Welcome back to This Week in Biotech by Biotech Blueprint, edition 119, covering biotech and pharma news from September 18th to 24th, 2026.
🎙️ Biotech Blueprint brings you weekly video updates on the latest biotech and pharma news, plus in-depth podcast interviews with industry leaders. You can find us on YouTube, Spotify, and Apple Podcasts. If you want to be featured on our podcast, please get in touch.
📩 For consulting inquiries, please email me.
👥 If you haven’t already, connect with me on LinkedIn.
🛒 And make sure to check out our MERCH STORE.
VIDEO SUMMARY
THIS WEEK’S KEY TAKEAWAYS 🔑
Several Phase 3 wins were reported this week, but details were missing. Roche and Ionis reported that sefaxersen cut protein in the urine of IgA nephropathy patients, with no percentage. Amgen said dazodalibep produced a “statistically significant and clinically meaningful” improvement in Sjögren’s disease, with no score. Merck said its eye drug beat Lucentis in diabetic macular edema, with no letters of vision. Otsuka and Ionis said ulefnersen improved function and survival in a rare inherited form of ALS, with no hazard ratio. Merck rose 1% on an announcement containing no numbers.
The one company that actually disclosed data, including two anaphylaxis cases (1 more than placebo), was Celldex. I don’t think anaphylaxis will be a big problem for this drug long term. Dupixent carries an anaphylaxis warning and it is one of the biggest drugs in the world. But the stock fell 13% anyway. The market is currently punishing disclosures and giving a free pass to keeping numbers to yourself. However, this won’t last long because all numbers that are being withheld will have to be disclosed at medical meetings within months, and stocks will get repriced. I am just pointing out this new trend among larger companies that somewhat feels disingenuous.
The second theme is that manufacturing, not biology, is a constraint on the most advanced medicines right now. Telix paid $1.65 billion upfront for ITM, whose lead drug was rejected by the FDA in August for manufacturing and facility problems and nothing else. Ultragenyx’s Sanfilippo gene therapy was approved at $3.95 million, a year after its own manufacturing rejection. And Fyodor Urnov revealed that the base editing drug that saved Baby KJ cost $1.65 million to make, roughly double of earlier estimates. On top of all that, Anthropic said 950 Claude agents found a new CRISPR-like enzyme family in 21 hours, and gene editing stocks fell up to 11.6%. Nobody has actually shown what this system does yet, and Anthropic says so themselves. So while I agree that discovery is getting cheaper by the month and 21 hours is very impressive, the market reaction seems a little premature. Manufacturing is where the real bottleneck is now, and that’s what needs more attention.
Third, the gap between Lilly and Novo keeps widening. This week, Lilly got a second once weekly basal insulin approved and signed its fourth China discovery alliance since January 2025, worth up to $3.35 billion with InnoCare. Novo held its capital markets day and told investors to expect growth ‘broadly in line with industry peers.’ For a company that spent five years growing faster than almost anyone in pharma, that’s a significant downgrade. Shares fell about 5%.
BIOTECH/PHARMA NEWS 🧬
🔹 The FDA approved Lilly’s Onswik (insulin efsitora alfa) on Sept. 24 for adults with type 2 diabetes, cutting basal injections from about 365 a year to 52. It’s the second weekly basal insulin in the US, six months after Novo’s Awiqli in March. The four trial QWINT program covered more than 3,400 adults. In QWINT-2, insulin-naive patients lowered A1C, the standard three month blood sugar measure, by 1.34% versus 1.26% on daily degludec, with clinically significant or severe hypoglycemia at 0.58 versus 0.45 events per patient year (rate ratio 1.30, 95% CI 0.94 to 1.78) and no severe episodes on efsitora. QWINT-3 showed a greater A1C drop than degludec at one year. It isn’t approved in type 1 diabetes, where severe hypoglycemia risk was higher. Price wasn’t disclosed. Separately, Lilly committed $100 million upfront and near term, plus up to $3.25B in milestones, to InnoCare (Beijing-based company) for compounds against up to five undisclosed targets. Lilly shares rose about 3%.
