Skip to main content
NCBI home page
PMC Canada Author Manuscripts logoLink to PMC Canada Author Manuscripts
. Author manuscript; available in PMC: 2013 Sep 5.
Published in final edited form as: Epilepsia. 2009 Mar 27;50(7):1780–1792. doi: 10.1111/j.1528-1167.2009.02067.x

High frequency oscillations (80–500 Hz) in the preictal period in patients with focal seizures

Julia Jacobs 1, Rina Zelmann 1, Jeffrey Jirsch 1, Rahul Chander 1, Claude-Édouard Châtillon 1, François Dubeau 1, Jean Gotman 1
PMCID: PMC3764053  CAMSID: CAMS3402  PMID: 19400871

Summary

Purpose

Intracranial depth macroelectrode recordings from patients with focal seizures demonstrate interictal and ictal high frequency oscillations (HFOs, 80–500 Hz). These HFOs are more frequent in the seizure-onset zone (SOZ) and reported to be linked to seizure genesis. We evaluated whether HFO activity changes in a systematic way during the preictal period.

Methods

Fifteen minutes of preictal intracranial electroencephalography (EEG) recordings were evaluated in seven consecutive patients with well-defined SOZ. EEG was filtered at 500 Hz and sampled at 2,000 Hz. Ripples (80–250 Hz) and fast ripples (250–500 Hz) were visually marked, and spectral analysis was performed in seizure-onset as well as nonseizure-onset channels. Linear regressions fitted to the power trends corresponding to intervals of 1, 5, and 15 min before the seizure onset was calculated.

Results

Total rates of HFOs were significantly higher in the SOZ than outside. Preictal increases and decreases in HFO rates and band power could be detected in all patients, and they were not limited to the SOZs. These measures were very variable, and nosystematic trends were observed when comparing patients or seizures in the same patient.

Discussion

High frequencies in the range of 80–500 Hz are present during the preictal period and are more prominent in the SOZ. They do not change in a systematic way before seizure onset for the horizons we tested. The 80–500 Hz band may be used for the localization of seizure-onset areas but may be more difficult to use for seizure prediction purposes.

Keywords: Intracranial EEG, Epilepsy, Ripples, Fast ripples, Seizure prediction

Approximately one-third of epileptic patients have inadequate seizure control through anticonvulsant medications. For these patients the unforeseeable way in which seizures occur represents the most disabling and dangerous aspect of their disease. Even if seizures seem to occur suddenly and without warning in most patients, there has been a longstanding discussion on whether seizures may be building up slowly in the hours or minutes before the clinical event (Le Van Quyen et al., 1999; Lehnertz et al., 1999; Litt et al., 2001). Some patients describe unspecific premonitory symptoms hours before the seizure (Schulze-Bonhage et al., 2006; Haut et al., 2007), and several studies detected electroencephalography (EEG) changes during the preictal period (for review see Mormann et al., 2007). However, the presence and time-frame of consistent EEG changes prior to seizures remain uncertain and no method of seizure prediction can consistently predict seizures in different patients (Schelter et al., 2006; Mormann et al., 2007).

The conventional range of EEG analysis involves frequencies below 100 Hz, but studies over the last decade suggest that localized higher frequencies may also be important. Frequencies above 100 Hz have been extensively characterized in human epileptic mesial temporal structures using depth microelectrodes. Whereas 100–200 Hz oscillations appear related to physiologic memory processing, higher frequencies, between 200 and 500 Hz, are associated with epileptogenic tissue (Bragin et al., 1999). These high frequency oscillations (HFOs) have been termed ripples (80–250 Hz) and fast ripples (250–500 Hz).

Recently depth macroelectrodes and spectral and visual analysis techniques have also revealed focal HFOs in humans during interictal (Urrestarazu et al., 2006, 2007; Jacobs et al., 2008; Worrell et al., 2008) and ictal recordings (Jirsch et al., 2006; Ochi et al., 2007; Ramachandrannair et al., 2008). Discrete HFOs occurred mainly in regions of seizure onset and rarely in regions of secondary spread in mesial temporal as well as neocortical seizures. Moreover, no ictal high frequency activities occurred in the seizures of patients with poor localization (Jirsch et al., 2006).

In animal studies, a clear relationship between the presence of HFOs and their degree of activity with spontaneous seizures could be shown (Bragin et al., 2004). In an in vitro model of low-Mg2+ seizures, an increase of HFOs preceded seizure activity (Khosravani et al., 2005). In rats, an increase of ripple and fast ripple bands could be observed within the dentate gyrus 1 s before seizure onset (Bragin et al., 2005). The behavior of HFOs in the preictal period in epilepsy patients has been evaluated in only one study (Khosravani et al., 2008), which found changes in the seconds preceding seizures. The possibility that HFOs may reflect basic epileptogenic processes and changes during the preictal period bears scientific and clinical interests. For instance, potential therapies involving EEG-triggered anticonvulsant injections or electrical stimulation could result in seizure control. Such antiseizure therapies may have greater chance of success if seizures could be predicted from the interictal EEG (Elger, 2001; Osorio et al., 2005).

We hypothesized that HFOs in the range of 80–500 Hz measured using depth macroelectrodes in patients with intractable temporal and extratemporal epilepsy change in frequency of occurrence during the preictal state. Visual and spectral analyses techniques were used to assess HFO rates and band power in the 15 min prior to seizure onset. These two methods were chosen to detect changes in high frequency power in general but also in distinct HFOs. Distinct HFOs are very short events and, therefore, a change in their rate may not be detectable with spectral analysis.

Methods

Patient selection

Between September 2004 and July 2005, 12 patients underwent intracranial electrode implantation in the epilepsy unit of the Montreal Neurological Hospital. Electrode placement was tailored to the clinical history, seizure semiology, results of surface EEG investigation, neuroimaging, and neuropsychological testing. The same neurologist analyzed all spontaneous seizures—regardless of sampling frequency—recorded during the implantation period, and identified seizure onset and propagation areas for each seizure. For this study, patients with a clearly defined seizure-onset zone (SOZ) were selected, having at least one complete seizure recorded with a 2,000 Hz sampling frequency. Ictal events were excluded if they were not associated with clinical manifestations. The Montreal Neurological Hospital Research Ethics Committee approved this study and written informed consent was obtained from all participants.

