People think in Bayesian terms all the time: we use prior information and the evidence at hand to make decisions in our day-to-day lives. And it is this adaptive thinking that can be so useful in clinical trials, for example, in oncology trials where the standard of care might evolve during the course of the trial. In his recent webinar, “Novel Methods of Trial Design” Prof. Yuan Ji, serving as Cytel’s executive advisor, discusses Bayesian methods in early-phase oncology trials. Let’s take a closer look.

Bayesian Statistical Methods

In statistics, there are two different philosophies — frequentist and Bayesian. Bayesian statistics have a much older history than frequentist methods; they were first developed by the English minister Thomas Bayes in the 1700s. Bayesian statistics allow us to make probabilistic inferences on the parameter of interest, which is missing in a traditional frequentist approach. The Bayesian approach allows every new piece of data to serve as evidence to update a hypothesis called a “prior.” A “posterior” is then the result of the prior being updated considering this new evidence. This approach makes Bayesian methods ideal for small sample trials, clinical trials using historical data and real-world solutions, or clinical trials that require flexible learning.

Bayesian Methods in Early-Phase Oncology Trials

In recent years, early-phase trials in oncology have gotten more sophisticated and powerful. For example, with Project Optimus, the FDA is making efforts to reform the dose optimization and dose selection paradigm in oncology drug development. In this new era of dose optimization, maximum tolerated dose (MTD) is no longer considered the optimal dose. Instead, Bayesian adaptive dose-finding designs, such as the eff-tox designs that combine efficacy and toxicity outcomes, are preferred to find the optimal dose. The joint i3 + 3 (Ji3 + 3) design for eff-tox dose finding is used for cell and gene therapies as it considers both safety and efficacy outcomes in making dosing recommendations.

Multiple Cohort Expansion (MuCE) is one of the more innovative elements of Bayesian designs. It is an advanced Bayesian approach superior to Simon’s 2-stage design for expansion cohort trials and master protocols. Methods like MuCE are especially important in oncology where several doses and indications must be tested for successful completion of early phase trials.

In his webinar, Professor Ji also shares the many advantages of using Bayesian methods in confirmatory settings. He addresses the benefits of using statistical software in enabling the work of statisticians more efficiently.

Overall, in oncology, where standard of care might evolve during the course of a clinical trial, statisticians trained in Bayesian methods can help salvage data collected during a clinical trial.

Webinar by Professor Yuan Ji of The University of Chicago

Cytel is partnering with LIDE Biotech to conduct a webinar series that explores the use of co-clinical trials in expediting drug development for cancer patients. During the second webinar in this series, Professor Ji shares his insights on the relevance of Bayesian statistics in clinical trial designs and presents basic concepts of statistical modeling and decision-making. He also talks about advanced novel designs for early-phase and late-phase clinical trials. For early-phase oncology trials, Professor Ji lists a few recent developments for dose optimization and for late-phase trials, he discusses examples of fundamental issues of statistical errors.

Share Via