Post-Doctoral Fellow University of California, San Francisco San Francisco, California
Rationale: A Responsive Neurostimulation (RNS) System that delivers focal electrical stimulation to the brain in response to seizures is gaining traction as a long-term therapy for patients with medically-refractory epilepsy. This therapy reduces the rate of seizure occurrence by an average 75% over nine years. However, the magnitude and time-scale of therapeutic benefit is highly variable and difficult to track – particularly in patients with rare seizure occurrence. Rational biomarkers that track therapeutic efficacy based on electrophysiological assay of patient background state would inform a more continuous treatment strategy. Recent studies have demonstrated that networks from which seizures begin may be localizable based on interictal functional connectivity derived from spontaneous intracranial neural recordings. In this study, we tested the hypothesis that patients who are better responders to RNS therapy exhibit a stereotypic reorganization of connectivity that differs from patients who are poor responders to therapy. Methods: To evaluate the relationship between interictal network physiology and clinical outcome to RNS therapy, we analyzed long-term, intracranial EEG data during seizure-free periods from 42 patients with medically-refractory epilepsy (n=23 neocortical-onset; n=18 mesial temporal-onset). We evaluated signatures of variability in neural signal fluctuation, operationalized as the line-length, and functional connectivity, operationalized as frequency-dependent phase coherence, as predictors of outcome. Specifically, we used a combination of unsupervised machine learning and linear regression to relate change in the constituent components of a patient’s interictal network during RNS therapy to patient outcome (Figure 1). Results: We found evidence of reorganized functional connectivity over time, specific to epilepsy localization and degree of seizure reduction. Patients whose RNS targeted neocortical or mesial temporal structures were more likely to have better outcome with increased network connectivity within either a broad frequency range above 15 Hz (beta and gamma band) or a narrow frequency band between 8-15 Hz (alpha band), respectively. Leave-one-out cross-validated regression models demonstrated the capability to predict percent reduction in a patient’s seizures based on the magnitude increase in frequency-specific network connectivity within 12 months of RNS implant (Figure 2). Conclusions: RNS System is capable of indirectly modulating the interictal background electrophysiology. These electrophysiological changes are reflected in network connectivity, the degree to which is associated with the percent reduction in seizures since the implant of the RNS System. Our findings reveal that RNS may impact epilepsy electrophysiology beyond the impending seizure, either directly or indirectly, and these changes are significantly correlated with clinical outcome to therapy. These results offer new candidate biomarkers to objectively and continuously track, and potentially prognosticate, the efficacy of RNS therapy alongside current gold-standard measures of patient-reported outcome. Funding: Please list any funding that was received in support of this abstract.: None Click here to view image/table
