Ivonescimab Beat Keytruda on Overall Survival in China as Ultragenyx’s Angelman Phase 3 Failed and GSK Bet Up to $1.3B on Hutchmed’s KRAS Cancer Drug (September 4, 2026)

Last updated: September 4, 2026

Akeso’s ivonescimab showed a statistically significant overall survival benefit over pembrolizumab monotherapy in the Chinese HARMONi-2 trial, with numbers withheld for a conference. Ultragenyx’s Angelman Phase 3 failed outright, and GSK licensed Hutchmed’s preclinical KRAS drug for $110 million upfront.

This page covers the ivonescimab overall survival result and what it does and does not change before November 14, the Ultragenyx Aspire failure, the GSK and Hutchmed deal terms, AbbVie’s fully disclosed CERVINO win, the FDA’s released Replimune review documents, a heavy medtech day, and the full log of how this week’s delivered catalysts traded.

What did the HARMONi-2 overall survival analysis show?

Ivonescimab monotherapy produced a statistically significant improvement in overall survival compared with pembrolizumab monotherapy at this analysis of HARMONi-2, per Summit Therapeutics’ September 2 release. No hazard ratio, medians, or p value were disclosed; the data are being held for a medical conference.

HARMONi-2 (AK112-303) is a randomized, double blind Phase 3 trial conducted in China by Akeso, testing ivonescimab, a PD-1 and VEGF bispecific antibody, head to head against pembrolizumab (Keytruda) as monotherapy in first line, PD-L1 positive, locally advanced or metastatic non small cell lung cancer. The trial made headlines in 2024 when ivonescimab roughly halved the risk of progression, with a progression free survival hazard ratio of 0.51 (95% CI 0.38 to 0.69, p<0.0001). Overall survival was the missing piece: a 2025 interim analysis requested by Chinese health authorities showed only a positive trend. This analysis, per the company, crosses into statistical significance.

The caveats matter. This is a single region trial run in China by Akeso, the comparator is pembrolizumab alone rather than a chemotherapy combination, and the release prints no effect sizes. Summit shares closed at $17.13 on Thursday, up 17.3 percent from Wednesday’s $14.60, in the first full session after the release.

Does the HARMONi-2 win change Summit’s November 14 FDA decision?

Not directly. Summit’s US application rests on a different trial in a different population: HARMONi, in previously treated EGFR mutated non small cell lung cancer, with a PDUFA goal date of November 14, 2026. HARMONi-2 supports the mechanism but is not the reviewed dataset.

The HARMONi package carries its own history. The primary overall survival analysis in April 2025 showed a hazard ratio of 0.79 that missed statistical significance (p=0.057), a September 2025 cut reached nominal significance (HR 0.78, nominal p=0.0332), and a June 2026 update put the hazard ratio at 0.76 in both the full population and the Western subgroup, without statistical significance and with medians withheld. Fierce Pharma reported in July that the updated analysis likely constitutes a major amendment that could push the decision three months past November 14. That reporting has not been confirmed by the FDA or the company.

Why did Ultragenyx’s Angelman syndrome Phase 3 fail?

Apazunersen (GTX-102) showed no difference from placebo. The Phase 3 Aspire study missed its primary endpoint, change from baseline in the Bayley-4 cognitive raw score, and its key secondary endpoint, net response on the Multidomain Responder Index, per Ultragenyx’s September 2 release.

The release, issued after Wednesday’s close, prints no effect sizes and states there were no differences between treated and control groups on either measure. Safety was consistent with Phase 1/2. Ultragenyx said it will evaluate the program and decide its disposition, and separately that it will assess planned operations to define and implement significant expense reductions. The company estimates Angelman syndrome affects approximately 60,000 people in commercially accessible geographies.

The timing is unforgiving. Ultragenyx’s UX111 gene therapy for Sanfilippo syndrome type A has an FDA action date of September 19, and the company now goes into that decision announcing cost cuts. Shares lost 44.0 percent Thursday, closing at $14.85 against Wednesday’s $26.53.

What did GSK pay for Hutchmed’s HMPL-A830?

