Authors :
Presenting Author: Matthew Pesce, BS – Boston Children's Hospital
Jeffrey Bolton, MD – Neurologist, Department of Neurology, Boston Children's Hospital; Ellen Grant, MD – Director of the Fetal-Neonatal Neuroimaging and Developmental Science Center, Fetal-Neonatal Neuroimaging and Developmental Science Center, Boston Children's Hospital; Joseph Madsen, MD – Director of Epilepsy Surgery, Department of Neurosurgery, Boston Children's Hospital; Navaneeth Makaram, PHD – Postdoctoral Research Fellow, Fetal-Neonatal Neuroimaging and Developmental Science Center, Boston Children's Hospital; Christos Papadelis, PHD – Director of Research, Jane and John Justin Neurosciences Center, Cook Children’s Health Care System; Phillip Pearl, MD – Director of Epilepsy and Clinical Neurophysiology, Department of Neurology, Boston Children's Hospital; Alexander Rotenberg, MD, PhD – Epileptologist; Director of the Neuromodulation Program, Department of Neurology, Boston Children's Hospital; Scellig Stone, MD, PhD, FRCSC – Neurosurgical Director, Department of Neurosurgery, Boston Children's Hospital; Eleonora Tamilia, PHD – Instructor of Pediatrics, Fetal-Neonatal Neuroimaging and Developmental Science Center, Boston Children's Hospital
Rationale:
The gold standard to estimate the epileptogenic zone (EZ) in drug-resistant epilepsy (DRE) is localizing the seizure onset with intracranial EEG (iEEG), requiring many days of recordings. Estimating the EZ via interictal data would largely benefit presurgical planning, but an undisputable interictal EZ biomarker has yet to be identified.
A traditional approach involves locating the irritative-zone (IZ: contacts with spikes) or HFO-zone (HFOZ: contacts with HFOs). Yet, these are local biomarkers, while DRE is now considered a distributed brain network abnormality, where Functional Connectivity (FC) can outline hyperconnected seizure networks.
We present novel iEEG interictal measures to estimate the EZ by identifying spike and HFO hubs: we integrate traditional spike and HFO rates (IZ, HFOZ) with FC estimation between contacts (functional-network). We aim to assess whether integrating information from spike and HFO hubs allows to estimate the EZ and predict postsurgical outcome in children.
Methods:
We studied iEEG (5 min, Fig 1A) of 40 epilepsy surgery patients with known Engel outcome. For each patient, we:
Computed the rates of spikes, ripples and fast ripples per iEEG contact (Fig 1B);
Ran FC analysis for spikes and HFOs, obtaining three FC matrices - Fig 1C;
Estimated new combined measures of “hubness” that quantify the spike or HFO rate of each contact plus the rates of the connected contacts (Fig 1D).
To assess the presurgical value of our interictal measures, we compared (
Wilcoxon sign-rank) values inside and outside the resection (defined by postop-MRI) in good and poor outcome patients. We performed logistic regression to predict postsurgical outcome integrating spike/HFO rates and network measures (estimated outside the resection). ROC curve and confusion matrix were generated with the regression outputs.
Results:
Traditional spike rate was higher inside than outside the resection in both good and poor outcomes (Fig 2A), suggesting that it would likely locate areas that were already resected in poor outcomes. Traditional HFO rates did not show differences (p >0.05; Fig 2B).
Spike connectivity instead was higher inside than outside resection in good (p< 0.001) but not in poor outcomes (Fig 2A). The same results were found for ripple connectivity (Fig 2B) and fast ripple integrated measure of hubness (Fig 2C). This suggests the potential of the network measures to pinpoint areas outside resection in poor outcome cases.