EyePoint Crashes 67% After Duravyu Fails Phase 3 in Wet AMD

EyePoint spent years telling this industry it could break the injection cycle in wet AMD. The market believed it. This morning the stock crashed sixty-seven percent.

And AstraZeneca killed a Phase 3 bispecific in lung cancer on the same day it won two others, argenx moved a myositis subtype that has never had an approved therapy, and Cytokinetics took Bristol Myers to court. Busy Monday. Let's get into it.

Welcome to The Pharma Closeout for Monday, August 17th. I'm Alex Mercer.

And I'm Maya Patel.

Sixty-seven percent. That's the whole story in one number — EyePoint crashed sixty-seven percent Monday after Duravyu missed its primary endpoint in the LUGANO Phase 3 in wet AMD, head-to-head against aflibercept. And what makes today brutal rather than merely bad is what the company did next. They went to an ad hoc analysis to pull something usable out of the dataset. When you're presenting an analysis nobody prespecified, you've already told the market what you think of the prespecified one.

Say the comparator again.

Aflibercept. Standard of care. Patients.

Then the durability story was never what this trial was asking. That's the part I want people to sit with, because the durability actually delivered — patients went eight months or more without a supplemental anti-VEGF injection. Treatment burden dropped substantially. More than half were supplement-free at Week 56. Every one of those numbers is real. None of them survive an eye chart that doesn't hold.

Durability was the product. Vision was the gate. They built the whole company on the wrong side of that sentence.

I'd push on how you just framed that. The read circulating this morning is "the durability platform failed." It didn't. What failed is an assumption — that a retina specialist will trade any visual acuity shortfall for fewer office visits. That's a narrower lesson than "platform dead," and it's a far more expensive one for everyone building six-month implants and one-and-done gene therapy behind them.

I'll give you the platform read — that's too broad, you're right. But I'm holding the commercial conclusion, because there's a line in this dataset that does travel across the category. Nine patients in the Duravyu arm had vision loss. Zero in the aflibercept arm. The company characterized those nine as unrelated.

Nine versus zero.

Nine versus zero. Adjudicate it however you like — a retina specialist reads that column first.

Then the synthesis is neither of ours. The platform isn't dead, the durability isn't fake, but the risk tolerance in this indication just got re-priced for every sustained-delivery program in the field. Think about what the comparator is here. It's a drug that works, patients keep their vision, and the only cost is the visit schedule. An imbalance like that in the experimental arm — even adjudicated as unrelated — is the kind of thing that shapes an advisory committee discussion and then shapes a label, long before anyone gets to a payer conversation. The endpoint miss is one trial. The imbalance is a class question.

Which is the part that should be in someone's Monday deck. The competitive read: aflibercept just beat the most credible durability challenger in wet AMD in a head-to-head, and the anti-VEGF franchise is more durable than the street was modeling on Friday. If your model assumed injection burden collapses at the end of the decade, that assumption needs a new date on it — and so does everything you built downstream of it.

And watch which analysis EyePoint leads with next. There's a version where a signal that strong supports a narrow, well-selected population and a second pivotal gets financed. There's another version where the safety column closes that door before anyone argues statistics. The company's next disclosure tells you which conversation they think they can still win.

That's the failure story. Here's a company buying the other side of the decade. Sandoz struck a three hundred and twenty-two million dollar deal with Shanghai Henlius Biotechnology — three biosimilars in early development, milestone-loaded, with a hundred and a half million up front. Henlius develops, manufactures, and keeps China. Sandoz commercializes everywhere else. And the structural read is the interesting part: Sandoz just used a China in-licensing deal to add assets it wasn't going to build itself, taking its biosimilar pipeline from thirty-nine to a potential forty-six against the biologics loss-of-exclusivity wave. Western generics is now sourcing its next decade out of Chinese biotechnology.

The price isn't what I'd look at. The selection is. Repatha, Benlysta, Erbitux.

Which is a strange trio on paper.

It's a strange trio until you notice Erbitux lost patent protection a decade ago and still has no biosimilar on the market — the compound is complex enough that it deterred everyone who looked at it. So Sandoz isn't buying three copies of anything. They're buying one late-decade cardiovascular opportunity, one lupus asset, and one manufacturing bet the rest of the field walked away from. That third one is the tell about who they think Henlius is.

Staying on the deal side — Slate Medicines is combining with Fulcrum Therapeutics in a reverse merger, with two hundred and forty-five million dollars in private placement financing alongside it. Fulcrum discontinued its sickle cell program in June after regulators raised concerns about secondary blood cancer risk. Slate started up this year with a Series A and a PACAP-targeted migraine candidate licensed out of China. Fulcrum shareholders get five percent of the combined company plus a cash dividend. Runway into 2029.

