32: Moving Clinical Trial Goalposts

In this episode of "In the Interim…", Dr. Scott Berry and Dr. Kert Viele analyze how regulatory, editorial, and science community standards often impose additional, inconsistent requirements for novel methods in clinical trial design, rarely applied to standard approaches. Examples from oncology, enrichment trials, platform studies, and endpoint analysis illustrate how adaptive and Bayesian designs are frequently subject to higher scrutiny, shifting metrics, or distinct evidentiary demands. The episode covers technical and regulatory issues, such as the selective application of Type 1 error controls, evolving multiplicity guidance, and challenges in ethical reasoning with adaptive allocation. Scott and Kert frame the discussion with empirical comparisons and advocate for the use of clinical trial simulation to ensure fair, metric-driven evaluation of both novel and legacy designs.

Key Highlights:

● Oncology combination therapy trial with Bayesian borrowing facing heightened regulatory caution versus single-arm historical controls.

● Hierarchical versus pooled analysis in enrichment/basket trials, with focus on error definitions and subgroup effects that have always existed.

● ICH E20 guidance potentially discourages use of enrichment by imposing new subgroup comparison burdens absent from standard trials.

● Platform trial multiplicity rules contrasted with parallel single-arm trials; regulatory stance continues to evolve.

● Ethical debate on adaptive allocation: questioning rationale behind adaptive randomizing may be ethically challenging, but fixed allocation is okay despite same interim data.

● Critical review of explicit utility weighting in the DAWN trial, despite alternative methods having the same issues