Stimulation of STN impairs aspects of cognitive control in PD
Citation Manager Formats
Make Comment
See Comments

Abstract
Objective: To test the hypothesis that subthalamic nucleus (STN) stimulation in Parkinson disease (PD) patients affects working memory and response inhibition performance, particularly under conditions of high demand on cognitive control.
Methods: To test this hypothesis, spatial working memory (spatial delayed response [SDR]) and response inhibition (Go–No–Go [GNG]) tasks requiring varying levels of cognitive control were administered to patients with PD with previously implanted bilateral STN stimulators (n = 24). Patients did not take PD medications overnight. Data were collected while bilateral stimulators were on and off, counterbalancing the order across subjects.
Results: On the SDR task, STN stimulation decreased patients’ working memory performance under a high but not low memory load condition (effect of stimulator condition on high load only and condition × load interaction, p < 0.05). On the GNG task, STN stimulation reduced discriminability on a high but not medium inhibition condition (effect of stimulator condition on high inhibition level only, p = 0.05; condition × inhibition level interaction, p = 0.07).
Conclusion: STN stimulation reduces working memory and response inhibition performance under conditions of greater challenge to cognitive control despite significant improvement of motor function.
Despite the effectiveness of deep brain stimulation (DBS) of the subthalamic nucleus (STN) for motor symptoms of Parkinson disease (PD),1,2⇓ there has been concern about cognitive impairments following the procedure, as the surgery requires electrode penetrations through the frontal lobe.3 Some studies have reported that surgery decreased performance on tasks that depend on prefrontal cortex4,5⇓ and require cognitive control, the active monitoring and manipulation of information in response to internal goals.6 However, presurgical vs postsurgical comparisons can be confounded by effects of time, changes in medication, practice effects on clinical neuropsychological tasks, and the absence vs presence of STN stimulation.
Through its connections to prefrontal cortex, the STN may play an important role in tasks requiring cognitive control, such as working memory and response inhibition.7-9⇓⇓ Thus, it is possible that STN stimulation itself alters performance on these tasks independent of surgical effects. Previous studies of the effects of STN stimulation on cognition have reported conflicting results.10-13⇓⇓⇓ This inconsistency may depend in part on task difficulty and level of demand on prefrontally mediated cognitive control processes.
Materials and methods.
Subjects.
This study was approved by the Institutional Review Board at Washington University School of Medicine, and all participants gave informed consent. Twenty-four patients with PD with previously implanted bilateral STN stimulators were studied. Patients met the diagnostic criteria for clinically definite PD.14 The surgical implantation of stimulators (Medtronic model 3389 DBS leads, Minneapolis, MN) targeted STN with a technique that combines conventional stereotactic planning using formulas with reference to the anterior–posterior commissural line, visual targeting on T2-weighted MRI, frame-based targeting using computerized methods (Medtronic Stealth Station, Framelink IV), and microelectrode recording.15 Intraoperative test stimulation confirmed optimal location of electrodes, similar to other published methods.15 The degree of subsequent clinical benefit achieved by stimulation, as measured by change in Unified Parkinson’s Disease Rating Scale (UPDRS) Motor subscores, is comparable with other centers.16 Exclusionary criteria included a history of neurologic events or diagnoses other than PD or dementia on clinical exam prior to surgery. Patients had to be at least 2 months post STN stimulator implantation to allow time for programming to achieve good clinical benefit for each subject.
The 24 subjects (8 female, 16 male) had been diagnosed with PD for 12.5 years on average and were tested between 2 and 15 months following STN stimulator implantation. See table 1. All subjects except 1 were taking levodopa/carbidopa daily, and 13 of the 24 were taking additional PD medications (amantadine, tolcapone, trihexyphenidyl, pergolide, pramipexole, ropinirole, entacapone, selegiline). All subjects except one were right handed. Subjects used their dominant hands to respond on the computer.
Table 1 Clinical and demographic variables
Protocol.
