Parkinson’s disease (PD) is a progressive neurodegenerative disorder that primarily affects movement, caused by a neuronal loss of dopamine-producing neurons in the substantia nigra and resultant degeneration of dopaminergic pathways in the basal ganglia. The cardinal symptoms include tremors, muscle rigidity, bradykinesia, and balance problems. Although there is currently no cure, various treatments, including medication and physical therapy, can help manage the symptoms and improve quality of life. Postural instability is a common and disabling symptom of PD, affecting about 20% of patients at onset and rising to 90% with disease progression (1); it is thought to result from changes in static posture and postural reflexes.

Gait dysfunction has emerged as the greatest challenge in PD management, likely due to its paroxysmal nature and the complex, interconnected motor and sensory circuits underlying the different domains of postural control (2, 3). Postural control strategies rely on a complex integration of visual, vestibular and proprioceptive somatosensory inputs. Sensory processing in PD is disturbed at the subcortical and cortical levels (see Takakusakifor et al. (4) for review). An insufficient supply of cholinergic and monoaminergic input impairs sensory integration within the thalamus (5). Furthermore, the reduced cholinergic supply from the basal forebrain, determines a deficit in cortical processing of proprioception, (6, 7), vestibular graviception (8, 9), and visual sensation (10) in the parietotemporal cortex.

Recent papers have demonstrated that the inability to efficaciously utilize vestibular information to retain upright stance is a major determinant of imbalance in PD, independent from visual and somatosensory processing (11, 12). Since postural instability has been associated with disease progression and higher Unified Parkinson’s Disease Rating Scale (UPDRS) scores (13–15), therapeutic approaches aimed at enhancing the ability to utilize vestibular information could also be beneficial in addressing motor impairments in PD.

The present study aims to explore whether an upper limb aerobic exercise, performed using a device that primarily targets the axial muscles of the cervical-dorsal spine, rebalancing proprioceptive input from neck muscles (16), can improve postural control and, consequently, motor symptoms in individuals with PD. Specifically, the effects of two exercise modes—supervised exercise (SE) and not supervised, voluntary exercise (VE)—were analyzed.

Supervised exercise, is comparable to forced exercise, a form of aerobic activity that has been receiving increasing attention in the research literature, particularly in studies on PD (17–23). It involves the assistance of an external source or person to help individuals exercise at higher intensities or for longer durations than they would voluntarily consider possible. The primary distinction between the SE and VE is that the former enhances the patient’s voluntary effort to achieve a higher exercise rate, whereas VE is performed at the patient’s self-selected, typically lower, pace.

It has been suggested that contradictory results in rehabilitative treatments may be due to differences in the rate or intensity of exercise performed (17). This hypothesis indirectly highlights the distinction between VE and SE, with SE pushing patients beyond their voluntary limits.

We used the computed dynamic posturographic EquiTest® System (NeuroCom Int. Inc., Clackamas, OR, United States) (24) to assess postural stability—the ability to control the center of mass in relation to one’s base of support during both static and dynamic tasks.

We recruited 28 patients with PD, 14 of whom underwent training with SE and 14 with VE. The average age of the two groups was not significantly different: 66.9 ± 7.5 and 68.6 ± 9.1, respectively.

All patients were taking L-Dopa at varying doses (range 400 mg–750 mg) and/or dopamine agonist.

The patients performed about 50–60 repetitions for each series. It is important to underline that after learning the correct trajectories, the patients’ spontaneous voluntary rate was no more than 30–40 repetitions per series; consequently, the number of repetitions performed by the VE group was generally lower than that of the SE group.

At time 0, before training with Angel’s Wing, the Hoehn and Yahr and UPDRS Part III scale scores did not differ between the two groups. For both scales, there was a statistically significant improvement at the end of the 8 weeks of training, but only in the SE group: p = 0.031 (Hoehn and Yahr scale), p = 0.007 (UPDRS Part III scale) (Table 1).

In Figure 3 and Table 1, the results of the sensory analysis descriptive data are reported. The statistical analysis shows a significant improvement in vestibular SOT values in subjects belonging to the SE group (p = 0.017), with a post-training percent improvement of 17.1% (Figure 3E). Visual and somatosensory scores did not change post-training in either group. The composite subscore of the PD subjects was significantly higher post-training compared to pre-training, but only in the SE group (p = 0.0001) (Figure 3D).

A mixed-model ANOVA was conducted to examine the effects of Time (pre vs. post) and Group (SE vs. VE) on the vestibular SOT values. Results revealed a significant main effect of Time (F(1,24)=5.2, p = 0.031, partial η² = 0.18), indicating that values changed from pre-to post-measurements overall. There was also a significant main effect of Group (F(1,24)=8.7, p = 0.007, partial η² = 0.27), showing that SE and VE groups differed in their overall vestibular SOT scores. Finally, the interaction between Time and Group was significant (F(1,24)=4.8, p = 0.038, partial η² = 0.17), suggesting that the change over time differed between the two groups.

The difference between the pre-training and post-training values of the Vestibular SOT in the SE group was analyzed for correlations with Visual SOT, Somatosensory SOT, age, and clinical scores of the subjects; however, no significant correlations were found.

No statistically significant correlation was found between UPDRS scores at baseline (time 0) and Vestibular SOT values.

Vestibular efficacy is recognized as a key factor in abnormal postural control in PD, regardless of the extent of nigrostriatal degeneration (12). This fact may partly explain why the balance impairment in PD patients is poorly responsive to dopaminergic medications (3).

