Industry watchers who cheered on Summit Therapeutics’ claim that the first global phase 3 trial of its Akeso-partnered PD-1xVEGF drug showed consistent benefit between China and Western populations may be disappointed, although hopes for an approval remain alive thanks to newly calculated patient survival data from a longer follow-up.
As Summit disclosed in May, a combination of ivonescimab and chemotherapy significantly reduced the risk of disease progression or death by 48% at the primary analysis of HARMONi, which is the first global phase 3 trial for the closely watched PD-1xVEGF antibody.
At that time, Summit said the data showed “consistency of the magnitude” of the progression-free survival (PFS) benefit between patients enrolled in and outside Asia. The use of the word “consistency,” plus the similarity of the 48% rate compared to the 52% risk reduction previously observed in a subgroup of similar patients enrolled in the China-only HARMONi-A trial and later included in the global HARMONi analysis, led to positive reviews from several Wall Street analysts.
“[T]oday’s press release reports a best possible outcome between ivonescimab’s consistency of effect across the two different populations,” Cantor Fitzgerald analyst Eric Schmidt said in a May 30 note.
However, detailed results presented at the 2025 World Conference on Lung Cancer seem to depict a less rosy picture.
While investigators of HARMONi still maintained in a presentation that the PFS benefit was consistent across all predefined subgroups, a longer-term follow-up showed that, in European and North American patients, ivonescimab’s PFS improvement was 33% (hazard ratio 95% CI: 0.45-1.00), which is notably lower than the 45% recorded by Chinese patients. The risk reduction for the overall trial was 43% at the longer follow-up.
This is compared to the primary analysis—which had shorter follow-up—when only 18 PFS events occurred in the ex-China population, posting a massive 70% improvement for the experimental arm in this subgroup. By comparison, with 257 PFS events accrued, the Chinese population saw a 44% benefit from the ivonescimab regimen.
The subgroups were different sizes, with 165 patients in the ex-China subpopulation and 273 patients in the China cohort.
The 12-percentage-point PFS difference between the two cohorts in the long-term follow-up exceeded the 10-percentage-point threshold that Jefferies analyst Clara Dong, Ph.D., set for determining a “similar” showing and entered a territory that Leerink Partners analyst Daina Graybosch, Ph.D., labeled “really bad.” Dong set that margin in a Sept. 3 note about the primary analysis, and Graybosch laid out hers in a Sept. 5 interview with Fierce Pharma, both before seeing the detailed HARMONi data.
To Graybosch, a gap within 5 percentage points would be an easy call, and 5 to 10 points could cause some debate.
By itself, a 33% PFS benefit is decent and can be considered clinically meaningful in an oncology trial. The overall 48% PFS benefit recorded in the primary analysis meant that the trial did hit one of two primary endpoints.
“But that’s not the only question here,” Graybosh said. “The question, a broader one, is, [with] all the other phase 3 results we’re seeing in China, how applicable are they to outside of China?”
That’s why Graybosch argued that a difference bigger than 10 percentage points will be quite disappointing.
But Summit’s chief business and strategy officer, Dave Gancarz, pointed to a fresh-from-the-oven update to HARMONi’s overall survival analysis with a data cut in September, arguing that ivonescimab’s results were indeed consistent between the Chinese and Western patients.
Confusion around overall survival
While PFS met its overall mark in HARMONi, ivonescimab has disappointed when it comes to overall survival (OS), missing the trial’s other dual primary endpoint.
Summit already telegraphed the negative survival result in May. But the original wording of Summit and Akeso’s communication left many investors and analysts with the initial impression that the trial may still have a chance to reach statistical significance on OS in the future.
“Although statistical significance for OS was not yet reached at the time of the announcement, a promising trend toward OS benefit was observed,” Akeso said in a Aug. 26 release about its half-year performance.
However, the presentation at WCLC confirms that HARMONi missed OS at its final analysis. Confirming what Summit had previously disclosed, the death risk was 21% (HR 95% CI: 0.62-1.01) in favor of ivonescimab and chemo, with a p value of 0.057, while the statistical significance bar was set at a p value of 0.0448.
That means, even though Summit may take another look at OS, the company cannot claim statistical significance on an exploratory analysis because it was not prospectively baked in to the formal statistical plan, Graybosch told Fierce.
The final OS analysis was triggered by 261 deaths in the 438-person trial. At that time, 122 patients (56%) in the ivonescimab-chemo arm and 140 (64%) in the chemo group had died. The median OS was 16.8 months and 14 months, respectively.
Missing OS was particularly bad for ivonescimab because the FDA has told Summit that a statistically significant OS benefit is necessary to support an approval in second-line EGFR-mutated NSCLC.
Nevertheless, an FDA nod is not completely out of the question, Graybosch added. And newly updated OS data at WCLC suggest that ivonescimab may indeed still stand a chance.