On the study day, subjects refrained from taking any PD medications for at least 12 hours prior to testing. Cognitive testing was done with either both STN stimulators on or with both stimulators off, and the order of these two conditions was randomly selected for each subject. Patients and cognitive examiners were not told the order of stimulator conditions. Stimulator settings were changed at least 40 minutes prior to testing.
Cognitive testing took 20 minutes for each condition. Patients sat in a chair facing a computer screen and performed two computerized tasks that require cognitive control: working memory and response inhibition. Each task included two conditions with varying levels of demand on cognitive control processes.
Spatial delayed response task.
The spatial delayed response (SDR) task17 is an experimentally derived working memory task that has been closely linked to lateral prefrontal cortex functioning in animals and humans.18-22⇓⇓⇓⇓ Working memory tasks require individuals to maintain, monitor, and use internal information to guide behavior and thus engage cognitive control processes. Working memory load was varied to require different levels of these control processes. Working memory is thought to be a fundamental cognitive skill that supports many other more complex “executive” functions such as planning and organizing complex behavior, reading comprehension, fluid intelligence, and the capacity to draw inferences.23,24⇓ PD patients are impaired on verbal and spatial working memory tasks, even at the beginning stages of their disease25-27⇓⇓ (perhaps because of altered caudate-modulated basal ganglia output or changes in mesocortical dopaminergic pathways).28 Poor working memory performance correlates with increased severity of PD26 and with worse Mini-Mental State Examination (MMSE) scores.27 Cognitive impairment as measured by the MMSE in PD is strongly related to decreased quality of life.29
In this task, subjects focused on a central fixation cross on a computer screen placed approximately 60 cm away from the subject. While fixated, either one or two cues (each 1 cm in diameter) appeared for 150 milliseconds in any of 32 possible unmarked locations at an 11.5-cm radius from the central fixation. A delay period (5 or 15 seconds) was then imposed. During the delay, subjects performed a continuous performance task in which a series of geometric shapes (triangle, square, and diamond) appeared in place of the fixation cross (1,000-millisecond duration, 750- to 1,250-millisecond intertrial interval). Subjects pressed the spacebar whenever the diamond shape appeared. Reaction times and accuracy rates were recorded. After the delay, the fixation cue returned, and subjects pointed on the computer screen where they remembered seeing the cue(s). Responses were measured in X and Y coordinates and compared with the actual location of the cue. Delay trials and trials with no mnemonic load (cue-present trials) were presented in random order. On the cue-present trials, the cue was present during the response phase. This set of trials gave an indication of subjects’ pointing accuracy. Mean error in millimeters (distance between recall and actual target) was calculated for each subject for each type of trial. There were either one or two cues to be remembered on each trial. In the two-cue condition, both locations were presented simultaneously, and in the recall phase, subjects pointed to both locations, in any order desired. Forty experimental trials were presented: 20 with only one cue presented and 20 with two cues presented. Trials were blocked by number of cues. Within each block, conditions (cue present, 5-second delay, 15-second delay) were presented in random order. The order of blocks was counterbalanced across subjects. Subjects performed four cue-present trials and eight trials per delay for each block.
Go–No–Go task.
The Go–No–Go (GNG) task assessed the ability to inhibit a prepotent response under conditions of low or high prepotent response strength30-32⇓⇓ and requires active cognitive control processes such as conflict monitoring. Response inhibition skills are thought to play a role in the effortful control of behavior and the guidance of behavior by internal rather than external cues. These skills are necessary for optimal performance on a variety of complex tasks including working memory, set shifting, and/or maintenance.30-33⇓⇓⇓ PD patients are impaired on tasks emphasizing response inhibition,33 perhaps owing to altered basal ganglia output to anterior cingulate cortex or altered mesocortical dopaminergic pathways. Poor performance significantly predicts later cognitive decline and correlates with overall UPDRS scores.33,34⇓
This task involved monitoring a visual display while single uppercase letters were presented one at a time interspersed with the number 5 (250-millisecond duration, 1,000-millisecond intertrial interval). In this task, participants were instructed to push a target response button at the occurrence of every letter but to withhold a response when the number 5 was presented. Target frequency (percentage of trials where a button press was required, e.g., letters) was manipulated in a blocked fashion. There were two levels of target frequency (medium = 50%; high = 83%). The high target frequency block is designed to produce a strong prepotent response (e.g., press the button) and so is more challenging for response inhibition skills. One block at each frequency level was performed with the order randomly determined for each subject. Each block contained 150 trials. Reaction times and accuracy rates were recorded.