Using the SOT within the EquiTest system, we demonstrated that patients with PD and balance impairments who participated in an 8-week active SE training program exhibited improved utilization of vestibular input and showed significantly greater improvements in clinical scale scores compared to those who underwent VE training.

The difference between the pre-training and post-training values of the Vestibular SOT in the SE group did not correlate with Visual SOT, Somatosensory SOT, age, and clinical scores of the subjects. Consistent with previous research (30), the Vestibular SOT does not correlate with the initial severity of PD.

SOT creates sensory-conflict conditions, rendering the visual and/or proprioceptive balance control systems unreliable. As a result, it serves as a useful tool in rehabilitation and clinical research for assessing postural stability. The use of the SOT to study postural alterations in PD is not a novelty. Huh et al. (11) were the first to use the SOT to investigate the vestibular contributions to postural control in patients with PD, demonstrating that those with freezing of gait exhibited significantly impaired postural sensory processing, particularly an inability to utilize vestibular information. The same conclusion was reached by Bohen et al. (12), who also utilized SOT.

In our study, the improvement observed in the first group of our patients, those who underwent training through SE, cannot be attributed to any form of vestibular ‘rehabilitation,’ as no vestibular stimulation occurred under our training conditions (upper limb exercise in a seated position). There is evidence that neck proprioceptive input enhances the vestibular-evoked perception of body rotation (31). This is supported by findings that the cerebellar interpositus nucleus integrates both vestibular and neck proprioceptive signal (32). In this regard, exercise with the Angel’s Wings device, by promoting relaxation of the upper trapezius muscle and correcting the posture of the cervico-dorsal spine (16), could help rebalance proprioceptive input from the neck muscles, thereby contributing to the improvement of the vestibular function. However, the lack of improvement in patients who performed the training with Angel’s Wings voluntarily, without a trainer and likely not at their full potential, suggests that the anatomo/physiological changes above described, may not be entirely responsible for the clinical/vestibular improvement.

The most plausible hypothesis is that, similarly to what has been demonstrated in other studies on patients with PD, it is the SE itself that led to the improvement (17, 18).

When the subjects performed an exercise with a trainer (that is a SE), the exercise rate is mechanically increased to help participants achieve and maintain a pace greater than their preferred voluntary exercise rate, requiring them to be actively engaged. The exercise we used corresponds to a forced exercise as the exercise was performed at a rate higher than what our patients would spontaneously consider their maximum, facilitated by continuous guidance from a trainer.

Our result regarding the improvement in the ability to utilize vestibular information after exercise does not represent the true novelty of this work, as this data has already been demonstrated in other studies (11, 12); rather, the novelty lies in the fact that the result was achieved through a SE performed with the upper limbs. In fact, almost all the data in the literature refers to forced exercise involving the lower limbs, particularly those performed with a modified bicycle.

Forced exercise may alter cortical excitability in patients with PD by increasing both the quantity (faster motor activity) and consistency (lower variability) of afferent information compared to VE (17). It has been proposed the forced exercise determines causes hyperstimulation of peripheral sensory receptors (such as joint receptors, muscle spindles, and Golgi tendon organs), enhancing the afferent sensory input to cortical and subcortical regions, leading to increased cortical activity and motor output (18, 19, 22). Indeed, a previous study utilizing the same upper limb exercise training demonstrated changes in the recruitment efficiency of the corticomotor pathway, accompanied by an improvement in the UPDRS-III score (33).

With respect to VE, engaging subjects in SE leads to the activation of a greater number of peripheral sensory receptors, primarily the Golgi tendon organs. This helps enhance sensory afferent neural activity, which in turn stimulates central processing of information to improve motor control and performance (19, 21).

It is important to note in this context that, in patients with PD undergoing forced exercise, neuroimaging studies have confirmed that this exercise can trigger adaptive/neuroplastic brain changes (23, 34). Cerebral blood flow analysis and functional connectivity studies have shown widespread post-exercise changes in activation patterns compared to baseline, involving the supplementary motor area, primary motor cortex, basal ganglia, thalamus, and cerebellum (23). The similarity in the pattern and extent of brain activation observed on fMRI between forced exercise and antiparkinsonian drugs suggests a potential shared underlying mechanism of action in the brain that contributes to symptomatic relief. Lastly, based on findings from animal studies, forced exercise may facilitate the release of neurotrophic factors such as GDNF or BDNF, which are thought to contribute to improved motor function (35). The lack of clinical improvement in the group of patients who underwent voluntary exercise (VE), despite achieving postural correction, might be explained by the absence of the aforementioned neuroplastic/neurotrophic changes.

A limitation of our study is the relatively small sample size; consequently, we cannot completely rule out selection bias, which may limit the generalizability of our findings.

Another limitation is the lack of long-term follow-up, which prevents us from assessing the duration of the observed effects. Finally, we used the original UPDRS scale; therefore, this should be taken into account when comparing our results with studies employing the MDS-UPDRS scale.

In conclusion, we have demonstrated that SE performed with the upper limbs while seated can improve the utilization of vestibular information and, consequently, could enhance motor performance in patients with PD. The same results were not achieved through the same exercise performed voluntarily. Active assistive SE performed with upper limbs, should therefore be considered a crucial treatment methodology for patients with PD, especially those with motor limitations in the lower limbs.