Dose finding using the biased coin up-and-down design and isotonic regression

Area of Science:

• Biostatistics
• Clinical Trial Design
• Pharmacometrics

Background:

• Accurate dose-finding is crucial for drug development.
• The up-and-down biased coin design (BCD) is a method for dose escalation.
• Estimating the target dose with a specific toxicity rate is a key challenge.

Purpose of the Study:

• To investigate five estimators for the target dose within the BCD framework.
• To compare their performance in terms of accuracy and efficiency.
• To identify the most effective dose-finding estimator for the BCD.

Main Methods:

• Utilized maximum likelihood, weighted least squares, sample averages, and isotonic regression for dose estimation.
• Employed the up-and-down biased coin design (BCD) for dose escalation.
• Evaluated estimators based on mean square error and the average number of subjects needed for convergence.

Main Results:

• A linearly interpolated isotonic regression estimate demonstrated strong performance.
• This method was found to be simple to derive and implement.
• It performed comparably or superiorly to other estimators in most studied scenarios.

Conclusions:

• The linearly interpolated isotonic regression estimator is a recommended approach for dose-finding with the BCD.
• This method offers a balance of simplicity and statistical efficiency.
• It aids in achieving prespecified toxicity rates in clinical trials.