Abstracts

Rationale: To compare the value of 5HT1A receptor PET and FDG-PET for temporal lobectomy planning.Methods: 41 patients had surgery with at least one year follow-up. PET was performed with 18F-FCWAY for 5HT1A receptor binding estimation and 18F-FDG for glucose metabolism (CMRglc). After partial volume correction, 5HT1A receptor binding and CMRglc were measured in regions drawn on coregistered MRI. Mean regional asymmetry values, and the number of regions exceeding mean +/- 2 standard deviation asymmetry in 16 healthy volunteers were compared between seizure-free (SF) and not seizure-free (NSF) patients using SYSTAT (Systat Software, Inc. Chicago Ill). Results: Twenty-six of 41 (63%) patients were SF since surgery. SF patients had lower group mean 5HT1A binding in hippocampus (p=0.05), amygdala (p=0.02), and parahippocampal gyrus (p=0.05). SF patients had more individual regions with reduced binding ipsilateral than contralateral to the resection (p<0.02); there was a trend for patients with no contralateral binding reduction to be more likely to be SF. The positive predictive value for SF outcome was 71% with any ipsilateral binding reduction and 77% with more ipsilateral than contralateral regions with reduced binding. SF patients had mean CMRglc reduced in amygdala (p=0.07), and significantly more individual hypometabolic regions (p<0.01). The positive predictive value for SF outcome was 76% with any ipsilateral hypometabolism and 77% with more ipsilateral than contralateral hypometabolic regions. 15 of 41 patients had normal MRI; 10 were SF. SF patients had lower group mean 5HT1A binding in hippocampus (p<0.07), amygdala (p<0.01), parahippocampal and fusiform gyrus (p<0.03). There was a trend for SF patients to have more individual regions with reduced binding ipsilateral than contralateral to the resection. The positive predictive value for SF outcome was 86% with any ipsilateral binding reduction and 100% with more ipsilateral than contralateral regions with reduced binding. No region had significant group mean CMRglc reductions. There was a trend for SF patients to have more individual regions with reduced CMRglc ipsilateral to the resection. The positive predictive value for SF outcome was 69% with any ipsilateral hypometabolism and 70% with more ipsilateral than contralateral hypometabolic regions. Conclusions: Our study suggests that 5HT1A receptor PET does not have a general advantage over FDG-PET for prediction of outcome after temporal lobectomy, but might be more valuable if MRI is normal.