Abstracts

Rationale: Temporal lobe epilepsy (TLE), the most common drug-resistant epilepsy, is characterized by hippocampal sclerosis. Previous studies have shown distributed alterations of structure and function beyond the hippocampus, collectively establishing TLE as a system-level disorder (1). Yet, the underlying interactions between cortical structure and networks are still poorly understood. Classical neuroanatomical studies suggest that the human brain is hierarchically organized with information flowing from unimodal sensory regions through heteromodal areas to reach higher-order transmodal cortices (2). Recently, MR-based stepwise functional connectivity (SFC;3) has allowed the study of interactions and transitions between systems, thereby recapitulating this hierarchical view. Our aim was to map interactions of large-scale networks in TLE. We hypothesized that information processing is altered along the cortical hierarchy and dysfunction is coupled to structural anomalies. Methods: We studied 72 TLE patients (24 male, age=369) and 41 age and sex-matched healthy controls using high-resolution 3D T1-weighted and 2D resting-state fMRI (rs-fMRI) acquired at 3 Tesla. For each subject, after parcellating the brain into 1,020 regions(4), we sampled rs-fMRI timeseries along the mid-thickness cortical surface and computed pairwise Pearson’s correlations between parcels to generate connectivity matrices (1,020×1,020; thresholded at top 10 percentile per row). We used the SFC technique to calculate the number of connections (degrees) linking seed regions (unimodal sensory areas) to the rest of the brain at connection lengths or steps, ranging from 0 (i.e., the seed) to 100 steps away(5). After normalizing SFC degrees based on the Yeo-Krienen’s networks(6), we calculated connectivity peaks (to quantify connectivity strength) and areas under the curve (AUC; to quantify transitions). Linear models assessed group differences. Finally, we compared point-wise differences in cortical thickness along the cortical hierarchy trajectory. Findings were corrected for multiple comparisons using FDR at q<0.05. Results: In line with previous reports, SFC in healthy controls (Fig. 1A, upper panels) transitioned from unimodal sensory cortices to heteromodal areas (steps 50 to 60), to converge to transmodal areas of the DMN (steps 60 to 70). Conversely, in TLE (Fig. 1A, lower panels), connectivity was segregated in unimodal sensory areas (until step 60), with limited links to heteromodal regions, and slowly converged to DMN and limbic networks. Figure 1B illustrates the trajectories of the SFC in relation to the Yeo-Krienen’s functional networks. Compared to controls, in TLE AUCs were wider in primary sensory, dorsal attention (DA) and limbic networks, and narrower in frontoparietal (FPN) and DMN (pFDR<0.01; Fig. 1C). Likewise, SFC peaks were higher in DA and limbic networks, and lower in salient, FPN and DMN (pFDR<0.01; Fig. 1C). While cortical thickness gradually increased along the functional hierarchy in both groups, TLE presented with marked atrophy in unimodal sensory areas (pFDR<0.05). Conclusions: TLE is associated with disrupted organization of the cortical hierarchy characterized by functional segregation of morphologically abnormal unimodal and heteromodal cortices. These alterations are likely at the basis of multi-system cognitive dysfunction. Reference1.Caciagli L, et al. J. Front. Neurosci (2014)2.Mesulam M-M. J. Neurol (1998)3.Sepulcre J, et al. J. Neurosci (2012)4.Cammoun L, et al. J. Neurosci Methods (2012)5.Carmona S, et al. Hum. Brain Mapp (2015)6.Yeo BT, et al. J. Neurophysiol (2011) Funding: CIHR (MOP-57840, MOP-123520)CIHR (MOP-57840, MOP-123520)