Why Are Approved Drugs Withdrawn?

Before the Conclusion | After Approval: Drugs Reconsidered · Series introduction

A drug has been approved, doctors have prescribed it, and patients have used it for some time. Later, a larger trial fails to produce the expected results, or a serious harm gradually emerges, and regulators begin discussing withdrawal of approval. Faced with such news, people naturally ask: if problems were found later, why was the drug allowed onto the market in the first place?

The question deserves investigation. To answer it, however, we first need to establish what actually happened afterward.

Sometimes we obtain new evidence that did not exist at the time. Sometimes the original positive findings fail to withstand more thorough testing. Sometimes an investigation finds that the evidence originally submitted was itself unreliable. All these events may lead to withdrawal, but they involve different scientific questions and different responsibilities.

Understanding these distinctions can help us judge what a withdrawal story actually tells us. It also concerns a question closer to patients' lives: why do we trust a treatment today, and under what circumstances should we change that decision tomorrow?

The evidence for approval always has a scope

We often say a drug “passed FDA approval,” which can sound like a permanent endorsement of its overall quality. In fact, approval always concerns a particular disease, particular patients, and specified doses and methods of use. It rests on the evidence available at the time and the resulting assessment of benefits and risks.

That assessment cannot encompass everything the future will bring. Preapproval clinical trials provide important information, but they are limited by patient numbers, eligibility criteria, and observation time. A less common harm may become apparent only after more people receive treatment. A risk associated with long-term use may require longer follow-up to identify. The FDA therefore includes postmarket monitoring as part of drug development, rather than treating market entry as the end of research.[1]

But “trials have limitations” cannot explain every problem. An unforeseeable risk and a signal that had already appeared but was downplayed or inadequately investigated require separate scrutiny. Determining which applies to a particular withdrawal requires returning to the information available then and examining how those involved handled it.

First identify what changed the judgment

One situation is that the research supporting approval loses credibility. If researchers learn treatment assignments and then selectively revise endpoint assessments, the eventual difference may include the effects of human selection. The first problem then concerns how the evidence was produced, rather than simply how large the drug's effect was.

A second situation is that subsequent research fails to confirm clinical benefit. An early, small study may show a positive signal that later research does not reproduce. That result alone does not establish fraud. It requires us to reassess how likely the drug is to help patients and whether the original degree of confidence was sufficient.

A third situation is that new harms change the balance of benefits and risks. A drug may still relieve pain, control inflammation, or delay disease progression, yet an additional serious risk makes its previous use no longer justifiable. The question is whether the benefits patients receive are enough to offset the costs.

These situations can overlap. When the evidence for efficacy is weakened, risks that were already present become harder to accept because the benefit used to justify patients taking those risks has also become uncertain.

The following five stories show how these judgments can change.

Select a case or question to open the full article. On a phone, the table can be scrolled horizontally.

“Market withdrawal,” “withdrawal of approval,” and “withdrawal of an indication” in this table need to be understood separately. Sales may stop at a different time from the legal withdrawal of approval. Removing a use for one disease does not mean removing every use of the drug. For people receiving treatment, the country concerned, the patients affected, and any transition arrangements also matter.

Why approval sometimes comes before additional evidence

For rapidly worsening diseases with limited treatment options, waiting for stronger results has a real cost. Another year's wait may mean that some patients lose their chance of treatment. Drug approval therefore always faces a difficult task: reducing ineffective treatment and harm while also accounting for losses caused by waiting.

Accelerated approval is one approach to this problem. It allows eligible drugs to receive earlier approval based on a surrogate endpoint or an intermediate clinical endpoint considered “reasonably likely to predict clinical benefit,” with subsequent studies required to verify the expected benefit. Approval may be withdrawn if verification fails.[2]

However, drugs withdrawn after entering the market have not all used accelerated approval. Relyvrio received traditional approval. Its story concerns the use of regulatory flexibility to accept greater residual uncertainty in a serious disease. The two pathways cannot be treated as the same system simply because both can involve “market entry first, new results later.”[5,10]

Accepting some uncertainty requires corresponding arrangements for research and correction. When subsequent trials are completed, whether their findings are disclosed on time, and how people act on unfavorable evidence all influence what earlier access ultimately brings patients. We must assess both the approval decision and whether responsibilities for continued verification were fulfilled afterward.

A patient's improvement and a drug's additional contribution

Withdrawal disputes often feature accounts that are difficult to ignore. Some patients feel their disease has stabilized; some can finally reduce other treatments; others believe a drug has bought them time. These experiences can tell us what patients value and which changes warrant further study.

