Abstracts

Rationale: Between 2 and 5% of all children experience at least one seizure during childhood. The recurrence risk is around 40% within the following two years, whereas the incidence of epilepsy is 1%. Traditionally, interictal epileptiform discharges (IEDs) in electroencephalogram (EEG) have been used as biomarker for estimating the risk of further seizures or epilepsy. However, poor sensitivity limits its prognostic value. High frequency Oscillations (HFOs), have gained importance as biomarkers for epileptogenic regions in intracranial EEG. Very recently, ripples within spikes, detected from scalp EEG have shown to be a specific biomarker for seizure risk in epilepsy with centrotemporal spikes. The aim of our study is to test the use of scalp HFO as a biomarker for seizure recurrence risk after a first unprovoked seizure in childhood and compare its pathological significance to IED. Methods: Of a prospective cohort of children after a first unprovoked seizure, all EEG recorded within 72 hours after the first seizure were visually evaluated through two independent reviewers, blinded for the outcome, using the Software „Harmonie Monitoring System“ (Stellate Systems Inc., Montréal, Canada). Patients were divided in two groups according to outcome within one year of follow up that was either diagnosis of epilepsy according to ILAE guidelines or no diagnosis of epilepsy. Frequency of spikes, HFOs and spike-ripples was compared between those two groups using unpaired t-test. Results: Between 07/01/2017 and 08/31/2018, 76 patients were screened after a suspected first seizure, of those 56 could be included (table 1). Twenty patients were excluded due to unambiguity of diagnosis (n=16) and 4 were lost to follow up. As of 05/31/2019, 27 (48.2%) were diagnosed with epilepsy. Overall, scalp HFOs have been detected in 43 (76.8%), spikes in 22 (39.3%) and spike-ripples in 19 (33.9%) of the patients. Frequency of ripples, spikes and spike-ripples are displayed for both outcome groups separately in figure 1. Ripples and spike-ripples were found significantly more frequently in patients diagnosed with epilepsy than in patients not diagnosed with epilepsy (p<0.0001 for ripples and p=0.04 for spike-ripples) whereas the difference in spike frequency was not statistically significant (p=0.38). The median (range) ratio of spike-ripples to total spike count was 0.21 (0-1) in patients diagnosed with epilepsy and 0.005 (0-0.06) in patients not diagnosed with epilepsy (p=0.0002). Spikes showed a specificity of 69%, ripples of 52% and spike-ripples of 79% for epilepsy, whereas a sensitivity of 48%, 96% and 79% were found for each marker respectively. Conclusions: We were able to demonstrate that out of a cohort of prospectively followed patients after a first unprovoked seizure in childhood, patients that were diagnosed within epilepsy within one year showed significantly more frequent ripples and spike-ripples in early EEG whereas there was no difference in spike frequency. Comparing the diagnostic accuracy in the presence of at least one spike-ripple or at least one spike, we found comparable sensitivity and specificity, but this does not consider the much higher frequency of ripples and spikes ripples in the epilepsy group. Limiting factors in our study were small number of patients and limited follow up period of one year. Development of an automated detection engine for scalp HFOs would allow larger study populations and make clinical use more feasible. Nevertheless, our study suggests that ripples and spike-ripples in scalp EEG recorded shortly after a first unprovoked seizure were found to be more suitable biomarkers than spikes for establishing a prognosis for future development of epilepsy. Funding: No funding