How Cognitive Reserve should Influence Rehabilitation Choices using Virtual Reality in Parkinson's Disease

Virtual reality (VR) is used in the rehabilitation of patients with Parkinson's disease (PD) in several studies. In VR trials, the motor, physical characteristics, and the degree of the disease are often well defined, while PD cognitive reserve is not. This systematic review was performed to define a cognitive profile for patients with PD who could best benefit from using VR to enhance functional motor aspects during rehabilitation. PubMed, Cochrane Library, Scopus, and Web of Sciences databases were analysed to identify randomized clinical trials (RCT) and randomized pilot trials that addressed the rehabilitation of motor symptoms in subjects with PD using VR. The included studies used Mini-Mental State Examination (MMSE) or Montreal Cognitive Assessment (MoCA) to evaluate the cognitive aspect. Only articles written in English and with full texts were considered. The risk of bias from all included studies was assessed based on the Cochrane risk-of-bias tool and the PRISMA guideline was considered. Eighteen articles were eligible for review, including three randomized pilot trials. All studies aimed to evaluate the effect of VR on the motor aspects typically affected by PD (balance, postural control, risk of falls, walking, and reaching). The most widely adopted approach has been nonimmersive VR, except for one study that used immersive VR. Both the benefits of physical activity on the motor symptoms of patients with PD and the impact of cognitive reserve during the rehabilitation of these patients were highlighted. The analysis of the results allowed us to outline the ideal cognitive profile of patients with PD who can benefit from the effects of rehabilitation using VR.


Introduction
Parkinson's disease (PD) is a chronic progressive neurodegenerative disease, characterized by the loss of dopaminergic neurons in the pars compacta of the substantia nigra and the accumulation of alpha-synuclein aggregates in specific regions of the brain stem, spinal cord, and cerebral cortex [1]. e estimated prevalence of PD in industrialized countries is 0.3% in the general population (1% in people over the age of 65 and 3% over 80 years), with incidence rates from 11 to 19 per 100,000 people each year [2,3].
Patients with PD may also have affected cognitive functions, particularly global cognitive performance (as measured by the MMSE screening test) and behavioral deficits that affect aspects of social and community life [4]. e most common cognitive symptoms are deficit of attention and executive functions (working memory, planning, and inhibition), difficulties in episodic memory, verbal fluency, and visuospatial and visuoperceptual abilities [5].
Cognitive reserve (CR) is a theoretical construct that describes differences in individuals' susceptibility to cognitive, functional, or clinical decline due to ageing or neurological disease [6].
is concept is fundamental in neurodegenerative disorders such as PD, considering the severity of motor and cognitive disability and the functional impact on daily life [6]. Higher levels of CR are thought to be related to delayed disease onset and higher cognitive performance [7], and higher CR was associated with a better performance on the MMSE, thus confirming the protective role of CR on global cognitive functioning.
CR cannot be measured directly; it encompasses several different factors, including genetics, environment, education, occupational demands, lifetime experiences, and mental stimulation [8,9].
Studies in the literature show that the level of education and physical activity, especially aerobic, and cognitive activities, reducing the loss of brain mass and strengthening compensatory circuits, have protective effects on the brain [10]. e study by Koerts et al. highlighted the relationship between CR and impairment of executive functions, that is, cognition skills; they pointed out that patients with PD who have high premorbid intellectual capacity show fewer cognitive deficits than patients with low premorbid capacity [11]. e complex management of PD can be achieved through a calibrated combination of drug therapy and rehabilitation. Physiotherapy aims to maximize the quality of movement and promote functional independence and general fitness in patients with PD, minimizing secondary complications of the disease [12].
From a rehabilitation perspective, virtual reality (VR) represents an alternative, noninvasive therapeutic modality, often used in association with conventional rehabilitation, to cope with the degenerative characteristics of PD. Furthermore, VR is more captivating for patients with high CR, as Pazzaglia et al. pointed, and the therapeutic exercise is perceived as more exciting and fun by having visual and auditory feedback contextual to the movement [13]. e following are the two main categories of VR: immersive, which allows a more direct experience of virtually generated environments, and nonimmersive, which allows a subject to observe, through a standard high-resolution monitor, a virtual environment with which he/she can interact through interfaces, such as keyboards and controllers [14].
Various studies, considering the premises to integrate cognitive and motor aspects, propose multimodal rehabilitation approaches that combine motor training with cognitive stimuli through technologies and virtual reality for patients with PD. e common goal is to create an enriched environment capable of stimulating different cognitive aspects, involving the subjects with a more playful approach [15][16][17].
Often, the physical characteristics and disease stage of PD patients, included in VR trials, are well defined; the same does not happen regarding their cognitive profiles. is could erroneously suggest that VR-associated motor rehabilitation may be useful to all patients with PD regardless of the degree of cognitive reserve. About 25% of patients with PD, especially after the age of 70, may experience mild cognitive impairment or dementia. In addition, in patients with PD, dual-task rehabilitation exercises or multimodal activities are not always recommended, especially in the presence of cognitive or complex tasks, which can lead to freezing of gait, loss of balance, and increased falls, all due to attention deficits and the reduction of automatisms and psychomotor speed in patients with PD [18].
From this premise, our hypothesis was that VR may be effective in patients with PD who respond to a specific neuro-cognitive profile. On the other hand, in PD patients with inadequate cognitive reserve, VR may not have the same therapeutic efficacy as suggested by Imbimbo et al., where patients with a higher cognitive reserve benefited more from the VR treatment. In contrast, patients with low cognitive reserve could achieve better results by following a traditional rehabilitation program [19]. erefore, the cognitive reserve of PD patients could indicate the disease's evolution and help clinicians choose the most suitable rehabilitation strategy [14].
As a result, the considerations addressed so far lead to the goal of this systematic review, which is investigating to what extent neuro-motor rehabilitation with VR is useful for improving the motor aspects in patients with PD in relation to the cognitive reserve.

