Mapping of Spike Propagation Reveals Effective Connectivity and Predicts Surgical Outcome in Epilepsy
Abstract number :
1.329
Submission category :
9. Surgery / 9B. Pediatrics
Year :
2022
Submission ID :
2204011
Source :
www.aesnet.org
Presentation date :
12/3/2022 12:00:00 PM
Published date :
Nov 22, 2022, 05:22 AM
Authors :
Margherita A.G. Matarrese, MSc – Università Campus Bio-Medico di Roma; Alessandro Loppini, PhD – Engineering – Università Campus Bio-Medico di Roma; Eleonora Tamilia, PhD – Boston Children's Hospital; M Scott Perry, MD – Cook Children's Health Care System; Joseph R. Madsen, MD – Boston Children's Hospital; Phillip Pearl, MD – Boston Children's Hospital; Simonetta Filippi, M.Sc. – Università Campus Bio-Medico di Roma; Christos Papadelis, PhD – Cook Children's Health Care System
This abstract is a recipient of the Grass Young Investigator Award
This abstract has been invited to present during the Pediatric Epilepsy Highlights platform session
Rationale: The best available treatment option for children with drug resistant epilepsy (DRE) of focal onset is resective neurosurgery, which requires a biomarker that identifies the epileptogenic zone with high precision. Interictal spikes are considered the main interictal biomarkers of epilepsy, but they suffer from low specificity mostly due to their propagation across large brain areas forming networks. Here, we aim to reveal the relationship between the spatiotemporal propagation of spikes and effective connectivity among onset and areas of spread and assess their surgical prognostic value.
Methods: We retrospectively analyzed interictal intracranial electroencephalography (iEEG) data from 43 children (25 with Engel 1, 1 year after surgery) with DRE who underwent resective neurosurgery. Interictal spikes and their propagation were detected at the iEEG contact’s level (Figure 1A). Using electric source imaging on iEEG data (Figure 1B), we mapped the spatiotemporal propagation of spikes in the source domain with dynamic statistical parametric mapping (Figure 1C), and identified three source regions of interest: onset, early spread, and late spread zone (Figure 1D). For each region, we: (1) extracted a time series of the reconstructed electrical activity; (2) estimated the direction of information flow using Granger Causality (Figure 1E); and (3) computed the overlap percentage with the resection volume (ORES), the flow direction strength (FDS), and their average (ORES +FDSout/FDSin). Finally, we assessed whether resection of the source regions predicts outcome via Fisher exact test and whether information flow improves the predictive performance.
Results: We found a higher number of events where spikes were propagating for patients with good vs. poor outcome (p=0.02; 122 [IQR: 68.75-220.25] vs. 44.5 [22-135]). ORES of onset, early spread, and late spread were higher in good vs. poor outcome (onset: p< 0.001, 96 vs 14%; early: p=0.002, 86 vs 14%; late: p=0.002, 59 vs 7%). In patients with good outcomes, ORES of onset was higher compared to areas of spread (early: p=0.01; late: p=0.001) (Figure 2A-C). We also observed a predominant direction of information flow from onset to late spread for 63% of patients (Figure 2B), and a higher outward than inward FDS from the onset for patients with good outcome (p=0. 02, 0.72 [0.60-0.84] vs. 0.59 [0.35-0.72]) (Figure 2D). Resection of spike onset and early spread predicted outcome (onset: p< 0.001, PPV=93%, NPV=61%; early: p=0.01, PPV=91%, NPV=53%), while resection of late spread did not (Figure 2F). Combining the ORES with the FDS ratio increased the predictive performances of these parameters taken individually [PPV of 94% and NPV of 65% (p < 0.001)] (Figure 2E).
Surgery