Presurgical magnetoencephalography-derived network control metrics distinguish successful hippocampal resection vs sparing in temporal lobe epilepsy.

OBJECTIVE

Surgical decision-making in temporal lobe epilepsy (TLE) faces a critical challenge in determining whether the hippocampus can be safely spared during anterior temporal resection, particularly when surgery involves the language-dominant hemisphere. We investigated whether presurgical network control metrics derived from magnetoencephalography (MEG) can differentiate patients who achieved seizure freedom with hippocampal resection (HR) from those who achieved seizure freedom with hippocampal sparing (HS).

METHODS

We analyzed presurgical spike-free interictal MEG in 25 TLE patients with seizure freedom after anterior temporal resection (19 with hippocampal resection, 6 with sparing). Functional networks constructed from MEG data were partitioned into communities using a Louvain-based consensus clustering algorithm. Within identified hippocampal-related communities from each patient, we applied control centrality analysis to quantify the influence of each node on community synchronizability. Group differences were assessed using Fisher's exact tests with false discovery rate (FDR) correction applied within each frequency band and hemisphere. Leave-one-subject-out (LOSO) sensitivity analysis assessed the robustness of findings to individual subjects.

RESULTS

After FDR correction, synchronization of the contralateral anterior superior temporal sulcus (STS) in the HS group at beta band emerged as the primary finding (p = .001, q = .02). This finding showed 100% stability in LOSO and persisted in a subgroup analysis excluding patients with hippocampal seizure onset (p = .004). At p < .05, additional region-frequency differences formed a coherent anatomic pattern: the HR group showed higher synchronization fractions within ipsilateral limbic structures, whereas the HS group showed higher synchronization in the ipsilateral posterior hippocampus and contralateral temporal regions with 100% LOSO stability.

SIGNIFICANCE: Presurgical MEG-based network control analysis identified contralateral anterior STS synchronization as a candidate biomarker for hippocampal sparing eligibility in TLE. This proof-of-concept finding warrants prospective validation as potential tools to support individualized surgical planning beyond localization-based approaches.