SEMINAR – September 13, 2024

Abstract

Multi-arm trials are gaining interest in practice given the statistical and logistical advantages that they can offer. The standard approach is to use the fixed (throughout the trial) allocation ratio, but there is a call for making it adaptive and skewing the allocation of patients towards better performing arms. This is motivated by the goal of providing the most benefit to the patients in the trial while maximizing the information collected on the most promising arms. However, among other challenges, it is well-known that these approaches might suffer from low statistical power. We present a response-adaptive design which explicitly allows to control the trade-off between the number of patients allocated to the “optimal” arm and the statistical power. Such a balance is con- trolled through the calibration of a tuning parameter, and we explore various strategies to effectively perform it. We consider the general setting of a normally distributed endpoint and the design that targets a desirable value of that endpoint. The proposed allocation rule naturally arises from a weighted version of Shannon’s differential entropy, a context-dependent information measure which gives a greater weight to those treatment arms which have characteristics close to the pre-specified clinical target. We also introduce a simulation-based hypothesis testing procedure to assess whether the best performing treatment arm is significantly superior to the second best. This emphasizes a primary focus on selecting the optimal arm rather than on a comparison to the control. A simulation study highlights the potential advantage of the proposed class of designs over the considered competitors in an early Phase IIa proof-of-concept oncology clinical trial.

About the Speaker

Gianmarco Caruso is a post-doctoral researcher at MRC Biostatistics Unit (University of Cambridge, UK) working on efficient design of clinical trials. He is specializing in Bayesian statistics applied to clinical trials methodology (e.g., response-adaptive designs, dynamical borrowing, prior elicitation) and quantitative ecology (e.g., capture-recapture models). He completed his PhD in Methodological Statistics at the Sapienza University of Rome (Italy) in May 2023, and holds a Master’s degree in Applied Mathematics from PSL University (Paris Dauphine, France) as well as a Master and Bachelor’s degree in Statistics from Sapienza University of Rome.

Event Organizer

Jonathan L. Auerbach
Assistant Professor, Department of Statistics
College of Engineering and Computing
George Mason University

Ben Seiyon Lee
Assistant Professor, Department of Statistics
College of Engineering and Computing
George Mason University