Eliza Ward: Hey. You doing alright? Because I've been staring at this FDA story for an hour and I'm still — okay, I'm still not sure I have the sequence right.
Brian Reed: Which part tripped you up?
Eliza Ward: The firing. Tracy Beth Høeg — she's acting director of CDER — signs a formal withdrawal letter to Amgen on April 27th. The letter says Tavneos lacks substantial evidence of effectiveness. Eighteen days later, she's fired.
Brian Reed: Eighteen days.
Eliza Ward: After she declined to resign. That's the confirmed sequence. What I can't — I mean, causation is a different question.
Brian Reed: Right — but the part that doesn't fit is: Tavneos was approved in October 2021 for ANCA-associated vasculitis. So you've got five years of a legal, marketed drug. And now the original developer, ChemoCentryx, is accused of manipulating the ADVOCATE trial data that got it approved in the first place.
Eliza Ward: And Amgen bought ChemoCentryx for $3.8 billion in 2022. A year after approval.
Brian Reed: So what does Amgen say now — are they contesting the withdrawal?
Eliza Ward: They're fighting it, yeah. Amgen commissioned Duke Clinical Research Institute to do an independent re-adjudication of the ADVOCATE endpoints. That's their answer. The manipulation question first, because the mechanics of it are actually the story.
Brian Reed: Walk me through what actually happened in the trial.
Eliza Ward: Imagine a teacher grades a final exam, and the class average isn't quite high enough to pass. Then someone who already knows the average goes back and quietly re-grades nine students' answers upward. That's November 2019. Unblinded ChemoCentryx personnel — people who knew which patients were in which arm — re-adjudicated primary endpoint assessments for nine patients. Nine.
Brian Reed: Nine patients out of a rare-disease trial — and that moved statistical significance?
Eliza Ward: That's the thing. The margin was that thin. Nine re-adjudications flipped a non-significant result. Which — I mean, that tells you something about the underlying data even before you get to the fraud question.
Brian Reed: And the ADVOCATE trial was the only pivotal trial. So those nine patients weren't a rounding error in a larger case. They were — actually, the FDA's withdrawal proposal says it was the sole study establishing substantial evidence of effectiveness. So the manipulation wasn't eating into a margin. It was the entire foundation.
Eliza Ward: Right. Rare-disease policy is what made that possible — FDA can approve on a single trial when patient populations are too small to run two. That's not a loophole. That's the deliberate design. But it means the leverage on any one study is enormous.
Brian Reed: And the NEJM ran those ADVOCATE results in 2021 and didn't retract until July 2026 — which is three months after Høeg's withdrawal letter. The EMA separately moved to revoke European authorization on the same data integrity findings. So what moved the NEJM? Did the FDA investigation make the case, or did they wait for the EMA?
Eliza Ward: Honestly — we don't know the sequencing inside the NEJM. That's a real gap. What we know is the retraction followed the FDA findings. Whether it led or lagged the EMA decision, I can't say with confidence.
Brian Reed: But hang on — there's a framing going around that I want to push on, because I keep seeing it. People are saying the data fraud killed eight people. And I don't think that holds.
Eliza Ward: Yeah, I've seen that too. And it's — actually, no, it's wrong in a specific way.
Brian Reed: Because the eight deaths come from the FDA drug safety communication — March 31st, 2026. Seventy-six cases of drug-induced liver injury, 54 hospitalizations, 8 deaths. That's a hepatotoxicity signal. That's real-world exposure across 25,000 patient-years. It has nothing to do with what ChemoCentryx personnel did to nine endpoint assessments in November 2019.
Eliza Ward: They're on completely different timelines. A patient who woke up jaundiced in early 2026 — liver biopsy confirms vanishing bile duct syndrome, which is one of the seven biopsy-confirmed VBDS cases in that communication — that harm exists whether or not the ADVOCATE trial was manipulated. Those are two separate questions.
