Introduction: Responsive neurostimulation (RNS) is an FDA approved intervention for epilepsy patients who are medication resistant and are poor candidates for surgical removal of brain tissue at the seizure onset zone. RNS treatment involves the chronic implant of two electrode leads that continuously monitor neural activity near the seizure onset zone, and respond with electrical stimulation when a seizure is detected. For most patients, RNS has led to a significant reduction in seizure frequency over the course of 2-3 years, however not all patients respond positively and patients rarely become seizure free. RNS therapy is difficult to optimize as the mechanism of action is currently unknown, and the selection of implant location and stimulation and detection parameters remains largely trial and error. The primary objective of this study is to implement a patient-specific volume conduction model to determine the electrical impact of RNS stimulation settings on the seizure onset zone (SOZ), and then associate this with patient outcome over time. 

Materials & Methods: Pre-implant MRI images, electrode coordinates, and RNS stimulation parameters were obtained from 6 epilepsy patients undergoing RNS therapy after intracranial EEG (iEEG) monitoring in the EMU. A whole-brain distributed dipole model was generated with the reconstructed preoperative MRI scan of each patient using Brainstorm software. The electric potential induced at the SOZ by stimulation from the RNS leads was calculated over time for each patient, accounting for the changing stimulation intensity and configuration of active RNS electrodes. The magnitude of the potential field at the SOZ was correlated with two measures of seizure reduction outcomes (patient reported and device recorded) over time at both the individual and group levels. 

Results & Discussion: We discovered that proximity of RNS electrode leads to the SOZ, as measured by leadfield volume conduction magnitude, did not have a significant association with a patient’s seizure reduction for either outcome measure, after correction for multiple comparisons. Our results suggest that the short-term impact of stimulation at the SOZ may not be a primary factor contributing to seizure reduction in RNS patients and that surgical planning of RNS lead targets should not be constrained by the SOZ site. Validation on a larger group of patients should be conducted before these results influence clinical practice. Our results are corroborated by the theory that epilepsy is a network disorder and the RNS device serves to modulate that network by strengthening inhibitory connections. Prior work suggests that it is more important to locate RNS electrodes near particular brain structures rather than the SOZ.

Conclusions: In this proof-of-concept study, we have developed a method for simulating patient-specific RNS lead activation parameters, and hope to use our technique in future studies with an expanded dataset to uncover alternative anatomical implant target locations that better correspond with seizure reduction outcomes.