COMET: An Active Surveillance Trial in Low Risk DCIS, Challenges and Integration with National Cancer Database

• Activity Number: 154 - Integrating Real World Data with Clinical Trials: Opportunities and Challenges
• Type: Topic Contributed
• Date/Time: Tuesday, August 4, 2020 : 10:00 AM to 11:50 AM
• Sponsor: Caucus for Women in Statistics
• Author(s): Terry Hyslop*
• Companies: Biostat and Bioinformatics
• Keywords: non-inferiority; trial; observational; instrumental variable; bias; integration

Abstract

The COMET trial is a Phase 3 randomized non-inferiority trial of active monitoring to operative therapy in low risk Ductal Carcinoma In-Situ (DCIS) patients. ~20% of 50000 women diagnosed annually will go on to invasive cancer, with most treated according to invasive cancer guidelines; surgery, radiation and endocrine therapy, with potential for overtreatment. Meaningful clinical outcomes will be in 10 yrs, thus, we developed an observational study by selecting a treatment-stratified sample from 1,330 CoC-accredited centers, over-sampling for monitoring. We accessed detailed data on treatments, outcomes; and present some challenges of this observational cohort. In the randomized trial, patients may be reluctant to “accept” their randomization. These rates are higher and imbalanced, across arms leading to concern that the trial is more observational/preference-based than truly randomized. We assess bias, diminishment of power, and focus on instrumental variables to overcome the bias of ITT. We use the observational study as potential prior information for simulations and will integrate with the trial to enhance estimates of long-term outcomes.