In the race to forge the next Keytruda from PD-(L)1xVEGF bispecific antibodies, Pfizer and the BioNTech-Bristol Myers Squibb alliance are split on how to position overall survival—the gold-standard oncology endpoint—in their first-line non-small cell lung cancer (NSCLC) clinical trials.
While Pfizer insists on making overall survival (OS) as at least one of the primary endpoints across all its phase 3 trials for 3SBio-partnered PF‑08634404, including in the Symbiotic-Lung-01 trial for first-line NSCLC, BioNTech and BMS recently revised the design for their Rosetta-Lung02 study of pumitamig to make progression-free survival (PFS) the single primary endpoint.
BioNTech and BMS’ decision came as a surprise. Merck & Co.’s Keytruda got its chemo combo approvals in first-line NSCLC with trials that use OS and PFS as primary endpoints, and both Pfizer and Summit Therapeutics are following the same strategy in their PD-1xVEGF phase 3s.
Given Keytruda’s established standard-of-care status, anything less than a statistically significant OS benefit versus the entrenched blockbuster, even if approvable, would likely not revolutionize the U.S. market. That’s why a key debate persists as to whether PD-(L)1xVEGF bispecifics can truly replace Keytruda despite strong head-to-head PFS data against the PD-1 king.
The new Rosetta-lung02 design means that, “with a PFS readout, we can already sit down with regulators if we hit the PFS, and we can start an accelerated approval path,” Özlem Türeci, M.D., co-founder and chief medical officer of BioNTech, said in an interview with Fierce on the sidelines of the 2026 American Society of Clinical Oncology annual meeting.
“What this also ensures is that our entire alpha is on PFS,” Türeci continued. “And I think this is accepted as the earliest and most sensitive acceptable endpoint for a phase 3 study for this compound class.”
Türeci stressed that OS remains a key secondary endpoint that will be part of any review package with the FDA, and that the study is powered for OS analysis. Besides the endpoint adjustment, the partners also increased the trial size from 982 to 1,260 patients to give it sufficient statistical power.
“It is simply how you test PFS first, and then OS after,” Cristian Massacesi, M.D., BMS’ chief medical officer, said in a separate interview with Fierce, also noting that OS will be formally tested in a predefined analysis.
This design allows BioNTech and BMS to have a faster readout to at least get in the FDA’s door without having to wait for a mature OS readout, Türeci explained. In the heated race among PD-(L)1xVEGF bispecifics, detailed data are important. Another factor that will determine the winner is “speed of execution,” she said.
Starting a potential regulatory filing conversation earlier was the exact rationale behind Summit’s recent decision to add an interim PFS analysis to the squamous cohort of the Harmoni-3 trial for Akeso-partnered ivonescimab in first-line NSCLC, although that plan didn’t pan out as the trial is now continuing to its final analysis without hitting its interim goal. But OS is still a co-primary endpoint in Harmoni-3.
BioNTech and BMS made PFS the sole primary endpoint after discussions with the FDA, Türeci said. However, some recent controversies have suggested that the FDA may not always be as aligned on regulatory paths as drugmakers indicate.
“The FDA would tell you that if they don’t agree in particular with a selected endpoint, they would tell you very clearly,” Türeci said.
Fully aware of BioNTech and BMS’ method, Pfizer is staying course with OS as a co-primary endpoint in its Symbiotic-Lung-01, the company’s oncology chief development officer, Johanna Bendell, M.D., told Fierce on the sidelines of ASCO 2026.
“We still see that overall survival is very important, not only for patient impact and understanding the impact of the medicine that you’re developing, but also from a regulatory standpoint. There’s an importance to show that overall survival.”
In lung cancer, having a statistically significant OS is “really important” because it reinforces doctors’ confidence that PFS benefits are indeed meaningful, David Spigel, M.D., president and chief medical officer at Sarah Cannon Research Institute and an ASCO expert in lung cancer, told Fierce on the sidelines of the meeting.
Would hitting OS as a secondary endpoint “as robust scientifically as a primary endpoint? I would say no,” he said. “But the difference is probably very small, and it’s still very meaningful clinically.”
In first-line NSCLC, PFS alone is likely not enough for regulators in the U.S. “unless you had something with it, like a secondary strong OS endpoint or quality of life endpoint in an unmet medical need,” he said.
Julie Brahmer, M.D., director of the thoracic oncology program at the Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins, said she understands that “drug development is going so fast these days, in order to stay ahead of the next wave, you need some early readouts” such as PFS. But clinicians will decide whether to use a drug approved this way based on the options in front of them, she said in a separate interview with Fierce.
If two PD-(L)1xVEGF drugs both have statistically significant OS in first-line NSCLC, Brahmer said she would prefer the one that hit it as a primary endpoint versus secondary.
Updated NSCLC results at ASCO
Debate over statistical designs aside, competition in the immuno-oncology space will largely come down to data. At this year’s ASCO meeting, BMS and BioNTech provided results from the phase 2 dose optimization portion of Rosetta-Lung02, the first look at a PD-(L)1xVEGF agent in first-line NSCLC in a global population.
At an April 13, 2026, data cutoff, among 40 evaluable patients with a median follow-up of 9 months, pumutamig plus chemotherapy showed a confirmed objective response rate of 57.1% in patients with nonsquamous NSCLC and 68.4% with squamous NSCLC. The unconfirmed ORR was 71.4% and 73.7%, respectively.
The results are better among patients who took the lower, 1400 mg dose, which BioNTech/BMS have decided to move into the phase 3 portion. There, the confirmed ORR was 63.6% in nonsquamous patients and 72.2% in the squamous group.
Both Türeci and Massacesi highlighted results in the PD-L1-negative population, which do not benefit much from anti-PD-1/L1 agents alone. Across the two dosage strengths, pumitamig and chemo’s confirmed ORR reached 47.6% in PD-L1-negative patients, 77.8% in PD-L1-low and 100% in PD-L1-high.
“This confirms our strategy for phase 3, that we have an all-comers study, because we want to defragmentize first-line non-small cell lung cancer,” Türeci said.
As the patient populations get smaller with subgroups, Türeci said she was encouraged that even with a relatively higher percentage of PD-L1-negative patients, pumitamig’s ORR results look competitive. In a China phase 2 trial, ivonescimab plus chemo achieved a 62% ORR in an inclusive first-line NSCLC population.
With not enough follow-up time as of the cutoff, the phase 2 portion of Rosetta-Lung02 currently did not have PFS or OS data to report.
Also at ASCO, an update from an early-stage Chinese study linked single-agent PF-08634404, at its FDA-aligned dose of 10 mg/kg, to a 67.6% confirmed ORR in first-line, PD-L1-positive NSCLC. The ORR was 75% among squamous patients and 63.6% in nonsquamous.
So far, the results among the three leading agents are largely similar. Two differences might set pumitamig apart due to its PD-L1 component versus PD-1, BMS’ Massacesi noted, although he cautioned that “we still don’t know how [they will] translate clinically.”
PD-L1 is more present in the tumor microenvironment than PD-1, which could give pumitamig an advantage in terms of synergy with the VEGF component. PD-L1 inhibition is also historically associated with fewer and less severe cases of immune-related pneumonitis compared with PD-1 inhibition.
For its part, Pfizer’s Bendell noted that PF-08634404 is differentiated preclinically with more potent anti-angiogenic effect.
Clinically, Bendell highlighted the combination potential of PF-08634404 with Pfizer’s internal ADCs, especially its vedotin-based ADCs, because preclinical data suggest that “the vedotin ADCs may be more synergistic with immunotherapies” compared with Topoisomerase 1 payloads.
