Editor’s Briefing: September 17, 2026

In Short. Novartis stopped its TREM2 ALS candidate after a phase 2 miss, the fourth clinical stumble in a month, and paid Sironax $125 million for a blood-brain-barrier delivery platform; Longeveron’s pediatric stem-cell phase 2b missed its primary and the company is cutting costs while reviewing “all options”; Roche put $36.5 million upfront (up to $1 billion) into Dualitas’ bispecific immune screen; Vera’s Trutakna Origin 3 finals showed kidney-function stabilization and a 76% cut in composite progression risk as IgAN rivalry with Otsuka’s Voyxact intensifies; NICE finally cleared Enhertu for NHS use in HER2-low advanced breast cancer; Ori locked a potential $120 million automated cell-therapy manufacturing pact while Cellares tested virus-free CAR-T automation; and Novo tapped Anthropic’s Claude Science as capital and franchise fights (Legend’s new CEO, BMS–Ono vs Amgen on Opdivo biosimilars) filled the midweek tape.

---

Novartis: ALS stop, BBB buy

Novartis discontinued VHB937 (lifonebart), a TREM2 stabilizer/activator, in amyotrophic lateral sclerosis after the phase 2 ASTRALS trial in 251 people with early-stage disease missed primary and secondary endpoints. The primary compared the drug with placebo on a composite of survival without ventilator need and ALSFRS-R change; secondaries included respiratory decline and neurofilament light. A phase 2 Alzheimer’s study of the same molecule continues. BioSpace counted the ALS halt as Novartis’ fourth clinical blow in under a month, after pelacarsen’s phase 3 cardiovascular miss, del-desiran’s phase 3 myotonic dystrophy type 1 miss, and a pause of multiple CAR-T studies after three patient deaths. On the same midweek tape, Novartis exercised its option on Sironax’s large-molecule brain-delivery platform for $125 million at close (the 2025 option had sketched up to $175 million in upfront and near-term payments). Sironax keeps rights to use the platform on certain assets and says proceeds will advance three early clinical programs. Novartis neuroscience leads have argued publicly that crossing the blood-brain barrier will need a toolbox, not a single fix. (Fierce Biotech; BioSpace; Fierce Biotech; BioPharma Dive)

Longeveron: Elpis II miss, “all options”

Longeveron’s Elpis II phase 2b of mesenchymal stem-cell therapy laromestrocel in 40 infants with hypoplastic left heart syndrome (adjunct to stage 2 palliative surgery) missed its Month-12 primary on right-ventricular ejection fraction versus control. Exploratory signals (no treated deaths vs one control; 12 vs 19 major adverse cardiovascular events) did not clear statistical significance; the company reported no treatment-emergent adverse events attributed to the product. Shares fell about 60% after hours. Longeveron ended June with $10.1 million cash (runway into 4Q as of that guidance), said it will implement cash-conservation measures, hire a strategic advisor, and review “all options,” while still seeking funds to push laromestrocel in aging-related frailty after an earlier frailty phase 2b also missed its six-month primary. (Fierce Biotech)

Dualitas screen, Anthropic on Novo’s bench

Roche is paying Dualitas Therapeutics $36.5 million upfront, with Dualitas eligible for up to $1 billion in milestones plus royalties, to screen more than 300,000 potential bispecific antibody combinations for immune disease. Roche can advance a limited set; Dualitas may later develop what Roche leaves. Separately, Novo said it will collaborate with Anthropic to apply Claude Science to R&D workflows and to strengthen software engineering, without disclosing financial terms, framed under CEO Mike Doustdar’s push to become “the world’s most AI-driven healthcare company” and with stated data-governance and human-oversight guardrails. The pact sits beside Novo’s earlier OpenAI operational work and a crowded Big Pharma AI deal lane (Nvidia, Google Cloud, and others). (BioPharma Dive; BioPharma Dive)

IgAN rivalry and Enhertu’s NHS door

Vera’s Origin 3 final efficacy analysis in adults with primary IgA nephropathy at risk of progression showed an annualized eGFR decline of 0.6 mL/min/1.73m² over two years on Trutakna (atacicept), inside the KDIGO 2025 physiologic-loss band of about 1 mL/min/1.73m² per year, and set up a bid to convert July’s accelerated approval to traditional approval. Cross-trial, Voyxact’s Visionary showed a 0.3 mL/min/1.73m² annualized gain; Trutakna’s placebo-adjusted eGFR effect (~5 mL/min/1.73m²) was numerically similar to Voyxact’s (~4.5). The sharper differentiator Fierce flagged: a 76% reduction in composite kidney-disease progression risk, with no Trutakna patient on long dialysis, transplant, or death through two years versus eight on placebo. Trutakna blocks APRIL and BAFF; Voyxact hits APRIL; Vertex’s povetacicept awaits an IgAN decision by November 30. Vera also cited more than 350 patient start forms in the first ten weeks post-accelerated approval. In parallel, NICE recommended routine NHS England use of AstraZeneca/Daiichi Sankyo’s Enhertu in HER2-low advanced or metastatic breast cancer after chemotherapy progression, reversing a 2024 cost-effectiveness rejection; NHS England said access starts Thursday. Destiny-Breast04 underpinned the guidance (PFS 9.9 vs 5.1 months; OS 23.4 vs 16.8 months vs physician’s-choice chemo). NICE linked the commercial solution partly to higher U.K. cost-effectiveness ceilings under the recent U.S.–U.K. pharma trade arrangement. (Fierce Pharma; Fierce Pharma)

