Neural Fragility as an iEEG Biomarker of the Seizure Onset Zone
Abstract number :
3.096
Submission category :
2. Translational Research / 2C. Biomarkers
Year :
2021
Submission ID :
1825773
Source :
www.aesnet.org
Presentation date :
12/6/2021 12:00:00 PM
Published date :
Nov 22, 2021, 06:50 AM
Authors :
Patrick Myers, BS, MSE - Johns Hopkins University; Adam Li, BS, MS - Johns Hopkins University; Chester Huynh, BS - Johns Hopkins University; Zachary Fitzgerald, BS - Cleveland Clinic; Iahn Cajigas, MD, PhD - University of Miami Miller School of Medicine; Damian Brusko, MD - University of Miami Miller School of Medicine; Jonathan Jagid, MD - University of Miami Miller School of Medicine; Angel Claudio, MD - University of Miami Miller School of Medicine; Andres Kanner, MD - University of Miami Miller School of Medicine; Jennifer Hopp, MD - University of Maryland Medical Center; Stephanie Chen, MD - University of Maryland Medical Center; Jennifer Haagensen, DO, MS - University of Maryland Medical Center; Emily Johnson, MD - Johns Hopkins Hospital; William Anderson, MA, MD, PhD - Johns Hopkins Hospital; Nathan Crone, MD - Johns Hopkins Hospital; Sara Inati, MD - NINDS, NIH; Kareem Zaghloul, MD, PhD - NINDS, NIH; Juan Bulacio, MD - Cleveland Clinic; Jorge Gonzalez-Martinez, MD, PhD - University of Pittsburgh Medical Center; Sridevi Sarma, PhD - Johns Hopkins University
Rationale: Over 15 million epilepsy patients worldwide have drug-resistant epilepsy (DRE). For these DRE patients who have focal epilepsy, successful surgical treatment requires complete removal, or disconnection of the seizure onset zone (SOZ), brain region(s) where seizures originate. Unfortunately, surgical success rates vary between 30%-70% because no clinically validated biological marker of the SOZ exists. We develop and retrospectively validate a new intracranial EEG (iEEG) marker - neural fragility - in a retrospective analysis of 91 patients by using neural fragility of the annotated SOZ as a metric to predict surgical outcomes. We hypothesize that fragility will be high in the clinically annotated SOZ for surgical success cases and low in surgical failure cases.
Methods: iEEG recordings from 91 DRE patients who later underwent resective surgery (44 successes and 47 failures) were collected retrospectively from: Johns Hopkins Hospital, the National Institute of Health, Cleveland Clinic, University of Maryland Medical Center, and Jackson Memorial Hospital, University of Miami. We calculated the neural fragility, a metric of imbalance of a node in a dynamic network, for each patient. We compared distributions of fragility values inside the clinically annotated SOZ against outside the SOZ via a one-sided Mann-Whitney U test. We then predicted the outcome of a surgical resection based on the agreement of a hypothesized SOZ generated from fragility values and the clinically annotated SOZ. To validate fragility as a predictor of the SOZ, we benchmarked the performance of fragility against 20 other features, 6 frequency based and 14 graph features.
Results: Fragility was calculated around seizure events and used to generate spatio-temporal heatmaps. The more red a row of the heatmap is, the more fragile the corresponding brain region is. This presentation allows for observation of a large amount of complex data in a compact and interpretable manner. Figure 1 demonstrates typical fragility maps for surgical success and failure cases. In success cases, there is a high degree of agreement between highly fragile regions and annotated SOZ contacts as shown in patient 1. Failure cases often had one of two presentations: either there were highly fragile regions outside of the SOZ (patient 26) or all regions appear highly fragile (patient 40). When used in a classifier to predict surgical outcome, fragility outperforms all other features tested, as shown in Figure 2.
Conclusions: Fragility predicts 43 of the 47 surgical failures, with an overall prediction accuracy of 76%. This performance is a major improvement over to the 48% accuracy of clinicians (successful outcomes). In most surgical failure cases, we can identify an untreated area with high fragility. Neural fragility’s predictive power as an SOZ marker outperformed 20 of the most common proposed metrics. These results indicate neural fragility may serve as a biomarker of the SOZ in iEEG recordings, potentially aiding clinicians in neurosurgical planning.
Funding: Please list any funding that was received in support of this abstract.: NIH T32 EB003383, NSF GRFP (DGE-1746891), Arcs Chapter Scholarship, Whitaker Fellowship and the Chateaubriand Fellowship.
Translational Research