Garrit Jentsch, Rupert Austin, Aaron Hayman, Fernando Ortega, Helen Walker and Joachim Grevel
BAST Inc Ltd, Loughborough, UK
Objectives: Recent years have witnessed the emergence of various Model Based Drug Development (MBDD) strategies all of which centre around a continuum of mathematical statistical models that capture the developing project from the patho-physiology of the disease and the pharmacology of the drug to the conduct of clinical studies. The model continuum is maintained to simulate experiments and clinical studies in order to optimise their design and to quantitatively assess their chance of success.
Methods: Computational frameworks supporting a MBDD strategy are very rare. We have started to develop an agent-based simulation environment that allows us to integrate the model continuum in a single framework. The first prototype evaluates the operating characteristics of a simulated trial. The underlying statistical models are formulated in NMTRAN (ADVAN13) but simulations are executed without the use of NONMEM.
Results: Two case studies are presented. The first study demonstrates the use of the clinical trial simulator (CTS) from the perspective of a payer assessing a new treatment by evaluating whether a randomly selected subject from an active treatment group experiences a response that marks an improvement over the response of a randomly selected subject from a control group. The target is set at improvement in 85% of the cases. A biomarker indicative of disease severity is simulated, and a probability of achieving the target value (PTV) of 0.9 is determined.
The second case study deals with the development of a new drug against a disease whose severity is measured by a composite score. Using data of a phase II trial the CTS is used to identify the dose level and observation period that should be used in a confirmatory trial. Treatment is successful, if 50% of the patients exhibit a disease score lying below a certain threshold. The outcome of each simulated trial is evaluated with a proportion z-test at a significance level of 0.05. Simulations reveal that only one of the proposed dose levels leads to a PTV > 0.8. In order for the probability of success of the trial to approach the PTV more than 500 subjects would need to be studied.
Conclusions: The case studies demonstrate how the CTS evaluates the operating characteristics and success chances of a clinical trial and how it guides study design optimisation.
Reference: PAGE 23 (2014) Abstr 3047 [www.page-meeting.org/?abstract=3047]
Poster: Methodology - Other topics