EEG recording

Intracranial implantations consisted of combined depth and epidural electrodes placed according to the method of Olivier et al. (1994) with frameless stereotactic image guidance (SSN Neuronavigation System, Mississauga, ON, Canada) through percutaneous and skull twist drill holes. Intracranial depth electrodes were manufactured on site from stainless steel wires, with nine contacts on each electrode and a 5-mm intercontact distance. Details of the electrode construction have been described previously (Jacobs et al., 2008). Effective surface area of each single deep contact was 0.85 mm2, whereas each of the eight more superficial contacts was 0.80 mm2. Depth electrodes were targeted individually toward the presumed SOZ. Common temporal lobe targets were the amygdala/uncus (LA/RA: left/right amygdala), anterior hippocampus (LH/RH), and parahippocampus (LPH/RPH). The exact electrode positioning of all patients are shown in Table 1.

Table 1.

Clinical information

Patients MRI Electrode positions Seizure onset
1 R OF and anterior insular FCD L AC, PC, OF
R AC, PC, OF, Le		 R OF6-8
2 L T atrophy L A, HC, PH, OF, T (temporal pole) L A1-2, HC1-2, PH1-2
3 R HC atrophy L A, HC, PH, 1.: 2.:
R A, HC, PH, T (R angular gyrus) R A1-2, PH1-2 L A1-2, HC1-2, PH1-2
4 L F FCD L C OF, Le, AF (perilesional)
R C, OF,		 L Le 3-5, AF1-3
5 No lesion R A, HC, PH, OF, C R A 1-3, HC1-2, PH 1-2
6 Porencephalic cyst L CP area L A, T (first temporal), OF, AC, Le (in gliosis), F (perilesional), OP L AC 3-5, F 1-3
7 R T atrophy L A, HC,-PH, A, HC, PH R A1-2, HC1-2, PH1-2
Open in a new tab

This table gives the clinical information about MRI findings, electrode positions, and seizure onset.

A, amygdala; AC, anterior cingulate gyrus; AF, anterior frontal; C, cingulate gyrus; CP, centroparietal; F, frontal; FCD, focal cortical dysplasia; HC, hippocampus; L, left; Le, lesion; OF, orbitofrontal; OP, occipitoparietal; PC, posterior cingulate; PH, parahippocampus; R, right; T, temporal.

Patients’ antiepileptic drugs were reduced and intracranial EEG was monitored continuously for spontaneous seizures during typical 2-week implantation periods. EEG telemetry signal was recorded digitally with a 128-channel Harmonie system (Stellate, Montreal, Quebec, Canada) with filter settings of 0.1 and 500 Hz and a sampling frequency of 2,000 Hz for continuous periods lasting 24–96 h. The analyses reported below were performed using a bipolar montage consisting of consecutive contacts on each depth electrode (for example in electrode LA, we examined channels LA1-LA2, LA2-LA3… LA8-LA9). Channels exhibiting artifacts longer than 30 s around the time of seizures were eliminated from analysis (typically the most superficial channels), and shorter artifacts were marked and also excluded from the analysis.

Data analysis

Seizure and channel selection

One typical seizure per patient was selected for visual and spectral analyses. A typical seizure was defined as the patient’s most frequent seizure in regard to clinical and EEG semiology. The first such seizure during the monitoring was selected for analysis, provided it showed a good technical quality and no prolonged artifacts.

All seizure-onset channels were selected for analysis in each patient (Table 2). In addition, we selected 5–16 “control” channels clearly located outside the SOZ during the analyzed seizure; the number of channels depended on the number of electrodes implanted. These control channels were located in the homologous contralateral areas to the region of seizure onset if available, or else in adjacent regions. In the patients in whom two seizure types occurred, 1–2 channels within this second seizure onset area, which was not involved in the onset of the analyzed seizure, were additionally selected. This first seizure period was used for visual analysis of HFOs (see below).

Table 2.

Number of selected channels and seizures

Patients No. selected contacts No. selected SOZ contacts No. seizures recorded No. analyzed seizures
1 12 2 5 1
2 21 5 5 4
3 15 5 1. SOZ: 7, 2. SOZ: 1 4
4 11 4 14 4
5 12 5 9 3
6 8 3 2 1
7 12 5 4 1
Open in a new tab

This table provides information about the number of selected seizures and channels. Up to four seizures were analyzed per patient. The table gives only the number of clinical seizures, as some patients had additional electrographic events, which were not used for this study.

Up to three other seizures were selected and analyzed only with spectral analysis. If more than four seizures with good EEG quality were recorded on 2,000 Hz, the first four were analyzed. For these seizures the same channel selection was used. Because the visual marking of distinct HFOs is time-consuming, we did not mark the preictal periods for these additional seizures visually.

Visual marking of HFO events

A determination of seizure onset was made visually from the unfiltered EEG viewed at 10 s/page (33 mm/s). A preictal period of 15 min was inspected for the presence of visually identifiable HFOs. Channels were displayed with the maximum time resolution to visualize HFOs, which corresponded to approximately 0.6 s across the computer monitor (1,200 samples of a signal sampled at 2,000 Hz). Two versions of the same EEG were displayed side by side, one with a high-pass filter at 80 Hz and another with a high-pass filter at 250 Hz using finite impulse response (FIR) filters to eliminate ringing. The computer display was split vertically, and ripples were marked on one side using the 80-Hz high-pass filter and fast ripples on the other side with the 250-Hz high-pass filter. A ripple was marked if an event was clearly visible on the side of the 80-Hz filter and did not occur or show the same shape on the side of the 250-Hz filter, as it is defined as a distinct event between 80 and 250 Hz. An event was identified as a fast ripple if it was visible in the 250-Hz filter. Ripples and fast ripples were regarded as such only if their amplitude was clearly higher than the baseline of the investigated channel and they consisted of at least four consecutive oscillations. This method was used rather conservatively and HFOs in doubt in regard to their difference to baseline were not marked. Spikes were also marked on an unfiltered copy of the EEG and with the standard time-scale to allow the calculation of co-occurrence between HFOs and spikes. The data were marked by two reviewers (JJ and CC), one marking the EEG from time point −15 to 0, and the other reviewing the marked data in the opposite way. Both reviewers reviewed the data together and decided on a common way to mark if the second reviewer disagreed with the previous markings.

