Abstracts

This study investigates the value of preoperative [F-18]fluorodeoxyglucose (FDG) brain PET imaging in predicting outcome with respect to seizure control in epilepsy patients who undergo temporal lobe surgery. Results obtained using visual image inspection are compared with those of statistical mapping.
Eighteen epilepsy patients (age 20-59 years) with discordant semiology, EEG, and MR findings underwent standard FDG PET brain imaging preoperatively. For statistical mapping, individual PET images were compared to those of normal adult controls (n=22, age 37 [plusmn] 9 years) using pixelwize Z-score mapping (abnormal pixels with p[lt]0.05) following anatomic standardizaton and extraction of gray matter activity using 3-dimensional stereotactic surface projections (3D-SSP). A metabolic pattern typical of temporal lobe epilepsy (TLE) was established separately from a group of patients who underwent temporal lobectomy with Engel Class I outcome. Based on PET patterns of metabolic activity, individual cases in the study group were categorized as having a) findings typical of TLE (metabolic reductions in the medial and lateral temporal cortices, inferior frontal cortex, and thalamus) and b) atypical of TLE (abnormalities not limited to the above structures). Conventional transaxial PET images were interepreted visually by multiple observers and compared with results of statistical mapping. Post-operative outcomes following resection of temporal cortices suspected to have seizure foci were classified as seizure-free, worthwhile improvement ([gt]75% seizure reduction), or no substantial improvement ([lt]75% reduction).
All 5 patients with statistical PET patterns consistent with typical TLE became seizure-free postoperatively after corresponding temporal lobe resection. Visual reading correctly identified the affected temporal lobe in 3 of 5 patients. Of 13 patients having statistical PET patterns atypical for TLE, none became seizure-free, and 9 of the 13 had no substantial improvement regardless of the side resected (chi-square p[lt]0.01 for group difference). Of these, 3 patients who had diffuse cortical metabolic abnormalities (Z score [gt]5) atypical for TLE had no substantial improvement postoperatively. Visual reading identified these cortical abnormalities in only one patient. MRI findings showed no temporal abnormality in 3 of 5 patients with seizure-free outcome and 7 of 9 patients with no substantial improvement.
Refractory epilepsy patients who demonstrated FDG PET findings atypical for TLE had significantly less postoperative improvement in seizure control as compared to those having PET patterns typical for TLE. Use of statistical mapping improves predictive outcome in patients with findings atypical for TLE as compared to visual PET image interpretation. FDG PET may be useful in identifying patients who may not benefit from temporal lobe resection.
[Supported by: University of Washington Radiology Research Fund]