Relating Global Cognition With Upper-Extremity Motor Skill Retention in Individuals With Mild-to-Moderate Parkinson's Disease

Introduction

Despite clear evidence of deficits in upper extremity motor control and dexterity in Parkinson's disease (PD) (1, 2) that meaningfully impact on one's activities of daily living (3), most rehabilitation research and clinical practice for PD focuses on gait and balance problems. When prescribed, however, motor rehabilitation can improve upper extremity movement patterns and physical function (4, 5), depending on one's ability to learn and retain novel motor skills. While individuals with PD may benefit from physical rehabilitation, they demonstrate slower learning rates (6) and learn to a lesser extent (7) than individuals without PD, yet longer-term skill retention remains unclear (8). In light of this, some people with PD show marked gains following therapeutic intervention, while others do not [e.g. (4), see also (9)]. The ability to predict therapeutic responsiveness could help therapists streamline and personalize treatments. However, most predictive tools or models of post-intervention motor outcomes are time- and cost-intensive [e.g., annual clinical measures (10), neuroimaging (11), etc.].

In contrast, cognitive assessment may be a feasible, brief, and relatively inexpensive tool for gaining insight to an individual's motor learning capacity [see (12)]. Global cognitive status has been shown to predict gains in gait speed following standard-of-care physical therapy independent of primary diagnosis (13, 14). With respect to PD specifically, physical therapy combined with cognitive training may be more efficacious in improving reactive postural adjustments (i.e., responses to perturbations) and motor symptoms than physical therapy alone (15). Furthermore, lower-extremity physical therapy (i.e., aerobic exercises, treadmill training) has been shown to improve global cognition as well in people with PD (16), although combined physical and cognitive therapy may be more beneficial (17). However, the relationship between global cognitive measures and upper-extremity improvements in PD has not been explored. Empirically, visuospatial function has been linked with upper limb motor learning in both younger (18–21) and older adults (22–26) without PD, and since visuospatial deficits can occur with PD (27–29), this may help explain why people with PD tend to learn motor skills slower and to a lesser extent than those without PD. However, the extent to which cognitive impairment (global or specific) interferes with upper extremity motor learning in individuals with PD remains unknown.

In a recent randomized clinical trial in individuals with mild-to-moderate PD (clinicaltrial.gov registration number NCT02600858) (30), motor practice while “on” dopamine replacement medication (i.e., levodopa) improved 48-h retention of a functional upper extremity motor task compared to practice “off” dopamine replacement medication. The purpose of the present study was to perform secondary analyses of these data to evaluate whether cognition was related to skill learning in the upper extremity. Based on previous findings, it was hypothesized that cognitive functioning would be related to longer-term retention of an upper extremity motor task, where better cognition would be associated with more skill retention.

Methods

Participants

Twenty-three adults aged ≥50 years old with a confirmed diagnosis of PD were included in this secondary analysis of data from a previously published randomized clinical trial (clinicaltrials.gov registration number NCT02600858) (30). Inclusion criteria included idiopathic PD diagnosed by a neurologist, age 50–80 years, in Hoehn and Yahr stages 1–3, and had been on a stable antiparkinsonian medication regime for 1 month prior to pretest assessment as well as throughout the study. Exclusion criteria included prior surgical treatment of PD (e.g., deep brain stimulation), dementia [Montreal Cognitive Assessment (31) (MoCA) < 18] (32), and the presence of concomitant neurological conditions. Included participants must have been taking dopamine replacement medications. The clinical trial protocol required half the participants (n = 12) to complete upper extremity motor training while continuing to take their prescribed dose of levodopa medication; the other half (n = 11) skipped their first dose of medication each day of motor training such that they were “off” medication following overnight withdrawal. These participants took their remaining daily doses after they completed the motor training each morning. Details regarding dopamine medication and other participant characteristics have been previously reported (30).

Global cognition was measured using the MoCA, a brief cognitive screening tool in which scores range from zero to 30; a score of 26 (or higher) is considered to be normal cognitive functioning, as defined by the publisher (31). To evaluate upper extremity dexterity, participants completed the Nine-hole peg test (33) (a timed clinical measure of dexterity) and another timed experimental upper extremity dexterity task that simulates buttoning a shirt unimanually (24, 34, 35); for both these tasks faster trial times indicate better performance. Participants were also tested by a trained examiner with the motor subsection of the Movement Disorder Society-Unified Parkinson's Disease Rating Scale (MDS-UPDRS) (total range of scores = 0–152) (36). To evaluate depressive symptoms, participants completed the Geriatric Depression Scale (GDS) Short Form (37), a self-report rating tool consisting of 15 items and a score of four or lower is considered normal. Participants self-reported hand dominance. All participant characteristic data, including MoCA score, were collected in an initial visit while the participants were “on” their prescribed dose of dopamine replacement medication, regardless of which group they were randomized to (“on” vs. “off” medication).

