Main page > Innovation > Flexible designs > PhRMA adaptive design working group

This page is maintained by the PhRMA adaptive design working group (Alex Dmitrienko).

Adaptive design lecture series.

Testing and estimation in adaptive group sequential designs with treatment selection presented by Prof. Martin Posch (Medical University of Vienna) [Download slides].

Integrating selection and confirmation phases into a single trial can expedite the development of new treatments. We consider adaptive group sequential trials to compare several treatments (or doses) with a control. In an interim analysis one or more of the considered treatments are selected to be continued in the further stages. The choice of treatments depends on the data accumulated so far. This includes typically not solely data on the primary endpoint but also data on secondary endpoints, safety data and also data from external sources as, e.g., other trials running in parallel. Adaptive designs based on combination tests are efficient tools to implement treatment selection without compromising the multiple type I error rate (Bauer and Kieser, 1999; Hommel, 2001). A crucial feature of these designs is that the treatment selection rule need not be specified in advance. For the sake of credibility, in clinical trials one should, however, specify the major options for treatment selection in the protocol. Besides hypotheses testing, we discuss the properties of point estimates and the construction of confidence intervals for designs with adaptive treatment selection (Posch et al. 2005).

Bauer, P. and Köhne, K. (1994). Evaluation of experiments with adaptive interim analyses. Biometrics. 50, 1029-1041.
Müller, HH. and Schäfer, H. (2004). A general statistical principle for changing a design any time during the course of a trial. Statistics In Medicine. 23, 2497-2508.
Posch M., König F., Branson M., Brannath W. Dunger-Baldauf C. and Bauer P. (2005). Testing and estimation in flexible group sequential designs with adaptive treatment selection. Statistics in Medicine. 24, 3697-714.
Hommel, G. (2001). Adaptive modifications of hypotheses after an interim analysis. 43, 581-589

Supply issues in adaptive clinical trials presented by Mr. Tom Parke (Tessella) [Download slides].

Adaptive designs make drug supply during the trial much harder and if they feel they can't supply your trial you won't be able to run it. So what are the key issues for drug supply and how can we address them:
1. There are more treatment arms - how do we supply more doses?
2. Arms may be dropped/introduced or arms may become more/less likely to be allocated.
3. How do we know how much of each dose we will need make/package?
4. During the trial, how do we know which doses to ship?

Adaptive designs for clinical trials: Insightfully innovative or irrelevantly impractical presented by Prof. Stuart Pocock (London School of Hygiene and Tropical Medicine) [Download slides].

Drug supply for adaptive trials presented by Dr. Nitin Patel (Co-founder and Chairman, Cytel Inc) [Download abstract] [Download slides].

Interim analyses with multiple primary endpoints – Application to an HIV vaccine trial presented by Dr. Devan Mehrotra (Merck) [Download slides].

Published research on group sequential methods and adaptive designs has largely focused on clinical trials with a single primary endpoint. In this presentation, we will discuss interim analysis strategies when there are two primary endpoints and the goal is to maximize the number of endpoints for which statistical significance can be demonstrated in a timely manner while ensuring control of the overall type I error rate. To illustrate the key ideas, we will describe an interim analysis strategy that was developed and implemented for a placebo-controlled HIV vaccine test-of-concept efficacy trial with dual primary endpoints: HIV infection status (infected/uninfected) and post-infection HIV RNA set-point (for those who become HIV infected). The proposed strategy can be extended to more general settings, including trials with more than two treatment arms and/or with than two primary endpoints.

Population enrichment within a group sequential design presented by Dr. Cyrus Mehta (Cytel Inc) [Download slides].

We present a method for combining group sequential stopping rules with population enrichment within the framework of a confirmatory clinical trial. The approach is motivated by the need for greater efficiency in large cardiology trials where event rates are low and efficacy gains relative to the current standard of care are likely to be small. Since such trials typically involve sample sizes in the thousands, group sequential designs with early stopping for benefit or futility are commonly adopted. The chances of success with such trials can be greatly enhanced if patient eligibility is restricted to a subpopulation, selected from analysis of interim data that is sensitive to the new treatment.Statistical methodology for enriching an ongoing group sequential trial in this manner, possibly accompanied by a data dependent sample size increase, will be presented. The approach is general and can be applied to late stage clinical trials in other therapeutic areas besides cardiology.

Sample size recalculation in internal pilot study designs: A review presented by Prof. Tim Friede (University of Warwick) [Download slides].

The adequacy of sample size is important to clinical trials. In the planning phase of a trial, however, the investigators are often quite uncertain about the sizes of parameters which are needed for sample size calculations. A solution to this problem is mid-course recalculation of the sample size during the ongoing trial. In internal pilot study designs, nuisance parameters are estimated on the basis of interim data and the sample size is adjusted accordingly. This review attempts to give an overview on the available methods.

Comparing adaptive designs and the classical, group sequential approach to clinical trial design presented by Prof. Chris Jennison (University of Bath).

We shall consider the objectives of adaptive and group sequential designs and compare these approaches in two applications. The first case study concerns sample size re-estimation in response to interim estimates of the treatment effect. Our conclusion here is that group sequential methods provide an adequate mechanism to curtail or prolong a study as information accrues about the primary endpoint. In the second case study we focus on adaptation of the patient population in response to interim results and see tangible benefits from the adaptive approach: moreover, the issues here are beyond the remit of group sequential tests. The talk will conclude with an attempt to generalise from these two examples.