Search Strategy.
is systematic review included articles published in the last 10 years (from 2011 to July 2021), according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines [20], and evaluated studies related to the rehabilitation of patients with PD using VR in the following databases: PubMed, Cochrane Library, Scopus, and Web of Science.
Different combinations of the following MeSH terms were used to select the articles: (Parkinson OR Parkinson's Disease) AND virtual reality AND (rehabilitation OR training OR exercise). e reference lists for most of the relevant studies were scanned for additional citations. Country, author, affiliated institution, and enrolment period data were extracted and reviewed to identify and exclude duplicate publications using the same cohort. Any disagreement regarding accepting full-text articles was resolved by discussion until a consensus was reached.

Study Eligibility Criteria.
Our target was randomized clinical trials (RCTs) and randomized pilot studies (full text in English) and studies evaluating cognitive aspects using the Mini Mental State Examination (MMSE) [21] or Montreal Cognitive Assessment (MoCA) [22] and that deal with the motor aspects of rehabilitation through virtual reality; the motor aspects were "balance," "falls," "ambulation," "postural control," and "reaching." MMSE is a short exam, used to evaluate the patient's neuro-cognitive performance by administering a few questions to test orientation, memory, attention, calculation, and language. e total score is between 0 and 30; a score ≥ 24 indicates normal values. e MoCA acts as a quick screening for mild cognitive impairment. It evaluates different cognitive domains: attention and concentration, executive functions, memory, language, visuo-constructive skills, abstraction, calculation, and orientation. e maximum possible score is 30; a score ≥26 is considered normal.
Studies that adopted one or both of the previously mentioned scales were preferred to facilitate the analysis of the results. In this way, based on similar and comparable data, it was possible to consider the cognitive characteristics of the patients.
Studies other than RCTs and those that administered VR treatment to patients with neurological conditions other than PD (stroke and multiple sclerosis) were excluded.

Studies Quality Evaluation.
e methodological quality of the studies was assessed using the PEDro scale [23]. Studies with scores ≥ 9 were of "excellent" quality. Studies with scores from 6 to 8 were considered "good," studies with scores from 4 to 5, "fair," and those with scores ≤4, "poor." e risk of bias was also assessed for each RCT using the Cochrane risk-of-bias tool [24]. e main domains were evaluated in the following sequence: (1) selection bias (generation of randomized sequences and allocation concealment); (2) reporting bias; (3) performance bias (blindness of participants and staff); (4) detection bias (blinding of the evaluation of results); (5) attrition bias (incomplete outcome data, such as those due to dropouts); (6) other sources of bias. e scores for each domain of bias and the final score for the risk of systematic bias were classified as low, high, or unclear risk.

