Basket trials involve patients from multiple histologies expressing the same target and they have quick emerged as one of the most popular to develop precision medicine regimens. Since all patients express the same target and receive the same intervention there are efficiencies to be gained by allowing information sharing between baskets formed by histologies. This can either be done using a frequentist aggregation schema or through a Bayesian hierarchical model. Regardless of the choice of method for information sharing certain statistical issues need to be addressed in all basket trials. These include control of family-wise Type I error rate, definition of power under the multi-dimensional alternative space, and calibration of competing designs for a fair comparison.
An efficient basket trial design.
Cunanan KM, Iasonos A, Shen R, Begg CB, Gönen M. Stat Med. 2017 May 10;36(10):1568-1579. doi: 10.1002/sim.7227. Epub 2017 Jan 18.
Basket Trials in Oncology: A Trade-Off Between Complexity and Efficiency. Cunanan KM, Gonen M, Shen R, Hyman DM, Riely GJ, Begg CB, Iasonos A. J Clin Oncol. 2017 Jan 20;35(3):271-273. doi: 10.1200/JCO.2016.69.9751. Epub 2016 Nov 28.