Brian Reed: Which is — I mean, that actually makes both problems worse to separate out. Not better.
Brian Reed: Because if you conflate them, you let the fraud question absorb all the moral weight — and then the hepatotoxicity signal, which is its own documented disaster, gets treated like a footnote. Seventy-four serious outcomes. That's not a footnote. But it also means the fraud question stands independently — patients were exposed to a drug whose efficacy was, let me see, unproven on legitimate data. That's the harm from the manipulation. Not the liver injuries.
Eliza Ward: Wait — and that's actually the sharper charge against Amgen. Not that the drug poisoned people through fraud. It's that the efficacy was never established, so 25,000 patient-years of exposure happened on a foundation we can't verify.
Brian Reed: And whether the Duke re-adjudication changes any of that — that's the part we haven't gotten to yet, and it matters a lot for what comes next.
Eliza Ward: And we don't know what it found. That's — the Duke re-adjudication is the crux of Amgen's whole defense, and the outcome just isn't in the available reporting.
Brian Reed: Because the July 2026 hearing — the FDA scheduled it, and then, from what we can tell, nothing comes out the other side. No reported outcome. Tavneos is still in some kind of limbo where the FDA wants it off the market and Amgen is saying the data remain valid, that approval reflected a totality of evidence. Those are Amgen's words. And both things are simultaneously true right now.
Eliza Ward: Which is — I mean, that's actually unusual. The FDA proposes withdrawal, the company fights back with an independent re-adjudication commissioned from Duke Clinical Research Institute, and the regulatory status just... sits there.
Brian Reed: And while it sits there — this is the part that I can't shake — 25,000 patient-years of exposure happened. Which probably means somewhere between three and five thousand actual people. Not patient-years. People. And nobody is tracking publicly what happened to the ones who were mid-treatment when the fraud came to light.
Eliza Ward: Three to five thousand.
Brian Reed: Let me see if I can make that concrete. A rheumatologist — say, somewhere without a major academic center nearby — she's got a patient with microscopic polyangiitis. She prescribed Tavneos because it was FDA-approved, CDER-cleared, legal. Now it's, what, June 2026? The drug is still technically on the market. Does she keep the patient on it? Does she switch? Switch to what? The guidance isn't there.
Eliza Ward: And CDER — the same division Høeg was running — is the one that approved it in 2021 and is now the one proposing withdrawal. That internal tension is real. The institution is essentially in conflict with its own prior decision.
Brian Reed: If withdrawal actually happens, we're looking at October 2021 to sometime in late 2026 at minimum. Five years. And the EMA moved separately — same data integrity findings — so the European side is already trending toward revocation. Does the FDA follow, or does Duke's re-adjudication give Amgen enough runway to stall further?
Eliza Ward: That's the only concrete signal to watch. What Duke actually found, and whether the July hearing produced any formal record. Everything else is inference until that lands.
Brian Reed: And the official reason Høeg was fired hasn't been confirmed. That's not a detail. That's the center of the retaliation question. We have sequence. Eighteen days. We don't have causation from any official source.
Eliza Ward: Which is — I mean, that's where I'd want to be careful. If the EMA moved on the same findings, and the NEJM retracted on the same findings, and the FDA investigation was thorough enough to produce a formal withdrawal letter — why is U.S. regulatory status still unresolved? That gap is real and I don't have a clean answer for it.
Brian Reed: So the honest version is: either Tavneos gets withdrawn — and it will have taken over five years, cost an official her position, and required a journal retraction to get there — or it stays approved. And if it stays, that's Amgen's Duke re-adjudication holding the line. Which would mean documented manipulation can be defended indefinitely if you have the resources to commission your own independent review.
Eliza Ward: That's the structural question. Not just for Tavneos. What changes in how CDER handles post-market integrity so this path — five years, one whistleblower, a firing — isn't the only route from manipulation to withdrawal?
Brian Reed: And we don't know what Duke found. That's where it sits.