A remaining shot at approval
Following the May announcement, Summit conducted another OS analysis with longer follow-up of the ex-China population. With a fresh data cutoff earlier this month, the death risk reduction from the ivonescimab regimen has expanded to 22% (HR 95% CI: 0.62-0.98) with an improved nominal p value of 0.0332. The number was 16% (HR 95% CI: 0.53-1.32) for the North American and EU group, or 30% (HR 95% CI: 0.38-1.30) for North America specifically, and 24% (HR 95% CI: 0.58-0.99) for Chinese patients. Again, none of these data carry statistical significance.
In its second-quarter report a few days ago, Akeso said that the China-only HARMONi-A detected a “statistically significant and clinically meaningful” OS at its final analysis. The positive readout gave several analysts optimism that data in HARMONi’s Western patients may improve in the future, too.
As the Chinese proportion (62% of entire trial population) of HARMONi came directly from patients in the earlier-concluded HARMONi-A study who had previously failed on a third-generation EGFR inhibitor, the longer follow-up was only for the Western population. The median follow-up for North American and European patients was 13.7 months for the update, versus 9.2 months when the final OS analysis happened. By comparison, Chinese patients had a median 32.7 months of follow up.
In the new analysis, the median OS for the control arm was 14 months for both Chinese and Western subgroups. The median OS was 17 months, not reached and 16.7 months for Western, North America and Chinese patients, respectively.
With the longer follow-up, “we’re seeing a similar pattern taking place in the West versus the East,” Summit’s Gancarz said in an interview with Fierce Pharma.
What the U.S. regulator will really be looking for is how the ex-China subpopulation performs, which is not statistically powered in HARMONi to begin with. Therefore, to Leerink Partners’ Graybosch, if ivonescimab shows a meaningful OS benefit in the Western subgroup with longer follow-up, combined with a robust PFS, the FDA may still be persuaded.
“What’s important to the FDA is already something that you’re not doing statistics on,” Graybosch said.
Summit intends to seek an FDA approval but will consider the exact timing, the company said in May. In arguing that the OS data still need to mature at that time, the California biotech pointed to how the median follow-up time for Western patients was less than the median OS.
Because the data cut was very recent, Summit has not decided whether this is the dataset that it will bring to the FDA for an approval, Gancarz said.
"We’re very encouraged by this data,” Gancarz said. “What becomes the next steps—we just haven’t gotten together to answer that question, because we just completed the [quality control] and analysis here.”
Again, that China-vs-Western consistency question
While Summit is encouraged by the new OS data, the readout still invites a question.
If Summit used HARMONi-A as a benchmark to design HARMONi while assuming ivonescimab would perform similarly between Chinese and Western patients, how did HARMONi reach the trigger of its final OS analysis so quickly that the 9-month follow-up for the all-important Western population was clearly too short?
One possible explanation would be that deaths in the Western population happened faster than expected in the first few months before ivonescimab could demonstrate its therapeutic effect, analyst Graybosch hypothesized without seeing the data. But Summit’s Gancarz told Fierce that the trial had recorded fewer death events in the West than planned when it hit the prespecified number.
A Fierce analysis of the OS curves showed some early differences between the two trials.
In HARMONi, ivonescimab’s benefit started to show after nine months. In HARMONi-A, separation of the survival curves between the experimental and control arms took about half the time, before month five, according to data presented last year after a median follow-up of 17.6 months. By month nine, the ivonescimab-chemo arm already had a 4.4-percentage-point absolute advantage over chemo alone in terms of patients’ probability of survival. The information was similar when HARMONi-A performed its first interim OS analysis with a data cutoff in June 2023 after a median follow-up of 10.2 months. Summit officially kicked off HARMONi in May 2023.
Gancarz argued that a 4.4-percentage-point difference is not that big to warrant any alarm over the general consistency of ivonescimab’s data between China and Western populations.
“I think the reality is that the Chinese patients are not completely identical to the North American patients,” and that includes the fact that about 14% of HARMONi-A patients only received an earlier-generation EGFR inhibitor, rather than a third-generation drug in HARMONi, Summit’s clinical development head, Jack West, M.D., said in a joint interview with Gancarz.
As to the short follow-up time for the Western population, West pointed to slow enrollment of HARMONi partly because trial activities were not yet back to normal after the pandemic.
“Summit was a completely unknown company, and ivonescimab was a completely unknown drug in the U.S. and globally, and that it just took longer than anticipated to get a toehold and start generating data,” West said.
Now the extra five months of follow-up is “really accounting for what would have been appropriate had the accrual and site activations and everything” progressed at an ideal pace, he added.
“The positive results from the HARMONi study underscore the global applicability of ivonescimab and demonstrates the potential benefit ivonescimab has to bring to patients around the world, including the United States,” Bob Duggan, chair and co-CEO of Summit, said in a Sept. 7 press release.