Sometimes a new candidate drug for a pediatric study has already been tested on adults for safety and efficacy. We know that the drug is likely to work quite differently in children, but we do not know the degree to which the effects will be different. As a result, a conventional approach is to discard much of the information that has already gathered during studies of adults, and then to start from scratch with a pediatric trial.
Taking this approach has at least one unfortunate consequence. It means throwing out information that could also be relevant for the pediatric study. Such information may then have to replicate during the new study, meaning that a pediatric drug will take even longer to get to market and to reach young patients.
A more modern clinical trial design can utilize the information from the adult trials as Bayesian priors – information that is updated during the course of a trial. Bayesian methods have a pragmatic alternative to the problems created by a conventional design: no information needs to be thrown out unless new information arrives to supplant it. This method is known as Bayesian updating and always works off of existing information.
Such new information can achieve the same well-powered trials that a conventional method might employ. However, since a portion of the information confirmed by a conventional trial already exists, the trial can be significantly faster.
Recently, Cytel designed a Bayesian pediatric trial for a customer and combined it with a three look adaptive design. This reduced the sample size by 30% of patient costs, and also reduced trial times by 20% - 40% (depending on early stopping.) Most importantly, however, the results maintained the type-1 error rate, even while achieving significant benefits for both patients and sponsors.
Pediatric trials are particularly well suited to benefit from Bayesian designs. A great deal of prior information exists for pediatric drugs that have correlating adult therapeutics. Leveraging such prior information can spare the need to start from scratch for testing a therapeutic on which there is already significant information available.
To learn more about Bayesian trials read this post from Cytel Senior Director of Consulting, Pantelis Vlachos: Frequentist? Time for an update!
Related Items of Interest
Adaptive Design and Bayesian Statistics: 5 Years Later (Podcast)
5 Reasons to Invest in Bayesian Dose-Escalation Methods
Bayesian Trial Designs are Safe, Efficient & Flexible: A Review of Published Phase 1 Studies
Bayesian-Bandit Adaptive Designs for Rare Disease Drug Development
