Author of

• 12648

  +

  P3.11 - Poster Session 3 - NSCLC Novel Therapies (ID 211)

  • Event: WCLC 2013
  • Type: Poster Session
  • Track: Medical Oncology
  • Presentations: 1
  • Moderators:
  • Coordinates: 10/30/2013, 09:30 - 16:30, Exhibit Hall, Ground Level

  • 12664

    +

    P3.11-016 - Using Statistical Models to Improve Phase II Study Designs (ID 1580)

    09:30 - 09:30  |  Author(s): V.A.M. André
    • Abstract

    Background
    Single arm phase II studies have traditionally been conducted in oncology, but they rely completely on historical information to assess treatment benefits and are associated with high bias and false-positive error rates. This has contributed unsustainably high phase III clinical trial failure rates in oncology. On the other hand, conventional randomized controlled trials (RCT) involve more patients, higher cost and longer timelines. Thus, innovative statistical methods have been developed to try to address this challenge. An area of focus has been the use of statistical models to improve the design of phase II studies to better mimic what is intended for the phase III design, and therefore to predict its outcomes with greater precision.

    Methods
    Two statistical innovations have been applied to our phase II portfolio. Firstly, the Bayesian Augmented Control (BAC) design has been implemented to gain the benefits of using historical information (“borrowing”) as well as using randomization to a concurrent control arm. The BAC design utilizes models to summarize existing knowledge on the standard of care (SOC) in a defined patient population. For example, patients can be “borrowed” from a previous well-designed phase III RCT which established the SOC. The extent of “borrowing” depends on the similarity between the response in the new control patients and the historical patients, which is evaluated by pre-specified models. Secondly, Change in Tumour Size (CTS) from baseline to the end of Cycle 2 has been incorporated as an endpoint in our studies because this endpoint correlates strongly with progression-free survival (PFS) and overall survival (OS) in certain tumor types such as Non-Small Cell Lung Cancer (NSCLC). The approach consists in using tumor size measurements as a continuous variable, rather than a categorical endpoint based on Response Evaluation Criteria In Solid Tumors (RECIST), for assessing anti-tumor activity.

    Results
    The BAC design enables randomized controlled trials with smaller sample sizes, yet maintains statistical power. It also allows disproportionate enrollment to the experimental arm, which is often attractive to investigators and patients seeking access to novel therapies.

    Conclusion
    We believe that our phase II data package is now more informative whilst still meeting the logistical needs. It is hoped that this will facilitate a better Phase III prediction, which will increase our Phase III success rate. To date, insufficient phase III trials which were designed based on these improved phase II trials have yet been completed to be able to evaluate whether our phase III success rate has ultimately improved.