Follow-up update | October 31, 2026
On October 29, the court granted final approval to a $69 million securities settlement. The funds have been placed in escrow; actual payment to investors has not yet been verified. This is not an FDA withdrawal hearing and does not resolve compensation for patient injuries. The original article below retains its information date.
Before the Conclusion | After Approval: Medicines Reconsidered, Case One
A drug's pivotal clinical trial has run into problems, the paper has been retracted, and the regulator has proposed withdrawing approval. Intuitively, the story might seem close to its end. But in the United States, avacopan's story continues: doctors can still prescribe it, some patients believe it has helped them reduce steroids, and the company has submitted new analyses in an effort to keep it available.
If the original trial was unreliable, why might patients still feel they benefited? If some people really do benefit, why is the regulator still seeking withdrawal?
Understanding this case requires us to consider patients' improvement, the drug's additional contribution, and the trustworthiness of the research process together. These questions are related, but none can stand in for the others.
It sought to address a real treatment problem
Avacopan, marketed as Tavneos, is used for granulomatosis with polyangiitis and microscopic polyangiitis, forms of severe active ANCA-associated vasculitis. These diseases can damage organs including the kidneys and lungs. Treatment must both control inflammation and limit the harms of long-term therapy.
Glucocorticoids can suppress inflammation rapidly, but may bring infection, bone, and metabolic burdens. Developers hoped that blocking the complement C5a receptor and reducing the amplification of inflammation involving neutrophils could help patients maintain disease control while using fewer steroids. Experimental research supported this biological pathway, but clinical comparisons were still needed to establish how much additional benefit patients would receive.[1,12]
The pivotal ADVOCATE trial randomized 331 patients. Both groups received background immunosuppressive treatment. One group also received avacopan for 52 weeks; the other received a protocol-specified prednisone taper, with study-provided prednisone stopped by week 21. Additional steroids were permitted in specified circumstances. The experimental group was therefore not entirely steroid-free, and the control group did not stop all treatment after 20 weeks.[1,3]
The steroid reduction here was, first of all, a planned treatment strategy. We cannot count taking fewer steroids as proof that the new drug works, then use that “proof” to explain why tapering is possible. What needs to be tested is whether disease remains controlled with that reduction, and whether the reduction in steroid harm is sufficient to offset the risks of the added treatment.
How could problems arise while central adjudicators remained blinded?
One important trial endpoint was sustained remission at week 52: patients had to be in remission at both weeks 26 and 52, with no relapse in between, and meet rules on steroid use during a period before assessment. This judgment involved not only the Birmingham Vasculitis Activity Score, or BVAS, but also whether abnormalities reflected active disease, previous damage, or other causes.[1,2]
Central adjudication was intended to apply consistent standards to cases from different hospitals and allow experts to assess them without knowing treatment assignments. Yet it still required clinical interpretation and depended on which information was submitted and which cases were referred for reassessment.
At the FDA's 2021 advisory committee meeting, the agency had already pointed out that a reanalysis using local investigators' BVAS assessments no longer supported statistical superiority at week 52. The discussion also concerned how central-adjudication forms handled persistent activity. This showed that the conclusion was sensitive to the assessment method, but a methodological difference alone did not establish manipulation. Central adjudication was the prespecified primary method and could also correct problems in local scoring.[2]
The later investigation examined a different part of the process. According to CDER's documents, the first database lock, unblinding, and analysis in 2019 did not produce a result meeting the superiority threshold. Company personnel who knew treatment assignments then selected nine patients for readjudication. Five patients in the avacopan group changed from “not in sustained remission” to “in sustained remission,” taking the analysis across the statistical threshold. The first analysis and the related process were not accurately disclosed.[3]
This must not be merged with the 2021 methodological debate as though they were the same evidence. One concerned the robustness of the conclusion; the other concerned changes to endpoints after the results were known.
