Distributed functional cross-network connectivity of the epileptogenic zone predicts late seizure relapse in temporal lobe epilepsy.
TL;DR
Long-term outcomes in temporal lobe epilepsy depend not only on accurate localization and removal of epileptogenic tissue but also on how its connectivity is distributed across brain networks before surgery, which may represent a biomarker of late relapse.
Abstract
Objective
Although many patients with temporal lobe epilepsy achieve seizure freedom in the first post-operative year, the likelihood of long-term seizure freedom declines substantially over time. Late seizure relapse may reflect ongoing epileptogenicity, potentially driven by persistent epileptic activity arising from the epileptogenic zone. We hypothesized that its more distributed connectivity profile across brain networks would distinguish patients experiencing late relapse.
Methods
We prospectively enrolled 209 patients with drug-resistant temporal lobe epilepsy across six centers who underwent destructive surgery, pre-operative diffusion-weighted magnetic resonance imaging (MRI), and/or resting-state functional MRI. To isolate network mechanisms related to late relapse rather than early surgical failure, we focused on patients who achieved seizure freedom after the first post-operative year. After two post-operative years, these patients were stratified into long-term seizure freedom (SF-SF) and late relapse (SF-NSF). We delineated post-operative resections to localize epileptogenic zones and quantified their network properties using graph-theory metrics-betweenness centrality and participation coefficient. Patient graph-theory metrics were standardized relative to normative connectomes derived from 362 healthy controls. We evaluated associations between patient graph-theory metrics and long-term outcomes using Firth's bias-reduced regressions and permutation testing.
Results
Fifty-four patients with diffusion-weighted MRI (SF-SF = 37, SF-NSF = 17) and 30 patients with functional MRI (SF-SF = 19, SF-NSF = 11) met the inclusion criteria. Greater functional MRI-derived participation coefficient of resections was associated with late relapse, reflecting a more distributed connectivity profile across networks. In a post hoc analysis, lower diffusion-weighted MRI-derived betweenness centrality of the epileptogenic hippocampus was associated with late relapse. Effects were confirmed by patient-level and anatomy-preserving permutations.
Significance
Long-term outcomes in temporal lobe epilepsy depend not only on accurate localization and removal of epileptogenic tissue but also on how its connectivity is distributed across brain networks before surgery. Pre-operative cross-network connectivity of the epileptogenic zone may represent a biomarker of late relapse.