Ictal and interictal high frequency oscillations in patients with focal epilepsy

Abstract

Objective: High frequency oscillations (HFOs) can be recorded with depth electrodes in focal epilepsy patients. They occur during seizures and interictally and seem important in seizure genesis. We investigated whether interictal and ictal HFOs occur in the same regions and how they relate to epileptiform spikes.

Methods: In 25 patients, spikes, ripples (80-250 Hz) and fast ripples (FR: 250-500 Hz) and their co-occurrences were marked during interictal slow wave sleep (5-10 min), during 10 pre-ictal seconds and 5s following seizure onset. We compared occurrence and spatial distribution between these periods.

Results: HFOs and spikes increased from interictal to ictal periods: the percentage of time occupied by ripples increased from 2.3% to 6.5%, FR from 0.2% to 0.8%, spikes from 1.1% to 4.8%. HFOs increased from interictal to pre-ictal periods in contrast to spikes. Spikes were in different channels in the interictal, pre-ictal and ictal periods whereas HFOs largely remained in the same channels.

Conclusions: HFOs remain confined to the same, possibly epileptogenic, area, during interictal and ictal periods, while spikes are more widespread during seizures than interictally.

Significance: Ictal and interictal HFOs represent the same (epileptogenic) area and are probably similar phenomena.

Fig. 1

Channels with potential muscle artefacts were excluded from the analysis. This was done by reviewing the SEEG at normal time scale with a filter of 80 Hz together with the available epidural, ECG and EMG channels. Muscle artifact can be recognized as a simultaneous high frequency artifact over channels that are potentially outside of the brain, like channels LS6-7 and above, LC5-6 and above and RS5-6 and above in this example. Another clue could be obtained by filtering at lower frequencies as well. If still in doubt, the signal was reviewed at a timescale showing all samples. Muscle artifact shows a less sinusoid shape than HFOs and the frequency spectrum shows relatively more frequencies (Otsubo et al., 2008). Whenever there was doubt, the channel was excluded.

Fig. 2

Example of calculating the ranking distance between interictal and ictal periods in one patient for ripples, FRs and spikes. A: channels are ranked according to the percentage of time occupied by the events (color is used here to facilitate channel identification; it does not represent percentage of time). Then, arrows indicate the channels for which rankings are different between the two periods, in this case interictal and ictal, and these differences are summed (6×, 6×, 24×). B: same concept as in A except that channels are weighted by their percentage of time occupied and arrows are weighted by their point of origin to allow large changes to influence the outcome more than small changes. So, an equal change in ranking can lead to a greater ranking distance if the relative differences of the changes are higher (in this example ripples compared to FRs: both have six changes in ranking, but the weighted ranking distances are, respectively 0.29 and 0.18). LH1-2, RH1-2, RH2-3 etc. are channel names (Table 1).

Fig. 3

Sum of percentages of time (over all channels) occupied by ripples, FRs, spikes, ripples without spikes (R_isol) and FRs without spikes (FR_isol), during the interictal, pre-ictal and ictal periods. The darker grey represents the sum of events in the SOZ-channels (in total 119 channels), while the lighter grey represents the sum of events in the channels outside the SOZ (in total 683 channels). For HFOs there is a significant increase from interictal to pre-ictal and from pre-ictal to ictal, but spikes show a significant decrease pre-ictally. The pre-ictal to ictal increase is greatest for the spikes and relatively low for the HFOs without spikes, especially outside the SOZ.

Fig. 4

mean RKD between rankings of channels for the percentage of time occupied by events between interictal and ictal, interictal and pre-ictal, pre-ictal and ictal. A *means that the value is significantly greater than 0.2 with z-test (p < 0.05).

Fig. 5

Bar graphs of interictal, pre-ictal and ictal amounts (percentage of time) of ripples, FRs and spikes, summed for all 25 patients after separating the channels into groups of “channels with interictal events”, “channels without interictal but with ictal events” (rates below 0.2) and “patients with only ictal events” (not shown). The “channels with interictal events” were ranked for each patient from highest to lowest and the summed up for the interictal as well as the pre-ictal and ictal events. Example: on one patient, channels A and B show interictal spikes with the highest percentage of occupancy in channel A; in a second patient, channels D, C and E show interictal spikes in descending order (and no spikes in channels A and B), the graph would sum up the percentages of channels A and D (rank 1), channels B and C (rank 2) and 0 and E (rank 3). The last bar shows the summation of the “channels with only ictal events”. This bar does also contain interictal events because rates were below 0.2, so not 0. The first light grey bars represents the interictal, the second darker bars the pre-ictal and the last black bars the ictal period. The distribution of spikes over channels differs more between the ictal and interictal period than the distribution of HFOs: the channels that show most HFOs interictally also show most HFOs ictally; rankings are parallel, unlike for spikes.