Dual-Task Performance and Brain Morphologic Characteristics in Parkinson’s Disease

Individuals with Parkinson’s disease (PD) have impaired automaticity, the ability to execute tasks without attention directed toward the task [1]. As a result, these individuals have limited dual-task (DT) performance, which is the skilled execution of two tasks simultaneously [1]. During daily activities, this ability is necessary for daily functions (e.g., ambulation or transfers) as they often are completed in the context of walking and attending to a secondary task like talking, manipulation of objects like a cell phone, or monitoring the environment [2]. It is hypothesized that many individuals with PD are impaired in this ability due to insufficient capacity to automatically process a task, as well as limited attentional resources [1]. Individuals with PD possess a lower capacity for automaticity and also poor coordination, balance deficits, bradykinesia/hypokinesia, freezing of gait, and cognitive deficits all of which worsen when confronted with complex or challenging tasks and environments, such as DTs [1, 3].

Performance of motor tasks, including gait, is less consistent and more effortful for individuals with PD when simultaneously required to perform a secondary motor or cognitive task [4]. Prior studies have shown that the performance of a DT leads to alterations in the activity levels in the prefrontal cortex (PFC), premotor cortex (PMC), and supplementary motor area (SMA) in healthy adults as well as those with neurologic disease [5, 6]. Among individuals with PD, similar to healthy adults and individuals with other neurologic diseases, brain activity in the PFC, PMC, and SMA is increased during DT paradigms [7, 8]. Additionally, specific syndromes in PD, such as freezing of gait or mild cognitive impairment, have been linked to the interaction and combined deterioration between motor, cognitive, and limbic functions [9, 10]. Specifically, cortical thicknesses in sensorimotor regions have been found to predict DT cost on gait speed among individuals without freezing of gait, while cortical thicknesses in frontal and parietal areas predict the same in individuals with freezing of gait [10]. Furthermore, dissimilar patterns of subcortical degeneration have been shown to explain differences in symptoms between tremor dominant and postural instability/gait difficulty subtypes such as gait impairment and cognition [11]. These symptoms exert a strong influence on the course of PD progression and the quality of life of those living with this disease. A comprehensive understanding of the neurologic underpinnings of DT abilities may provide insights leading to specifically targeted rehabilitation for individuals with these debilitating syndromes [12]. However, the brain morphology that may accompany or precipitate these impairments in DT performance among individuals with PD has not been fully described. Therefore, identifying regions that differ between individuals that are impaired and those that perform similar to healthy individuals could also provide targets for developing interventions that could attenuate the loss of automaticity during the progression of PD [13]. To this end, the purpose of this study was to determine the differences in cortical thickness and subcortical volumes between individuals who are high DT (highDT) performers and low DT (lowDT) performers in PD. It was hypothesized that the highDT group would exhibit significantly different patterns of volume and cortical thickness in the PFC, SMA, and entorhinal cortex, and that the cortical thickness of these regions would correlate with DT performance. Additionally, it was hypothesized that the highDT group would exhibit significant different patterns of volume in subcortical regions including the caudate, putamen, brainstem, and cerebellar cortex compared to lowDT group.

Exploratory analyses of volumetric magnetic resonance imaging (MRI) data were obtained from 42 people with PD and 47 older adults without a neurological diagnosis. These individuals were recruited as part of an observational study at the Cleveland Clinic Lou Ruvo Center for Brain Health [14]. Data obtained and analyzed included the following: demographics (age, sex, race, ethnicity), disease severity (Movement Disorder Society – Unified Parkinson’s Disease Rating Scale [MDS-UPDRS], Hoehn and Yahr stage, levodopa equivalent daily dose [LEDD], time since diagnosis, Montreal Cognitive Assessment [MoCA]), single-task (ST) and DT performance obtained during DT timed up and go (TUG) utilizing a serial subtraction task, and cortical volumes obtained from volumetric MRI on a 3T Siemens Skyra scanner. All data were obtained with participants in the clinically defined “ON” state. The “ON” state is defined as when PD medications begin to take effect with patients experiencing period of good symptom control.

