Deep brain stimulation of the subthalamic nucleus improves cognitive flexibility but impairs response inhibition in Parkinson disease

(1)

Deep Brain Stimulation of the Subthalamic Nucleus Improves Cognitive Flexibility but Impairs Response Inhibition in Parkinson Disease

Karsten Witt, MD; Ulrich Pulkowski, MD; Jan Herzog, MD; Delia Lorenz, MD;

Wolfgang Hamel, MD; Gu¨nther Deuschl, MD; Paul Krack, MD

Background:Deep brain stimulation of the subthalamic nucleus (STN) improves motor symptoms of Par- kinson disease. Although several studies have assessed cognitive functions before surgery and after long-term STN stimulation, only a few have assessed patients while stimulation is on and off to more specifically address the short- term cognitive effects of STN deep brain stimulation.

Objective:To examine the short-term effects of STN stimulation on several tests sensitive to executive function and the long-term effects of STN stimulation on a global cognitive scale.

Design:Twenty-three patients with Parkinson disease were tested 6 to 12 months after surgery with STN stimulation switched on and off in a random order while taking their regular medication. The Unified Parkinson’s Dis- ease Rating Scale motor score was also rated in the on and off stimulation condition. The neuropsychological battery included digit span, verbal fluency, Stroop color

test, and random number generation in a single- and dual- task condition.

Results:Short-term stimulation improved the results on the Random Number Generation Task, requiring suppression of habitual responses, but induced more errors in the interference task of the Stroop color test. Digit span, verbal fluency, and dual-task performance results did not change. There was a significant correlation (r=0.47,P=.02) between improved performance on the Random Number Generation Task and impaired response inhibition in the Stroop interference condition. A preoperative to postoperative comparison showed no changes in global cognitive function with long-term STN deep brain stimulation.

Conclusions:Short-term STN stimulation improves cognitive flexibility (giving up habitual responses) but impairs response inhibition. Long-term STN stimulation does not change global cognitive function.

Arch Neurol. 2004;61:697-700

L

^ONG^-TERM DEEP BRAIN STIMU- lation (DBS) of the subthalamic nucleus (STN) is an effective treatment of motor symptoms in patients with advanced Parkinson disease (PD). Spe- cific effects on the cognitive domain are still debatable. When comparing cognitive function before and after surgery, several variables can interfere: (1) the surgi- cal procedure, (2) the reduction in dopaminergic medication, (3) DBS itself, and (4) postoperative mood changes. The influence of these factors could be avoided by testing patients with stimulation on and off. Until now, only 2 studies examined PD patients with stimulation on and off. Pil- lon et al¹showed that STN DBS improves the speed in parts A and B of the Trail- Making Test, the word and color condition of the Stroop color test, the reaction time performance, and subscores of tests sensitive for spatial working memory. Ja- hanshahi et al²found, in a small series of

patients, that STN DBS improves cognitive flexibility, attention, and working memory but worsens conditional associative learning and performance on the interference condition of the Stroop color test.

The present study more specifically investigates the effects of short-term STN stimulation on executive function in a larger series of patients and documents the effects of long-term STN stimulation on more global cognitive functioning.

METHODS PATIENTS

Twenty-three PD patients (6 women and 17 men), with a mean ± SD age of 57.4 ± 5.8 years, were examined 6 to 12 months (mean ± SD, 8.8 ± 1.6 months) after bilateral electrode implantation in the STN for DBS. All patients had advanced PD, with a mean ± SD disease dura- tion of 15.1 ± 5.5 years.

ORIGINAL CONTRIBUTION

From the Departments of Neurology (Drs Witt, Pulkowski, Herzog, Lorenz, Deuschl, and Krack) and Neurosurgery (Dr Hamel), Christian-Albrechts-University Kiel, Kiel, Germany.

