Abstracts

Rationale: Recent studies on temporal lobe epilepsy (TLE) with mesial temporal sclerosis (MTS) have demonstrated widespread grey and white matter (GM/WM) abnormalities within and beyond the affected temporal lobe. However, the topological properties of the TLE anatomical brain network remain largely unknown. A recent study has shown altered network organization in TLE patients using regional cortical thickness correlation (Bernhardt BC et al, Cereb Cortex 2011, 21: 2147-57). The purpose of this study was to investigate the topological organization of the WM brain network constructed from diffusion tensor image (DTI) tractography in TLE patients with left MTS using graph theoretical analysis. Methods: Whole brain DTI and T1 images were acquired from 15 TLE and left MTS (TLE+LMTS) patients (39±12 years, 8 male) and 21 controls (37±12 years, 13 male) on a 1.5T Siemens scanner. Anatomical automatic labeling atlas in standard space was transformed into each individual space (SPM) and used to parcellate the cortical GM into 78 regions. Whole brain WM fibers were obtained by deterministic tractography from DTI (ExploreDTI). The number of fibers connecting any two of the 78 cortical regions and their corresponding mean fractional anisotropy was calculated to construct a weighted WM connectivity matrix (78x78) for each subject. Graph theoretical analysis was used to measure the network topological properties and multiple regression model was used to compare those parameters between groups controlling for age and sex. The network parameters with significant between-group differences were correlated with disease duration and age of seizure onset in patients. Results: The WM network of the TLE+LMTS patients showed significant decreased clustering coefficient (p=0.004), global efficiency (p=0.01) and local efficiency (p=0.019) and increased characteristic shortest path length (p=0.0004) versus controls. Further analysis revealed 12 cortical regions with significantly decreased regional efficiency in TLE+LMTS patients that are mainly in the left temporal lobe (inferior/middle/superior temporal gyri, temporal pole, fusiform and parahippocampal gyri) along with left inferior frontal gyrus, inferior parietal and supplementary motor area and right angular gyrus. Regional efficiency of left inferior temporal gyrus was negatively correlated with disease duration (r=-0.785, p=0.002) and positively correlated with age of seizure onset (r=0.587, p=0.045) and regional efficiency of left parahippocampal gyrus was positively correlated with age of seizure onset (r=0.763, p=0.004). Conclusions: The WM network of TLE+LMTS patients was associated with decreased local and global interconnectivity and increased cost of information propagation mainly in the ipsilateral temporal lobe as well as some areas of left frontal lobe and bilateral parietal lobes. The ipsilateral temporal lobe of patients with long disease duration and early seizure onset tend to have poorer connectivity with other brain regions. Our findings provide evidence of disrupted topological organization of the WM structural network in TLE+LMTS patients.