Capture of Longitudinal Change in Real-Life Walking in Cerebellar Ataxia Increases Patient Relevance and Effect Size

Abstract

Background: With disease-modifying drugs for degenerative ataxias on the horizon, ecologically valid measures of gait performance that can detect patient-relevant changes in trial-like time frames are highly warranted.

Objectives: In this 2-year longitudinal study, we aimed to unravel ataxic gait measures sensitive to longitudinal changes in patients' real lives using wearable sensors.

Methods: We assessed longitudinal gait changes of 26 participants with degenerative cerebellar disease (Scale for the Assessment and Rating of Ataxia [SARA]: 9.4 ± 4.1) at baseline, 1-year, and 2-year follow-up using three body-worn inertial sensors in two conditions: (1) laboratory-based walking (LBW); and (2) real-life walking (RLW). In RLW, a context-sensitive analysis was performed by selecting comparable walking bouts according to macroscopic gait characteristics. Gait analysis focused on measures of spatio-temporal variability, particularly stride length variability, lateral step deviation, and a compound measure of spatial variability (SPCmp).

Results: Gait variability measures showed high test-retest reliability in both walking conditions (intraclass correlation coefficient [ICC], ≥0.82). Cross-sectional analyses revealed high correlations of gait measures with ataxia severity (SARA, effect size ρ ≥ 0.75); and with patients' subjective balance confidence (Activity-specific Balance Confidence scale [ABC]: ρ ≥ 0.71). Although SARA showed longitudinal changes only after 2 years, the gait measure SPCmp revealed changes after 1 year with high effect size (r_prb = 0.80). Sample size estimation for the gait measure SPCmp showed a required cohort size of n = 42 participants (n = 38; spinocerebellar ataxias [SCA]_1/2/3 subgroup) to detect a 50% reduction in progression at 1 year with a hypothetical intervention, compared to n = 147 for SARA at 2 years.

Conclusions: Because of their ecological validity and larger effect sizes, real-life gait characteristics represent promising performance measures as outcomes for future treatment trials. © 2025 The Author(s). Movement Disorders published by Wiley Periodicals LLC on behalf of International Parkinson and Movement Disorder Society.

Keywords: biomarker; cerebellar ataxia; digital health; real‐life walking; wearable sensors.

FIG. 1

(A) Illustration of the mapping procedure of corresponding real‐life walking (RLW) trials at baseline (blue) and the 1‐year follow‐up (red). (B) Relationship between longitudinal differences (∆ = 1‐year follow‐up – baseline) differences in mean bout length and lateral step deviation for the degenerative cerebellar disease (DCD) participants (non‐matched: red; matched: blue). In the non‐matched condition, there is a significant correlation between the longitudinal difference in lateral step deviation and the difference in mean bout length (r = 0.59, P = 0.0034**). (C,D) Radar plots illustrating cross‐sectional and longitudinal differences on Scale for the Assessment and Rating of Ataxia, Activity‐specific Balance Confidence scale, and four gait parameters for the gait conditions laboratory‐based walking LBW (C) and RLW (D): speed, spatial variability compound measure (SPCmp), coronal range of motion variability (CorRoM_SD), and foot angle variability (ToeOutAngle_SD). Cross‐sectional differences can be seen by comparison of healthy controls at baseline (HC_BL) and the DCD group at baseline (DCD_BL). Given are average values for each group. Longitudinal progression can be seen comparing DCD participants at baseline (DCD_BL) and follow‐up (FU) assessments (DCD_FU1, DCD_FU2) as well as for healthy controls (HC_BL and HC_FU2). [Color figure can be viewed at wileyonlinelibrary.com]

FIG. 2

(A) Within‐subject changes between baseline and follow‐up assessments for the group of degenerative cerebellar disease (DCD) participants. Shown are longitudinal differences in Scale for the Assessment and Rating of Ataxia (SARA) (upper panel) and the spatial variability compound measure (SPCmp) (lower panels). For both measures, the left panel shows the change between baseline and 1‐year follow‐up (FU1), and the right panel the change between baseline and 2‐year follow‐up (FU2). In all panels, SARA scores of individual cerebellar participants are color‐coded. Black dotted line = mean change across all participants. Stars indicate significant differences between time points (* ≡ P < 0.05, ** ≡ P < 0.0083 Bonferroni‐corrected, *** ≡ P < 0.001). (B) Sample size estimates were made for future treatment trials that showed different levels of progression reduction for the various outcome measures: SARA as well as the gait measures SPCmp and stride length variability (StrideLcv) in the laboratory (LBW) and real life (RLW). The estimated number of participants per study arm is plotted against the hypothesized therapeutic effect on reducing 1‐year progression or 2‐year progression in DCD patients or the SCA_1/2/3 subgroup (SCA_1/2/3), respectively. Solid lines indicate 1‐year estimates, and dashed lines indicate 2‐year estimates. [Color figure can be viewed at wileyonlinelibrary.com]