Synthetic control arms – use of RWE in clinical trials
Randomized clinical trials (RCTs) are the ‘gold standard’ for evaluating the efficacy and safety of drugs and treatments.
However, RCTs have a number of limitations, including large recruitment numbers needed, participants’ fear they will end up receiving a placebo, time and cost. In recent years, there has been a growing interest in restructuring RCTs’ framework, particularly in finding alternative approaches to collecting comparison data. This means that rather than having patients receiving the placebo or the standard-of-care drug, placebo arms are modelled using information from real-world data that has previously been collected. These synthetic control arms use data collected from sources including electronic health records (EHR), medical claims, data generated from fitness trackers or home medical equipment, disease registries, laboratory test results, and historical clinical trial data.
The concept grew with the ability to store and manipulate large datasets and gained much attention from both pharmaceutical companies and regulators, such as the FDA and EMA. In 2017, the FDA approved Merck’s Bavencio (avelumab) for the treatment of metastatic Merkel cell carcinoma, based on a single-arm trial and a synthetic comparator arm which used historical control of matched patients. Roche used synthetic control data to expand access to Alecensa (alectinib), a treatment for non-small-cell lung cancer. The use of synthetic control data accelerated the access to Alecensa for patients in 20 European countries. Amgen used historical data from patient records in the US and the EU for the approval of Blincyto (blinatumomab) for the treatment of a rare form of leukemia.
The benefits of using synthetic control arms for the pharmaceutical industry and patients are numerous: lower trial costs, reduced delays and bringing therapies faster to patients. However, it is important to acknowledge some limitations. The generation of synthetic controls can be challenging: the data may be difficult to collect or of low quality, the data has to be integrated from multiple sources which requires collaborating with multiple stakeholders. However, there are certain situations where the benefits outweigh the risks, e.g. in case of rare diseases and diseases with high unmet need, particularly with the increasing number of patient subpopulations defined by specific genetic mutations or biomarkers. Furthermore, the use of synthetic controls is beneficial in areas where control group performance is well characterized historically, and results are generally consistent from trial to trial, where standard-of-care is well defined and stable, or where there are objective endpoints that are easy to measure.
The use of real-world evidence for generating synthetic control arms is still in its early days. New technologies, such as machine learning, will be instrumental to extract relevant information, ensure quality of data, and ultimately enable us to see the full potential of this approach. While synthetic control arms will never completely replace RCTs, there is great value in leveraging real-world evidence. As with any new approach, particular care is needed to achieve high-level accuracy and all stakeholders involved must work together to define standardized evaluation criteria for ensuring quality.