Durvalumab Plus FLOT Wins the Survival Race Pembrolizumab Lost

Adding durvalumab (Imfinzi) to perioperative FLOT cut the risk of death by 22% in resectable gastric and gastroesophageal junction adenocarcinoma. The phase 3 MATTERHORN final overall-survival analysis, published September 9, 2026, is the first statistically clear survival win for immunotherapy in this curative-intent setting.

Pembrolizumab already ran a similar perioperative design and did not clear the overall-survival bar. The new paper also shows the gain is not even: node-positive patients drive it, and node-negative patients do not.

The 7-Point Survival Gap at Three Years

MATTERHORN randomised 948 previously untreated adults with stage II to IVa resectable gastric or gastroesophageal junction adenocarcinoma, 474 to durvalumab plus FLOT and 474 to placebo plus FLOT. Investigators enrolled 1,258 people from November 17, 2020, to September 2, 2022, at 147 centres in 20 countries across Asia, Europe, and the Americas. Median age was 62. Seventy-two percent were men, 68% had a gastric primary, and 70% had clinically node-positive disease.

At a data cutoff of September 1, 2025, after about 43 months of follow-up, 160 of 474 patients (34%) in the durvalumab group had died, against 192 of 474 (41%) on placebo. Overall-survival maturity was 37%. Median overall survival was not reached in either arm. The hazard ratio was 0.78 (95% CI 0.63 to 0.96; p=0.021), which cleared the prespecified threshold of p<0.0499.

MATTERHORN MAIN RESULTS

Event-free survival, the primary end point, had already been met, with a hazard ratio of 0.71 (95% CI 0.58 to 0.86; p<0.001). Median event-free survival was not reached with durvalumab and was 32.8 months with placebo. The 24-month event-free rate was 67.4% versus 58.5%. One-year event-free survival was 78.2% versus 74.0%.

Among patients with evaluable central pathology, a pathological complete response was reached in 91 of 385 (24%) on durvalumab versus 34 of 372 (9%) on placebo. Those are the same 91 and 34 complete responses counted over all 474 patients per arm in the intention-to-treat rate of 19.2% versus 7.2%. Post-hoc event-free analyses still favoured durvalumab after a complete response (HR 0.29), a major pathological response (HR 0.32), and any pathological response (HR 0.60). The complete-response comparison rested on 11 events, and the authors warned that the interval is wide.

Node-Positive Patients Drive the Overall Win

About 70% of the trial, 659 of 948 patients, had clinically node-positive disease at randomisation. In that group the overall-survival hazard ratio was 0.73 (95% CI 0.57 to 0.94). In node-negative disease the hazard ratio was 1.01 (95% CI 0.67 to 1.51). A treatment-by-nodal-status interaction test was not statistically significant (p=0.19), so the split is a signal, not a proven interaction. It is still the number that will follow these patients into clinic.

OVERALL SURVIVAL BY SUBGROUP

PD-L1 tumor area positivity of at least 1% covered 90% of the study, 853 of 948 patients, scored on the VENTANA SP263 assay. Both TAP groups carried an OS hazard ratio of 0.79. The TAP-negative interval (0.41 to 1.50) crosses 1.0 and sits on only 10% of the sample. A post-hoc look at histology found a diffuse-type hazard ratio of 0.98 (95% CI 0.68 to 1.40) and an intestinal-type hazard ratio of 0.76 (95% CI 0.56 to 1.05).

Josep Tabernero, MD, head of medical oncology at Vall d’Hebron University Hospital in Barcelona and a senior MATTERHORN investigator, said the regimen is “the first immunotherapy strategy to demonstrate prolonged survival in patients with early-stage gastric or gastroesophageal junction cancer, and point to a new standard of care.” He also put the disease in context: complete resection remains the cornerstone, FLOT is the Western peri-operative chemo, and five-year survival is still 33% in the United States and 25% globally.

GI oncologists digesting the paper have treated the 7-point three-year gap as the clinic number and have kept returning to those two flat curves, diffuse-type and PD-L1-negative, because the US and EU labels do not carve them out. Practices that follow the intention-to-treat win will give durvalumab to patients whose own subgroup has not shown a survival difference.

Why Pembrolizumab Missed Overall Survival

MATTERHORN did not walk into an empty field. KEYNOTE-585 tested perioperative pembrolizumab with chemo in the same disease and reported its main findings in 2024, with a later overall-survival update after a median 59.9 months of follow-up. The main cohort of 804 patients used cisplatin plus capecitabine or cisplatin plus fluorouracil. A separate 203-patient cohort used FLOT. Pathological complete response rose. Overall survival did not clear statistical significance.