(REPRINTED) ARCH NEUROL / VOL 61, MAY 2004 WWW.ARCHNEUROL.COM 697

Downloaded From: http://archneur.jamanetwork.com/pdfaccess.ashx?url=/data/journals/neur/13744/ by a Universite de Geneve User on 04/27/2017

(2)

Dementia (Mattis³Dementia Rating Scale score,⬍130) was an exclusion criterion. Quadripolar electrodes (Medtronic, Min- neapolis, Minn) were implanted using stereotactic magnetic resonance imaging and intraoperative electrophysiology, as previ- ously described.⁴The patients were retested 12 months later using the Mattis Dementia Rating Scale. At testing, the stimulation characteristics were as follows: mean ± SD pulse width, 63 ± 13 µs; mean ± SD frequency, 139 ± 15 Hz; and mean ± SD stimulation voltage, 3.2 ± 0.4 V. All electrodes were monopo- lar, using 1 electrode contact. Postoperatively, patients re- ceived a mean levodopa equivalence dosage of 746 mg. Of the tested patients, 19 (82.6%) were taking a dopaminergic agonists. Five patients were taking only dopaminergic agonists (3, cabergoline; and 2, pergolide mesylate), and 4 were taking levodopa only. Fourteen patients were taking a combination of levodopa and dopamine agonists (9, cabergoline; and 5, pergolide). All patients gave their informed consent. The pro- tocol was approved by the local ethical committee of Christian- Albrechts-University Kiel.

TEST PROCEDURE

Patients were taking their usual antiparkinsonian medication while stimulation was on and off. The conditions of stimulation were set 30 minutes before beginning the neurological and neuropsychological examination. The conditions of stimulation were counterbalanced in a pseudorandom order. The motor score of the Unified Parkinson’s Disease Rating Scale (UPDRS) (part III) was evaluated with stimulation on and off.

Digit span forward and backward was carried out accord- ing to the administration of the Wechsler Adult Intelligence Scale.⁵ Digit span scores forward and backward were analyzed separately.

Verbal fluency⁶was tested using the categories female or male, first name, and animals or plants in parallel forms for 1 minute each. The order of categories and consonants was pseu- dorandomized. For literal fluency, words beginning with “K”

and “N” or “L” and “M” were taken. Each fluency task was scored separately. A shortened Stroop color test⁷contained 4 trials:

(1) reading words (blue, yellow, green, and red) printed in black ink, (2) reading color dots for simple color naming, (3) interference condition reading words (blue, yellow, green, and red) printed in ink of different colors, and (4) interference condition naming the color of the ink of the written words (blue, yellow, green, and red). For each trial, a test containing 36 items was used. The number of errors and the time to finish the test was scored for each trial separately.

For the Random Number Generation Task (RNGT), subjects were instructed to generate series of 100 numbers of the digits 1 to 10 in a random fashion paced by a tone (1 Hz). The concept of randomness was explained using standard proce- dures (ie, using instruction based on an analogy of selecting and replacing numbered table tennis balls from a shoe carton).

An example trial was undertaken. If the subject had not un- derstood the instructions, several simple examples of randomness were given and the examiners (K.W. and U.P.) and the subject executed a test run. The total time taken to generate 100 items was noted. In a dual-task condition, patients gener- ated digits randomly and simultaneously sorted a stack of 100 mixed blue and red cards. The digits were recorded and analyzed by different random measurements; the Evans⁸Random Number Generation Index is a first-order measure of randomness. This index varies between 0 and 1; the higher the index, the less random the series. Finally, we calculated the counting scores. These indices measure the tendency to count in ascend- ing or descending series.⁹Count score 1 (CS1) and count score 2 (CS2) measure the tendency to count in steps of 1 and 2, re- spectively. The CS1 was increased by one²if one counting step (eg, 2-3 or 7-6) was observed. The sequence length is squared

to give higher weights to runs of longer sequences. The total count score is the sum of CS1 and CS2 and reflects all counting tendencies in steps of 1 and 2. Subjects with a higher counting score are unable to suppress habitual counting tendencies.

STATISTICAL ANALYSIS

A nonparametric analysis (Wilcoxon signed rank test) was performed on each of the variables comparing the stimulation on and stimulation off conditions. Furthermore, a correlation within neuropsychological changes (changes in digit span, verbal fluency, Stroop color test performance, and different measurements of the RNGT) and between neuropsychological and motor changes (UPDRS score) was performed (Spearman rank correlation analysis).

RESULTS

The mean ± SD UPDRS motor score while taking usual medication was 18.8 ± 9.4, and it decreased to 11.5 ± 7.4 after turning on the stimulators (z= 4.20,P= .01).

