Shan Pan (1), Lisa Stamp (2), Stephen Duffull (1)
(1) School of Pharmacy, University of Otago, Dunedin, New Zealand (2) Department of Medicine, University of Otago, Christchurch, New Zealand
Objectives: Rheumatoid arthritis (RA) is a chronic autoimmune disorder that is characterised by inflammation in joint synovial membrane [1]. For effective control of RA and hence the prevention of joint destruction, multiple drugs including the gold-standard treatment methotrexate (MTX) are used in routine clinical care. The objective of this study is to determine whether it is possible to delineate the treatment effect of MTX in the presence of other RA treatments under an observational cohort study using clinical trial simulation.
Methods: The Early Arthritis Observational Study at Christchurch Hospital was established in 2004. Data on treatment history and disease activity score in 28 joints (DAS28) were available for 146 patients. In total 14 RA treatments including MTX were used and categorised based on the onset of action (i.e. rapid, intermediate and delayed). Individual patients’ treatment history in the observational study was extracted using MATLAB® (R2013b) and the time course of DAS28 was simulated based on the extracted treatment history. It was assumed that individual DAS28 (DAS28ij) changed from the baseline (DAS28ø,i) due to both treatment effects (Eij) and the natural history of disease (NHij) considering the measurement error (εij), as given by
DAS28ij = DAS28ø,i – Eij + NHij + εij .
Here subscripts i and j represent the jth clinical visit in the ith individual. Eij is considered to be nonlinear and additive, and NHij is linear and symptomatic [2].
Model parameters were derived from the data, the literature and expert opinion. The simulation performance was evaluated using local polynomial regression (LOESS). Treatment effects of individual categories were explored under different study designs and were estimated through stochastic simulation-estimation in NONMEM® 7.2.
Results: The LOESS curves in both observed and simulated DAS28 profiles had similar trends after patients being stratified into high and low disease baseline groups. Under slow RA progression, parameters pertaining to RA treatment effects were precisely estimated. However, the rapid disease progression was found to confound the treatment effects irrespective of study design.
Conclusions: Under the observational study design, MTX treatment effect was able to be delineated when the RA disease progression was slow. In future, a more mechanistic approach using population kinetic-pharmacodynamic modelling may be developed to learn about the treatment effect and optimal doses of MTX.
References:
[1] Cohen S et al. Arthritis Rheum. 2002;46(3): 614-24.
[2] Holford N. Br J Clin Pharmacol. 2015;79(1): 18-27.
Reference: PAGE 25 (2016) Abstr 3694 [www.page-meeting.org/?abstract=3694]
Poster: Drug/Disease modeling - Other topics