Inferential Seamless Designs for Treatment Selection during a Phase II/III Study in Oncology

In drug development, clinical trials are a cornerstone of evaluating new treatments. However, traditional designs often face significant challenges in efficiency, particularly when assessing multiple treatment options. The emergence of Multi-Arm Multi-Stage (MAMS) designs, such as the "Drop-the-Losers" (DtL) approach, offers innovative solutions by combining multiple hypotheses within a single trial framework. This thesis evaluates the DtL approach, focusing on its application in Phase II/III oncology trials, where identifying promising treatments early is critical. This work extends existing methodologies by incorporating binary and survival endpoints, addressing limitations of earlier designs which focused primarily on continuous outcomes. Using simulation studies, the statistical performance of the DtL design was assessed across diverse clinical scenarios. Key metrics included Type I error control, statistical power and the Monte Carlo Standard Error (MCSE) under varying endpoint distributions. Simulations involved endpoints such as progression-free survival (PFS). Results demonstrated that the DtL design effiectively balances statistical rigor and efficiency. The design maintained robust Type I error control while ensuring high power in detecting clinically relevant effects. Scenarios with harmful or ineffective treatments highlighted the ethical advantage of eliminating suboptimal options early, minimizing patient exposure to ineffective therapies. One limitation identified is that long-term survival endpoints may not be mature enough to supportearly treatment selection analyses. Updating the endpoint of interest over time could align with clinical practices, starting with objective response rate at the first analysis, transitioning to Benefit-Risk scores or PFS at the second analysis, and concluding with overall survival at the final analysis. However, such an approach requires careful handling of correlations between drug effects across these endpoints, since selecting treatments based on a poor surrogate endpoint. This thesis provides a comprehensive framework for implementing DtL designs in clinical trials, demonstrating its potential to accelerate treatment evaluation in oncology while upholding ethical and scientific standards. This methodology contributes to the ongoing evolution of designs, underscoring their value in modern clinical research.