Table 1displays the results of the neuropsychological testing. There were no significant changes between the on and off stimulation condition for the digit span and verbal fluency scores.

After turning the stimulators on, patients made more errors in the interference condition of the Stroop color test. The speed of this test was not altered significantly in any condition.

In the single-task condition of the RNGT, the CS1 and the total count score decreased significantly after turning on the stimulators. The time to generate 100 items was similar. In the dual-task condition of the RNGT, the CS1 and the CS2 improved with stimulation, but these changes were not significant. To analyze the influence of STN DBS on dual-task performance, we compared the difference of single- and dual-task performance of the CS1 and CS2 with and without DBS. There was no significant influence of the stimulation on dual-task performance.

The correlation analysis (Spearman rank correlation) between the positive change on the CS1 and the worsening in the Stroop color test (error rate) in the interference conditions showed a significant positive correlation (r=0.47,P=.02). The correlation analysis (Spearman rank correlation) between the positive change on the total count score and the worsening in the Stroop color test (error rate) in the interference conditions showed a significant positive correlation too (r=0.59,P=.01). The more errors in the interference condition of the Stroop color test were made after turning on the stimulators, the more patients gave up their counting tendencies in the RNGT. There were no further significant correlations between neurological (UPDRS score) and neuropsychological changes.

To assess global changes of cognitive functioning, the Mattis Dementia Rating Scale was repeated 12 months after electrode implantation. The results are shown in Table 2. No significant changes concerning the subitems of the Mattis Dementia Rating Scale were observed.

COMMENT

Subthalamic nucleus stimulation led to mild improvement in the UPDRS motor score and to some minor

(3)

changes in cognitive abilities while patients took their usual medication. Therefore, DBS was effective in the motor and cognitive domains. This study demonstrates a significant improvement of counting scores in the RNGT after turning on the stimulation. On the other hand, patients made more errors in the interference condition of the Stroop color test in the stimulation on condition. A further analysis of these changes caused by the STN stimulation showed a correlation between improved RNGT performance and impaired response inhibition in the Stroop interference condition. Concerning the effects of STN DBS on global cognitive functioning after 12 months, we found no significant changes. These results are in line with those of other studies^1,10examining relatively young PD patients after STN stimulation.

In everyday life, we are used to counting in ascend- ing or descending series, mostly in steps of 1 or 2, but we never produce series of digits in a random fashion.

To produce a perfect random series of digits, one has to give up the habit of counting. Higher counting rates were shown by different patient groups with a dysexecutive syndrome (patients with frontal lesions⁹and patients with PD¹¹). Recently, an imaging study was performed, dem- onstrating the human cortical network subserving random number generation. Generating numbers randomly activates cortical areas involved in executive functioning (the dorsolateral prefrontal cortex, the an- terior cingulate cortex, and the biparietal and cerebellar cortex).¹²In summary, (1) theoretical consideration about the task, (2) human lesion studies, and (3) data from functional imaging give evidence that executive control is necessary for generating digits randomly. Different variables describe the cognitive abilities necessary for successful random generation of numbers. The Evans Ran- dom Number Generation Index reflects monitoring capacities⁹that did not change significantly under STN Table 1. Results of the Neuropsychological Battery With Stimulation Off and On

