Determination of Interim Go/No-Go Criteria with Simulations of Longitudinal Continuous Data in a Phase 2 Clinical Trial of a Neurodegenerative Disorder

• Activity Number: 337 - Interim Monitoring and Analyses: Two-Stage, Multi-Stage, and Group Sequential Designs
• Type: Contributed
• Date/Time: Tuesday, August 1, 2017 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Weining Z Robieson* and Greg Cicconetti and Deli Wang
• Companies: AbbVie Inc. and Abbvie and AbbVie, Inc.
• Keywords: interim analysis ; simulation ; futility ; Go/No-Go criteria

Abstract

A study design task for this trial with continuous, 52-week change from baseline primary endpoint was to determine the number and timing of interim analyses (IAs) for Go/No-Go decisions. Trial idiosyncrasies led to consideration of two to four potential IAs. The late endpoint incentivized consideration of earlier visits at IAs. A mixed model repeated measure analysis is run at each IA using observed data and future visits for ongoing subjects imputed based on a linear disease progression assumption. Imputed Week 52 effect size, based on the placebo corrected contrast and imputed estimate of standard deviation, was chosen as the metric for interim Go/No-Go criteria. Trials were simulated for a number of longitudinal response scenarios. Go/No-Go criteria were based on whether the imputed effect size exceeded respective thresholds. No available statistics software is able to design a trial utilizing all available longitudinal continuous data for interim analyses. We will discuss our innovative approach for identifying decision rules and considerations that helped determine the final rule from a set of rules that met the desired characteristics.