top of page

FDA Told Summit It Needs Overall Survival. The Trial Missed by 0.057. November 14 Will Show How Firm That Line Is.

Writer: Lexim AI
Lexim AI
Nov 4
6 min read

On Saturday, November 14, FDA's goal date arrives for Summit Therapeutics' biologics license application for ivonescimab, a PD-1 and VEGF bispecific antibody, combined with chemotherapy in EGFR-mutated non-small cell lung cancer after progression on a tyrosine kinase inhibitor. If approved, it would be Summit's first marketed product and, according to the company's materials, ivonescimab's first Western regulatory authorization.

The drug's biology is not the interesting regulatory question. The interesting question is a sentence in Summit's own SEC filings: FDA told the company that a statistically significant overall survival benefit is necessary to support marketing authorization in this setting, and that the progression-free survival results alone may not be enough. The pivotal trial's overall survival result missed statistical significance by a narrow margin. FDA accepted the application for full review anyway. How the agency resolves that tension will be useful to every sponsor who has ever held a stated requirement and a near-miss at the same time.


How HARMONi was built

HARMONi (NCT06396065) was a randomized, double-blind, placebo-controlled phase 3 trial at 114 sites in Asia, Europe, and North America. It randomized 438 patients 1:1 (219 per arm) with locally advanced or metastatic EGFR-mutated nonsquamous NSCLC whose disease had progressed after a third-generation EGFR tyrosine kinase inhibitor such as osimertinib. Both arms received pemetrexed and carboplatin every three weeks for four cycles, followed by pemetrexed maintenance, with the experimental arm adding ivonescimab 20 mg/kg and the control arm adding placebo. About a quarter of patients in each arm had brain metastases at baseline.

Two design features matter for the regulatory question. First, PFS assessed by an independent review committee and OS were dual primary endpoints, so both carried formal weight from the start. Second, enrollment began in China and was later expanded to Europe and North America, which produced a trial with 273 Asian and 165 Western patients, roughly 38% Western, in which follow-up matured at very different speeds in the two groups.


What the HARMONi data show

HARMONi had two primary endpoints, progression-free survival (PFS) and overall survival (OS). The PFS endpoint was met: median PFS was 6.8 months with ivonescimab plus chemotherapy versus 4.4 months with placebo plus chemotherapy (hazard ratio 0.52). The OS endpoint was not. Median OS was 16.8 months versus 14.0 months, with a hazard ratio of 0.79 (95% confidence interval 0.62 to 1.01; p=0.057). The confidence interval just touches 1.

Safety was described by the company as acceptable and manageable. Grade 3 or higher treatment-related adverse events occurred in 50.0% of ivonescimab patients versus 42.2% on chemotherapy alone. Discontinuations due to treatment-related adverse events were 7.3% versus 5.0%, and 0.9% of ivonescimab patients had grade 3 or higher treatment-related hemorrhagic events.

Since the primary analysis, Summit has reported updated OS data that it says continue to trend favorably, including in Western patients, and says it shared that analysis with FDA. The primary analysis was published in The Lancet Oncology in August 2026. Separately, Akeso's HARMONi-2 trial, run in China against pembrolizumab monotherapy, reported a median OS of 30.8 versus 22.6 months (hazard ratio 0.73; p=0.009). That is a different setting and population, and the current application rests on HARMONi.


The regional question

The Western subgroup is where follow-up timing complicates the picture. At the primary analysis, the OS hazard ratio in Western patients was about 0.98, but their median follow-up was only 9.2 months, shorter than the median survival in either arm, while Asian patients had a median follow-up of 32.7 months. A hazard ratio estimated on immature data in one region and mature data in another can look like a regional difference when part of the gap is a difference in maturity.

The June 30, 2026 data cutoff addresses that directly. With Western median follow-up extended to 23.2 months, the Western hazard ratio improved to 0.76, matching the 0.76 reported for the full 438-patient population at that cutoff, while the Asian data remained those from an April 12, 2025 cutoff. Summit made this analysis available to FDA. Two cautions apply. The primary analysis is the pre-specified one, and the updated analysis, however reassuring, was not the test the trial was designed around. And a regulator reviewing a multiregional trial will look for the effect to hold in the population relevant to U.S. practice, which is why the maturity of the Western data matters for this review.


