FDA Publishes Ivonescimab Rejection Letter, Revealing 93% of Trial Data Came From China
The United States Food and Drug Administration has published its own written reasons for refusing to approve Summit Therapeutics’ lung cancer drug ivonescimab, and the letter reveals a problem that goes beyond the survival data that dominated the first headlines. According to the agency, 93 per cent of the events used to measure whether the drug delayed the disease occurred in patients enrolled in China, leaving it unable to say whether the benefit would apply to patients in the United States.
The complete response letter, dated 13 November 2026, was published on the FDA’s website this week as part of the agency’s programme of publishing rejection letters, and was reported by Fierce Pharma and STAT. Summit had announced the rejection itself three days earlier, on 15 November, saying the agency had declined to approve ivonescimab in combination with platinum chemotherapy for patients with EGFR-mutated, locally advanced or metastatic non-squamous non-small cell lung cancer whose disease had progressed after treatment with an EGFR tyrosine kinase inhibitor, commonly known as a TKI. Reuters reported at the time that the agency had cited the failure of the late-stage HARMONi trial to show a statistically significant improvement in overall survival.
The published letter, which runs to the full reasoning behind that decision, lists four clinical and statistical deficiencies. The first, and the one that has drawn the most attention from drug developers and trial specialists, concerns the geography of the evidence. The FDA said the applicability of the trial’s prespecified primary endpoint, progression-free survival, to the US population “cannot be established as 93% of events (257 of 275 PFS events) occurred in patients enrolled in China”. Chinese patients made up about 60 per cent of the people enrolled in the trial, yet accounted for almost all of the cancer progressions and deaths that the trial measured.
The second deficiency concerns the size of the benefit. The HARMONi trial did meet its progression-free survival endpoint: patients who received ivonescimab plus chemotherapy went a median of 6.8 months before their cancer worsened or they died, compared with 4.4 months for patients who received chemotherapy plus a placebo, a hazard ratio of 0.52. But the FDA described that as a modest improvement, and noted that the final overall survival analysis did not reach statistical significance, with no substantial difference in the proportion of patients whose tumours shrank. In the primary survival analysis, carried out in April 2025, patients on ivonescimab lived a median of 16.8 months against 14.0 months on chemotherapy alone, a hazard ratio of 0.79 with a p-value of 0.057, just short of the statistical bar the trial had set for itself. Tumours shrank substantially in 44.7 per cent of patients on the drug, against 34.2 per cent on chemotherapy alone.
The third deficiency concerns the longer follow-up data that Summit offered in support of its application. The company had argued that updated analyses, with data running to June 2026, showed a roughly 24 per cent lower risk of death on ivonescimab, with a hazard ratio of 0.76 in both the global population and the Western subgroup. The FDA called those analyses exploratory only, because they were carried out after the event and were not part of the trial’s prespecified plan. The agency added that the survival analyses were weakened by different data cut-off dates across regions, which it attributed to follow-up ending earlier for patients enrolled in China, where clinical sites closed during 2025.
The fourth deficiency is the balance of benefit against risk. The FDA concluded that the modest gain in progression-free survival does not outweigh the risks of adding ivonescimab to chemotherapy, including serious immune-related side effects and side effects linked to blocking VEGF, the protein that helps tumours build a blood supply. In the trial, serious treatment-related side effects occurred in 28.0 per cent of patients on the ivonescimab regimen, against 15.1 per cent on chemotherapy alone, according to reports of the letter’s contents.
How the trial became so lopsided
The letter also records how the imbalance arose, and that history matters because it shows the agency had warned Summit long before the decision. The trial was originally designed as a single-country study enrolling only in China, and was later modified to add North America and the European Union. Enrolment in China ran from January to November 2022, but enrolment in North America and the EU did not begin until May 2023, nearly a year and a half later. The FDA said it had raised concerns about this enrolment approach at a meeting in March 2023, and again at the pre-filing meeting in May 2025, when it told the company that a statistically significant survival benefit would be necessary for approval.
The practical consequence of that staggered start was that Chinese patients were followed for far longer, and so contributed almost all of the events the trial needed to measure its endpoints. By the time the primary survival analysis was performed in April 2025, patients in North America and Europe had been followed for a median of only 9.2 months. The trial’s design therefore answered a question about Chinese patients with confidence, and a question about Western patients only tentatively, and the agency was not willing to bridge that gap with after-the-event analyses.
