Aiming to bolster ivonescimab’s case ahead of a key FDA decision, Summit Therapeutics has provided updated overall survival data from the phase 3 HARMONi trial for its PD-1xVEGF bispecific antibody.
According to the new analysis, performed with a data cut-off in June 2026, the overall survival (OS) hazard ratio in the Western subgroup of 165 patients improved to 0.76, compared to 0.84 at a September 2025 cut-off and 0.98 at the trial’s primary analysis in April 2025.
The latest OS readout indicates a 24% reduction in the risk of death for ivonescimab plus chemotherapy versus chemo alone in this population of previously treated EGFR-mutated nonsquamous non-small cell lung cancer. The results do not bear statistical significance.
The new analysis also brings ivonescimab’s OS hazard ratio to the same level as the entire trial group of 438 patients, Summit noted. Previously, the full study’s OS hazard ratio was 0.78 in September and 0.79 for the primary analysis, neither of which was statistically significant.
The results have been shared with the FDA, Summit said. The FDA is reviewing Summit’s application, which is the first in the U.S. for both ivonescimab and the PD-1xVEGF class, with a target decision date of Nov. 14, 2026. The updated analysis likely constitutes a major amendment, which may trigger a three-month extension of the agency’s review period.
“The HARMONi results we announced today are a new analysis that was just recently made available to the FDA,” a Summit spokesperson said in a statement to Fierce. “We think it’s important context for our submission.”
Summit’s stock price rose about 3.1% Wednesday morning as of publication time.
Summit’s FDA application was risky after HARMONi missed its OS endpoint, even as the agency explicitly requested a statistically significant OS to support a filing. However, a go-ahead is still possible, given that the FDA has in the past approved at least two agents in the previously-treated EGFR NSCLC setting without statistically significant OS: Johnson & Johnson’s Rybrevant in 2024 and AstraZeneca and Daiichi Sankyo’s Datroway last year.
Summit has also described HARMONi’s OS miss more as an unfortunate technical failure than a reflection of ivonescimab's usefulness.
HARMONi rolled over a subgroup of Chinese patients from Summit partner Akeso’s HARMONi-A trial who had previously failed on third-generation EGFR inhibitors. Because enrollment of the Western population went slower than expected, the study reached its predetermined primary analysis when only a few deaths were recorded in the Western subgroup.
During the primary analysis, the median follow-up of the Western subgroup was 9.2 months, shorter than the median OS length of 16.8 months for the entire trial group. For the September 2025 data cut, the median Western follow-up was 13.7 months, while the median OS remained unchanged.
For the current June 2026 cut, with a median Western follow-up of 23.2 months, “most western patients have now discontinued treatment or completed two years of treatment,” Summit said in its July 22 release.
The median OS numbers were not disclosed Wednesday. In a note, Summit said additional information from HARMONi, including median OS and confidence intervals for hazard ratios, is intended to be presented at an upcoming medical conference.
“The positive results from this latest overall survival analysis in the global HARMONi study continue to support the translation of the therapeutic profile of ivonescimab across the globe, including western patients from North America and Europe,” Maky Zanganeh, president and co-CEO of Summit, said in a July 22 statement.
Meanwhile, investors’ focus remains on the global HARMONi-3 readout for ivonescimab in first-line NSCLC. Summit shares suffered a bloodbath in May after the study surprisingly missed statistical significance on progression-free survival during an interim analysis. The company has said it expects final PFS and interim OS readouts this year.
In its Friday statement, Summit’s spokesperson said there’s no update on HARMONi-3 timing at this point.