AstraZeneca's Wainua Missed in ATTR-CM: The Stabilizer Problem

AstraZeneca and Ionis spent a hundred and forty weeks proving their drug does exactly what it was designed to do — and lost the trial anyway.

Meanwhile Mounjaro picked a fight with a drug that already works and won it narrowly, BioNTech pulled a cancer vaccine after more deaths in the vaccine arm, and Novartis is committing outcomes-trial money to a target Novo just failed at. Let's take the week apart.

Welcome to The Pharma Closeout Week in Review for Sunday, August thirtieth. I'm Alex Mercer.

And I'm Maya Patel. A lot happened this week that deserves a second look. Let's get into it.

So — Wainua. AstraZeneca and Ionis, the CARDIO-TTRansform post-mortem in Munich, and here is the number that carries the whole week: three hundred eighty-one primary endpoint events in the Wainua arm against three hundred ninety-two on placebo. Cumulative cardiovascular death and recurrent cardiovascular events, through a hundred and forty weeks. Published simultaneously in the New England Journal.

Say that gap again.

Eleven events. That's the separation after nearly three years, in a disease that affects something like three hundred to five hundred thousand people worldwide. The miss itself broke in July and it genuinely shocked the street — Jefferies called it an outright failure at the time. Munich is where AstraZeneca had to stand up and explain it.

And the explanation is the story. Fifty-seven percent of these patients came in already on a background stabilizer. Another twenty-four percent started one during the trial. So you're asking whether a silencer adds cardiovascular benefit on top of stabilizer therapy — in a study where roughly four out of five patients ended up on stabilizer therapy, including the control arm.

So the placebo arm stopped being a placebo arm mid-trial.

It did. And this is the piece I'd underline for anyone running late-stage cardiology: the standard of care moved underneath the study while the study was running. That twenty-four percent who initiated mid-trial aren't a protocol footnote — they are a treated control group the design never planned for. Put a cumulative recurrent-events composite on top of that, in a population old enough that competing risk from non-cardiovascular death eats into the analysis, and you can see why the investigators also ran a version folding in death from any cause. The endpoint was carrying a lot of weight under conditions nobody anticipated.

AstraZeneca's read is that the medicine performed. Mina Makar, who runs their cardiovascular, renal and metabolism group, told Fierce the study performed very well — that the drug knocks down ATTR exactly as expected, and that its efficacy as a monotherapy is strong on quality of life, on six-minute walk, on cardiovascular events. Their position is: this wasn't a molecule failure. It was an add-on question the trial couldn't answer.

I'm going to push, because you took that frame too cleanly. "Strong as a monotherapy" describes a subgroup nobody randomized. The patients who stayed off stabilizers weren't assigned to stay off them — their physicians made a different call, and those physicians had reasons. Take a post hoc slice of a failed trial and call it the drug's real performance, and you've built a marketing claim, not evidence.

I'll concede that. A non-randomized subgroup is not evidence the way an intent-to-treat result is evidence — that's a fair hit. But I'd hold the wider point: nothing in this dataset says the biology is wrong. The knockdown happened. What failed was the additive step.

Then the honest version is narrower than where either of us started. This isn't a biology failure and it isn't a quiet win. It's a sequencing failure. The window to test a silencer on its own in this disease closed while the trial was still enrolling — by the time the answer arrived, the question had changed underneath it.

And that's what makes this expensive for the whole class, not just for AstraZeneca and Ionis. Two weeks ago on this show we said Munich would decide whether stabilizer-first was a quarter or a structure. Munich answered — from a direction nobody was watching. Every silencer program in ATTR cardiomyopathy now has to explain what it adds to a patient already on background therapy. The trial built to produce that evidence didn't produce it.

The label-first read sharpens it further. Nobody writes a broad add-on indication off this dataset. So the commercial question stops being silencer versus stabilizer and becomes: what do you add, to whom, and when. And I want to be careful here, because there is randomized evidence pointed at that question — it just isn't AstraZeneca's. At the same congress, Alnylam presented a prespecified HELIOS-B subgroup analysis of vutrisiran, split by whether patients were on tafamidis at baseline — two hundred fifty-nine of six hundred fifty-four randomized and treated patients — published the same day in the Journal of the American College of Cardiology. Alnylam's read is that the benefit held with or without a stabilizer underneath it.

Which is the competitive read. If you're running commercial or competitive intelligence in ATTR, stop modelling a two-class race and start modelling combination sequencing — because whoever generates real add-on data in a stabilizer-treated population owns the conversation. And right now the only randomized read on that question belongs to the competitor, not to the trial that was built to produce it.

Hold the caveat alongside it, though. A prespecified subgroup of one trial is the strongest evidence anyone has on the add-on question, and it is still a subgroup of one trial. Nobody has run the study designed to answer it — and that's the thread that keeps me up. Can that study even be run now? Enrolling ATTR cardiomyopathy patients off stabilizer therapy in 2026 is a different ethical and recruitment proposition than it was when this study opened. AstraZeneca hasn't said what comes next for the program. Until somebody solves the design problem, the whole class is stuck arguing about subgroups.

Take the week whole and one thing decided every major story — not the molecule. The comparator.

Start with the cleanest version of that, because Lilly chose theirs deliberately. The FDA approved Mounjaro on Friday to lower the risk of major adverse cardiovascular events in adults with type 2 diabetes at high risk. Cardiovascular death, non-fatal heart attack, non-fatal stroke.

