Shaonan Wang (1), Chiara Piana (1), Vincent Dubois (1), Muriel Boulton (1), Roberta Bursi (1)
(1) Grünenthal GmbH, Aachen, Germany
Objectives: The aim of this study was to explore joint modelling of pain intensity and informative dropout data from escalating doses of a reference drug and placebo in patients with moderate to severe chronic pain.
Methods: Nonlinear mixed effect modelling (NONMEM) was used to describe exposures of the reference drug and their relationship to average daily pain intensity measured using Numerical Rating Scale change from baseline (NRSCHB). A linear correlation between NRSCHB and probability of dropout was introduced into a time to event model in order to link efficacy and time to dropout [1]. Joint modelling of NRSCHB and dropout was simultaneously performed using NONMEM 7.2. Posterior predictive check (PPC) for NRSCHB was generated by simultaneously simulating NRSCHB and time to dropout. Once a subject was simulated to reach the time to dropout, no further NRSCHB was simulated in that subject
Results: A concentration-NRSCHB relationship was identified for the reference drug and was best described by an Emax function. A Weibull model was used to describe the probability of dropout, where the hazard function was dependent on treatment, titration and subject visit time. The PPC for NRSCHB obtained from the joint model with informative dropout showed a better fitting to observations compared to the PPC obtained from the NRSCHB model alone without dropout.
Conclusions: A joint model of NRSCHB and time to dropout was developed and adequately described the observed NRSCHB and Kaplan-Meier curve. By assuming an Emax exposure-NRSCHB relationship and by taking into account informative dropout, the joint model was capable of providing a realistic prediction of this trial as well as an opportunity to simulate future trials.
References:
[1] Bjornsson, M.A. et al, Br J Clin Pharmacol 71:6, 899-906
Reference: PAGE 25 (2016) Abstr 5790 [www.page-meeting.org/?abstract=5790]
Poster: Drug/Disease modeling - Other topics