Abstracts

Rationale: Although prior work has evaluated the predictablity of epilepsy surgical outcome by proton spectroscopic imaging (1H-MRSI) and FDG-PET in mesial temporal lobe epilepsy (MTLE), none has examined the same group of patients with both imaging modalities. Methods: MRSI and FDG-PET were acquired prospectively in eighteen patients with unilateral MTLE defined by video and ictal EEG. Hippocampal 1H-MRSI was acquired at 4.1T using a 2D point resolved spectroscopy (PRESS)(point resolved) pulse sequence. Hippocampal regions of interest were manually drawn on the FDG-PET scans and normalized to whole brain counts. All anterior mesial temporal resections were performed by the same surgeon using a standard subpial resection technique. Surgical outcome data were available for at least one year in all patients. Outcome was defined as seizure-free or not (including patients both with and without persistent auras still considered to be seizure-free). Prediction of seizure-free outcome was tested with Fisher's Exact test using statisticlly significant lateralization of the imaging modalities. Results: Neither 1H-MRSI, FDG-PET, or the combination of the two was predictive of seizure-free outcome. This was the case whether or not persistent auras were included. However, 1H-MRSI was nearly predictiive of seizure-free outcome (P=0.0586) when patients with persistent auras were included as seizure-free. Discussion: In contrast to relative hippocampal atrophy, lateralized 1H-MRSI and FDG-PET measures from in the hippocampus are not predictive of seizure-free outcome. This finding supports the hypothesis that these in vivo imaging methods reflect different functional or metabolic disturbances epileptogenic tissue that are both independent of structural abnormality, abnormality that typically predicts successful surgery in partial epilepsy.