🔹 Telix agreed to buy ITM Isotope Technologies Munich for $1.65B upfront plus up to $700M in milestones. What Telix is buying is a rejected drug and a supply chain. ITM-11 (lutetium-177 edotreotide) won its Phase 3 COMPETE trial in gastroenteropancreatic neuroendocrine tumors, then received a complete response letter on August 7, citing only manufacturing and third party facility issues. Not safety, nor efficacy. A $100M milestone requires approval by Dec 31, 2027. The sharper point is the supply chain. ITM supplies the lutetium-177 used in Pluvicto, so Novartis’s isotope vendor is now owned by a competitor aiming directly at Lutathera, two weeks after the first generic Lutathera equivalent was approved. A day later Novartis licensed a preclinical radioligand from China’s BoomRay for up to $900M.
🔹 The FDA’s Molecular and Clinical Genetics Panel voted 7-2 with one abstention on Sept 23 that the benefits of GRAIL’s Galleri multi cancer blood test outweigh its risks for adults 50+. On safety the vote was 10-0, and on effectiveness it was a much closer 6-4. The 70k person randomized NHS-Galleri trial missed its primary endpoint, a reduction in stage 3 and 4 cancers (incidence rate ratio 1.03, p=0.63), while showing 14% fewer stage 4 diagnoses. In the 25,490 person PATHFINDER 2 study, the test flagged 0.54% of participants, with a positive predictive value of 60.3%, meaning about 4 in 10 positives led to a workup that found no cancer. Twelve month sensitivity was 39.3% across all cancers and 69.8% for twelve prespecified deadly ones, with specificity of 99.64%. The panel essentially endorsed a test that finds more cancers without proving it saves lives. Those aren’t the same thing, and the distinction is the central argument in cancer screening. GRAIL shares rose 12.3% to a 52 wk high. The premarket approval decision comes “in the coming months,” and coverage decisions will be argued over the stage 3-4 miss.
🔹 The FDA granted full approval on Sept. 17 to Ultragenyx’s Fayuvi (rebisufligene etisparvovec), the first treatment for Sanfilippo syndrome type A, a fatal childhood neurodegenerative disease with an average lifespan of about 15 years. The one time AAV9 gene therapy delivers a working SGSH gene. Approval rested on the Transpher A study against an external natural history control, and came after a July 2025 rejection over manufacturing. The list price is $3.95M,, justified against more than $8M in lifetime care costs. Jefferies models only about $250M in peak sales, and the second priority review voucher Ultragenyx received, worth about $200M at market rates, is nearly as valuable as the drug’s forecast annual revenue. The stock rose 13%. Six days later, Fyodor Urnov said the personalized base editor made for Baby KJ cost $1.65 million to manufacture, roughly double earlier estimates. For this to scale into N=1 platforms, that number has to fall by an order of magnitude.
🔹 Anthropic said 950 Claude agents, running for 21 h and processing about 210M tokens, narrowed more than 200k reverse transcriptases in public DNA databases to 3,500 unusual candidates and then to 20, identifying a family it calls array-associated reverse transcriptases. Wet lab work in its biosafety level 1-2 lab confirmed the associated repeat arrays produce short RNAs, which is what a guided DNA-cutting or editing system would need. The preprint, posted September 23, isn’t peer reviewed, and Anthropic says the system’s function is unproven and hasn’t been linked to programmable editing. Feng Zhang called it “genuinely intriguing.” The market treated it as a threat anyway, and CRISPR Therapeutics fell 5.5%, Beam 6.0% and Prime Medicine 11.6%. Gene-editing companies are valued on delivery, manufacturing and clinical data, not on owning a particular enzyme. What the result does show is that AI can now scan enormous genetic databases fast and cheap, which makes enzyme discovery less of a competitive advantage.
