Evaluating the Loss of Efficiency for Promising Zone Designs Compared to Group Sequential Designs in the Setting of Time-to-Event Data

• Activity Number: 534
• Type: Contributed
• Date/Time: Wednesday, August 7, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Martin King*+
• Companies: AbbVie
• Keywords: promising zone ; adaptive design ; group sequential design ; late-stage design

Abstract

The "promising zone" design (PZD) is an alternative to the traditional group sequential design (GSD) for late-stage studies. PZDs allow the sample size to be increased based on the conditional power (CP), assessed at an interim analysis. If interim results are "unfavorable" (low CP; for example, < 30%) or "favorable" (high CP, e.g., >90%), the PZD trial continues as originally planned. If interim results are "promising" (e.g., CP of 30-90%), the sample size is increased according to a pre-specified algorithm.

The VALOR trial is a PZD in the setting of time-to-event data (Mehta, et al., ASCO 2012). We compare the VALOR trial PZD to various GSDs with respect to operating characteristics of: expected duration, expected cost, and power of the study. We find that GSDs can be designed that shorten the expected duration, reduce the expected cost, and/or increase the power vs. the PZD. Sensitivity analyses suggest these results are robust to design elements of the PZD (e.g., enrollment speed, rate of endpoints). We conclude that in the time-to-event setting, a PZD should be preferred over a GSD only if there is a specific reason to sacrifice efficiency compared to that achievable by a GSD.