Abstracts

Rationale: High frequency oscillations (HFOs), i.e. ripples (80-250Hz) and fast ripples (FR, 250-500Hz), are novel EEG markers of epileptic tissue. Retrospective studies using various methods for intracranial EEG recording and HFO identification suggest that removal of HFO-generating tissue correlates with a seizure free postsurgical outcome. In this study, three centers using different techniques for recording intracranial EEG prospectively collected data hypothesizing that HFOs can be used as markers of epileptic tissue, independently of patient population and recording technique. Methods: Between March 2011 and June 2012, the Epilepsy Unit of the Montreal Neurological Institute (Canada), the Epilepsy Center Freiburg (Germany) and Pediatric Epilepsy Program of the University of California Los Angeles (USA) each included up to 20 consecutive patients undergoing intracranial EEG. Ten minutes of interictal EEG were analyzed. An HFO detector was trained in the first minute (in which HFOs and baselines were visually marked) for each individual channel and the parameters obtained were used in the remaining minutes. For channels without events during the training period, default parameters per center were obtained from additional 3 patients not considered in this study. Rates were calculated for ripples and FR. Removed brain areas were identified using postsurgical MRI. The ratio of ripple and FR rates in removed vs. remaining channels as well as the ratio of the number of removed vs. remaining channels with ripple, FR or in the SOZ were calculated. After a minimum of 12 months the postsurgical outcome according to Engel was classified by a physician unaware of the previously calculated ratios. Spearman rho was used for correlations between all values and the outcome. Results: A total of 52 patients were included. Of these, 24 were investigated with chronic intracranial depth electrodes, 8 with chronic subdural grids and 20 with intra-operative electro-corticography. 29 patients were in Engel class 1, 7 in Engel class 2, 9 in Engel class 3 and 8 in Engel class 4. Removal of the majority of ripple and FR events was significantly correlated with better surgical outcome (Ripple: p=0.007, rho = -0.41; fast ripple: p=0.001, rho=-0.47; Figure 1). Removing channels with HFOs was also associated with a more favorable surgical outcome (Ripples: p=0.02, rho=-0.37; Fast ripples: p=0.02, rho: -0.36 Figure 2). No significant correlation was found between removal of SOZ channels and postsurgical outcome (Figure 2). Conclusions: This study replicated the previously described correlation between the removal of HFO-generating tissue and postsurgical seizure outcome with data collected prospectively in independent centers. This observation is promising for clinical applications in two ways: first, it is likely that similar findings are applicable to most clinical situations in other centers and independent of patient populations and implantation technique. Second, we showed that a uniform way of semiautomatic analysis was feasible in different settings.