An Adaptive Seamless Phase II/III Design for Alzheimer’s Trials with Subpopulation Selection
*Richann Liu, Pfizer  

Keywords: Seamless design, Subpopulation selection, Phase II/III, Alzheimer’s, CDR-SB

Seamless phase II/III combined design is attractive as it offers the possibility of faster clinical development, overall cost saving and more efficient use of information. An operationally & inferentially adaptive seamless phase II/III Alzheimer’s trial design is proposed and discussed, where stage I of the design is for dose selection with 3 or 4 doses and placebo and stage II is dose confirmation with 1 or 2 doses and placebo. The primary objective is to establish efficacy and safety of the drug compared to placebo and the primary efficacy endpoint will be change from baseline in Clinical Dementia Rating-Sum of Boxes (CDR-SB) at week 104. Prospectively defined subgroup will be enrolled in the study with the hypothesis of particular benefit. An intermediate endpoint during stage I will be used to ensure stage II includes the most appropriate patient population assuming strong correlation between intermediate and primary endpoint. Hybrid Bayesian-frequentist framework (Speigelhalter, O’Hagan) was utilized for interim decision making and final analysis. Design calibration for clinical and statistical parameters has been conducted through extensive modelling & simulations to explore operating characteristics under different scenarios.