What Did US FDA Do Wrong In Its Review Of Aducanumab AdCom Members Have A List

The US Food and Drug Administration faced a barrage of criticism from members of its advisory committee over its positive review of Biogen, Inc.’s aducanumab Alzheimer’s disease drug. They objected to the way the agency laid out data in a joint briefing document with Biogen, saying FDA weighted the findings in favor of the sponsor and downplayed the discordant results of two Phase III studies.

At its 6 November meeting, the Peripheral and Central Nervous System Drugs Advisory Committee voted that Biogen’s single positive Phase III study could not provide primary evidence of effectiveness for aducanumab for treatment of Alzheimer’s disease. The vote was nearly unanimous, with 10 members voting against use of Study 302 to support approvability and one member casting an “uncertain” vote. (See sidebar.)

The meeting was starkly different from typical advisory committee meetings, as it put FDA at the center of the debate over the data. The agency had taken the extraordinary action of collaborating with Biogen in analyzing the data after Biogen terminated the research program for aducanumab. (Also see "USA FDA Gives Biogen Big Hand In Effort To Get Its Alzheimer’s Treatment On The Market" - Pink Sheet, 4 Nov, 2020.)

FDA concluded that despite the negative results of an identically designed Phase III trial, Study 301, the successful Study 302 was sufficient to demonstrate effectiveness. Committee members criticized the conclusions in the briefing document.

“I’m highly critical of the fact that the FDA presentation today was so heavily weighted to just giving the same conclusions that the sponsor did.” – Scott Emerson, University of Washington

Caleb Alexander, Johns Hopkins University Bloomberg School of Public Health, questioned the basic framing of the briefing materials. “The bottom line is that I find the materials that the FDA has provided strikingly incongruent and I have a hard time understanding, after carefully reviewing what I thought was a very well done and well-articulated biostatistical review, which convincingly argued the evidence was ‘at best compellingly conflicted’ how the FDA could conclude that there is substantial evidence of effectiveness and in particular that Study 302 provides ‘a robust and persuasive study,’” he stated.

“It just seems to me like the audio and the video on the TV are out of sync. And there are literally a dozen different red threads that suggest concerns about the consistency of the evidence,” Alexander said.

“There is no question that all of this was just terrifically one-sided,” Scott Emerson, University of Washington, said. “I’m highly critical of the fact that the FDA presentation today was so heavily weighted to just giving the same conclusions that the sponsor did.”

“I was very disturbed by some of the FDA’s interpretation of 301 by starting out with the assumption that the treatment works and now trying to say, why did we get no results in 301,” he stated. “I was very, very, very disturbed by some of the analyses that were considered. . . I just didn’t understand why there seemed to be this kind of unilateral effort to discredit one study.”

Negative Statistical Review

Emerson, a biostatistics expert, was added to the panel as a temporary voting member to consider the aducanumab data. He was one of the most vocal critics during the committee discussion.

FDA’s statistical reviewer Tristan Massie disagreed with the agency’s conclusions. In a memo included as an appendix in the briefing document, he said Biogen’s biologics license application does not present just a single, strong study. Rather, he said, “we have a second large adequate well-controlled study that directly contradicts the first and is not even close to significance.” He was joined by two concurring reviewers. (Also see "Biogen’s Aducanumab: One Positive Phase III Trial Is Good Enough For Demonstrating Efficacy, US FDA Says" - Pink Sheet, 4 Nov, 2020.)

In contrast, FDA’s clinical reviewer Kevin Krudys said Study 302 met its primary and secondary endpoints and appeared exceptionally persuasive. He concluded that Biogen had provided substantial evidence of effectiveness to support approval.

FDA asked the committee to discuss the impact of the results of Study 301 on the considerations of the results of Study 302. Madhav Thambisetty, National Institute on Aging, said the discordant perceptions on this question were aptly summarized in the discordance between the two reviewers. “To paraphrase Dr. Tristan Massie, if you have two and you take the best and pretend like it’s the only one, your estimate is likely biased.”

FDA did not address the statistical review during its presentations. While Massie briefly responded to a couple of questions, he was not asked to expound on his analysis of the data.

Does Dose-Ranging Study Support Effectiveness?

FDA also asked the committee if an early-phase dose-ranging study, Study 103, which suggested a dose-response, provided supportive evidence of the effectiveness of aducanumab. Seven members voted that it did not and four voted they were uncertain.

Alexander said the agency seemed to selectively “use lines of evidence from 103 and 301 to support the findings of 302 at the expense of calling out any number of lines of evidence that call into question the findings of 302.”

Aaron Kesselheim, professor at Harvard Medical School and faculty member at Brigham and Women’s Hospital, said Study 103 was not designed to provide supportive evidence but was designed for evaluation of other things. “That’s why in 103 you’re getting efficacy results that seem very discordant from the efficacy results that you see in 302,” he said.

“In no sense would I regard 103 [as] going to be in place of another confirmatory study," Emerson said. Study 103 "doesn’t make me relax criteria regarding 302."

FDA did not ask the committee whether Biogen should conduct another trial, a question overhanging the review. But at the end of the meeting, Emerson posed the question: “What additional study would I want to see?” He said he would like a randomized withdrawal of the planned dose. He said he was not certain how long participants should be treated prior to withdrawal, but noted a comment by one of the speakers at the open public hearing session that they could quickly tell if the drug was having an effect.

