Abstracts

Rationale: Assessment of memory is important for presurgical evaluation. One noninvasive method to study this is through use of established preoperative fMRI language tasks. FMRI may predict postoperative memory outcome in adults with TLE explaining 10% of the variance. We sought to utilize language fMRI in children to directly probe activation of structures important for memory. We examined if a language task would reliably elicit hippocampal activation in a pediatric epilepsy population. We also compared lateralization of hippocampal activation during this language task between patients and typically developing (TD) controls.Methods: We examined children with left hemisphere focal epilepsy and normal MRI ages 4-12 (n=14; mean age=10.1; SD=2.4) and TD controls (n=23; mean age=8.6, SD=2.4). We used 3T EPI BOLD fMRI with an age-level adjusted auditory description decision task (reverse speech control) known to engage the frontal-temporal language network. Imaging processing and statistical analyses were conducted in SPM8. Image normalization and segmentation was done using subjects high-resolution T1 images with the Dartel Toolbox in SPM8, then applying those parameters to the EPI images. The region of interest (ROI) for bilateral hippocampi was based on the Wake Forest PickAtlas. We calculated a laterality index (LI) for the hippocampal ROI using the LI Toolbox. Results: Individual data analysis showed 91% (21 of 23) of the TD children had activation in the hippocampus, while 71% (10 of 14) of pediatric patients demonstrated activation in the hippocampus (both at p=0.05, uncorrected). The group analysis for the controls showed a cluster of 144 voxels in the left hippocampus (p=0.05, uncorrected), while no significant clusters of activation survived for the patient group map. Controls had greater activation than patients when directly contrasted (two-sample t test, p=0.001, uncorrected). Hippocampal LI was not different between the two groups (p=0.424, Mean LI: Patients = -0.01, TD= -0.01). Categorical distribution of laterality was similar (Patients: 7 bilateral hippocampal activation, 4 left, and 3 right; TD: 11 bilateral, 6 left, and 6 right). Conclusions: Our results demonstrate greater incidence of hippocampal activation during a semantic decision task in TD children than in children with epilepsy. Lower incidence of individual activation of the hippocampus in epilepsy patients may be due to disruption of hippocampal networks. The insignificant epilepsy group results also suggest greater variability of activation within the hippocampus for children with epilepsy. Both TD children and individuals with epilepsy show similar patterns of lateralization for hippocampal activation, with the majority of individuals demonstrating bilateral activation. This bilateral hippocampal activation may be indicative of less material-specific lateralization of memory functions in children, and help to explain why children do not consistently demonstrate material-specific memory deficits following temporal lobe surgery.