FDA, Gottlieb Get Real About Real-world Evidence

Implementation of a new framework would secure RWE’s role in the drug-approval process.

When Roche last year launched its bid to acquire Flatiron Health and its treasure trove of oncology electronic health record data, Daniel O’Day, then CEO of Roche’s pharmaceutical business (and now in the top job at Gilead), said: “We believe that regulatory-grade, real-world evidence is a key ingredient to accelerate the development of, and access to, new cancer treatments.”

Ten months later, FDA Commissioner Scott Gottlieb issued the agency’s framework for using real-world evidence (RWE) in the review and approval of new indications for existing drugs and biologic agents. It was an important step toward fulfilling a central part of the 21st Century Cures Act, which President Obama signed in the waning days of his administration.

Here we are in early 2019, and how quickly and firmly the FDA and the pharmaceutical industry will embrace RWE and its raw material, real-world data (RWD), remains an open question. If that embrace is ever to get beyond gestures and lip service, the FDA will need to change its rules and culture. Moreover, the agency, drugmakers, and others will have to figure out how to marshal millions if not billions (or even trillions?) of data points from EHR systems, claims data, and patient registries into analytics that actually mean something.

Definitions from the FDA framework

Real-world data. Data relating to patient health status and/or the delivery of health care routinely collected from a variety of sources.

Real-world evidence. Documented clinical outcomes of potential benefits or risks of a medical product derived from analysis of RWD.

The 21st Century Cures Act was a product of legis­lative dealmaking that combined increases in research funding (especially for cancer) with changes in how drugs are to be approved. More specifically, with regard to drug approval, the law says the FDA must investigate the application of RWD to generate RWE to assign new uses for already-approved drugs and support post-approval requirements.

The draft framework that Gottlieb unveiled last December is a move toward putting some meat on those bones. It spells out eight steps the FDA will take, including guidance on how to assess the reliability and relevance of RWD from medical claims, EHRs, and other sources. Before the framework, the FDA had issued guidelines in July 2018 for how industry should use EHR data in clinical investigations.

“Historically, the FDA has been very reluctant to look at real-world evidence except for post-approval surveillance,” says Anirban Basu of the University of Washington. He sees the FDA’s new RWE framework as a step forward.

Anirban Basu, a health economist who studies comparative health outcomes and policy at the University of Washington and who’s written about RWE for the Health Affairs blog, sees the framework as a step forward. “Historically,” he says, “the FDA has been very reluctant to look at real-world evidence except for post-approval surveillance, and yet especially in the last 10 years tons of real-world evidence and real-world data have been generated.”

The influence that FDA rules have in shaping drug industry practices and priorities can’t be overestimated. Chris Hogg, a former pharma strategist who is now chief commercial officer at Propeller Health, a digital health company, says the framework “really is a sea change in how we develop and test new products and services.”

But at this point, the FDA may be playing catch-up with the industry it regulates. The industry has been ramping up its RWE efforts in anticipation of the agency rolling out the welcome mat. In a 2018 Deloitte survey, 90% of drug industry respondents reported that they’ve either set up or are investing in building RWE capabilities to evaluate their products, and 70% said they’re bringing those capabilities in-house.

A real-world example of an RWD trial

The framework for real-world evidence that the FDA published in December gives the ADAPTABLE trial as an example of a real-world data study. The trial, which started in April 2016, is randomizing 20,000 patients (on clinicaltrials.gov it says 15,000) to one of two doses of aspirin, 81 mg and 325 mg. The participants must have a history of myocardial infarction or known atherosclerotic disease but otherwise inclusion and exclusion criteria are being kept to a minimum. The investigators are using EHRs and claims data to capture the results for the primary (death, hospitalization for myocardial infarction, others) and secondary (revascularization procedures, hospitalization for serious bleeding) endpoints.

The 21st Century Cures Act and RWE are not without their critics. In the name of modernization, the law lowered drug approval standards, they say. Some have waggishly suggested that RWE be changed to RBE—“really bad evidence”—because there’s just too much inconsistency and too many gaps in the data that RWE depends upon, they say.

Retooling the regulatory process

Traditionally, the FDA has required drugmakers to put their products through a progression of trials, from preclinical testing on through phase 1, 2, and 3 trials. Lately, though, the agency has approved an increasing number of drugs, especially oncology drugs, in an expedited fashion, based only on results from early-phase trials. Those expedited approvals are mainly for drugs that target life-threatening and rare diseases and that have the potential to be significantly more effective than existing treatments. Zykadia (ceritinib), for example, was approved in 2014 based mainly on evidence from a phase 1 trial.

Taking RWE beyond the framework goal of adding new indications for existing drugs, what impact could it have on the regulatory drug review model? Basu has put forth such a model that essentially permits limited commercial distribution after phase 2 but with randomization of study participants. His idea is that a manufacturer would file a safety-based new drug application to obtain limited market access with what he calls “only-in-research labeling” over a limited period—two years, say. This new category of approval would allow physicians to use the drug in a real-world trial that would, in effect, supplant the existing phase 3. The drug would be tested in the wild, so to speak, with RWD collected within the tightly organized environment of a randomized controlled trial. It would be up to providers to use the drug and plans to decide how, or if, to pay for them.