Analysis.
To determine effects of bilateral stimulation on performance, data were analyzed with repeated measures general linear models and paired t-tests. Level of demand on cognitive control processes and stimulator condition were repeated measures for both tasks. On the SDR task, mean error for the 15-second delay was the dependent variable. This task condition was likely to have the greatest sensitivity (e.g., smallest chance of ceiling effects). Number of cues (memory load) was the variable representing cognitive control demand. For the GNG task, a discriminability index (Pr = hits—false alarms)35 was the dependent variable, and target frequency (high vs medium) represented cognitive control demand. Secondary analyses were also conducted on other task variables using repeated measures general linear models or paired t-tests. Significance was set at p = 0.05 for tests of the hypothesis that cognitive control demand and stimulation condition would affect performance.
Results.
Subjects.
Twenty-four subjects were tested on both tasks. One subject’s SDR data were invalid owing to a technical failure, so data from 23 subjects were analyzed. Two subjects did not complete the GNG task in the off condition and another subject’s GNG data were invalid owing to a technical failure, so data from 21 subjects were analyzed.
All subjects had significant clinical improvement with stimulators on (off medications overnight; UPDRS Motor subscale scores; 50% decrease on average; paired t-test, p < 0.001) (see table 1). Variables for stimulators on the left and right side of the brain were statistically comparable (amplitude in volts, left mean = 3.2, SD = 0.5, right mean = 3.2, SD = 0.5; rate in pulses per second, left mean = 185, SD = 0; right mean = 185, SD = 0; pulse width in microseconds, left mean = 67, SD = 19, right mean = 70, SD = 20; impedance in ohms, left mean = 1,065, SD = 264, right mean = 1,087, SD = 256; paired t-test, p > 0.20). Fourteen subjects were tested first while stimulators were off, and 10 were tested first while stimulators were on.
Analyses.
SDR test.
There was an interaction between memory load and stimulator condition (F[1,22] = 11.3, p = 0.003). Performance became less accurate (paired t-test, t = −2.2, p = 0.04) with stimulation in the two-cue condition but did not change in the one-cue condition (paired t-test, t = 0.22, p = 0.82) (figure 1A). In addition, the main effect of memory load was significant (two-cue performance worse than one-cue overall, F[1,22] = 35.4, p < 0.001; in the off condition, p = 0.001, and in the on condition, p < 0.001, paired t-tests), but main effect of stimulator condition was not (F[1,22] = 0.76, p = 0.39). The difference between subjects who were tested off first and subjects who were tested on first on change in the two-cue, 15-second delay condition was not significant (t-test, t = −0.90, p = 0.38).
Figure 1. Effects of subthalamic nucleus (STN) stimulation on cognitive control. (A) Mean ± SEM spatial delayed response (SDR) performance (0 = perfect recall) on the 15-second delay, one-cue, and two-cue trials with bilateral stimulators off and on. Turning stimulators on decreased accuracy on the two-cue condition (increased recall error). (B) Mean ± SEM Go–No–Go (GNG) discriminability (1 = perfect performance) on medium and high target frequency conditions with bilateral stimulators off and on. Turning stimulators on decreased discriminability on the high target condition only (decreased accuracy).
