Adaptive Clinical Trials: What Are They and Why Do They Matter?
Traditional clinical trials follow a fixed plan from start to finish. But what if a trial could change course based on data collected along the way? Adaptive designs allow exactly that, making trials faster, more efficient, and more ethical.
This video from Clarity Clinical Solutions explains how adaptive designs work and why they are becoming standard.
What an adaptive design is
An adaptive design is a clinical trial that uses accumulating data to modify aspects of the study without undermining its validity. The adaptations are not ad hoc. They are pre-specified in the protocol.
At pre-planned points, an independent committee looks at the data and may continue as planned, stop early, modify doses, or change the sample size. The key principle: adaptive designs do not mean the trial is unreliable. They mean the trial is designed to use information efficiently.
Why adapt?
Three reasons:
- Efficiency. Adaptive trials can reach answers faster by dropping ineffective arms early or combining phases.
- Ethics. Fewer patients are exposed to ineffective treatments, and more receive the better treatment.
- Flexibility. When initial assumptions are wrong, adaptive designs can adjust without restarting.
Types of adaptive designs
Sample size re-estimation adjusts the number of patients based on observed variability. Sample size calculations rely on assumptions about variability and effect size, which are often wrong. At an interim analysis, the blinded data is used to re-estimate the required sample size. This was used in over 50% of recent cardiovascular outcome trials.
Adaptive dose finding adjusts doses during the trial. The continual reassessment method adjusts the dose for each new cohort based on observed toxicity. Bayesian dose-finding uses prior knowledge to estimate the optimal dose. Adaptive dose-finding can reduce phase 1 duration by 30 to 50% and expose 40% fewer patients to subtherapeutic doses.
Seamless phase 2/3 designs combine dose-finding and confirmation into one continuous trial. The trial starts with multiple dose arms. At an interim analysis, the best dose continues into phase 3 while existing patients contribute data to the final analysis. This saves 12 to 18 months of development time.
Multi-arm multi-stage (MAMS) designs test multiple treatments simultaneously against one shared control. At each interim analysis, arms that show insufficient activity are dropped. Instead of running five separate trials, MAMS runs one trial with shared controls. The STAMPEDE trial for prostate cancer tested six treatment strategies against one control, identifying two effective treatments and dropping four.
Response adaptive randomization changes the randomization ratio based on which arm is performing better. As data accumulates, more patients get the better treatment, up to 80 to 90%. This is called a "play the winner" design. It was famously used in the ECMO trial for respiratory failure, where the adaptive ratio reached 90% in favor of ECMO.
Drop the loser designs use futility analysis to eliminate ineffective treatments early. If a drug is unlikely to show a meaningful benefit, that arm is terminated. The opposite, early success stopping, stops the trial if a treatment is so clearly effective that continuing would be unethical. These designs use alpha-spending functions to preserve the overall false positive rate.
What regulators say
The FDA published comprehensive guidance on adaptive designs in 2019 (FDA: Adaptive Designs for Clinical Trials of Drugs and Biologics). Key conditions:
- Adaptations must be pre-specified in the protocol
- The overall type 1 error rate must be controlled
- The FDA strongly encourages early dialogue
In 2023, about 35% of new drug applications to the FDA incorporated at least one adaptive design element, up from less than 5% in 2000.
Famous examples
I-SPY 2 is the most famous adaptive trial, an ongoing phase 2 platform trial for high-risk breast cancer. It uses a MAMS design with response adaptive randomization. Multiple drugs are tested simultaneously against a shared control. Promising drugs are graduated to phase 3, ineffective ones are dropped. I-SPY 2 has screened over 20 agents and cut the time to identify effective drugs by two to three years.
The RECOVERY trial changed the pandemic. Launched in March 2020, it was a pragmatic, adaptive platform trial testing multiple COVID-19 treatments. It enrolled 47,000 patients across 200 hospitals. Dexamethasone was shown to reduce mortality by one-third. Hydroxychloroquine and lopinavir were dropped for futility. The adaptive design found effective treatments in months, not years.
The challenges
Adaptive designs are not free. They come with:
- Statistical complexity. They require sophisticated methods to control error rates.
- Operational challenges. Real-time data collection and fast analyses require advanced infrastructure.
- Operational bias risk. If sites learn interim results, it can bias enrollment.
- Regulatory uncertainty. Some sponsors still prefer traditional designs for certainty.
The bottom line
Adaptive designs use accumulating data to make trials faster, more efficient, and more ethical. From I-SPY 2 in breast cancer to RECOVERY in the pandemic, they have proven they can find answers in months, not years. The FDA supports them, and their use is growing quickly.
The future of clinical trials is adaptive.
This article is based on the Clarity Clinical Solutions video "Adaptive Clinical Trials: What Are They and Why Do They Matter?" Watch it here: Adaptive Clinical Trials: What Are They and Why Do They Matter?
References
- Clarity Clinical Solutions — "Adaptive Clinical Trials: What Are They and Why Do They Matter?" (video, source of the design types and trial examples). https://www.youtube.com/watch?v=jO3fyMlmTHc
- FDA — Adaptive Designs for Clinical Trials of Drugs and Biologics (2019 guidance). https://www.fda.gov/regulatory-information/search-fda-guidance-documents/adaptive-design-clinical-trials-drugs-and-biologics-guidance-industry