Abstracts

Rationale: Epilepsy research has increasingly focused on cortical, subcortical and functional brain abnormalities observed in patients. However, the understanding of how these parameters relate to each other and to patient s neuropsychological performance remains limited. In this study, we explored the relationship among hippocampal functional connectivity (FC), cortical volume, white matter diffusivity (WMD) and gray-white matter contrast (GWMC) across patients with temporal lobe epilepsy (TLE) and healthy controls. Starting from a universe of parameters related to default mode network integrity and connectivity, we distilled our data into 4 factors representing the gray matter (GM), WMD, hippocampal FC and GWMC characteristics in the whole brain, and quantified i) the correlation among these factors ii) the ability of these factors to explain the variation in naming and verbal learning/memory performance (i.e., Boston Naming Test (BNT), Auditory Naming Test (ANT), Verbal Paired Associates (VPA) Test).Methods: T1-, diffusion-weighted MRI and fMRI data were collected from 14 healthy controls and 14 patients with TLE. Functional connectivity was derived from hippocampal seed-based analysis of event-related activation in a semantic decision task. Each dimension was separately reduced to its principal component, where KMO (a measure of sampling adequacy) were restricted to > 0.75 and individual parameter loadings were restricted to > 0.8 to ensure meaningful results with a small sample. Resulting principal components representing GM, WM, FC and GWMC were correlated with each other and standardized scores from BNT, ANT and VPA-2 (delayed recall).Results: Factors representing GM area and GWMC showed significant Pearson s correlation (r = .709; p < .001) while those representing WMD and hippocampal FC appeared correlated (r = .440, p < .05). Spearman s correlations confirmed the former to be a linear relationship (p < .001) while the latter appeared to be non-linear and only approaching significance (Rho = .337, p = .07). These factors explained 56.7% of the variation in VPA-2 scores; 32.4% of the variation on BNT scores and 35.7% of the variation in ANT scores. GM and FC were significant in predicting ANT performance, while WM and FC were significant in predicting BNT and VPA-2 performance.Conclusions: Structural and functional measures including GM area, WMD, GWMC and hippocampal FC are unlikely to represent distinct, orthogonal vectors of disease in individuals with epilepsy. Instead, they are likely to be driven by complex pathological processes that we cannot measure directly, and therefore likely to be correlated among themselves. Our results suggest that FC may be more related to WM integrity, while GWMC more closely follows GM changes, which is consistent with previous reports. Yet each of these measures plays a distinct role in explaining verbal memory and language performance in TLE. This underlines the importance of combining and contrasting structural and functional measures to better predict cognitive performance in TLE.