The interaction between memory load, stimulator condition, and delay length was significant when all delay conditions (cue present, 5 and 15 seconds) were considered (F[2,21] = 5.3, p = 0.01). When the cue-present and 5-second delay conditions were considered separately, the main effects of stimulator condition and memory load and the interaction between the two variables were not significant (cue present: interaction, F[1,22] = 0.63, p = 0.44; main effect of condition, F[1,22] = 0.01, p = 0.93; main effect of memory load, F[1,22] = 1.87, p = 0.19); 5 seconds: interaction, F[1,22] = 0.5, p = 0.82; main effect of condition, F[1,22] = 3.2, p = 0.09; main effect of memory load, F[1,22] = 2.25, p = 0.15). In addition, there was no effect of stimulator condition or memory load on the delay period task performance in terms of correct hits or median reaction times (F[1,19] < 1.00, p > 0.33).
GNG test.
The interaction between target frequency and condition on GNG discriminability was not significant (F[1,20] = 3.6, p = 0.07), nor was the main effect of condition (F[1,20] = 2.82, p = 0.11). The main effect of target frequency was significant (high target frequency performance worse than medium target frequency overall, F[1,20] = 30.9, p < 0.001; in the off condition, p = 0.02, and in the on condition, p < 0.001, paired t-tests). To test the primary hypothesis further, each target frequency level was considered separately. Stimulator condition did have an effect on GNG discriminability for the high target frequency condition (paired t-test, t = 2.06, p = 0.05) but not for the less demanding medium target frequency GNG condition (paired t-test, t = −0.56, p = 0.58) (see figure 1B).
The main effect of stimulator condition and the interaction between stimulator condition and target frequency on median reaction times (correct hits only) were not significant (stimulator condition: F[1,21] = 1.78, p = 0.20; target frequency: F[1,21] = 0.63, p = 0.44; table 2). There was a main effect of target frequency on reaction time, with faster reaction times in the high target frequency condition (F[1,21] = 32.71, p < 0.001). In post hoc paired t-tests, this effect was significant in both the off (t = −3.6, p = 0.002) and the on (t = −4.6, p < 0.001) stimulator conditions.
Table 2 Mean (SD) for selected cognitive variables from the SDR and GNG tasks
As an exploratory analysis, correlations were performed between individual change in reaction times and accuracy on the GNG task across stimulator conditions. Stimulator-related reductions in GNG discriminability correlated with reductions in reaction time in the high but not medium target frequency level (high: r = 0.50, p = 0.02; medium: r = −12, p = 0.62) (figure 2). Decreases in GNG discriminability appeared to reflect primarily decreased nontarget accuracy (increased false alarms), as change in discriminability correlated with change in nontarget accuracy (r = 0.94, p < 0.001) but not change in target accuracy (r = −0.08, p = 0.73).
Figure 2. Relationship between reaction time and discriminability on the Go–No–Go (GNG) task across stimulator conditions. Percentage change in GNG discriminability across stimulator conditions correlated significantly with percentage change in reaction time across stimulator conditions.
There were no differences in high target performance variables between subjects who were tested off first and subjects who were tested on first (t-tests, p > 0.08). The correlation between change in GNG high target discriminability and change in the SDR 15-second two-cue condition was not significant (r = 0.28, p = 0.23). Correlations between change in UPDRS Motor rating and change in GNG high target discriminability or the SDR 15-second two-cue condition were not significant (GNG: r = −0.20, p = 0.38; SDR: r = 0.04, p = 0.84).
Discussion.
Bilateral STN stimulation decreased memory performance on the SDR task when memory load was highest and decreased discriminability on the GNG task when demands on inhibitory control were greatest. These data are consistent with the hypothesis that STN stimulation interferes with patients’ ability to handle higher demands on cognitive control processes. This performance pattern could be caused by a common mechanism such as decreased conflict monitoring or by separate cognitive control mechanisms.
