An adaptive Bayesian hierarchical design approach for a multi-histology oncology trial targeting specific pathways and genetic signatures rather than a histological subtype
View Presentation View Presentation
*Ohad Amit, GlaxoSmithKline  Allison Florance, GlaxoSmithKline  Elizabeth Krachey, GlaxoSmithKline 

Keywords: Bayesian, simulation, oncology

Recent advances in the understanding of cancer biology have led to the development of therapeutics that target specific pathways and genetic signatures rather than a histological subtype. Multi-histology oncology trials enroll subjects on the basis of the therapeutic target and pathway of interest rather than the specific tumor type. This allows for effective investigation of multiple rare cancers within a single trial. Methodology for an adaptive Bayesian hierarchical design is examined for a Phase 2, one arm, multi-histology trial. Ongoing interim analyses are incorporated to assess futility and success for each histology. Simulation work is used to develop interim and final decision rules, and select a hierarchical model with an appropriate amount of borrowing across histologies. Operating characteristics are examined to study the robustness of the design to various trial parameters such as sample sizes and response rates. This design is also compared to that of more traditional study designs.