Bayesian Multi-Arm De-Intensification Designs

TL;DR

This paper presents a Bayesian sequential design for evaluating de-intensified cancer therapies, focusing on toxicity and efficacy, with applications in HPV-associated oropharynx cancer to identify safer treatment options with similar effectiveness.

Contribution

It introduces a novel Bayesian multi-arm de-intensification design that models toxicity and efficacy for adaptive decision-making in cancer trials.

Findings

Effective early termination for inferior treatments

Successful identification of non-inferior de-intensified therapies

Validated design through simulations and real data

Abstract

In recent years new cancer treatments improved survival in multiple histologies. Some of these therapeutics, and in particular treatment combinations, are often associated with severe treatment-related adverse events (AEs). It is therefore important to identify alternative de-intensified therapies, for example dose-reduced therapies, with reduced AEs and similar efficacy. We introduce a sequential design for multi-arm de-intensification studies. The design evaluates multiple de-intensified therapies at different dose levels, one at the time, based on modeling of toxicity and efficacy endpoints. We study the utility of the design in oropharynx cancer de-intensification studies. We use a Bayesian nonparametric model for efficacy and toxicity outcomes to define decision rules at interim and final analysis. Interim decisions include early termination of the study due to inferior survival of experimental arms compared to the standard of care (SOC), and transitions from one de-intensified treatment arm to another with a further reduced dose when there is sufficient evidence of non-inferior survival. We evaluate the operating characteristics of the design using simulations and data from recent de-intensification studies in human papillomavirus (HPV)-associated oropharynx cancer.