Stimulation of non-motor subthalamic nucleus impairs selective response inhibition via prefrontal connectivity.

Abstract

Given the inconsistent results in the past, there is an ongoing debate whether and how deep brain stimulation in the subthalamic nucleus modifies cognitive control processes like response inhibition in persons with Parkinson's disease. In this study, we examined how the location of the stimulation volume within the subthalamic nucleus affects the performance in an antisaccade task but also how its structural connectivity is related to response inhibition. Antisaccade error rates and latencies were collected in 14 participants on and off deep brain stimulation in a randomized order. Stimulation volumes were computed based on patient-specific lead localizations using preoperative MRI and postoperative CT scans. Structural connectivity of the stimulation volumes with pre-defined cortical oculomotor control regions as well as whole-brain connectivity was estimated using a normative connectome. We showed that the detrimental effect of deep brain stimulation on response inhibition, measured as antisaccade error rate, depended upon the magnitude of the intersection of volumes of activated tissue with the non-motor subregion of the subthalamic nucleus and on its structural connectivity with regions of the prefrontal oculomotor network including bilateral frontal eye fields and right anterior cingulate cortex. Our results corroborate previous recommendations for avoidance of stimulation in the ventromedial non-motor subregion of the subthalamic nucleus which connects to the prefrontal cortex to prevent stimulation-induced impulsivity. Furthermore, antisaccades were initiated faster with deep brain stimulation when the stimulation volume was connected to fibres passing the subthalamic nucleus laterally and projecting onto the prefrontal cortex, indicating that improvement of voluntary saccade generation with deep brain stimulation may be an off-target effect driven by stimulation of corticotectal fibres directly projecting from the frontal and supplementary eye fields onto brainstem gaze control areas. Taken together, these findings could help implement individualized circuit-based deep brain stimulation strategies that avoid impulsive side effects while improving voluntary oculomotor control.