A member of the committee recused from the meeting, David Knopman, professor of neurology at Mayo Clinic and associate director of Mayo's Alzheimer's Disease Research Center, said in a recent article in the journal Alzheimer's & Dementia that Biogen should conduct another Phase III trial with high-dose aducanumab. Knopman was a site investigator in Biogen's Phase III aducanumab trials.

FDA: No Intent To Ignore Study 301

FDA clinical reviewer Krudys addressed questions raised in discussion of Study 103 as to what constitutes supportive evidence of aducanumab’s effectiveness.

He said the main issue was to try to provide clarity in the questions posed to the committee so they could comment on the data as presented in the briefing document.

“It seems to be causing a lot of consternation and that’s unfortunate,” Krudys said. “It wasn’t intended to provide an artificial exclusion of any of the data” but to take the arguments made in the briefing book layer by layer.

The intent was not to ignore Study 301, he added, but to consider to what degree, “if one were assured about what happened in 301,” there is strength in 302.

Impact Of Unblinding

In Study 302, high-dose aducanumab (10 mg/kg) resulted in a statistically significant 22% reduction (p=0.0120) in clinical decline compared to placebo at week 78 on the Clinical Dementia Rating-Sum of Boxes score, which was the primary outcome measure.

In a presentation to the committee, Samantha Budd Haeberlein, Biogen senior VP and head of the neuordegeneration development unit, said aducanumab “potentially prolongs patients’ independence by several months, even a few years, as demonstrated in our long-term study." She said it is a "stepping stone for our next advances.”

Post hoc analyses conducted by FDA and Biogen to understand the discordant results for the high-dose groups in the two Phase III studies suggested the high-dose group in Study 301 had twice as many rapid-progressors – participants having unusually rapid decline than each of the other groups in both studies.

Both studies had sporadic unblinding for dose management of amyloid-related imaging abnormalities (ARIA), a spectrum of imaging findings detected on brain magnetic resonance imaging. These findings include brain edema and brain microhemorrhage.

Thambisetty voiced “significant concerns about characterizing Study 302 as being either robustly positive as described in the meeting material, or as charactering it as a homerun,” as FDA’s Billy Dunn, director of the Office of Neuroscience, referred to it during his presentation. Thambisetty specifically cited the potential impact of unblinding on assessment of aducanumab’s effectiveness.

“I think this is such an important application to get right because of the overwhelming imperative for new treatments and the precedent that’s set and what the application’s review conveys to the scientific and clinical communities about the evidentiary threshold for approval in this disease space.” – Caleb Alexander, Johns Hopkins University

“I think that’s a huge concern with this study,” he said. Noting that 35% of all patients exposed to the drug developed ARIA, he said it is conceivable that patients who are given a diagnosis of ARIA and then subjected to very intense serial MRI surveillance, are going to be unblinded to the treatment.

“What makes this especially concerning is that the primary endpoint, which is CDR-Sum of Boxes scores, is entirely dependent upon subjective information that is provided to the rater by the patient and the caregiver.”

Joel Perlmutter, Washington University School of Medicine, said the differential blinding in people getting the high dose and the retrospective definition of rapid progressors is a concern to him. “These all raise questions and even if we don’t see statistical difference on unblinding, when you add these things up they can together cumulatively be an issue.”

Important ‘To Get Right’

Committee members expressed the need for further analysis of the data to answer remaining questions.

“I feel to some extent the clock has run out and we haven’t been able to ask questions. The FDA just gave us conclusions and not results,” Alexander said. “I would like to get into some details here about the totality of evidence and about the conclusions that the FDA seems to be reaching and about the incongruous materials that have been provided and the dozens of questions that we really haven’t had a chance to ask the FDA about.”

“And in particular, I’d like to query the FDA about any number of concerns that their own statistical reviewer has identified and that I have not heard either an adequate response from the sponsor, but more importantly from the FDA, regarding their own interpretation of those reviews,” Anderson said.

At another point, Alexander noted how crucial FDA’s review of aducanumab is for future Alzheimer’s disease drugs.

“I think this is such an important application to get right because of the overwhelming imperative for new treatments and the precedent that’s set and what the application’s review conveys to the scientific and clinical communities about the evidentiary threshold for approval in this disease space,” Alexander stated.

Industry representative Michael Gold, VP Development Neurosciences at AbbVie, also noted the significance of FDA’s action on aducanumab, which if approved would be the first disease-modifying treatment to slow the clinical decline associated with Alzheimer’s disease.

“I think it is an acutely important decision that can have repercussions across clinical research enterprises [in] industry in terms of how we interpret these kind of studies and what is the standard of evidence,” Gold said.

As for the unsuccessful Phase III study, he said, “It is important to be sort of respectful of the fact that 301 was well designed, well conducted, well executed. There’s no evidence that it somehow was defective in any way, shape or form.”

Perlmutter noted that this Alzheimer’s treatment addresses a huge unmet need. “But I also think that if we approve something where the data is not strong, that we have a risk of delaying good treatment, effective treatment for more than a couple of years, for many years,” he said. “And I think there’s a huge danger in approving something that turns out not to be effective.”