In April 2020, while most industry focused on the COVID-19 pandemic, the FDA granted an unusual approval for a rare pediatric disease. Based on the Phase II Sprint trial, regulators determined that selumetinib demonstrated a clinically meaningful improvement in overall response rate for children with neurofibromatosis type 1 (NF1). However, SPRINT had no placebo control.
Instead, SPRINT used an external control arm, also known as a synthetic arm, comprised of 50 patients from two previous studies. The first set of historical data was a previous natural history study of NF1, and the second was the placebo arm of a previous clinical trial in NF1 for an entirely different drug.
The final analysis compared all 50 patients receiving selumetinib in SPRINT to 50 matched patients from the two historical studies. Selumetinib, which AstraZeneca and Merck market as Koselugo, is now the first FDA-approved treatment for NF1.
“The selumetinib approval happened at the height of the pandemic, so not enough people paid close attention,” says Dr Vivek Subbiah, a clinical trialist at MD Anderson in Houston, Texas. “This was a prime example of how external controls can generate regulatory approval.”
Going forward, experts say external arms could replace placebo arms in more and more RCTs. In a trial with an external control arm, patients receiving treatment are matched with patients from historical studies in the same indication and with similar baseline characteristics. As a result, trials can measure the relative safety and efficacy of a treatment to normal disease progression, without necessarily needing a placebo arm.
Though external control arms could bolster trials of the future, experts say randomised, placebo-controlled trials are still the gold standard of clinical evidence. External control arms require meticulous data-matching along demographic and disease characteristics to achieve statistical rigour, they add.
“External arms are becoming more common in the pharmaceutical sector, but it’s important to know these are not silver bullets,” says Arnaub Chatterjee, senior vice president of Medidata . “You have to be incredibly judicious on how you use synthetic controls and where it makes sense.”
When to use external control arms
In some cancers and rare diseases like NF1, the patient population is so small that a placebo arm is too costly or unrealistic to recruit, Subbiah explains. For cancers with no approved therapies and constantly evolving standards of care, placebos can even become unethical, he adds.
In addition, external control arms could work well in indications with large pools of existing standardised RCT data, such as pain trials or depression trials, adds Kathryn Hall, PhD, a drug development researcher at Brigham and Woman’s Hospital. However, placebo-controlled trials should always be used for viral or bacterial infections, where only placebos can account for disease proliferation, she adds.
Although regulators have demonstrated growing interest in external control arms, they usually will not accept this trial design when placebo patients are readily available and easy to recruit, Chatterjee adds. But outside the regulatory setting, sponsors are using external arms for internal validation and early proof-of-concept studies across many different indications, he notes.
To date, the majority of external control arms have been in oncology, according to GlobalData’s Clinical Trials Database. Some 28% of drug trials utilizing external arms over the past decade have been in oncology indications, followed by 14% in genetic disorders. Meanwhile, infectious diseases accounted for 11% of external control arm trials. GlobalData is the parent company of Clinical Trials Arena.
Comparing apples to oranges?
The main challenge of external control arms is harmonizing data across a wide range of previous trials, Hall explains. When selecting previous clinical trial data, details like the duration of the placebo arm or the number of patient drop-outs can significantly affect results, she adds. Without rigorous matching and statistical analysis, comparing a treatment group to an external control is like “comparing apples to oranges,” she says.
When generating a pool of patients for an external control arm, sponsors should take patient-level data from multiple previous clinical trials that meet stringent selection criteria, Chatterjee adds. External control patients can also come from real-world data (RWD) sources, such as electronic health records or patient databases, he notes. However, RWD is often unstructured and ripe with missing datapoints, requiring high confidence in patient comparability, he says.
Meanwhile, some sponsors are pursuing hybrid approaches to external control arms, which supplement a smaller placebo arm with added patients from historical trials, Chatterjee explains. At Medidata, Chatterjee worked with Medicenna Therapeutics to design a Phase III trial in glioblastoma using this hybrid approach. Meanwhile, other glioblastoma studies have used the “window of opportunity” Phase 0 design to streamline drug development.
To increase the feasibility of external control arms, regulators and industry should come together to standardise trial designs for specific indications, Hall says. For example, all hypertension trials could agree to measure blood pressure at specific intervals over an eight week period, making it easier for future trials to draw from previous placebo pools, she notes.
However, Hall says trial standardization will be costly, requiring significant buy-in from major pharmaceutical players. With tensions over primary endpoint tensions in disease areas ranging from plaque psoriasis to amyotrophic lateral sclerosis (ALS), widespread trial harmonization could prove a tall task.
Will pharma breakthrough?
Though the selumetinib approval was a major step toward new trial approaches, the field is a long way away from widespread implementation of external control arms, Subbiah says. Randomised, placebo-controlled trials will always play a major role in drug development, but they cannot answer every major clinical question, he adds.
Though the prevailing strategies for drug development are logically sound, the field should still think bigger for strategies to revolutionise clinical trial design, Hall explains. External control arms, she says, could be the new idea that the field is looking for.
“Clinical trials in general need some kind of major shift,” she says. “Whether it’s a new genomic angle or a different model for using placebos in trials, we’re in need of a breakthrough idea.”
- External control arms can replace placebo arms with historical clinical trial data, and sometimes other RWD, that meets stringent patient matching criteria.
- External control arms are most common in oncology and rare diseases, where placebo arms are often cost-prohibitive or unethical. Meanwhile, placebo-controlled trials are still the gold standard for major indications with readily available trial participants and for infectious diseases.
- The greatest challenge for external control arms is harmonizing data across different historical clinical trials and pools of RWD. Regulators and industry should make a conceited effort to standardise trial designs to facilitate future external control arms.
Previously, Clinical Trials Arena broke down the trends, sponsors, and strategies driving growth in real-world evidence studies.