GSK is paying $110 million upfront for worldwide rights outside Greater China to HMPL-A830, a preclinical antibody conjugate aimed at KRAS driven cancers, with up to $1.185 billion in development, regulatory and commercial milestones, for a total of up to $1.295 billion plus tiered royalties.

HMPL-A830 is what Hutchmed calls an Antibody Targeted Therapy Conjugate: a selective KRAS small molecule inhibitor payload conjugated to an anti EGFR antibody, designed to deliver KRAS inhibition preferentially to tumors expressing EGFR. Target indications are colorectal, pancreatic and lung cancer. The asset has not yet entered the clinic; Hutchmed expects a first trial in the second half of 2026 and will lead the global Phase 1 program, while GSK takes development and commercialization beyond that, everywhere except Mainland China, Hong Kong, Macau and Taiwan.

The cash share stands out. At $110 million on a $1.295 billion total, GSK is paying 8.5 percent of the headline in guaranteed money for a preclinical asset. For context from this ledger’s recent entries, Roche paid Simcere $75 million on an up to $1.53 billion total for a B cell trispecific, a 4.9 percent cash share, and Alteogen’s platform deals have carried upfronts near 1 to 2 percent of headline. Hutchmed’s Nasdaq listed ADRs closed at $14.11, up 17.2 percent; GSK’s ADRs closed at $50.87, up 0.9 percent.

The same day showed the other end of the market. Pfizer licensed PF-08046031, a discontinued Phase 1 CD228 directed antibody drug conjugate inherited from Seagen, to Medicus Pharma for $12 million upfront and a $15 million first anniversary payment, with more than $1 billion in milestones and a $2 million Pfizer investment in Medicus, per Fierce Biotech’s account of the deal. A shelved Phase 1 asset moves for about one percent of its biobucks in cash; a hot preclinical mechanism moves for 8.5 percent.

What did AbbVie’s etentamig show in the CERVINO trial?

Etentamig, a BCMA by CD3 bispecific T cell engager, produced a 74.0 percent response rate versus 45.7 percent for standard therapy and cut progression risk 60 percent in triple class exposed relapsed or refractory multiple myeloma, per AbbVie’s September 3 topline release, which disclosed the numbers immediately.

In CERVINO, patients with at least two prior lines were randomized against investigator’s choice of carfilzomib plus dexamethasone, elotuzumab plus pomalidomide and dexamethasone, or selinexor plus bortezomib and dexamethasone. The objective response rate was 74.0 percent (95% CI 67.25 to 79.97) versus 45.7 percent (95% CI 38.59 to 52.91, p<0.0001). Progression free survival carried a hazard ratio of 0.40 (95% CI 0.29 to 0.54, p<0.0001), with medians to come. Twelve month overall survival was 87.9 percent versus 72.0 percent (HR 0.48, nominal p=0.0012). Cytokine release syndrome hit 28.3 percent of etentamig patients, predominantly grade 1 with no grade 3 or higher; grade 3/4 infections ran 27.7 percent versus 19.2 percent; discontinuations for adverse events favored etentamig at 3.6 percent versus 9.6 percent. Full results are scheduled for September 25, and AbbVie will discuss next steps with regulators. AbbVie closed down 0.6 percent at $260.21.

What do the FDA’s Replimune review documents reveal?

Per BioSpace’s September 3 reading of newly released review documents, the FDA’s primary review team concluded Tudriqev should be rejected a third time, and a Center for Biologics Evaluation and Research director overruled that conclusion in an August 6 memo citing the advisory committee’s 10 to 3 vote.

The review team found Replimune had failed to address key deficiencies from the two prior complete response letters, echoing the July position that the data package was insufficient to conclude substantial evidence of effectiveness. The director’s memo, as reported, cited the totality of evidence, acknowledged the remaining uncertainties in full, and leaned on verification through the ongoing IGNYTE-3 confirmatory trial and the regulatory flexibility of the accelerated approval pathway. BioSpace did not name the director, and neither will we. When Tudriqev was approved in August we wrote that hostile briefing documents create variance, not verdicts; the documents now show the verdict was decided one level above the people who wrote the briefing.