Cell-therapy plants and quality alarms

Oribiotech signed a ten-year pact with an unnamed commercial-stage biopharma to put its Iro closed automated cell-therapy platform into manufacturing, potentially worth $120 million in services, milestones, equipment, and consumables. Ori says Iro holds an FDA Advanced Manufacturing Technology designation, has 23 partners, and has dosed a first GMP patient; the microwave-oven-scale units stack for capacity. The deal lands as rival Cellares, still absorbing ~168 job cuts after losing a large pharma customer (reported as BMS), partners with Taiwan’s GenomeFrontier to evaluate virus-free CAR-T (GF-CART01) production on Cellares’ Cell Shuttle via electroporation transfection rather than viral vectors. On the quality side, BioProcess International highlighted Northwestern’s Reese Richardson report of nearly 19,000 doctored commercial-antibody assay images implicating more than 15 vendors (Thermo Fisher among the largest exposures); of more than 50 manipulated-image products checked in open validation communities, 83% failed at least one listed application. BioSpace separately reported an FDA warning letter to Bausch & Lomb’s Tampa sterile site (phones in aseptic areas, inadequate environmental monitoring, persistent out-of-limit microbes), sent September 4 and posted September 15; the company said no halt or recall is required and pointed to ongoing remediation. (Fierce Pharma; Fierce Pharma; BioProcess International; BioSpace)

Franchise moves: Legend’s CEO, Opdivo’s copycats

Legend Biotech named Ingrid Zhang, a 15-year Novartis veteran and most recently chief commercial officer for Novartis’ ex-U.S. international business, as permanent CEO effective September 15, succeeding interim Alan Bash (back as Carvykti president after Ying Huang’s July exit). Carvykti did $657 million in 2Q 2026 sales; Legend posted a first adjusted quarterly profit of $63 million. Zhang’s package includes a $950,000 base, 90% bonus target, at least $10 million in restricted stock, and a $150,000 signing bonus. Downstream of PD-1 franchises, BMS and Ono sued Amgen in Delaware over the proposed Opdivo biosimilar ABP 206; Amgen says an FDA decision is expected by year-end and that it still aims to be in the first wave, while BMS has pointed investors to a U.S. copycat entry floor around 2028. (Fierce Pharma; Fierce Pharma)

Public-health sidebar

STAT’s Morning Rounds marked four U.S. measles-related deaths this year, more than in any year since 1992, against Pennsylvania’s outbreak of about 693 cases across 38 counties; the CDC had not yet registered measles as an underlying cause pending National Center for Health Statistics coding, even as HHS commentary disputed measles’ role in earlier deaths. (STAT News)

What to watch

pdufa.bio (as of September 17) and today’s regulatory packet still put Ultragenyx UX111 (Sanfilippo A) on September 19 and Merck Winrevair HYPERION label work on September 21, then Mirum/Incyte zilurgisertib (FOP) on September 26. Ameluz BCC and Scholar Rock Isembyld (apitegromab) now show as decided/approved on the calendar against their original late-September goal dates; Mim8 (denecimig) remains listed upcoming without a clean date in the anonymous API slice. October watches: Roche Tecentriq adjuvant MSI-H/stage III colon (October 9), Merck/Daiichi I-DXd in extensive-stage SCLC (October 10), Roche Enspryng in thyroid eye disease (October 15). Fierce’s midweek reports still treat Telix Pixclara as approved; the packet’s TLX Awaiting flag (days-to-decision negative) looks stale. Wednesday’s PDUFA VIII stakeholder meeting is now in the rear-view; Trial Blazer expedited-IND applications remain open through October 30. Connect’s COPD rademikibart readout stays a month-end clinical gate.

Open mechanism and franchise questions from this tape: whether TREM2 remains a workable ALS bet after ASTRALS, or mainly an Alzheimer’s microglial hypothesis; whether Sironax-class BBB shuttles change which large molecules Novartis can actually dose into CNS; whether Dualitas-scale bispecific combinatorial screens yield differentiated immune biology or mostly optionality; whether BAFF/APRIL dual blockade (Trutakna) beats APRIL-only on hard kidney outcomes in practice, not just cross-trial eGFR; whether automated, virus-free, or AMT-designated cell-therapy platforms compress COGS enough to matter before capacity and comparability do; and whether Opdivo’s first-wave biosimilars arrive on Amgen’s year-end FDA clock or on BMS’s 2028 patent floor.