How a failed ALS drug provides a blueprint for the FDA to create a ‘Conditional Withdrawal’ pathway

Earlier this month, Amylyx’s ALS drug Relyvrio was shown to be ineffective in a clinical trial, prompting the company to uphold a commitment it had made to the FDA during its advisory committee meeting to withdraw its drug’s approval status. But while the company’s actions were strictly voluntary, an analysis by AgencyIQ shows that the FDA might already have the authority it needs to establish a limited “Conditional Withdrawal” pathway that could be used by the regulator in certain circumstances to promote good behavior by product sponsors.

How do things work now?

In general, FDA approval isn’t “conditional.” While some foreign regulators, like the European Medicines Agency (EMA) and Health Canada, can confer “conditional” approval for human drugs, the FDA cannot. Further, the FDA does not require companies to withdraw products should the confirmatory study that follows an accelerated approval ultimately fail. In sum, at the time the FDA approves a drug, there are relatively few conditions placed upon the approval decision.
The FDA generally has just one tool to impose conditions on the approval of a product: A Postmarketing Required study, also known as a PMR. The agency can only impose PMRs in a limited set of circumstances, such as for drugs granted accelerated approval or when the FDA wishes to assess a known or suspected serious risk. From these studies, the FDA is provided with confirmatory data it can use to determine if a drug is actually effective or safe. If the confirmatory study fails to show the expected benefit or demonstrates safety concerns, the FDA may then request that the sponsor to withdraw the product.
The FDA may also later require that a sponsor withdraw a product from the market if it feels that its original decision to approve is no longer appropriate. The FDA’s withdrawal procedures depend on the product’s original regulatory approval pathway, with separate processes for products granted accelerated approval or traditional approval. While the FDA may require products to be withdrawn for many different reasons, it most often does so for reasons of safety or efficacy.
However, the process of withdrawing a product from the market is difficult – especially if the sponsor of the product chooses to fight the FDA’s decision to withdraw. In the end, the FDA can pursue a withdrawal with the sponsor’s voluntary consent, which occurs in the majority of cases; rarely, the agency will pursue this course without the sponsor’s consent, which happens rarely. A company’s contesting the FDA’s decision to withdraw a drug may drag that process out for months – and even years, depending on the product, its indication and its regulatory history. [ Read AgencyIQ’s analysis of how FDA has authority to withdraw drugs from the market if they fail to confirm benefit.]
In context, the FDA often is asked to review drugs or biologics for serious conditions with unmet need for which the product offers a slight, uncertain or questionable benefit to patients. In these cases, regulators may have serious doubts about whether approval is fully justified based on the available data supporting clinical benefit, but their decision may fall on the side of making a product available to patients who have extremely limited (or no) options.
For example, in September 2022, the FDA approved Amylyx Pharmaceuicals’ Relyvrio for the treatment of amyotrophic lateral sclerosis (ALS) in adults. While the FDA noted that the analysis provided by the sponsor did “not persuasively demonstrate effectiveness from a statistical perspective,” it also explained that, “given the serious and life-threatening nature of ALS and the substantial unmet need, it is appropriate to exercise regulatory flexibility in considering these results.”

What’s the context?

