Skip to content

Pink Sheet – US FDA Wants More Examples Of Real-World Data Use

Pink Sheet – US FDA Wants More Examples Of Real-World Data Use

Executive Summary

CDER Director Woodcock says for regulatory purposes, using real-world data likely will remain a case-by-case decision.


The US FDA does not appear ready to add indications to drug labels solely based on real-world evidence, even if the correlations to clinical trial data are strong, but still wants the experiments to be done.


Ultimately, the more attempts at making the leap that are seen, the easier it will be for FDA to write guidance and determine best practices, so down the road real-world evidence eventually can be used to support labeling or other aspects of drug development.


One step towards building supporting data came from a pilot project initiated by the Friends of Cancer Research. FOCR convened six data partners to determine whether real-world evidence can be used to develop “an early perspective on real-world outcomes, as defined by real-world endpoints from [electronic health records] and claims data, and whether these data correlate to overall survival in the context of randomized control trials,” according to an white paper outlining the methods and results.


The pilot reviewed data from real-world endpoints from several sources looking at the outcomes that could be evaluated for advanced non-small-cell lung cancer (aNSCLC) patients treated with immune checkpoint inhibitors.


The results indicated several real-world endpoint data correlated with median overall survival measured in immune checkpoint inhibitor clinical trials, suggesting that there may be some promise in eventually using the data for regulatory purposes.


Center for Drug Evaluation and Research Director Janet Woodcock congratulated the group for the work during a July 10 conference on the project, but was not ready to accept its findings and the methods used for regulatory use. But she said “what we need is more examples.”


“I think that there are legitimate questions to ask,” Woodcock said of the FOCR pilot project’s results. “Everything is going to be very case-specific, but it is really important that people do these things and that we start learning from them.”


“Whether or not this winds up in the label … is probably not as important as that we learn what the limitations are, we understand how to approach this, how to correct for those to move forward,” she added.


Woodcock reiterated to the Pink Sheet after the conference that decisions on using real-world evidence likely will be made case-by-case. She said the evidence still must be convincing and experience sufficient that a treatment effect is present.


The question FDA must answer is “can you make a causal inference” from the data, Woodcock said.


Jennifer Malin, senior medical director for oncology and genetics at UnitedHealthCare, also said that the value of real-world evidence still will be measured by the intended use of the treatment.


“The issue comes down to, What’s the use case?” Malin said during the conference. “Would that level of data be sufficient if it were a question that there’s no reason you couldn’t do a randomized controlled trial? No.”


“You want the best level of evidence that can be obtained to answer important questions,” she added. “The answer is, It depends.”


FDA will try to provide more clarity with its upcoming guidance on real-world evidence, including a framework for its use to support some drug approvals and postmarketing requirements. (Also see “Real-World Evidence: Advice, Principles And Examples Emerge From FDA” – Pink Sheet, 7 Oct, 2017.)

Several EHR Endpoints Correlate With OS

The pilot project considered several EHR and claims-generated data points, including overall survival, time to next treatment, time to treatment discontinuation, progression event, real-world progression-free survival, real-world time to progression and index date. (See box for definitions.)

Data Point Definitions in FOCR Pilot Project

  • Overall Survival: length of time from the date study treatment was initiated to the date of death or proxied by time to disenrollment.
  • Time to Next Treatment: Length of time from the date study treatment initiated to next systemic treatment.
  • Time to Treatment Discontinuation: Length of time from the date of PD-(L)1 regimen initiation to the date of treatment discontinuation.
  • Progression Event: Distinct episode in which the treating clinician concludes there has been a growth or worsening in the aNSCLC.
  • Real-World Progression-Free Survival: Length of time from the date the patient initiates the PD-(L)1 regimen to the date a progression event is evident and documented in the EHR or the patient dies.
  • Real-World Time to Progression: Length of time from the date PD-(L)1 treatment is initiated to the date a progression event is documented, excluding death.
  • Index Date: The earliest PD-(L)1 initiation in the advanced setting anchored to start of the immune-checkpoint inhibitor-containing regimen.

FOCR and its data partners concluded that there was a high level of shared characteristics among the data sets even though sample sizes, data capture and sources were different, “demonstrating the feasibility of identifying aNSCLC patients treated with immune checkpoint inhibitors from diverse [real-world evidence] sources,” according to the white paper.


They also said in the paper that “several extractable endpoints from EHR and claims data correlate with OS,” although further validation is necessary to determine if the endpoints are reliable surrogates for overall survival outside a clinical trial, and that survival among patients as reviewed in EHRs and claims data “fall within the range of median OS values observed in several immune checkpoint inhibitor trials.”


Eric Rubin, Merck Research Laboratories senior VP and therapeutic area head for oncology early development, said as it is shown that useful data can be extracted from these types of sources, interest in potentially using it in drug development increases.


“We focused on designing traditional trials with development,” Rubin said. “I don’t think we’ve gotten as far along as actually proposing as part of the development process a real-world study, but one could envision such a thing given the data we’ve seen today.”


Indeed, much of the work was done in only two months and was admittedly not as extensive as it could have been. But John Penrod, Bristol-Myers Squibb Co. group director of worldwide health economics and outcomes research in oncology, said the data “may accelerate the exploration of good hypotheses and the development of more helpful treatments for patients.”


More immediately, the results also could allow a wider patient population into formal clinical trials through enhanced eligibility criteria and calm fears that a wider variety of patients would dampen the treatment effect. (Also see “Real-World Evidence Could Help Relax Clinical Trial Enrollment Criteria” – Pink Sheet, 4 Jul, 2018.)

‘Heading in the Right Direction’ With RWE

Of the companies represented at the conference, it seemed FDA was the farthest along in terms of incorporating real-world evidence use.


Rajeshwari Sridhara, director of the CDER Office of Biostatistics Division of Biometrics V, said the agency on a scale of one to 10 may be at a nine or 10 for real world data use in rare diseases. But she added that it may only at a one for real world evidence use in common diseases, especially if a traditional clinical trial could be done to show efficacy.


“I think we are heading in the right direction,” she said. “It’s hard to put a number on some of these because each disease is different and how do you weigh that?”


Drug developers at the conference placed their use of real world data more in the middle of the scale. Rubin said Merck Research Labs likely is about a three or four for using it heading toward an application submission, but for post-approval reimbursement it is likely at a nine. Penrod said BMS is similar – in the development space informing choices, the company is about a five, but after that stage is at an eight or nine and rising.


More work already is underway using real-world evidence to replicate clinical trial data. A group including Brigham and Women’s Hospital, Harvard Medical School and the health care data company Aetion wants to replicate 30 randomized clinical trials using claims databases. (Also see “Real-World Data Could Get Boost From Trial Replication Project” – Pink Sheet, 26 Apr, 2018.)


It is one of several FDA-sponsored real-world evidence projects. (Also see “Real World Evidence Benefits, Limits Explored In US FDA Demonstrations” – Pink Sheet, 29 Oct, 2017.)…