Lesion-related epilepsy maps to a common brain network
Frederic Schaper1, Janne Nordberg2, Alexander Cohen3, Christopher Lin4, Joey Hsu5, Andreas Horn4, Michael Ferguson4, Shan Siddiqi4, William Drew4, Louis Soussand6, Anderson Winkler7, Marta Simo8, Jordi Bruna9, Sylvain Rheims10, Marc Guenot10, Marco Bucci11, Lauri Nummenmaa11, Julie Staals12, Albert Colon13, Linda Ackermans14, Ellen Bubrick4, Jurriaan Peters15, Ona Wu16, Natalia Rost17, Jordan Grafman18, Hal Blumenfeld19, Yasin Temel14, Rob Rouhl14, Juho Joutsa2, Michael Fox20
1Brigham and Women's Hospital, Harvard Medical School, 2Turku Hospital, 3Boston Children's Hospital, 4Brigham and Women's Hospital, 5Pitt University, 6AP-HP, 7NIH, 8L’Hospitalet del Llobregat, 9Hospital Universitari De Bellvitge - ICO Duran I Reynals, 10Hospices Civils de Lyon, 11University of Turku, 12AzM/ Academic Hospital Maastricht, 13Kempenhaeghe, 14Maastricht University, 15Boston Childrens Hospital, 16MGH Stroke Research Center, 17Massachusetts General Hospital, 18Northwestern University, 19Yale University Sch of Med, 20Brigham and Women's Hospital / Harvard Medical School
Objective:
To test whether lesions related to epilepsy map to a common brain network.
Background:

Focal epilepsy is commonly caused by brain lesions, but it remains unclear why some lesion locations result in epilepsy while others do not. One possibility is that some brain regions or networks are more vulnerable than others. Identifying the brain regions or networks related to epilepsy could inform prognosis and guide interventions.

Design/Methods:

Lesion locations from patients with ischemic stroke-related epilepsy (n = 76) and control lesions (n = 625) were mapped to a common brain atlas. Traditional lesion mapping methods were used to test for associations between epilepsy and lesion location. Next, we computed the brain network functionally connected to each lesion location using human brain connectome data (n = 1000). Network connections associated with epilepsy were identified. Generalizability to other lesion types was assessed using independent datasets of four different lesion etiologies (n = 772) and a leave-one-dataset-out analysis. These connections were then used to generate a brain network map that best encompasses lesion locations related to epilepsy. Finally, we tested whether thalamic deep brain stimulation sites that improve seizure control were connected to this same network (n = 30).

Results:

We found that lesions associated with epilepsy occurred in multiple heterogenous locations spanning different lobes and vascular territories. However, these heterogenous lesion locations were part of a common brain network defined by functional connectivity to the basal ganglia and cerebellum (Vmax = 6.8, P < 0.001). Functional connectivity to these regions was associated with the risk of epilepsy across different lesion types (χ2 = 205.3, P < 0.001) and with therapeutic response to thalamic deep brain stimulation (r = 0.63, P < 0.005).

Conclusions:

Lesion locations related to epilepsy map to a common brain network defined by functional connectivity to the basal ganglia and cerebellum.

10.1212/WNL.0000000000203957