Abstracts

Rationale: Hippocampal sclerosis (HS) subtype is an important predictor of surgical outcome in temporal lobe epilepsy (TLE). However, it is currently not possible to diagnose HS subtypes (which hippocampal subfields are involved) with conventional in vivo magnetic resonance imaging (MRI). High-resolution (1 mm isotropic) diffusion tensor imaging (DTI) of the hippocampus has shown patterns of hippocampal subfield diffusion abnormalities in TLE patients which were consistent with HS subtype on surgical histology [1]. Longitudinal imaging is required to determine the stability of these subhippocampal diffusion changes over time in both patients (including post-surgery) and controls.

Methods: Sixteen TLE patients (mean age 43±13 years) and 9 healthy controls (mean age 40±13 years) underwent 2 high-resolution DTI scans, 2.3±0.4 years apart. Diffusion images were acquired on a Siemens Prisma 3T with single shot 2D EPI, 20 slices aligned along long axis of hippocampus, 1x 1 x 1 mm³ resolution, 10 averages of 10 gradient directions at b=500 s/mm² and 10 b0s acquired in 5:18 minutes. Patients were subdivided into non-HS (n=5), unilateral HS (n=9) and bilateral HS (n=2) based on clinical MRI. Six patients (1 non-HS, 5 unilateral HS) had undergone anterior temporal lobe resection after their initial scan providing post-surgical histology (type 1 HS in 2 and type 2 HS in 4 subjects; one of which was reported in our previous study [1]). The hippocampus was manually segmented on mean diffusion weighted images in native space to yield volume, mean diffusivity (MD) and fractional anisotropy (FA) which are presented individually for each hippocampi. Regional MD maps (color coded) were compared between the scans to assess consistency of regions of abnormal elevated MD.

Results: Repeated measures ANOVA showed significant effect of hippocampus group (healthy, non-HS, HS and contralateral hippocampus of surgical TLE) for volume (p < 0.001), FA (p=0.002), and MD (p < 0.001). At both scans, sclerotic hippocampus had 22-44% higher MD, 5-12% lower FA and 50-67% lower volume, compared to controls and non-HS hippocampi (Fig. 1). No significant effect of time or interaction of time was observed for any group. Good correspondence was observed between the regions with higher MD and hippocampal subfields pathology in 5 of 6 surgical patients. MD elevated hotspots were present in the ipsilateral hippocampus of unilateral HS TLE and were detected consistently at both scans ~2.5 years apart (Fig. 2 B). The contralateral hippocampus of the TLE surgical group showed little to no observable regional MD change from the pre-surgical scan (Fig. 2 C).