Authors :
Presenting Author: Ke Xie, MSc – McGill University
Jessica Royer, PsyD – McGill University
Ella Sahlas, BS – McGill University
Alexander Ngo, BS – McGill University
Thaera Arafat, MD – Montreal Neurological Institute and Hospital, McGill University
Judy Chen, BS – McGill University
Yigu Zhou, BS – McGill University
Raul Rodriguez-Cruces, PhD – McGill University
Raluca Pana, MD – McGill University
Andrea Bernasconi, MD – McGill University
Neda Bernasconi, MD, PhD – McGill University
Zhiqiang Zhang, MD – Department of Diagnostic Radiology, Jinling Hospital, Medical School of Nanjing University
Luis Concha, MD, PhD – Universidad Nacional Autónoma de Mexico
Boris Bernhardt, PhD – McGill University
Rationale:
Temporal lobe epilepsy (TLE) is the most common pharmaco-resistant focal epilepsy in adults, affecting over 50 million people worldwide. Accumulated evidence suggests that cortical thinning in certain neocortical regions is a crucial structural MRI characteristic of TLE. Additionally, aberrant functional activity in TLE at the local and whole-brain levels is increasingly revealed via resting-state fMRI. In this study, we employed multi-site resting-state fMRI data and several analysis strategies to investigate functional imbalances in patients with TLE at the individual level and across different scales.
Methods: The study analyzed 223 adults with unilateral TLE and 258 age- and sex-matched healthy controls collected from three sites (
Fig 1a). Three functional properties across different spatial scales were calculated from resting-state fMRI time series at the node level: temporal variability, regional homogeneity, and degree centrality (
Fig 1b). A W-score (analogous to a z-score adjusted for age and sex) was calculated for each patient and brain region as follows: W-score = (raw - predicted)/RSD, indicating the extent to which each brain region for that individual is atypical relative to the norm. We used a cut-off score of 2 standard deviations (|W|
≥ 1.96) to isolate regions with ‘significant’ atypicality. We then established a prevalence map, showing the percentage of TLE patients with markedly atypical W-scores in each brain region (
Fig 1c).
Results:
Across all brain regions, the median prevalence of atypical temporal variability, regional homogeneity, and degree centrality in TLE was 10.3%, 7.6%, and 6.3%, respectively (Fig 2a, 2b). As such, in each brain region, ~10.3%, 7.6%, and 6.3% of patients were considered outliers on an individual scale, respectively. A systemic assessment of the three measures using Mahalanobis distance revealed diffuse atypicality of W-score in TLE (mean prevalence ± SD = 25% ± 8% [9% - 53%]), with a higher prevalence in the bilateral temporal lobes (Fig 2c). This finding is replicated and highly consistent across three sites (rho > 0.39, Pspin < 0.001). Additionally, a higher prevalence of atypical W-scores was observed in TLE patients with hippocampal atrophy in the ipsilateral temporal cortex compared to those with normal hippocampal volume. Similarly, TLE patients with a longer illness duration exhibited a higher prevalence of atypical W-scores in the ipsilateral temporal, parietal, and insula cortices than those with a shorter disease duration.