The visually marked data were divided into 30-s nonoverlapping segments. The rates of visually marked ripples, fast ripples, and spikes were counted for each segment using a MATLAB program. A regression analysis was performed to fit straight lines to the proportion of each 30-s segment that was occupied by HFOs. Linear trends 5 and 15 min before the seizure onset were considered. No trend was calculated for the 1-min periods as this would have only consisted of two 30-s segments. The significance level was set at p < 0.01. A slope significantly different from 0 implied a linear trend in the progression of the proportion of time occupied by HFOs as the seizure approached.

Spectral analysis of high-frequency bands in preictal sections

EEG data from the same selected channels, during a preictal period of up to 15 min, were selected for spectral analysis. This was performed for a maximum of four seizures per patient. Interictal epileptic discharges (spikes) were marked and excluded from the spectral analysis, as were segments with artifacts. Their data were replaced by linear interpolation of the previous and following spectra. The power trends in ripple (80–250 Hz), and fast ripple (250–500 Hz) bands were computed as functions of time.

The power trend for each band was computed using the Short Time Fourier Transform technique (Thakor & Tong, 2004). The preictal data was divided into 1.024 s nonover-lapping segments. Each segment was multiplied by a Blackman window, and the Fast Fourier transform calculated. By calculating the power of each band for each segment, a power trend as a function of time was obtained. Using regression, three straight lines were fitted to the power trends corresponding to preictal duration of 1, 5, and 15 min. The significance level was set at p < 0.01 in a first step and results of this analysis are presented in this study. Analyses were, however, repeated at p < 0.05 to ensure that no systematic changes were missed by the more conservative statistical analysis. A slope significantly different from 0 implied a linear trend in the progression of the power trend. These patterns of significant regressions were screened for any changes that developed consistently across SOZ and non-SOZ channels as the seizure onset approached.

Statistical analysis

The total rates of ripples and fast ripples were calculated for each channel and the mean rates of channels inside and outside the SOZ were compared. This first analysis did not look at temporal changes in HFO rates during the preictal period.

To evaluate temporal changes in HFO rates prior to the seizure, first, results of significant regression toward the onset of the seizure were studied comparing results from visual and spectral analysis for the first seizure of each patient. Results from the different patients were then compared to find a common trend in the behavior of HFOs before the seizure onset. Finally, the spectra of different seizures of the same patient were analyzed to find common patterns of high frequency power before the seizures of the same individual.

Results

Seven patients were included in this study, as they showed a well-defined focal seizure onset. A seizure onset over the mesial temporal structures was observed in four patients and over neocortical areas in three (Table 1). In one patient (patient 3), two SOZ areas could be identified, both focal and well defined.

Comparing rates: SOZ versus non-SOZ

The rates of ripples, fast ripples, and spikes were significantly higher in the SOZ (ripples: 18.2 ± 24.6/min, fast ripples: 9.1 ± 17.4, spikes: 13.8 ± 15) than outside [ripples: 3.7 ± 10.3/min, F(1,1837): 238, p < 0.001; fast ripples: 1.4 ± 4.5, F(1,1837): 164, p < 0.001; spikes: 2.8 ± 5.7, F(1,1837): 385, p < 0.001] when looking at all channels of all patients. Sixty-six percent of ripples and 57% of fast ripples occurred outside of spikes and again their rates were significantly higher in the SOZ (ripples alone: 10.7 ± 16.1/min, fast ripples alone: 4.4 ± 8.2/min) than outside [ripples alone: 2.9 9.7/min, F(1, 1378): 117.4, p < 0.001, fast ripples alone: 1.0 ± 4.2, F(1,1378): 99.2, p < 0.001].

At the single patient level, rates of ripples, fast ripples, and spikes were also significantly higher in the SOZ than outside for six of the seven patients: one showed no significant difference for ripples (Patient 7), another for fast ripples (Patient 1), and a third for spikes (Patient 3). Therefore, at least two of three event types showed significant differences in all seven patients.

Preictal changes in HFOs

Patients are reported separately, as they showed very variable results. Reported regressions were above the significance level of p < 0.01. Analysis using the significance level of p < 0.05 revealed very few additional significant positive and negative trends prior to the seizure, but did not reveal any systematic changes. Changes in the rates of spikes are not described in detail, as this was not the scope of our research, but no systematic increases or decreases were observed.

Patient 1

This patient had a right orbitofrontal and discrete anterior insular focal cortical dysplasia and an orbitofrontal seizure onset; all seizures derived from the dysplastic area (Suppl. 1) (Fig. 1). In the first seizure, visual marking showed an increase in ripple rates in two SOZ channels as well as three non-SOZ channels during 15 min and a decrease in one non-SOZ channel during the same period. Spectral analysis showed increase in the ripple band in all but one and in the fast ripple band in all but two channels 15 min prior to seizure onset (Fig. 2). However, seven channels showed a decrease in the ripple and three a decrease in the fast ripple band in the 5-min period. In the 1-min period, five channels showed an increase in the ripple band and three an increase in the fast ripple band; three others showed a decrease in the fast ripple band.

Figure 1.

Figure 1

Open in a new tab

Patient 1. Example of interictal and seizure onset recordings. (A) Electroencephalography (EEG) segment at seizure onset. Rhythmic spikes can be observed on contacts ROF6–7 and 7–8, in red. The red line represents the time of seizure onset. Spiking activity visible before that line in channels ROF6–7 and 7–8 represented the patient’s baseline activity in the interictal state, which can also be seen in C. The patient had a small dysplastic lesion and all seizures started focally in the two marked contacts. (B) Propagation of ictal activity after 10 s to contacts LAC1-2 and 2–3. (C–F) Typical preictal EEG segments in this patient. (C) Interictal EEG unfiltered and with normal time scale. (D) Gray segment of part C extended in time, with unchanged filters and gain; no HFOs are visible. (E) High pass filter of 80 Hz is applied to the same gray section and the gain is considerably increased, revealing ripple oscillations (green) on ROF6–7, 7–8 and F8–9. (F) Same EEG segment filtered at 250 Hz, showing fast ripples (blue) in the two seizure onset channels.

