Life Sciences R&D & Medical

Why external control arms should be in your clinical trial toolbox

By Manu Garg, Qin Ye, Abhinav Bansal, and Esha Halabe

Nov. 15, 2021 | Article | 6-minute read

Why external control arms should be in your clinical trial toolbox

Imagine a patient with a malignant brain tumor and multiple cardiac comorbidities who enrolls in a clinical trial to receive a new experimental treatment developed to prevent tumor-related seizures. As part of the patient’s enrollment, her presence is mandatory during follow-up visits at the trial site. During COVID-19, that means sitting in waiting rooms alongside potentially sick people, putting her life in jeopardy; even so, she feels that the risk is worth it. But when the trial is complete, she finds out that she was part of the randomized control arm and received a placebo.


In cases such as this one, where a randomized control trial (RCT) is less than ideal, trial sponsors should consider a different way to collect comparison data. One way to do this is by creating an external control arm (ECA), which substitutes a traditional placebo or control group with real-world data (RWD) collected from existing medical records, disease registries, patient-generated data, claims forms or other data sets. Under this arrangement, the cancer patient with comorbidities—and many people like her—could  gain access to new treatments.


ECAs, sometimes also referred to as synthetic control arms, can help accelerate timelines and reduce the cost of clinical research, which can be a key differentiator in a competitive market. They can enable clinical research in oncology and rare diseases where conventional RCTs aren’t feasible or ethical and help bring innovative therapies to the market faster. They also can shorten research and development processes. Even regulators are seeing such use cases for ECAs favorably, especially when used to complement single-arm trials. Most importantly, ECAs help reduce the burden placed on high-risk patients that enroll in clinical trials hoping to receive the experimental drug, only to miss out.

When the gold standard falls short

RCTs are considered the gold standard of clinical evidence and there are very good reasons for that—especially when it’s feasible to enroll a high volume of patients across the treatment and control arms. RCTs are designed to capture detailed information about a patient’s condition and provide a reliable method for testing the effects of a new therapy on patients.


However, with advancements in our understanding of disease heterogeneity and precision medicine, the population of eligible trial patients is becoming smaller. It may be difficult to enroll enough patients to support a fully powered RCT in studies that evaluate treatments for rare conditions. What’s more, in uniformly fatal diseases, it may be unethical to put patients in the control arm. To overcome these challenges, clinical researchers are increasingly evaluating ECAs.

ECAs hasten R&D timelines

Roche effectively applied this trial design in a study that evaluated the comparative effectiveness of the anaplastic lymphoma kinase blocker alectinib versus ceritinib using a real-world, data-enabled ECA. Roche derived the alectinib treatment arm by extracting existing data from two phase II studies. For the ceritinib treatment arm, Roche applied inclusion and exclusion criteria to data pulled from an electronic medical record database.


“Rather than waiting for phase III results, Roche used a synthetic control arm of 67 patients to provide the necessary evidence of relative performance. The decision to use a synthetic control arm advanced coverage of alectinib by 18 months in 20 European countries,” STAT News reported. They also reported that Amgen gained accelerated approval in the EU for blinatumomab for the treatment of a rare form of leukemia, using an ECA.


“In the absence of comparative evidence from RCTs, external controls are often used to bridge the gap of providing comparative evidence using direct adjusted comparisons,” noted the authors of an analysis of Roche’s alectinib versus ceritinib in the Journal of Comparative Effectiveness Research.

Growing regulatory support for real-world data

As sponsors evaluate the feasibility of an ECA-powered trial and develop study protocol, engaging with regulators early on is key the acceptance of  such trials. The 21st Century Cures Act, which was signed into law in late 2016, was a major milestone in this direction. The Cures Act builds on the FDA's ongoing work to incorporate the perspectives of patients into the FDA’s decision-making process during the development of drugs, biologics and devices. Additionally, it enhanced the industry’s ability to incorporate real-world evidence into clinical trial design. After the law was enacted, the FDA released guidelines that provide clarity on what would be acceptable use of RWD as part of clinical evidence.


Earlier this year, the FDA issued additional guidance about the use of RWD in regulatory decision-making, as part of its ongoing Real World Evidence Program. The guidance discusses several aspects surrounding the use of electronic health records and medical claims in clinical studies, such as the selection of appropriate data sources, development and validation of study design elements and data provenance and quality. The guidance also underscores the need for sponsors to push data partners to meet the demands of regulatory-grade RWD, which includes ensuring that data are accurate, complete, traceable and that there’s a sufficient number of representative patients for the study. Additionally, the guidance provides more clarity on the elements to be included in protocols and statistical analysis plans.

Medical record granularity, AI boost data availability

Another aspect that sponsors should pay careful attention to when designing a trial using ECAs is the availability of RWD obtained through patient medical records. ZS’s experience with a broad array of real-word data sets has been that some data points are more readily available in medical records than others. For example, baseline organ function tests and standard of care biomarkers are well documented in medical records because they’re captured for a broad range of patients. However, data on HIV tests, hepatitis, cardiac function and investigational biomarkers aren’t usually tested for in the general population and are thus captured more sparsely in routine care. Despite these challenges, the granularity of information has been consistently improving over the past decade and can meet the requirements of many studies.


Further, with the advancements in natural language processing, advanced machine learning and artificial intelligence, it’s increasingly feasible for investigators to curate physician notes reliably, giving them a deep understanding of patient’s medical history. These techniques allow them to derive additional insights about a patient that aren’t apparent in a record’s structured data, such as whether they have controlled brain metastasis or if they’re stable on concomitant medications that require careful monitoring.

ECA data requires careful approach

The validity of ECA patients lies in their exchangeability with the internal trial patients. Exchangeability refers to how well the comparator group provides an approximation of the treatment group’s disease experience. Exchangeability can be adversely affected by selection bias and factors such as misclassification, improper measurement and missing data. To help account for these influences, data scientists have developed frameworks that help evaluate sources of potential bias. One of those frameworks describes statistical methods that can help alleviate bias in studies that use ECAs.


Despite the promise shown by recent successful implementation of ECAs, it’s analytically challenging to develop ECAs that provide a true comparator to the treatment arm. Doing so requires a carefully designed process, from study conceptualization and execution of a statistical analysis plan to thoughtful consideration of data sources.

Keeping ECAs on the table

When asked, “What’s the optimal role of ECAs in clinical research?” I suspect that many in our field would say, “Use RCTs to the extent possible and then use ECAs for everything else.”


However, it’s my hope that the industry can become sensitive to the notion circumstances should dictate the approach sponsors take. And the processes I’ve outlined above—such as determining when a RCT creates an undue patient burden and thoroughly vetting RWD sources—were designed to help future trial sponsors better identify opportunities for an ECA.


Figuring out the right balance here promises to be a difficult and dynamic process, but it’s a challenging problem that’s well worth solving.

Add insights to your inbox

We’ll send you content you’ll want to read – and put to use.