Stochastically curtailed phase II clinical trials

Area of Science:

• Biostatistics
• Clinical Trial Design
• Pharmacoeconomics

Background:

• Phase II clinical trials are crucial for evaluating new treatments, balancing the need to identify efficacy with minimizing patient exposure to ineffective therapies.
• Existing designs like Simon's minimax and optimal designs address early termination but may be improved.
• Statistical decision-making in early-phase trials requires robust methods to handle uncertainty in treatment response proportions.

Purpose of the Study:

• To propose and evaluate three novel statistical designs for phase II clinical trials based on stochastic curtailment and conditional power.
• To compare these new designs against established methods like Simon's minimax and optimal designs.
• To assess the performance of the proposed designs regarding early termination opportunities, expected sample size, and type I/II errors.

Main Methods:

• Development of three stochastically curtailed (SC) designs: SC binomial tests, SC Simon's optimal design, and SC Simon's minimax design.
• Utilizing conditional power as the basis for stochastic curtailment decisions.
• Comparative analysis of the proposed SC designs with Simon's traditional minimax and optimal designs, focusing on key performance metrics.

Main Results:

• The study compares the number of opportunities for early study termination across all designs.
• Expected sample sizes under the null hypothesis (p <= p(0)) are evaluated for each design.
• Effective type I and type II error rates are contrasted between the proposed SC methods and traditional designs.

Conclusions:

• The proposed stochastically curtailed designs offer alternative approaches to phase II trial management.
• These methods provide tools for more efficient trial conduct by optimizing early stopping rules.
• Graphical monitoring tools are presented to aid in the practical application of these stochastic curtailment designs in ongoing trials.