What FDA has said, and what it has done

Two things from Summit's filings sit side by side. First, FDA's stated position that OS must be statistically significant here. Second, FDA's acceptance of the application in January 2026, with a plan to perform a complete review, including mid-cycle and wrap-up meetings, and, absent major deficiencies, to review proposed labeling before the goal date. Accepting a filing is not a signal about the outcome, and the company's own view is that the data support a favorable benefit-risk profile despite the OS result. But the sequence shows FDA is reviewing the application on its merits rather than declining it on the endpoint alone.


Two readings of the same number

A second regulatory idea bears on how to read p=0.057. In August 2025, FDA issued draft guidance titled Approaches to Assessment of Overall Survival in Oncology Clinical Trials (docket FDA-2024-D-5850). It recommends that randomized oncology trials be designed to assess OS so that potential harm can be evaluated, with an emphasis on OS as a pre-specified safety endpoint when it is not the primary efficacy endpoint, and it addresses design issues such as limiting protocol-specified crossover, which can complicate interpretation of OS. The guidance is still a draft and focuses mainly on trials where OS is not the primary endpoint, whereas HARMONi made OS a dual primary.

Still, it offers a useful lens. Read as a safety check, the HARMONi OS result is reassuring: the point estimate favors ivonescimab (hazard ratio 0.79, a 2.8-month difference in medians), and the interval's upper bound sits at 1.01, so there is no signal that adding the drug shortens survival. Read as an efficacy test against the requirement FDA stated, it is a miss, because the trial did not cross its significance threshold. The regulatory question is whether reassurance on harm, a met PFS endpoint, and a consistent updated trend can stand in for a statistically significant OS result that FDA said it needed. That is a judgment about benefit-risk and evidence standards, not a calculation, and FDA is the one making it.

The PFS result is large: a hazard ratio of 0.52 with an interval (0.41 to 0.66) far from 1. But FDA's stated concern in this setting is one PFS cannot settle by itself, whether delaying progression translates into longer life, which is the question OS exists to answer.


What the decision will test

There are three outcomes worth distinguishing, and each carries a different message for other programs.

  • Approval. It would suggest that, in a multiregional trial with a consistent OS trend and a met PFS endpoint, a near-miss on a dual primary OS endpoint can be weighed in the benefit-risk judgment rather than treated as disqualifying. Other sponsors would study the labeling language and any postmarketing requirements closely.

  • Complete response. It would reaffirm that a requirement FDA states in writing during review is the bar, even when the shortfall is small. Sponsors would read the letter for what FDA says additional evidence would look like, such as longer OS follow-up.

  • Something in between. A narrower indication, postmarketing commitments, or other action would show where FDA sees the line when evidence is promising but incomplete.

No outside party can predict which of these will occur, and this article does not try to. The more useful exercise is to notice what each would teach.


What sponsors can take from this now

  • Treat written FDA statements as the working bar. If FDA tells you a particular endpoint must be statistically significant, plan as though it means that, and revisit it in a Type B or C meeting if your data may fall short.

  • Think about OS early. When OS is a dual primary endpoint, power, follow-up duration, and subsequent-therapy effects deserve as much attention as the PFS design.

  • Know your regional story. In multiregional trials, consistency across regions and in the population relevant to the U.S. will be part of the review conversation.

  • Prepare for more than one outcome. Labeling, postmarketing studies, and communication plans look different after an approval, a narrow approval, and a complete response.


The decision is due this Saturday, though FDA can act before a goal date that falls on a weekend. Whatever it decides, the details of how it reasons about a p-value of 0.057 will be worth reading.


Sources: Summit Therapeutics, Forms 10-K (FY2025) and 10-Q (June 30, 2026), and Form 8-K disclosures; Summit Therapeutics press release, July 22, 2026; Drugs.com, Ivonescimab FDA Approval Status; The Pharma Letter, Summit Therapeutics profile; reported results for HARMONi and HARMONi-2 from Summit and Akeso; HARMONi primary analysis, Lancet Oncology 2026;27(9):1094-1109; updated HARMONi OS analysis, WCLC 2026 (Passaro et al.); OncLive, ILCN, and Targeted Oncology coverage of HARMONi; FDA draft guidance, Approaches to Assessment of Overall Survival in Oncology Clinical Trials (August 2025; Federal Register notice, August 19, 2025).

Comments


bottom of page