“If you think about how that new drug will be used by physicians across patients, how a managed care plan is going to cover it, how much it should be paying for that, that information from the existing phase 3 trial structure is extremely uninformative,” says Basu.

Evidence about a drug from a randomized controlled trial may not translate into quotidian clinical experience for several reasons, Basu points out. Participants in a trial are carefully selected. The people who actually use a drug may be radically different in any number of realms: age, access to health care, comorbidities, overall health. Moreover, trials are short, partly because they are expensive, so they may not yield much information about the long-term effects of a new medication.

For a long time, drug approvals and assessments have been viewed as binary events, yea or nay. Basu says “you can’t make a one-time decision” if you really want to generate data and evidence about how a medication is doing in actual use. “You have to allow use of these drugs for some time and then come back and see how they’re working in the real world, and then make a final decision on approval.”

The huge volume of outcomes data now available brings RWE closer to reality, says Chris Hogg of Propeller Health. You can enroll people in studies quickly.

The huge volume of outcomes data puts RWE within reach, Propeller’s Hogg says. “When you have so much data, you can enroll people in studies so quickly that it frees you up to ask a lot of questions that you’d like to ask, but that couldn’t get done in an expensive phase 2 study.”

Getting serious

One sign that the FDA may be serious about RWE was the appointment late last year of Amy Abernethy, one of the founders of Flatiron Health, as principal deputy FDA commissioner. “She brings a lot of new perspectives with real-world data and evidence and decision making to the FDA that may not have existed in the past,” says Basu.

Recent moves by the FDA make Andrew Matzkin of Health Advances hopeful that the agency is serious about RWE. “I think they’re getting there.”

“I think they’re getting there,” says Andrew Matzkin, a health care strategy consultant at Health Advances. He’s followed the agency’s RWE efforts for digital health and medical devices, for which it issued guidance in 2017. “It’s the FDA, so it’s going to take quite a bit of time, but the guidance that they’ve issued for software and medical devices and the details emerging about the precertification pilot program are really, really encouraging.” The precertification program that the FDA launched in 2017 for digital health products is designed so companies and their technologies are approved rather than each product getting the OK.

Potential role for health plans

Matzkin can see health plans deriving big benefits from RWE in the regulatory realm, even though plans themselves won’t play a direct role in the regulatory process. The bridge from regulatory to coverage determinations consists of high-quality data and analytics at scale. “Once that exists,” he says “it becomes easier to implement a concept that payers and the life science industry have talked about for a decade or more: outcomes-based or value-based contracts for drugs.”

Insurers have been reluctant to develop or sign those contracts partly because of a lack of information about the efficacy and safety of drugs outside of RCTs.

Just who is going to collect the RWD, asks Jason Shafrin of Precision Health Economics. “Is it the health plan or is it going to be the drug companies?”

Health economist Jason Shafrin at Precision Health Economics says formularies need to change so that they are patient-centered and adjust to how people respond to a medication. Shafrin’s previous research on cancer therapies, published in the journal Value Health in 2017, found that factors such as a clinical trial’s patient population, geographic location, sample size, outcome measures and other factors can influence how well a treatment will work in the real world.

That’s where the data come in. “If something is very high value for 999 out of 1,000 people, then more restrictive formularies may make sense,” Shafrin says. “But if real-world data tells you the highest-value drug works best for 600 out of 1,000 people, that’s a little more problematic, since 400 people won’t be getting their preferred treatment.” Getting that kind of data can be expensive. “But that is helpful if you want the highest-value treatment for each patient rather than just the average,” he says. That, however, raises a question—and a potential conflict—about who’s going to collect that RWD. “Is it the health plan or is it going to be the drug companies?” asks Shafrin.

Another potential conflict: Physicians want to know if a drug is safe and if it works before they prescribe it. Despite their limitations, phase 2 and 3 trials are designed to produce the kind of evidence that will pass FDA muster and satisfy physicians. Payers, though, want information about how the drug is doing (or will do) once physicians start to prescribe. That difference is why payers are interested in moving toward pragmatic trials—that is, trials designed to evaluate the effectiveness of inter­ventions in real-life routine practice conditions, Shafrin says. “A drug may work under perfect conditions, but will it work in specific patient populations?”

Basu sees RWE as potentially giving insurers a bigger role in evaluating drugs. It should put them in a position to determine how, or if, to cover drugs that have cleared phase 2 but are in the evaluation phase of what is now phase 3. Randomization is the cornerstone of drug research; to determine if a drug truly works, it has to be randomly given to some patients and denied to others, then researchers compare outcomes of the two groups. Health plans should have a role in determining what patients get these drugs, Basu says. “They have to understand that these drugs are being approved under the only-in-research designation for only specific randomly selected patients within the target populations. They can cover them if they want.”

Now the FDA can begin the heavy lifting of writing rules to support RWE.