W 04: A Value-Driven Decision Making for Drug Development Strategy

Poster Presenter

Masanori Ito

Senior Director, Head of Advanced Analytics, Advanced Informatics and Analytics
Astellas Pharma Inc.
Japan

Objectives

I focus on optimizing the drug development program by quantitative trade-off analysis. We show how to choose the good strategy (e.g. one indication first vs. multiple indications at a time) and to optimize the study designs at each phase to maximize the expected net present value (eNPV).

Method

ORAL PRESENTATION SCHEDULED: Session 2A at 9:40 - 9:50

Patel et al. (2013) proposed the value-driven framework to optimize sample sizes and trial schedules for Phase III. I propose to optimize the whole development program in one compound by minimizing the risk and maximizing the eNPV from much more macro viewpoint.

Results

I presented a method to compute the value of different scenarios while considering the uncertainty into the parameters of drug development based on the various scenario simulations. Best scenario was selected based on the estimated PoS (probability of success) and eNPV. Simulation results showed that the good strategy has changed depending on the settings of success probability, cost, and net present value for a new compound in our simulation. The sensitivity analyses clarified which factors have impact on the eNPV in each setting. I developed the cumulative function of eNPV for each scenario with PoS and it was very useful to see the operating characteristics of each strategy option.

Conclusion

In order to optimize the productivity of drug development in a pharmaceutical company, it is critical to consider various options of development program and evaluate the value of each strategy quantitatively. Statistical scenario simulation is a good approach to solve the complex multi-dimensional trade-off problems without any human cognitive bias of decision makers. The scenario simulations judged the aggressive strategy (e.g. develop multiple indications at a time) was better than standard plan (e.g. develop unique indication first) in some settings. It was a seemingly counterintuitive result and therefore I found that such a quantitative analysis is critical to support the good decision making. I concluded that a quantitative decision making approach that is available from various scenario simulations is critical by maximizing eNPV and minimizing risk at each drug development phase.