Before the September 2022 FDA approval of Amylyx’s Relyvrio, another potential withdrawal model was raised by that company’s Chief Executive Officer during a meeting of the FDA’s Peripheral and Central Nervous System Drugs Advisory Committee. [ Read AgencyIQ’s coverage of that meeting here.] The FDA had convened that meeting – a highly unusual second advisory committee that followed an earlier meeting with a negative vote – following the sponsor’s submission of new analyses (but not new data) from its Phase 2 CENTAUR study, which Amylyx argued supported approval, but about which the FDA’s statisticians were skeptical.
The firm also had an ongoing Phase 3 randomized, placebo-controlled trial, called PHOENIX, that was still ongoing at the time the FDA was reviewing the drug for approval. The trial aimed to enroll 600 ALS patients and follow their survival and change from baseline on a functional rating scale, with a data readout planned for early 2024. This pending readout from a larger, Phase 3 study troubled reviewers tasked with making a decision on Phase 2 CENTAUR data: “This places the Agency in a challenging situation of potentially making a regulatory decision that may not be subsequently aligned with the results of the ongoing study,” FDA wrote in its briefing.
The FDA posed a key question at that second meeting: What would Amylyx do if the PHOENIX study showed that Relyvrio was no more effective than the placebo? As BILLY DUNN, then-director of the FDA’s Office of Neuroscience, argued at the meeting, Amylyx had already committed to Canadian regulators that it would withdraw Relyvrio if the PHOENIX study failed. Canada approved Relyvrio under the country’s conditional Notice of Compliance with Conditions pathway, which is based on “promising evidence” rather than the FDA’s approval standard of “substantial evidence,” allowing withdrawal of approval if CENTAUR failed or went uncompleted. In his opening remarks, Dunn said of Relyvrio’s Canadian approval status that “Arguably, that should make it easy for the company to make a similar public statement concerning the prospect of an approval of the current application.” Dunn then called on Amylyx to provide a “clear understanding” of its intent for Relyvrio should the PHOENIX study fail.
Immediately following Dunn’s introductory remarks, Amylyx Pharmaceuticals co-CEO JUSTIN KLEE affirmed that the company was prepared to withdraw Relyvrio if it failed the PHOENIX study. “To be clear, if PHOENIX is not successful, we will do what is right for patients — which includes voluntarily removing the product from the market,” he said.
But this was a verbal commitment by the company – not an ironclad agreement – which was quickly picked up on by the FDA’s Dunn. “The comment by the CEO of the company [Klee] at today’s meeting is not something that is a substitute or a replacement in any way for what we discussed with you, which was our regulatory authority to withdraw approval of the drugs,” Dunn said in addressing the committee. “Those two things are distinct and unrelated in that manner.”
Though Dunn said this request wasn’t to be interpreted as a bargain for approval, it was clearly an important contextual factor for reviewers. Committee member BRIAN TRAYNOR, a senior NIH investigator, said he was “struck by the public statement of the CEO on behalf of the company, convincing them to voluntarily withdraw the drug if the Phase 3 [trial] is negative, and in addition, the existence of an established method for the FDA to withdraw it if need be….” Another committee member, neurologist LIANA APOSTOLOVA of the Indiana University School of Medicine, said she was “somewhat reassured that if an approval is issued, it can be withdrawn in the future….”
Ultimately, the FDA’s Advisory Committee voted in favor of the drug on the second go-around. The committee voted 7-2 that the “available evidence of effectiveness [was] sufficient to support approval” of Relyvrio in patients with ALS, and the FDA ultimately came to the same conclusion and approved the drug using its standard review pathway (i.e., not accelerated approval).
However, neither the FDA’s approval letter nor the FDA’s review documents made any mention of Klee’s commitment at the advisory committee hearing. Notably, the approval letter also made no mention of the PHOENIX study – or any other efficacy-related confirmatory study. As AgencyIQ noted at the time of approval, whatever Klee said at the advisory committee meeting, there was no assurance that Amylyx would actually uphold its promise if the PHOENIX study failed, as it was under no obligation to do so.

What’s new?

In early March 2024, Amylyx announced the results of the PHOENIX Trial, which showed that Relyvrio “did not meet its primary endpoint of reaching statistical significance (p=0.667) as measured by change from baseline in the Revised Amyotrophic Lateral Sclerosis Functional Rating Scale (ALSFRS-R) total score at Week 48, nor was there statistical significance seen in secondary endpoints.” At the time the topline results were announced, the company said it was committed to engaging with regulatory authorities and making “informed decisions,” which “may include voluntarily withdrawing [Relyvrio] from the market.”
Then, on April 4, 2024 Amylyx announced that it had “started a process with the FDA and Health Canada to voluntarily discontinue the marketing authorizations for Relyvrio […] and remove the product from the market in the U.S. and Canada” based on the results of PHOENIX.
The company’s decision to withdraw its product won praise from some regulatory experts and media commentators. “Amylyx did the right thing by withdrawing its ALS drug. That’s rare and worthy of praise,” declared a headline in STAT News. “They walked the talk, and that’s fantastic,” said former Alnylam CEO JOHN MARAGANORE to Endpoints News. “This is something that we don’t see very often,” said STAT News journalist ADAM FEUERSTEIN in a podcast on the announcement. “We don’t see companies voluntarily withdrawing drugs from the market. This is one of those examples of putting patients first, ahead of business considerations.”
While the company is still pursuing other indications for the drug, including in Wolfram syndrome and progressive supranuclear palsy ( PSP), the development program for Relyvrio in ALS seems to have ended with the PHOENIX failure.

What are the implications?

