Niels Hendrickx (1), Alzahra Hamdan (2), Andrew C. Hooker (2), Xiaomei Chen (2), Rebecca Schüle (3,4, 5), ARCA study group, Andreas Traschütz (3,4), Matthis Synofzik (3,4), Mats O. Karlsson (2), France Mentré (1), Emmanuelle Comets (1,6) and the EVIDENCE-RND consortium
(1) Université Paris Cité, IAME, Inserm, F-75018, Paris, France. (2) Pharmacometrics Research Group, Department of Pharmacy, Uppsala University, Uppsala, Sweden (3) Division Translational Genomics of Neurodegenerative Diseases, Hertie Institute for Clinical Brain Research (HIH), University of Tübingen, Tübingen, Germany; (4) German Center for Neurodegenerative Diseases (DZNE), Tübingen, Germany. (5) Division of Neurodegenerative Diseases, Department of Neurology, Heidelberg University Hospital, Germany (6) Univ Rennes, Inserm, EHESP, Irset - UMR_S 1085, 35000, Rennes, France.
Objectives:
Genetic cerebellar ataxias are progressive rare neurological diseases affecting the cerebellum, often with multi-systemic damage to other neurological systems, causing debilitating impairment of gait, balance, speech, and fine motor skills. More than 100 ataxia diseases are autosomal-recessive cerebellar ataxias (ARCAs). Patient’s disease severity is evaluated with the Scale for the Assessment and Rating of Ataxia (SARA) score, a composite clinical score (0-40), which can be modelled using total score (TS) [1], or through Item Response Theory (IRT) models [2]. There are currently no disease modifying drugs for most ARCAs, and promising treatment trials with robust designs are needed [3]. To guide optimal trial design for disease-modifying treatments in rare neurological diseases, this work aimed to study the influence of the choice of model (TS- vs IRT-based), inclusion criteria and design on the power and type 1 error of simulated clinical treatment trials for a hypothetical disease modifying effect. To benefit from a trial, patients with neurodegenerative diseases should ideally be enrolled during the early or mid-stages of their disease [4]. However, the rate of progression is slow and varies with the time since symptom onset (TSO) [1] (0.1-1 point/year). Thus, TSO is also investigated as one of the inclusion criteria.
Methods:
We use a four-parameter logistic model (TS) [1] fit on 173 patients with a specific genetic ataxia – Autosomal-Recessive Spastic Ataxia Charlevoix Saguenay (ARSACS) from the ARCA registry [5], describing the SARA score versus TSO. This model was modified to fit a drug effect slowing the disease progression rate. The second simulation model is a longitudinal IRT model fit on the 173 ARSACS patients [2].
First, inclusion criteria based on TSO at the start of the study were investigated with a parallel design: early (0-10 years), mid-stage (10-20 years), late (20-30 years) and heterogeneous (0-30 years) inclusion. Datasets were simulated under the TS and the IRT model with a drug effect parameter slowing the disease progression rate by 0% (for type 1 error) and 50% (for power). Second, other designs were investigated with heterogeneous inclusion, simulated using the TS model with the same drug effects: a cross-over design [6], and a delayed start design [7]. In each scenario, 500 trials were simulated, with 100 patients in each, with 1:1 allocation, a trial duration of 5 years with a measurement every 6 months. The simulated datasets were analysed with the simulation model (IRT or TS), a linear model and the TS model with a fixed amplitude parameter. The empirical type 1 error was evaluated as well as the power corrected by choosing the threshold for statistical significance such that the type I error is 5% for that threshold.
Parameters were estimated with the SAEM or Laplace method using NONMEM 7.5.1.
Results:
For the parallel design, when simulating under the TS model, the type 1 error was controlled in every inclusion criterion, and inflated for the linear model in late and heterogeneous inclusion. When simulating under the IRT model, the type 1 error was controlled for the IRT model but inflated for the TS model. In both cases, the type 1 error inflation could be due to model misspecification. In both simulations, the power was the highest for the late and heterogeneous inclusion (resp. 77%, 80% for IRT, simulated under IRT and 98%, 88% for TS, simulated under TS). In early inclusion, the power when simulating under IRT was higher than when simulated under TS (43% for the TS, 72% for the IRT vs 17% when simulating and estimating under TS). The other designs showed controlled type 1 error but lower power (57% for cross-over and 74% for delayed start design).
Conclusions:
For a parallel design, the late and heterogeneous inclusion criterion yielded the highest power. Power and type I error were highly dependent on the simulation model and the lack/presence of any misspecifications in the analysis model. Crossover and delayed start designs were shown to be less powerful than a parallel design, since the disease progression is slow and observed for a shorter duration under treatment or placebo (compared to a parallel design), so the power is more sensitive to duration than to sample size.
Acknowledgements
This work was supported by the European Joint Programme on Rare Diseases (EJPRD) (Grant Agreement n°825575) WP20 Innovation Statistics project “EVIDENCE-RND” (to F.M. R.S, and M.S), the EJPRD PROSPAX consortium, further supported by the Clinician Scientist program “PRECISE.net” (to A.T, M.S. and R.S.). We also thank Lionel de la Tribouille for the use of the CATIBioMed calculus facility.
References:
[1] Hendrickx N, Mentré F, Schüle R, Gagnon C, ARCA study group, Evidence-RND consortium et al. . Predicting individual disease progression including parameter uncertainty in rare neurodegenerative diseases: the example of Autosomal-Recessive Spastic Ataxia Charlevoix Saguenay (ARSACS), PAGE 31 (2023) Abstr 10518 [www.page-meeting.org/?abstract=10518]
[2] Hamdan A, Hooker AC, Chen X, Traschütz A, Schüle R, Synofzik M, et al. Item Response Theory Analysis of the Scale for the Assessment and Rating of Ataxia in Autosomal Recessive Cerebellar Ataxias [Internet]. Abstr 10626; 2023; PAGE 31. Available from: www.page-meeting.org/?abstract=10626
[3] Salem IH, Beaudin M, Klein CJ, Dupré N. Treatment and Management of Autosomal Recessive Cerebellar Ataxias: Current Advances and Future Perspectives. CNS Neurol Disord Drug Targets. 2023;22(5):678-697. doi:10.2174/1871527321666220418114846
[4] Benatar M, Wuu J, McHutchison C, Postuma RB, Boeve BF, Petersen R, et al. Preventing amyotrophic lateral sclerosis: insights from pre-symptomatic neurodegenerative diseases. Brain. 2022 Jan 1;145(1):27–44.
[5] Traschütz A, Reich S, Adarmes AD, Anheim M, Ashrafi MR, Baets J, et al. The ARCA Registry: A Collaborative Global Platform for Advancing Trial Readiness in Autosomal Recessive Cerebellar Ataxias. Front Neurol. 2021;12:677551.
[6] Sibbald B, Roberts C. Understanding controlled trials. Crossover trials. BMJ. 1998 Jun 6;316(7146):1719. doi: 10.1136/bmj.316.7146.1719. PMID: 9614025; PMCID: PMC1113275.
[7] Wang D, Robieson W, Zhao J, Wiener C, Koch G. Statistical considerations in a delayed-start design to demonstrate disease modification effect in neurodegenerative disorders. Pharm Stat. 2019;18(4):407-419. doi:10.1002/pst.1931
Reference: PAGE 32 (2024) Abstr 11057 [www.page-meeting.org/?abstract=11057]
Poster: Methodology - Other topics