Executive Function and Dopamine Response in Parkinson’s disease Freezing of Gait

Abstract

Background:

This investigation examined whether aspects of attention and executive functioning differed between Parkinson’s Disease (PD) patients with freezing of gait (FOG) based on responsiveness to dopamine. We also explored association of cognition with FOG severity and gait metrics.

Methods:

Fifty-four individuals with PD completed the study protocol: 17 without freezing (PDC), 23 with dopa-responsive FOG (RFOG), and 14 with dopa-unresponsive (URFOG). Standardized neuropsychological tests assessed attention (focused and sustained), psychomotor speed, and set-switching (time and errors). FOG severity was measured using the new FOG Questionnaire (nFOG-Q). Metrics from timed up and go (TUG) tasks were obtained while “on” and “off” dopamine, with and without dual cognitive tasks.

Results:

After controlling for clinical and demographic factors, analysis of covariance revealed a significant between-group difference for set-switching errors; planned contrasts revealed increased set-switching errors in URFOG relative to RFOG and PD control groups. Groups were not different in other cognitive domains. FOG severity was modestly associated with set-switching errors in RFOG but not URFOG. TUG performances while “on” were associated with set-switching errors in PD controls, and with focused attention in RFOG.

Conclusion:

PD patients with dopa-unresponsive FOG are more prone to set-switching errors than those who respond to treatment. Furthermore, executive function appears relevant to FOG severity only in patients who show dopamine response. Together, these findings suggest disruption of a common dopamine-mediated pathway for FOG and ability to monitor rules while alternating cognitive processes. Consideration of dopa-response could be useful in characterizing cohorts and treating FOG in PD.

Introduction

Freezing of gait (FOG) is a debilitating condition, experienced by the majority of patients with PD^1,2,3. It is described as the inability to produce effective stepping in spite of the intention to do so⁴. The pathophysiology of FOG is poorly understood. For some PwP, FOG is responsive to dopamine replacement therapy and occurs only during “off” fluctuations; for others, FOG does not respond to dopamine and can manifest at any time. This distinction has led to characterization of pharmacologic FOG subtypes according to dopamine response⁵. A study by Factor et al found more severe executive dysfunction when FOG was not dopa-responsive⁶. Group differences were particualry notable for set-switching ability, as assessed Trails B, a timed test of psychomotor sequencing requiring altneration between number and letter targets in ascending order. Completion time on Trails B has also been shown to be related to subjective⁷ and objective measures of FOG severity⁸. Though not examined with respect to dopamine response, this association may vary given the well-established role dopamine also plays in executive function in PD⁹ While these findings suggest specific relevance of set-switching to FOG, a more recent study found that differences between PD groups with and without FOG on Trails B were no longer statistically significant (p=.054) after controlling for disease severity¹⁰; however, unlike the prior studies where participants were tested at their subjective “on” state, in this study cognitive testing was performed while “off” medication. This design is useful in examining the effect of dopamine removal on cognition in treated PD, but performance is more likely to be confounded by motor impairment, particularly for tasks like Trails B that require motoric responses. Additionally, FOG groups were not examined separately according to dopamine response.

Taken together, prior research suggests that consideration of dopamine response may be critical in examining associations between executive functioning and FOG. Characterization of this relationship with respect to dopamine response could prove useful in clarifying the pathophysiology of FOG and direct appropriate interventions. Thus, the objective of the current study was to further elucidate previously reported findings between executive functioning and FOG phenomenon in PD. Specifically, we examined whether dopamine responsiveness in FOG was associated with set-switching and related domains of attention and psychomotor speed that contribute to task performance. We also examined whether associations between cognitive functioning and subjective FOG severity ratings differed according to dopamine response. Finally, associations between cognition and objective gait measures were explored.

