The FDA should require negative studies of a new drug to be included in the product’s labeling—not just those with beneficial findings—to prevent physicians and consumers from being misled about the drug’s safety and efficacy.
This conclusion comes from researchers who examined the drug’s long history of rejections before the FDA finally approved gepirone extended release (Exxua) for major depressive disorder in 2023, despite limited evidence of effectiveness. The findings appear in JAMA Psychiatry.
The agency based its decision on two trials that showed a small but statistically significant benefit, while apparently ignoring 11 other studies that found the drug no better than placebo, according to Erick Turner, MD, of Oregon Health & Science University, and colleagues.
The gepirone label “represents a cherry‑picked subset of studies,” Turner told MedPage Today. “For those, [the FDA] relied on statistical significance as opposed to clinical significance … with a very tiny margin of advantage over placebo: less than a half a point difference on the Hamilton Depression Rating Scale.”
“Unfortunately, the gepirone ER product label mentions only the two statistically significant trials and excludes the 11 unsuccessful efficacy trials, and the lack of clinically meaningful effects,” they wrote. Adverse events were aggregated only from those two trials, not from the others.
“Product labeling should transparently report all adequate and well‑controlled trials relating to an FDA‑approved indication, not just those with positive outcomes, so that clinicians can make better‑informed prescribing decisions,” the researchers noted.
Gepirone is also expensive, costing $1,500 per month without insurance, and a Medicare Part D plan check showed Exxua’s annual cost could reach the 2026 cap of $2,100.
Turner, a former medical officer in the FDA’s neuropharmacological drugs division, was part of the panel that rejected gepirone for the first of at least four times in 1999, before leaving the agency in 2001. He was not involved in the 2023 decision.
When approval finally came in 2023, Turner said he was astonished. “I said, ‘Wait a minute. Isn’t this the same drug?’ … and saw how convoluted its path to approval was, how long it took.”
He likened the difference between statistical and clinical significance to a patient whose blood pressure is lowered from 180/120 mm Hg to 175/115 mm Hg—a statistically significant change that remains clinically inadequate.
The authors noted that many physicians misunderstand the FDA approval process. Seventy percent mistakenly believe the FDA must find both clinical and statistical significance before granting approval, and 73 percent think new drugs must be as effective as older ones, while 40 percent think they must be more effective—none of which are formal requirements.
“This lack of insight into the data underlying FDA‑approved drugs may help explain why clinicians tend to overestimate interventions’ benefits and underestimate their harms,” they wrote.
Robert Steinbrook, MD, health research group director for Public Citizen, who was not involved in the study, said the gepirone case “speaks to the need for the FDA to consider this example as an impetus for revising its guidance” on legal and regulatory standards for drug approval.
When the final decision was made in 2023, it came down to the FDA’s belief that gepirone’s application “had met the legal regulatory standard for approval, which was different than consideration of the totality of the evidence, and whether this is a good clinical drug,” Steinbrook told MedPage Today. “That’s the problem. The FDA seems to feel that its hands were tied.”
The researchers’ work “does make a strong case for more inclusive labeling,” he added.
Adriane Fugh‑Berman, MD, of Georgetown University Medical Center and the watchdog group PharmedOut, wondered how many other drugs have been approved based on “unimpressive clinical evidence but that people think are safe and effective.”
“Drug companies are invested in exaggerating benefits and minimizing perceptions of harm, and right now, the labels are helping them out. The FDA should be encouraging better information,” she told MedPage Today.
Asked whether mental‑health or pain drugs have been approved despite negative studies, Turner, an expert on publication bias, said it was hard to know.
“We do know that psychiatric drugs, particularly antidepressants and drugs for anxiety, roughly half are positive so that means you have a lot of negative studies. Thus there’s the potential for publication bias by the sponsor, who is not publishing every study,” or submitting them to the FDA for review, he told MedPage Today.
Compounding the issue is that study endpoints for these conditions are often subjective, like depression or pain. “You can’t really measure how much someone has,” he said.