Early-phase trials of disease-modifying therapies in Parkinson’s disease (PD) typically use the MDS-UPDRS Part III motor score as the primary outcome for motor function. The authors note that this endpoint often necessitates either large sample sizes or relatively long treatment durations (commonly 12–18 months), even where substantial treatment effects are assumed. The principal reason is that early PD progression is relatively slow compared with the variability inherent in MDS-UPDRS Part III assessments, which reduces the endpoint’s responsiveness to change over typical trial intervals.
Sample size calculations and responsiveness estimates were derived from publicly available placebo-arm summary statistics from four recent, large Phase 2 trials of early-stage PD: SPARK, PASADENA, PADOVA and LUMA. The manuscript cites ClinicalTrials.gov disclosures and peer-reviewed or conference sources associated with each trial, including NEJM reports for SPARK and PASADENA, an npj Digital Medicine report for a related score used in PASADENA, a Lancet publication for PADOVA, and a conference poster for LUMA. All extracted values are reported in the manuscript tables and supplementary material. The analysis used only aggregate summary data; no individual participant data were accessed and no new data collection was performed.
The investigators implemented a responsiveness-based approach to compare the performance of the MDS-UPDRS Part III with alternative sensor-derived endpoints. Using the placebo-arm progression observed in the cited trials, they estimated the sample size required to detect a range of assumed proportional treatment effects on MDS-UPDRS Part III progression. They then applied a relative-responsiveness framework to translate those estimates to alternative endpoints, allowing direct comparison of projected sample sizes under the assumption that proportional treatment effects on progression are equivalent across endpoints.
The alternative endpoints evaluated were smartphone- and sensor-derived Motor Functional Composites. Two variants are described:
These composites were developed by Indivi AG, the employer/contractor organization for all authors, and form the basis of the comparative responsiveness and sample size analyses.
Applying the responsiveness-based calculations to the placebo-arm data from SPARK, PASADENA, PADOVA and LUMA, the authors report that powering a one-year trial using MFC 1 or MFC 2 was projected to reduce the required sample size by more than half compared with trials using MDS-UPDRS Part III as the primary endpoint, assuming equivalent proportional treatment effects. The reduction is presented as a projection based on relative responsiveness; specific numeric sample size figures and the underlying extracted values are reported in the manuscript tables (Table I and Supplementary Table S1).
The authors emphasize that the MDS-UPDRS Part III’s slower responsiveness in early PD, driven by slow disease progression relative to measurement variability, underlies the larger sample size or longer durations typically required when that scale is used in isolation.
The sample size calculations and the responsiveness-based translation method were implemented in a publicly accessible, interactive application. The manuscript provides the URL for the application, allowing users to explore projected sample sizes under different assumptions and to apply the approach to alternative parameter inputs. All values extracted from the cited trial sources that were used in the analyses are reported in the manuscript tables and supplementary materials.
The work was funded by Indivi AG. All authors are employees or contractors of Indivi AG, which developed the Motor Functional Composites (MFC 1 and MFC 2) and the sample size calculator application evaluated in the paper; this relationship is disclosed as a competing interest. The analysis used only publicly available aggregate summary statistics; institutional review board or ethics approvals specific to new data collection were not required because no individual-level data were accessed. The manuscript lists the trial registrations and source references for the aggregate data used.
Note: Specific numeric sample size estimates and the detailed extracted values are reported in the manuscript (Table I and Supplementary Table S1) and are accessible via the paper and the interactive application; those numeric details are not restated here beyond the projections described in the source.