This study is a secondary analysis of data from a prior study, which was conducted under Institutional Review Board approval. Additionally, each participant provided consent prior to their initiation of the parent study from which these data originate. Inclusion criteria were the following: neurologist-diagnosed PD as per the UK Brain Bank criteria [15] and between the ages of 50 and 85 years old. Participants were excluded if they had other neurologic diagnoses (e.g., stroke, multiple sclerosis), significant orthopedic injury or surgery in the prior 6 months that negatively affected gait or balance, history of deep brain stimulator placement, or an implant that was not MRI compatible. Participants who were left-handed (n = 3) were not included in the analysis to reduce the impact of variations in brain lateralization between individuals who are right-handed and those who are left-handed [16]. Additionally, 3 participants were excluded as they were missing values for variables included in the analyses. A total of 34 participants were included in substantive analyses and dichotomized into the lowDT group (n = 11) or highDT group (n = 23) by comparing their combined DT effect (cDTE) (explained below), a measure of automaticity and DT capacity, to a previously determined mean value for healthy older adults (μ = −74.2%). This cut point was selected as it provided a unique ability to interpret findings relative to healthy individuals of similar age that had minimal to no DT impairment [17]. If a participant’s cDTE was greater than the mean value for healthy older adults, they were included in the highDT group. If a participant’s cDTE value was below the mean, they were included in the lowDT group. All measurements were done in the clinically defined ON PD medication state on the same day that the MRI scan was completed.

The Hoehn and Yahr scale, MDS-UPDRS part III, and disease duration were extracted as measures of disease severity. The stage of functional disability in PD was defined with the Hoehn and Yahr scale [18]. Additionally, MDS-UPDRS part III was used to clinically determine disease severity [19, 20].

The MoCA was used as a measure of global cognition. It is a brief screening tool for global cognition that was designed to be sensitive to mild cognitive impairment [21]. Additionally, data from more detailed cognitive testing were also extracted including the trail making test parts A and B, judgment of line orientation, and symbol digit modality test. These measures were used to describe executive function, processing speed, inhibition control, and visuospatial perception [22‒24].

Depression, anxiety, and apathy were assessed, respectively, via the Geriatric Depression Scale (GDS), State Trait Anxiety Index (STAI), and single dichotomous response item. The GDS is a reliable and valid measure used to identify depression in older adults [25]. The STAI is considered a reliable and nonspecific measure of negative affectivity in an individual [26].

The brain volumes of each participant were obtained from a 3T MRI (Skyra, Siemens Medical Solutions USA Inc., Malvern, PA, USA) with semiautomatic segmentation and quantification of brain regions via the FreeSurfer Image Analysis Suite, version 6.0 (http://surfer.nmr.mgh.harvard.edu/) [27, 28]. Additional details regarding FreeSurfer processing can be found in online supplementary material 1 (for all online suppl. material, see https://doi.org/10.1159/000540393). MRI scans were conducted immediately before the DT assessment. Head motion during measurement was not quantified, but all scans were visually inspected for quality assurance in FreeSurfer. If any motion was encountered during extraction, the boundaries were manually corrected or the participant was discarded, depending on the degree of the artifact. Consequently, scans from 34 individuals with PD and 47 healthy controls were included after excluding two participants with PD due to segmentation errors. FreeSurfer brain volume measurements were extracted for the different regions of interest (ROIs). The ROIs that were selected for this study were based on findings in the following previous studies: caudate volume [29, 30], putamen volume [31], PFC thickness [29], PMC thickness [6], SMA thickness [30], cerebellar cortex volume [30], primary somatosensory (S1) cortex thickness [32], primary motor (M1) cortex thickness [30, 31], entorhinal cortex [33‒35], and brainstem volume [6, 9, 31, 33, 34, 36]. Mean cerebral cortical thicknesses were also included [6, 9, 31, 33, 34, 36]. Given that DT included a mobility task using the lower extremities, the paracentral lobule cortex thickness was also included.

