Deep Brain Stimulation and Cognitive Outcomes Among Patients With Parkinson’s Disease: A Historical Cohort Study

Parkinson’s disease is a neurodegenerative disease characterized by loss of dopaminergic neurons in the substantia nigra and development of Lewy bodies in dopaminergic neurons. Parkinson’s disease manifests with both motor and nonmotor symptoms, including resting tremor, rigidity, bradykinesia, stooped posture, neurobehavioral disorders, cognitive impairment, and autonomic dysfunction (1). Current treatments include pharmacotherapy, nonpharmacological alternatives, and surgery in the form of lesioning, as well as deep brain stimulation (DBS) (1).

DBS involves implantation of an electrode in a subcortical brain structure that provides high-frequency electrical stimulation (2, 3). DBS has proven to be particularly useful in the treatment of moderate to advanced Parkinson’s disease, but only an estimated 10%−20% of patients with advanced disease may be appropriate candidates for DBS (4). Studies have shown that DBS in patients with Parkinson’s disease may be superior to pharmacological treatment (5–7) but may increase the likelihood of decline in neuropsychological function (8). The globus pallidus pars interna (GPi) and subthalamic nucleus (STN) are effective subcortical targets for DBS (3, 4, 9), and their effects on motor and nonmotor outcomes have been discussed in the literature (10, 11). For example, both bilateral GPi DBS and STN DBS are reported to have equal efficacy on Parkinson’s disease motor symptoms (12–14). However, to date, only a few studies have investigated the effect of GPi DBS and STN DBS on cognitive function, yielding conflicting results. Although there was no difference in cognition in some studies (9, 15, 16), others found cognitive and behavioral changes after STN DBS (17–19). One large study including Parkinson’s disease patients from Veteran’s Affairs hospitals and university hospitals reported that patients in the STN group had a greater decline on the Processing Speed Index of the WAIS-III scale (14); another large study found no significant differences between the two groups on composite scores of cognition, mood, and behavioral changes (16). A third study found cognitive changes as a complication but mentioned it as an aspect that needed to be further addressed (13). Hence, the question of cognitive changes or increased risk of worsening in certain domains of cognitive function following DBS remains unresolved.

The aim of this study was to build upon this literature and add to it more specific cognitive data regarding the effect of both GPi and STN DBS on cognitive outcomes. One research group concluded that the increased postoperative delirium and confusion in STN stimulation patients is consistent with some increased vulnerability of the tissues surrounding the STN to DBS surgery (13). It has also been noted that the STN is a compact target, and the region is a critical crossroads for many important functional brain circuits. Therefore, the spread of current in this area is more likely to cause adverse effects (4). For these reasons, we specifically tested two hypotheses: Intragroup change: There is a decrease in cognitive test scores between baseline and 6-month follow-up in the STN DBS group but not the GPi DBS group. Intergroup difference: There is a difference in the change of cognitive test scores from baseline to 6-month follow-up between the two groups.

Methods

Study Participants, Design, and Setting

This historical cohort study included 29 patients with Parkinson’s disease who underwent surgery at the DBS Center of Mayo Clinic, Scottsdale, Arizona, between January 1, 2002, and December 31, 2015; 12 patients underwent bilateral GPi DBS and 17 patients underwent bilateral STN DBS by one neurosurgeon (MKL). The inclusion criteria were a diagnosis of Parkinson’s disease and available neuropsychological testing at baseline and 6 months post-DBS (follow-up). The results from the neuropsychological testing were also used to rule out dementia, psychosis, or treatment-refractory severe depression. The study was approved by the Institutional Review Board at Mayo Clinic.

Neuropsychological Tests

Patients completed a neuropsychological test battery before undergoing DBS and at a mean of 6.7 months post-DBS (range: 3–15 months). All tests were administered by a board-certified neuropsychologist. The test battery was selected for clinical purposes and included four neuropsychological tests or indices to reflect the three cognitive domains of interest in this study: language: Boston Naming Test raw score (20) and WAIS Verbal Comprehension Index (WAIS-VCI; composite score of information, similarities, and vocabulary subtests) (21, 22); attention and concentration: WAIS Working Memory Index (WAIS-WMI; composite score of arithmetic and digit span subtests) (21, 22); and processing speed: WAIS Processing Speed Index (WAIS-PSI; composite score of coding and symbol search subtests) (21, 22). For all tests, a higher score indicates a better cognitive performance. Deidentified data on neuropsychological test scores from each study participant were obtained from electronic medical records. All patients underwent the Boston Naming Test and completed WAIS-III.

