Abstracts

Rationale: Genetic generalized epilepsy (GGE) is classified into several syndromes based on clinical criteria and characteristic electroencephalographic (EEG) signatures. The clinical criteria are distinct, but the EEG differences are less well defined. In the International League Against Epilepsy (ILAE) classification, spike-wave discharge frequency with normal background is taken into consideration in differentiating GGE syndromes. Yet, this distinction is not explicit. Childhood absence epilepsy (CAE) and juvenile absence epilepsy (JAE) are clinically characterized by predominant absence seizures. Both conditions demonstrate paroxysms of generalized spike-wave discharges on the EEG. We hypothesized that duration of generalized paroxysms differs between CAE and JAE. Methods: We performed a cross-sectional study of consecutive patients with CAE and JAE recruited from two epilepsy clinics, diagnosed and classified according to the ILAE criteria. All had 24-hour ambulatory EEG recordings. All EEGs were visually analyzed by the same investigator. Epileptiform discharges were classified as generalized paroxysms (GP) (duration ?-2 sec) and fragments (duration < 2 sec). The duration of each discharge was manually measured in seconds. We tabulated the duration of each GP of all 24-hour EEGs. Additionally, we calculated the mean GP duration for each individual patient followed by Mann-Whitney U Test to explore the significance of difference in GP durations between JAE and CAE. We employed receiver operating characteristics (ROC) curves to study the overall diagnostic accuracy of GP duration at multiple thresholds to distinguish JAE from CAE. ROC curve of GP duration was constructed by plotting sensitivity against 1-specificity at a range of thresholds. We used area under the curve (AUC) as a measure of overall diagnostic accuracy. The AUC was interpreted as follows; 0.5, discrimination of JAE from CAE is no better than chance; 0.6 -0.69, poor discrimination; 0.7-0.79 fair discrimination; 0.8-0.89, good discrimination; and 0.9-1, outstanding discrimination. We constructed ROC curves at two levels; using mean GP duration of each subject as test variable (subject-level) and the duration of each unit of GP from all subjects (paroxysm-level). The data analyses were performed with IBM SPSS (version 21) statistical software package (IBM Corporation, New York, USA). Results: We studied a total of 987 paroxysms (CAE-197, JAE-790). There were 15 CAE and 34 JAE patients in the cohort with mean ages of 27.89.2 and 2811.0 respectively. 93.3% of CAE patients and 82.4 JAE patients were on antiepileptic drug therapy at the time of EEG recording. The median durations of seizure-freedom in CAE and JAE groups were 90 and 165 days respectively. The median duration of GP at subject-level was significantly higher (p=0.04) in JAE (3.2 sec) compared with CAE (2.6 sec). The ROC curves yielded AUCs of 0.75 (95% CI 0.57 to 0.92) at subject-level and 0.71 (95% CI 0.68 to 0.74) at paroxysm-level (Figure 1). With increasing GP duration, diagnostic specificity for JAE increased and sensitivity decreased. Accordingly, a GP lasting ?-2.9 seconds can be diagnosed as JAE with 66% sensitivity and 67% specificity (Table 1). Conclusions: Our results indicate that GP duration is a fairly accurate test to distinguish JAE from CAE. The ROC curve analysis also provides different thresholds of GP durations with varying sensitivities and specificities. These differences between CAE and JAE may be reflective of changes in intrinsic epileptogenicity. Funding: nil