Upper-Extremity Motor Training

As described previously (30), the motor training protocol required participants to complete a familiarization trial, then 50 training trials each day for three consecutive days. More details regarding the motor task are provided below. Participants were then re-tested 2 and 9 days later. Two-day follow-up was the stated primary outcome of this clinical trial and was therefore reported previously; thus, only the longer-term nine-day follow-up was included in this analysis.

The motor task used in this study was designed to mimic an activity of daily living [i.e., feeding (38)]. This task has been validated against subjective and objective measures of daily functioning in a cognitively impaired sample (39). The experimental apparatus was comprised of three “target” cups placed 16 cm from a center “home” cup at 45, 90, and 135 degrees (Figure 1). Participants were asked to use a plastic spoon held in their non-dominant hand to collect two raw kidney beans from the home cup and transport them to one of the three target cups. The non-dominant hand was used to ensure the task was not overlearned and to avoid potential confounds of a ceiling effect (40). Participants were instructed to move first to the target cup ipsilateral to the non-dominant hand, then to the middle cup, then to the contralateral cup, repeating this pattern four more times. Thus, each trial consisted of 15 reaches. The primary measure of performance was trial time, which began when the participant picked up the spoon and ended when they completed all reaching movements and placed the spoon back onto the table; thus, lower trial times indicated better performance. Dropping beans, transporting an incorrect amount, or moving to the wrong target were counted as errors, and the participant could not continue until the error was corrected, therefore errors contributed to longer trial times. Participants were not provided with performance feedback but could explore different movement strategies to optimize performance [i.e., discovery learning (41)]. As noted previously, each training session consisted of 50 trials (i.e., 750 reaches per session), and participants completed three training sessions over 3 days (1/day), totaling 2,250 reaches. This dose of training was selected based on previous feasibility and efficacy studies in other clinical and healthy populations (42, 43).

Figure 1. Participants used their non-dominant hand to complete a reaching task that simulates feeding oneself; participants use a spoon to select only two beans from the center “home” cup and deposit them into target cups. One trial consisted of 15 repetitions (i.e., five arcs to each of the three target cups). This figure was adapted from “Dexterity and Reaching Motor Tasks” by MRL Laboratory is licensed under CC BY 2.0.

Statistical Analysis

JMP Pro 14.0 (SAS Institute Inc., Cary NC) was used for all statistical analyses. To examine whether global cognition was related to the amount of learned motor skill, MoCA scores were included in a multiple linear regression model as a predictor of nine-day follow-up performance (α = 0.05), along with baseline motor performance, age, MDS-UPDRS Motor subsection score, and GDS as covariates. The MDS-UPDRS Motor subsection score and GDS were included to control for severity of PD motor signs and depressive symptoms, respectively. All continuous variables (age, MoCA, baseline motor performance, and 9-day follow-up) were normally distributed, as determined by Shapiro-Wilk tests. Furthermore, a partial correlation matrix indicated minimal collinearity between predictors, with only a moderate correlation between age and MoCA (r = −0.21). In addition, quantile range analysis indicated no outliers for any of the continuous variables. Thus, assumptions for linear regression were tested and met.

Baseline motor performance was measured as the first trial of the first motor training session. Similarly, follow-up was measured as a single trial performed 9 days after the last training session. Although we did not have a specific hypothesis regarding the effect of dopamine replacement medication on learning, we also included the variable of group (“on” vs. “off” medication) as a covariate to control for any confounds of dopamine replacement medication status on the primary outcome. We note that baseline motor performance was not different between the medication groups (p = 0.28), consistent with results from the primary analysis of this clinical trial (30). We also note that based on results from the primary analysis of this clinical trial [see Table 2 in original publication (30)], we were sufficiently powered to detect significant differences between the two timepoints (Cohen's d = 0.58; effect-size r = 0.27).

Results

Individual characteristics are provided in Table 1. Participants demonstrated mild PD symptoms and disease severity (median Hoehn and Yahr stage = 2, not shown in table). Using their non-dominant hand, participants completed the Nine-hole peg test and experimental dexterity task in 25.38 ± 4.31 and 102.81 ± 54.37 s (mean ± SD), respectively. Results for the Nine-hole peg test were consistent with previously reported values in PD (33). In addition, participants were bradykinetic, taking twice as long to complete the dexterity task as healthy older adults from previously reported data (35). MoCA scores ranged between 23 and 30, indicating that some participants were within the normal range of cognition while others fell below [based on (31)].

