Traditional rule-based approaches to dose escalation such as 3+3 are widely used in early clinical development. They can be appealing due to the simplicity of execution. However, estimates produced may be highly variable and the targeting of true Maximum Tolerated Dose may be poor.  Bayesian dose escalation approaches in early phase trials can offer an effective alternative to determining the maximum tolerable dose of a new drug more quickly, as well as ensuring that all of the information available to trial clinicians is taken into account so that the patients enrolled in the trial receive the best possible treatment.

However, there has been some historic reluctance to use these methods due to concerns about aggressive dose escalation, as well as the inherent complexities of prior specifications and computational difficulties.  Nowadays, the availability of powerful simulation software, and increased familiarity means such designs are becoming more popular. Indeed, several companies are now incorporating Bayesian methods for many of their early phase oncology trials.

The Bayesian Logistic Regression Model( BLRM) is one particular Bayesian method.  Its use in early phase dose escalation trials was pioneered in a paper by Neuenschwander et al (1) and is incorporated as a methodology within Cytel’s East Escalate for both single and dual agent trials.

The BLRM is a two-parameter dose-toxicity model which allows inferential statements about the probabilities of a DLT to be made at each dose level. After each patient cohort, information is derived from the posterior distribution of the model parameters.  This then allows the clinical team to decide on a sensible dose for the next patient cohort.

Notes:

1) Neuenschwander, B., Branson, M., and Gsponer, T. (2008). Critical Aspects of the Bayesian Approach to Phase I Cancer Trials. Statistics in Medicine, 27(13), 2420‚Äì2439. doi: 10.1002/sim.3230.

Share Via