Abstracts

Rationale: The medial frontal cortex is often involved in frontal lobe seizures. Cortical regions subserving neurological or cognitive function can be spatially reorganised in the context of chronic epilepsy and abnormalities of white matter tracts can be found, even when conventional structural MRI is normal. Using Diffusion Tensor Imaging (DTI) and tractography, we investigated how frontal lobe epilepsy (FLE) affects the anatomical connectivity of the SMA as defined by motor fMRI. Methods: We studied 36 patients with FLE and 16 healthy controls. FLE was diagnosed as left FLE (L-FLE) or right FLE (R-FLE) on the basis of non-invasive electro-clinical and radiological data. No patient had a visible lesion near the supplementary motor area (SMA) on MRI. All subjects underwent a visually-cued motor functional MRI (fMRI) task followed by DTI on a 3T MR scanner. Statistical parametric mapping of fMRI data yielded bilateral activation in the medial frontal lobe. After coregistration of the fMRI space to the DTI space, the gyrus corresponding to maximal SMA activation was identified in each hemisphere and a seed region (sphere of 6mm radius) was placed in the underlying white matter. Probabilistic tractography was run from every seed voxel using PiCo (Parker, G. J., H. A. Haroon, et al.; 2003; J Magn Reson Imaging 18(2): 242-54) to create connectivity maps that were thresholded to keep voxels with a probability of connection >10%. Mean Fractional Anisotropy (FA) and Mean Diffusivity (MD) of the tracts were calculated. Paired t-tests were used to assess asymmetry in FA and MD in controls. A one-way ANOVA was used to compare results across hemispheres (within-subject effect : L-SMA, R-SMA) with post-hoc Dunnett t-tests to compare results across groups (between-subject effect : controls, L-FLE, R-FLE). Results: In controls, mean FA was greater in the left hemisphere (mean 0.37 vs 0.34, p=0.02). There was no interhemispheric difference in MD. The asymmetry in FA was confirmed in the ANOVA which showed a significant effect of the hemisphere (L-SMA > R-SMA, p=0.001) but not of the group (controls vs L-FLE vs R-FLE, p=0.241). There was no significant interaction between hemisphere and group. Conclusions: Our results suggest a stronger SMA connectivity in the left hemisphere. In FLE, the anatomical connectivity of the functionally active SMA appears preserved. This has important implications for epilepsy surgery of the medial frontal cortex.