Alumis TYK2 Pill Fails Lupus: What's Left of the Platform & Kazia's PI3K Inhibitor Cuts CRC Tumor Burden 52%

Alumis lost fifty-five percent of its value today. The TYK2 pill missed every endpoint it was measured against in lupus — and then the CEO got on the call and said there's a clear path forward.

Roche wrote another China check, Novartis finally broke the liver curse that's haunted BTK drugs in MS, and Kazia's paxalisib data landed in colorectal cancer. Busy Tuesday. Let's get into it.

Welcome to The Pharma Closeout for Tuesday, September 1st. I'm Alex Mercer.

And I'm Maya Patel.

So, Alumis. Fifty-five percent — that's the drop today, per BioPharma Dive, and it nearly erases every gain the company made this year. The drug is envudeucitinib. In a mid-stage systemic lupus study it missed all of its main and secondary objectives. And the reason this is a story and not a bulletin is what happened next: CEO Martin Babler told investors there's a clear path forward and no significant change to the development timeline.

The path exists. Whether it's clear is a different question.

Go ahead.

What they're pointing at is the high interferon gene signature subgroup — robust responses there, and Alumis says that's a majority of moderate-to-severe SLE patients. Fine. But read the justification carefully. Alumis said the effect in that subgroup looked similar to what Sotyktu showed. So the argument for spending more time is that a next-generation, more potent TYK2 blocker matched the first-generation molecule in the one population where TYK2 was always going to work. That isn't differentiation. That's the class ceiling, described politely.

I'm going to push back on the framing, because the setup matters here. Alumis's valuation reset in January on psoriasis data — that's the franchise. Lupus was the expansion story. It's the thing that turned a crowded psoriasis asset into a platform. So the question isn't whether the biology died today. It's whether an enriched Phase 3 in lupus is the best use of the next dollar they spend.

Credit where it's due. That interferon subgroup was prespecified, not fished out after the fact. But Alumis also said those patients turned out to be under-represented in the trial, which is their explanation for the overall miss. So the study wasn't built to answer the question they now want it to answer. A prespecified subgroup is a hypothesis with a plan. It is still not a result, and the agency will treat it exactly that way when Alumis sits down with them.

OK, I'll give you that. And here's where it lands: they're still filing in psoriasis by year-end, so the near-term catalyst is intact. What the market deleted today wasn't the company. It was the option on everything past psoriasis.

And the option got repriced for the whole field, not just Alumis. Takeda paid four billion for a TYK2 blocker out of Nimbus. Per BioPharma Dive, nobody in this class has yet shown these drugs do anything outside psoriasis and psoriatic arthritis — other TYK2 inhibitors have already struggled in lupus and in IBD. Stifel's Alex Thompson called it a show-me story. That's the label the entire class is wearing tonight.

The competitive read: if you're running commercial strategy on an oral immunology asset priced on indication expansion, your discount rate changed this morning. Psoriasis is a real market. TYK2 as an autoimmune platform is now something you prove in a randomized trial, not argue in a deck.

And the thing I'd actually watch isn't the Phase 3. It's whether the FDA lets them use the interferon signature as an enrollment criterion at all. Because if the agency says run it broad, this program doesn't have a path. It has a wish.

Alumis is what happens when an immunology bet goes wrong in public. Roche made a quieter one today. It licensed a preclinical trispecific antibody from Simcere Zaiming — seventy-five million dollars upfront, up to one point five three billion in total payments, plus tiered royalties. The molecule is SIM0660: a CD3-engaging arm paired with binding domains against two B-cell antigens, CD79a and CD19.

Seventy-five up front on a preclinical asset tells you what Roche thinks it's buying. That's not conviction. That's rent on an option.

It's also the third Asia deal in about a week. Over a billion into DualityBio's ADC platform recently, up to two point three billion on Hanmi's early obesity asset before that.

The money's the least interesting part. The target pair is the point. Hitting CD79a and CD19 together aims squarely at patients who've already failed CD20- or CD19-directed therapy — and Simcere flags autoimmune as the second use case, which is where the whole B-cell depletion field is heading. That's the same thesis CAR-T is chasing, in a molecule you can actually manufacture at scale.

Which is the read for BD teams. Simcere says this is their sixth out-licensing deal, more than six point one billion in aggregate potential value across Roche, AbbVie, Ipsen and Boehringer. A Chinese platform company isn't an opportunistic source anymore — it's a repeat counterparty. If your sourcing strategy still treats China as a one-off, you're a cycle behind the people you're bidding against.

