When we see news of a new cancer drug, our attention is easily drawn first to the most striking number: how much tumors shrank, how long progression was delayed or how much the risk of death fell. But understanding a study also requires knowing what treatment the other group received.
If the control group already received an effective immunotherapy–chemotherapy combination and the new drug still produced an additional improvement, the result answers a specific, important question: how much further can we go from the progress already made?
Akeso's phase 3 HARMONi-GI1 study, presented at ESMO 2026, offers an example. Conducted in China in first-line advanced biliary tract cancer, it compared ivonescimab plus chemotherapy with durvalumab plus chemotherapy and met its prespecified overall survival primary endpoint. [1][2]
Why does the comparator matter?
Biliary tract cancer encompasses malignancies arising in the bile ducts, gallbladder and related sites, with differences between sites and molecular characteristics. Clinical research continues to explore how outcomes can be improved in advanced disease.
Combining immunotherapy and chemotherapy has already become an important treatment approach in this field. Durvalumab, used in HARMONi-GI1's control group, is a PD-L1 inhibitor. Research already supports the clinical use of its combination with chemotherapy. [1]
That context needs to remain visible when we read these results. The study tested the difference between two complete regimens, on a foundation that already included immunotherapy.
The reference point for evaluating new drugs changes as medicine advances. A trial's choice of comparator affects both the question it can answer and the relevance of its findings to patients today. Understanding the control treatment helps us judge the weight of a positive result more accurately.
What additional action does ivonescimab offer?
Ivonescimab is a bispecific antibody targeting both PD-1 and VEGF. PD-1 participates in the regulation of inhibitory signals in immune cells; blocking this pathway can, under suitable conditions, strengthen antitumor immune responses. VEGF is involved in blood-vessel formation and is also related to the local tumor immune environment.
By combining these actions in one molecule, researchers hope to influence both immune suppression and the tumor's vascular environment. Original mechanistic research also observed that binding to the two target types could mutually enhance some binding and functional activities, providing experimental support for the molecular design. [3]
But a clear distinction must be maintained between mechanistic research and clinical outcomes. Cell and animal experiments can explain “why it is worth trying”; randomized studies in patients can then test “what it actually achieves.”
A bispecific antibody recognizing two targets does not automatically mean twice the treatment effect. Nor can a number be extracted from this trial to establish how much benefit came from the VEGF component alone. The immunotherapy components also differed between groups. The trial compared complete regimens and did not isolate the effects of each component.
What did this phase 3 study show?
HARMONi-GI1 was randomized, double-blind and multicenter. In its prespecified interim analysis, median overall survival was 14.9 months with ivonescimab plus chemotherapy versus 12.1 months with durvalumab plus chemotherapy; the hazard ratio for death was 0.75. The company also reported improvements in secondary endpoints, including progression-free survival. [1]
These overall survival results support an additional benefit from the new regimen in the population studied. The 14.9 and 12.1 months are medians of the groups' survival distributions. Their difference does not mean that every patient lives an additional 2.8 months. Nor is a hazard ratio the “proportion of patients cured.”
What deserves attention is that overall survival was specified in advance as the primary endpoint, and the comparator against which that improvement was achieved. This allows a more specific discussion of the bispecific combination's clinical potential in advanced biliary tract cancer.
These results also have a defined scope. Biliary tract cancer is not a single uniform disease. Overall findings need to be understood alongside eligibility criteria, subtypes, prior and subsequent treatment and other information. Whether a number observed in a subgroup supports a reliable conclusion also depends on whether testing was prespecified and the sample was large enough.
How should risk be assessed when another pathway is added?
Bringing immune-related and VEGF-related actions into one molecule means that safety assessment must also cover the corresponding risks. An ingenious mechanistic design still needs testing with patient data.
The company reported serious adverse events in 43.4% and 43.7% of the two groups. [1] This measure describes events that meet the definition of “serious.” It is not the total of all side effects and cannot replace data on grade ≥3 events, bleeding or treatment discontinuation.
Safety assessment also requires knowing exactly what happened, when it occurred, whether it resolved and how it affected treatment and daily life. Similar overall proportions of serious events still require a detailed reading of individual risks and treatment exposure. Average results do not mean that every patient has the same ability to tolerate treatment.
This brings a positive study back to the most practical clinical judgment: how much benefit can be added, what burden must be accepted, and which patients are more likely to gain from that tradeoff?
What needs to be tested next?
This study provides important evidence for ivonescimab in biliary tract cancer. As follow-up, regulatory assessment and studies in more regions progress, attention will continue to focus on long-term survival, specific safety outcomes, quality of life and applicability across healthcare settings.
We can respect what a study conducted in China has already established while leaving questions about patient characteristics, care pathways and access in other regions to subsequent evidence. International use requires specific evaluation, as any new treatment does when it reaches more patients.
HARMONi-GI1 is especially useful for understanding a study's frame of reference. Hiding the control group and retaining only “the new drug improved outcomes” removes an important part of the result's meaning. Knowing what treatment the control group already received makes it clearer where the additional progress occurred.
As effective treatments multiply, the direction of medical progress also becomes more specific: to keep finding additional benefits, supported by evidence and tolerable for patients, on top of existing treatment.
References
- Akeso: Original announcement of phase 3 HARMONi-GI1 overall survival results, 2026-10-25. Includes study design, overall efficacy, safety and further development information.
- Summit Therapeutics: Original announcement of results from the HARMONi-GI1 study in China, 2026-10-25. Provides the partner's account of the findings and where the study was conducted.
- Original research on ivonescimab's molecular design and cooperative binding mechanism, 2025. Title: Design of a fragment crystallizable-engineered tetravalent bispecific antibody targeting programmed cell death-1 and vascular endothelial growth factor with cooperative biological effects. This paper provides mechanistic and preclinical evidence; it does not substitute for the phase 3 clinical results.
Evidence cutoff: October 27, 2026. Clinical results are based mainly on original company material released during the meeting. Full publications and further follow-up remain worth following. This article is for public science education. Consult an oncologist about specific treatment.
This is Before the Conclusion.
What does a news item, a number or a seemingly good result really mean? Follow Before the Conclusion as we use medicine and science to understand the reasoning and tradeoffs behind the evidence, and keep asking questions.
Text and image usage statement
The text of this article is original work by Before the Conclusion. You are welcome to share a link to this webpage. Copying, reposting, republishing screenshots and commercial use are prohibited without permission. Any separately supplied accompanying illustrations are subject to the same restrictions and may not be used individually without permission. Image-production and source information is given in the relevant captions.