Abstracts

Rationale: A wide range of procedures, including interictal and ictal EEG recording, MRI and FDG-PET has been used to identify the seizure focus and predict the chance of seizure freedom after surgery in patients with temporal lobe epilepsy. Some studies have risks including radiation exposure and injury due to seizures, as well as considerable expense. The incremental value of performing multiple studies to attempt to predict seizure freedom after surgery is uncertain.Methods: We performed a retrospective analysis of presurgical, surgical, and postsurgical data for 99 consecutive anterior temporal lobectomy patients (mean age 32 +/- 9.5 years; range 13 to 56 years). Mean age of seizure onset was 11.5 +/- 8.3 years. Mean follow up was 70 months. All patients had surface interictal EEG and ictal video-EEG monitoring, 97 MRI brain imaging, 68 FDG-PET, and 41 extra-operative intracranial EEG monitoring. We recorded surgery outcome at one year and at the most recent follow-up visit. Data analysis was performed with Systat (Systat Inc., Chicago Ill). Results: Seventy-six percent of patients were free of seizures impairing consciousness at one year, and 75% at most recent follow-up. Sixty-four patients had either lateralizing or localizing MRI findings corresponding to the resected seizure focus. Forty had lateralizing or localizing (18F)-FDG-PET. Seventy-one and 72 patients had interictal and ictal surface EEG localizing or lateralizing findings, respectively. Only interictal surface EEG was significantly associated with seizure freedom at both one year and most recent follow-up (p<0.01). There was a trend for FDG-PET (p<0.10). MRI, surface ictal EEG, and invasive EEG did not have a significant relation to outcome. Surface ictal EEG had the highest sensitivity for seizure-free outcome, but FDG-PET the highest specificity. Combining studies did not improve results. Conclusions: Our study shows the difficulty of assessing the value of diagnostic procedures in a clinical care model. Failure to show a significant relation of MRI to outcome may have been due to a high percentage of patients with normal MRI who did well. Selection bias may have influenced assessment of the value of video-EEG monitoring, since only a small number of patients with negative findings had surgery. However, our data suggest preoperative localizing studies have considerable redundancy, and that ictal video-EEG monitoring may not be needed for all patients considered for temporal lobectomy.