With notable therapies from Biogen, Sarepta and MacroGenics failing to show efficacy in pivotal or confirmatory trials, experts question the use of biomarker evidence for approval while one former regulator insists that a “failed trial is not a failed drug.”
What do Biogen’s amyotrophic lateral sclerosis drug Qalsody, MacroGenics’ breast cancer drug Margenza and Sarepta’s exon skippers Vyondys 53 and Amondys 45 for Duchenne muscular dystrophy have in common? They all failed pivotal or confirmatory trials. But most were ultimately fully approved by the FDA or remain on the market.
To make their case for approval or continued marketing, the companies have pointed to a range of supporting evidence, including positive biomarker data, the success of other primary endpoints or, in the case of Sarepta’s exon skippers, “substantial real-world evidence” and a “positive safety profile.”
They’re far from alone. A 2023 study published in JAMA Internal Medicine found that 21 of the 210 new drugs approved by the FDA between 2018 and 2021 failed to meet at least one primary efficacy endpoint in a pivotal trial.
Joe Ross, professor of medicine and public health at Yale School of Medicine and co-author of the study, told BioSpace he was “surprised to find that about 10% of all approvals are made despite the primary efficacy endpoint being null.” But, he added, “perhaps it isn’t surprising because the FDA has such flexibility in reviewing a marketing application.”
The most notable case—highlighted by the authors—is Biogen’s Aduhelm (aducanumab), greenlit under the FDA’s accelerated approval pathway in June 2021 as the first Alzheimer’s drug to treat the underlying cause of the disease. It was also among the most controversial approvals of the past decade, leading to the resignation of three prominent FDA reviewers.
In March 2019, Biogen and partner Eisai halted the global Phase III ENGAGE and EMERGE trials of Aduhelm after it became clear the anti-amyloid drug would not succeed. Then, a few months later, the companies reported that EMERGE had in fact hit the primary endpoint at the highest dose, demonstrating a significant decrease in clinical decline. But the FDA based its approval on a biomarker: Aduhelm’s ability to reduce amyloid beta plaques in the brain.
“It’s a challenging conundrum,” Ross said. “Particularly by the time you reach the pivotal trial phase for a new drug . . . lots of people have invested lots of time and money into bringing that product to market. I think everybody wants it to be a success.”
This desire isn’t limited to the company and the patients who could benefit from the drug, he said. “The FDA wants it to be successful. They put out press releases for every new drug they approve, not for every new drug they don’t approve.”
Harpreet Singh, chief medical officer at clinical research organization Precision for Medicine, was formerly a division director of oncology under now-retiring Center for Drug Evaluation and Research head Richard Pazdur. “One thing that Rick Pazdur taught us that has really stuck with me over the years is that a failed trial is not a failed drug,” she told BioSpace.
Burden of Proof
There are many reasons that trials fail, Singh said. In the 2023 JAMA study, “success of other pivotal studies” was the number one reason listed under the FDA’s rationale for approval despite a failed pivotal trial at 62%, followed by “benefit suggested by secondary or exploratory endpoints” at 48% and “post-hoc effectiveness analysis” at 33%.
For Singh, one of the biggest reasons for a trial failure is endpoint selection. “Did we pick the appropriate endpoint for the therapeutic area, for the therapeutic class?”
A “very classic example” of this is found in immunotherapy, she said. Between 2018 and 2021, Singh touched over 50 approvals, “and I think in doing that, you see that as a class, immunotherapy was not suited to shorter-term endpoints like response rate and progression-free survival,” she noted. “Rather, you are really going to see the best benefit of immunotherapy on the overall survival endpoint.”
Patient selection is usually the second major reason why trials fail, Singh said. As the drug development process expands out into a large, randomized, multi-regional trial, “you see a large amount of heterogeneity, so the efficacy that we see sometimes slides a little bit” from where it was in the single-arm setting.
Finally, she said, operational issues can be a factor. “Now, being CMO of a CRO who runs trials, I can tell you firsthand that operational issues certainly are critical to a trial’s success.”
In these less clear-cut cases, where one or more pivotal endpoints are missed, Ross explained, the FDA may turn to clinical outcomes that suggest a benefit. “Maybe it didn’t hit the p-value, but it’s close,” he said. “At least the evidence is aligning towards clinical benefit.”
Where the calculus becomes trickier for Ross is biomarkers. “When it’s just a change in a biomarker and even that is not statistically significant, that’s really worrying to me,” he said, adding that the biomarkers used by the FDA over the years have been shown to be not very well correlated with clinical outcomes. “Maybe at the time when the first approval was made, there was some suggested evidence, and then FDA felt obligated to sort of continue to use that biomarker because of the precedent it established.”
Ross pointed to response rate and progression-free survival (PFS) in cancer drug development. “We know pretty conclusively that response rate and progression-free survival do not correlate well with patients’ symptom or functional status, or overall survival rate, but we continue to use those markers over and over,” he said. Notably, the FDA in August issued draft guidance emphasizing that drug developers should prioritize using overall survival (OS) as the primary endpoint in their studies—a campaign Center for Biologics Evaluation and Research Director Vinay Prasad has vigorously pursued.
When To Let Go?
Another study finding that surprised Ross was that despite missing these primary endpoints in a pivotal trial, a confirmatory study was only mandated or requested by the FDA about a third of the time, or for seven of 21 approvals.
While Ross was open to considering mixed results for an approval, he was resolute that the confirmatory trial must support the drug’s benefit for it to remain on the market. Regarding Sarepta’s Vyondys 53 and Amondys 45, he said, “I think that they should be removed from the market.”
“The FDA showed a lot of flexibility, one might even say leniency, by approving those drugs at the time that they did with the expectation the confirmatory studies would be done,” he continued. “And I think the company should remove them from the market. The company at some point should, for lack of a better term, face the music. The drugs were not effective.”
Vyondys 53 was approved based on data showing it increased levels of dystrophin, a crucial muscle protein lacking in patients with DMD, after 48 weeks. Amondys 45 satisfied the biological endpoint of micro-dystrophin protein expression at 12 weeks post-treatment, according to Sarepta’s press release at the time. Neither therapy met the confirmatory ESSENCE trial’s primary endpoint of an individual’s ability to ascend a flight of stairs at 96 weeks.
Still, as of Nov. 3, Sarepta said it was planning to meet with the FDA to discuss flipping the accelerated approval to a traditional approval.
Overall, Singh said nuance is key in these situations. “I don’t think that failed endpoints suggest failed trials,” she reiterated. “I think you really have to unpack what that looks like.” She referred to the June 2021 approval of Amgen’s Lumakras (sotorasib)—the first KRAS-targeting drug ever authorized by the FDA—to help make her case.
Lumakras’ accelerated approval, over which Singh presided, was granted based on the drug’s success in the single-arm Phase II CodeBreak 100 study. However, in October 2023, an advisory committee voted against granting full approval of the drug, saying PFS data from the Phase III confirmatory CodeBreaK 200 study could not be reliably interpreted.
“Essentially, there was so much enthusiasm about the novel drug that there ended up being asymmetric drop out of the control arm . . . and ultimately rendered the trial results uninterpretable,” Singh explained. Singh and Pazdur “certainly had the option to pursue a path to withdraw that drug from the market,” she said, but because of the high unmet need and the “robust response rate”—treatment with Lumakras led to a confirmed objective response rate of 36%—they elected not to.
“I think that’s a beautiful example of a failed trial really not representing a failed drug,” she said.
