ABSTRACT
This chapter discusses some challenges and misconceptions on designing combination trials and provides practical guidance and designs to practitioners for conducting drug combination trials. A major challenge in designing combination trials is that combinations are only partially ordered according to their toxicity probabilities. Drug combination therapy has become the mainstream approach to cancer treatment. Numerous designs have been proposed to find a single maximum tolerated dose (MTD) for drug combinations. Riviere found only one trial that used a new model-based design in their review of 162 drug combination trials. Illustration of the linearization approach is to find a single MTD for drug combination trials using a zigzag or linear path. Unlike the linearization approach that requires users to select a specific ordered path in the dose combination matrix, the Bayesian optimal interval design (BOIN) drug combination design provides a simple, well-performing method to find a single MTD directly in the two-dimensional dose matrix.