TWO PERIOPERATIVE IMMUNOTHERAPY TRIALS

In KEYNOTE-585’s main cohort, pathological complete response was 12.9% with pembrolizumab versus 2.0% with placebo. Event-free survival had a hazard ratio of 0.81 (p=0.0198) and missed its prespecified significance threshold of p=0.0178. Final median overall survival was 71.8 months versus 55.7 months, hazard ratio 0.86 (95% CI 0.71 to 1.06). In the FLOT cohort, pathological complete response was 17.0% versus 6.8%, and the overall-survival hazard ratio was 1.04, with 24-month overall survival of 72% versus 73%.

MATTERHORN put every patient on FLOT, the stronger Western triplet, and powered the study for event-free survival first, then overall survival. KEYNOTE-585 built its main analysis on cisplatin doublets and left FLOT in a 203-patient safety cohort that could not carry an overall-survival claim. The MATTERHORN authors, writing in context, said KEYNOTE-585 and ATTRACTION-5 had shown that combining anti-PD-1 therapy with chemo in non-metastatic disease was feasible, but that neither trial showed a definitive overall-survival improvement. Durvalumab, an anti-PD-L1 antibody, is the drug that crossed that line, on FLOT, in 948 patients.

FLOT’s Path From German Hospitals to a Global Label

FLOT earned this job in FLOT4, the German phase 2/3 trial led by Salah-Eddin Al-Batran, MD, of Krankenhaus Nordwest in Frankfurt. That study randomised 716 patients with resectable gastric or gastroesophageal junction adenocarcinoma to perioperative FLOT or to ECF/ECX, the older epirubicin-cisplatin-fluoropyrimidine triplet. Overall survival favoured FLOT, hazard ratio 0.77, with a median overall survival of 50 months versus 35 months. Five-year overall survival was 45% versus 36%, and three-year overall survival was 57% versus 48%. Pathological complete response was 16% versus 6%, and margin-free resection was 85% versus 78%.

The 15-month median survival gain made FLOT the Western default. Recurrence stayed common. AstraZeneca notes that about one in four patients who undergo surgery develop recurrent disease within a year, and that gastric cancer remains the fifth most common cancer worldwide and the fifth-highest cause of cancer death, with nearly one million people diagnosed each year and about 660,000 deaths in 2022.

The American Cancer Society estimates 31,510 new cases of stomach cancer in the United States for 2026, and about 10,740 deaths. In 2024, AstraZeneca counted roughly 6,500 drug-treated patients in the US with early or locally advanced gastric or gastroesophageal junction cancer. Yelena Y. Janjigian, MD, who leads gastrointestinal oncology at Memorial Sloan Kettering Cancer Center and is MATTERHORN’s principal investigator, has said the trial was the first phase 3 study to take FLOT global, including in Asian centres where a two-drug combination or surgery then adjuvant chemo had been the norm. Nineteen percent of MATTERHORN, 180 patients, came from Asia. An accompanying commentary on the paper noted there was no accrual from China, which accounts for about 40 to 50 percent of gastric cancer cases worldwide.

THE ROAD TO THE LABEL

1. 2010 to 2015: FLOT4 enrols 716 patients in Germany and later shows FLOT beating ECF/ECX on overall survival (HR 0.77).
2. 2024: KEYNOTE-585 reports a pathological complete-response gain without a statistically significant overall-survival result.
3. June 2025: MATTERHORN meets event-free survival (HR 0.71) in a New England Journal of Medicine paper and an ASCO plenary.
4. November 25, 2025: FDA approves perioperative durvalumab with FLOT, then durvalumab alone.
5. March 16, 2026: The European Commission approves the same regimen.
6. September 9, 2026: The Lancet publishes the final overall-survival analysis (HR 0.78) and event-free survival by pathological response.

Each step of that chain bought a similar relative death reduction, roughly 20 to 25 percent, stacked on the last standard rather than replacing it.

The Extra 10 Cycles After Surgery

Patients received durvalumab 1,500 mg or placebo every 4 weeks plus FLOT every 2 weeks for 4 cycles, 2 before surgery and 2 after, then durvalumab or placebo every 4 weeks for 10 more cycles. The FDA label matches that schedule. The agency also stated, in plain language, that the trial was not designed to isolate the effect of durvalumab in the neoadjuvant phase from the adjuvant phase. The extra year of antibody after the knife is part of the approved package and has never been tested on its own.

HOW THE REGIMEN IS GIVEN

• Before surgery: Two cycles of durvalumab 1,500 mg every 4 weeks with FLOT every 2 weeks.
• After surgery: Two further cycles of durvalumab with FLOT, then durvalumab alone every 4 weeks for 10 cycles.
• Cap: Treatment continues until progression, recurrence, or unacceptable toxicity, or a maximum of 12 cycles after surgery.

