Abstract
To demonstrate the potential of real-world evidence generated from real-world data, Foundation Medicine and Flatiron Health teamed up to create a large-scale clinico-genomic database. The database, which links clinical and genomic data gathered from patients with non-small cell lung cancer treated in community practices, was validated in a recent study.
Oncologists have been grappling with if and when real-world evidence (RWE)—generated by analyzing real-world data (RWD) gathered outside of traditional clinical trials—should be used in cancer research. A recent study offers one example of how this could work: Researchers demonstrated that clinical and genomic data gathered from patients with non–small cell lung cancer (NSCLC) treated in community practices can be successfully combined into a large-scale, validated database (JAMA 2019;321:1391–9).
The project linked clinical data from electronic health records of 28,998 patients treated at 275 cancer clinics in Flatiron Health's network to genomic data from testing performed by Foundation Medicine, including 4,064 patients with NSCLC who were the focus of the study. “We've seen the work we can do with genomic data,” explains Gaurav Singal, MD, Foundation Medicine's chief data officer. “Flatiron had been doing an amazing amount of work validating and building really high-quality clinical datasets, but neither one of us had the other. By bringing our forces together, we felt that we could really advance the field.”
The next task, Singal says, was to validate the combined dataset, so he and his colleagues cross-checked the genomic and clinical characteristics of patients in their database with those described in The Cancer Genome Atlas and large, multicenter trials. Then they also confirmed that their prognostic data were consistent with those collected in other trials.
Next, they compared treatment results to those in other studies. In keeping with previous findings, targeted therapies extended overall survival (OS) in patients with mutations in genes such as EGFR and ALK. Additionally, they saw a positive correlation between tumor mutation burden and OS in patients treated with PD-1/PD-L1 inhibitors. “That's the foundational layer that lets us say, ‘Now we can use this database to develop even better biomarkers for immunotherapy,’” Singal explains.
Singal and his colleagues now plan to use the database to research drug resistance and biomarkers, as well as to explore whether RWE can be factored into treatment decisions—and how it might improve the efficiency of drug development and approvals.
In the last category, the FDA is making progress. In March, the agency released a framework outlining plans for integrating RWE into regulatory decisions (available at www.fda.gov/media/120060/download), and in April it used RWD from post-marketing reports and electronic health records to expand the indication for palbociclib(Ibrance; Pfizer).
“This study nicely demonstrates that linking complex real-world datasets in a [Health Insurance Portability and Accountability Act]–compliant manner is feasible and can provide unique insights not otherwise easily obtainable from traditional clinical trials,” says Sean Khozin, MD, MPH, associate director of the FDA Oncology Center of Excellence, who was not involved in the research. He adds that such insights “can allow for better personalization of treatment decisions … while supporting new hypotheses that can be tested in a more controlled clinical environment.”
In an editorial, Ethan Basch, MD, of the University of North Carolina Lineberger Comprehensive Cancer Center in Chapel Hill, and Deborah Schrag, MD, MPH, of Dana-Farber Cancer Institute in Boston, MA, note that the study highlights how RWD can broaden understanding of patient responses to treatments, measure the effectiveness of novel therapies, and inform clinical trial design (JAMA 2019;321:1359–60). However, they also point out that the study reveals the “challenges faced by this field as it develops,” including missing or incomplete information, possible biases in clinical decisions, logistical constraints of large-scale data collection, and questions of privacy and consent.
Singal views the database, and, by extension, RWD, not as a replacement for randomized, controlled clinical trials, but as an additional option: “There are always going to be differences between clinical trial data and real-world data, but I think this dataset is another now-validated tool that we can deploy for the right problems.” –Catherine Caruso
For more news on cancer research, visit Cancer Discovery online at http://cancerdiscovery.aacrjournals.org/CDNews.