Epilepsia © ILAE

Figure 2.

Figure 2

Open in a new tab

Patient 1. Example of positive HFO trends toward the seizure onset. Two seizure-onset zone (SOZ) channels and two non-SOZ channels of a seizure analyzed with the visual and spectral methods are shown. Top: Trend from visual analysis. The proportion of the 30-s bins occupied by the visually identified (A) ripples (R) and (B) fast ripple (FR) is plotted, together with the linear trends whenever they were significant (p < 0.01), for 15 min (green) and 5 min (red) before seizure onset (t = 0). Because R and FR have different usual durations, the scales for R and FR are different. A positive linear trend 15 min prior to the seizure can be observed in the two SOZ channels and in LAC1-2. Middle: Trend for spectral analysis. Power spectra normalized with respect to the first 30 s, for (A) R band (80–250 Hz) and (B) FR band (250–500 Hz). A significant positive linear trend (p < 0.01) can be observed in R and FR bands for 15 min (green) before seizure onset in all the channels. Channel LOF7-8 presented also a negative significant trend in the R band 5 min (red) prior to onset. In the last preictal minute (yellow) the SOZ onset channels showed a positive trend. Bottom: Table showing the significant trends for all channels in patient 1, seizure 1. “+15” indicates a positive correlation in the 15-min period, and “−15” a negative correlation for the same period. The channels with light gray background are the ones presented above.

Epilepsia © ILAE

Three other seizures were analyzed, resulting in four seizures subjected to spectral analysis. They showed again variable preictal HFO patterns (Table S1), but three of the four seizures showed an increase in the ripple band in two SOZ channels 1 min before the seizure. This was not specific, however, and ripple band increases were seen in additional but variable other channels in all seizures. Three of four seizures also showed, as described previously, a very widespread increase in ripple and fast ripple band over the 15-min period involving nearly all channels. In this regard this patient’s seizures showed a more uniform pattern than all other patients. However, one seizure showed a contrary pattern, with a widespread decrease in ripple and fast ripple band in the 15-min period.

Patient 2

This patient had left temporal lobe atrophy and ipsilateral mesial temporal seizure onset with propagation to the temporal neocortex (Suppl. 2). Visual analysis of HFOs showed a decrease in ripple rates over 15 min in two non-SOZ channels and no other significant changes. Spectral analysis (Fig. 3) showed a decrease in ripple band in four non-SOZ channels and an increase in one, in the 15-min analysis. In the fast ripple band, a decrease in three non-SOZ channels and an increase in one were observed in the same time. During the 5-min period, a decrease in ripple band in four and an increase in three other channels were observed. Fast ripple band changes were seen in four channels as a decrease and three others as an increase. Finally, in the 1-min period, ripple band showed an increase in eight channels and a decrease in one, whereas fast ripple band showed an increase in six channels and a decrease in two. No increase and decrease of HFO bands showed a clear distribution, although changes were different for SOZ channels compared to non-SOZ channels.

Figure 3.

Figure 3

Open in a new tab

Patient 2. Example of a patient with no clear pattern of trend before seizure onset. As in Figure 2, Top: Trend from visual analysis; Middle: Spectral Trend, and Bottom: Table showing the significant trends for all channels. In the channels presented, no significant trends were obtained for the visually marked R and FR, even though the rate of events was high (see LHC1-2). No consistent pattern of change can be identified, as there were significant increases as well as decreases for the different durations (1, 5, and 15 min) before seizure onset and because only some channels presented significant trends.

Epilepsia © ILAE

One additional seizure was analyzed (Table S2). It showed a more widespread increase of both frequency bands within 15 min and a decrease in only two channels. In general the pattern of changes was very different during the second seizure, as a decrease in both frequency bands could be seen in most channels within the minute prior to seizure onset.

Patient 3

Seven of the eight seizures recorded in this patient originated in the right mesial temporal structures (Suppl. 3). One, not considered for this study, originated from the contralateral side. Visual analysis of the first seizure showed an increase in ripple rates in one SOZ channel and an increase in fast ripple rates in two (Table S3). No other changes were observed. Spectral increases in the ripple band were seen in 10 channels and in the fast ripple band in seven, including all SOZ channels, during the 15-min period. In the 5 min prior to the seizure, increase in one channel and decrease in one were observed for the ripple band and increase in two channels and decrease in three for the fast ripple band, including SOZ and non-SOZ channels. Decreases in three channels for the ripple band and in five for the fast ripple band were seen during the last minute prior to the seizure.

Three additional seizures with the same SOZ were analyzed (Fig. 4). They showed very variable patterns consisting of either increases or decreases in ripple or fast ripple bands during different time periods. No difference between SOZ and non-SOZ channels was observed.

Figure 4.

Figure 4

Open in a new tab

Patient 3. Comparisons of all the analyzed seizures. Top: The spectral analysis for FR band is shown for one SOZ channel and one non-SOZ channel for all seizures. For the SOZ channel, there was a negative trend 15 min before the seizure onset in two seizures but a positive trend in another. In the non-SOZ channel, there was an increase in three seizures 1 min before the onset, but a decrease in one. Bottom: Table showing channels with positive and negative trends that are in the SOZ and non-SOZ for each of the seizures of this patient. The channels with light gray background are the ones presented in detail.

Epilepsia © ILAE

Patient 4

Fourteen seizures were recorded, and the seizure onset was in the inner contacts of the anterior frontal electrode and the middle contacts of the electrode aiming at the focal cortical dysplasia in 12 seizures (Suppl. 4). Two of the 14 seizures had a more widespread onset, but they were not included in this study, as they were not recorded with a sampling rate of 2,000 Hz. All analyzed seizures, therefore, had uniform onset. Visual markings showed a significant increase of ripples toward the seizure in only the 15-min window in one of the SOZ channels and no other significant regressions (Table S4). Spectral analysis, however, showed an increase in the ripple band in one SOZ channel and in fast ripple band in three, over 15 min, as well as an increase in the ripple and fast ripple bands in two SOZ channels over 5 min. Two other SOZ channels showed a decrease in ripple and fast ripple bands within these 5 min. No changes in the spectra could be observed in the last minute before the seizure. Changes were limited to the SOZ channels.

