Abstracts

Rationale: High frequency oscillations (HFO, ripples: 80–250 Hz, fast ripples: 250–500 Hz) are new promising EEG markers of epileptic tissue and could therefore be useful in epilepsy surgery. While some studies suggest that the amount of HFO generating tissue that is removed correlates with the postsurgical seizure outcome, others failed to demonstrate that HFO reception can predict seizure outcome in individual patients. One challenge in the identification of HFO is the differentiation between physiological and epileptic HFO. In mesiotemporal structures, physiological ripples can be identified by analyzing their co-occurrence with scalp EEG sleep spindles. This study aims to investigate if co-occurence of HFO and sleep spindles is typical for mesiotemporal structures or could also be seen in the neocortex. Additionally, we hypothesize that ripples co-occurring with epileptic spikes are predominately found in the seizure onset zone (SOZ) while physiological spindle ripples do not show this preference. Methods: We examined consecutive 16 patients who underwent chronic intracranial EEG with mesiotemporal implantation and simultaneous scalp EEG at Freiburg Epilepsy Center. EEG sampling rate was 2 kHz. For each patient a 1-hour slow wave sleep EEG segment was chosen. Ripples and spikes were automatically detected in intracranial and sleep spindles in scalp EEG with the mossdet detectors of Lachner-Piza. Ripples were divided into three subgroups: ripples coincident with scalp sleep spindles (SpR), ripples coincident with epileptic spikes in the same contacts (SpikeR), and ripples not coincident with other events (NoncR). Ripple rates were compared between SOZ and non-SOZ areas using a Kruskal Wallis tests (α <0.05). Results: 1,157 contacts in mesiotemporal lobe (MTL) structures (amygdala: 25 contacts, hippocampus: 43, parahippocampus: 12) and neocortex (temporal: 700, frontal: 296, parietal: 57, occipital: 24) were analyzed. Most patients had the majority SOZ contacts in the MTL or the temporal neocortex (TNC). Sleep spindles, spikes, and the three ripple subgroups could be seen in all patients (total amount of ripples: n=822,304; SpR: n=89,690; SpikeR: 170,005; NoncR: 523,566). The highest average rate of SpR per contact per minute was found in the frontal neocortex (1.63/min), followed by the hippocampus (1.42/min), the amygdala (1.40/min), the TNC (1.18/min), the parietal lobe (1.03/min), the occipital lobe (0.90/min), and parahippocampal structures (0.68/min). The boxplot comparisons (including only MTL and TNC contacts) showed significantly higher SpikeR rates in SOZ contacts than in Non-SOZ contacts (p<0.001, α<0.05), whereas SpR showed lower rates and only small differences between SOZ and non-SOZ (see fig. 1). In non-SOZ contacts, SpR showed lower rates in all MTL structures and the TNC in comparison to SpikeR and NonCR, with the highest rate of SpR in the Amygdala (see Fig.2; p<0.001, α<0.05). Conclusions: As previously described, SpikeR are highly linked to SOZ areas and represent most likely epileptic events. In contrast no link between SpR and the SOZ could be found in concordance with the idea that these events are physiological. A link between ripples and spindles could not only be shown for mesio-temporal events as described before but also for ripples generated over other brain regions. Identifying ripple events that co-occur with sleep spindles therefore might be a useful tool to improve precision of epileptic events in the HFO analysis. It however remains to be proven whether the differentiation of physiological and epileptic HFO actually improves the prediction of seizure outcome in individual patients. Funding: Jonas Bruder is funded by the DFG grant JA 1725/4-1.