SC08: Quantitative Decision-Making for Staging up to Phase 3 Clinical Development

Quantitative decision-making (QDM) in drug development involves pre-specifying a set of criteria for actions of "go," "no go," and sometimes "gray" and evaluating the operating characteristics of these criteria. A key application of QDM is to make decisions for the next stage of development based on available data. While decisions can be based on various success criteria such as regulatory approval, health technology assessment (HTA) outcomes, or commercial viability, common scenarios focus on parameters that characterize the treatment effect or an estimate thereof or the probability of technical success. This short course will cover the foundations of QDM in clinical development. It will include non-Bayesian approaches, such as those based on the sampling distribution of the parameter estimate(s) of interest, and Bayesian approaches, including those based on the posterior distribution of the parameter(s) of interest and the posterior predictive distribution of a function of future data. Topics will include defining a critical success factor (CSF) for a proof-of-concept (PoC) and a phase 2 study, using predictive models with frequentist and Bayesian approach to evaluate CSF, and employing novel estimators to enhance estimation efficiency. The course will also address challenges in rare disease development, where small, often uncontrolled studies form a basis for stage up to phase 3 randomized controlled pivotal study(ies). A closely related issue is that, following QDM, directly using phase 2 study results to plan phase 3 studies is subject to selection bias, potentially leading to over-estimation of the treatment effect. This issue can be further exacerbated with additional endpoint and/or subgroup selection, as well as potential violation of pre-determined QDM rule in light of seemingly promising post hoc analyses of phase 2 data. The course will also cover the theory and methods for addressing selection bias when planning phase 3 based on positive phase 2 results.