|
Abstract:
|
Typically phase II cancer trial is single-arm with tumor response rate as the primary endpoint, and two-stage designs are used. However, in many situations investigators are poorly served by such conventional methodology, primarily for two reasons: 1) there might be high uncertainty in targeting the alternative hypothesis, and 2) it is impossible to estimate the treatment effect as compared to historical control, since the treatment effect is confounded with differences in trials. An adaptive two-stage design with an active control is proposed to tackle the problems, in which the information from both treatment and control arms in the first stage is utilized to either reject or continue testing the study treatment with either an optimistic or a skeptic target response rate, based on outcomes from stage I. The characteristics of the proposed design are evaluated by simulation studies.
|
