As the clinical development landscape changes rapidly, so does the interest in ways trial sponsors can accelerate research and development programs while reducing inefficiencies, time and related cost. According to ClinicalTrials.gov, in 2020 alone, there were nearly 400 trials using adaptive designs. In comparison, from 2005 to 2018, there were less than 300 trials in total using these approaches.

The rapid increase in adaptive trials is likely tied to the FDA guidance on complex and innovative trials. Adaptive trials preserve the statistical validity of the analysis at the conclusion of the trial, even with the trial adapting to incoming data. Because these trials can respond to the data, sponsors are able to make better use of the data, which increases study efficiencies and accelerates treatment development, while also maintaining scientific and statistical integrity.

Adaptive designs can be applied across all phases of clinical research, but depending on the stage of clinical development, certain designs may be more useful than others. Guidance from experienced biostatisticians is critical to pairing the adaptive design option with development phase and trial objectives. We will briefly talk about some common adaptive design types, and their uses.

Adaptive design options

• Group sequential (typically for Phase 2 or 3 studies): This design has been accepted by regulatory agencies for decades and has its statistical properties worked out in closed form; in this design, the number of patients is not set in advance, but is allowed to expand based on interim results of the trial. The interim analysis or analyses will occur at some intermediate point(s) in the recruitment process. Each interim allows for one of 3 options:

1.  Stop the trial for futility

2.  Stop the trial for efficacy

3.  Continue the trial with some sample size, either determined in advance or based on the data.


In this design, molecules that do not show therapeutic benefit or are showing sizeable benefit can stop early, and only those therapeutics where the information is equivocal need to continue. For molecules that are nontherapeutic or supertherapeutic , we can stop the study early thereby not spending resources unnecessarily. When the data are equivocal, this design will determine the right size to achieve the desired power, thereby reducing risk of trial failure.
• Continuous reassessment model (CRM) (for Phase 1 dose finding): The CRM estimates the maximum tolerated dose (MTD) of new therapies. It has been shown to be more accurate in targeting MTD than the long-used 3+3 design. Additionally, this type of study has shown to secure more study participants at or close to MTD than the 3+3 design. A strong competitor is the I3+3 design, which mimics the operating characteristics of CRM while maintaining the simplicity of 3+3 designs. There are a host of competing design options that can also be employed. We recommend optimizing the design for the assumed dose-toxicity relationship. Often, we want designs that perform well over a range of possible dose-toxicity relationships.
• Seamless Phase 2/3 or 1/2: In combining both dose selection and confirmation phases in one trial, this design provides easier transition from learning phases into the confirmation phase. In the earlier phase, a number of investigational doses (say 3 for Phase 2 portion of the 2/3 design) are investigated, and the most promising is carried into the latter confirmatory phase. This design can be either operationally seamless, which is statistically simpler, or fully seamless. The fully seamless option allows for sharing of information from the earlier phase with the latter phase to use the information more efficiently. However, the cost is that  needs to be adjusted to compensate.
• Dose ranging (Phase 1 or 2): This design allows study teams to estimate the dose-response relationship efficiently. A primary objective is finding the best dose to carry into the next phase. Key advantage: If the assumptions on the dose-response relationship are incorrect, this design will compensate, whereas a fixed design will not. Our experience indicates that even in the case where our team’s assumptions are correct, this design still performs competitively with a fixed design. Heads, you win; tails, you win.

There are several new but underutilized adaptive design studies that are available to sponsors and worth consideration:

• Precision dosing: Further enhancing dose finding, precision dosing trials predict a patient’s response to a particular dose based on their individual characteristics and disease state. As data are collected, updates are made to better estimate the right dose for individual patients, reducing variability, and allowing for quicker readouts. This design requires involvement of pharmacokinetics modeling.
• External comparator: This design allows sponsors to use aggregated, real world data from external sources to strengthen comparative context into studies, which helps reduce sample sizes and potentially accelerates informed decision making.

IQVIA’s statisticians have been designing and executing adaptive designs for over 20 years. Though adaptive trial design may take longer in upfront planning, through high level expertise and tech-enabled solutions, these models can help reduce overall timelines and provide seamless transitions from phases of development while ensuring adaptations made in the process are helping to maximize the probably of trial success, better meeting patient needs and sponsor goals.

Please contact GlobalFSPGTM@iqvia.com directly to learn more about IQVIA capabilities specific to adaptive trial design, please visit iqvia.com/biostatistics