What happened in medtech on September 3?

A busy day for a vertical that had gone quiet: Medtronic’s Cornerstone Robotics partnership, two orthopedic robotics acquisitions, an FDA warning letter for Fresenius Medical Care, and two fresh cybersecurity disclosures, while Boston Scientific’s restoration target reached the end of its promised week unconfirmed.

Where does the Boston Scientific cyberattack stand?

The company’s last public update remains September 1 at 8:44 pm ET, which targeted partial restoration of shipping for some products “this week.” As of Thursday evening September 3, no newer update had been posted to the company’s incident page, so that target is unconfirmed at the week’s end.

The September 1 update also said the timeline for full restoration is not yet known. Boston Scientific closed down 2.9 percent at $46.95 Thursday; market data, no cause assigned. The stock has now given back ground in each session since the August 26 disclosure week began, and a quantified 8-K remains the thing to watch.

How did the week’s delivered catalysts trade?

The tally we promised Monday is final: of the thirteen dated, promised catalysts delivered from Friday August 28 through Thursday September 3, exactly one sponsor closed higher on delivery day. That is the entire empirical case for underwriting the label, not the headline.

Alongside that ledger, the withholding pattern sharpened. The four positive closes logged across the no numbers series are JAZZ up 2.5 percent, Novartis up 6.0 percent, Teva up 3.6 percent, and now Summit up 17.3 percent, every one on a release that declared direction and withheld magnitude. The two companies that printed a miss this week, Alumis and Ultragenyx, lost 56.6 percent and 44.0 percent. AbbVie printed a clean win with full numbers and closed slightly lower.

How did Thursday’s biotech and pharma stocks close?

Exchange closes for September 3, with Wednesday’s close for comparison. Market data only; no cause assigned except where a company disclosure landed the same session.

Amylyx logged a fifth consecutive close below the $35.50 offering price, with the $500.2 million offering’s closing still unannounced as of Thursday evening; the greenshoe window forces disclosure by roughly September 18. Novo Nordisk’s ADRs added 1.6 percent to $47.51.

Frequently asked questions

What is ivonescimab and how does it work?

Ivonescimab is a bispecific antibody developed by Akeso and licensed to Summit Therapeutics for major Western markets. It blocks PD-1, the checkpoint Keytruda targets, and VEGF, the angiogenesis driver Avastin targets, in a single molecule.

Did ivonescimab really beat Keytruda?

In HARMONi-2, run in China, ivonescimab monotherapy beat pembrolizumab monotherapy with statistical significance on progression free survival in 2024 and, per the September 2 release, on overall survival at this analysis. The OS effect size has not been disclosed.

Why does it matter that HARMONi-2 was conducted only in China?

The FDA has historically wanted multiregional evidence before acting on single country trials. Summit’s US filing therefore rests on the global HARMONi study in a different population, which is why the China OS win does not automatically translate into US approval odds.

When will the actual HARMONi-2 survival numbers be released?

Summit and Akeso say the analysis will be presented at an upcoming medical conference, without naming which. Under the withholding pattern we have tracked since August, the congress presentation becomes its own catalyst date, because magnitude reprices even when direction is known.

What is Summit’s November 14 PDUFA date for?

It is the FDA goal date for ivonescimab plus chemotherapy in EGFR mutated non small cell lung cancer after progression on a third generation EGFR TKI, based on the HARMONi trial. July reporting flagged that an updated survival analysis may qualify as a major amendment and push the date three months.

What is apazunersen and why did the Aspire trial fail?

Apazunersen (GTX-102) is an antisense oligonucleotide designed to reactivate the silenced paternal UBE3A gene in Angelman syndrome. Aspire showed no difference from placebo on cognition or the multidomain responder index; Ultragenyx has not yet said why, and printed no effect sizes.

Does the Angelman failure affect Ultragenyx’s September 19 FDA decision?

UX111 for Sanfilippo syndrome type A is a separate program under separate review, so the science is unaffected. The balance sheet is shared: Ultragenyx announced significant expense reductions two weeks before its most important near term approval decision.