Summit has not yet said publicly when, or even whether, it will resubmit, although Citi analysts said in a note on 16 November, after speaking with the company’s management, that a resubmission is being contemplated, according to Fierce Pharma. The FDA’s letter recommends a new adequate and well-controlled multiregional trial showing a statistically significant and clinically meaningful improvement in overall survival, progression-free survival or another endpoint measuring clinical benefit, sufficient to outweigh the drug’s toxicity, and asks the company to meet the agency before starting it to discuss the design and endpoints. Under federal rules for biologics, Summit must resubmit or take another available action within one year of the letter, or the agency may treat the lack of a response as a request to withdraw the application. The letter also asks for a safety update and English translations of approved foreign labelling with any resubmission.
What ivonescimab is, and why so much was riding on it
Ivonescimab is a first-in-class bispecific antibody that attacks cancer in two ways at once. It blocks PD-1, a protein that damps down the immune response, and it blocks VEGF-A, the protein that helps tumours grow the blood vessels they need to survive. It was invented by the Chinese biotech company Akeso, which holds the rights in China, while Summit holds the rights in the United States and other licensed territories. The drug is already a commercial product in China: it received its first approval there in May 2024, in combination with chemotherapy, for the same group of patients whose disease had progressed after TKI therapy, on the basis of the HARMONi-A trial, a China-only study. That study, published in the Journal of the American Medical Association, showed a 54 per cent reduction in the risk of disease progression, and its final analysis later showed a statistically significant survival benefit, with the risk of death 26 per cent lower on ivonescimab. In April 2025 the drug was also approved in China as a first-line treatment for patients with PD-L1-positive lung cancer, after a head-to-head trial in which it beat pembrolizumab, the world’s best-selling cancer immunotherapy, and in August 2026 it was approved there in combination with chemotherapy for first-line squamous lung cancer.
That record in China is part of why the American rejection stung. Summit’s shares fell 16.4 per cent on the Monday after the announcement, and about 10 per cent more on Tuesday, according to Fierce Pharma. The drug had been one of the most closely watched experimental cancer medicines in the industry, not only because of its unusual dual mechanism, but because it embodied a new model of drug development: a medicine invented and first proven in China, then carried into the American market by a US partner. That model is now facing a harder question than its supporters expected, about whether evidence generated mostly in one country can support approval in another.
The patients at the centre of the decision are a defined and difficult group. Non-small cell lung cancer accounts for the large majority of lung cancers, and lung cancer remains the leading cause of cancer death worldwide. Patients whose tumours carry EGFR mutations are usually treated first with a targeted TKI drug, and when that stops working, the options narrow sharply to chemotherapy. According to Summit’s own account of the rejection, no FDA-approved regimen has demonstrated a statistically significant overall survival benefit in this setting. That is the unmet need ivonescimab was meant to address, and it is why the agency’s demand for a new trial, rather than an approval on the evidence available, is a real setback for those patients as well as for the company.
Analysis: why progression-free survival was not enough
The most important lesson of the ivonescimab letter is about what counts as proof in cancer medicine, and it is a lesson that reaches well beyond one drug. Progression-free survival, the time before a cancer grows or the patient dies, is a legitimate and widely used measure, and the HARMONi trial met it convincingly, cutting the risk of progression or death by almost half. But the FDA has always treated overall survival, how long patients actually live, as the gold standard, and it has grown steadily more insistent on it for drugs that are added to chemotherapy in advanced disease, where the extra toxicity is real and the extra time is measured in months.
The agency’s reasoning is not hard to follow. A drug that delays a scan showing growth, but does not demonstrably extend life, and does not shrink tumours much more often than chemotherapy alone, while causing serious side effects in more than a quarter of patients, has to clear a high bar to justify its place in treatment. The FDA’s letter says, in effect, that ivonescimab did not clear it. The company’s counter-argument, that longer follow-up eventually showed a survival benefit in both Western and Asian patients, was not dismissed out of hand, but it was treated as what it was: an analysis done after the results were known, with different cut-off dates in different regions, and therefore not the reliable evidence the agency requires to approve a medicine.
