FDA Appears to Mostly Follow Its Drug Advisory Committees’ Advice

The FDA tends to agree with the recommendations of its drug advisory committees most of the time — especially if the committee votes in favor of approving a drug — but it is also convening advisory committees less often than before, a qualitative study showed.

During the years 2010 to 2021, the agency approved 142 of the 147 drugs for which its advisory committees recommended initial approvals and 33 out of the 36 drugs for which committees recommended approval for supplemental indications, reported Aaron S. Kesselheim, MD, JD, MPH, of Brigham and Women’s Hospital and Harvard Medical School in Boston, and colleagues.

The FDA didn’t approve 40 of the 60 drugs for which committees voted against approval and 18 of the 21 drugs for which committees recommended against approving supplemental indications, they noted in JAMA Health Forum.

Advisory committees were also convened less frequently as time went on, from a high of 50 in 2012 to a low of 18 in 2020 and in 2021.

“In cases of discordance, the FDA was more likely to favor approval than were its advisors, and the FDA declined to reconvene advisors before most approvals of drugs that had received a negative vote,” Kesselheim and team wrote. “Reforms are necessary to clarify the FDA’s commitment to independent expert advice as an integral part of its regulation of prescription drugs.”

Because of increasing concerns about the way the FDA uses its advisory committees, the researchers reviewed all advisory committee meetings from 2010 through 2021 related to prescription drug regulation with the goal of assessing how often advisors were convened, for what purposes, and how often the FDA acted in accordance with advisory committee recommendations.

During the study period, the agency convened 409 human drug advisory committee meetings. The Oncologic Drugs Advisory Committee met the most often (78 meetings, or 19% of the total), followed by the Endocrinologic and Metabolic Drugs Advisory Committee (11%) and the Anesthetic and Analgesic Drug Products Advisory Committee (10%).

More than half of all meetings (54%) included votes on initial drug approvals, 14% included votes on supplemental indications, 7% on safety actions, 2% on the withdrawal of a drug granted accelerated approval, 3% on a different regulatory action, and 20% did not vote on any regulatory action.

As to the FDA’s actions, a positive recommendation for an initial approval was associated with a significantly increased likelihood of approval (HR 9.42, 95% CI 5.99-14.80, P<0.001). Of the drugs that were approved, the median time between a positive vote and approval was 74 days, while the median time between a negative vote and approval was 700 days.

The issue here is what the FDA really wants to learn from its advisory committees, noted Genevieve Kanter, PhD, of the University of Southern California in Los Angeles, in an accompanying editorial. “The asymmetry in the FDA’s response to approval [versus] non-approval recommendations says something very important about the agency’s view of advisory committees,” she wrote.

“On its face, the chief question that the FDA poses to advisory committees appears to be, ‘Should we approve this drug?’ What these findings tell us is that the FDA is not really asking that,” Kanter continued. “Instead, the agency is asking, ‘What is preventing this drug from being approved right now?’ or perhaps, ‘What are different ways we can get this drug approved?'”

“This pro-approval bias suggests that the FDA is using advisory committees to identify barriers to approval and guide the agency toward easier paths for approval,” Kanter said, citing as an example what happened with the Alzheimer’s disease drug aducanumab (Aduhelm), “where the advisory committee was roundly convinced that the drug had no effect on key clinical end points. Seeing that pivotal study findings on the clinical end points were unpersuasive, the agency found an alternative route to approval using a surrogate end point and did not reconvene its implacable advisory committee.”

“Pro-approval bias need not be problematic as long as it reflects the public’s priorities of speedy access to new treatments and high(er) risk tolerance,” she concluded. “If, on the other hand, it reflects the influence of pharma, this bias is deeply concerning,” especially in light of recent developments suggesting political arm twisting, as well as the diminished frequency of advisory committee meetings, among other issues.

In the end, the FDA “can only enhance its credibility by taking its advisory committees seriously,” wrote Kanter. “Let us hope that the FDA does not fall prey to actions that confer shortsighted political gains at the expense of a more important lasting legacy.”

For this qualitative study, Kesselheim and colleagues used a manual review of meeting summaries prepared by FDA staff for the 18 human drug advisory committees active at any time from January 2010 through December 2021.

Outcomes of votes on regulatory questions were determined by using meeting minutes. The committee recommendation was defined as the majority outcome of the vote. Corresponding FDA actions were determined using a variety of sources, including FDA announcements and press releases, drug labels and approvals at Drugs@FDA, and internet searches for industry publications and company press releases. Alignment of FDA action with advisory votes for new drugs and indications was judged as of 1 year after the vote was held and as of Nov. 30, 2022.

Meetings were excluded if they were not held due to cancellation or postponement. If a committee was asked to discuss two drugs in 1 day, such as in separate morning and afternoon sessions, the sessions were counted as two meetings occurring on the same day, so that each recommendation could be counted separately, the researchers explained.

Limitations to the study included the fact that votes on questions of initial approval “often did not include language of specific indications, so we were unable to systematically compare approved indications with advisory recommendations for initial approvals,” Kesselheim and team noted. “Second, although most voting panelists are experts, some represent patients or other consumer or industry interests. We did not account for how non-expert panelists with voting authority may have impacted recommendations.”