Saskia Uppenkamp
Zach Weinberg, left, and Nat Turner, the founders of Flatiron Health.
- In 2016, the FDA and Flatiron Health, an oncology technology company, started a collaboration to figure out what could be learned using "real world" clinical data.
- That "real world" data comes from cancer patients' health records that Flatiron analyzes.
- In data presented at an upcoming conference, the FDA and Flatiron will show how this data could one day inform how we treat cancer.
The FDA is turning to unconventional sources of data to get a better picture of how drugs it's approved are being used in real life.
In 2016, the FDA teamed up with Flatiron Health, a $1.2 billion New York-based healthcare technology startup that collects clinical data from cancer patients - such as what medications a patient has taken and how he or she has responded to them.
Amy Abernethy, Flatiron's chief medical officer, told Business Insider that the goal of the collaboration is to figure out what you can learn from real world data. Right now, that data is locked up in electronic health records, which tend to be organized to help bill insurers, rather than detail, say, the rate at which the tumor is shrinking, in a way that could be pulled out for a study.
Now, we're getting the first look at that answer in data presented as part of the American Society of Clinical Oncology annual conference, which starts Friday.
It's the first time Flatiron, which was founded in 2012, is presenting at ASCO, Abernethy said. In total, the company's database was used in 14 studies presented at the conference.
The FDA and Flatiron Health have been collaborating to explore how the FDA might be able to use real-world data (that clinical information Flatiron collects) to inform how drugs the agency approves are used outside of the clinical trials that are the basis for their approval.
For one of their first studies, the two looked into how cancer immunotherapy drugs, specifically PD-1 checkpoint inhibitors, are being used in the real world in patients with non-small-cell lung cancer.
Dr. Sean Khozin, one of the authors on the study and a senior medical officer at the FDA, told Business Insider that Flatiron is just one of many companies the agency is working with to check out how alternative sources of data can be used to better inform the agency.
Pulling from Flatiron's database, which compiles clinical data taken out of de-identified electronic health records, the researchers found that patients who used one of two checkpoint inhibitors - pembrolizumab, (otherwise known as Keytruda), and nivolumab, (Opdivo) - had low rates of being tested for PD-L1 expression. For those taking pembrolizumab, there needs to be a certain threshold of PD-L1 expressed on the tumor for treatment, according to the FDA's label. And of those who were tested, it rarely happened using the test the FDA had approved as a companion diagnostic.
The findings suggest that the therapies aren't used quite like how they're approved by the FDA, which could make a difference in how the treatments pan out.
Besides randomized controlled clinical trials, there are a number of ways researchers can study how a particular cancer treatment is playing out in the real world: there are retrospective studies, observational studies, and even prospective research done using patient registries.
Abernethy said Flatiron's process is similar to a registry trial, but because data from over the years is already available in the database, the trials don't take as long as they might otherwise.
The results weren't necessarily surprising, Abernethy said, but it does set the stage for further research.
"It starts to set a line in the sand to understand what's happening in the community and why," Abernethy said.
Going forward, the data Flatiron and the FDA analyze could help fine-tune cancer treatment guidelines. For example, if the cancer patients in the Flatiron database that are using the checkpoint inhibitors tend to get better results when they're older, that could suggest that doctors should more actively consider putting them on that course of treatment than they might have based on the clinical trial data alone.
And, ultimately, the hope is that more electronic health record data becomes useable for researchers looking to answer questions about how drugs are being used in the real world. Khozin said that since the data in electronic health records isn't as easy to get, especially at a large enough scale.
But, Khozin said, he thinks someday soon that will begin to change, at which point insights like the one the FDA found with Flatiron will be easier to access.