DT performance was evaluated by having each participant execute a motor task and a cognitive task concurrently. Each task was tested both on its own and while simultaneously doing the other task without preferencing attention to one aspect over the other. Performance of both ST and DT trials was used to calculate automaticity and DT performance. Participants were instructed to “walk as quickly as you can safely” and/or “complete the subtraction task as quickly as you can accurately” in both the ST and DT conditions. The methods of calculating automaticity and DT performance are outlined below.

ST motor performance was assessed by having participants complete a single trial of the TUG test. Motor performance was quantified as seconds to complete the TUG [37].

Participants then completed a serial subtraction task while seated to assess their ST cognitive performance. Participants were instructed to perform serial subtraction by three from an arbitrary number between 80 and 100 for 20 s. The number of correct responses in the serial subtraction was recorded, and the correct response rate (the average number of seconds per correct response) was calculated [17].

To assess DT performance, participants completed the TUG again while simultaneously completing the serial subtraction task, described below as DT-TUG [37]. To minimize learning effects, the participants were instructed to begin at a different number between 80 and 100. DT motor performance was quantified as seconds to complete the DT-TUG. DT cognitive performance was quantified as correct response rate (the average number of seconds per correct response) using the time in seconds to complete the DT-TUG [17].

A negative score indicates poorer DT performance (DT interference) when compared to ST performance. On the contrary, a positive score shows improved DT performance (DT facilitation) under DT conditions when compared to ST performance [38]. As noted above, participants were placed in the lowDT group if their cDTE indicated a decline in performance as compared to ST performance of greater than 74.2%.

All analyses were conducted using SPSS 29.0 (IBM SPSS Statistics for Windows, Armonk, NY, USA: IBM Corp) with α at 0.05. Groups were compared on demographics and disease severity using independent sample t tests (for continuous variables) or χ² tests (for categorical or nominal variables). Shapiro-Wilk tests and Levene’s tests were conducted and indicated a lack of normal distribution or homogeneity of variances. Due to this, the substantive analyses proceeded with nonparametric testing using Quade’s ANCOVA to compare cortical thicknesses and brain volumes between the highDT and lowDT groups, while controlling for age, sex, MDS-UPDRS part III, total intracranial volume, and LEDD. Between-group effect sizes (ηp²) were also estimated for each hypothesized ROI. Subsequently, Quade’s ANCOVA was used to compare cortical thicknesses and brain volumes of the ROIs between highDT, lowDT, and a healthy comparison group of older adults while controlling for age, sex, and total intracranial volume. Benjamini-Hochberg corrections were applied to account for multiple comparisons. To further understand the relationship between brain morphology and DT performance, a hierarchical multiple linear regression analysis was conducted. The dependent variable in the model was cDTE. Block 1 of the model included covariates (age, MDS-UPDRS part III, sex, total intracranial volume, and LEDD). In block 2 of the model, the ROIs were entered in a stepwise fashion, with an inclusion threshold of p = 0.05.

No differences were found between DT groups on demographics, cognition, or neurobehavioral characteristics (p_s > 0.113) (see Table 1). The groups differed on HY with lowDT group having a greater proportion of participants in stage three compared to the highDT group having a greater proportion in stage two (p = 0.012). However, the groups did not differ on MDS-UPDRS part III (p = 0.116) or time since diagnosis (p = 0.567). Unsurprisingly, the DT groups differed in measures of DT performance (DTEs and mAAI) and ST performance, used in the calculation of DT effects (Table 1). When comparing the healthy group to the DT groups, the groups differed on sex with the healthy group having a greater proportion of females compared to either DT group (online suppl. material 2).

Notably, the highDT group had thicker cerebral cortices in both the left (p = 0.001) and right (p < 0.001) hemispheres grossly. Among the ROIs analyzed, the highDT group had thicker S1 cortices in the right hemisphere (p < 0.001), M1 cortices bilaterally (p < 0.001, left; p = 0.001, right), and SMA cortices bilaterally (p = 0.001, left; p < 0.001, right) (Fig. 1; Table 2). There were no significant differences between groups in any of the subcortical ROIs. Effect size estimates (ηp²) for the differences in cortical thicknesses or brain volumes between the highDT and lowDT groups for each hypothesized ROI are reported in Table 2. Broadly, effect sizes for hypothesized cortical ROIs had moderate to large effect sizes, while subcortical ROIs tended to have moderate to small effect sizes.

