Abstracts

Rationale: Seizure semiology is widely accepted as being an important in the presurgical evaluation of patients being considered for resective epilepsy surgery. Whereas individual lateralizing signs (LS) have been previously evaluated for inter-rater reliability and positive predictive value (PPV) they have not been studied in aggregate as patients may exhibit more than one lateralizing sign. Furthermore, the value of semiology to localize the ictal onset zone within a hemisphere has not been measured objectively. Our aim was to objectively study the lateralizing and localizing value of seizure semiology. Methods: 73 consecutive patients who underwent resective epilepsy surgery between 1999-2007 with Engel class Ia seizure outcome at 1 year were included. Patients who had multilobar resection, hemispherectomy and those younger than 12 years old were excluded. Three observers (PK, AA, DS) were asked to analyze video files blinded to clinical data. They were asked to lateralize seizures based on all available LS including description of auras and neurological deficits. In addition they attempted to localize the seizures to one lobe and within each lobe based on predefined criteria. Agreement of 2 or more observers was required in order to score a particular sign or assign localization. The lateralizing and localizing value of semiology was compared with findings from other presurgical tests (interictal/ictal EEG, MRI, PET). Results: Table 1 shows the interobserver agreement (Kappa index) and PPV for various lateralizing signs. 19/30 (63%) temporal lobe seizures were correctly lateralized (k=0.64) with correct sublobar localization in 27/30 (90%) patients, k=0.6. Sublobar localization was correct in 12/28 (43%) of patients, k=0.24 (sublobar localization not possible in 2 patients). 20/27 (74%) frontal lobe seizures were correctly lateralized (k=0.55). Lobar localization was correct in 21/27 (78%) patients, k=0.41 and sublobar localization was correct in 5/25(20%) patients, k=0.36 (sublobar classification was not possible in 2 patients). 7/8 (87%) parietal lobe patients were correctly lateralized, k=0.83. Lobar localization was correct in 3/8 (37%) patients, k=0.5. 7/8 (87%) occipital lobe patients were correctly lateralized (k=0.67). Lobar localization was correct in 7/8 (87%) patients, k=0.67. Figure 1 shows the lateralizing and localizing value of seizure semiology compared with other diagnostic modalities. Conclusions: Seizure semiology has good lateralizing value particularly in extratemporal epilepsy. Version, focal clonic movements and Figure 4 posturing had high inter-rater agreement and PPV. Localization was highest in temporal and frontal lobe seizures followed by occipital and least for parietal lobe seizures. Semiology appears to have comparable value in terms of lateralization and localization with other diagnostic modalities (MRI, EEG, PET, or SPECT).