Taneja A(1),Danhof M(1 ), Della Pasqua OE(1,2)
(1) Division of Pharmacology Leiden Academic Centre for Drug Research Leiden University The Netherlands (2) Clinical Pharmacology Modelling & Simulation GlaxoSmithKline Stockley Park UK.
Objectives: Pain response in clinical trials relies on the use of clinical scales of pain intensity and relief. In contrast to current practice, here we show how a model-based approach can be used to determine the most appropriate dose range in a Phase III trial in patients with rheumatoid arthritis. Using a new selective COX-2 inhibitor as paradigm compound we illustrate how prostaglandin inhibition can be correlated to pain relief (core set measures of rheumatoid arthritis).
Methods: Data from a phase IIb clinical study in 540 patients was available for the analysis. Nonlinear mixed effects modelling was used to characterise the pain response, as defined by the ACRN scale. Model development included the evaluation of placebo and drug effect terms. Â The use of a Weibull function in conjunction with an Imax model was found to describe the placebo and drug effects appropriately. To enable identification of patient subgroups during model building and evaluate a putative correlation between target inhibition and pain relief, patients were categorised into responders and non-responders, based on the magnitude of changes from baseline (i.e., >25% decrease in pain). These data were then linked to a previously published model of prostaglandin (PG) inhibition using a non-parametric smoothing function. Modelling and simulation was performed in NONMEM v7.2, whilst data manipulation, graphical and statistical summaries were performed in R.
Results: The longitudinal response model described the data adequately, with the number of responders increasing from 54 to 69% with ascending dose levels (i.e., from 0, 10, 35 to 50mg). Different potency (IC50) estimates were obtained for each subgroup, i.e. responders and non-responders). Â Target (>80%) inhibition was predicted to be reached at doses greater than 250mg, whilst the median ACRN drop >50% occurred at a dose >100mg.
Conclusions: The assumption of an underlying exposure-response model for the clinical changes in ACRN allowed us to describe the time course of pain response in rheumatoid arthritis patients and make inferences about the heterogeneity in response. Despite the challenges in sampling and analysing biomarkers in Phase III studies, our exercise illustrates how a model-based approach can be used to better substantiate the dose rationale of compounds for the treatment of chronic pain.
References:
[1] Pilla Reddy, V., Kozielska, M., Johnson, M., Vermeulen, A., de Greef, R., Liu, J., Groothuis, G.M., Danhof, M., and Proost, J.H. Structural models describing placebo treatment effects in schizophrenia and other neuropsychiatric disorders. Clin Pharmacokinet 50, 429-450.
[2] Dionne, R.A., Bartoshuk, L., Mogil, J., and Witter, J. (2005). Individual responder analyses for pain: does one pain scale fit all? Trends Pharmacol Sci 26, 125-130
Reference: PAGE 22 () Abstr 2813 [www.page-meeting.org/?abstract=2813]
Poster: Other Drug/Disease Modelling