When comparing this seizure with three other seizures, no clear similarities could be found in the spectra. There were decreases as well as increases in high-frequency band power before the seizures in SOZ and non-SOZ channels.

Patient 5

This patient without visible lesion had a very uniform pattern of focal seizure onset in right mesial temporal structures (Suppl. 5). The visually marked fast ripples showed a significant decrease in the 15 min before the seizure in two SOZ channels and two non-SOZ channels, whereas ripples showed the same in three non-SOZ channels (Table S5). Spectral bands for ripples and fast ripples showed several channels inside and outside the SOZ, with an increase in band power within 15 min before the seizure. Four non-SOZ channels showed a decrease in ripple band, and three non-SOZ channels also showed an increase in the fast ripple band in the 5 min prior to the seizure. Five channels showed increases and two decreases in the ripple band during the last minute.

Two additional seizures were analyzed. The only common finding in all seizures was an increase in spectral power for ripples and fast ripples in the 15 min before onset in the two SOZ channels recording from the amygdala. All other changes seemed random, and no common pattern could be identified.

Patient 6

Two seizures were recorded, both deriving from gliotic tissue anteriorly to a centroparietal porencephalic cyst (Suppl. 6). Visual analysis did not show any significant changes in HFOs prior to the seizure. Spectral analysis showed an increase in the ripple band in three channels and a decrease in fast ripple band in two other channels within 15 min (Table S6). In one channel an increase in ripple band and in four channels and increase in the fast ripple band was observed 5 min prior to the seizure. No spectral change in the minute prior to the seizure was seen. No difference between SOZ and non-SOZ was seen. Only one seizure was recorded with 2,000 Hz sampling and no comparison to other seizures was possible.

Patient 7

Seizures originated from both mesial temporal areas, but all clinical events originated in the right mesial temporal structures (Suppl. 7). Only one was recorded with a sampling rate of 2,000 Hz. Visual HFO marking showed an increase in ripple rates in the 5 min before the seizure in the left hippocampus, a non-SOZ channel during this seizure (Table S7). Spectral analysis showed a ripple band decrease in one channel and increases in two during the 15 min period. Fast ripple band decreased in four and increased in one channel in the same period. During the 5 min prior to the seizure, ripple and fast ripple bands decreased in three channels each. The ripple band increased in one channel and decreased in one channel within the 1-min period and the fast ripple band increased in three and decreased in two channels. There was no visible difference for all these changes between SOZ and non-SOZ channels. No further seizures could be analyzed.

Comparison of patients and conclusion

When comparing the preictal changes in HFO rates and spectra for the different patients, no systematic change could be found. Visual marking of the distinct events (ripples and fast ripples) led in general to fewer significant changes than spectral analysis. Even if some patients showed a clear increase in HFO rates and others a clear decrease, no consistent pattern was identifiable. This was also the case even when trying to compare the four patients with mesial temporal lobe seizures. Moreover, in the patients in whom several seizures were analyzed, no consistent change or pattern could be observed across seizures. Finally, there were no differences in the distribution of preictal HFO increase or decrease between SOZ and non-SOZ channels. In conclusion, no systematic change in HFO rates or high-frequency power was found, especially no systematic increase in SOZ channels.

Discussion

This study does not support the hypothesis that HFOs change in a predictive manner during the preictal state. The changes observed seemed more likely to represent spontaneous fluctuations in HFO rates and band power, as they were variable between patients and between different seizures of the same patient. A secondary finding of this study was the demonstration that rates of HFOs during the preictal period were significantly higher in SOZ than outside, confirming findings for interictal periods of slow wave sleep (Jacobs et al., 2008).

Should preictal changes be expected? There are indications that mechanisms of seizure generation are variable depending on the type of epilepsy (Lopes da Silva et al., 2003). The mechanisms leading to the occurrence of seizures remain unclear, and several hypotheses have been made regarding the transition from interictal to ictal activity. One hypothesis states that seizures are initiated by abnormally discharging neurons, which show high bursts of activity and the ability to recruit neighboring neurons. This process leads to an abnormal synchronization of firing neighboring neurons, a loss of surround inhibition, and finally to sustained and highly synchronized discharges of large neuronal groups, called a seizure (Yaari & Beck, 2002). Especially in patients with focal cortical onset, a gradual change from interictal to ictal state has been suggested, which should then be detectable with EEG (Lopes da Silva et al., 2003). All patients in the present study had focal seizures, and gradual preictal EEG changes could, therefore, have been expected.

The patients were selected independent of the anatomic area of seizure onset, the etiology of the disease, and the frequency of their habitual seizures. It remains possible that we could not detect consistent preictal changes because HFO rates can only be used as a seizure predictor in certain types of epilepsy. However, patients in our study whose seizures were similar in etiology and electrographic ictal pattern had no similarities in preictal patterns. In addition, patterns of preictal HFO occurrence were not even similar when comparing several similar seizures of the same patient. In particular, the finding of negative and positive regressions in the rates of HFOs prior to seizures seems to suggest that the regressions reflect random variations in HFO rate and cannot be used for seizure prediction purposes.