Methods

Participants:

All patients included in this study were recruited from the Movement Disorders clinic at the Medical University of South Carolina and gave informed written consent. The study was conducted with approval from the institutional review board. All participants met UK Brain Bank diagnostic criteria for PD. Exclusion criteria included deep brain stimulation implantation, a score of <26 on the mini-mental status examination at screening and gait dysfunction other than from PD. Each subject was characterized as PD-control (no FOG), dopa-responsive FOG, or dopa-unresponsive FOG based on their response to item 1 from the new freezing of gait questionnaire (nFOG-Q)¹¹ to confirm FOG, and item 14 from Part II of the Unified Parkinson’s Disease Rating Scale (UPDRS) administered in the ON and OFF dopaminergic medication state, and confirmed by motor assessment by a movement disorders neurologist, in order to determine dopa-response. Dopa-response was defined as an improvement of more than one point on item 14 of the UPDRS, part II. Final FOG status was determined by agreement between subject and neurologist. Five subjects screen failed: two had MMSE <26, two had unclear freezing, and one was too severe to complete laboratory assessments. Three subjects withdrew due to distance or time limitations to complete the study. Three subjects were removed from analysis: two were later found to have developed an atypical parkinsonian syndrome and one (in control group) was found to have freezing. Fifty-four (54) individuals completed the study (17 without freezing, 23 with dopa-responsive FOG, and 14 with dopa-unresponsive FOG).

Behavioral Assessments:

Behavioral assessments were collected and included the new nFOG-Q, and the UPDRS parts 1-4 in the ON and OFF state as a measure of PD motor severity. OFF assessments were made at least 12 hours OFF all dopaminergic medication, ON assessments were made at least 30 minutes after taking the first dose of medications on the same day. PD-controls only had ON assessments as they did not have motor fluctuations. Subjects completed two timed up and go (TUG) trials on the GaitRite digital walkway (CIR Systems, Franklin NY) in the OFF state, which were averaged to yield the single task TUG time variable. Specifically, they were asked to stand up, walk over GaitRite mat, step off the GaitRite onto the M² walkway, turn around a cone set at the center of the M² (54 inches to the center of the cone from the leading edge of the M2/GaitRite interface), and walk back to the chair. The instructions for the walking task were identical to what is commonplace during the TUG¹². This was followed by two more trials in the OFF state with a dual task (serial 7s and every other letter of the alphabet), which were averaged to yield the TUG time dual task variable. The assessment was repeated in the ON state. Turn duration was calculated as the time when subjects stepped off of the GaitRite mat onto the M² mat, where they turned around a cone placed at a predetermined distance at the center of the mat to standardize the distance of the turn¹³.

Participants completed a battery of standardized neuropsychological tests in a single session during the subjective “on” state. The test battery included subtests from the Wechsler Memory Scale, fourth edition (WMS-IV)¹⁴, Neuropsychological Assessment Battery (NAB)¹⁵, and Delis-Kaplan Executive Function System (D-KEFS)¹⁶. Domains included the following measures:

• Focused Attention: WMS-IV Digit Span; NAB Numbers & Letters B Efficiency, NAB Numbers & Letters C Efficiency

• Sustained Attention: NAB Numbers & Letters A Errors, NAB Numbers & Letters A Time

• Psychomotor Speed: D-KEFS Trail Making Numbers, D-KEFS Trail Making Letters, D-KEFS Color Naming, D-KEFS Word-Reading

• Set-Switching Time: D-KEFS Trail Making Number-Letter Switching, D-KEFS Color-Word Interference Switching

• Set-Switching Errors: D-KEFS Trail Making Number-Letter Switching, D-KEFS Color-Word Interference Switching

Composite scores for each domain were derived using the average rank of raw scores (lower is better). Gender distribution across groups was assessed via Chi-square; other demographic and clinical characteristics were examined using one-way ANOVAs. The Quade approach to analysis of covariance (ANCOVA) of ranks¹⁷ was used to compare groups on composite measures while controlling for demographic and clinical factors. All statistical analyses were performed using SPSS^® version 25.

Results

Sample Characteristics:

As shown in Table 1, groups were found to vary on a number of relevant clinical and demographic factors. Compared to PD controls, the dopa-responsive group had lower education, t(52)=3.74, p<.0005, longer disease duration t(52)=2.66, p=.01, higher UDPRS Part III scores, t(52)=2.72, p=.009, and higher Levodopa Equivalent Daily Dose (LEDD), t(52)=4.68, p<.0005. The dopa-unresponsive group also had higher UPDRS Part III scores t(52)=2.05, p=.045, and greater LEDD than the PD controls t(52)=3.50, p=.001. The dopa-unresponsive group was older than the dopa-responsive group, t(52)=2.56, p=.013. Global cognitive functioning (MMSE) was not significantly different between groups. Gender composition was not significantly different between groups, X²=.071, p=.965. FOG severity was also not significantly different between the dopa-responsive and unresponsive groups.

