Deep brain stimulation of the subthalamic nucleus: A two-edged sword

Chronic high frequency stimulation of the subthalamic nucleus (STN) of the basal ganglia is a highly effective treatment for Parkinson’s Disease (PD). Such deep brain stimulation is thought to suppress spontaneous, including pathological, activity in the basal ganglia [1–5]. Equally, however, it must also remove any residual physiological functioning in these key motor structures, and yet there is paradoxically little evidence to suggest that the motor action of the limbs is in any way further impaired by stimulation or even focal ablation of the STN in PD patients [6–8]. This has led to the influential hypothesis that the human basal ganglia are not necessary for simple limb movements, but are particularly involved in more subtle and complex functions, such as the promotion of motor flexibility, that are not readily revealed by standard tests of motor behavior [6]. Here we show that the basal ganglia are involved in processing simple limb movements in the human, by separating the effects of deep brain stimulation on pathological and physiological activities based on baseline task performance. Our hypothesis was straightforward. Patients that, at the time of study, have performance in a simple motor task that is compromised by PD will improve with deep brain stimulation, in tandem with the suppression of pathological activity by stimulation [1–5]. In contrast, in those patients with relatively intact task-related processing, as evidenced by a baseline performance within normal limits, deep brain stimulation would be expected to suppress physiological processing and thereby impair performance. We tested rapid repetitive depression of a keyboard key with the forefinger as our simple voluntary movement. Thirteen patients with PD (see Table S1 in the Supplemental data available online) and chronically implanted STN electrodes were studied after overnight withdrawal of anti-parkinsonian medication, although the long action of many of the drugs used to treat PD meant that patients were still likely to have been partially treated when assessed. Each hand was tested separately with and without deep brain stimulation at 130 Hz and the percentage change in tapping speed during deep brain stimulation calculated (see Supplemental Experimental Procedures). There was a striking correlation between baseline performance in the task and percentage change with deep brain stimulation, whereby those sides showing the best performance without deep brain stimulation actually slowed during stimulation, whereas those with poorer performance got quicker in the task (Figure 1A). Next, we divided sides into two groups according to whether or not tapping performance off deep brain stimulation was within the normal limits established on 12 sides in six healthy age-matched subjects (see Supplemental Experimental Procedures). Tapping rates were confirmed to improve on those sides where performance was compromised by parkinsonism at the time of study, but deteriorated on those sides in which tapping rates were within normal limits, whether absolute (Figure 1B) or percentage change (Figure 1C) tapping rates were considered. Finally, we confirmed the reproducibility of our findings by repeating the experiment on seven patients (13 sides) with PD (see Supplemental Table S1). Again there was a negative correlation (r = –0.558, p = 0.048) between the percentage change in tapping rate of each hand with deep brain stimulation and tapping rate prior to onset of deep brain stimulation. More importantly, tapping rates deteriorated on those six sides in which performance was within normal limits without deep brain stimulation, whether absolute (tapping rate 154.7 ± 3.3 off deep brain stimulation, 144.5 ± 5.5 during deep brain stimulation, p = 0.02, unpaired two-tailed t-test) or percentage change (mean %