Abstracts

Rationale: Structural atrophy indicated by decreases in gray matter compared to controls has been detected across the brain in temporal lobe epilepsy (TLE) using MRI. TLE has also been associated with changes in functional connectivity using resting functional MRI. We hypothesized that widespread networks affected by TLE would include regions where gray matter decreases and functional connectivity are linearly related. Our goal was to identify individual brain regions where gray matter decreases correlated with functional connectivity in a group of left TLE patients, and to determine whether any of these regions formed synchronous multiregional networks unique to left TLE. Methods: In a group of 15 electrographically determined left TLE patients (35 ± 10 yrs, 10 F, 1 left-handed), we performed 3T MRI imaging including T1-weighted structural (ultra-fast gradient echo, 1 mm3) and resting functional scans (3.75 x 3.75 x 5 mm, TE=35 ms, TR = 2 s, 200 dynamics). The T1-weighted images were used to perform optimized voxel-based morphometry [1] to quantify voxel-wise gray matter concentration (GMC). The functional MRI images were used to calculate voxel-wise functional connectivity (FC) maps to the left hippocampus (LH) and to the left thalamus (LThal) using the general linear model [2]. Regions of linear correlation between the GMC and FC map to the LH were found using biological parametric mapping (BPM) [3], a linear modeling method for comparing two image sets. The BPM was repeated to find linear correlations between the GMC and FC map to the LThal. Finally, we looked for FC connections among these regions that were correlated with each other to form multiregional networks. To do this we assumed that the LH to LThal connection was the hub of the network and determined which other connections were correlated with it across the group. We also compared the FC of the same connections in 15 healthy controls. Results: Twenty regions had GMC correlated with FC to either the LH or LThal. We found 5 of these connections correlated with the connection LH - LThal (conn A) across the patient group (Figure 1). These connections include: the LH and the anterior cerebellum (conn B), LH and the right temporal lobe (conn C), LThal and the midcingulate gyrus (conn D), LThal and the left postcentral gyrus (conn E) and LThal and the right postcentral gyrus (conn F). This means that in patients where the strength of connection A was high, the strength of these other connections were also high; and those in which the strength of connection A was low, these others were low. This relationship was not found in controls. Conclusions: We performed voxel-wise regression to determine regions of correlated GMC decreases and FC in left TLE. From these regions we identified a multiregional network of synchronous connections unique to the patient group which may reflect seizure propagation in left TLE. NIH T32 EB001628, R01 NS055822 and UL1 RR024975-01 [1] Ashburner J. 2000. NeuroImage 11(6):805-821. [2] Morgan VL. 2012 Epilepsia 52(9):1741-1749. [3] Casanova R. 2007. NeuroImage 34(1):137-143.