Abstracts

Rationale: Responsive neurostimulation (RNS) is a promising treatment for drug-resistant focal epilepsy; however, clinical outcomes are highly variable across individuals. The therapeutic mechanism of RNS likely involves modulatory effects on brain networks; however, with no known biomarkers that predict clinical response, patient selection remains empiric. In this study, we aim to determine whether functional brain connectivity measured non-invasively prior to device implantation predicts clinical response to RNS therapy.

Methods: This is a single-center retrospective cohort study of participants with drug-resistant focal epilepsy who underwent magnetoencephalography (MEG) and were subsequently implanted with an RNS device between August 15, 2014 and March 31, 2020. Of 29 total participants, 26 participants had stimulation enabled and were included in the analysis. Stimulation was enabled for a median (interquartile range) duration of stimulation of 45.0 (IQR 38.0-54.0) months across all participants. Functional connectivity (FC) across multiple spatial scales (global, hemispheric, and lobar) were computed from pre-implantation resting-state MEG. FC were normalized against spatial maps of healthy individuals and investigated as predictors of clinical response, defined as percent change in self-reported seizure frequency at last clinic-visit relative to pre-RNS baseline. Area under the receiver operating characteristic (ROC) curve (AUC) quantified the performance of FC measures in classifying participants as responders ( >50% reduction in seizure frequency) or non-responders (≤50%).

Results: Demographics of responders (N=20) and non-responders (N=6) were not statistically different. Non-responders had a lower baseline seizure frequency than responders (p=0.017). Mean global FC in the alpha and beta frequency bands were lower in non-responders, as compared to responders (alpha, p=0.041; beta, p=0.012), enabling classification of RNS responder status (AUC 0.783 [95% CI, 0.588-0.979], 0.850 [95% CI, 0.702-0.998], and 0.908 [95% CI, 0.793-1] for alpha, beta, and combined frequency bands, respectively). Mean global, hemispheric, and lobar FC in the alpha range correlated with seizure frequency reduction (global, p=0.008; hemispheric, p=0.017; lobar, p=0.043).

Conclusions: FC is a personalized, non-invasively measured biomarker that has the potential to predict clinical response to RNS therapy prior to device implant. With further validation, this biomarker may facilitate identification of patients who are most likely to benefit from RNS. These findings also support an emerging view that the therapeutic mechanism of RNS depends on network-level effects in the brain.

Funding: Please list any funding that was received in support of this abstract.: Research reported in this publication was supported by the National Center for Advancing Translational Sciences of the NIH under Award Number (5TL1TR001871-05 to JMF). Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the NIH.