Abstracts

Rationale: The most commonly observed MRI findings in temporal lobe epilepsy (TLE) are hippocampal atrophy and T2 signal hyperintensity, the presence of either of which is a sign of hippocampal sclerosis (HS). However, most TLE patients have no obvious hippocampal MRI abnormalities. The hippocampus has many fine structural features that may be seen on high-resolution MR images, but they may be difficult to appreciate and therefore are easily overlooked. Asymmetries of these fine structural details may be evidence of underlying hippocampal abnormalities and thus may suggest the site of seizure onset in TLE patients. An asymmetric loss of differentiation of the laminar hippocampal internal architecture (HIA) has been suggested as a possible sign of hippocampal sclerosis, as has an asymmetric loss of digitation of the hippocampal head (DIG) (Figure 1), but these findings have not been well described in the literature. We examine the relationship between laterality of seizure onset in TLE patients and HIA, DIG, and asymmetric loss of dentation of the ventral surface of the hippocampus (DENT) (Figure 1). Methods: Fifty-four patients were identified who had video-EEG proven unilateral TLE and temporal lobe protocol MRI scans with high-resolution coronal T2-weighted images that were free of significant artifact. Scans were reviewed by a single reviewer blinded to side of seizure onset. Each coronal slice through the body of the hippocampus was evaluated on each side according to a benchmarked 4-point scale of clarity of HIA from 1 = "no differentiation of HIA" to 4 = "Very Clear differentiation of HIA", and an average HIA clarity score was calculated for each side. Coronal images through the hippocampal head were evaluated on each side for the number of visible digitations from 1 to 3. On oblique sagittal images, the degree of dentation of the ventral aspect of the hippocampus was scored on a 4 point scale for each side. Asymmetry scores were calculated for each measure by subtracting left from right. Evidence of HS was also recorded. Asymmetry scores were used in single and multiple logistic regression models to predict laterality of seizure onset for patients without evidence of HS (HS-), patients with evidence of HS (HS ), and both groups combined. Results: When the measures were considered individually, asymmetries of HIA and DIG were each significantly predictive of the side of seizure onset in the entire group and both subgroups (Table 1). An asymmetry of DENT was a significant predictor in the whole group and the HS subgroup, but showed only a trend (p=0.10) in the HS- subgroup. In multivariable models, HIA asymmetry was shown to be the strongest predictor. DIG significantly contributed to the model of the whole group but did not in the HS- subgroup (p=0.09), and the sample size of the HS subgroup did not allow a confident estimate of significance of the multivariable model. DENT did not significantly add to the multivariable model of the combined group or either subgroup. Conclusions: Asymmetries of HIA, DIG, and DENT are subtle imaging findings that may be seen in TLE and indicate the laterality of seizure onset.