Cognitive sequelae of subthalamic nucleus deep brain stimulation in Parkinson's disease: a meta-analysis

Summary

Background

Deep brain stimulation of the subthalamic nucleus (STN DBS) is an increasingly common treatment for Parkinson's disease. Qualitative reviews have concluded that diminished verbal fluency is common after STN DBS, but that changes in global cognitive abilities, attention, executive functions, and memory are only inconsistently observed and, when present, often nominal or transient. We did a quantitative meta-analysis to improve understanding of the variability and clinical significance of cognitive dysfunction after STN DBS.

Methods

We searched MedLine, PsycLIT, and ISI Web of Science electronic databases for articles published between 1990 and 2006, and extracted information about number of patients, exclusion criteria, confirmation of target by microelectrode recording, verification of electrode placement via radiographic means, stimulation parameters, assessment time points, assessment measures, whether patients were on levodopa or dopaminomimetics, and summary statistics needed for computation of effect sizes. We used the random-effects meta-analytical model to assess continuous outcomes before and after STN DBS.

Findings

Of 40 neuropsychological studies identified, 28 cohort studies (including 612 patients) were eligible for inclusion in the meta-analysis. After adjusting for heterogeneity of variance in study effect sizes, the random effects meta-analysis revealed significant, albeit small, declines in executive functions and verbal learning and memory. Moderate declines were only reported in semantic (Cohen's d 0·73) and phonemic verbal fluency (0·51). Changes in verbal fluency were not related to patient age, disease duration, stimulation parameters, or change in dopaminomimetic dose after surgery.

Interpretation

STN DBS, in selected patients, seems relatively safe from a cognitive standpoint. However, difficulty in identification of factors underlying changes in verbal fluency draws attention to the need for uniform and detailed reporting of patient selection, demographic, disease, treatment, surgical, stimulation, and clinical outcome parameters.

Introduction

Deep-brain stimulation of the subthalamic nucleus (STN DBS) is an effective treatment for advanced Parkinson's disease.1, 2, 3, 4, 5 The procedure alleviates tremor, rigidity, bradykinesia, and levodopa-induced dyskinesia—the latter probably as a consequence of the reduction in dopaminergic medication after surgery.1, 6, 7 STN DBS can also slow disease progression,⁸ although the procedure's putative neuroprotective effect is uncertain.9, 10

Although the effectiveness of the procedure in the treatment of motor symptoms is accepted, its effect on non-motor symptoms is less clear. The potential dissociability of the motor and neurobehavioural effects is implied by the observation that although health-related quality of life is improved by the procedure,11, 12, 13, 14, 15 these salutary effects can be limited to, or be more evident in, physical aspects of quality of life, such as mobility and bodily discomfort.¹⁶ Changes in quality of life are moderated by changes in depression,¹⁷ which might improve after surgery, occur de novo after deep-brain stimulation, or may be associated with recurrence or exacerbation of a pre-existing condition.¹⁸ Depression could be related to deep-brain stimulation, reduced dopaminergic medication after surgery, or psychosocial factors.19, 20

Increased recognition that Parkinson's disease, traditionally regarded as a movement disorder, is also associated with cognitive, behavioural, and emotional changes21, 22 correlated with fronto-striatal circuit dysfunction,23, 24 has generated interest in the non-motor effects of deep-brain stimulation. Several qualitative reviews of the psychiatric20, 25, 26, 27 and cognitive aspects28, 29, 30, 31, 32 of deep-brain stimulation are an indication of this interest.

Cognitive and mood disturbances seem to be reported more frequently as side-effects of stimulation of the subthalamic nucleus than of the pallidum.³³ This observation could implicate extraneous factors, such as the greater frequency with which the subthalamic nucleus is targeted or greater attention to and awareness of neurobehavioral issues in historically later studies, which tend to assess subthalamic nucleus rather than pallidal stimulation.²⁶ However, direct comparisons between unilateral pallidotomy and bilateral STN DBS34, 35 and bilateral pallidal and STN DBS36, 37, 38 consistently reveal an increased probability of cognitive and behavioural adverse effects after STN DBS. Whether this finding relates to the subthalamic nucleus being a smaller target with motor, associative, and limbic circuits lying in close proximity to each other, and thus there being a heightened risk of electrode misplacement and current spread to non-motor circuits remains speculative.

Empirical details of the effect of STN DBS on cognitive and neurobehavioural functions are only beginning to emerge. In a qualitative review of initial studies of the neuropsychological sequelae of STN DBS and subthalamotomy, Woods and colleagues³² concluded that the most consistently reported findings were reductions in verbal fluency and improvements in self-reported symptoms of depression. Indeed, mild to moderate declines in verbal fluency that are persistent and evident even 3 years after surgery are reported in 30–50% of patients after the procedure.³⁹ Reports of changes in global cognitive functioning, memory, attention, and executive functions are less common and severe cognitive impairments are seen in fewer than 1–2% of patients.³⁷ By contrast, mild to moderate declines of circumscribed scope are reported in 20% of patients.6, 37 A potential difficulty in the interpretation and reconciliation of discordant findings about the nature and extent of neurobehavioural changes ensuing from the procedure is that a host of factors other than stimulation per se can be associated with cognitive alterations, including, for example, selection criteria for and characteristics of patients, surgical experience, operative complications, comorbid disorders, and medication changes. Furthermore, most studies had small sample sizes and might have had inadequate power to identify effects other than very large postsurgical cognitive changes. Specifically, the median sample size of the studies of STN DBS reviewed by Woods and colleagues³² was ten (range one to 63; all single-group, pretest–post-test designs).

In view of the limitations in sample size and the number of inconsistencies in the published work for the extent and duration of possible changes in episodic memory, attention, executive functions, and verbal fluency after STN DBS, Woods and colleagues⁴⁰ assessed the power of studies that looked at cognitive effects of the procedure. Their analysis showed that only two of 30 studies reviewed had adequate power (above 80%) to detect large cognitive effects, and that none had sufficient power to detect cognitive changes associated with conventionally small or medium effect sizes. This noteworthy lack of statistical power and attendant high level of type 2 error risk could adversely affect clinical decision-making by potentially overestimating the neurobehavioural safety of the procedure and thus needs addressing.

Until large-scale studies on the cognitive effects of the procedure are published, statistical meta-analyses provide estimates of a population effect size across independent studies in the interim. Meta-analyses increase statistical power to detect true non-zero population effects by lowering the standard error and consequently narrowing the CIs associated with the population effect size estimate.⁴¹ Hence, a quantitative meta-analysis might enable a better understanding than a qualitative review of the variability and clinical significance of cognitive dysfunction subsequent to STN DBS for Parkinson's disease. We sought to examine the extent of changes related to STN DBS in eight domains of neuropsychological functioning with a meta-analysis.