr., Wilson CL, et al. Hippocampal and entorhinal cortex high-frequency oscillations (100–500 Hz) in human epileptic brain and in kainic acid–treated rats with chronic seizures. Epilepsia 1999;40:12737.
  • 20
    Staba RJ, Wilson CL, Bragin A, et al. High-frequency oscillations recorded in human medial temporal lobe during sleep. Ann Neurol 2004;56:10815.
  • 21
    Bragin A, Wilson CL, Staba RJ, et al. Interictal high-frequency oscillations (80-500 Hz) in the human epileptic brain: entorhinal cortex. Ann Neurol 2002;52:40715.
  • 22
    Engel J Jr., Wilson C, Bragin A. Advances in understanding the process of epileptogenesis based on patient material: what can the patient tell us? Epilepsia 2003;44(suppl 12):6071.
  • 23
    Lasztoczi B, Antal K, Nyikos L, et al. High-frequency synaptic input contributes to seizure initiation in the low-[Mg2+] model of epilepsy. Eur J Neurosci 2004;19:136172.
  • 24
    Grenier F, Timofeev I, Steriade M. Neocortical very fast oscillations (ripples, 80-200 Hz) during seizures: intracellular correlates. J Neurophysiol 2003;89:84152.
  • 25
    Traub RD, Whittington MA, Buhl EH, et al. A possible role for gap junctions in generation of very fast EEG oscillations preceding the onset of, and perhaps initiating, seizures. Epilepsia 2001;42:15370.
  • 26
    Traub RD, Michelson-Law H, Bibbig AE, et al. Gap junctions, fast oscillations and the initiation of seizures. Adv Exp Med Biol 2004;548:11022.
  • 27
    Gutnick MJ, Connors BW and Prince DA. Mechanisms of neocortical epileptogenesis in vitro. J Neurophysiol 1982;48:132135.
  • 28
    Hwa GG, Avoli M, Oliver A, Villemure JG. Bicuculline-induced epileptogenesis in the human neocortex maintained in vitro. Exp Brain Res 1991;83:32939.
  • 29
    Lebovitz RM. Autorhythmicity of spontaneous interictal spike discharge at hippocampal penicillin foci. Brain Res 1979;172:3555.
  • 30
    Wilson CL, Khan SU, Engel J Jr., et al. Paired pulse suppression and facilitation in human epileptogenic hippocampal formation. Epilepsy Res 1998;31:21130.
  • 31
    Wilson CL and Engel J Jr. Electrical stimulation of the human epileptic limbic cortex. Adv Neurol 1993;63:10313.
  • 32
    De Curtis M, Tassi L, Lo RG, et al. Increased discharge threshold after an interictal spike in human focal epilepsy. Eur J Neurosci 2005;22:297176.
  • 33
    Barbarosie M and Avoli M. CA3-driven hippocampal-entorhinal loop controls rather than sustains in vitro limbic seizures. J Neurosci 1997;17:930814.
  • 34
    De Curtis M, Librizzi L, Biella G. Discharge threshold is enhanced for several seconds after a single interictal spike in a model of focal epileptogenesis. Eur J Neurosci 2001;14:1748.
  • 35
    De Curtis M, Avanzini G. Interictal spikes in focal epileptogenesis. Prog Neurobiol 2001;63:54167.
  • 36
    Ayala GF, Dichter M, Gumnit RJ, et al. Genesis of epileptic interictal spikes. New knowledge of cortical feedback systems suggests a neurophysiological explanation of brief paroxysms. Brain Res 1973;52:117.
  • 37
    Olivier A, Germano IM, Cukiert A, et al. Frameless stereotaxy for surgery of the epilepsies: preliminary experience. Technical note. J Neurosurg 1994;81:62933.
  • 38
    Gasser T, Bacher P, Mocks J. Transformations towards the normal distribution of broad band spectral parameters of the EEG. Electroencephalogr Clin Neurophysiol 1982;53:11924.
  • 39
    Neckelmann D, Amzica F, Steriade M. Changes in neuronal conductance during different components of cortically generated spike-wave seizures. Neuroscience 2000;96:47585.
  • 40
    Prince DA. Inhibition in “epileptic” neurons. Exp Neurol 1968;21:30721.
  • 41
    Ylinen A, Bragin A, Nadasdy Z, et al. Sharp wave-associated high-frequency oscillation (200 Hz) in the intact hippocampus: network and intracellular mechanisms. J Neurosci 1995;15(1 Pt 1):3046.
  • 42
    Dzhala VI, Staley KJ. Mechanisms of fast ripples in the hippocampus. J Neurosci 2004;24:8896906.
  • 43
    McIntyre DC, Poulter MO, Gilby K. Kindling: some old and some new. Epilepsy Res 2002;50:7992.
  • 44
    McIntyre DC. The kindling phenomenon. In: PitkanenA, SchwartzkroinPA, MoshéSL, eds. Models of seizures and epilepsy. Amsterdam : Elsevier Academic Press, 2006.
  • 45
    Benini R, D'Antuono M, Pralong E, et al. Involvement of amygdala networks in epileptiform synchronization in vitro. Neuroscience 2003;120:7584.
  • 46
    Feindel W, Rasmussen T. Temporal lobectomy with amygdalectomy and minimal hippocampal resection: review of 100 cases. Can J Neurol Sci 1991;18(suppl 4):6035.
  • 47
    Fish DR, Gloor P, Quesney FL, et al. Clinical responses to electrical brain stimulation of the temporal and frontal lobes in patients with epilepsy. Pathophysiological implications. Brain 1993; 116(Pt 2):397414.
  • 48
    D'Antuono M, De Guzman P, Kano T, et al. Ripple activity in the dentate gyrus of dishinibited hippocampus-entorhinal cortex slices. J Neurosci Res 2005;80:92103.
  • 49
    Barth DS. Submillisecond synchronization of fast electrical oscillations in neocortex. J Neurosci 2003;23:250210.
  • 50
    Maier N, Nimmrich V, Draguhn A. Cellular and network mechanisms underlying spontaneous sharp wave-ripple complexes in mouse hippocampal slices. J Physiol 2003; 550(Pt 3):87387.