Additionally, differences in cortical thickness and brain volumes between highDT, lowDT, and healthy groups for each hypothesized ROI were tested. Of note, left M1 cortex thickness (p = 0.002), left PFC thickness (p < 0.001), and right SMA thickness (p = 0.003) differed between groups (see online suppl. material 3).

The hierarchal linear regression model found that the included covariates (block 1 – age, sex, total intracranial volume, MDS-UPDRS part III, and LEDD) did not relate to cDTE (p = 0.967). In block 2, the left paracentral lobule thickness was selected stepwise and significantly improved the model fit to the data (ΔF = 7.400, p = 0.011) such that the addition of the paracentral lobule thickness to the model improved the variance explained in cDTE by 20.8% (see Fig. 2). Among the other ROIs, only the left S1 cortex thickness met the threshold for inclusion following block 1 (p = 0.048); however, it did not meet criteria for inclusion in the model after the left paracentral lobule was added to the model.

Overall, the results of this study indicate that DT performance in PD may be related to cortical thickness of the sensorimotor and SMA cortices. Specifically, those with better DT performance (highDT group) had larger right S1 cortices, bilateral M1 cortices, and bilateral SMAs. Additionally, average cortical thickness in both hemispheres was greater in the highDT group. We initially hypothesized greater subcortical volumes in highDT performers given the cognitive control required for DT performance, but interestingly, no differences were found between groups on volumes of any subcortical ROIs. Interestingly, this continued to hold true when comparing to healthy older adults. PD pathology directly affects the brainstem, caudate, and putamen and indirectly affects the cerebellum and cerebral white matter throughout its development before PD diagnosis and during the disease post-diagnosis [39]. Given this, it is possible that the effects of PD on these regions drive individuals to compensate with other cortical regions to maintain DT performance. It is also possible that any morphologic changes in these regions associated with DT performance occur early in disease development. This notion is supported by the evidence that DT capacity is limited early in the prodromal state of PD [40]. Of note, while the findings indicate some mixed lateralization of the results, with the highDT group possessing larger S1 cortices only in the right hemisphere (as noted above), bilateral differences for each of the regions were present before correction for multiple comparisons. This loss of statistical difference may be due to insufficient sample size. Given this, we will focus on the regions that were found to be different between groups with less emphasis on laterality.

Statistically significant findings for the cerebral cortex thickness of the entire left and right hemispheres were observed. These were expected, as it has been shown that poorer DT capacity is associated with smaller brain volumes in other populations [41]. The findings of this study extend prior studies and suggest widespread neural involvement in DT performance. This is supported by an extensive body of literature using functional neuroimaging during DTs, which indicates the involvement of multiple diverse regions throughout the cerebral cortex [1, 7, 8].

The findings of this study indicated that individuals in this study with better DT performance had thicker SMA and sensorimotor cortices. Thicker SMA cortices in highDT performers may reflect their role in working memory and the performance of learned voluntary movements [42]. Previous studies have demonstrated increased activation localized to the SMA during DT conditions when compared to ST walking due to its role in motor preparation, planning, and sensorimotor integration [5]. The results of our study extend these findings. Additionally, there is strong evidence from mobile neuroimaging studies to support the role of the SMA in dual-tasking among individuals with PD [7]. Given the complexity of the DT paradigm utilized in the current study, it is unsurprising that SMA thickness was related to DT performance. Additionally, the finding that the highDT group had thicker S1 and M1 cortices could reflect their role in receiving and organizing afferent inputs and generating neural impulses that produce and regulate movement [36]. Greater thickness in both M1 and S1 may allow for greater capacity to integrate sensory information to execute the appropriate motor response.

