2021 - Online - In the cloud

PAGE 2021: Methodology - New Modelling Approaches
Moustafa M. A. Ibrahim

Enhancing Efficiency and Decision Quality in Parkinson’s Disease Drug Trials for LRRK2 Programs Using Item Response Modelling

Moustafa M. A. Ibrahim (1), Elodie L. Plan (1), Siv Jönsson (1), Chao Chen (2), Mats O. Karlsson (1)

(1) Pharmetheus (2) GlaxoSmithKline PLC

Background and Objectives: Owing to the slow and variable progression of Parkinson’s disease (PD) symptoms, clinical trials aiming at identifying drug effects can be long and costly. The aim of this work was to develop a framework enhancing efficiency in PD drug effect characterization, through prior evaluation of the study design, for trials in patients with the LRRK2 gene mutation, characterized by a slow disease progression (DP).

Methods: Data were obtained from PPMI [1] and included observations up to five years, from the Genetic PD, the Genetic Registry PD, and the DeNovo PD cohorts. Individual, longitudinal, item-level scores of Part I (non-motor experiences of daily living), Part II (motor experiences of daily living), and Part III (motor examinations) from the MDS-UPDRS [2], a PD assessment instrument, were extracted.
An Item Response Theory (IRT) model was built. First, item characteristic curves (ICCs) were estimated, treating the baseline visit as the reference, and all other visits as new individuals. Subsequently, with ICCs fixed to their estimated values, data reassigned per individual, and using estimated latent variable (LV) values as dependent variable, a longitudinal DP model was developed for each of the three parts. Finally, covariate effects from factors like age, sex, cohort, time from diagnosis, and LRRK2 mutation status, were evaluated on the LV parameters, using stepwise covariate model building procedure, while handling missingness in LRRK2 through a mixture probability.
A 2-year delayed-start design [3] was simulated. 1000 early diagnosed virtual patients, presenting the LRRK2 gene mutation, were equally assigned to either an early start arm (active treatment all along) or a delayed start arm (placebo during 1 year, followed by active treatment). Hypothetical placebo, symptomatic and disease-modifying drug effects were implemented after mapping their parameters from item score scale to their corresponding LV estimates [4,5]. Series of simulations (n=500) were performed. Endpoints of interest were the LV and the total score in each of the 2 arms at end of phase 1 (week 52) and end of phase 2 (week 104).

Results: Data from 895 patients with a baseline MDS-UPDRS of 19, 1-74 (median, 5th-95th percentiles) were analysed. The DP model described jointly the data from the three parts of the MDS-UPDRS scale in the three cohorts of the database. Transforming the non-tremor items in part III from left-side/right-side tests into best-side/worst-side tests resulted in a better description of the data. A linear progression of the LV for each part was characterized. While identified predictors for the baselines were disease duration for all parts, and age for part III baseline, the factor found to affect the slopes of the three parts was LRRK2 gene mutation. Visual predictive checks of the final model adequately mimicked the time course of the three parts. Simulations displayed an initial reversible placebo effect in both arms at the beginning of each phase, with an alleviated disease progression in the early start arm already in phase 1 and in the delayed start arm only in phase 2. The added noise between the theoretical LV profiles and the corresponding total score could be visualized.

Conclusions: We successfully developed a disease model for PD subjects with LRRK2 gene mutation. A novel design was simulated with hypothetical placebo and drug effects. This framework allows the estimation of the simulated scores with relevant analyses methods, with widespread applicability in evaluating the impact of different study design features, e.g., study duration, population, sample size, interim analyses, and assessment schedules.



References:
[1] The Parkinson’s Progression Markers Initiative (PPMI) database (www.ppmi-info.org/data). For up-to-date information on the study, visit www.ppmi-info.org.
[2] Goetz CG, Tilley BC, Shaftman SR, Stebbins GT, Fahn S, Martinez-Martin P, Poewe W, Sampaio C, Stern MB, Dodel R and et al. Movement disorder society-sponsored revision of the unified parkinsons disease rating scale (mds-updrs): Scale presentation and clinimetric testing results. Movement Disorders. 2008, 23: 2129–2170.
[3] Rascol O, Fitzer-Attas CJ, Hauser R, Jankovic J, Lang A, Langston JW, Melamed E, Poewe W, Stocchi F, Tolosa E, Eyal E, Weiss YM and Olanow CW. A double-blind, delayed-start trial of rasagiline in Parkinson’s disease (the ADAGIO study): prespecified and post-hoc analyses of the need for additional therapies, changes in UPDRS scores, and non-motor outcomes. Lancet Neurol 2011, 10: 415–423.
[4] Sheng, Y., Zhou, X., Yang, S., Ma, P., & Chen, C. Modelling item scores of Unified Parkinson's Disease rating scale part III for greater Trial efficiency. Br J Clin Pharmacol. 2021 Feb 12. Epub ahead of print.
[5] Wellhagen, G.J., Karlsson, M.O. & Kjellsson, M.C. Comparison of Precision and Accuracy of Five Methods to Analyse Total Score Data. AAPS J 2021, 23:9.


Reference: PAGE 29 (2021) Abstr 9857 [www.page-meeting.org/?abstract=9857]
Poster: Methodology - New Modelling Approaches
Click to open PDF poster/presentation (click to open)
Top