Table 1.

Clinical and demographic characteristics by group, mean (SD)

Domain	PD Controls (n=17)	Dopa- responsive FOG (n=23)	Unresponsive FOG (n=14)
Age	68.18 (5.79)	64.87 (6.28)*	70.47 (7.37)**
Education	17.82 (2.51)	14.57 (2.17)	16.73 (3.62)
Gender (male)	13	16	12
Disease Duration	5.76 (3.62)	9.78 (4.62)*	7.33 (5.85)*
LEDD (mg)	517 (333)	1298 (601)*	1165 (564)*
UPDRS Part-III Total	19.24 (6.76)	26.04 (7.66)*	24.93 (8.26)
MMSE Total	29.00 (1.17)	28.59 (1.47)	28.14 (1.03)

^*Significantly different from PD Controls, p<.05

^**Statistically different from Dopa-responsive FOG

Primary Analyses:

To determine whether groups differed in cognitive domains, ANCOVA of ranks was performed controlling for all clinical and demographic factors in which differences were observed between groups (i.e., age, education, disease duration, UPDRS Part III ON scores, and LEDD). Linear regression of covariates on cognitive domain ranks was first used to derive standardized residuals; these residuals were then evaluated as adjusted ranks in one-way ANOVAs. As shown in Table 2, a statistically significant group difference was observed for set-switching errors, F(2,52)=3.764, p=.03. Planned contrasts revealed that this was driven by worse set-switching error ranks in the dopa unresponsive group relative to the PD control group, t(52)=2.67, p=.01, and the dopa responsive group t(52)=2.03, p=.048; the PD control and dopa responsive groups were not significantly different. No statistically significant group differences or trends were observed in the other cognitive domains.

Table 2.

Covariate-adjusted ranks by group (higher values indicate better function)

Domain	PD Controls (n=17)	Dopa-Responsive FOG (n=23)	Unresponsive FOG (n=14)	p-value
Focused Attention	2.22 (8.35)	.67 (11.28)	−3.56 (13.01)	ns
Sustained Attention	−.95 (10.73)	.51 (9.56)	.30 (12.27)	ns
Speed	−2.41 (10.58)	1.00 (12.27)	1.19 (12.97)	ns
Set-Switching Time	−1.28 (14.28)	−1.02 (12.33)	3.01 (12.49)	ns
Set-Switching Errors	4.00 (11.73)	1.06 (10.39)	−6.16 (10.04)	p=.03

Pearson product-moment correlations examined relationships between adjusted cognitive domain ranks and total score from the nFOG-Q. The only cognitive domain associated with nFOG-Q was set-switching errors, r=−.411, p=.01. When examined separately according to dopamine response, association between nFOG-Q and set-switching errors was statistically significant for the dopa-responsive group, r=.−447, p=.033, but not the dopa-unresponsive group, r=.−095, p=.735. Group differences in this relationship are illustrated in Figure 1. No other statistically significant associations between nFOG-Q and cognitive domains emerged when response groups were examined separately.

Figure 1.

Associations between covariate-adjusted set-switching error rank (higher indicate better functioning) and freezing of gait severity (higher scores indicate greater severity) according to dopamine response.

Secondary Analyses:

Several of the gait metrics were found to be strongly skewed and not normally distributed. Group differences on gait metrics were therefore examined using non-parametric Mann-Whitney U-Tests (see table 3). Across FOG groups, the PD Control group performed the TUG tasks faster while “on” medication. The RFOG evinced better performance for TUG Time-to-Turn and TUG Dual-Tasking measures than the URFOG group while “on” medication, but not during the “off” examination.

Table 3.

Gait measures PD patients with in dopa-responsive (RFOG) and unresponsive (URFOG) freezing.