Variable

Stimulation Off*

Stimulation On*

z Score

2-Tailed PValue Digit span

Forward 7.69 ± 2.12 7.61 ± 2.50 0.05 .96

Backward 4.91 ± 1.76 5.04 ± 1.77 0.69 .49

Verbal fluency Category

First run 9.43 ± 3.54 9.78 ± 4.13 0.28 .78

Second run 9.13 ± 4.43 8.70 ± 4.24 1.74 .08

Literal

First run 17.17 ± 5.02 15.65 ± 5.22 0.30 .78

Second run 15.74 ± 6.61 16.04 ± 6.71 0.51 .54

Stroop color test

Words printed in black ink

No. of errors 0.04 ± 0.21 0.00 ± 0.00 1.00 .32

Time, s 16.52 ± 2.69 16.57 ± 3.76 0.51 .61

Color dots for color naming

No. of errors 0.04 ± 0.21 0.22 ± 0.42 1.63 .10

Time, s 26.00 ± 6.42 25.07 ± 5.57 0.91 .36

Interference condition Reading words

No. of errors 0.43 ± 0.90 0.47 ± 1.16 0.30 .76

Time, s 24.15 ± 10.36 21.72 ± 9.40 1.90 .06

Naming colors

No. of errors 0.78 ± 1.44 1.74 ± 1.66 2.84 .004

Time, s 55.17 ± 15.60 58.91 ± 15.86 1.72 .08

Random Number Generation Task Single task

Random Number Generation Index 0.37 ± 0.02 0.36 ± 0.01 0.15 .88

Null score 48.04 ± 7.78 48.55 ± 6.10 0.77 .44

Counting score

1 78.04 ± 41.32 50.74 ± 28.23 2.86 .004

2 37.35 ± 44.10 29.30 ± 12.47 0.17 .99

Total 115.39 ± 51.39 80.04 ± 34.65 2.81 .005

Time, s 136.45 ± 33.91 135.90 ± 41.05 0.06 .95

Dual task

Random Number Generation Index 0.41 ± 0.03 0.38 ± 0.02 1.20 .23

Null score 51.14 ± 8.18 49.90 ± 6.63 0.69 .49

Counting score

1 90.83 ± 58.92 78.96 ± 47.68 0.67 .50

2 33.04 ± 21.72 26.09 ± 14.15 1.46 .14

Total 123.87 ± 68.16 105.05 ± 52.03 1.87 .06

Time, s 188.66 ± 58.17 191.84 ± 84.39 1.12 .26

*Data are given as mean ± SD.

(4)

stimulation in our study. Digit span forward and re- verse assess working memory capacities too. These data showed no significant change with STN stimulation. The RNGT requires suppression of natural preferences for counting in series. This feature of the RNGT is a classic executive function that can be modified by STN stimulation. We confirm the findings of Jahanshahi et al,²who demonstrated decreased (improved) counting tendencies after switching on STN stimulators in a group of 7 PD patients. The counting scores are sensitive to the rate of production,⁹but in our study, the rate of production did not vary between the 2 stimulation conditions. Thus, the changes in the counting scores were not the result of an altered production rate.

With stimulation on, patients made significantly more errors in the interference condition of the Stroop color test, whereas the speed was not altered. This find- ing is in line with that of a previous study²but not with the findings of other studies.^1,10

Our patients did not spontaneously complain of symptoms comparable with premature response in ac- tivities of everyday living. These results are in line with those of Saint-Cyr et al.¹³In their study, a “disinhibition” 6 months after surgery was reported by caregivers, but not by patients, whereas 12 months after surgery, the caregivers noted no more disinhibition but an increase in apathy. This disinhibited behavior may only be a short- lasting effect, analogous to the dyskinesia induced after STN surgery, that also tends to disappear with time after an STN lesion or while undergoing long-term STN stimulation.⁴This deficit can be demonstrated with sensitive instruments like the Stroop color test but gener- ally does not impair the patients in everyday life. There was no significant correlation between motor improvement and neuropsychological changes due to STN DBS.

Regarding this lack of correlation between neuropsychological and motor data in our study, one major point that must be considered is the ongoing medication of our patients. Of the tested patients, 82.6% (19/23) took agonists, mostly cabergoline, a long-acting agonist with a half- life of 60 hours. A medication withdrawal overnight would not exclude dopaminergic influence, and, therefore, we assessed the patients while they were taking their usual medication. The ongoing medication may mask the full effect of the STN DBS, which might alter the motor rat-

ing and further correlation analysis. Furthermore, the correlation analysis showed a significant correlation between a worsening in Stroop color test performance (error rate in the interference condition) and an improvement in counting scores of the RNGT due to STN stimulation. These effects may be induced by stimulation of the associative territory of the STN.

In our study, STN DBS did not improve a dual cognitive task. Simultaneous cognitive task performance is a dopamine-related function.¹⁴Because our patients were undergoing dopaminergic treatment, our results are still preliminary. This issue should be investigated in further studies using motor and nonmotor tasks simultaneously with more concurrent attention-demanding tasks.

