Quantitative Gait and Balance Outcomes for Ataxia Trials: Consensus Recommendations by the Ataxia Global Initiative Working Group on Digital-Motor Biomarkers

Abstract

With disease-modifying drugs on the horizon for degenerative ataxias, ecologically valid, finely granulated, digital health measures are highly warranted to augment clinical and patient-reported outcome measures. Gait and balance disturbances most often present as the first signs of degenerative cerebellar ataxia and are the most reported disabling features in disease progression. Thus, digital gait and balance measures constitute promising and relevant performance outcomes for clinical trials.This narrative review with embedded consensus will describe evidence for the sensitivity of digital gait and balance measures for evaluating ataxia severity and progression, propose a consensus protocol for establishing gait and balance metrics in natural history studies and clinical trials, and discuss relevant issues for their use as performance outcomes.

Keywords: Cerebellar ataxia; Digital motor performance marker; Gait and posture.

Fig. 1

(A) Area under the ROC (receiver operating characteristic) Curve (AUC) in descending order for each gait measure discriminating people with spinocerebellar ataxia (SCA) from healthy controls (HC). (B) Pearson correlation of the four most discriminative gait measures with clinical SARA scores related to the ataxia severity of each subtype of SCA 1,2,3 and 6 (adapted from [21])

Fig. 2

Correlations between the step length coefficient of variation (CV) and the falls/year (A) and SARA scores (B) in 17 ataxic participants. Pearson’s R coefficient (R) and significance (p) are reported (adapted from [74])

Fig. 3

Recording of sway path (SP) in anteroposterior and lateral direction and the calculated sway direction histogram (SDH). (A) Normal subject. (B) Predominantly lateral sway and very large sway eyes closed, in a patient with Friedreich’s ataxia. Adapted from [80]

Fig. 4

(A) Representative examples of representative statokinesiograms (postural sway path) during a 30-s, feet-together, eyes-open stance in a healthy control individual, an individual with pre-ataxic SCA6 and an individual with manifest SCA6. (B) Both sway ellipse area and sway mean velocity are correlated with severity of ataxia, as measured by the SARA in SCA 1,2,3 and 6 [85]

Fig. 5

Postural sway abnormalities in pre-ataxic SCA2 participants (Pre-SCA2) in comparison to healthy participants (HC) for a stance task with feet together and for tandem stance. Shown are stance measures jerks (A) and Path Length (B). ns: P > 0.0013 (after Bonferroni correction); **, P < 0.005; ***, P < 0.0005; Adapted from [23]

Fig. 6

Relationship between body sway and estimated time to disease onset for pre-ataxic mutation carriers in different stance tasks. Shown are relationships for genetically-based estimates of onset according to [97]. Each circle represents one participant. Body sway (length of sway path) was determined in three different stance conditions: (A) feet closed (Romberg test, RB) and eyes open; (B) feet closed (Romberg) and eyes closed; (C) feet closed (Romberg test, RB) and eyes closed on a foam cushion (mattress). P-values indicate significant correlations between durations to estimated disease onset and body sway. Reprinted with permission from [20]

Fig. 7

Differences between subgroups of participants with cerebellar ataxia (CA) stratified according to gait and posture ataxia severity as determined by the SARA_p&g subscore [reprinted by permission from [36]]. Subgroups: CA_Mild: SARA_g&p [0:2], CA_Mod: SARA_g&p= [3–4], CA_Sev: SARA_g&p [5–6]. Shown are group differences for constrained lab-based walking and real-life walking). LatStepDev and the compound measure of spatial variability were sensitive in distinguishing these severity subgroups also during real-life walking

Fig. 8

Illustration of different ways to show sensitivity to changes in ataxia severity. In most studies, cross-sectional analysis (blue) has been performed to show sensitivity to ataxia severity by correlations of balance and gait digital measures with clinical ataxia scores like the SARA, the FARS or the number of falls. These correlations with clinical ataxia scores are strongly influenced by the range of disease severity (range of observations [122]). Longitudinal (red): To serve as valid performance measure in ataxia intervention trials, these gait measures need to prove their sensitivity to individual longitudinal change over short time-spans (e.g. 1 year). In addition, the target population in clinical trials will most likely not encompass the full range of disease severity, but will be limited to, for example, mild-to-moderate disease

Fig. 9

(A) Longitudinal analyses of 1-year follow-up assessments: Within-subject changes between baseline and 1-year follow-up for a SCA 3 group. Upper panel: Within-subject changes in the SARA score and the gait measures of lateral sway and Stride length CV in the slow walking condition from baseline (BL) at the 1-year follow-up (FU). Lower panel: Within-subject changes between baseline and 1-year follow-up represented as delta (∆). In all panels, SARA scores of individual participants with cerebellar ataxia are colour coded. Black dotted line = mean change across all participants. The stars indicate significant differences between timepoints (*≡ p<0.05, **≡ p<0.0083 Bonferroni-corrected, ***≡ p<0.001). Effect sizes r_prb were determined by matched-pairs rank biserial correlation. (B) Sample size estimations were performed for future intervention trials showing different levels of reduction in progression levels for the different outcome measures: SARA, lateral sway and stride length variability in the walking conditions with preferred and slow speed. The estimated number of participants per study arm is plotted over the assumed therapeutic effect for lowering the 1-year progression in SCA3 (reprinted from [25] with permission)

Fig. 10

Examples of how a variety of concepts of interest cascade from a single meaningful aspect of health across select conditions and clinical populations (adapted from [182])