Measure	RFOG	URFOG	Group Difference
nFOGQ	20.12 (5.12)	23.06 (3.44)	p.048
On TUG	24.93 (8.56)	27.81 (7.29)	p=.169
On Turn	5.67 (3.13)	6.71 (2.65)	p=.048
On Dual TUG	30.68 (14.21)	83.88 (173.61)	p=.036
On Dual Turn	7.02 (4.80)	20.42 (40.00)	p=.011
Off TUG	49.40 (37.93)	30.75 (9.11)	ns
Off Turn	12.66 (11.43)	7.45 (3.86)	ns
Off Dual TUG	92.91 (107.09)	78.50 (144.71)	ns
Off Dual Turn	21.74 (22.85)	26.10 (60.36)	ns

Spearman-Brown correlation analysis examined relationships between adjusted cognitive domain ranks and gait metrics separately by group, with lower values indicating better performance for cognition and gait. Increased set-switching errors in PD-Controls were associated with increased TUG Time to Turn, rho=.515, p=.035, and overall TUG while dual tasking, rho=.522, p=.032. In the ON state, worse focused attention in the RFOG group was associated with increased TUG Time to Turn while dual-tasking, rho=.512, p=.012. There were no statistically significant associations between cognitive and gait measures in the URFOG group, while “on” medication, or in either group during the “off” examination.

Discussion

Findings from this study highlight the relevance of executive functions, as measured by set-switching, to FOG behavior in PD. Our findings confirm previous reports of a significant association between FOG and executive dysfunction, with dopa-unresponsive FOG evincing greater cognitive impairment⁶. We expand this work by showing a significant association between executive dysfunction and FOG severity in the dopa-responsive group and relating cognition to relevant gait metrics. In prior studies set-switching was measured only with respect to completion time for a test of psychomotor sequencing (i.e., Trails B). In this study we found that set-switching errors, but not completion time, during two different set-switching tests (D-KEFS Trail Making and Color-Word Interference) were associated with FOG severity. Set-switching errors were also found to be related to objective gait measures in the PD control group, but not the FOG groups, although in the RFOG subgroup there was an association between focused attention and time-to-turn while dual-tasking during the “on” state.

In the dopa-responsive group, as executive function worsened, FOG worsened, implying that executive function was protective in this group. Alternatively, FOG may emerge in RFOG as executive functions reach a threshold of dysfunction. Notably, this pattern was not observed in the dopa-unresponsive group. Given both groups did not differ significantly in baseline cognitive (MMSE) and motor (UPRDS III) performance or duration of disease, our findings suggest divergent neuropathologic disruptions between these subtypes of FOG.

In the 2014 study by Factor et al.⁶, set-switching, as measured by the additional time burden on Trail Making part B (switching) relative to Part A (sequencing) was also found to be worse in dopa unresponsive FOG compared to dopa responsive FOG and controls. Frequency for errors during this task was not reported. In our sample, the composite for errors but not completion time for set-switching tasks was different between groups. These findings are not considered discrepant, however, as different forms of the Trail Making Test were used (i.e., D-KEFS versus Halstead-Reitan) and our composite also included the D-KEFS Color-Word Switching task. Factor et al. also observed a significant difference on the Wisconsin Card Sorting Test (WCST) with both FOG groups completing less successful categories than PD controls. The WCST requires an individual to alternate response sets to complete categories and is untimed. While not reported in the paper, it can be assumed that the FOG groups in that sample would have also committed increased errors due to set loss or perseverative responding.

While impaired set-switching deficits have been widely reported in PD with and without FOG, the current findings highlight the importance of considering the accuracy in addition to time. First, set loss errors provide a measure of cognitive functioning that is independent of motoric slowing and other disturbances that impact completion time¹⁸. Additionally, errors represent a breakdown of rule monitoring and mental control that is not mitigated by more cautious approach. In the present study, increased error rate for dopa-unresponsive FOG suggests disruption of a dopaminergic pathway. Moreover, association between FOG severity and error rates in the dopa-responsive group suggests the possibility of a common dopaminergic pathway underlying both the ability to monitor rules and the FOG phenomenon itself.

The classic model for the role of dopamine in executive functions in PD is the ‘dopamine overdose’ hypothesis first proposed by Gotham, Brown and Marsden in 1988¹⁹. Under this theory the relatively preserved ventral striatum can be overstimulated in the setting of dopamine replacement leading to overstimulation of limbic system in turn leading to impairment in executive functions mediated by the limbic and orbitofrontal system like learning, and risk taking. This theory has been tested and found some empirical support in the literature²⁰ as well as in the clinical setting with the advent of impulse control disorders observed particularly in relationship with dopamine agonists²¹. Interestingly, dopamine agonists have also been implicated in the generation of FOG^22-23. The notion that dopamine can interfere with PFC function was supported by a recent study showing reduced performance on dual cognitive tasks in the ON state vs. OFF state, despite increased pre-frontal cortex (PFC) activation in the ON-state²⁴. As discussed by Westbrook and Braver²⁵, dopaminergic tone in the prefrontal cortex is associated with working memory, while dopamine in the striatum is associated with motivation to exert cognitive effort. Performance on all the tests administered in this study required cognitive effort, but only the set-switching tasks demanded working memory to accurately alternate responses. Taken together, our findings suggest dopaminergic pathways in the prefrontal cortex play a critical role in both motor and cognitive deficits in FOG.