Statistical Methods

Because of the small sample size of this study (N=29) and the fact that our data were not normally distributed, we conducted statistical analyses using nonparametric tests. A chi-square test (Fisher’s exact test) was used to compare the sex distribution of each group. Wilcoxon rank-sum tests (Mann-Whitney U test) were used to compare demographic characteristics (i.e., mean age, mean disease duration, mean time between surgery and follow-up test, and baseline scores on all four neuropsychological tests) between GPi DBS and STN DBS patients. We also conducted Wilcoxon rank-sum tests to compare the mean neuropsychological test scores between baseline (before DBS) and follow-up (6 months post-DBS) separately for both STN DBS and GPi DBS groups (intragroup change), as well as to compare cognitive changes between the two groups (i.e., we calculated the difference between baseline and follow-up scores for each test and compared the delta between STN DBS and GPi DBS groups; intergroup difference). The statistical analyses were conducted using the conventional two-tailed alpha level of 0.05 and performed with IBM SPSS Statistics version 22.

Results

There was no significant difference between patients in the GPi DBS group and the STN DBS group in terms of age, sex, disease duration, time between surgery and follow-up tests, or baseline scores on any of the four tests (Table 1). Among GPi DBS patients, there was no significant difference between baseline and follow-up scores on any of the neuropsychological tests (Table 2). STN DBS patients had significantly lower scores at 6-month follow-up as compared with baseline on the WAIS-PSI (mean [SD], 91.47 [10.42] versus 81.65 [12.03]; p=0.03; Table 2). Additionally, we observed that the STN DBS group scored lower on the WAIS-VCI at follow-up compared with baseline (mean delta, –3.41 [SD=5.91]), whereas the GPi DBS group scored higher (mean delta, 4.83 [SD=10.15]), and this difference in delta was significant (p=0.008) (Table 3).

Discussion

In this historical cohort study, we observed that patients with STN DBS performed significantly worse at 6 months post-DBS as compared with baseline in processing speed. This change may be regarded as clinically meaningful; the WAIS-PSI score declined from average at baseline to low average at follow-up. However, only longitudinal follow-up with repeated tests augmented by clinical data can provide a definitive answer regarding clinical significance. In patients who underwent GPi DBS, there was no significant difference between baseline and follow-up score for any of the cognitive tests, and all three WAIS index scores at baseline and follow-up were in the average range. Furthermore, when we compared the change between baseline and follow-up scores between the two groups, we observed a significant difference between the score change in the GPi DBS group and the STN DBS group on the WAIS-VCI. Patients who underwent STN DBS scored on average 4 points lower at 6-month follow-up as compared with baseline, whereas patients who underwent GPi DBS scored on average 4.5 points higher at 6-month follow-up as compared with baseline. In addition, we observed a trend for a greater difference between baseline and follow-up score in the STN DBS as compared with GPi DBS group on the WAIS-WMI and WAIS-PSI, though not significant.