There’s an argument that Amylyx’s move to voluntarily withdrawn in the U.S. (alongside its required withdrawal in Canada) should be considered a basic expectation – not “worthy of praise”. As a best practice, a company’s decision to withdraw a drug that did not meet its pre-specified benefit outcomes should be the minimum expectation to protect both patients, payers and taxpayers alike. Amylyx reported more than $380 million in net product revenue in 2023 from Relyvrio before the PHOENIX readout showing it was no better than placebo for ALS patients. Further, Amylyx’s initial press release in March 2024 did not necessarily confirm that they would withdraw the product; instead, the statement said that the company was considering options, “which may include withdrawing Relyvrio from the market.” (Emphasis added)
That language raises an important question about the nature of Amylyx’s commitment: Had the company identified any subgroups in the PHOENIX study who did see benefit, would it have fought the FDA to keep Relyvrio on the market for these groups, as other companies have in the past? Or, the NIH’s Bryan Traynor pointed out at the advisory committee hearing, what would have happened if Amylyx had appointed a new CEO who decided not to honor co-CEO Klee’s verbal commitment? As noted above, this was not a legally binding commitment on the part of the company.
But Amylyx’s approach may be instructive to other companies. Klee’s verbal agreement to withdraw Relyvrio if the PHOENIX study failed clearly had an impact on voting members of the FDA’s Advisory Committee, and perhaps on the FDA itself. In the future, other firms could use this commitment-based approach, if their products are intended to treat serious conditions with unmet need, are generally safe, but have marginal or questionable efficacy data to support an approval. However, the present lack of a mechanism for a binding commitment from companies is perhaps the biggest limiting factor for the FDA’s to accept such a method. Dunn’s request that Amylyx go on the record about withdrawal may have helped place public pressure on the company to fulfill an ultimately non-binding commitment.
This spotlights the differences between the experiences with Health Canada and the U.S. FDA: While the FDA must make difficult decisions about whether it can rely on non-binding promises from sponsors, Health Canada does have the authority to allow for conditional approval – as it did in the case of Relyvrio.
What, exactly, are FDA’s authorities on conditional approval? Right now, the FDA only has explicit statutory authority to grant conditional approval in only one product area: animal drugs. This authority was implemented under the Minor Use and Minor Species Animal Health Act in 2004. However, the same authority does not exist for human drug or medical device products.

