Seizure Onset and Propagative Zones Exhibit Distinct Local Structure-function Coupling: A Combined Diffusion MRI and SEEG Study
Abstract number :
1.122
Submission category :
2. Translational Research / 2C. Biomarkers
Year :
2022
Submission ID :
2204354
Source :
www.aesnet.org
Presentation date :
12/3/2022 12:00:00 PM
Published date :
Nov 22, 2022, 05:24 AM
Authors :
Graham Johnson, MD/PhD Student – Vanderbilt University; Derek Doss, BS – Vanderbilt University; Jasmine Jiang, BS – Vanderbilt University; Saramati Narasimhan, PhD – Vanderbilt University; Danika Paulo, MD – Vanderbilt University Medical Center; Victoria Morgan, PhD – Vanderbilt University; Sarah Bick, MD – Vanderbilt University Medical Center; Dario Englot, MD/PhD – Vanderbilt University Medical Center
Rationale: Increasing evidence has suggested that identification of regions involved in early seizure propagation (“Propagation Zones,” PZ) is important to predict seizure freedom after epilepsy surgery (Andrews et al. JAMA Neurology. 2019;76(4):462). Resting-state connectivity analyses using stereo-electroencephalography (SEEG) have shown promise in efficiently characterizing seizure onset zones (SOZ), but have found difficulty in distinguishing both SOZs and PZs from non-involved brain regions. Recently, evidence has suggested that ictal structure-function coupling can be used to delineate brain regions important in seizure dynamics (Shah et al. Brain. 2019;142(7):1955-1972). Thus, we investigated if resting-state structure-function coupling differed between SOZs, PZs, and non-involved regions and could be used to efficaciously characterize the ictal onset and early propagative epileptic network without ictal recordings.
Methods: We calculated the SEEG partial directed coherence of 26 consented patients with focal epilepsy undergoing presurgical evaluation at Vanderbilt University Medical Center. Using preoperative diffusion MRI, we then implemented a custom technique to obtain structural connectivity metrics between SEEG contacts. We calculated the structural connectivity of SOZs, PZ (spread within 10 seconds), and non-involved regions over a range of Euclidean distances from 5 to 80 mm. Next, we calculated the structure-function coupling over the same distances. Finally, we generated models using a support vector machine to classify SOZs, PZs, and non-involved regions using (1) functional connectivity only, and (2) functional and structural connectivity.
Results: SOZs and PZs exhibit comparably high local structural connectivity compared to non-involved regions despite SOZs demonstrating significantly greater functional connectivity to both PZs and non-involved regions (Figure 1). PZs and non-involved regions do not demonstrate a significant difference in functional connectivity despite a significant difference in structural connectivity. However, PZs exhibit significantly higher local structure-function coupling to that of non-involved regions, with SOZs exhibiting the highest local structure-function coupling. A support vector machine to classify SOZs, PZs, and non-involved regions at the SEEG contact level was able to significantly increase model accuracy by incorporating local structure-function coupling (Figure 2).
Conclusions: The comparably increased structural connectivity of SOZs and PZs is possibly due to the SEEG implantation scheme being biased towards clinically predicted seizure onset and propagation networks. Regardless, SOZs and PZs demonstrate a distinct local structure-function coupling to that of non-involved regions and each other. This distinct coupling profile can be used to accurately classify SOZs, PZs and non-involved regions. These findings could be used clinically to efficiently characterize the epileptic network using only brief resting-state recordings.
Funding: NINDS R01NS112252, NINDS R01NS075270, NINDS R01NS110130, NINDS R01NS108445, F31NS120401, NIGMS T32GM007347
Translational Research