Statistical Considerations in Developing Master Protocols in Oncology
Shenghui Tang, FDA  *Lijun Zhang, FDA 

Keywords: master protocol, biomarker, oncology

A master protocol can include multiple diseases, multiple treatments, or multiple molecular markers. It could be an umbrella trial, a cloud trial, or basket trial. A study with a master protocol can screen patient for multiple biomarkers and could lower screen failure rate effectively. In 2014, Lung-MAP, the lung Master Protocol, was launched and it has 4 sub-studies which include 4 novel agents targeting either PI3K, CDK4/6, FGFR, or c-MET for the 2nd-line treatment (those who have progressed on platinum-based doublet chemotherapy) of advanced squamous NSCLC. Patients with none of these abnormalities will be enrolled onto the anti-PDL1 study. Each sub-study includes a phase II component and the primary objective of the phase II component is to evaluate if there is sufficient evidence to continue to the phase III component by comparing progression-free survival. Statistical issues in developing master protocols will be discussed, which include sub-study assignment for patients who has more than one biomarker, estimation of treatment effect under different sub-study assignments when biomarker prognostic and/or predictive effect sizes are different, and etc.