Although the results of this study did not reveal systematic changes in high-frequency power prior to seizures and could not demonstrate the usefulness of high-frequency recordings for seizure prediction purposes, longer pre-ictal periods would have to be analyzed in a larger number of patients and seizures to generalize these preliminary findings (Mormann et al., 2007). In fact, changes in the 60–100 Hz frequency band prior to seizures have been reported (Worrell et al., 2004) in a small number of patients. To analyze HFOs over longer periods, an automatic detection of the oscillations would be necessary. Methods to detect distinct HFOs have been proposed (Staba et al., 2002; Nelson et al., 2006; Gardner et al., 2007), but we did not find them adequate for our requirements; an automatic detector should not only be able to detect HFOs within variable baselines, but also to distinguish between HFOs co-occurring with spikes and those outside spikes. Visual analysis and manual marking of HFOs in this study had two advantages over automated methods. First, it could adjust for baseline changes, which in our experience are often significant, especially as depth electrodes record from different brain regions (hippocampus, cortex, white matter). Second, we could show that the proportion of HFOs within and outside spikes remains stable preictally and changes in HFO rates, therefore, did not result from changes in spike rates (and oscillations due to filtering of fast spikes). Visual marking of HFOs, however, is time-consuming (Urrestarazu et al., 2007) and we had to limit the analysis to 15 min. This marking also raises questions on the objectivity of marked events as well as intra- and interrater reliability. An important aspect of this study was to ensure that the marking of events was stable over the preictal period and did not change over time. For this purpose the EEG segments were marked by one reviewer from time point −15 to 0 and re-reviewed by a second in the reverse direction.

During the marking process, reviewers agreed to mark specific events, disregarding events that were marginally different from the channel’s baseline. However, objective measures of reliability, such as the calculation of Cohen’s Kappa, may be valuable in the future. The authors addressed this question in an independent study on inter-ictal data and found good agreement between reviewers (average Cohen’s Kappa: 0.7; Zelmann et al., 2008). We, therefore, believe that the data presented in this study reflects an accurate measure of HFO rates. An automatic detection of HFOs within and outside spikes over a longer period may eventually provide more detailed and accurate information on preictal changes.

Not only longer but also shorter periods prior to the seizure may reveal trends not noted here. We limited our analysis to two 30-s bins in the minute before the seizure onset. Some studies looking at preictal activity analyzed shorter periods in the range of several seconds before the seizure to detect preictal changes (Bragin et al., 2005), one finding a change in the high frequency band 8 s prior to seizure onset (Khosravani et al., 2008). The exact definition of the seizure onset time, however, might be difficult and subjective. For this reason the analysis of shorter periods prior to the seizure risks look at ictal instead of preictal activity.

In principle, preictal changes of distinct HFOs rates, high frequency power, or both are possible. We, therefore, performed spectral analysis for high frequency power in addition to visual analysis of HFOs. In this analysis we excluded the 1-s bins that included spikes to avoid detecting changes in the spectra caused by changes in spike rate. For this reason and because the rates of HFOs and power spectra are not directly comparable, it is not surprising that the results of spectral and visual analyses differed: Although the visually marked data only analyzed distinct oscillations clearly separable from the background EEG and occupying a very small fraction of time (Fig. 1), spectral analysis detects sustained changes in high-frequency power.

Ripples occur as physiologic oscillations during memory consolidation and are also found in pathologic tissue (Draguhn et al., 2000; Bragin et al., 2004; Jacobs et al., 2008). Fast ripples have been more closely linked to epileptogenicity and the SOZ (Staba et al., 2002, 2007). They are not generated in all lesional brain tissue but appear specific to areas able to generate seizures (Dubeau et al., 2007). It could have been expected that preictal changes within neuronal networks, such as reduced inhibition and increased synchronicity of neighboring neurons, would result in changes in HFO rates. These changes would be limited to areas of seizure onset in the preictal period and then could be present in areas of propagation during the seizure. Jirsch et al. (2006) observed large number of ictal HFOs in the seizure onset areas and later in areas of propagation. This could also be supported from findings in an in vitro model of low-Mg2+ spontaneous seizures in rats, which observed a preictal change in the subripple (0–100 Hz), ripple (100–200 Hz), and fast ripple (200–300 Hz) bands (Khosravani et al., 2005). Studies in rats, which showed increases in high frequency band seconds prior to seizures, are not easily comparable to our study (Bragin et al., 2005). To date there are no reports describing distinct HFO rate increases during the preictal period, and this was absent in our findings as well. Instead, we found that HFO rates during the preictal period follow spontaneous fluctuations, as might be expected from a period of uncontrolled state of consciousness (Bagshaw et al., 2008). Moreover, the changes in HFOs were not limited to the SOZ and were also found in distant non-SOZ areas.

Moving from the temporal to the spatial distribution, HFOs showed the same characteristics during the preictal period as those during the interictal period (Jacobs et al., 2008). The difference in HFO rates between SOZ areas and areas outside the SOZ was highly significant. It is not surprising that rates in general were a bit lower than in our previous studies because most seizures in this study occurred during wakefulness, and our other studies (Urrestarazu et al., 2007; Jacobs et al., 2008) were performed during slow wave sleep, when rates are highest (Staba et al., 2004; Bagshaw et al., 2008).

No changes in the proportion of HFOs co-occurring with spikes as well as no systematic changes in the rates of spikes were observed in the preictal period. This is not surprising, since most studies did not show preictal changes in spiking rates (Gotman & Marciani, 1985; Gotman & Koffler, 1989; Spencer et al., 2008). HFOs in the preictal period in some regards seem to behave very similarly to spikes. They occur very clearly and more frequently in the areas generating seizures, and they are not more likely to occur before a seizure than at other times, whether or not they co-occur with spikes.

In conclusion, our study suggests that HFOs may not be good indicators for the transition from interictal to ictal states during the minutes prior to seizures and hence may not directly be involved in the mechanisms of seizure generation.

Supplementary Material

S1
S2
S3
S4
S5
S6
S7

Acknowledgments

This project was supported by grant MOP-10189 from the Canadian Institutes of Health Research.

Footnotes

We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines. J Gotman is a major shareholder of Stellate.

Disclosure: None of the other authors has any conflict of interest to disclose.

Supporting Information

Additional Supporting Information may be found in the online version of this article:

Tables S1–S7: Tables for the results of each patient and all analyzed seizures can be found as supplementary data.

