Amgen's $3.7 Billion Drug Is About to Face a Jury of Scientists
Amgen paid $3.7 billion for a drug the FDA now says was approved based on manipulated trial data. The company's response: hire Duke to re-crunch the numbers. Whether that's enough to save Tavneos could set a precedent for the entire industry.
Nine patients. That's how many it allegedly took to flip a failed drug trial into an FDA approval worth billions.
Now Amgen is betting that an independent team from Duke University can prove the drug actually works, before regulators pull it off the market for good.
The Drug That Launched a $3.7 Billion Bet
In October 2022, Amgen closed its acquisition of ChemoCentryx for roughly $3.7 billion in cash. The crown jewel: Tavneos (avacopan), a first-in-class pill for a brutal autoimmune condition called ANCA-associated vasculitis, where the immune system attacks blood vessels in the kidneys, lungs, and other organs.
Tavneos had earned FDA approval a year earlier, in October 2021, based on a single pivotal trial called ADVOCATE. The pitch was compelling. Patients on Tavneos could dramatically reduce their dependence on high-dose steroids, which cause a cascade of nasty side effects from bone loss to diabetes. In ADVOCATE, 65.7% of Tavneos patients maintained remission at one year, compared with 54.9% on the standard steroid taper.
For a rare disease with few good options, that looked like a win. Amgen swooped in and bought the whole company.
Then the FDA started pulling at threads.
The Accusation: Cooked Books, Hidden Results
On April 27, 2026, the FDA's drug evaluation center (CDER) dropped a bombshell: it formally proposed to withdraw Tavneos from the U.S. market. The agency didn't mince words. It accused the original applicant of submitting "untrue statements of material fact" in the drug application.
The alleged scheme is almost comically brazen. According to FDA, the original analysis of ADVOCATE's primary endpoint (the main thing the trial was designed to prove) did not show the drug worked. But after seeing those disappointing results, unblinded study personnel went back and re-reviewed just 9 out of 331 patients, cherry-picking cases to re-adjudicate. They then submitted only the favorable re-analysis to the FDA.
Get tomorrow's biotech intelligence before your competitors.
Join thousands of biotech professionals who start their day with our free, daily briefing.
The original, unfavorable analysis? Never disclosed.
Think of it like a teacher who grades your test, sees you failed, then quietly changes nine answers and submits the new score. Except the test is worth $3.7 billion and people's health is on the line.
FDA now says it "can no longer" consider ADVOCATE a credible demonstration that Tavneos works. And the agency didn't stop there: it also flagged serious post-marketing liver injuries, including a rare and severe condition called vanishing bile duct syndrome. According to FDA communications, there have been 76 cases of drug-induced liver injury, 54 hospitalizations, and 8 deaths potentially linked to the drug. In Japan, roughly 20 deaths among approximately 8,500 treated patients prompted a recommendation to halt new prescriptions.
Amgen's Hail Mary: Call in Duke
Faced with the potential loss of its multibillion-dollar asset, Amgen declined FDA's request to voluntarily pull Tavneos. Instead, the company is fighting back with an unusual weapon: an independent, blinded re-analysis of the ADVOCATE data by the Duke Clinical Research Institute (DCRI).
The review started in February 2026. In a June 1 letter to the FDA, Amgen disclosed the effort and committed to submitting the results by June 29, which coincides with the close of the public comment period on FDA's withdrawal proposal.
The strategy is clever. By commissioning a respected academic institution to re-crunch the numbers from scratch (blinded, using prespecified rules), Amgen is essentially saying: "Don't take our word for it. Don't even take ChemoCentryx's word for it. Take Duke's."
Amgen also publicly stated it is "not aware of any issues with the underlying patient data," pushing back on the manipulation narrative. On safety, the company argues there have been no known U.S. deaths linked to Tavneos liver injury, though FDA's global data tells a darker story.
Why This Might Not Be Enough
The Duke review could theoretically show that Tavneos works even without the disputed post-hoc changes. If a fully independent, blinded re-adjudication still demonstrates a meaningful treatment effect, that would be a significant data point in Amgen's favor.
But there's a problem regulators care about more than just numbers: trust.
FDA's case isn't solely about whether the drug works. It's about the fact that unfavorable data was hidden and favorable data was manufactured through selective re-review. That's a process violation, and no amount of after-the-fact re-analysis can un-ring that bell. Even if Duke's numbers look good, the agency could argue that the integrity of the original submission was fatally compromised.
It's the difference between being acquitted because you're innocent and being acquitted on a technicality. The FDA may not see a clean Duke analysis as full exoneration.
The original 2021 advisory committee was already split on whether ADVOCATE's data justified approval, and the FDA's primary review team at the time did not recommend approval. The drug squeaked through once. Doing it again, with data integrity allegations hanging overhead, will be much harder.
The Bigger Picture: A Precedent in the Making
Regulatory watchers are paying close attention to this case for reasons that go well beyond one drug. If Amgen succeeds in using an external academic re-analysis to fend off a withdrawal, it could create a playbook for other companies facing similar scrutiny. Got a data integrity problem? Commission an independent review and hope the new numbers are strong enough to override the original sin.
On the other hand, if FDA proceeds with withdrawal regardless of the Duke findings, it sends an equally powerful message: hiding unfavorable data from regulators has consequences that can't be fixed retroactively, even by a $3.7 billion acquirer armed with Duke statisticians.
For patients with ANCA-associated vasculitis, the stakes are personal. Tavneos represented a real step forward in reducing the steroid burden that makes treatment almost as miserable as the disease itself. Losing access would be a genuine setback, especially for patients already on the drug.
What Happens Next
Tavneos remains on the U.S. market while the regulatory process unfolds. The public comment period closes June 29, 2026, the same deadline Amgen set for submitting the Duke analysis. After that, FDA will review all submissions and decide whether to press forward with withdrawal, modify its position, or negotiate some middle ground (think: restricted labeling, a risk management program, or mandatory liver monitoring).
Analysts broadly see material downside risk for Tavneos in the U.S., treating the Duke review as providing some "option value" rather than a likely path to vindication. The consensus framing: this is legacy risk from the ChemoCentryx acquisition, not a signal of broader problems at Amgen.
But for the rare disease community, and for anyone who cares about how drugs get approved and stay approved, this case is one to watch closely. The answer to a deceptively simple question hangs in the balance: can good data redeem a bad process?
Clinical & Regulatory
4 min read
Roche Just Crashed the Obesity Party With a Number Nobody Expected
Roche's obesity drug enicepatide posted 22.5% weight loss in 48 weeks, landing squarely in Zepbound territory and turning Roche from a latecomer into a legitimate contender. The real question: can it compete with what Lilly and Novo are cooking up next?
