Digital Gait Measures Capture 1‐Year Progression in Early‐Stage Spinocerebellar Ataxia Type 2

Abstract

BackgroundWith disease‐modifying drugs in reach for cerebellar ataxias, fine‐grained digital health measures are highly warranted to complement clinical and patient‐reported outcome measures in upcoming treatment trials and treatment monitoring. These measures need to demonstrate sensitivity to capture change, in particular in the early stages of the disease.ObjectiveOur aim is to unravel gait measures sensitive to longitudinal change in the—particularly trial‐relevant—early stage of spinocerebellar ataxia type 2 (SCA2).MethodsWe performed a multicenter longitudinal study with combined cross‐sectional and 1‐year interval longitudinal analysis in early‐stage SCA2 participants (n = 23, including nine pre‐ataxic expansion carriers; median, ATXN2 CAG repeat expansion 38 ± 2; median, Scale for the Assessment and Rating of Ataxia [SARA] score 4.8 ± 4.3). Gait was assessed using three wearable motion sensors during a 2‐minute walk, with analyses focused on gait measures of spatio‐temporal variability that have shown sensitivity to ataxia severity (eg, lateral step deviation).ResultsWe found significant changes for gait measures between baseline and 1‐year follow‐up with large effect sizes (lateral step deviation P = 0.0001, effect size rprb = 0.78), whereas the SARA score showed no change (P = 0.67). Sample size estimation indicates a required cohort size of n = 43 to detect a 50% reduction in natural progression. Test–retest reliability and minimal detectable change analysis confirm the accuracy of detecting 50% of the identified 1‐year change.ConclusionsGait measures assessed by wearable sensors can capture natural progression in early‐stage SCA2 within just 1 year—in contrast to a clinical ataxia outcome. Lateral step deviation represents a promising outcome measure for upcoming multicenter interventional trials, particularly in the early stages of cerebellar ataxia. © 2024 The Authors. Movement Disorders published by Wiley Periodicals LLC on behalf of International Parkinson and Movement Disorder Society.

Sensor-free motion registration and automated movement evaluation: Leveraging machine learning for clinical gait analysis in ataxia disorders

“…Obviously, but nevertheless important to note is, that the selection of stage-dependent features tailored to the respective severity level of a participant further improved the detectability of progressive longitudinal changes. This is in line with previous studies in SCA2 with body-worn sensors (Seemann et al, 2024). In addition to the superiority in detecting subtle and longitudinal changes, the automated analysis was more accurate in the reproduction of onsite ratings compared to a human baseline.…”

“…step width while simultaneously decreased step length, increased variability in the placement and general trajectories of the feet have been described either from sensor-based gait assessments or with a multi-camera system. (Buckley et al, 2018; Ilg and Timmann, 2013; Kadirvelu et al, 2023; Seemann et al, 2024; Serrao and Conte, 2018). While this needs particular prerequisites, video-based markerless motion capturing has been used to study gait disturbances on a rodent model of ataxia (Lang et al, 2020).…”