Abstracts

Rationale: Approximately 30% of individuals with temporal lobe epilepsy (TLE) develop chronic pharmacoresistant epilepsy. Although resection of the affected temporal lobe is the most effective treatment to arrest refractory seizures, postsurgical seizures can persist in up to 40% of surgical candidates. Thus, predicting downstream consequences of resective surgery on the brain’s connectome, and how this reorganization relates to postoperative seizure outcome, remains a clinical challenge. Here, we assessed connectome reconfigurations following surgery in TLE patients using manifold representations of structural brain networks and examined associations with hippocampal atrophy and clinical variables.

Methods: Participants. We studied 25 adult patients with drug-resistant TLE (13 males, mean age±SD=25.2±7.5 years). Patients underwent longitudinal T1w and diffusion MRI and underwent anterior temporal lobectomy between both time points. Postoperative histopathology confirmed hippocampal sclerosis in every patient. Postsurgical seizure outcome was determined according to Engel’s modified classification.

Connectome organization. Subject-specific structural connectome gradients were generated, aligned to a normative template (Fig. 1a­–b), normalized, and sorted into ipsilateral/contralateral to the focus. Subcortical gradient values were determined based on their connectivity allegiance to a gradient cluster (Fig. 1c–d). Surgical cavities were automatically segmented from pre- and postoperative T1w MRIs. Cavity masks were excluded from all analyses (Fig. 1e).

Results: Multivariate mixed effects models, performed on the aggregate of the first three structural gradients, assessed pre- to postoperative connectome changes. Following surgery, individuals with TLE showed profound and widespread gradient alterations affecting regions surrounding the resection site (p_FDR< 2✕10^–6) and contralateral fronto-occipital cortices (p_FDR< 4✕10^–4; Fig. 2a).

To simplify multidimensional gradient changes, we computed the Euclidean distance between the template center and each data point in the 3D gradient manifold and assessed longitudinal pre- to postoperative changes. Postoperative contraction was observed in contralateral temporo-limbic cortex, hippocampus, and amygdala (p_FDR< 6✕10^–6) while expansion was observed in ipsilateral fronto-occipital cortices (p_FDR< 0.005; Fig. 2b).

To conceptualize patterns of manifold contraction/expansion with respect to markers of hippocampal disconnectivity and clinical factors, we correlated postoperative deformation values with (i) hippocampal subfield atrophy, (ii) postsurgical seizure outcome, (iii) epilepsy duration, and (iv) seizure frequency. The severity of preoperative ipsilateral hippocampal atrophy across all subfields was related to greater postoperative contraction in areas of surgery-induced deformations (Fig. 2c). Favourable surgical outcome was associated with greater postoperative temporo-limbic contraction and fronto-occipital expansion (Fig. 2c).

Conclusions: Our findings highlight the utility of network neuroscience approaches to contextualize downstream connectome disruptions with clinical and standard neuroimaging parameters.

Funding: Please list any funding that was received in support of this abstract.: FRQS, CIHR.