Podcast and newsletter

Podcast Transcript

Bayesian Statistics is a particular approach of applying probability to statistical problems. This approach starts with a summary of our prior beliefs based on the relevant, available information. When we collect new data, for example in the course of a clinical trial, information from these data is combined with our prior beliefs to provide our current beliefs in terms of probabilities. In contrast, traditional or classical statistical approaches to decision-making are based on only the new data and do not incorporate any prior beliefs. Bayesian statistics can be used in practically all situations in which traditional statistical approaches are used and may have advantages. For example, when experts from various disciplines have determined that there is high-quality, relevant information external to a clinical trial, these methods may allow studies to be completed more quickly and with fewer participants, and it is easier to adapt the design of a Bayesian trial based on the accumulated information compared with a traditional trial.

While the development of Bayesian statistical methods began centuries ago, recent increases in computing power have made it more feasible to implement these statistical approaches. By the end of the second quarter of FY 2024, the FDA expects to convene a public workshop to discuss aspects of complex adaptive, Bayesian, and other novel clinical trial designs. By the end of FY 2025, FDA also anticipates publishing draft guidance on the use of Bayesian methodology in clinical trials of drugs and biologics.

At FDA, we are seeing increased use of Bayesian statistical methods in pediatric drug development. Since many cases of pediatric drug development occur after the demonstration that the drug is safe and effective in adults, Bayesian statistics can incorporate the information from adults that can be considered in understanding the effects of a drug in children.

A recent example is an asthma product that was discussed at an FDA advisory committee on Nov 8, 2022. The primary clinical study enrolled both pediatric and adult patients to estimate the rate of severe asthma exacerbations relative to placebo. Traditionally, we would look at the results for the populations separately and in a single pooled analysis. However, in our evaluation of this product, Bayesian methods allowed us to borrow variable amounts of information obtained from adults and to evaluate the dependence of the results on the amount borrowed and to ultimately make more informed decisions.

We are also seeing more Bayesian designs in dose finding trials for drugs and biologics, particularly in oncology where the primary goal is to understand the tolerability of the product and to identify a maximum tolerated dose (MTD). Here, Bayesian designs allow much more flexibility in the design and dosing in the trial and can improve the accuracy with which the MTD is estimated and the efficiency of the study by linking the estimation of toxicities across doses.

Ultra-rare diseases are another area where traditionally designed trials are particularly challenging given the extremely limited patient population sizes. Bayesian methods provide two key advantages here: the ability to incorporate prior information and the ability to adapt the design more easily.

Bayesian approaches using hierarchical models, i.e., multi-level statistical models describing how the data elements relate to one another, are particularly useful for assessing how well a drug works in particular subgroups of patients (defined for example by age or race). Analyses based on these models can provide estimates of drug effects in these subgroups that are generally more accurate than the estimates one obtains by analyzing each subgroup in isolation.

Sponsors are encouraged to meet with FDA when considering a Bayesian design. The Complex Innovative Designs (CID) Paired Meeting Program, established under PDUFA VI to support the goal of facilitating and advancing the use of complex adaptive, Bayesian, and other novel clinical trial designs, offers sponsors whose meeting requests are granted the opportunity for increased interaction with FDA staff to discuss their proposed CID approach.Thus far, the selected submissions in the CID Paired Meeting Program have all utilized a Bayesian framework. This is not surprising as Bayesian approaches may be well-suited for some complex innovative designs because they can provide flexibility in the design and analysis of a trial. In addition, Bayesian inference may be appropriate in settings where multiple sources of evidence are considered, such as has been proposed in some of the selected submissions.

Resources

• Impact Story: Using innovative statistical approaches to provide the most reliable treatment outcomes information to patients and clinicians
• Complex Innovative Trial Design Meeting Program
• Bayesian_Methods_in_Human_Drug_and_Biological_Products_Development_in_CDER_and_CBER
• Perspectives_on_informative_Bayesian_methods_in_pediatrics