Abstracts

Predicting Surgical Outcome in Temporal Lobe Epilepsy Utilizing Non-Invasive Markers with Functional Imaging and Scalp EEG

Abstract number : 1.311
Submission category : 9. Surgery / 9A. Adult
Year : 2021
Submission ID : 1825544
Source : www.aesnet.org
Presentation date : 12/4/2021 12:00:00 PM
Published date : Nov 22, 2021, 06:43 AM

Authors :
Layton Lamsam, MD - Yale University School of Medicine - Department of Neurosurgery; Jennifer Percy, MD - Department of Neurology, Yale School of Medicine; John Chiari, BS - Department of Neurosurgery, Yale School of Medicine; Jagriti Arora, MS - Department of Radiology and Biomedical Imaging, Yale School of Medicine; Colin Young, MD - Department of Radiology and Biomedical Imaging, Yale School of Medicine; Mauricio Mandel Brigido, MD, PhD - Department of Neurosurgery, Yale School of Medicine; Todd Constable, MD - Department of Radiology and Biomedical Imaging, Yale School of Medicine; Ming-Kai Chen, MD - Department of Radiology and Biomedical Imaging, Yale School of Medicine; Eyiyemisi Damisah, MD - Department of Neurosurgery, Yale School of Medicine; Dennis Spencer, MD - Department of Neurosurgery, Yale School of Medicine; Pue Farooque, DO - Department of Neurology, Yale School of Medicine

Rationale: Our group recently published on predictors of surgical failure in patients undergoing temporal lobectomy and found that rapid spread of seizures outside of resection margins (< 10 seconds) on intracranial electroencephalography (iEEG) was highly predictive of surgical failure.1 Given these findings, we sought to evaluate the predictive value of non-invasive markers (scalp EEG and functional imaging) on surgical outcome after temporal lobectomy.

Methods: The Yale Epilepsy Databank was reviewed for temporal lobectomy patients whose pre-surgical evaluation consisted of scalp EEG, quantitative positron emission tomography (qPET), and resting state functional magnetic resonance imaging (fMRI) between 2007 – 2019. Post-surgical outcome (at a minimum of one-year follow-up) was classified as good (International League Against Epilepsy (ILAE) Class 1 – 2) and poor (ILAE Class 3 – 5). Scalp interictal and ictal EEG was reviewed independently by two epileptologists and seizure spread was classified as lateralized (involvement of two or more lobes) or contralateral (involvement of the opposite hemisphere). Flurodeoxyglucose-PET (FDG-PET) was compared to a group of controls without epilepsy to generate qPET with z-scores for all anatomic regions. Resting state fMRI data was compared to age-matched controls to create z-scores for all anatomic regions. Univariate significance tests were used to select candidate features (threshold p < 0.05), which were then centered and scaled. Elastic-net regression was used for classification of surgical outcome. Leave-one-out cross-validation (LOOCV) was used for training on the first 25 consecutive patients. The model was internally validated on the last 10 consecutive patients.
Surgery