Abstracts

Rationale: In a recent study of 60 epilepsy patients undergoing left anterior temporal lobe (ATL) resection (Binder et al., Epilepsia, 2008;in press), we showed that postoperative verbal memory change on a list-learning test was predicted by preoperative language lateralization as measured by fMRI (r = -.43, p <.001). Here we compare this method with an fMRI task that activates the anterior hippocampus bilaterally in healthy adults (Binder et al., Epilepsia, 2005;46:1061-70). Methods: 65 consecutively encountered TLE patients (30 left and 35 right) underwent ATL resection, preoperative language fMRI, preoperative hippocampal fMRI, and verbal memory testing before and 6 months after surgery. Language fMRI used a semantic decision - tone decision contrast. A language lateralization index (LI) was computed for each patient based on the volume of activation in each hemisphere. Hippocampal fMRI used a depth-of-processing (meaningful vs. nonsense) contrast with complex visual scenes. An anterior hippocampal LI was computed based on volume of activation in an anterior hippocampus region of interest. Consistent long-term recall (CLTR) on the selective reminding test (SRT) was used as an index of verbal memory. Change scores on this test were computed as the difference between postoperative and preoperative scores. Results: Approximately 30% of the left ATL patients declined significantly on the CLTR index, compared to only 5% of the right ATL group. Preoperative language LI predicted CLTR change scores in the left ATL patients (r = -0.39, p = .02). Anterior hippocampal activation was detected preoperatively in 29/30 left and 34/35 right ATL patients. The preoperative anterior hippocampal LI was significantly different in the left (mean = -.32) and right (mean = +.20) ATL groups (p = .001) and correctly predicted the side of seizure focus in 70% of cases. Despite this evidence of functional lateralization, the anterior hippocampal LI was not correlated with CLTR change in the left ATL patients (r = -0.04, n.s.) or with change on any other SRT index. Conclusions: Prior efforts to predict memory outcome in ATL patients with fMRI have focused mainly on asymmetry of activation in the medial temporal lobe. Our results show that hippocampal activation asymmetry is correlated with the side of seizure focus but is not a good predictor of memory outcome. We propose that this activation, like the asymmetry observed in many Wada memory protocols, is insufficiently material specific. These visual encoding tests measure overall asymmetry of the episodic memory system but do not distinguish between verbal and nonverbal encoding processes. Language lateralization is a more accurate predictor of outcome because it is more closely tied to lateralization of verbal episodic memory encoding processes.