What is an adaptive design?
An adaptive clinical trial design is defined as a study which includes a sequence of interim analyses to enable sponsors to dynamically modify the course of a trial. Broadly speaking there are two different types of adaptive designs, rigid and flexible designs. In rigid adaptive design studies rules for adaptation are completely specified in the protocol. A traditional interim analysis is an example of a rigid adaptive design. Advantages of such designs include
• Potential modifications are approved up front by regulatory authorities and ethics committees.
• No need to file protocol amendments
• Logistics for changing treatments, doses etc. can be planned up front
• Credibility of results is maintained, especially with “firewalls” where only a limited number of people have access to results
• Broad regulatory acceptance
Flexible adaptive designs consist of studies where there are no fixed rules for adaptations. Adaptations are driven by the observed data and the overall drug development objectives for the study medication under consideration, not through protocol design. Key advantages of flexible adaptive designs include complete flexibility to react to unanticipated events, and options exist to introduce any new doses, change endpoints etc.
• Ad hoc changes based on unblinded data may jeopardize the credibility of the study
• Increase risk of adapting trials too early, thereby jeopardising the overall study findings
• Questionable regulatory acceptance
What is the motivation for an adaptive design?
Adaptive design methodology has the potential to increase the efficiency of trial design and minimise exposure to potentially harmful and un-efficacious experimental treatments. From a scientific perspective, key motivators include ability to drop ineffective doses, improve understanding of the disease process (eg, change visit schedules or dosing regimens) and improve the measure of disease outcome (eg, change endpoint). Further examples are given in Phillips and Keene1.
There are also commercial advantages associated with adaptive designs. These include ability to respond to changes in market (eg, approval of a competitor product) and increased investment via reducing the risk (eg, increase power and sample size).
Opportunities for interim trial design modifications, and the prerequisites, problems and pitfalls that need to be considered as soon as any kind of flexibility is introduced into confirmatory clinical trials is discussed in published regulatory guidelines2. It is important to note that adaptive designs should not be viewed as a means to alleviate the burden of rigorous planning of clinical studies nor, as the saying goes, to be able to look for unimportant effects when the study fails to provide the desired results.
What are the key issues when designing an adaptive trial?
Numerous concerns have been highlighted regarding the use of adaptive designs, as discussed by Flemming.3 These range from loss of statistical efficiency, interpretability of results, risks associated with basing design changes on unreliable interim estimates of efficacy, and issues regarding compromising the integrity and credibility of the trial. This later point is of particular concern to regulators. However, Data Monitoring Committees (DMC) are potentially well placed to address the issue and take a leading role in the implementation of adaptive design studies.
What is a Data Monitoring Committee?
As discussed in the CHMP guideline on Data Monitoring Committees4, a DMC is a group of independent experts external to a study who have responsibility for assessing the scientific and ethical integrity of the trial. Their primary role is to protect patient safety. Sponsors and/or investigators are responsible for the trial. DMCs make recommendations to the sponsor. Implementation of any recommendations is solely the responsibility of the sponsor.
A DMC typically comprises physicians, statisticians and/or ethicists. Rarely is a Project Manager involved. Frequently the Project Management tasks are picked up by the Biostatistician, more on this later. Finally for most DMCs an independent statistician is appointed to provide unblinded information relating to the study, but they are not typically voting members of the DMC.
What role do DMCs have in implementing adaptive designs?
DMCs are ideally positioned for implementing rigid adaptive designs, where the adaptations are pre-planned and defined in the protocol. Members of the committee have the technical expertise to implement such changes and can provide assurance of the scientific validity of any adaptation.
For flexible adaptive design studies an increased skill set is required. To implement flexible adaptive design studies strategic issues relating to the drug development program need to be considered, not simply the scientific validity of a single study. That is, the probability of regulatory success needs to be considered prior to undertaking any adaptation. Also the commercial implications of the adaptation need to be understood. For example the adaptation needs to be considered in context of the target product profile and the associated financial returns. Subsequently, implementation of flexible adaptive designs are probably better suited for internal sponsor committees or Contract Research Organisations. If an internal sponsor committee is utilised it is critical that a different team than the one executing the study is used in order to safeguard the scientific integrity of the study results. Even with such safeguards, concerns may still exist about the credibility of the data. On the other hand, would sponsors be willing to offer Contract Research Organisations a blank cheque to implement adaptations requiring additional resources in such a way that keeps sponsors completely blinded?
Should Project Managers be DMC members?
Traditionally the independent statistician has been utilised to provide logistical support for DMCs; that is, organization of meetings, scheduling etc. However, this is not cost effective, nor best use of their skill sets. Within organizations how often are experienced professionals used to sort out missed flights, wrong contact numbers, unsent documents. A rethink is needed. An obvious solution is to routinely include project management staff as a member of the DMC. Project Managers are professional co-ordinators.
Adaptive designs allow for major design modifications of ongoing clinical trials based on the observed data. DMC are ideally placed to implement rigid adaptive designs, where the focus is on technical expertise. Internal sponsor committees or Contract Research Organisations may be better suited to implement flexible adaptive designs where a broader range of skills is required. Inclusion of Project Managers will make more efficient DMCs.
Alan J Phillips(email@example.com) is Vice President of Biostatistics and Programming, Europe at ICON Clinical Research. Charles Du Mond is Senior Vice-President, US Biostatistics and Programming at ICON Clinical Research Inc.
1. Phillips A.J. and Keene O.N. (2006) Adaptive Design for Pivotal Trials: Discussion Points from the PSI Adaptive Design Expert group. Pharmaceutical Statistics, 5, 61-66.
2. Reflection Paper on Methodological Issues in Confirmatory Clinical Trials with Flexible Design and Analysis Plan.
3. Flemming T.R. (2006) Standard vs. adaptive monitoring procedures: A commentary. Statistics in Medicine, 25: 3305-3312