Abstracts

Rationale: High frequency oscillations (HFO) can be divided into Ripples between 80-250Hz and Fast Ripples (FR) above 250Hz. Both are new EEG markers of epileptogenicity. The occurrence of HFOs is closely linked to epileptogenic areas, and the removal of areas with HFOs correlates with postsurgical outcome. Ripples however can also occur physiologically during memory consolidation processes in medial temporal-lobe structures. The proportion of physiological and pathological HFOs recorded in EEGs from these structures is unclear, as frequency analysis is insufficient for separating the two types of events. We investigated the correlation between Ripple and FR rates and memory performance. The hypothesis is that physiological ripples correlate positively with memory performance.Methods: All patients investigated between 2004-2010 with bilateral medial temporal electrodes and with an IQ allowing for memory testing were included in this study. Intracranial EEG was filtered at 500Hz and recorded at 2000Hz with depth electrodes. Ripples and FRs were visually marked by two independent reviewers during 5 minutes of slow wave sleep, and rates of HFOs were calculated for each channel. Patients underwent three verbal and three nonverbal memory tests. They were grouped into severely impaired, impaired, borderline or intact categories for verbal and nonverbal memory. We calculated a Spearman correlation between HFO rates in the hippocampi and the memory category, and compared HFO rates in each hippocampus with the corresponding (verbal-left,nonverbal-right) memory result using ANOVA (p=0.05). Separate analyses were performed for seizure onset zone (SOZ) and non-SOZ channels.Results: 20 patients were included; 10 had bilateral, 5 unilateral and 5 no memory impairment. Verbal memory was affected in 9 and nonverbal memory in 14 patients. There was no correlation between HFO rates and memory performance in the SOZ (Fig 1). This was true for general memory and for separate verbal and nonverbal memory results. There was however a significant negative correlation between the overall memory performance and rates of ripples (r=-0.64, p =0.005) outside the SOZ (fig. 2). Similarly, the ANOVA comparison of rates showed significantly higher rates of HFOs in areas with the poor memory performance compared to intact memory (p=0.002).Conclusions: Our results suggest that the majority of HFOs are not linked to memory function and are therefore likely pathological, which is concordant with their link to epileptogenicity. The findings were restricted to areas outside the SOZ. The absence of negative correlation between memory performance and HFO rates in the SOZ could be explained by HFO rates in the SOZ being generally so high that differences between areas with remaining and impaired memory function cannot be seen. These results are reassuring in regard to the use of HFOs as a marker of epileptogenicity in temporal-lobe epilepsy.