There is a second lesson, about trial design, and it may prove the more consequential one for the industry. The FDA’s letter is a rare, detailed statement of how the agency reads a trial that began in one country and widened later. It does not object to foreign data as such. It objects to a trial in which the distribution of events, not just the total number of patients, is so skewed that the result effectively describes one population while being offered as evidence about another. The agency pointed to its own guidance on multiregional trials, including the international ICH E17 standard, as the reference point. Sponsors planning trials with large non-US enrolment will read this letter as a worked example of the question reviewers will ask, and the answer they will expect is a trial designed from the start to enrol patients in several regions at the same time, with follow-up that ends on the same date everywhere.
That has immediate implications for a whole class of medicines. In recent years, American and European companies have increasingly licensed drugs invented in China, where trials can be run faster and at lower cost, and then sought approval at home on the strength of that evidence. The FDA has already signalled scepticism about this approach in other cases, and the ivonescimab letter is now its clearest statement of the terms on which such evidence will be judged. It is not a ban on foreign data. It is a demand that the data be generated in a way that lets the agency see the drug working in the population that will actually receive it.
Why the publication itself matters
The third point worth noting is the publication of the letter at all. Until recently, complete response letters, the documents in which the FDA tells a company why its drug cannot be approved, were treated as confidential, and the public learned the agency’s reasoning only through the company’s own, often selective, account. The FDA has begun publishing the letters itself, with trade secrets and personal information removed, and the ivonescimab letter is one of the early examples. The difference is visible in this case: Summit’s announcement emphasised the survival question, while the published letter shows that the geography of the trial data was at least as important, and that the agency had been raising it for more than three years.
That transparency cuts both ways. It gives patients, doctors and investors a fuller picture of why a drug was refused, and it gives other companies a detailed guide to what the agency expects. It also removes the shelter that companies once had to present a rejection in the most favourable light. For an industry that has often complained that the FDA’s decisions are opaque, the new openness is an answer, and the ivonescimab letter is a demonstration of what it looks like in practice.
The decision also lands in a week in which the FDA has shown a firmer hand in other areas. On Friday, the agency declined to approve Capricor Therapeutics’ cell therapy for Duchenne muscular dystrophy, citing serious data integrity problems in a key trial, including unauthorised changes to the analysis plan. The two decisions are unrelated in their details, but together they suggest a regulator that is willing to refuse applications, and to say why in public, even when the medicines involved address serious unmet needs. That is a difficult posture for patients to hear, but it is the posture the agency is paid to hold.
What to watch next
Several things will determine whether ivonescimab ever reaches American patients. The first is the HARMONi-3 trial, which is comparing ivonescimab plus chemotherapy with pembrolizumab plus chemotherapy as a first-line treatment for non-small cell lung cancer. Results for squamous disease could arrive by the end of the year, with non-squamous data due in the first half of 2027, according to trade-press reports. A strong result there, in a trial designed from the start as a global study, would do more for the drug’s American prospects than any argument about the HARMONi data.
The second is what Summit decides about the new trial the FDA has asked for. Designing and running a multiregional phase 3 trial in lung cancer takes years and costs hundreds of millions of dollars, and the company will have to weigh that against the size of the American market for this particular group of patients, and against the progress of its other trials. The one-year clock in the letter forces a decision, but it does not force a trial: the company can also withdraw the application, or narrow its ambitions to a different group of patients.
The third is the wider regulatory picture. If other companies with China-first drugs read this letter and redesign their trials, the ivonescimab decision will have changed how medicines are developed, not just whether one medicine is sold. If they do not, more rejections of the same kind are likely. Either way, the letter has set a standard that will be quoted back to the agency, and to the industry, for years: a trial that proves a drug works in one country does not, by itself, prove it works in another, and a delay in progression is not the same as a longer life.
Sources
- US Food and Drug Administration, Complete Response Letter, BLA 761524, 13 November 2026 (via openFDA)
- STAT, FDA discloses reasons for rejecting Summit drug, 18 November 2026
- Fierce Pharma, FDA’s complete response letter for Summit’s ivonescimab cites China-heavy data beyond survival miss, 17 November 2026
- CASRAI, FDA publishes ivonescimab rejection letter, 19 November 2026
- Akeso / PR Newswire, Ivonescimab’s updated label shows positive dual results in PFS and OS from HARMONi-A study