Results from this investigation also support the notion of pharmacologic subtypes of FOG with distinct underlying cognitive profiles. This finding, together with previous work on pharmacologic subtypes of FOG⁶, as well as phenomenologic subtypes, and more recently described phenotypic²⁶ and neural substrate subtypes²⁷ can help to explain the heterogeneity that has been observed in this patient population. Such heterogeneity can explain diverging and even contradicting results reported in imaging studies^28-30 and clinical trials in FOG^30-33. Recent work has reported improvement in FOG reporting following deep brain stimulation (DBS) of the subthalamic nucleus³⁴. However, these reports can only be applied to this specific population which includes non-demented, dopa-responsive patients. Other studies including largely dopa-unresponsive cohorts may yield different results in therapeutic trials. Similarly, if dopa-response dictates underlying cognitive profiles, dopa-response may also be associated with changes in underlying neural signatures.

Dopa-responsiveness to freezing also has implications for treatment. For instance, optimizing dopa-response and even DBS is often beneficial for patients with RFOG. For URFOG, enhancing cholinergic or noradrenergic function may prove effective. Prior studies of acetylcholinesterase inhibitors^35,36 as well as stimulants^13,31,32,37 for treatment of FOG have yielded conflicting results. A recent RCT showed that a cognitive training program reduced FOG during “on” but not “off” assessments³⁸. While the investigators did not classify participants according dopamine response, this would suggest that at least for URFOG, cognitive training may provide some benefit. Future studies of these non-dopaminergic therapies for FOG would benefit from characterization of treatment cohorts and targeting the URFOG subtype.

Limitations:

Several limitations must be considered when interpreting the present results. While FOG severity was rated using a validated questionnaire and gait metrics associated with freezing were quantified, frequency and duration of freezing events were not objectively assessed in the study. Given that the phenomenon of freezing is episodic and variable within and between days, remote wearable sensors would be required to obtain this data. Such devices are commercially available and could prove useful in future studies³⁹. Although the ON-UPDRS scores for the FOG and PD-control groups were very similar, and the PD-control group did not fluctuate, we can only infer that their OFF UPDRS score are unlikely to be different, since only ON UPDRS scores were collected.

Given that the study is cross-sectional, it remains unknown whether changes over time in set-switching parallel FOG severity. Interpretation of results from this study is also limited to PD patients without evidence of dementia. As such, it is unclear whether the nuanced association between set-switching and freezing in RFOG would remain in the context of more severe cognitive impairment, or if this would generalize across other cognitive domains as well.

In this study gait was tested on and off medication, while cognition was only tested while “on”, as it was believed that performance on cognitive tests during the off state in this population would be heavily confounded by motor impairment. We are therefore unable to speculate whether differences between dopa-response groups in set-switching and other cognitive domains persist in the off state, or if cognition is more improved with medication in one group than another. The use of untimed tests without graphomotor demands (e.g., Digit Span or Wisconsin Cart Sorting Test) would be required to address this question. If the dopa-responsive group showed greater improvement in cognition during the on state, this would further support our hypothesis of a common dopaminergic pathway. Future studies addressing this possibility are planned.

Conclusion:

We conclude that dopaminergic pathways play a key role in facilitating higher level compensatory responses that occur in patients with FOG, and as dopaminergic pathways degenerate executive function is no longer an effective compensatory response. Our findings highlight the importance of considering accuracy in addition to time when evaluating executive functions, as well as considering dopa-response in the characterization and classification of FOG.

Highlights.

• Freezing of gait in Parkinson’s disease is associated with executive dysfunction

• Executive function is worse when freezing does not respond to dopamine

• When dopamine responsive, freezing severity and executive function are correlated

• Findings suggest a common dopamine pathway for freezing and executive function