Five percent. That's what a discontinued pipeline is worth to the people buying the listing.

It's a cash shell with a Nasdaq ticker, and Slate priced it exactly that way.

The clinical bet underneath it is more interesting than the structure. Going at PACAP rather than CGRP, specifically in patients who don't respond to approved treatments, is a wager that non-responders are mechanistically distinct — not just undertreated or underdosed. That is a real hypothesis, and it's testable. SLTE-1009 starts Phase 1 in Australia with topline expected mid-2027, and the bispecific moves toward Phase 2a behind it. If the hypothesis is wrong, this financing bought enough runway to find out cleanly. Most reverse mergers don't buy that.

On the regulatory and clinical side, the readout that earns the most attention today is argenx. ALKIVIA hit its primary endpoint — Vyvgart Hytrulo delivered a fifteen point four point greater improvement in mean Total Improvement Score at Week 52 versus placebo in the combined myositis population. Patients, p-value of point zero zero one one. But the number isn't the story. Separation from placebo emerged at Week 4 and held through a full year, and it held while patients were being tapered off steroids.

Explain why the taper changes what that score means.

Because an improvement score achieved while patients stay on chronic corticosteroids is a much softer claim than the same score achieved while you're pulling them off. That's my one explainer today. The subtype split is where this gets genuinely interesting. IMNM hit — argenx says this is the first Phase 3 to show statistically significant improvement in disease activity in a subtype with no approved therapy at all. Dermatomyositis showed a comparable magnitude of improvement and did not reach statistical significance in that smaller cohort.

So significance lands in the subtype with no competitors, and misses in the one with roughly twice the U.S. patient population.

Which is precisely the label conversation. Consistent effect size across both, significance in one.

And those are two completely different businesses.

They are. A combined myositis indication is a serious franchise expansion on top of the existing FcRn business. A restriction to IMNM is an elegant orphan launch into genuine unmet need — and nothing more than that. The trial result is settled. The commercial outcome isn't, and it won't be until the label language is written.

Also worth noting on the regulatory side — AstraZeneca had a day that cut both ways. Positive Phase 3 for Enhertu in HER2-mutant non-small cell lung cancer, improving progression-free survival versus standard of care. The Tagrisso-Orpathys combination showed both PFS and overall survival improvement in MET-amplified, post-Tagrisso EGFR-mutated disease. Then they terminated the Phase 3 of volrustomig, their PD-1/CTLA-4 bispecific, after interim monitors concluded it wasn't going to improve survival against Keytruda plus chemotherapy.

The termination is the more informative event of the three.

Because of where they ran it.

Because of exactly where they ran it. That trial enrolled PD-L1-low patients — the population where a second checkpoint should add the most. Failing there isn't a dosing problem you fix in the next protocol amendment. And In the first human study of this drug, one enrollee stopped because of side effects., which is why Leerink has been questioning whether there's a therapeutic window at all. AstraZeneca is proceeding with mesothelioma, cervical, and head and neck. Those readouts now carry the entire dual-checkpoint thesis for a company with an eighty billion dollar sales goal for 2030 that leans partly on this class.

And out of China — InnoCare cleared approval to start a head-to-head registrational Phase III. Mesutoclax plus azacitidine against venetoclax plus azacitidine, treatment-naive AML in elderly or unfit patients, overall survival as the primary endpoint.

Powering superiority on overall survival against venetoclax is not a hedge — that's a company saying it doesn't need a placebo arm to make its case. Chinese developers are increasingly skipping the comfortable design and going straight at the incumbent. If that reads out positive, the pressure lands on the standard of care, not on the challenger, and it lands in the largest AML treatment population there is.

Looking ahead — the catalysts worth calendaring all came out of today. Detailed ALKIVIA data goes to an upcoming medical meeting, and that's where the dermatomyositis cohort gets properly litigated. Slate's Phase 1 topline is expected mid-2027, with the merger targeted to close by year end.

I'd watch EyePoint's next communication ahead of any of those. A company that leads with an ad hoc analysis on the day of a miss is telling you what it believes its remaining leverage is. If the next disclosure centers on the safety adjudication rather than the efficacy, that's a management team handling a class problem, not a trial problem — and that difference decides whether anyone writes a check for the second pivotal.

The one I'll be sitting with tonight is IMNM. A subtype where patients have had nothing approved, and a Phase 3 that finally moved. That detailed readout is worth waiting for. Have a good evening.

And that is your Pharma Closeout for Monday, August 17th — EyePoint down sixty-seven percent, argenx moving a disease that has never had a therapy, AstraZeneca winning twice and losing once before lunch, and Sandoz quietly buying its next decade out of Shanghai. One session, four completely different bets on where this industry goes. Follow the show on Spotify, Apple Podcasts, or wherever you listen — tomorrow's briefing lands on its own.