PS55: Allocating for Robustness in Platform Trials: A Regulatory Case Study

Platform trials promise large gains in efficiency over standard trial designs, often 30-50% or higher from the use of shared controls, adaptive stopping, and reinvestment of saved resources. These trials, however, often come with additional assumptions. The FDA draft master protocol guidance dedicates significance discussion to assumptions arising from "two tier" randomization, where patients are allocated to a cohort (a treatment and its matched control) and then to active or control within a cohort. In this setting, usually dictated by operational concerns with blinding and consent, there are concerns about the poolability of controls. To date, many of the allocation procedures proposed in platform trials (such as the "sqrt(k)" allocation discussed in the adaptive guidance) are aimed at maximizing the efficiency of the pooled analysis and may actually amplify, rather than mitigate, the impact of these assumptions. Is this optimal when assumptions are being investigated and sensitivities are needed such as using "matched controls" (controls within an individual cohort)? In this session we will review these issues and discuss allocation methods when both efficiency and robustness are desired. This session will discuss a platform trial that has been extensively discussed with FDA, including initial concerns about the poolability of controls and a response to change the allocation scheme that produces sizeable increases in the power of the sensitivity analyses. This methodology is applicable to all platform trials employing a "two tier" randomization scheme and may allow for maintaining much of the efficiency of platform trials while minimizing concerns about assumptions. This session will consist of three talks. The first is a description of randomization schemes in platform trials and the required assumptions under different allocation systems. The second talk is a case study of the trial referenced above, including the originally proposed allocation scheme and the changes made to address regulatory concerns. The third talk will be from an FDA regulator discussed the advantages and disadvantages of the proposed allocation methodology.