The central adjudicators receiving the cases still did not know the treatment assignments or that an initial analysis had already taken place. The retraction notice also stated that two academic authors had previously been unaware of this process. We therefore cannot assume that all the adjudicators knowingly cooperated.[3,4]
But the experts' lack of knowledge cannot remove the effect of case selection. Imagine that after an examination, someone who knows the scores sends only selected papers that might gain points for remarking. Even if the examiner cannot see the students' names, that cannot undo the prior selection.
The regulator's findings and the company's response
On April 27, 2026, the FDA's Center for Drug Evaluation and Research, CDER, proposed withdrawing approval. It found that the process constituted data manipulation and that the marketing application contained untrue statements of material fact. The company disputed this finding, arguing that the procedures corrected the application of endpoint rules, that the underlying clinical data had not been altered, and that adjudicators remained blinded.[3,6]
We need to distinguish the regulator's explicit findings, the company's rebuttal, and whether a court has issued a judgment on the merits of the conduct. We cannot reduce them all to “each side has its own account,” nor expand a regulatory document into a criminal conviction of every research participant.
The case also shows that data integrity extends beyond numbers on laboratory reports. How medical records become “endpoint met” or “endpoint not met” is part of the production of research evidence too. The timing of database lock and unblinding, who can initiate readjudication, and whether a complete, verifiable record exists all affect whether results can be trusted.
How much evidence can a new adjudication restore?
Amgen subsequently commissioned a team at the Duke Clinical Research Institute to conduct an independent, blinded readjudication of the original trial's primary endpoints. It reexamined records from the same patients; it did not conduct a new randomized trial.[5]
The public preprint reported sustained remission rates at week 52 of 61.4% versus 52.4%, with an adjusted between-group difference of 9.8 percentage points and a 95% confidence interval from −0.3 to 19.9 percentage points. That interval allows for some additional benefit, but also for no additional benefit. The result met the original noninferiority margin but did not meet the superiority criterion.[5]
The new analysis therefore cannot be translated as “proven ineffective.” Noninferiority and superiority answer different questions. The former asks whether, within a prespecified margin, the new regimen is not worse by more than the amount allowed. The latter requires sufficient evidence that it is better. Meeting noninferiority does not, on its own, establish whether the margin was appropriate or whether the comparator regimen could adequately test the new drug's additional effect.
The parties also interpret the approval requirements differently. CDER emphasizes that it had required superiority for sustained remission at week 52 as early as 2016. Amgen emphasizes that the eventual approval was for adjunctive treatment, without a claim of completely replacing steroids, and that the evidence should be assessed as a whole. The readjudication results warrant discussion, but do not by themselves remove the original trial's data-integrity problems.[3,6]
How much favorable evidence remains for a drug, and whether the available material can continue to support approval, are related but distinct judgments. This is why even a more favorable reassessment may be unable to restore the original study's suitability as a basis for approval.
Why improvement in routine care cannot settle the dispute directly
The company also cites routine-care data. An Amgen-funded public preprint summarized 71 real-world observational studies, 53 of them conference abstracts, and reported favorable signals of disease control.[7]
These data can show that many patients receiving regimens containing avacopan reached remission. But background immunosuppressive treatment, doctors' decisions about who starts the new medicine, and which patients complete follow-up all influence the results.
Without a sufficiently comparable control group, we observe “what happened after patients received this regimen,” but have greater difficulty establishing “how much of the benefit came specifically from adding avacopan.” Increasing the number of studies does not automatically remove comparison problems they share.
Routine-care studies and readjudication of the original trial may also involve different patients, follow-up, and definitions of remission. We cannot directly subtract the strictly defined sustained remission rate in a randomized trial from a high remission rate in a review to decide which data are “more real.”
Patient experience, clinicians' observations, and comparative research each provide important information. Respecting patients' experiences of improvement is compatible with requiring a reliable estimate of additional efficacy.
Benefit to some patients and serious harm can coexist
Reviewing data through October 9, 2024, the FDA identified 76 cases of liver injury with reasonable evidence of a causal association with the medicine. Eight reported deaths, and seven involved biopsy-confirmed vanishing bile duct syndrome. This injury involves damage to and loss of bile ducts within the liver and can cause permanent harm.[8]
The number 76 is the count of identified cases, not the total number of people who took the drug. Dividing eight by 76 therefore cannot yield the overall mortality rate. But these cases also cannot simply be treated as unevaluated coincidences following treatment.
