Abstracts

Rationale: Automatic MRI morphometry has improved the understanding of medial temporal lobe epilepsy (MTLE) by showing the location and extent of extra-hippocampal gray matter atrophy, and its relationship with severity of epilepsy and cognitive impairment. The two most common forms of whole-brain MRI morphometry are voxel-based morphometry (VBM) and cortical thickness analysis (CTA). VBM relies on the voxel-wise quantification of the probability of gray matter volume (GMV); while CTA focuses on the tridimensional estimation of the distance between the gray/white matter transition and the pial surface. In theory, both quantify the regional content of gray matter, but results from both methods are consistently somewhat discrepant. We investigated morphometrical abnormalities associated with MTLE as measured by CTA and VBM in one sample of patients with MTLE. We compared both methods regarding the relationship between seizure-burden and structural abnormalities. We theorized that both methods may be sensitive to different features of atrophy. Methods: We studied 23 patients with MTLE and unilateral hippocampal atrophy, and 34 matched healthy controls. All subjects underwent MRI in a Philips 3T scanner yielding a high resolution T1 image with 1mm isotropic voxels. VBM involved iterative spatial normalization, segmentation, bias field correction, spatial smoothing (10mm), and mean GMV extraction from regions of interest (ROI) comprised in the Anatomic Automatic Labeling (AAL) atlas. Cortical thickness (CT) was computed using FreeSurfer, involving segmentation of white matter, tessellation of gray-white matter boundaries, inflation of the folded tessellation and automatic correction of topological defects, with subsequent calculation of CT. Mean regional CT estimates were computed for gyral-based ROIs derived from the AAL. MANOVA was used to assess regional differences between patients and controls. Simple correlation was used to investigate the relationship between seizure duration and CT and GMV. A level of statistical significance was set at p<0.05. Results: Patients with MTLE exhibited a significant decrease in GMV and CT (representative regions shown in Table 1). There was also a significant relationship between seizure duration and regional CT and GMV (Table 2). Notably, group-wise CT differences were not as prominent as GMV differences. Conversely, CT was more strongly related to duration of seizures. Conclusions: Both CTA and VBM revealed significant regional atrophy in patients with MTLE involving hippocampal, medial temporal and limbic structures. However, VBM results emphasized atrophy when subjects with MTLE were compared with controls. Conversely, CTA was more sensitive to atrophy related to duration of seizures. We postulate that these discrepancies can be due to (1) intrinsic methodological limitations or (2) distinct sensitivity of each method to different biological phenomena. We speculate that CT may reduce initially as a consequence of architectural changes of the cortex, which are followed by atrophy and reduction of GMV, when neuronal loss ensues.