Results
e initial search, carried out through electronic databases, produced 108 results. is search was complemented by a manual search of individual citations of systematic reviews and articles included in the review, identifying 11 additional studies. After duplicates were removed, the remaining 71 publications were reviewed according to their titles and abstracts. is led to the exclusion of 11 publications. Of the remaining 60, the full text of 11 studies could not be found (as they were posters and abstracts presented at conferences). e subsequent screening of the remaining complete texts allowed the identification of 15 publications relevant for the revision, and another three studies resulting from the selection of citations were added. Overall, 18 RCTs published in English were screened for inclusion ( Figure 1). e selected studies are shown in Table 1, which describes the type of VR and the protocols used, the results measured, the evaluation times, and the presence of adverse events during treatment.
According to the PEDro scale (Table 2), the mean methodological quality of the included RCTs was 6.1, indicating the overall good quality of the included studies. e risk of bias was considered low for 11 articles, while for the remaining seven articles, it was considered high (Table 3). e most frequent sources of potential bias were performance bias (related to participant and staff blinding), concealment of distribution in groups, the presence of uncompensated dropouts from analysis by intention to treat, and incomplete result data. Reports excluded (n = 11) Reports sought for retrieval (n = 60) Reports sought for retrieval (n = 11) Reports not retrieved (n = 11) Reports not retrieved (n = 0) Reports assessed for eligibility (n = 49) Reports assessed for eligibility (n = 11)    Parkinson's Disease 7    It should be noted that, among the studies included in the review, Del Din et al. [26], Mirelman et al. [35], and Pelosin et al. [36] included elderly patients with mild cognitive impairment; therefore, of the 1393 patients analysed in the studies, only 1052 were PD patients, with a mean age of 68.8 years and a mean disease duration of 8.47 years.
In all studies, the cognitive level of the patients was assessed using the MMSE and/or the MoCA. e inclusion of patients in the trials, except in Del Din et al.'s study [26], was defined using the Hoehn and Yahr scale (H&Y) [41] relating to the progression of the disease (Figures 2 and 3).
A synthesis study was conducted on the patients included in the various protocols to observe under which conditions VR can be used effectively. None of the studies examined showed the influence of cognitive reserve (or the analysis of the patient's cognitive profile) on the results. However, no study planned the treatment protocol with VR by comparing subjects with high cognitive reserve and groups with poor cognitive reserve.
All the studies included in the review aimed to evaluate the effects of VR-associated rehabilitation on the motor characteristics typically affected by PD. More specifically, seven trials [25, 27-29, 31, 37, 39] focused on improving balance; five studies [15,25,26,35,36] addressed the risk of falls reduction and their incidence.; seven articles [17, 26-28, 32, 34, 39] aimed to evaluate the effects of rehabilitation treatment on walking; three articles [28,38,40] dealt with the problem of postural control in patients with PD; only one article [30] focused on motor symptoms (assessed through the third section of the UMPRS scale) and one article [33] dealt with motor performance during reaching exercises. Unfortunately, it is impossible to compare the individual studies' results as different outcome measures were used.
Among the studies listed, two [32,40] evaluated the effects of VR on sensory integration, and three [16,34,36] examined the effect of VR on brain activation and cholinergic activity.

