After upstaging Merck & Co.’s Keytruda as a monotherapy for patients with first-line non-small cell lung cancer (NSCLC) in China, Akeso and Summit Therapeutics’ ivonescimab has shown it can delay tumor progression versus another PD-1 inhibitor as part of a combination with chemotherapy.
In the first phase 3 clinical trial in which a regimen has beaten the combination of a PD-1 inhibitor and chemotherapy in first-line NSCLC—with the caveat of China-only data—ivonescimab and chemo reduced the risk of progression or death by 40% compared with BeOne Medicines’ Tevimbra and chemo in patients with first-line squamous NSCLC.
The ivonescimab regimen extended the time patients lived without tumor progression by a median 4.2 months versus the Tevimbra-based therapy, reaching 11.1 months.
Results of the Akeso-run HARMONi-6 trial were presented during a presidential symposium at the 2025 European Society for Medical Oncology (ESMO) Congress in Berlin. As ivonescimab is the first product in the hot PD-(L)1xVEGF bispecific class, which holds potential to become the new cancer immunotherapy backbone, the drug’s phase 3 data will naturally be scrutinized under a microscope.
One key focus of a trial involving a PD-1/L1 inhibitor is how the drugs performed by PD-L1 expression subgroups. Results in HARMONi-6 favored ivonescimab across PD-L1 expression levels, according to the presentation.
In patients with PD-L1 expression below 1% total proportion score, ivonescimab’s progression-free survival (PFS) advantage reached 45%. The effect size was 36% in those with PD-L1 expression scores between 1% to 49%, and 29% in patients whose PD-L1 scored at least 50%.
The 40% PFS improvement in the overall population met investors’ expectations, which ranged from 35% to 40%, according to analysts’ notes from Leerink Partners and Jefferies ahead of Sunday’s presentation.
But the ESMO presentation included no information on overall survival (OS), a key secondary endpoint that would be even more important for understanding ivonescimab’s potential opportunity in the U.S.
The median follow-up for PFS was 10.3 months, and the OS endpoint remained immature at the data cutoff. Critics have argued that the absence of an OS report can indicate weakness because other companies have shared immature OS data at similar time points in NSCLC trials, Leerink analysts noted in an ESMO preview note.
The nondisclosure is not about the data being bad, but about them being immature, Dave Gancarz, chief business and strategy officer at Akeso’s U.S. partner Summit, said in an interview with Fierce Pharma.
Gancarz pointed to the duration of tumor response as an indicator of potential OS gains down the line. Among responders, ivonescimab-chemo takers’ responses lasted a median 11.2 months, versus 8.4 months for Tevimbra-chemo recipients. The objective response rate was also higher for the ivonescimab arm, at 75.9%, compared with 66.5% in the Tevimbra arm, with the Akeso/Summit drug again showing an advantage across PD-L1-negative and -positive subgroups.
Chinese authorities are reviewing Akeso’s application for the HARMONi-6 regimen in first-line squamous NSCLC and have not requested another OS data cut, Gancarz said.
But the Chinese regulatory agency is expected to do so based on its typical practice. For HARMONi-2, the drug’s high-profile Keytruda head-to-head trial, ivonescimab showed a 22.3% OS improvement over Keytruda, which did not bear statistical significance. After obtaining that additional, unplanned OS analysis, the Chinese agency in April granted the drug an approval as a monotherapy for the first-line treatment of PD-L1-positive NSCLC based on the drug showing a 49% PFS advantage in the study.
Following the HARMONi-2 win, the choice of Tevimbra-chemo as the comparator in HARMONi-6 has been called into question because Keytruda-chemo is the standard option and more relevant from a global perspective.
But HARMONi-6 is, after all, a Chinese study, and, for Akeso, the goal is to commercialize in China, Gancarz noted. In China, Tevimbra is one of the most popular PD-1 inhibitors, as all foreign-made options such as Keytruda only make up 15% of the Chinese PD-1/L1 market, according to an IQVIA report based on first-half 2025 data.
