Abstracts

Rationale: Temporal lobe epilepsy (TLE) is the most common form of localization-related epilepsy. While TLE originates in the medial temporal lobe, the thalamus is also implicated in seizure propagation and spread. Recent neuroimaging studies provide initial evidence that connectivity between the thalamus and MTL is impaired in TLE patients. Using 7T resting-state fMRI, we aimed to localize such connectivity deficits to specific subregions within the thalamus, in order to pinpoint the pathway of network disruption and identify a robust non-invasive biomarker for TLE. We hypothesized that the disruption would be localized to a subregion of the thalamus which exhibits high structural connectivity to the MTL. Methods: High-resolution 7T T1- and T2-weighted structural MRI and resting state BOLD-fMRI were acquired in 18 adult patients with unilateral TLE and 12 healthy adult controls. MTL subregions (hippocampus, parahippocampus, and amygdala) were segmented using the Anatomical Automatic Labeling (AAL) atlas (Tzourio-Mazoyer et al., 2002). Seven thalamic subregions regions per hemisphere were segmented using the Oxford Thalamic Atlas (Behrens, et al, 2003), which defines subregions based on predominant white-matter connectivity to seven cortical areas. Notably, the temporal-predominant thalamic subregion includes medial dorsal, pulvinar, and anterior nuclei which project to the MTL. Functional connectivity values from each thalamic subregion to the ipsilateral MTL were derived from Fisher-transformed Pearson correlations between the average pre-processed BOLD time series for each subregion. Results: Compared with controls, mean functional connectivity between the temporal-predominant thalamic subregion and the MTL was significantly decreased in the hemisphere ipsilateral to seizure onset (Mann Whitney U test, p = 0.02), but not in the contralateral hemisphere (Mann Whitney U test, p = 0.33). Subsequent analysis revealed that when averaging across all seven thalamic subregions, there was no significant difference in thalamo-MTL connectivity compared with controls, either ipsilateral or contralateral to seizure onset.  Conclusions: Our findings suggest that thalamo-temporal disruption in TLE is confined to the temporal-predominant subregion of the thalamus. This functional localization is consistent with our hypothesis based on known structural connectivity, and lends evidence to the notion that subregion-specific connectivity patterns can be more informative than measures based on the thalamus as a single entity. Our work can pave the way for future larger-scale studies using thalamic connectivity measures as predictive biomarkers of epileptic networks. Funding: NIH grant T32-EB009384, NIH grant K23-NS073801-01, the Mirowski Family Foundation, Thornton Foundation.