SURPASS-CVOT — more than thirteen thousand participants, thirty countries, more than four and a half years. Per Lilly's announcement, it's the first cardiovascular outcomes trial to compare two incretin medicines head-to-head instead of against placebo. Mounjaro versus Trulicity, a comparator that already carries proven cardiovascular benefit. Eight percent lower rate of MACE-3. Non-inferiority demonstrated. Superiority not established.

And that last clause is the whole strategic story. Lilly now has label parity with Novo on cardiovascular risk reduction — but the evidence underneath the two labels came from completely different worlds. Novo's came out of placebo. Lilly's came out of beating an active drug that works. Same sentence on the label. A very different conversation with a cardiologist.

Kenneth Custer said they set the higher bar deliberately. My read is that's not just rhetoric — it's a switching argument. You can't answer "why move this patient" with a placebo comparison.

If the label holds up in practice, yes — but I'd narrow what it buys them. It's a switching argument in the specialist conversation, not a population argument. And the population piece is the one I'd sit with: Lilly cites data that as many as one in three US adults with type 2 diabetes have undetected cardiovascular disease. This expansion matters less for the patients already flagged high risk than for the ones nobody has flagged yet — and nothing about a label finds those people.

Now flip the comparator problem completely the other way. BioNTech and Genentech terminated the Phase 2 of autogene cevumeran — their personalized mRNA cancer vaccine — in post-surgical stage two and three colorectal cancer. Numerical imbalance in overall survival. More deaths in the vaccine arm. The independent data safety monitoring board made the call.

In the adjuvant setting that's about as unforgiving as a signal gets.

Because the control arm does well.

The control arm does well by design — these are resected patients with no evident disease, so a monotherapy has nowhere to hide. And colorectal is the cold tumor the field has spent a decade failing to crack. Single-agent mRNA into that microenvironment was always the hardest version of this bet, and the DSMB doesn't recommend termination on an imbalance like that unless the direction was consistent.

The combination trial with immunotherapy continues, which tells you where the program's center of gravity just moved. And this lands a week after Moderna and Merck posted a positive Phase 3 in their own personalized vaccine program — so the category isn't the thing in question. The monotherapy strategy is.

Which is a bigger reframe than it sounds. If individualized neoantigen vaccines are combination assets rather than standalone ones, the commercial model, the partner economics and the manufacturing math all change at once — you're no longer building a product, you're building an add-on that has to justify itself against a backbone.

And then Novartis, walking straight toward a failure instead of away from it. Phase 2 TRANQUILITY data on pacibekitug, an IL-6 targeting antibody, in chronic kidney disease patients at high cardiovascular risk — sustained decreases in inflammatory biomarkers, dose-dependent response. They're proceeding to a cardiovascular outcomes trial anyway, despite Novo's ZEUS trial of ziltivekimab failing.

They paid one point four billion dollars for that antibody through Tourmaline Bio, so the incentive to keep going is not subtle.

Sure — but is the thesis wrong?

The thesis is exactly the argument we just spent ten minutes on with Wainua: they're betting the failure lived in the population and the trial, not in the target. That's defensible right up until it isn't. Biomarker suppression has never been the hard part of IL-6. Events are.

Which is the through-line for the entire week. Four programs, four comparator decisions — and in three of them what the drug was measured against did more to determine the headline than the drug did.

Fifty-seven and a half million dollars upfront for the dexmedetomidine sublingual film — acute agitation in schizophrenia and bipolar one or two, under FDA review for at-home use, with an action date of November fourteenth.

And the structure of the contingent piece is what actually tells you something. Up to sixty-seven and a half million more — but the full amount only pays if the product is approved by that November date. Slip to February twenty-seventh and it drops to fifty-five million. Later than that and it converts to sales-based milestones.

So Teva priced the calendar, not the asset.

They priced regulatory delay explicitly, which is more discipline than you usually see on a distressed asset with a near-term decision attached. And it's still an auction — court-supervised, higher bids possible, Teva collects a break fee if it loses.

The money isn't what makes it worth watching, though. An at-home acute agitation option sits in a gap that currently routes through emergency departments — that's the clinical argument, and it's why the asset had value even inside a bankruptcy. If that decision lands on time, the program a distressed seller couldn't finance becomes a launch inside a company that can, and this looks like the cheapest CNS deal anyone did this year.

The week ahead is defined by what hasn't landed rather than what's dated. Novartis has said an outcomes trial for pacibekitug is planned — not started. Watch the population they choose, because that choice is their entire answer to ZEUS.

And one loop we owe you. We flagged Novartis's pelacarsen cardiovascular outcomes readout as an event-driven second-half catalyst. It hasn't landed and there's no confirmed date — so it stays on the watchlist, not the calendar. I'd rather tell you that than pretend the clock ran out.

On the Teva bid, the sequence matters: bankruptcy court approval and the auction first, then November fourteenth. Two independent ways this deal changes shape before anyone books a dollar of revenue — and the second one is precisely what those contingent payments are written against.

The one I'll be reading closely is whatever Novartis says about how they define the pacibekitug population. After the week we just had, population choice is the whole ballgame. Have a good week, everyone. Let's go.

That wraps our Week in Review — a week where the comparator beat the molecule three times out of four. Wainua's failure in Munich. Mounjaro's cardiovascular label, earned against a drug that already works. BioNTech's vaccine pulled on a survival imbalance. And Teva buying a November decision for fifty-seven and a half million while nobody was looking. The ATTR question is the one to carry into the week: somebody has to run the trial nobody can design yet. Follow the show on Apple Podcasts, Spotify, or wherever you listen — a quick rating or share helps other listeners find us. You'll start every weekday with this.