Disease Progression in Multiple System Atrophy: The Value of Clinical Cohorts with Long Follow-Up
Alexandra Foubert-Samier and Cécile Proust-Lima contributed equally to this work.
Relevant conflict of interest/financial disclosures: The authors report no conflict of interest in line with the content of the manuscript.
Financial disclosures of all authors (for the preceding 12 months): T.S., M.L., C.H., and C.P.-L. have nothing to disclose. M.F. received honoraria to speak from BIAL, AbbVie, Orkyn, Consultancies for Bial and LVL Médical. A.P.-L. received honoraria from Almylam and Biohaven. P.P. received honoraria from IKT Laboratory and grant from the French Research Agency. W.G.M. has received fees for editorial activities with Elsevier, and consultancy fees from Lundbeck, Biohaven, Roche, Alterity, Servier, Inhibikase, Takeda, and Teva. O.R. received honoraria for scientific advice from companies developing novel therapies of biomarkers for MSA, including Lundbeck, Neuralight, ONO Pharma, Servier, Takeda, UCB, and received scientific grants for MSA from the French Ministry of Health, the French MSA Association ARAMISE, the MSA coalition, Lundbeck, ONO Pharma, Servier, Takeda. A.F.-S. received honoraria from Aguettant Laboratory.
Funding agencies: This work was funded by the French National Research Agency (Project DyMES—ANR-18-C36-0004-01) and the region Nouvelle-Aquitaine as part of the ESR project call 2021.
In their recent publication, Kühnel et al1 described the progression of multiple system atrophy (MSA) in the European MSA study group (EMSA-SG) cohort via an innovative disease progression model (DPM). DPMs are valuable longitudinal methods to describe MSA natural history while accounting for data uncertainty (delayed diagnosis, uncertain timing, heterogeneous staging).2 The mean trajectories of clinical progression are described along the homogeneous disease continuum (Fig. 1C) rather than the observed time since diagnosis (Fig. 1A) thanks to a temporal recalibration of progression according to an individual latent disease time, anchored to MSA disease stage at inclusion.
The population characteristics and the length of individual follow-up are critical in natural history studies and DPMs. Kühnel's study relied on 121 patients with rather advanced stage, outdated diagnosis criteria, and short follow-up of 2 years.1, 3 We replicated Kühnel's analysis on repeated Unified MSA Rating Scale sum scores I (activities of daily living) and II (motor examination) from the French MSA cohort4 (663 patients) with maximum follow-up of 11 years, consensus diagnosis criteria,5 and early stages at entry (see Supplementary Material Data S1 for details). MSA progression spanned a larger period than in the original paper (Fig. 1C) with mean time gaps at inclusion estimated at 3.6 and 9.1 years for moderately-dependent and helpless patients at inclusion, respectively (Fig. 1B right panels); and significant inter-patient differences (SD = 2.14 years). When restricting the sample to 2.5-year follow-up, these differences were smaller, especially among the most aggressively affected patients at entry, with estimates of 2.5 and 6.6 years (Fig. 1B left panels) and smaller inter-patient differences (SD = 0.79 years). This suggests that studies restricted to short-term follow-up overestimate the progression rate and underestimate inter-patient differences.
When applying DPMs, differences across stages should be carefully interpreted. They do not quantify the expected amount of time spent in each stage by a patient, but the time gap between patients entering the study at different stages. Estimating the duration spent in each disability stage requires specific modeling of disability over time.
- Sigmoid shape: based on generalized logistic function, progression trajectories are restricted to sigmoids leading to suboptimal fit of the data compared to data-driven approaches (see Data S1);
- Conditional markers' independence: individual recalibration requires the assumption that the latent disease time captures all the correlation between markers, which may be unlikely in practice;
- Homogeneity of progression: DPMs assume a unique mean profile of progression when subphenotypes of clinical progression may exist4;
-
Non-informative death: death is assumed to be predictable by the observed course of the markers when death caused by MSA may induce a more informative dropout to be jointly modeled.6
In conclusion, DPM constitutes a promising tool for disease study, but it needs to be interpreted with caution and calls for less stringent assumptions. The replication on the French MSA cohort highlights the importance of describing MSA progression based on long-term follow-up data and large cohorts to prevent too pessimistic projections and underestimation of sample sizes.
Acknowledgments
We would like to thank the French National Research Agency (Project DyMES-ANR-18-C36-0004-01) and the Nouvelle-Aquitaine region (Project AAPR2021A-2020-11937310) for their financial support that made this work possible. Several authors of this publication are members of the European Reference Network for Rare Neurological Diseases—Project ID no: 739510.
Author Roles
- 1 Research project: A. Conception, B. Organization, C. Execution;
- 2 Statistical analysis: A. Design, B. Execution, C. Review and critique;
- 3 Manuscript preparation: A. Writing of the first draft, B. Review and critique;
T.S.: 1A, 1B, 1C, 2A, 2B, 3A
M.F.: 2C, 3B
A.P.L.: 2C, 3B
M.L.: 2C, 3B
C.H.: 2C, 3B
P.P.: 2C, 3B
W.G.M.: 2C, 3B
O.R.: 2C, 3B
A.F.S.: 1A, 1B, 2A, 2C, 3A
C.P.L.: 1A, 1B, 2A, 2C, 3A
Open Research
Data Availability Statement
Research data are not shared.
