Changes in Functional Connectivity Relate to Modulation of Cognitive Control by Subthalamic Stimulation

Abstract

Subthalamic (STN) deep brain stimulation (DBS) in Parkinson's disease (PD) patients not only improves kinematic parameters of movement but also modulates cognitive control in the motor and non-motor domain, especially in situations of high conflict. The objective of this study was to investigate the relationship between DBS-induced changes in functional connectivity at rest and modulation of response- and movement inhibition by STN-DBS in a visuomotor task involving high conflict. During DBS ON and OFF conditions, we conducted a visuomotor task in 14 PD patients who previously underwent resting-state functional MRI (rs-fMRI) acquisitions DBS ON and OFF as part of a different study. In the task, participants had to move a cursor with a pen on a digital tablet either toward (automatic condition) or in the opposite direction (controlled condition) of a target. STN-DBS induced modulation of resting-state functional connectivity (RSFC) as a function of changes in behavior ON versus OFF DBS was estimated using link-wise network-based statistics. Behavioral results showed diminished reaction time adaptation and higher pen-to-target movement velocity under DBS. Reaction time reduction was associated with attenuated functional connectivity between cortical motor areas, basal ganglia, and thalamus. On the other hand, increased movement velocity ON DBS was associated with stronger pallido-thalamic connectivity. These findings suggest that decoupling of a motor cortico-basal ganglia network underlies impaired inhibitory control in PD patients undergoing subthalamic DBS and highlight the concept of functional network modulation through DBS.

Keywords: Parkinson's disease; cognitive control; deep brain stimulation; functional connectivity; subthalamic nucleus.

FIGURE 1

(A) In the visuomotor task, patients either performed “automatic” movements with a pen following a cursor to hit a (green) target or performed inverted movements in a “conflicting” condition to hit a (red) target (pen‐to‐cursor mapping inverted). (B) DBS electrode localizations of PD patients. Active contacts (red) during bipolar stimulation in the STN (orange) are highlighted and superimposed on a section of the BigBrain ultrahigh resolution model (Amunts et al. 2013). (C) Stimulation of bipolar contacts of the electrode in the STN (orange) produces an oval‐shaped volume of tissue activated (VTA).

FIGURE 2

Behavioral and VTA‐STN overlap results. (A) Reaction time change under DBS differed by condition: ON STN‐DBS, patients slowed down less for conflicting trials than OFF DBS, compared to the automatic condition. (B) Movements were faster ON than OFF DBS in both automatic and conflicting conditions and movements were faster in automatic than conflicting trials. (C) The amount of associative STN stimulated by the VTA correlated significantly with the effect of STN‐DBS on the percentage decrease in reaction time adaptation (see Equation 1) and (D) the amount of sensorimotor STN stimulated correlated with the increase in movement velocity under STN‐DBS. Upper and lower boundaries of the boxes represent 25% and 75%, respectively; whiskers extend to 1.5 time the interquartile range; *p < 0.05; **p < 0.01; n.s., not significant.

FIGURE 3

Correlations between functional connectivity and behavior under DBS. (A) DBS‐induced reduced response slowing in the conflicting condition correlated with a reduction of connectivity in a cortico‐thalamic‐basal ganglia network. In other words, the less patients are able to slow down reaction times during conflict, the less connected motor/premotor cortex is with (i) thalamus and (ii) parts of the basal ganglia and the less connected STN is with GPe. (B) Average movement velocity increased with DBS and this increase correlated with higher connectivity between thalamus and both pallidal segments. *p < 0.05; blue lines indicate reduction of functional connectivity; red lines indicate increase of functional connectivity. A schematic representation of these findings is shown in Figure S3 and the raw averaged connectivity matrices are presented in Figure S4.