A treatment can improve disease in some patients while also causing serious injury. Both belong in the same comparison: compared with other feasible treatments, how much additional benefit do patients gain from choosing a regimen containing avacopan, and how much additional risk do they face? When the size of the benefit becomes uncertain, the same risks require reassessment.
Why can it still be used in the United States when other regions have acted?
As of this article's verification date, the FDA website still listed proposed withdrawal. No published final US withdrawal order was located in the public materials searched for this article. The company has procedural opportunities to request a hearing and submit evidence, but a request does not mean that a formal hearing has been granted.[3]
The European Union revoked the authorization on August 4, 2026. In the United Kingdom, new patients could no longer start treatment from September 1, while existing patients were given six months to transition, with revocation planned for March 1, 2027. Decisions and transition arrangements in different jurisdictions cannot simply be substituted for one another.[9,10]
How to transition a stable patient safely and whether to start treatment in a new patient are clinical questions that need separate assessment. Continued availability for sale in the United States also does not mean doctors must wait until every legal procedure ends before revisiting the evidence and treatment options.
The case additionally involves securities litigation and patients' rights. A securities settlement addresses claims by specified investors; it does not automatically address injuries patients have already suffered. These procedures and the boundaries of compensation will be compared in the concluding article, After a Drug Is Withdrawn, Who Bears Patients' Losses?.[11]
The central question avacopan leaves us is how to move from “may have helped some people” to “sufficient to justify patients' continued exposure to risk.” A plausible mechanism can explain why research is worthwhile, and patients' improvement can suggest where value may lie. Reliable clinical recommendations also require knowing how large those benefits are, whom they apply to, and whether they outweigh the costs.
The next article turns to Relyvrio. There, the main difficulty is how much confidence early favorable results should give us when patients cannot afford to wait.
Principal sources
[1] ADVOCATE: Original study protocol. Avacopan and the prednisone taper were compared against a background of standard treatment.
[2] FDA, May 6, 2021: Advisory committee transcript. Includes differences between local scoring and central adjudication, persistent activity, and steroid-use rules.
[3] FDA, April 27, 2026: Proposed withdrawal announcement and investigation document.
[4] The New England Journal of Medicine, June 29, 2026: Retraction notice.
[5] The Duke team's Amgen-funded August 2026 preprint on primary-endpoint readjudication and September full analysis. These public versions concern the same readjudication and are not counted as two independent trials. This article interprets the findings through the confidence interval to avoid mixing one-sided and two-sided P-value thresholds.
[6] Amgen: Supporting material for its response and hearing request. Used to present the company's position.
[7] Ibiloye et al., June 2026: Preprint of the real-world systematic review. Funded by Amgen; study designs and publication formats are identified separately.
[8] FDA, March 31, 2026: Safety communication on serious liver injury.
[9] EMA: Medicine record and August 4, 2026 authorization revocation.
[10] MHRA: UK arrangements for new and existing patients and planned authorization-revocation date.
[11] Court-authorized website: ChemoCentryx securities litigation procedures and proposed-settlement information.
[12] Schreiber et al., JASN, 2009: Experimental research on the C5a receptor and ANCA-associated inflammation.
Sources checked through October 28, 2026. This article is for the exchange of medical and scientific information and is not individual medical advice. Discuss specific treatment changes with a doctor. Preprints, the company's position, regulatory findings, and judicial procedures are identified separately.
Before the Conclusion: there are still questions worth pursuing.
Medicine · Science · Ideas and Reading
Follow Before the Conclusion to stay curious and think independently together.
Text and image use statement
The text of this article was written originally by Before the Conclusion. You are welcome to share the article's web link. Without separate permission, please do not republish or excerpt its text by copying, reposting, publishing screenshots, or similar means. Unless otherwise indicated, its illustrations may not be separately republished, copied, modified, or used commercially. Rights in cited materials belong to their respective rights holders.