Our findings are in line with a few other studies that also reported a cognitive decline in STN DBS as compared with GPi DBS patients (13, 14, 17–19). For example, a multicenter study group observed a more pronounced decline on the Processing Speed Index of the WAIS-III in patients who underwent STN DBS as compared with GPi DBS. The investigators attributed this decline specifically to group differences on the digit symbol visuomotor subset (14). In another study, investigators reported that cognitive and behavioral changes were more common after STN than GPi implantation (13). A group from the Netherlands reported that STN DBS patients had a more pronounced cognitive decline when compared with controls based on decline in all verbal fluency measures, the Mattis Dementia Rating Scale, delayed recall of the Auditory-Verbal Learning Test, and Stroop Color Card and Color-Word Card tests at both 6- and 12-month follow-up (18). Investigators from the United States observed slightly worse Mattis Dementia Rating Scale scores in STN patients after 6 months and no change in GPi patients (17). Another large multicenter study noted the rate of adverse effects (which mostly consisted of cognitive impairment, psychiatric manifestations, and mood problems) after DBS surgery was higher in the STN group (19). Furthermore, a recently published review concluded that there might be a possible advantage of unilateral or bilateral GPi DBS over STN DBS with regard to cognitive outcomes (4). However, there are studies that reported conflicting findings. For example, researchers from the University of Florida found no differences between the changes in the STN DBS group and the changes in the GPi DBS group in combined letter and semantic verbal fluency at 7-month follow-up, although they noted a larger deterioration of letter verbal fluency task scores in STN DBS participants that did not reach the predefined level of significance (15). Similarly, researchers from the NSTAPS randomized control trial did not observe a difference between composite scores for cognition, mood, and behavioral effects at 12 months when comparing bilateral GPi DBS and STN DBS. This composite score included an assessment of a clinically significant worsening on three or more cognitive tests based on the reliable change index; the loss of professional activity, work, or job; the loss of an important relationship; or psychosis, depression, or anxiety for 3 months or longer as defined by the Mini International Neuropsychiatric Interview (16). In a follow-up analysis, the same group reported only small differences in cognition between GPi DBS and STN DBS at 12 months post-DBS. However, subgroup analyses demonstrated a significantly greater decline on tests measuring mental speed (including Stroop word reading, Stroop color naming, and WAIS similarities) in STN DBS as compared with GPi DBS patients (9). Their subgroup analysis finding is consistent with our finding, as we also observed the most pronounced decline between baseline and follow-up test scores for the STN DBS group on the Processing Speed Index of the WAIS.

The strength of our study is the use of established, validated neuropsychological tests that are widely used in both research and clinical practice. Moreover, the study was conducted at a center with high expertise in DBS. The selected time point of 6 months post-DBS was chosen so as to minimize the influence of cognitive worsening due to disease progression; the medication changes and DBS settings would have been stabilized by that time. The study is limited by the small sample size. However, the comparison between cognitive scores pre- and post-DBS adds more power to the study; even with the small sample size we were able to detect some statistically significant differences and marginally significant trends. These tests were not administered in medication off/DBS on state or vice versa but done in a practically defined on state (with their usual Parkinson’s disease medications and DBS settings), as it is more reflective of a real world day-to-day experience. This is a growing trend in the current DBS literature. Another potential shortcoming is that we did not assess for medication effects. Although the majority of patients in both groups were able to reduce the dose or number of Parkinson’s disease medications, other medications could in theory influence their cognitive function as well. However, there is no reason to believe that one group (STN or GPi) would be on significantly more medications than the other group. In addition, we did not investigate underlying mechanisms of the association between STN DBS and more pronounced cognitive decline. It has been hypothesized that STN is a critical crossroads for important functional brain circuits, and therefore the spread of current is more likely to result in higher adverse effects (4, 23). Finally, we are also unable to comment on the influence of lead location in the STN (given that STN is a smaller target), as we did not perform postoperative imaging. MR imaging post-DBS is performed if clinically indicated (e.g., inadequate response to DBS or unexpected side effects to DBS programming). As this was not the case, MR imaging post-DBS was not performed routinely.

In conclusion, we observed that patients with STN DBS performed significantly worse on follow-up on a test measuring processing speed. In addition, when compared with GPi DBS patients, STN DBS patients had lower mean scores (although not significant) on cognitive tests measuring language, attention and concentration, and processing speed at 6 months post-DBS compared with baseline. Our findings add to the growing body of evidence showing that the preferred target for DBS in patients with Parkinson’s disease might be GPi rather than STN, as this approach may be more likely to minimize potential negative effects on cognition. Clinicians may want to consider using this information to make an informed decision about which form of DBS is best for patients with Parkinson’s disease. Future studies in a larger sample, preferably with prospective cohort study design, are needed to confirm our findings.