Please note: Wiley-Blackwell not responsible for the content or functionality of any supporting information supplied by authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

References

  1. Bagshaw AP, Jacobs J, LeVan P, Dubeau F, Gotman J. Effect of sleep stage on inter-ictal high frequency oscillations recorded from depth macroelectrodes in patients with focal epilepsy. Epilepsia. 2008 Sep 17; doi: 10.1111/j.1528-1167.2008.01784.x. [Epub ahead of print] [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Bragin A, Engel J, Jr, Wilson CL, Fried I, Mathern GW. 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;40:127–137. doi: 10.1111/j.1528-1157.1999.tb02065.x. [DOI] [PubMed] [Google Scholar]
  3. Bragin A, Wilson CL, Almajano J, Mody I, Engel J., Jr High-frequency oscillations after status epilepticus: epileptogenesis and seizure genesis. Epilepsia. 2004;45:1017–1023. doi: 10.1111/j.0013-9580.2004.17004.x. [DOI] [PubMed] [Google Scholar]
  4. Bragin A, Azizyan A, Almajano J, Wilson CL, Engel J., Jr Analysis of chronic seizure onsets after intrahippocampal kainic acid injection in freely moving rats. Epilepsia. 2005;46:1592–1598. doi: 10.1111/j.1528-1167.2005.00268.x. [DOI] [PubMed] [Google Scholar]
  5. Draguhn A, Traub RD, Bibbig A, Schmitz D. Ripple (approximately 200-Hz) oscillations in temporal structures. J Clin Neurophysiol. 2000;17:361–376. doi: 10.1097/00004691-200007000-00003. [DOI] [PubMed] [Google Scholar]
  6. Dubeau F, Jacobs J, Chander R, Soulami L, Gotman J. Interictal Ripples and Fast Ripples in intracranial EEGs of patients with intractable lesional epilepsy. Epilepsia. 2007;45(Suppl 6):402. [Google Scholar]
  7. Elger CE. Future trends in epileptology. Curr Opin Neurol. 2001;14:185–186. doi: 10.1097/00019052-200104000-00008. [DOI] [PubMed] [Google Scholar]
  8. Gardner AB, Worrell GA, Marsh E, Dlugos D, Litt B. Human and automated detection of high-frequency oscillations in clinical intracranial EEG recordings. Clin Neurophysiol. 2007;118:1134–1143. doi: 10.1016/j.clinph.2006.12.019. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Gotman J, Marciani MG. Electroencephalographic spiking activity, drug levels, and seizure occurrence in epileptic patients. Ann Neurol. 1985;17:597–603. doi: 10.1002/ana.410170612. [DOI] [PubMed] [Google Scholar]
  10. Gotman J, Koffler DJ. Interictal spiking increases after seizures but does not after decrease in medication. Electroencephalogr Clin Neurophysiol. 1989;72:7–15. doi: 10.1016/0013-4694(89)90026-6. [DOI] [PubMed] [Google Scholar]
  11. Haut SR, Hall CB, Masur J, Lipton RB. Seizure occurrence: precipitants and prediction. Neurology. 2007;69:1905–1910. doi: 10.1212/01.wnl.0000278112.48285.84. [DOI] [PubMed] [Google Scholar]
  12. Jacobs J, Levan P, Chander R, Hall J, Dubeau F, Gotman J. Inte-rictal high-frequency oscillations (80–500 Hz) are an indicator of seizure onset areas independent of spikes in the human epileptic brain. Epilepsia. 2008;49:1893–1907. doi: 10.1111/j.1528-1167.2008.01656.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Jirsch JD, Urrestarazu E, LeVan P, Olivier A, Dubeau F, Gotman J. High-frequency oscillations during human focal seizures. Brain. 2006;129:1593–1608. doi: 10.1093/brain/awl085. [DOI] [PubMed] [Google Scholar]
  14. Khosravani H, Pinnegar CR, Mitchell JR, Bardakjian BL, Federico P, Carlen PL. Increased high-frequency oscillations precede in vitro low-Mg++ seizures. Epilepsia. 2005;46:1188–1197. doi: 10.1111/j.1528-1167.2005.65604.x. [DOI] [PubMed] [Google Scholar]
  15. Khosravani H, Mehrotra N, Rigby M, Hader WJ, Pinnegar CR, Pillay N, Wiebe S, Federico P. Spatial localization and time-dependant changes of electrographic high frequency oscillations in human temporal lobe epilepsy. Epilepsia. 2008 Aug 19; doi: 10.1111/j.1528-1167.2008.01761.x. [Epub ahead of print] [DOI] [PubMed] [Google Scholar]
  16. Lehnertz K, Widman G, Andrzejak R, Arnhold J, Elger CE. Is it possible to anticipate seizure onset by non-linear analysis of intracerebral EEG in human partial epilepsies? Rev Neurol. 1999;155:454–456. [PubMed] [Google Scholar]
  17. Le Van Quyen M, Martinerie J, Baulac M, Varela F. Anticipating epileptic seizures in real time by a non-linear analysis of similarity between EEG recordings. Neuroreport. 1999;13:10. doi: 10.1097/00001756-199907130-00028. [DOI] [PubMed] [Google Scholar]
  18. Litt B, Esteller R, Echauz J, D’Alessandro M, Shor R, Henry T, Pennell P, Epstein C, Bakay R, Dichter M, Vachtsevanos G. Epileptic seizures may begin hours in advance of clinical onset: a report of five patients. Neuron. 2001;30:51–64. doi: 10.1016/s0896-6273(01)00262-8. [DOI] [PubMed] [Google Scholar]
  19. Lopes da Silva F, Blanes W, Kalitzin SN, Parra J, Suffczynski P, Velis DN. Epilepsies as dynamical diseases of brain systems: basic models of the transition between normal and epileptic activity. Epilepsia. 2003;44(Suppl 12):72–83. doi: 10.1111/j.0013-9580.2003.12005.x. [DOI] [PubMed] [Google Scholar]
  20. Mormann F, Andrzejak RG, Elger CE, Lehnertz K. Seizure prediction: the long and winding road. Brain. 2007;130:314–333. doi: 10.1093/brain/awl241. [DOI] [PubMed] [Google Scholar]