Accepted for publication December 29, 2003.

Author contributions: Study concept and design(Drs Witt and Deuschl); acquisition of data (Drs Witt, Pulkowski, Herzog, Lorenz, Hamel, and Krack);analy- sis and interpretation of data(Drs Witt and Krack);draft- ing of the manuscript(Drs Witt, Herzog, Lorenz, and Krack);critical revision of the manuscript for important in- tellectual content(Drs Pulkowski, Hamel, and Deuschl);

statistical expertise(Drs Witt and Pulkowski);obtained funding(Drs Herzog, Lorenz, and Krack);administra- tive, technical, and material support(Drs Pulkowski, Her- zog, Lorenz, Hamel, and Krack);study supervision(Drs Deuschl and Krack).

Corresponding author: Gu¨nther Deuschl, MD, Depart- ment of Neurology, Christian-Albrechts-Universita¨t Kiel, Nie- mannsweg 147, D-24105 Kiel, Germany (e-mail:

g.deuschl@neurologie.uni-kiel.de).

REFERENCES

1. 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.

2. Jahanshahi M, Ardouin CM, Brown RG, et al. The impact of deep brain stimulation on executive function in Parkinson’s disease.Brain.2000;123:1142-1154.

3. Mattis S.Dementia Rating Scale.Odessa, Fla: Psychological Assessment Re- sources Inc; 1988.

4. Wenzelburger R, Koper F, Zhang BR, et al. Subthalamic nucleus stimulation for Parkinson’s disease preferentially improves akinesia of proximal arm movements compared to finger movements.Mov Disord.2003;18:1162-1169.

5. Wechsler D.WAIS-R Manual.San Antonio, Tex: Psychological Corp; 1987.

6. Benton AL. Differential behavioral effects in frontal lobe disease.Neuropsycho- logia.1968;6:53-60.

7. Perret E. The left frontal lobe of man and the suppression of habitual responses in verbal categorical behaviour.Neuropsychologia.1974;12:323-330.

8. Evans F. Monitoring attention deployment by random number generation: an index to measure subjective randomness.Bull Psychonom Soc.1987;12:35-38.

9. Spatt J, Goldenberg G. Components of random generation by normal subjects and patients with dysexecutive syndrome.Brain Cogn.1993;23:231-242.

10. Ardouin C, Pillon B, Peiffer E, et al. Bilateral subthalamic or pallidal stimulation for Parkinson’s disease affects neither memory nor executive functions: a con- secutive series of 62 patients.Ann Neurol.1999;46:217-223.

11. Brown RG, Soliveri P, Jahanshahi M. Executive processes in Parkinson’s disease: random number generation and response suppression.Neuropsycholo- gia.1998;36:1355-1362.

12. Daniels C, Witt K, Wolff S, Jansen O, Deuschl G. Rate dependency of the human cortical network subserving executive functions during generation of random number series: a functional magnetic resonance imaging study.Neurosci Lett.2003;

345:25-28.

13. Saint-Cyr JA, Trepanier LL, Kumar R, Lozano AM, Lang AE. Neuropsychological consequences of chronic bilateral stimulation of the subthalamic nucleus in Par- kinson’s disease.Brain.2000;123:2091-2108.

14. Malapani C, Pillon B, Dubois B, Agid Y. Impaired simultaneous cognitive task performance in Parkinson’s disease: a dopamine-related dysfunction.Neurol- ogy.1994;44:319-326.

Table 2. Neuropsychological Evaluation Results Before and 12 Months After STN DBS

Mattis Dementia Rating Scale Component

Before STN DBS*

After STN DBS*†

2-Tailed PValue

Total score 136.2 ± 2.9 134.7 ± 5.5 .25

Attention 35.7 ± 0.9 35.0 ± 2.1 .13

Initiation 34.7 ± 2.4 34.0 ± 2.9 .35

Construction 6.0 ± 0.0 5.9 ± 0.2 .32

Conceptualization 36.9 ± 1.6 37.1 ± 1.3 .55

Memory 22.9 ± 1.2 22.7 ± 2.5 .82

Abbreviations: DBS, deep brain stimulation; STN, subthalamic nucleus.

*Data are given as mean ± SD.

†Stimulation on.