Table 1. Individual participant characteristics (n = 23).

Overall, nine-day follow-up performance on the motor task was significantly faster (better) than that of baseline [one-sample t₍₄₃₎ = −3.13; p = 0.0016], as shown in Figure 2A. This indicates an overall effect of motor learning in this sample. Linear regression model results indicated that only MoCA score predicted 9-day follow-up performance (β = −2.76; 95% CI [−5.11, −0.39], p = 0.0248), such that higher MoCA scores were associated with faster (better) trial times 9 days post-training. This is further illustrated in Figure 2B, which shows baseline and follow-up data for each participant. Line colors indicate each participant's MoCA score (warmer colors = lower MoCA scores), such that warmer colors tended to cluster at slower trial times for 9-day follow-up, while cooler colors tended to cluster at faster trial times. Participant age (β = −0.18; 95% CI [−0.89, 0.52], p = 0.59), severity of PD motor signs (β = 0.26; 95% CI [−0.34, 0.86], p = 0.37), GDS (β = 0.71; 95% CI [−0.73, 2.15], p = 0.31) and medication status group (β = 2.83; 95% CI [−2.02, 7.68], p = 0.23) were not significantly related to follow-up motor performance. Since only MoCA score was a significant predictor of follow-up performance, results from a bivariate regression are provided in Figure 3 to further illustrate the negative relationship between the two variables (color gradient consistent with that of Figure 2B).

Figure 2. (A) Mean motor task performance at baseline and 9-day follow-up. On average, trial time (in seconds) was significantly faster at nine-day follow-up compared to baseline. *Indicates p = 0.0016. (B) Motor performance for each participant at baseline and 9-day follow-up. Line color indicates each participant's MoCA score, with warmer colors indicating lower MoCA scores and cooler colors indicating higher MoCA scores (range 23–30).

Figure 3. Scatterplot of motor performance at 9-day follow-up as a function of MoCA score. Bivariate linear correlation results are shown in figure. Color gradient corresponds to that in Figure 2.

Discussion

The purpose of this secondary analysis was to determine whether cognition was related to upper extremity motor skill learning in individuals with PD. Results indicated that MoCA score predicted follow-up performance of a functional upper extremity motor task 9 days after the last practice session, more so than baseline performance and age (regardless of “on”/“off” medication groups). These findings align with previous work that suggest cognitive testing can be used to predict rehabilitative outcomes (44, 45) and support that global cognition may be a useful tool to predict motor learning in clinical populations (46–48). Although the MoCA was used to evaluate global cognition in this study, there is existing evidence that supports the value of assessing specific cognitive domains as predictors of motor learning (24, 44, 49).

Given the high prevalence of cognitive impairment among people with PD (50, 51), global and specific cognitive measures should be considered to identify which and to what extent various cognitive impairments interfere with learning different skills. In terms of global cognitive measures, the MoCA may be particularly sensitive to screening cognitive deficits associated with PD compared to other global cognitive measures of cognition (i.e., the Mini Mental State Examination) (32). We acknowledge, however, that the MoCA is a rapid cognitive screen that does not thoroughly assess the function of each cognitive domain nor is it validated to measure the function of individual cognitive domains, which may be a limitation to this study. Thus, a more comprehensive battery of cognitive assessments would determine whether specific cognitive domains (or specific cognitive deficits) more closely predict motor skill retention in this population. For example, visuospatial deficits may interfere with upper-extremity learning (23–25, 52, 53), while fluid cognitive skills or executive function may interfere with lower-extremity learning (54). While these previous studies have not focused on motor learning in PD specifically, the effects of particular cognitive deficits may not be PD-specific but instead generalize to a number of older patient populations who may be receiving motor rehabilitation for a number of reasons (e.g., stroke, joint replacement). As such, the effect of cognitive impairment on motor rehabilitation is gaining interest, within and beyond PD (12, 55). Future studies in motor learning should consider a more comprehensive battery of cognitive tests, especially those that evaluate visuospatial and executive abilities, in order to identify evidence-based targets for adjuvant cognitive or non-invasive brain stimulation therapies [e.g. (15, 17, 56–58)] that can be administered prior to or during upper- and lower-extremity motor therapy for people with PD.

