Abstracts

Rationale: 2-Deoxy-2[¹⁸F]fluoro-D-glucose (FDG) PET is used to identify potentially epileptogenic brain regions during presurgical evaluation. In the interictal state, hypometabolic cortex is often presumed to encompass the epileptic focus, including the zone of seizure onset. However, hypometabolic regions are generally thought to overestimate epileptic regions, and the reported accuracy of FDG PET to identify epileptic cortex varies widely across studies. In this study, we have used FDG PET combined with intracranial EEG to identify factors associated with poor match between metabolic and ictal epileptiform abnormalities in neocortical epilepsy. Methods: Hypometabolic cortex was delineated by an objective, asymmetry-based approach (>10% decrease of FDG uptake) on the affected side in 36 consecutive young patients (age 8.5±5.4 years) who had abnormal interictal FDG PET and underwent epilepsy surgery due to intractable neocortical epilepsy. All patients had implantation of subdural electrode grids (total number of electrodes: 52-116) covering multiple lobes including all or most of the hypometabolic regions and adjacent areas. Electrodes with seizure onset were identified and correlated with location and extent of hypometabolic cortex using 3D surface-rendered MRI/PET/grid images. Results: In general, only a small portion of hypometabolic cortex showed seizure onset: hypometabolic areas were covered by 4.9 ± 5.5 electrodes with seizure onset vs. 17.0 ± 13.8 electrodes with no seizure onset (p<0.001). Almost half of the seizure onset electrodes (4.5±4.7 electrodes per patient) were located adjacent to hypometabolic regions. The seizure onset area was completely detected by FDG PET in 9/36 (25%) but entirely missed (but located in adjacent cortex) in 10 patients (28%); these latter patients had more recent seizures (within 12 h in all but one) before PET than those in whom hypometabolic cortex overlapped with the seizure onset region (p=0.002). Although the number of implanted electrodes was not different between left and right foci, focus side had a major impact on the extent of hypometabolism: an average of 27 (±16) electrodes on the left side (n=24) but only 10 (±10) on the right (n=12; p=0.002) covered hypometabolic cortex. Patients with left foci also had younger age at seizure onset (2.8 vs. 5.0 years; p=0.04) and less likely showed frequent spiking on PET-EEG (21% vs. 75%; p=0.046) than those with right foci. Age, duration of epilepsy and pathology had no significant impact on metabolism/EEG relationships. Conclusions: The major portion of hypometabolic cortex is often NOT involved in seizure generation, while seizure onset commonly occurs, partly or completely, in cortex adjacent to hypometabolic regions. Seizure(s) shortly (<12 h) before PET appear to aggravate this mismatch. Left hemispheric foci, young age at seizure onset and lack of frequent spiking on PET-EEG are associated with extensive non-epileptogenic cortical hypometabolism; consideration of these factors may improve interpretation of FDG PET findings for presurgical localization of neocortical epileptic foci.