Abstracts

Rationale: Our previous studies have suggested a spatial mismatch between the EEG-defined ictal focus and cortical hypometabolism on PET. In the present study, we used a novel multimodal imaging software to perform a detailed quantitative surface analysis of the spatial relationship between ictal intracranial EEG findings and location, extent and severity of cortical hypometabolism in neocortical epilepsy. Methods: Fourteen children (mean age: 9.0 years) with intractable neocortical epilepsy and abnormal interictal FDG PET underwent 2-stage epilepsy surgery. A total of 1199 subdural electrodes (64-106/patient) were placed covering the presumed epileptic cortex. Cortical hypometabolism was identified by landmark-constrained conformal surface mapping of the brain, using co-registered MRI/PET images. The cortical surface was parcellated into triangular finite cortical elements (FCEs, 256/hemisphere) that correspond spatially across subjects. FCEs with hypometabolism, based on lobe- and side-specific asymmetry thresholds (calculated as mean asymmetry indices [AIs] +2SD in normal controls) were defined in all patients. Subdural electrode locations were registered on the 3D brain surface from grid-X-ray images, and detailed spatial relationships between hypometabolic FCEs and ictal subdural EEG (seizure onset, rapid seizure spread) were analyzed. Ictal EEG findings were also evaluated in metabolic “borderzones”, defined by normometabolic FCEs directly surrounding hypometabolic regions; within 2 cm of hypometabolism. Results: The average number of FCEs showing hypometabolism was 41 (16% of the hemispheric surface, range: 5-133), and the average AI for these areas was 14.8%. Cortex underlying seizure onset electrodes showed more severe hypometabolism (mean AI: 10.8%) than cortical areas covered by non-onset electrodes (4.4%, p=0.062) but slightly less hypometabolism than the remaining (non-onset) hypometabolic areas (14.6%, n.s.). There was no significant difference between the extent of hypometabolic vs. normometabolic areas involved in seizure onset (mean number of FCEs: 4.6 vs. 4.9, respectively). Altogether 46±32% of onset electrodes were detected by FDG PET as hypometabolic. Interestingly, an additional 41% of the onset electrodes were located within the metabolic “borderzone” (figure), thus detecting a total of 87±29% of seizure onset electrodes. Electrodes showing seizure onset overlay hypometabolic areas more often than electrodes with seizure spread (77/163 vs. 25/116, p<0.001). Spread electrodes were located preferentially in metabolic "borderzones" rather than in the hypometabolic area itself (51% vs. 22%, p=0.008). Conclusions: These findings provide strong support to the concept that epileptogenic and hypometabolic neocortical areas only partially overlap. Seizure onset often extends from hypometabolic to adjacent normometabolic cortex, while early electrographic propagation prefers surrounding normometabolic regions. The clinical utility of FDG PET in guiding subdural electrode placement could be greatly enhanced by extending grid coverage to at least 2 cm beyond hypometabolic cortex.