In addition to providing empirical evidence as the groundwork for developing effective adjuvant therapies for motor rehabilitation in PD, this study offers clinicians a low-cost, easy-to-implement way to predict how responsive a person with PD might be to motor therapy. It is well-established that responsiveness to motor rehabilitation is often highly variable between PD patients [see 95% CIs in Robinson et al. (9)]. As such, the findings from the current study suggest that the MoCA may be a quick (~5–10 min) and simple way to predict how responsive a patient might be to upper-extremity training. This would inform therapists in how to streamline and tailor their treatments, and better allocate their time to activities that they know their patients will benefit from. Predictors of therapeutic responsiveness are already being explored outside PD using neuroimaging (59–61), neurophysiology (61–63), or genotyping (64–66), but these investigational methods are time- and cost-intensive, making them unfeasible for an allied health setting and out-of-pocket therapy.

In the published clinical trial (30), there was a modest effect of medication status during training (i.e., “on”/“off” medication while practicing the task) between baseline and 48-h follow-up task performance, such that the “on” medication group performed significantly better at this short-term retention period than did the “off” medication group. These results were interpreted to indicate that being “on” dopamine replacement medications may facilitate short term retention of motor skill. However, this secondary analysis indicates that the group difference was no longer present by the ninth day of retention, likely due to the modest effect of medication status on training previously observed. Instead, global cognition (which was not originally considered in the parent clinical trial) was a significant predictor of motor task performance well after training had been completed (9 days later), regardless of whether training had occurred “on” or “off” dopamine replacement medication. Indeed, dopamine replacement may be insufficient to offset the breadth of cognitive deficits associated with PD (67), and the short duration in which participants in the “off” group were withdrawn from their medication for training (relative to the 9-day duration of retention) may explain the lack of effect of group in this secondary analysis.

There are several limitations to this study. First, there was a limited range of MoCA scores in this sample, and more participants were in the normal range than below. Even though the MoCA is not a diagnostic tool (and is instead a cognitive screening tool), scores suggest that the majority of participants were cognitively intact, and more extensive neuropsychological testing would be necessary to determine whether participants with scores below the “normal” cut-off were in fact cognitively-impaired. This does not, however, take away from the findings and implications of this study, whereby rehabilitation-focused clinicians could still use the MoCA as a quick screening tool to better predict how patients might respond to upper-extremity skill training. Future studies will investigate a larger cohort of more cognitively-impaired individuals [such as people with PD who experience “freezing” (68–71)] to further test the generalizability of these findings. Second, a limitation of this study is that the participants in the “off” medication group resumed their regularly prescribed dopamine replacement therapy after training each day and throughout the 9-day retention period; thus, we are unable to discern potential effects of medication adherence or withdrawal on motor skill consolidation and retention. It is well-established that consolidation and retention are critical periods for motor learning, as well as acquisition (72). Third, we acknowledge that this study was not designed to directly test if global cognition would be predictive of clinical rehabilitation motor outcomes in individuals with PD, since it only evaluated the amount of skill retained over a period of 9 days. Performance of the functional upper extremity task used in this study has, however, been associated with subjective and objective measures of daily functioning in individuals diagnosed with Mild Cognitive Impairment (39), suggesting that the benefits of training may generalize to activities of daily living.

Conclusions

Our study supports the premise that cognitive impairments interfere with motor skill learning in PD, and provides the proof-of-principle that (1) cognitive screening may be a viable solution for personalizing motor rehabilitation for people with PD and (2) cognitive therapy and/or brain stimulation prior to, or concurrent with, motor training could enhance functional outcomes. Future mechanistic work should systematically test which specific cognitive domains are most relevant for different types of motor learning in PD to help inform targeted adjuvant cognitive or neurostimulation therapies that can enhance motor rehabilitation. For example, fluid cognition training may enhance gait adaptation, or non-invasive stimulation of parietal cortex could enhance functional upper-extremity training via visuospatial processes.

Data Availability Statement

Meta-data are available upon reasonable request. Requests to access these datasets should be directed to lee.dibble@hsc.utah.edu.

Ethics Statement

The studies involving human participants were reviewed and approved by University of Utah Institutional Review Board. The patients/participants provided their written informed consent to participate in this study.

Author Contributions

LD and SP organized the database. JLV and SS performed the statistical analysis. JLV wrote the first draft with assistance from CL-L. All authors contributed to conception and design of the study and contributed to manuscript revision, read, and approved the submitted version.

Funding

This work was supported in part by the National Institute on Aging at the National Institutes of Health (K01 AG047926 to SS and F31 AG062057 to JLV); University of Utah Office of Research (2015-16 grant), USA; the American Parkinson's Disease Association (2015-16 postdoctoral fellowship grant).

Conflict of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher's Note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

References