Proceedings of the University of Pennsylvania ninth annual conference on statistical issues in clinical trials: Where are we with adaptive clinical trial designs?

This issue of Clinical Trials contains the proceedings of the ninth annual symposium on statistical issues in clinical trials held at the University of Pennsylvania on 13 April 2016. This symposium considered special statistical issues in the design, conduct, and analysis of adaptive clinical trials. Clinical trials in some sense have, of course, always been adaptive—experimentation with human beings (all sensate beings, actually) requires being prepared to make changes if needed to ensure the safety of participants and the value of ultimate results. Early adaptations focused on trial termination, with development of sequential and multi-stage designs to allow more rapid acceptance or abandonment of therapies appearing definitively beneficial, harmful, or ineffective well prior to its planned completion.^1–3 Over the past 25 years, there has been an increasing interest in a broader range of adaptations, including increasing sample sizes based on interim comparisons, response-adaptive randomization, and incorporating accumulating information about the association of patient characteristics to treatment response into the treatment assignment process.^4–9 Many proposed designs are constructed within a Bayesian framework, requiring regular updates of models with information accrued from the ongoing trial; others are more sophisticated versions of the earlier multi-stage designs, constructed within the classical frequentist framework. Those developing such designs have argued that they are more efficient, as well as more ethical in that they can increase the proportion of trial participants assigned to treatments with the greatest likelihood of benefiting them, and more quickly remove ineffective or harmful products from further consideration. Others have been skeptical of both of these arguments and have raised concerns about the complexity of many of these designs, as well as the possibility that some adaptations may reveal too much information about interim comparative data. Regulatory agencies have been open to use of the new generation of adaptive designs in the early phases of drug development, but have been reluctant to accept them for trials intended to serve as the primary basis for product approval. ¹⁰ This conference was intended to provide an overview of the current thinking regarding adaptive designs. We invited speakers with substantial experience in designing, conducting, and analyzing adaptive clinical trials. As in prior conferences, panel discussants and members of the audience offered additional interesting perspectives on the issues. In this issue, you will find papers presented at the conference as well as edited transcripts of the panel and general discussions (Table 1).

OPEN IN VIEWER

Table 1.

2016 University of Pennsylvania conference on statistical issues in clinical trials: where are we with adaptive clinical trial designs?

Faculty	Institution	Topic
Frank Bretz, PhD	Novartis Pharmaceuticals	Adaptive Designs: The Swiss Army Knife Among Clinical Trial Designs?
Mahesh Parmar	University College London	Improving Outcomes for Patients As Rapidly As Possible: Some New Designs for Phase III Trials
Michael Proschan, PhD	NIAID	Blinded Adaptations: What Is and Is Not Okay?
Peter F. Thall, PhD	MD Anderson Cancer Center	Caveats for Outcome-Adaptive Randomization in Comparative Clinical Trials
Bruce W. Turnbull, PhD	Cornell University	Adaptive Trial Designs: A DSMB Perspective
Panelists	Institution
Rick Chappell, PhD	University of Wisconsin-Madison
Jason T. Connor, PhD	Berry Consultants
Angie DeMichele, MD, MSCE	University of Pennsylvania
Valerie Durkalski, PhD	Medical University of South Carolina
Steven Joffe, MD, MPH	University of Pennsylvania
Janet Wittes, PhD	Statistics Collaborative, Inc.

The Penn series of 1-day conferences on statistical issues in clinical trials is intended to provide intense focus on an area of statistical methodology of continued and/or emerging interest related to the design, conduct, and analysis of clinical trials. All conference proceedings except the first have been published in Clinical Trials.^11–18