Abstract
Introduction:
Quantitative MRI studies have consistently demonstrated that drug-resistant temporal lobe epilepsy (TLE) is associated with multilobar cortical thinning affecting lateral temporal and fronto-central regions (Bernhardt et al. 2010). While diffusion weighted imaging studies have also reported widespread white matter (WM) changes (Otte et al. 2012), their spatial relationship to cortical pathology is unclear, possibly given the targeted sampling of deep fibre tracts. Moreover, metrics of grey matter (GM) morphology and microstructural WM integrity have not been evaluated conjointly. Here, we mapped diffusion properties of the immediate subcortical WM and cortical GM thickness in a unified surface-based framework.
Methods:
We studied 61 drug-resistant TLE patients (34±9 years, 46 males; 31/30 left/right TLE) and 42 healthy controls (30±7 years, 21 males). Participants were scanned on a 3.0T Siemens TimTrio MRI. Based on T1-weighted MRI (3D-MPRAGE, 1x1x1mm3 isotropic voxels), we generated cortical surface models and measured cortical thickness across 80k vertices. We computed a Laplacian potential field between the cortex and ventricles, which guided the placement of subcortical surfaces at 1, 2 and 3 mm below the GM-WM boundary, with intrinsic vertex-correspondence to the overlying cortex. Based on co-registered diffusion MRI (64 directions, b=1000 s/mm2, 2x2x2mm3 isotropic voxels), we sampled vertex-wise mean diffusivity (MD) and fractional anisotropy (FA) on subcortical WM surface. Hemisphere-specific measurements of left and right TLE patients were normalized with respect to controls, and sorted into ipsilateral/contralateral to the seizure focus. Following correction for age and gender, we carried out vertex-wise t-tests to detect cortical thickness and subcortical WM diffusion differences between patients and controls across surfaces. Findings were corrected using random field theory at FWE<0.05.
Results:
Relative to controls, TLE patients presented with diffuse bilateral cortical thinning in anterior temporal, frontal, and centro-parietal regions; conversely, subcortical WM diffusion changes (MD increase and FA decrease) were more restricted, and largely limited to ipsilateral limbic cortices, including the parahippocampus, anterior cingulate, lateral temporal, and orbitofrontal regions (Fig 1). Diffusion abnormalities were similar at 1, 2, and 3 mm depth. Within the temporal lobe, MD and FA changes overlapped with cortical thinning in parahippocampal and lateral cortices; in the frontal lobe, overlaps were observed in dorsolateral prefrontal and frontopolar regions.
Conclusions:
Our study compared for the first time point-wise patterns of cortical and subcortical WM structural alterations in TLE and showed a differential distribution of anomalies. Predominantly ipsilateral limbic diffusion alterations may reflect gliosis and/or demyelination due to mesiotemporal deafferentation; on the other hand, extensive and multi-lobar bilateral GM atrophy is likely secondary to excitotoxicity in neocortical circuits affected by seizure spread.
