Combining digital outcomes of mobility with mobile brain imaging to predict future Parkinson’s disease

Introduction: Neurodegeneration represents a major health and economic crisis worldwide. Yet, transitioning from healthy brain aging to neurodegeneration is a prolonged process, including subtle pre-clinical features that, if identified, could lead to earlier diagnostic certainty. Individuals with isolated REM sleep Behaviour Disorder (iRBD) (i.e., those who enact their dreams) are 130 times more likely to eventually develop a synucleinopathy (i.e., approximately 50 % develop Parkinson’s disease (PD) versus dementia with Lewy bodies (DLB)) over a 10-15 year span, making iRBD the best current prodromal marker for predicting future neurodegeneration. However, iRBD alone does not reveal when a diagnosis can be made; individuals may be diagnosed in 5 or 15 years (Postuma et al., 2012). Whilst simple motor deficits in iRBD have been argued as one of the best predictors in iRBD for developing PD (Postuma et al., 2019), most of the evidence to date has evaluated walking speed, while neglecting more sensitive assessments such as postural stability, gait variability, gait asymmetry, and turning which are known to be impacted by pathology in brainstem regions. Recent work suggests that quantitative balance, gait and turning outcomes may be even more sensitive (Alibiglou et al., 2016; Ehgoetz Martens et al., 2019, 2020, 2022; MacKinnon et al., 2019; McDade et al., 2013). In fact, subtle balance and gait deficits can be exposed by challenging the motor system by manipulating sensory feedback or adding a concurrent cognitive task (dual task) (Ehgoetz Martens et al., 2019), which can interfere with compensation from prefrontal cognitive control networks often masking early motor signs due to declining motor automaticity. Quantitative, continuous monitoring with wearable technologies shows tremendous promise as well, however, the field is in its infancy since only a handful of studies have examined balance and gait deficits in iRBD using quantitative approaches beyond a stopwatch, and even fewer that have manipulated cognitive or sensory load or attempted to monitor individuals at home. Thus, a comprehensive set of sensitive and robust mobility assessments needs to be tested to reveal the optimal assessment that unmasks unique and subtle motor deficits in iRBD and determine their predictive utility for phenoconversion from iRBD to either PD or DLB. Only then can recommendations be made for the design of clinical trials to include sensitive motor assessments that best monitor and predict prodromal disease progression.Examining behaviour, however, does not always reveal the full picture, since the brain is well equipped to adapt and compensate as it undergoes progressive neurodegeneration. Several insights can be gained from examining the neural correlates of motor performance in early prodromal populations like iRBD. Functional near-infrared spectroscopy (fNIRS), an advanced technique to measure oxygenated hemoglobin levels (HbO2), can tolerate bodily movements, and is a non-invasive technique. This makes fNIRS suitable for both healthy and clinical populations to enable the assessment of changes in brain activation during gait to estimate cortical hemodynamic responses that occur in response to neural activity. While understanding the neural basis that underpins balance and gait deficits during the prodrome provides added insight into the neurodegenerative process, it is also instrumental for interpretating behaviour and making inferences for predictions for future disease progression.The objectives of this research are 1) to evaluate and optimize a comprehensive set of quantitative mobility assessments by comparing the performance outcomes in participants with iRBD to that of healthy control participants (HC); and 2) to understand early compensation related to prodromal neurodegeneration by evaluating cortical activation during motor performance using mobile neuroimaging in iRBD compared to PD and HCs.Methods: To evaluate objectives 1 and 2, 20 iRBD (67.8 years old), 27 age-matched HC participants (69.0 years old) were assessed in the clinic using wearable inertial measurement units (IMUs). After being outfitted with sensors, each participant underwent a motor and cognitive examination (UPDRS; MoCA). The participant was outfitted with a portable fNIRS system to measure HbO2 levels in the prefrontal cortex (PFC) while completing a comprehensive battery of validated balance and gait assessments. The clinical mobility assessment battery consisted of: (i) 30 second quiet stance with eyes open; (ii) 30 second quiet stance on a foam surface; (iii) 60 second alternating 360° turns test; (iv) 60 second alternating 360° turns test while performing the forward digit span (FDS, attention/working memory); (v) a self-paced 2 minute walk; (vi) a self-paced 2 minute walk while performing the forward digit span; (vii) a 2 minute dual-task walk while performing the oral Trail Making Test Part B (DT-TMB, executive function); (viii) a 2 minute dual-task walk while performing the auditory clock test (DT-AC, visuospatial); (ix) MiniBest test. A seated assessment of the forward digit span, Trail Making Test Part B, and auditory clock test were also completed to calculate dual-task interference. Participants also completed several questionnaires (Mayo sleep questionnaire, cognitive reserve index, life space assessment, movement specific reinvestment scale, hospital anxiety and depression scale, falls efficacy scale, activity-specific balance confidence scale). Different dual task walking trials were used to evaluate interference with attention, executive function, and visuospatial cognition. This cohort was also part of a larger sampler (23 iRBD, 61 HC) that completed remote continuous monitoring of gait over 7 days, where participants wore three IMUs (lumbar spine, bilateral feet) to collect measures of gait and turning during daily life. Postural sway outcomes of balance, spatiotemporal outcomes of gait, and characteristics of turning were examined during the clinical mobility assessment and free-living continuous monitoring assessment. Linear mixed models were used to assess whether the measures of interest differed between groups and conditions (for the clinical assessment), with age as a covariate.Results: Mediolateral jerk was significantly greater in the iRBD group than the healthy control group in the foam condition only (p = .047). Stride length variability was significantly greater in the DT-TMB compared to the single task condition in the iRBD group (p = .014) but not in the HC group (p = .12). Arm swing was lower (p = .041), turn duration (p = .0047) was longer, and turn speed (p = .041) was slower in the iRBD group than the HC group irrespective of walking condition. For other measures (gait speed, gait speed variability, step time, step time variability, swing time, swing time variability), gait worsened in the DT-TMB compared to all other conditions in both groups. In the continuous monitoring assessment, gait speed (p = .010) was slower, step time (p = .040) was longer, swing time (p = .00017) was lower, swing time variability (p = .018) was higher, turn duration (p = .045) was longer, and turn speed (p = .019) was slower in the iRBD group compared to the HC group. Analysis of fNIRS is currently ongoing.Conclusions: Early mobility deficits in iRBD require challenging the sensory and/or cognitive demands to elucidate differences from HCs. The oral trail making dual-task had the largest effect on gait variability in iRBD. In addition, examining quantitative mobility outcomes such as arm swing, turning duration along with monitoring free living mobility revealed novel group differences. Further analysis of PFC activity across mobility assessment conditions will provide insight as to the degree of compensation required to achieve comparable performance to HCs across various mobility tasks.