Authors :
Presenting Author: Donna Gift Cabalo, RN, MSc – McGill University
Jordan DeKraker, PhD, PostDoc – McGill University; Jessica Royer, PsyD, Ph.D candidate – McGill University; Shahin Tavakol, Ph.D candidate – McGill University; Ke Xie, Ph.D candidate – McGill University; Raul Rodriguez-Cruces, MD,Ph.D, PostDoc – McGill University; Andrea Bernasconi, MD,PhD – Montreal Neurological Institute and Hospital; neda Bernasconi, MD – Montreal Neurological Institute and Hospital; birgit Frauscher, MD – Montreal Neurological Institute and Hospital; Elizabeth Jefferies, PhD – York University; Jonathan Smallwood, PhD – Queens University; Lorenzo Caciagli, MD,PhD – University of Pennsylvania; Boris Bernhardt, PhD – McGill University
Rationale:
Temporal lobe epilepsy (TLE) is a common pharmaco-resistant epilepsy and is typified by hippocampal pathology. Research suggests that large-scale brain networks
1 may contribute to declarative memory deficits experienced by many patients. These networks can undergo reorganization and may underpin impairments in some patients but can also facilitate compensatory adaptation.
2 Here, we investigated the reorganization of declarative memory networks in TLE by analyzing signal flow between different brain regions.
Methods:
We studied 20 TLE patients (left/right=15/5, age=36 土 12 years; F/M: 10/10) and 60 healthy controls (age=34 土 8 years, F/M: 30/30). Participants underwent fMRI while performing episodic and semantic retrieval tasks (Figure 1A). To infer signal flow, we concatenated pre-processed neocortical
3 and hippocampal
4 data (200 parcels + 419 vertices/hemisphere) and computed subject and state-specific intrahemispheric effective connectivity (EC) using the
rDCM toolbox.
5 The model specified the design matrix as the experimental manipulations influencing state-specific neuronal activity, which resulted in 619x1 driving inputs (DI) and a 619x619 EC matrix/hemisphere (Figure 1B,C). We analyzed the resulting EC matrices, distinguishing efferent (outward) and afferent (inward) signal flow. Out/in-degree scores were computed by summing connectivity weights. Group differences were evaluated using surface-based univariate analysis, controlling for age and sex, and correcting for multiple comparisons. We also examined the association between memory impairments and atypical signal flow in both memory states.
Results:
In TLE, while no significant reorganization was observed in TLE for semantic memory states, there were significant alterations in signal flow during the episodic memory state as compared to neurotypicals. Notably, we observed contralateral changes in afferent signal flow and bilateral changes in efferent flow, in both hippocampal and neocortical regions (Figure 2A,B). Mirroring the brain alterations, we did not observe associations to behavioral measures of semantic cognition, but to scores of episodic memory function. Notably, while efferent signal flow was not associated with episodic memory performance, bilateral afferent signal flow from the neocortex to the hippocampus and ipsilateral afferent signal flow from the hippocampus to neocortex was negatively correlated with episodic memory scores (p
FDR< 0.1). (Figure 2C).