For many years, adaptive trial designs have suggested a way to make clinical trials more efficient. Adaptive designs present the prospect of dynamic protocols that might expand, contract, change direction, or stop based on real-time study results. The FDA endorsed adaptive designs in clinical trials as part of its Strategic Path Initiative in 2004. The agency then released a draft guidance document for adaptive designs in studies of drugs or biologics in 2010 and in May 2015 issued a companion document for devices. The content of this latest document reflects how the FDA seeks to balance the benefits of a well-designed adaptive trial against the concerns over a poorly-designed one.
Defining Adaptive Trial Design
In its May guidance, the FDA defines adaptive trial design as a study protocol that “allows for prospectively planned modifications based on accumulating study data without undermining the trial’s integrity and validity.” That phrase demonstrates some key features of how the FDA categorizes and views adaptive design strategies:
- An adaptive trial is dynamic by design; the protocol anticipates one or more changes in the trial’s direction or scope.
- The information for those changes will come from information generated within the study, not from external sources.
- The changes, if implemented, must not alter the scientific integrity of the research. Salient concerns for the FDA are decision points that could inflate positive responses, create false negative responses, or expose blinded data to the wrong people.
Progress of Adaptive Designs
A 2013 study calculated that about 20% of Phase III trials utilized some kind of adaptive design. According to the CSDD report, the most common adaptive strategy is “futility stopping,” where trials are cancelled if the treatment fails to reach a pre-determined level of effectiveness. The report added that sponsors have been slower to adopt more complicated designs, with less than 10% of clinical studies trying strategies such as “seamless” transitions from one stage of research to another or dynamically-designed dose-finding.
The article said that “uncertain commitment from regulatory agencies” may be preventing wider adoption of adaptive trial designs. If true, the FDA’s growing library of advice, including this latest discussion about device research, could allay that uncertainty.
Advantages of Adaptive Designs
In its draft guidance, the FDA states that well-designed adaptive trials can make studies shorter and more efficient. It provides some examples of how building in the potential to adapt can help a protocol:
- Early success: If a trial’s data shows success at an early, predetermined point, the FDA may grant approval with that smaller set of data, saving the time and expense of a longer trial.
- Early failure: Similarly, a sponsor can save important resources if the adaptive design allows the study to stop after falling short of a set target.
- Study size: Adaptive trials present the potential for better controlling how many people have to enroll in a study. If the trial’s statistical models can demonstrate an effect is large enough earlier than scheduled, enrollment can stop there and more participants are not needed.
Taking a lesson from experience, the 2013 Tufts CSDD article noted that one large research company saved $350 million over five years by using adaptive trial strategies.
And the FDA’s guidance argues that these savings in trial commitments have an ethically positive element as well. Strategies that can get a treatment approved sooner, shorten the time a participant has to spend in an ineffective treatment, or limit the number of people needed for an experiment all create a better environment for study participants.
Choosing Adaptive Designs or Strategies
One challenge for adaptive trials is finding the right strategy for the right trial. The FDA offered some examples of designs that could benefit device trials:
- Group Sequential Designs: This category includes futility stopping, the most common type of adaptive trial. A study can incorporate data checks at certain, sequenced points to see whether study results have reached targets for success or failure.
- Sample Size Adaptation: With proper methodologies, an adaptive trial design can check whether a study has recruited enough participants to demonstrate efficacy and possibly preempt further enrollment.
- Dropping a Treatment Arm: In studies with multiple arms, an adaptive protocol can plan to end one or more arms if they prove ineffective.
- Changing the Randomization Ratio: In a two-arm study, if a treatment appears to be significantly more effective than a control, pre-planning can anticipate shifting how many participants enroll in each.
- Planning to Adapt Based on the Total Information: Here an adaptive design can establish a certain amount of data as a study’s endpoint, rather than a point in time.
- Seamless Studies: For devices, the FDA suggests this design could work best in moving from a successful feasibility study to a pivotal study without approvals or participant recruitment. The FDA does consider this as a complicated strategy, and encourages careful coordination with the agency while setting it up.
The FDA does caution sponsors about some elements of using adaptive trials. Primarily, the FDA warns against result-based changes that could weaken a study’s merits.
Observer Biases: For adaptive designs, it’s clear from this guidance and its 2010 counterpart for drugs and biologics that the FDA will focus on the dangers of study biases or false signals. The FDA’s most prominent concern appears to be misinterpreting data as too successful (that is, a Type I, or false positive, error). Another area of attention is “operational bias,” where knowledge about study design, the study population, and ongoing results could affect how researchers conduct the study. The FDA has these concerns with any trial design, of course, but adaptive trials—with the possibility of such biases combining with study results to trigger an unwarranted change—seem to earn extra scrutiny.
Who approves the changes to the trial? The prospect of those biases leads directly to the critical issue of who decides that results merit a study change. The draft guidance suggests some possibilities for decision-makers, with the common feature that whoever approves the change should be isolated from the conduct of the study. The guidance document proposes a person or organization such as a data monitoring committee (DMC), an independent statistician, or a contract research organization (CRO).
The IRB’s Role in Adaptive Design Trials
The draft guidance for devices includes a section about an IRB’s oversight of an adaptive design study and raises two main issues: what to include in the consent process and how to manage protocol amendments.
The FDA suggests that with advance notice and coordination, an adaptive design-driven change does not have to mean a full protocol amendment from the IRB or the reconsenting of all participants. If an ethics committee is comfortable with adaptive design concepts in general and with the structure of a study in particular, then the anticipated shifts within the design could happen with minimal disruption. And a study might avoid widespread reconsenting requirements if the initial consent process includes a clear description of what the participant might experience during any studies in the trial.
Adaptive trial design has the potential to save time and resources, but it still presents challenges in design and implementation. If you are considering an adaptive study design, Quorum Review has experience with adaptive trial design, with device trials, and with the consent processes for each.
“Adaptive Clinical Trials for Overcoming Research Challenges,” News Medical, September 16, 2013; accessed 8/18/2015
Device Guidance, May 2015
Q&A on adaptive trials