In vivo mapping of cortical excitation-inhibition imbalance in temporal lobe epilepsy
Presented During:
Thursday, June 27, 2024: 11:30 AM - 12:45 PM
COEX
Room:
Grand Ballroom 101-102
Poster No:
365
Submission Type:
Abstract Submission
Authors:
Ke Xie1, Jessica Royer1, Raúl Rodriguez-Cruces1, Linda Horwood1, Alexander Ngo1, Hans Auer1, Ella Sahlas1, Judy Chen1, Yigu Zhou1, Sofie Valk2, Birgit Frauscher3, Raluca Pana1, Andrea Bernasconi1, Neda Bernasconi1, Boris Bernhardt1
Institutions:
1Montreal Neurological Institute and Hospital, Montreal, Canada, 2Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany, 3Duke University, Durham, USA
First Author:
Co-Author(s):
Introduction:
Excitation-inhibition (E:I) imbalance is posited as a fundamental pathophysiological mechanism in temporal lobe epilepsy (TLE).[1] However, previous evidence supporting this hypothesis has been primarily derived from experimental studies in non-human animals. This study aims to non-invasively elucidate the cortical pattern of E:I imbalance in TLE patients and explore its associations with disease severity and cognitive impairment.
Methods:
We studied 40 pharmaco-resistant TLE patients (17 males; age = 35.80±11.04 years; 27/13 left/right focus) and 40 age- and sex-matched healthy controls (19 males; 34.25±3.98 years). All participants underwent multimodal MRI at 3T, as well as global cognitive testing including the Montreal Cognitive Assessment (MoCA) and EpiTrack. A subset of participants underwent follow-up MRI scans and cognitive assessment. Node-wise Hurst exponent score, reflecting scale-free properties (i.e., 1/f slop) of fMRI signal and serving as a proxy for the overall E:I ratio within a given region,[2] was estimated via the univariate maximum likelihood method and discrete wavelet transform, modeling the resting-state fMRI timeseries as multivariate fractionally integrated processes. Quantitative and surface-wide between-group differences in Hurst exponent were assessed, with P-values adjusted for false discovery rate (FDR). Subsequently, we explored the relationship between TLE-related regional changes in Hurst exponent and microcircuit parameters estimated by connectome-informed biophysically computational simulations via a parametric mean-field model.[3] Finally, we examined associations with clinical and cognitive measures at baseline, as well as prospective cognitive decline after a 2-year follow-up.
Results:
In both cohorts, Hurst exponent scores exhibited a sensory-fugal distribution, being highest in the visual cortex, intermediate in the frontoparietal and default mode networks, and lowest in the paralimbic network (Fig. 1a, 1c), aligning with the sensory–fugal gradient of cytoarchitectural differentiation (healthy controls/TLE: rho = -0.41/-0.46, Pspin = 0.044/0.026). TLE patients had a significantly lower Hurst exponent score across the whole brain than healthy controls (Cohen's d = -0.75, P < 0.001), indicating an overall elevated E:I ratio. Surface-based analysis further revealed marked reductions in local Hurst exponent scores in bilateral temporal lobes, dorsolateral and dorsomedial prefrontal cortices, precuneus, fusiform, and occipital cortex in TLE compared to healthy controls (PFDR < 0.05, Fig 1b). When stratifying the topography into functional communities, pronounced effects were observed in the transmodal association system, such as the default mode, frontoparietal, and attention networks, as well as the visual system (Fig 1c). Computational models indicated that the degree of Hurst exponent changes was closely related to atypical increases in recurrent connection strength in TLE (rho = -0.22, Pspin = 0.015; Fig 1d). Finally, lower Hurst exponent scores in TLE patients were associated with longer disease duration (whole-brain, t = -1.91, P = 0.016; significant clusters, t = -2.02, P = 0.013) and poorer performance on both the MoCA (t = 2.33, P = 0.006; t = 2.93, P = 0.001) and EpiTrack tests (t = 2.72, P = 0.002; t = 3.06, P < 0.001). Moreover, in TLE patients, Hurst exponent scores declined significantly at the 2-year follow-up time point (Cohen's d = -0.84, P = 0.002; Cohen's d = -0.68, P = 0.006), mirroring the prospective decline in MoCA scores (t = 1.58, P = 0.035; t = 1.67, P = 0.031; Fig. 2).
·Fig. 1 | Hurst exponent reductions in TLE.
·Fig. 2 | Associations of Hurst exponent scores with clinical characteristics and behavioral assessments.
Conclusions:
In TLE, our finding of reduced Hurst exponent scores likely indicates widespread cortical excitation-inhibition changes, tilting the balance towards increased cortical excitability. These changes were found to increase with ongoing disease progression and more marked cognitive impairment, highlighting the potential of the Hurst exponent as a neuroimaging biomarker for TLE-related dysfunction.
Disorders of the Nervous System:
Neurodevelopmental/ Early Life (eg. ADHD, autism) 1
Modeling and Analysis Methods:
Diffusion MRI Modeling and Analysis
Neuroanatomy, Physiology, Metabolism and Neurotransmission:
Anatomy and Functional Systems
Physiology, Metabolism and Neurotransmission :
Cerebral Metabolism and Hemodynamics 2
Keywords:
Epilepsy
Modeling
MRI
Other - excitation-inhibition imbalance
1|2Indicates the priority used for review
Provide references using author date format
2. Trakoshis, S. et al. (2020), 'Intrinsic excitation-inhibition imbalance affects medial prefrontal cortex differently in autistic men versus women', eLife, vol. 9, p. e55684
3. Kong, X. et al. (2021), 'Sensory-motor cortices shape functional connectivity dynamics in the human brain', Nat Commun, vol. 12, no. 1, p. 6373