What is an Antibody Targeted Therapy Conjugate?

Hutchmed’s term for HMPL-A830’s design: instead of a chemotherapy payload, the anti EGFR antibody carries a selective KRAS inhibitor, aiming targeted inhibition at tumor cells that express EGFR. It is conceptually an antibody drug conjugate with a precision payload rather than a cytotoxic one.

How does GSK’s $110 million upfront compare with recent licensing deals?

At 8.5 percent of the up to $1.295 billion total, it is the richest cash share for a preclinical asset in our recent ledger, against 4.9 percent for Roche and Simcere’s trispecific and low single digits for platform access deals such as Alteogen’s.

What did AbbVie disclose about etentamig safety?

Cytokine release syndrome in 28.3 percent of patients, predominantly grade 1 with no grade 3 or higher events reported at topline, grade 3/4 infections in 27.7 percent versus 19.2 percent for standard therapy, and fewer discontinuations for adverse events than the control arm.

Who approved Replimune’s Tudriqev if the review team said no?

Per BioSpace’s reading of the released documents, a CBER director overruled the primary review team in an August 6 memo, citing the 10 to 3 advisory committee vote, the totality of evidence, and confirmation to come from the IGNYTE-3 trial. The outlet did not name the official.

Is Boston Scientific still shipping products?

Partially. The company targeted partial restoration of shipping for some products during the week of September 1 but has posted no public update since September 1 at 8:44 pm ET, and says the timeline for full restoration is not yet known.

What does the 1 of 13 delivered catalyst tally mean for investors?

Across thirteen dated catalysts delivered on schedule from August 28 through September 3, one sponsor closed higher on the day. Before a dated event, write down the narrowest outcome that still counts as a win and ask whether the stock is priced for that or for the headline.

What are the next dates to watch?

Lilly’s Sangamo platform purchase is expected to close September 4. Generate Biomedicines presents GB-0895 Phase 1 data at ERS September 7. Ultragenyx’s UX111 action date is September 19, Grail’s Galleri advisory panel is September 23, and AbbVie presents full CERVINO data September 25.

Sources

Primary sources: Summit Therapeutics release on the HARMONi-2 overall survival analysis (September 2, 2026); Summit Therapeutics release on the updated HARMONi analysis (July 22, 2026); Ultragenyx release on Phase 3 Aspire results (GlobeNewswire, September 2, 2026); Hutchmed release on the GSK licensing agreement (September 3, 2026); AbbVie release on Phase 3 CERVINO topline results (September 3, 2026); Amgen release on VESALIUS-CV mortality data (August 31, 2026); Medtronic release on the Cornerstone Robotics partnership (September 1, 2026); Boston Scientific cybersecurity incident page (updates through September 1, 2026, 8:44 pm ET, checked the evening of September 3).

Trade press and market data: BioSpace on the Replimune review documents (September 3, 2026); Endpoints News (corroborating headline); Fierce Pharma on the ivonescimab update and Fresenius warning letter (September 3, 2026); Fierce Biotech on the GSK and Hutchmed deal and the Pfizer and Medicus agreement (September 3, 2026); MedTech Dive on Medtronic, Stryker, Enovis, NovoCure and DaVita (September 3, 2026). Stock closes are exchange closes for September 3, 2026.

Related coverage

• TScan cut 75% of its staff to go in vivo as Novartis signed an up to $3.2B Alteogen deal and uniQure filed the first Huntington’s gene therapy BLA
• Novartis and BMS paused autoimmune CAR T trials after patient deaths as Roche paid $75M upfront for Simcere’s B cell trispecific
• Five delivered catalysts closed lower in one day as the FDA approved Mimrylo and Lisraya
• The week’s Binary Ledger tally: Amylyx priced an upsized $500.2M offering as the FDA approved Ultragenyx and Regeneron rare disease drugs
• Capricor’s deramiocel gets a November 22 PDUFA date as Roche bets up to $2.3B on Hanmi’s muscle preserving obesity drug