Notably, the left M1 cortex, left PFC, and right SMA cortex differed in the analyses of differences between DT groups and healthy older adults. Specifically, PFC thicknesses in both the highDT and lowDT groups were significantly different than in the healthy comparison group. Prior research confirms this finding, suggesting that PD affects this region in early stages of the disease [43]. Additionally, lowDT performers had thinner M1 and SMA cortices than both the highDT and healthy comparison groups. Providing further insight into these findings is the multiple linear regression analysis which indicated that the thickness of the paracentral lobule was related to DT performance in people with PD. Greater thickness in the paracentral region may signify more efficiency with motor execution in the lower extremities [44]. This could be related to better automaticity with lower extremity motor tasks and may explain its greater thickness in the highDT group. Additionally, given the findings of the regression analysis, the paracentral region may play a unique and integral role in the execution of DTs for people with PD. This is supported by previous studies that show the impact of PD on the sensorimotor cortex and the role of this brain region in DT performance [7, 36]. However, before this study, morphologic differences associated with DT performance had not been well described in individuals with PD. Taken together, the results of this study provide an initial understanding of the brain morphology changes among individuals with differing DT abilities. This can provide a vital first step toward the development of interventions that could minimize the loss of DT ability and automaticity alongside the progression of PD.

While this study provides insight into brain regions correlated with DT performance in PD, it is limited in several ways. First, this study was a secondary analysis; hence, the data were not designed or collected to answer the research questions posed in this study. Consequently, the relatively small sample size may be prone to type II errors. Additionally, participants were divided into lowDT and highDT groups based on the mean value of cDTE from a group of healthy subjects from a previous study, where values were variable (μ = −74.2 ± 131.3). While utilizing this cut point can be a weakness given the wide variability reported in the prior study, the ability to interpret the findings of this study relatively to DT ability of a comparable healthy group is enhanced by this approach. Further research is needed to establish more definitive methods for classifying individuals as highDT or lowDT performers. Another item to consider is the mismatch in time allowed to perform serial subtraction between ST and DT paradigms. While participants were given 20 s to complete the cognitive ST, they were constrained to the duration it took them to complete the DT-TUG to perform the cognitive portion of the DT. Differences in the time each task was performed could have influenced the correct response rate, highlighting a limitation in the methodology of this calculation. This study also examined only one aspect of DT performance (cDTE). Future work could investigate cortical thickness within other aspects of DT performance such as task-specific interference and attention allocation. Additionally, the sample only included people with mild to moderate PD. As such, these results may not be generalizable to individuals with advanced PD. Due to the cross-sectional nature of the study, we cannot conclude that progressive cortical thinning, associated with PD progression, caused the observed differences in DT performance. Future research should utilize longitudinal approaches to better understand the relationship between cortical thinning, DT performance, and PD progression. Specifically, investigating the relationship between brain volumes and other DT variables, including task-specific interference or attention allocation index, may provide valuable insights into the complex interplay between these factors. Future work would also benefit from comparison of DT neural correlates between individuals with healthy adults and those with PD. Last, studies utilizing mobile neuroimaging techniques (e.g., fNIRS, EEG) to study real-time brain function during DT activities in PD could provide further insights into the neural control that automatic and attentional processes contribute to DT performance.

The results of this study suggest a relationship between cortical thinning and automaticity in PD. Overall, DT performance may be related to broad and diverse cortical regions. Specifically, thinner SMA, S1, and M1 cortices were found in those with poorer DT performance in individuals with PD. The morphologic differences in these regions may account for worse DT performance due to decreased motor planning and execution. The influence of the progression of neural degradation throughout PD on DT performance warrants further research.

The authors thank the participants who completed this study.

This study protocol was reviewed and approved by the Institutional Review Board on ethical standards on human experimentation at the Cleveland Clinic, Approval No. 15-987. Written informed consent was obtained from participants to participate in this study.

The authors have no conflicts of interest to declare relevant to the content of this article.

Research reported in this publication was supported by the National Institute of General Medical Sciences of the National Institutes of Health under Award Number P20GM109025. Content is the sole responsibility of the authors and should not be taken to reflect the official views of the National Institutes of Health.

S.C. and J.L. both contributed to the conception and design of this project and the acquisition, analysis, and interpretation of the data. S.C. completed the initial draft, and Y.C., M.L. N.F., V.M, and J.L. critically reviewed the manuscript. All authors read and approved the final manuscript.

The data that support the findings of this study are openly available at nevadacntn.org.