Abstracts

Rationale: Seizure clustering may provide insight into the mechanisms of brain excitability and inhibition. One possible mechanism for clustering is that seizures occurring during a cluster may not be associated with the typical post-ictal inhibitory responses seen after isolated seizures. If this is true, then the terminal seizure in a cluster should differ from intra-cluster seizures by identifiable characteristics. To test this hypothesis, we examined the seizure characteristics of intra-cluster, terminal cluster and isolated seizures in an epilepsy monitoring database. Methods: Continuous EEG monitoring data for patients admitted to University Medical Center Freiburg 02/2002 - 11/2009 (Epilepsiae database) were analyzed. For each pair of consecutive seizures the inter-seizure interval (ISI) was calculated. Three definitions of clustering were based on ISI: at least 2 consecutive seizures with ISI of less than 2 hr, 4 hr or 8 hr. Using these definitions, we identified clusters of 2 or more seizures, and the terminal seizure in each cluster. Clustered seizures other than the terminal seizure are designated intra-cluster seizures. We examined characteristics of isolated seizures (not part of a cluster), intra-cluster seizures and terminal seizures. Seizures occurring within 24 hours of a rescue medication were excluded, as were episodes of status epilepticus. The relationship between a logarithmic transform of seizure duration (LSD), seizure classification (partial, generalized) and seizure type (intra-cluster, terminal cluster, isolated) was assessed utilizing T-test and Chi Square analysis. Results: The data consisted of 856 days of EEG monitoring from 96 patients (32.3 % intracranial monitoring). Three patients with seizure frequency >3SD from the mean were excluded. Of 1000 seizures available for the analysis, there were 200 8-hr clusters (average 3.8 seizures/cluster), 174 4-hr clusters (3.3 seizures/cluster) and 139 2-hr clusters (2.8 seizures/ cluster). There were no differences between terminal and isolated seizures in duration. Mean seizure duration for intra-cluster seizures using the 2 hr definition were significantly shorter (71 sec) than isolated or terminal seizures (89 sec) (p=0.005). Patterns of results were similar among the 3 definitions of clustering. Test statistics for LSD comparing the intra-cluster vs terminal / isolated seizures were significant for all definitions of clustering (p=0.000 for 2-hr definition, p=0.003 for 4-hr and p=0.043 for 8-hr definitions). Although partial seizures were shorter than generalized seizures, no correlation was found between seizure classification and seizure type in a cluster. Conclusions: A longer seizure duration predicts termination of a cluster and shorter seizures are more likely to sustain a cluster. Seizure classification (partial, generalized) was not associated with the tendency to cluster or terminate a cluster. This study suggests that a short seizure duration may not be sufficient to activate inhibitory mechanisms necessary to prevent recurrent seizures over short time frames. Acknowledgements: EU Project "Epilepsiae", Grant 211713