Adaptive Bayesian Designs for Phase I/Ii Clinical Trials(2/2)

In a conventional clinical trial for drug development, Phase I and Phase II trials are usually conducted to investigate the toxicity (or adverse effect) and short-term efficacy, respectively. The Phase I trial determines the maximum tolerated dose (MTD) under a specific dose limiting toxicity (DLT), while patients in the Phase II trial receive the MTD and then the short-term efficacy is tested. In general, a drug with allowable toxicity means that a subject after taking the drug has the probability of DLT response less than 1/3. Moreover, a drug has a required efficacy is that a subject after taking the drug has at least a specific threshold for the probability of short-term response. This two-year project develops adaptive Phase I/II clinical trials for binary responses of DLT and efficacy under some models of dose-toxicity curve and dose-efficacy curve, where Bayesian tests are constructed based on the posterior probability of the allowable toxicity and that of the required efficacy. At the first stage of the proposed adaptive clinical trial, the posterior probability of the allowable toxicity is sequentially computed to determine dose-up or dose-down. At the second stage with a batch experiment, the posterior probability of the allowable toxicity and that of the required efficacy are used to evaluate the safety and efficacy of the drug. In the first year, the project focuses on the case that, at stage two, the DLT response of subjects at stage one are still under monitoring. In the second year, the project considers the situation where both the DLT and efficacy responses of subjects at stage one are available at stage two. Hence, the proposed adaptive Phase I/II clinical trials can be used for the problem with late-onset toxicity and reach the conclusion of safety and efficacy of the drug in a shorter time period with less number of subjects.