These cognitive changes are not easily explained by changes in overall motor performance. Decreased accuracy on the SDR task and decreased response inhibition on the GNG task were found in the context of overall improved motor performance. In addition, these effects occurred on harder, but not easier, cognitive control conditions, despite equivalent motor demands (e.g., two-cue 15-second performance vs two-cue 5-second performance). Further, on a motor control condition (cue present) of the SDR task, patients did not perform differently across conditions. However, the effect of stimulation on GNG performance was weaker, and there was a more complicated relationship between motor change and GNG performance. Reaction times did not improve significantly with stimulation, but individual improvements in reaction time were related to decreased discriminability, which primarily reflected increased false alarm rates. These data suggest that shortening reaction times permit subjects to improve their ability to generate a response quickly when needed (Go trials) but impair their ability to inhibit this response when needed (No–Go trials). Thus, we hypothesize that whereas STN stimulation has a beneficial effect on neural circuits producing involuntary tremor, bradykinesia, and rigidity, it simultaneously interferes with neural circuits subserving the cognitive control or “braking” of voluntary motor behavior.36
It is unknown whether the degree of change in cognitive performance seen in this study (19% for SDR; 10% for GNG) has a direct impact on daily function in these patients. Based on the fundamental nature of the skills assessed by these tasks,24,37⇓ it seems possible that decreases in these skills would have downstream effects on more complex cognitive control tasks and tasks of everyday life. As yet, these hypotheses have not been directly tested in PD patients with STN stimulators. However, the cognitive effects in this study may be relevant to clinical reports of decreased attention and executive skills following STN stimulator placement.5 For example, one study4 found that STN surgery impaired tasks requiring mental manipulation (e.g., Backwards Digit Span, Trail Making B, a conditional associative learning test) and those that rely heavily on the imposition of strategies (e.g., verbal fluency, verbal and visual learning tasks). Notably, this study compared presurgical (no stimulation) with postsurgical (on stimulation) performance. Based on the current findings, it is possible that some of these surgical effects are actually due to the stimulation itself and not exclusively due to the implantation of electrodes. Consistent with this hypothesis, some reports found that turning STN stimulators on decreased performance under conditions of strong response conflict (Stroop).12,13⇓ Interestingly, rats with STN lesions also have impaired ability to inhibit responses under conditions of strong conflict.8,38,39⇓⇓ In contrast, two studies on PD patients report that turning on STN stimulators improved working memory.10,11⇓ However, one of the studies used a task comparable with the SDR one-cue condition in the current study, in which no change due to stimulation was seen and in which the demands on cognitive control are minimized. In addition, both studies tested patients while they were on levodopa, which may alter working memory performance.40,41⇓
The effect of STN stimulation on cognitive control processes could be mediated through multiple pathways that could be affected by stimulation in the region of the STN. Although the most dorsolateral two-thirds of the STN is “sensorimotor,” with afferents from motor and supplementary motor cortex and the external segment of the globus pallidus and efferents to sensorimotor regions of putamen, the most ventromedial and rostral portion of STN is “associative” or cognitive, with afferents from prefrontal regions associated with cognitive control functions (Brodmann areas 6, 8, and 9) and efferents to associative areas of caudate, putamen, and globus pallidus.9 In addition, the STN projects to the globus pallidus internus and the substantia nigra pars reticulata, which in turn projects directly and indirectly to dorsolateral prefrontal cortex (Brodmann areas 9 and 46)36,42⇓ and indirectly to anterior cingulate, both of which are associated with cognitive control.6
Although the surgical techniques and clinical benefit achieved in our sample are comparable with those in other centers,15,16⇓ some uncertainty remains regarding the precise location and field of stimulation of the active electrodes. The degree of cognitive change caused by STN stimulation may depend on variables that influence the spread of current to nonmotor regions of the STN, such as the span of the electrode array, angle of STN penetration, and stimulation parameters. Future studies that incorporate electrode placement and predictions of the spread and strength of current may be useful in understanding the anatomic pathways underlying cognitive effects of stimulation. In addition, further work needs to be done to understand how these cognitive changes relate to clinically used measures of cognitive function and whether they impact daily functioning and quality of life for these patients. Ultimately, this information may be important in the development of strategies to minimize the impact of stimulation on cognitive function and may provide new insights into understanding the role of basal ganglia–thalamocortical pathways that mediate aspects of cognitive control.