Tevimbra’s approvals in China and Europe in first-line squamous NSCLC were based on results from the phase 3 Rationale-307 trial. In that study, Tevimbra-paclitaxel-carboplatin showed a 47.6% PFS improvement over paclitaxel-carboplatin, while Tevimbra’s combo with nab-paclitaxel and carboplatin led to a 52.2% improvement. The median PFS for both Tevimbra regimens was the same at 7.6 months, versus 5.5 months for chemo alone. The data came after a median study follow-up of 8.6 months.
In the Keynote-407 trial that won Keytruda its first-line squamous NSCLC nods with chemo, the Merck regimen initially reported a 6.4-month median PFS (44% PFS improvement) after a median follow-up of 7.8 months. The figure later increased to eight months (43% PFS improvement) after 14.3 months of median follow-up.
Cross-trial comparisons can be difficult considering the various baseline patient characteristics such as PD-L1 expression distribution and region of enrollment. At least based on the above data, Tevimbra-chemo performed largely similarly to Keytruda-chemo in first-line squamous NSCLC in their separate registrational trials. And Gancarz pointed to a perception in the oncology field that PD-1 inhibitors are more similar than different.
“All of the discussions we’ve had with KOLs, treating physicians and so forth, each of them has kind of scoffed it [off],” Gancarz said, referring to any notion of a material difference between PD-1s. “They’re like, it’s a PD-1 […] That’s been one that people have been very accepting in terms of, that’s not going to be the judgment piece here in the trial.”
The 6.9-month median PFS that the BeOne regimen posted in HARMONi-6’s control arm slightly underperformed its 7.6-month Rationale-307 result, while Jefferies analysts set the “in-line” performance boundary at 7 to 7.5 months, according to an Oct. 8 note. The PD-L1 expression distribution between the two trials are different. While both trials had about 40% PD-L1-negative patients, the proportion of patients with PD-L1 expression between 1 and 49 was 25% in Tevimbra’s study, versus 40% for ivonescimab’s trial. PD-L1-high groups made up 35% and about 20% of the two trials, respectively.
Gancarz also touted ivonescimab’s safety profile. The drug’s bispecific construct has allowed it to reach patients who have not historically been able to receive PD-1/L1 inhibitors and VEGF medicines as separate drugs. In squamous NSCLC, VEGF inhibitors such as bevacizumab are not allowed because of potential life-threatening pulmonary hemorrhage.
In HARMONi-6, grade 3 or above treatment-related adverse events were more frequent for ivonescimab than Tevimbra, at 63.9% and 54.3%, respectively. But the rate of discontinuation was low, at 3.4% and 4.2% for the two meds, respectively. Side effects led to death in 3% of the ivonescimab population versus 3.8% for Tevimbra.
VEGF-related toxicities were also more frequent in the experimental arm, with 7.5% at grade 3 or above versus 2.3% for the control group. Rates of grade 3 or above proteinuria, hemorrhage and hypertension events were 2.3% versus 0%, 1.9% versus 0.8%, and 3% versus 1.1%.
Lately, attention on the ivonescimab program has shifted to Summit’s global HARMONi-3 trial, which is testing the PD-1xVEGF drug against Keytruda in their chemo combinations in first-line NSCLC, including both squamous and nonsquamous histologies.
Leerink analysts have raised concerns about that trial design because the efficacy bar for Keytruda-chemo in the two NSCLC subtypes is different, so the trial result may eventually by skewed by one group. Gancarz pointed to the trial’s 1,080-patient size, arguing that each histology will have enough representation for a meaningful analysis.
The HARMONi-6 readout comes after Summit last month reported results from the first global phase 3 data for ivonescimab. The study, HARMONi, tested ivonescimab and chemo in previously treated EGFR-mutant NSCLC.
The trial disappointed investors because of an OS data shortfall, even though a longer follow-up has shown promise. Summit has not disclosed a specific regulatory plan, even though it has said it intends to file for approval. Gancarz declined to comment on whether the company has decided to use the September data cut or a longer follow-up to pursue an FDA nod, but he said the company will “clarify that point in short order.”