Safety, reported at a December 20, 2024 cutoff, tracked the known profiles of both drugs. Any-grade adverse events occurred in 99% of patients in both groups. Grade 3 or 4 events occurred in 71.6% with durvalumab and 71.2% with placebo. Serious events were 48% versus 44%. Events leading to discontinuation of any trial treatment were 30% versus 23%. Any-grade immune-mediated events were 23% versus 7%. Adverse events with an outcome of death possibly related to any trial treatment occurred in 6 of 475 patients (1%) on durvalumab and 2 of 469 (less than 1%) on placebo. The share of patients who completed surgery was similar.

That is the trade. Grade 3 or 4 toxicity is already high on FLOT, and adding durvalumab did not raise it by much. It did raise immune-mediated events and the rate at which patients stopped some part of treatment. Oncologists reading the paper have started asking what the extra 10 monotherapy cycles add in the real world, because the registration study cannot answer that question.

Regulators Moved Months Before the Paper

The US Food and Drug Administration approved durvalumab with FLOT on November 25, 2025, as neoadjuvant and adjuvant treatment, followed by single-agent durvalumab, for adults with resectable gastric or gastroesophageal junction adenocarcinoma. The application had priority review, breakthrough designation, and orphan designation, and it ran under Project Orbis with regulators in Australia, Brazil, Canada, Israel, and Switzerland. The European Commission followed on March 16, 2026. Japan was still listed as under review in AstraZeneca’s March notice.

Today’s approval marks the first immunotherapy regimen approved in the neoadjuvant setting for gastric and gastroesophageal junction cancers. Nearly seven in 10 patients were alive at three years following treatment with the durvalumab-based perioperative regimen. This survival benefit, observed regardless of PD-L1 status, establishes a new standard of care in this curative-intent setting.

Yelena Y. Janjigian, MD, Memorial Sloan Kettering Cancer Center, AstraZeneca approval statement, November 25, 2025

Dave Fredrickson, AstraZeneca’s oncology haematology executive vice president, called it the third US approval for a perioperative Imfinzi regimen. The September paper does not reopen that decision. It fills in the overall-survival table, the pathological-response splits, and the nodal and histology curves that clinics will now argue over. A five-year overall-survival analysis is still planned. For node-negative and diffuse-type disease, that later look is the one that will decide whether the label’s all-comers language holds up in the groups that did not move.

Frequently Asked Questions

What Is FLOT Chemotherapy in the MATTERHORN Regimen?

FLOT is a four-drug intravenous triplet-plus-fluorouracil schedule given every 2 weeks: docetaxel 50 mg/m², oxaliplatin 85 mg/m², leucovorin 200 mg/m², and fluorouracil 2,600 mg/m² as a 24-hour infusion, the same doses used in FLOT4. MATTERHORN gave 2 cycles before surgery and 2 after, which is fewer FLOT cycles than the original FLOT4 plan of 4 before and 4 after, because durvalumab then continues alone.

Does the US Approval Require a PD-L1 Test?

No. The November 25, 2025, FDA indication has no PD-L1 cutoff and no nodal restriction. Ten percent of MATTERHORN patients had a TAP score under 1%, and that subgroup’s overall-survival interval crossed 1.0. Ninety percent scored TAP of at least 1% on the SP263 assay, a higher PD-L1-positive share than many gastric immunotherapy trials have enrolled.

Can the Extra Durvalumab After Surgery Be Dropped?

The label does not allow a tested shorter course. For patients weighing at least 30 kg, the recommended dose is 1,500 mg every 4 weeks with chemo for up to 4 cycles, then 1,500 mg alone every 4 weeks for up to 10 cycles. Below 30 kg the dose is 20 mg/kg on the same calendar. The FDA stated that MATTERHORN was not designed to isolate neoadjuvant from adjuvant durvalumab, so skipping the monotherapy year is an untested deviation, not a studied option.

How Did MATTERHORN Score a Pathological Complete Response?

Masked central review used modified Ryan criteria: no residual viable tumour cells in the primary tumour and in the resected lymph nodes, equal to 100% pathological regression. Major pathological response added the near-complete group with only single cells or rare small clusters. The event-free analysis inside the complete-response group was based on 11 events, which is why the authors treated that particular hazard ratio with caution.

Five-year survival from this cohort is still to come, and that later look is the one that will test whether node-negative and diffuse-type patients ever share the 7-point gap. Until then the approved regimen is FLOT plus durvalumab for resectable stage II to IVa disease, on an intention-to-treat win pembrolizumab never produced.

Disclaimer: This article is news reporting and analysis of a published clinical trial and of regulatory decisions. It is for information only and is not medical advice, a treatment recommendation, or a substitute for a consultation with an oncologist. Readers facing a gastric or gastroesophageal cancer diagnosis should discuss perioperative chemo and immunotherapy, including durvalumab plus FLOT, with a qualified medical oncologist and surgeon who have the full medical record, pathology, and staging. Figures, subgroup results, and approval status reflect the trial paper, company statements, and regulator notices cited here and may change as five-year survival data, other country reviews, and guideline updates appear.