But evidence is still needed to connect “improvement occurred after taking the drug” with “adding the drug caused additional improvement.” Patients may also be receiving other effective treatments, and disease can fluctuate naturally. Even continued deterioration after treatment may represent real benefit if it is slower than in an appropriate control group. What we need to compare is what different treatments produce when other conditions are as comparable as possible.

Safety requires similar distinctions. An event after taking a drug does not automatically prove the drug caused it. Researchers need to examine the clinical course, other possible causes, and comparative data to assess causality. The FDA explicitly distinguishes an “adverse event” reported after drug use from an “adverse reaction” caused by the drug.[11]

Patients' experiences and comparative research can both inform judgment, but they have different tasks. Respecting someone's improvement does not require turning their experience directly into an efficacy conclusion for every patient.

Questions remain after withdrawal

If a drug loses the basis for remaining on the market, regulators still have to address the circumstances patients are already living in. Who needs to change treatment? How can problems from abrupt discontinuation be avoided? Is there a population suited to restricted use? Which evidence is still worth collecting? These questions require specific arrangements. A withdrawal announcement is only one step.

Public assessment should also extend beyond the two judgments “regulators never make mistakes” and “no approval can be trusted.” More useful questions are whether the original conclusion faithfully reflected the available evidence, whether uncertainty was clearly communicated, whether new problems were promptly investigated, and whether the process of changing the decision was transparent.

Mylotarg especially reminds us that withdrawal can sometimes leave room for renewed research. A failed regimen may be replaced by a new one, but the change requires new evidence. The reasons for returning to the market must address, one by one, the problems that led to departure.

We also need to ask what happens to losses already incurred after a decision changes a drug's status. Continued supply or a switch to another drug addresses continuity of treatment. Refunds concern money already paid. Compensation for personal injury must address specific harm. Investor claims and government enforcement each have their own subjects. They may all concern the same drug, but none can substitute for another.

Withdrawal news therefore has two distinct follow-up questions: whether future patients will be better protected, and whether those already affected can obtain appropriate redress. The latter requires separate evidence and procedures. A change in the regulatory decision alone does not settle responsibility.

This series has seven articles: this introduction, five drug cases, and a closing article on patients' losses and accountability. Follow the questions in the table, then return to “After a Drug Is Withdrawn, Who Bears Patients' Losses?” to compare which arrangements have been implemented and which questions remain open.

We begin with avacopan, sold as Tavneos. Its original developer, ChemoCentryx, was acquired by Amgen in 2022.[13] This case raises a question more fundamental than disappointing later results: if the process that produced a pivotal study's conclusions can no longer be trusted, what can we rely on to decide whether patients should continue bearing its risks?

Main sources

[1] FDA: Postmarket drug safety monitoring.

[2] FDA: The Accelerated Approval Program and subsequent verification.

[3] FDA, April 27, 2026: Proposed withdrawal of Tavneos approval.

[4] EMA: Tavneos medicine record and the August 4, 2026 revocation of authorization.

[5] Amylyx, April 4, 2024: Relyvrio market withdrawal and patient transition arrangements.

[6] FDA / US Federal Register, August 29, 2025: Formal withdrawal of Relyvrio approval.

[7] EMA, October 6, 2004: Statement on the worldwide withdrawal of Vioxx.

[8] FDA, November 18, 2011: Decision to withdraw Avastin's breast cancer indication.

[9] FDA researchers, 2018: Evidence supporting Mylotarg's return to approval.

[10] FDA, September 27, 2023: ALS drug review background materials, identifying Relyvrio as a traditional approval.

[11] FDA: Understanding postmarket safety monitoring and adverse-event data.

[12] FDA / US Federal Register, November 28, 2011: Effective record of the formal withdrawal of Mylotarg approval. The company's 2010 commercial withdrawal and the formal withdrawal of approval in 2011 are different dates.

[13] Amgen, October 20, 2022: Completion of the ChemoCentryx acquisition, bringing Tavneos into Amgen's portfolio.

Sources checked through October 28, 2026. This article is for discussion of medical and scientific information and does not provide individual medical advice. Discuss specific treatment changes with your doctor.

Beyond the conclusion, there are still questions worth asking.

Medicine · Science · Ideas and reading

Follow Before the Conclusion to stay curious and think independently together.

Text and image usage

The text of this article was originally written by Before the Conclusion. You are welcome to share the article's webpage link. Without separate permission, do not republish or excerpt the text by copying, reposting, redistributing screenshots, or similar means. Unless otherwise marked, article illustrations may not be separately republished, copied, modified, or used commercially. Rights in cited materials belong to their respective rights holders.