From a China deal to a Swiss one. On the regulatory side, Novartis. Two Phase 3 trials of remibrutinib, the BTK inhibitor it already sells as Rhapsido for chronic hives, hit their goals in relapsing multiple sclerosis, beating Sanofi's Aubagio on relapse rate across nearly two thousand patients. Nice result. Not the headline.

Then what is?

Zero cases meeting Hy's Law. The two BTK rivals ahead of it in MS both ran into the liver. The FDA rejected Sanofi's tolebrutinib in non-relapsing secondary progressive MS and cited six Hy's Law cases in the complete response letter. Roche's fenebrutinib sat under a partial hold and has since reported one. Novartis ran the same class through the same organ in two thousand people and came out clean. That's the regulatory conversation. Efficacy gets it a filing.

And the street moved on it — shares up over four percent Tuesday, with Jefferies' Michael Leuchten floating more than three billion in peak MS sales. Novartis has been open about needing exactly this kind of asset against what's coming off patent.

Hold that number loosely until October.

Why?

Because Novartis didn't release detailed results. Disability progression showed a positive trend at three months, and nominal significance at six in a preplanned combined analysis of both trials. Nominal significance is not a powered win. And disability progression is precisely the axis where this drug has to compete with Ocrevus and Kesimpta — relapse rate gets you approved, disability data gets you positioned ahead of a biologic. Full results come at MSToronto in October.

All of this the same week the company paused its autoimmune CAR-T program.

Which tells you how Novartis is hedged across immunology right now — one modality on hold, another clearing Phase 3. For anyone building an oral MS franchise, the liver question that stalled this class for years may have just been answered by a molecule Novartis already sells. What's still open is the label, and that gets written on the disability data, not the relapse data.

One more before the calendar. Kazia Therapeutics put out preclinical data this morning on paxalisib in colorectal cancer. In an MSS/pMMR model, monotherapy cut tumor burden by fifty-two percent. And in a separate experiment, adding paxalisib to pembrolizumab cut tumor volume by a further fifty percent versus the immunotherapy alone.

MSS/pMMR is roughly eighty-five to ninety percent of metastatic colorectal cancer, and it's the population where checkpoint inhibitors have delivered essentially nothing. So Kazia is asking the right question: can you reprogram the microenvironment to restore immune visibility? It's just being answered in mice.

They say that to the company's knowledge, no other PI3K/mTOR inhibitor has shown meaningful activity in pMMR colorectal. So the differentiation claim is at least narrow enough to test.

Then look at how they've designed the test. Five arms: two paxalisib doses, each with and without pembrolizumab, against standard of care. And the primary endpoint is safety and tolerability. Progression-free survival, response rate, overall survival all sit as secondaries. That's a company telling you honestly this is a dose-and-signal study, not a proof of efficacy.

With enrollment starting first quarter of 2027.

So the near-term catalyst isn't data — it's whether the biomarker program they're building can identify responders before anyone gets randomized. If it can, that trial gets interesting fast. If it can't, five arms in an unselected population is how elegant preclinical biology quietly disappears.

Quick note before the calendar — if The Pharma Closeout is how you close out your pharma day, follow the show on Spotify or Apple Podcasts. Now, looking ahead — September's FDA calendar has levacetylleucine in ataxia-telangiectasia on the 19th, zilganersen in Alexander disease on the 22nd, and brepocitinib in dermatomyositis due sometime this month. And one loop we owe you: Novartis's pelacarsen cardiovascular outcomes data, which we flagged for the first half of this year, still hasn't landed. No confirmed new date.

The month I'd actually circle is October, for the full Remodel data at MSToronto. Put Alumis's FDA meeting next to it, whenever that lands.

Which question?

Whether an oral autoimmune drug is a franchise or a single indication. Both turn on data nobody outside the sponsor has seen yet. That's my favorite kind of month. Have a good evening.

And that is your Pharma Closeout for Tuesday, September 1st — Alumis down fifty-five percent and pivoting anyway, Roche back in China for a trispecific, Novartis finally clearing the liver hurdle in MS, and Kazia's paxalisib data in colorectal. Follow the show on Apple Podcasts, Spotify, or wherever you listen — a quick rating or share helps other listeners find us. Tomorrow's briefing lands on its own.