Three Nomograms and a Scale to Predict the Outcome of Epilepsy Surgery – a Multicenter Head-to-head Comparison and External Validation Study
Abstract number :
1.342
Submission category :
9. Surgery / 9C. All Ages
Year :
2022
Submission ID :
2204145
Source :
www.aesnet.org
Presentation date :
12/3/2022 12:00:00 PM
Published date :
Nov 22, 2022, 05:23 AM
Authors :
Levente Hadady, MD – University of Szeged; Beliz Hazan, PhD – Thomas Jefferson University; Péter Klivényi, MD, PhD – Professor, Department of Neurology, University of Szeged; Dániel Fabó, MD, PhD – National Institute of Mental Health, Neurology and Neurosurgery, Budapest, Hungary; Guido Rubboli, MD, PhD – Professor, Danish Epilepsy Center, Dianalund, Denmark; Michael Sperling, MD, PhD – Professor, Thomas Jefferson University; Sándor Beniczky, MD, PhD – Professor, Department of Neurophysiology, Danish Epilepsy Center, Dianalund, Denmark
Rationale: The aim of current study was to externally validate and to compare four different epilepsy surgery outcome prediction systems, three nomograms (Jehi et al. 2015, Fitzgerald et al. 2021) and the Epilepsy Surgery Grading Scale (ESGS) (Dugan et al. 2017).
Methods: A total of 1394 patients with therapy-resistant focal epilepsy, undergoing surgical evaluation in three different centers (Comprehensive Epilepsy Center, Thomas Jefferson University, Philadelphia, United States of America; Danish Epilepsy Center, Dianalund, Denmark; National Institute of Mental Health, Neurology and Neurosurgery, Budapest, Hungary) were retrospectively enrolled. Relevant data were extracted from the patients’ charts. Individual predictions were calculated using the nomograms, and patients were classified into the three groups according to the ESGS. Progression to surgery, postsurgical seizure freedom, and Engel Score were assessed at two and five years after surgery. We evaluated the nomograms using Harrel’s concordance statistic and calibration curves. A value below 0.5 indicates a poor model. A value of 0.5 indicates that the model is no better out classifying outcomes than random chance.
Results: A total of 1297 patients underwent epilepsy surgery following the evaluation. The models for complete seizure freedom had a c-statistics of 0.60 at two years with clinical features only, 0.592 including the scalp EEG and MRI findings, and 0.614 at five years. For an Engel score of 1 the c-statistics was 0.60 and 0.61 at two and five years, respectively. When, in contrary to the original nomogram, limiting the presence of focal-to-bilateral tonic-clonic seizures to the one year before the surgery, we acquired generally better fitting calibration curves and slightly higher concordance statistic values (mean improvement was 0.01, highest improvement was 0.2). With ESGS, we observed significant difference between grades 1 vs. 2, 1 vs. 3 and 2 vs. 3. (p = 0.001, p < 0.001 and p = 0.001 respectively) in the presurgical cohort. In the surgical group there was a significant difference between grade 1 vs. 3 and 2 vs. 3 (p = 0.022 and p = 0.003) but we did not observe a statistically significant difference between grades 1 vs. 2 (p = 0.47).
Surgery