An implied conditional approval pathway

However, the FDA may still have implicit authority to offer conditional approvals, according to an analysis by AgencyIQ. Here’s how it could work: Under the agency’s current withdrawal regulations, a drug’s sponsor may request that the FDA withdraw approval of a drug. “FDA will consider a written request for a withdrawal under this paragraph to be a waiver of an opportunity for hearing otherwise provided for in this section,” the regulation states. “Withdrawal of approval of an application or abbreviated application under this paragraph is without prejudice to refiling.” However, the regulation makes no mention of the FDA needing to accept this request, or to file it, immediately.
Theoretically, that means that a company could offer – or the FDA could request that sponsors provide – regulators a pre-filled request for withdrawal, and direct that FDA not file it until the pre-specified conditions for withdrawal are met. For example, a company in Amylyx’s position at an advisory committee hearing could instead offer a written, binding request for withdrawal to the FDA that would pre-specify the conditions under which a withdrawal would be initiated. Theoretically, the regulator could avoid filing the document until specific confirmatory data were available. In such a case, the FDA and the company might negotiate the specific data or conditions that might lead the withdrawal to be filed, or potentially a specific meeting that the company could have with the review team prior to the withdrawal action.
A theoretical Conditional Withdrawal pathway could help address some challenges – but create others. The FDA could employ regulatory flexibility in areas of high unmet need, but ensure that there are sufficient guardrails to ensure patient safety (including safety from spending on products that do not show benefit). However, regulators and sponsors often disagree about what constitutes a “failed” study – for example, a study may have several endpoints, or examine several subgroups. When positive results are seen in such subsets of trial data, firms may seek regular approval – but for a limited indication or population. At the very least, negotiations between the FDA and sponsors regarding what conditions should or should not result in withdrawal would be intense and high-stakes. There’s also possibility for misunderstanding or re-interpretation of the conditions for withdrawal at the time the full confirmatory dataset is made available.
The many challenges sponsors face in accrual and completion of confirmatory studies would still exist. When a product has entered the market and is available through regular channels or via compassionate use, patients may be reluctant to enroll in a clinical trial. Additionally, for rare and ultra-rare diseases, simply finding and enrolling sufficient patients to confirm benefit may be a challenge that would need to be anticipated in both the confirmatory trial’s statistical plan and any sponsor-FDA Conditional withdrawal agreement.
Like all approaches that lack specific statutory authority, this approach might be especially susceptible to legal challenges citing the recent “major questions” doctrine established under the Supreme Court case West Virginia v. EPA (2022), which found that Congress must generally authorize a regulatory agency to take a specific type of action, and that regulators may not use a “previously little-used backwater” of authority to create major new programs. [ Read AgencyIQ’s analysis of this case here.]
Ultimately, a conditional approval approach reliant upon FDA’s existing withdrawal authority is not all that dissimilar to the status quo, with one key difference. The FDA already has the authority to solicit companies to withdraw approvals. It may also pursue a withdrawal without the sponsor’s consent. If a company prospectively agrees to withdraw a drug, all that would change, process-wise, is the speed with which the FDA’s withdrawal would take place. Effectively, the company would give many of its current rights during the withdrawal process, such as a public hearing and a defined appeals process.
A regulatory approach of pre-defining future actions is also not without FDA precedent. Under the Consolidated Appropriations Act of 2023, FDA was granted explicit authority to authorize certain medical devices with a plan to monitor product performance over time, in expectation that certain changes would be made. The FDA refers to these as Pre-determined change control plans (PCCP). Under the FDA’s existing PCCP policy, the plans have primarily been used for products employing artificial intelligence or machine learning (AI/ML), but there’s also interest in using the policy for other technology types that may change over time, such as diagnostics or, potentially, products that might need changes to their sterilization methods. The core idea of a PCCP is to establish the baseline performance expectations of the product (which hold constant over time), as well as the sponsor’s plan to monitor post-authorization performance to ensure the product stays within those specifications. PCCPs also include the sponsor’s plan should the product fall outside its as-authorized specifications. By analogy, in a theoretical Conditional Withdrawal pathway, the sponsor’s commitment would be to meet specific goals in postmarketing studies, with a plan should those goals not be met. [ Read AgencyIQ’s analysis of the FDA’s Pre-Determined Change Control Plan draft guidance document.]
Some legislators have previously signaled that any FDA actions resembling Conditional Approval will require their explicit legislative sign off. In October 2018, three Democratic Senators sent a letter to the FDA, saying that its proposed Software Pre-Certification (Pre-Cert) pilot program was too conceptually similar to a “conditional approval” system, and therefore “not appropriate for human medical products.” Senator ELIZABETH WARREN (D-Mass.) followed up in 2019 with then-FDA Commissioner nominee STEPHEN HAHN. Warren pointed out that FDA has explicit authority to grant conditional approval for animal drugs and highlighted FDA’s lack of specific statutory authority to do so for human drugs. At the time, Warren stated “strong object[ion] to any expansion of the conditional approval pathway to human medical products.”
Some Republican lawmakers, however, have indicated support for Conditional approval: Republican Senator Mike Braun (R-Ill.) has repeatedly introduced legislation known as the Conditional Approval Act (now known as the Promising Pathway Act) which proposes the creation of a renewable two-year approval period for drugs based on a standard of approval of a finding of “substantial evidence of safety” and a “positive therapeutic outcome” from relevant early evidence.
A prospective conditional approach could also complement statutory authority that the FDA recently obtained from Congress after the passage of the Consolidated Appropriations Act of 2023. Under that law, provisions of which are more commonly referred to as the Food and Drug Omnibus Reform Act (FDORA), the FDA has the authority to require that confirmatory studies be “underway prior to approval, or within a specified time period after the date of approval, of the applicable product.” As more companies are forced by the FDA to start these studies prior to approval, regulators may have a better sense of which study endpoints would provide them with confirmatory data – and therefore, which ones could be relied upon for a Conditional Withdrawal. [ Read AgencyIQ’s analysis of the FDA’s new accelerated approval authorities and its impact on confirmatory studies.]
Relyvrio wasn’t the first example of a drug with a narrow case for approval, and it won’t be the last. Whether the FDA or legislators see value in the use of pre-defined promissory actions, however, may determine the frequency with which these approvals occur in the future.

To contact the author of this analysis, please contact Alexander Gaffney ( [email protected])
Featuring additional research and analysis from Kari Oakes and Laura DiAngelo
To contact the editor of this analysis, please contact Kari Oakes ( [email protected])