Discussion
is review aims to investigate to what extent neuromotor rehabilitation with VR is useful for improving the motor aspects in PD patients in relation to the cognitive reserve.
Most of the studies analysed in this review stated that VR associated with conventional rehabilitation produced better results, compared with rehabilitation alone, in terms of increasing motor characteristics, such as walking, balance. and postural stability, typically affected by PD. VR is a good rehabilitation option, especially when combined with conventional therapy, and seems more suitable in patients with a good cognitive reserve, measured indirectly with mean MMSE and MoCA scores of 27.94 ± 0.86 and 23.43 ± 2.04, respectively. Working with VR can be stimulating in patients with a high cognitive reserve as it is challenging, as Pazzaglia et al. pointed, where the exercises are perceived as interesting, motivating, and funny, providing immediate visual and auditory feedback [13].
In the literature, several studies consider VR an efficient tool for the rehabilitation of patients with PD. Endurance training, especially exercises performed on the treadmill, can improve balance, reduce gait disturbances, improve speed, stride length, and walking [42][43][44]. VR offers the opportunity to simulate immersive and controllable environments,  Author  1  2  3  4  5  6  7  8  9  10  11 Total score Bekkers et al. [25] N  with the possibility of customizing the rehabilitation treatment. However, within the review, there are several articles, eight specifically [17,25,29,32,33,37,39,40], which considered the effects of rehabilitation associated with VR on a par with those of conventional rehabilitation, suggesting that the use of VR could complement rehabilitation to increase motivation during treatment [17]. e homebased administration of VR could represent a valid alternative for subjects with PD with limited access to rehabilitation services [39].
Examples of exercises with VR carried out at home can be found in the studies of Gandolfi et al. [28] and van der Kolk et al. [30], which have associated VR with balance training and aerobic exercise, respectively.
It should be emphasized that the activity performed through VR guarantees good adherence to the treatment because, during the exercise, the integration of motor and cognitive skills is favored and reward circuits of the brain are stimulated [28,45], which increases the possibility for patients to choose to train at any time of the day.
However, it is necessary to consider the cognitive aspects of patients with PD and their complex motor picture. Gandolfi et al. [28] admit that their results study should not be generalized and applied in patients with significant cognitive decline, as VR could be risky. Indeed, the European Physiotherapy Guideline for Parkinson's disease [18] says that dual-task or multimodal therapeutic exercises are not always indicated for parkinsonian patients because they can lead to freezing, loss of balance, and increased falls, especially during complex cognitive tasks. For this reason, VR could only be used with patients that respond to a specific cognitive profile since it may not have the same therapeutic efficacy in others with reduced cognitive reserve.
is is confirmed by the study of van der Kolk et al. [30], in which the participants' cognitive level, considered normal (MoCA � 26, 3), did not prevent the occurrence of adverse events, such as arthralgia, back pain, and palpitations, related to VR treatment, as well as falls, heart problems, and musculoskeletal damage, even if not related to exercise. In addition, the same authors state that several patients, from the VR treatment group, have left the study due to the onset of technical problems; this suggests that the presence of supervision or assistance can help in some circumstances to continue a training session with VR.
Although VR shows numerous advantages (related to learning motor skills through repetitive practice, performance feedback, and motivation) [46], it also presents some critical issues for patients: insufficient perception of depth and lack of tactile feedback (which, the latter, can cause difficulties when performing virtual tasks that simulate reality) [47,48]. In addition, a recent systematic review results state that patients with advanced age may find VR games complicated or boring and may need supervision to complete the task undertaken [49]. Some of the studies described using VR rehabilitation programs have shown how resistance exercises, stretching, and cognitive rehabilitation can improve the patient's quality of life. is is because patient perceives themselves as an active part of treatment. However, the cognitive reserve was not considered, and its impact on rehabilitation was therefore not evaluated. However, we can infer from the review studies and the literature that the cognitive reserve should be considered in the evaluation phase of patients with PD to plan the optimal, tailored therapeutic approach.
Piccinini et al. [50] examined the influence of cognitive reserve on balance rehabilitation, using conventional therapy, in patients with PD. e results showed an improvement in balance, and regarding the relationship between the cognitive reserve and balance, the condition of patients with a lower cognitive reserve index (those with a lower level of education) improved more than that of patients with a high cognitive reserve index. It has been hypothesized that patients with better cognitive reserve should work on more stimulating mental tasks through approaches such as VR, dance, and technological tools. ey found an inverse correlation between the level of cognitive reserve and the improvement of balance in patients with PD undergoing traditional rehabilitation, which highlights the important role that life experience, education, and recreational activities play on the individual's ability to cope with a brain pathology.  Imbimbo et al. [19] examined the relationship between VR and the cognitive reserve in patients with PD. e exercises proposed were intended to improve coordination and balance. At the end of the study, it was observed that, in relation to the cognitive reserve, some patients, unlike others, showed no improvement. VR showed better result in patients with a medium/high cognitive reserve index; these PD patients are accustomed to the use of technology, unlike subjects with a lower level of cognitive reserve who were uncomfortable with this tool and may felt less stimulated to learn. e results of our review may confirm Piccinini et al.'s results [50], which had suggested using a more complex rehabilitation approach for patients with a higher cognitive reserve.
is study had a few limitations: only four databases were searched, and we acknowledge the possibility that we did not identify all relevant studies.

Conclusion
Most of the studies analysed in this review included subjects with an MMSE score ≥ 24 and a H&Y stage between 2 and 3. Rehabilitation associated with VR was proposed for patients with PD with a mean score (mean of averages) of 27.94 (SD � 0.86) and 23.43 (SD � 2.04) for MMSE and MoCA, respectively, which shows a normal or slightly reduced cognitive level (if we consider the cut-off of 26 for MoCA).
According to the disease progression state, patients with PD who underwent treatment with VR had, on average (average of averages), an H&Y stage of 2.5 (SD � 0.60), indicating a slight bilateral involvement of the disease with recovery of balance on the pull test.
In conclusion, the results of these studies show that VR is a useful strategy that improves motor aspects mainly affected by PD and is feasible for patients with a normal cognitive level and an H&Y's stage less than three. is innovative approach, excluding excessively strenuous activities, is feasible at home and should preferably be performed in the presence of a caregiver or supervision.
Data Availability e data supporting this systematic review are from previously published studies, which have been cited.