  21. Nelson R, Myers SM, Simonotto JD, Furman MD, Spano M, Norman WM, Liu Z, DeMarse TB, Carney PR, Ditto WL. Detection of high frequency oscillations with Teager energy in an animal model of limbic epilepsy. Conf Proc IEEE Eng Med Biol Soc. 2006;1:2578–2580. doi: 10.1109/IEMBS.2006.259694. [DOI] [PubMed] [Google Scholar]
  22. Ochi A, Otsubo H, Donner EJ, Elliott I, Iwata R, Funaki T, Akizuki Y, Akiyama T, Imai K, Rutka JT, Snead OC., 3rd Dynamic changes of ictal high-frequency oscillations in neo-cortical epilepsy: using multiple band frequency analysis. Epilepsia. 2007;48:286–296. doi: 10.1111/j.1528-1167.2007.00923.x. [DOI] [PubMed] [Google Scholar]
  23. Osorio I, Frei MG, Sunderam S, Giftakis J, Bhavaraju NC, Schaffner SF, Wilkinson SB. Automated seizure abatement in humans using electrical stimulation. Ann Neurol. 2005;57:258–268. doi: 10.1002/ana.20377. [DOI] [PubMed] [Google Scholar]
  24. Ramachandrannair R, Ochi A, Imai K, Benifla M, Akiyama T, Holowka S, Rutka JT, Snead OC, 3rd, Otsubo H. Epileptic spasms in older pediatric patients: MEG and ictal high-frequency oscillations suggest focal-onset seizures in a subset of epileptic spasms. Epilepsy Res. 2008;78:216–224. doi: 10.1016/j.eplepsyres.2007.12.007. [DOI] [PubMed] [Google Scholar]
  25. Schelter B, Winterhalder M, Maiwald T, Brandt A, Schad A, Timmer J, Schulze-Bonhage A. Do false predictions of seizures depend on the state of vigilance? A report from two seizure-prediction methods and proposed remedies. Epilepsia. 2006;47:2058–2070. doi: 10.1111/j.1528-1167.2006.00848.x. [DOI] [PubMed] [Google Scholar]
  26. Schulze-Bonhage A, Kurth C, Carius A, Steinhoff BJ, Mayer T. Seizure anticipation by patients with focal and generalized epilepsy: a multicentre assessment of premonitory symptoms. Epilepsy Res. 2006;70:83–88. doi: 10.1016/j.eplepsyres.2006.02.001. [DOI] [PubMed] [Google Scholar]
  27. Spencer SS, Goncharova II, Duckrow RB, Novotny EJ, Zaveri HP. Interictal spikes on intracranial recording: behavior, physiology, and implications. Epilepsia. 2008;49:1881–1892. doi: 10.1111/j.1528-1167.2008.01641.x. [DOI] [PubMed] [Google Scholar]
  28. Staba RJ, Wilson CL, Bragin A, Fried I, Engel J., Jr Quantitative analysis of high-frequency oscillations (80–500 Hz) recorded in human epileptic hippocampus and entorhinal cortex. J Neurophysiol. 2002;88:1743–1752. doi: 10.1152/jn.2002.88.4.1743. [DOI] [PubMed] [Google Scholar]
  29. Staba RJ, Wilson CL, Bragin A, Jhung D, Fried I, Engel J., Jr High-frequency oscillations recorded in human medial temporal lobe during sleep. Ann Neurol. 2004;56:108–115. doi: 10.1002/ana.20164. [DOI] [PubMed] [Google Scholar]
  30. Staba RJ, Frighetto L, Behnke EJ, Mathern GW, Fields T, Bragin A, Ogren J, Fried I, Wilson CL, Engel J., Jr Increased fast ripple to ripple ratios correlate with reduced hippocampal volumes and neuron loss in temporal lobe epilepsy patients. Epilepsia. 2007;48:2130–2138. doi: 10.1111/j.1528-1167.2007.01225.x. [DOI] [PubMed] [Google Scholar]
  31. Thakor NV, Tong S. Advances in quantitative electroencephalogram analysis methods. Annu Rev Biomed Eng. 2004;6:453–495. doi: 10.1146/annurev.bioeng.5.040202.121601. [DOI] [PubMed] [Google Scholar]
  32. Urrestarazu E, Jirsch JD, LeVan P, Hall J, Avoli M, Dubeau F, Gotman J. High-frequency intracerebral EEG activity (100–500 Hz) following interictal spikes. Epilepsia. 2006;47:1465–1476. doi: 10.1111/j.1528-1167.2006.00618.x. [DOI] [PubMed] [Google Scholar]
  33. Urrestarazu E, Chander R, Dubeau F, Gotman J. Interictal high-frequency oscillations (100–500 Hz) in the intracerebral EEG of epileptic patients. Brain. 2007;130:2354–2366. doi: 10.1093/brain/awm149. [DOI] [PubMed] [Google Scholar]
  34. Worrell GA, Parish L, Cranstoun SD, Jonas R, Baltuch G, Litt B. High-frequency oscillations and seizure generation in neocortical epilepsy. Brain. 2004;127:1496–1506. doi: 10.1093/brain/awh149. [DOI] [PubMed] [Google Scholar]
  35. Worrell GA, Gardner AB, Stead SM, Hu S, Goerss S, Cascino GJ, Meyer FB, Marsh R, Litt B. High-frequency oscillations in human temporal lobe: simultaneous microwire and clinical macroelectrode recordings. Brain. 2008;131:928–937. doi: 10.1093/brain/awn006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Yaari Y, Beck H. “Epileptic neurons” in temporal lobe epilepsy. Brain Pathol. 2002;12:234–239. doi: 10.1111/j.1750-3639.2002.tb00438.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Zelmann R, Zijlmans M, Jacobs J, Châtillon C-E, Gotman J. Improving the visual identification of high frequency oscillations [Abstract] Epilepsia Suppl. 2008;49:28–29. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

S1
NIHMS3402-supplement-S1.tif (570.3KB, tif)
S2
NIHMS3402-supplement-S2.tif (785.1KB, tif)
S3
NIHMS3402-supplement-S3.tif (860.9KB, tif)
S4
NIHMS3402-supplement-S4.tif (628.4KB, tif)
S5
NIHMS3402-supplement-S5.tif (694.9KB, tif)
S6
NIHMS3402-supplement-S6.tif (294.8KB, tif)
S7
NIHMS3402-supplement-S7.tif (482.8KB, tif)

RESOURCES