The FDA has typically used real-world evidence (RWE) as a way of monitoring safety issues post-approval, especially through the Sentinel Initiative, which started in response to the FDA Amendments Act of 2007.
Nearly a decade ago, President Obama signed the 21st Century Cures Act, which among other things, mandated that the agency explore broadening its use of RWE to include efficacy data used in regulatory decisions.
"The timing is right for this given that there are so many sources of real-world data now, especially including electronic health record data," Andrew Norden, chief medical officer at Cota Inc., told BioWorld Insight. "There is potential to use that data at a scale not possible before."
In December of last year, the FDA issued a framework outlining the considerations for evaluating RWE. Last month, the agency held a workshop to discuss the topic.
The movement has already seen isolated success. In April, the FDA approved the use of Ibrance (palbociclib, Pfizer Inc.), in combination with an aromatase inhibitor, in men with hormone receptor-positive, human epidermal growth factor receptor 2-negative advanced or metastatic breast cancer. The approval was based on RWE stored in the Iqvia Insurance database, Flatiron Health Breast Cancer database and Pfizer's global safety database.
Despite all the movement in the right direction, the FDA is still largely in prove-it mode when it comes to RWE and still exploring the advantages and challenges with using data gleaned from real-world sources.
There's a clear cost savings to using an existing data source compared to establishing a new custom-built dataset each and every time a company wants to answer a clinical question. And for diseases that affect limited populations, such as Ibrance in male breast cancer, the real-world dataset might be the only way to reasonably glean enough data to make an informed decision.
On the challenges side, "there's a real lack of interoperability, such that data sources that exist in the world don't generally talk to each other very effectively," Norden noted, adding that they also tend to be less structured, which can be a problem if the records don't include standardized assessments typically used as endpoints in clinical trials.
Real-world data sources can also be biased in their patient selection because there's typically a reason that a physician might choose to use one drug over another. "It's not been traditionally easy to tease that out and control for it when working with real-world data," Norden said, although he's seen databases become more granular over the years, which allows researchers to adjust for some of the risk factors.
In May, Boston-based Cota established a two-year research collaboration agreement with the FDA's Information Exchange and Data Transformation Program (INFORMED) to explore the use of RWE in oncology precision medicine.
Cota and INFORMED plan to start with early stage breast cancer, looking at prognostic indicators, especially molecular biomarkers, to determine if they can replicate the findings of multiyear clinical trials using only real-world data. The project is shooting for having output that can be published within a year.
Likewise, New York-based Flatiron Health Inc. is also working with INFORMED through a collaboration that started in 2016 and was renewed earlier this year for an additional two years. The project started with advanced non-small-cell lung cancer patients treated with immunotherapy. Data published in The Oncologist showed that patients in the real-world cohort had a shorter overall survival indexed to PD-1 inhibitor initiation than reported in clinical trials. The group also published a second paper in The Oncologist highlighting that real-world patients treated with PD-1 inhibitors tended to be older and more had smoking history relative to patients enrolled in the clinical trials.
Bringing the trial to the patient
Science 37 Inc. has taken a different approach to gathering RWE: It's helping drugmakers design and run clinical trials at virtual sites in patients' homes. By eliminating the traditional research center, the company has increased patient access to clinical trials, since they don't have to live near a clinical trial site, and has limited the burden on patients because they can interact with study investigators in their own home through telemedicine. Science 37 has also seen an increase in diversity of the socioeconomic status of patients participating in decentralized studies.
"The FDA is committed to supporting trials moving outside the boundaries of traditional clinical research at clinical research centers," Christopher Ceppi, chief product officer at Science 37, told BioWorld Insight.
But Science 37, of Los Angeles, has seen challenges getting some sponsors on board, especially when there might be a digital team at a company championing the idea, but the individual clinical team takes a not-in-my-study approach. "In cases where we've seen it work, we've seen alignments in the board level through executives through clinical operations and drug development," Ceppi said.
For larger studies, there can be a cost savings using decentralized studies, and for smaller studies, especially in rare populations, the advantage is primarily the ability to enroll enough patients. "Some studies can't be run under traditional trial design. Or the patient burden is going to be such that it's going to be excruciatingly hard to find patients," Ceppi said, pointing to studies where mobility is an issue for patients.
Moving the study from primarily collecting data at a research site to collecting it in the home increases the opportunity to gather more frequent data points, which can give better insight into the health status of the patient without a subjecting view. Ceppi called it the "holy grail" of trial design, although he noted that "we're not there yet."
Like health records, the use of technology to record data in real-world settings of clinical trials introduces new obstacles of interoperability between devices used in a study. Also not all technology, such as wearables, are designed under regulatory guidelines that would allow them to be used in clinical trials.
As the amount of data collected increases, there's also issues with scale and signal to noise. "One of our sponsors collected more data from one study than they had collected in the lifetime of their company," Ceppi recalled.