Andrijana Radivojevic, Martin Fink, Jean-Louis Steimer, Henning Schmidt
Novartis Pharma AG, Basel, Switzerland
Objectives: Population pharmacokinetic (popPK) modeling outlines a very significant area of Pharmacometrics, or Modeling & Simulation. PopPK activities need to be performed regularly during drug development – from first-in-human dose prediction until submission at the end of Phase III. Since such analyses currently can take a considerable amount of time, resources are bound and not readily available to support other aspects of model based drug development. Our aims were to examine how the process of building a popPK model can be supported in order to increase efficiency, to develop a user friendly implementation of a popPK toolbox as proof-of-concept, and to stimulate a scientific discussion about such an integrated approach.
Methods: Through an internal survey at Novartis Modeling&Simulation and review of published literature, the most typical popPK modeling scenarios (following the Pareto principle) were identified. Based on that information, a popPK workflow was defined, along with a standard dataset specification. A modular approach was chosen, allowing the modeler to use the workflow where possible and do additional assessments where needed.
Results: This investigation proposes a workflow for popPK analyses, including data standards, data exploratory and goodness-of-fit plots, model building, covariate search, model summary/comparison statistics. As a proof-of-concept, the workflow was implemented and documented within the user-friendly SBPOP Package [1]. It has already been tested internally at Novartis on analysis of Phase II and III data for several compounds.
Conclusions: While the literature offers guidelines on popPK model validation and reporting, recommendations on actual model building still remain limited. The given implementation of a step-wise popPK modeling workflow has been proven to bring a considerable gain in efficiency, but at the same time assuring compliance to regulatory and internal requirements, freeing the modeler from labor intensive repetitive coding tasks. The approach is easily extensible to sequential PKPD modeling. At the current stage, the discussed workflow and its implementation is meant to serve as a basis for scientific discussions on model building standards.
References:
[1] H. Schmidt. SBPOP Package: Efficient support for model based drug development – from mechanistic models to complex trial simulation, Poster, PAGE, 2013.
Reference: PAGE 22 () Abstr 2705 [www.page-meeting.org/?abstract=2705]
Poster: New Modelling Approaches