Measuring The Temporal Lobe Disconnection Needed for Seizure Control with Non-Invasive Network Modeling

Introduction: Despite concordant preoperative evaluation in patients with medically refractory temporal lobe epilepsy (TLE), seizures persist in one-third of patients after surgery. In prior studies modeling the epileptogenic network using resting state functional MRI (rsfMRI) combined with scalp electroencephalography (EEG), disconnection of a broadly distributed epileptogenic network after surgery correlated with higher rates of seizure freedom. However, it is not known what regions of interest (ROIs) are most implicated in that disconnection. We hypothesized that functional disconnection of the ROIs directly connected to the mesial temporal lobe would correlate with seizure freedom.

Methods: The epileptogenic network was modeled preoperatively non-invasively in 77 adult patients with TLE and co-registered with the Talairach atlas to subdivide the epileptogenic network into ROIs. Disconnection of the epileptogenic network in each ROI was measured by the percent change in the interconnectivity of the epileptogenic network that was co-localized in that ROI. The false-discovery rate method was used to correct for multiple comparisons.

Results: Demographic analysis did not reveal significant differences in preoperative characteristics between the seizure-free and not seizure-free groups. Next, disconnection of the epileptogenic network was compared between seizure-free and not seizure-free groups. Ipsilateral to the surgery side, the precuneus and cingulate cortex were disconnected to a higher degree in seizure-free patients. Contralateral to surgery, the amygdala, hippocampus, and posterior cingulate cortex were more disconnected in seizure-free patients.

Conclusion : In this study, functional disconnection of the mesial temporal lobe from the greater temporal lobe network was correlated to seizure control, which is conceptually self-evident but has not yet been shown using imaging techniques. Perhaps there is a redundant pattern of connectivity or more distributed hubness to explain why some networks are more resilient and therefore more difficult to disconnect. In future studies, we will attempt to model the post operative network disconnection by simulating ROI removal. Modeling a surgical resection and its likely impact on the temporal lobe network will be useful to surgeons to plan an approach that maximizes network disconnection and seizure control.