Two-stage designs for dose-finding trials with a biologic endpoint using stepwise tests

Area of Science:

• Clinical Trials
• Biostatistics
• Pharmacology

Background:

• Early phase dose-finding trials aim to identify the minimum effective dose (MED) for drugs.
• Traditional stepwise testing procedures are often single-stage, potentially leading to larger sample sizes.
• Monotone biologic endpoints, such as serum levels or gene expression, are common in these trials.

Purpose of the Study:

• To extend stepwise testing to a two-stage enrollment plan for dose-finding trials.
• To reduce sample size requirements through early stopping of unpromising doses (futility interim).
• To evaluate the performance of four proposed two-stage designs.

Main Methods:

• Development and examination of four novel two-stage dose-finding trial designs.
• Application of these designs to a hypothetical statin trial with four doses and a placebo.
• Calibration of design parameters and discussion of implementation strategies.

Main Results:

• A calibrated two-stage design reduced the average total sample size by 38% (from 125 to 78) compared to a one-stage test.
• Error rates and probability of correct selection were maintained comparably.
• The maximum total sample size slightly increased from 125 to 130.

Conclusions:

• Two-stage designs offer significant average sample size reductions in early phase dose-finding trials.
• These designs provide an efficient alternative to traditional one-stage methods for monotone biologic endpoints.
• The trade-off for reduced average sample size is a marginal increase in the maximum sample size.