Epileptogenic Network Nodes Generating Higher Frequency (350-600Hz) Fast Ripples Predicts Surgical Response in Patients with Drug Resistant Focal Epilepsy
Abstract number :
3.032
Submission category :
1. Basic Mechanisms / 1C. Electrophysiology/High frequency oscillations
Year :
2021
Submission ID :
1825781
Source :
www.aesnet.org
Presentation date :
12/6/2021 12:00:00 PM
Published date :
Nov 22, 2021, 06:50 AM
Authors :
Iren Orosz, MD - UCLA; Tomas Pastore, MS - University of Buenos Aires; Zachary Waldman, MS - Thomas Jefferson University; Richard Gorniak, MD - Thomas Jefferson University; Ashwini Sharan, MD - Thomas Jefferson University; Chengyuan Wu, MD, MS - Thomas Jefferson University; Itzhak Fried, MD PhD - UCLA; Diego Slezak, PhD - University of Buenos Aires; Jerome Engel, MD PhD - UCLA; Michael Sperling, MD - Thomas Jefferson University; Richard Staba, PhD - UCLA; Shennan Weiss, MD, PhD - SUNY Downstate, Kings County Hospital
Rationale: Seizure genesis involves coordinated activity from a network of inter-connected regions. The clinically defined seizure onset zone (SOZ) can be regarded as nodes in this network. High-frequency oscillations (HFOs; ripples 80-200 Hz; fast-ripples 200-600 Hz) occur during inter-ictal epochs and are generated at relatively higher rates in the seizure onset zone (SOZ). Unresected fast ripples have also been shown to predict a poor post-operative seizure outcome. In an effort to determine if HFOs can be used to properly define seizure generating networks, we sought to determine the HFO subtypes and HFO properties that best identify the SOZ, and predict response to surgery.
Methods: In 35 patients we applied an automated detector of sharp spikes, ripples and fast ripples to stereo-EEG recordings during non-REM sleep. Electrode locations were determined using atlas-based segmentation. We generated receiver operating curves (ROC) for classifying SOZ contacts using HFO rates, and used non-parametric statistics to compare HFO properties in the SOZ and non-SOZ.
Results: In surgical responders to resection and/or responsive nerve stimulation (RNS) implantation (n=21), the accuracy of most HFO type rates for classifying the SOZ was paradoxically worse, compared to patients who did not respond to surgery (n=7). However, fast ripples with a higher spectral content were found more often in the SOZ of patients who responded to surgery (p < 0.0001), but not in non-responders. Fast ripple rates classified the SOZ better in the responders as compared to non-responders after including only events >350 Hz. Not all brain regions generated fast ripples > 350 Hz.
Conclusions: We show that when the SOZ and brain networks generating fast ripples > 350Hz overlap resective surgery or RNS will reduce seizure burden, but when these networks do not overlap surgical response will be poor.
Funding: Please list any funding that was received in support of this abstract.: 1K23NS094633-01A1.
Basic Mechanisms