🔹 Celldex’s EMBARQ CSU1 and CSU2 trials randomized 1,939 patients with chronic spontaneous urticaria, meaning recurring hives and itch with no identifiable trigger, to barzolvolimab at 150 mg every four weeks, 300 mg every eight weeks, or placebo. The drug is an anti-KIT antibody that depletes mast cells, the immune cells that release histamine. It hit the primary endpoint on weekly urticaria activity score at week 12 in both trials, plus complete response at weeks 12 and 24. Complete response, meaning no hives and no itch, reached about 50% versus 15-18% on placebo at 24 weeks. In patients who had failed Xolair, the current standard, the 150 mg arms reached 55.3% and 41.7% complete response at week 12 versus 9.3% and 15.1%. That’s a clean signal in the population with the fewest options. The problem is two life threatening anaphylaxis cases on drug, with one more on placebo, plus neutropenia and hair and skin color changes. The stock fell 13% on the day. Cantor models $4.8B peak sales and filing is planned for 2027. Dupixent carries an anaphylaxis warning and sells billions, so this doesn’t automatically sink the drug, but it probably means a first dose monitoring requirement on the label.
🔹 Roche and Ionis’s Imagination trial hit its primary endpoint at a prespecified week 37 interim analysis in IgA nephropathy, a kidney disease where immune deposits damage the filtering units. Monthly self injected sefaxersen, an antisense drug that blocks production of complement factor B, reduced 24 hr urine protein-to-creatinine ratio versus placebo. Roche disclosed no magnitude. Safety matched earlier trials. The study stays blinded to week 105 for kidney function, and Roche will seek accelerated approval on proteinuria. The mechanism is the same as Novartis’s Fabhalta, which is a twice daily pill that can raise cholesterol, while sefaxersen showed no cholesterol effect in Phase 2. The field is crowded. Fabhalta, Otsuka’s Voyxact and Vera’s Trutakna are already approved, and Vertex’s povetacicept is pending. Ionis is due up to $400M in milestones. Roche says €1-2B peak, while analyst models sat near €400M by 2030 before the readout. Without the proteinuria percentage there’s no way to rank it against Fabhalta.
🔹 Viking Therapeutics’ 180 patient Phase 2 obesity study ran a 21 week weekly induction at 15 to 22.5 mg, producing 16-19% weight loss, then switched patients to less frequent shots for 12 weeks. Patients kept 82-97% of lost weight depending on strategy, about 90% on every other week dosing and 85% on monthly dosing. Only five patients discontinued for side effects across all arms, and GI effects didn’t differ meaningfully during maintenance. Leerink called it “stellar,” and the stock rose as much as 36%. There are two caveats though. Twelve weeks is a short maintenance window, and there was no comparison against simply staying on the weekly drug. Viking then did the rational thing and sold stock into the rally, pricing 7.86 million shares at $35+ $225M of 2% convertible notes due 2032, upsized to about $500M. Shares fell 13.4% on the pricing. Still early and still small, but it keeps alive the idea that monthly maintenance dosing is where independent obesity players can compete with Lilly and Novo.
Have a great rest of your week and thanks for reading Biotech Blueprint!
DISCLAIMER: This publication is for informational and educational purposes only and does not constitute investment, legal, medical, or tax advice, a solicitation, or an offer to buy/sell any security. Information is believed reliable but no warranty is made as to accuracy or completeness; views may change without notice. Do your own research and consult qualified professionals.
DISCLOSURE & CONFLICTS: The author may hold positions in securities mentioned and may change positions at any time without notice. No compensation is received from companies mentioned, and there are no known material business relationships unless explicitly stated. Content may reference clinical data and regulatory events; always consult primary sources and a licensed clinician for medical decisions. Past performance is not indicative of future results.
LIABILITY: Use of this content is at your own risk. The author assumes no responsibility for any losses arising from reliance on this material.
Front cover image source: Matt Chinworth cen.acs.org