Acknowledgments
Supported by the Greater St. Louis Chapter of the American Parkinson’s Disease Association (APDA), NIH (NS41248, NS41509), American Academy of Neurology, APDA Advanced Center for Parkinson’s Disease Research at Washington University, and the Barnes–Jewish Hospital Foundation (Jack Buck Fund for Parkinson’s Disease Research).
The authors thank Deanna Barch, PhD, and Todd Braver, PhD, for use of the GNG task and for helpful advice and Dana Perantie for expert assistance.
Footnotes
Dr. Revilla received partial fellowship funding in excess of $10,000 from Medtronic, the manufacturer of the implanted stimulators. Dr. Dowling has acted as a paid consultant for Medtronic.
- Received September 10, 2003.
- Accepted in final form December 11, 2003.
References
- ↵
- ↵Vingerhoets FJG, Villenmure JG, Temperli P, et al. Subthalamic DBS replaces levodopa in Parkinson’s disease: two-year follow-up. Neurology. 2002; 58: 396–401.
- ↵
- ↵
- ↵
- ↵
- ↵Chudasama Y, Baunez C, Robbins TW. Functional disconnection of the medial prefrontal cortex and subthalamic nucleus in attentional performance: evidence for corticosubthalamic interaction. J Neurosci. 2003; 23: 5477–5485.
- ↵
- ↵
- ↵Pillon B, Ardouin C, Damier P, et al. Neuropsychological changes between “off” and “on” STN or GPi stimulation in Parkinson’s disease. Neurology. 2000; 55: 411–418.
- ↵Rivaud–Pechoux S, Vermersch A, Gaymard B, et al. Improvement of memory guided saccades in parkinsonian patients by high frequency subthalamic nucleus stimulation. J Neurol Neurosurg Psychiatry. 2000; 68: 381–384.
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵Goldman–Rakic PS, Funahashi S, Bruce CJ. Neocortical memory circuits. Cold Spring Harbor Symp Quant Biol. 1990; 55: 1025–1038.
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵Mahieux F, Fenelon G, Flahault A, et al. Neuropsychological prediction of dementia in Parkinson’s disease. J Neurol Neurosurg Psychiatry. 1998; 64: 178–183.
- ↵
- ↵
- ↵
- ↵
- ↵Baunez C, Nieoullon A, Amalric M. In a rat model of parkinsonism, lesions of the subthalamic nucleus reverse increases of reaction time but induce a dramatic premature responding deficit. J Neurosci. 1995; 15: 6531–6541.
- ↵
- ↵
- ↵Middleton FA, Strick PL. Anatomical evidence for cerebellar and basal ganglia involvement in higher cognitive function. Science. 1994; 266: 458–461.
Letters: Rapid online correspondence
REQUIREMENTS
You must ensure that your Disclosures have been updated within the previous six months. Please go to our Submission Site to add or update your Disclosure information.
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.
You May Also be Interested in
Dr. Babak Hooshmand and Dr. David Smith
► Watch
Related Articles
- No related articles found.
Alert Me
Recommended articles
-
Articles
Subthalamic nucleus stimulation modulates afferent inhibition in Parkinson diseaseA. Sailer, D. I. Cunic, G. O. Paradiso et al.Neurology, January 29, 2007 -
Articles
Effects of subthalamic nucleus stimulation and levodopa on freezing of gait in Parkinson diseaseM. U. Ferraye, B. Debû, V. Fraix et al.Neurology, April 14, 2008 -
Articles
Comparative effects of unilateral and bilateral subthalamic nucleus deep brain stimulationR. Kumar, A.M. Lozano, E. Sime et al.Neurology, August 01, 1999 -
Articles
Neuropsychological effects of bilateral STN stimulation in Parkinson diseaseA controlled studyH.M.M. Smeding, J. D. Speelman, M. Koning-Haanstra et al.Neurology, June 26, 2006