Bispecific Antibodies in Solid Tumours: Biomarkers, Toxicity & the Pipeline

Bispecific Antibodies in Solid Tumours: Biomarkers, Toxicity & the Pipeline

When most people hear “bispecific antibody” in oncology they think of T cell engagers. That half of the field is real, and we cover it separately in our analysis of T cell engagers in solid tumours. This article is about the other half: bispecifics that do not redirect T cells at all, but instead hit two receptors on the tumour, or two immune or angiogenic pathways at once.

That non-TCE group now includes two FDA-approved medicines in solid tumours (amivantamab and zanidatamab), two more approved in China (ivonescimab and cadonilimab), a US decision on ivonescimab due on 14 November 2026, and a wave of PD-(L)1 x VEGF licensing deals that has drawn in Pfizer, Bristol Myers Squibb, Merck and AbbVie. For biomarker and diagnostics teams, these molecules raise a question that monotherapies never did: when a drug has two targets, which one do you test for?

Key takeaways

• Non-TCE bispecific antibodies in solid tumours fall into two groups: receptor-targeting bispecifics (amivantamab, EGFR x MET; zanidatamab, biparatopic HER2) and immuno-oncology bispecifics (PD-1 x VEGF, PD-1 x CTLA-4, PD-1 x TIGIT).
• Approved receptor-targeting bispecifics are selected on one marker only: EGFR mutation for amivantamab (not MET) and HER2 by IHC/ISH for zanidatamab.
• Ivonescimab (PD-1 x VEGF) has three approved indications in China and beat pembrolizumab on overall survival in PD-L1-positive NSCLC in HARMONi-2. Its first FDA decision is due on 14 November 2026.
• Not every IO bispecific has succeeded: AstraZeneca stopped the phase 3 eVOLVE-Lung02 trial of volrustomig (PD-1 x CTLA-4) in August 2026.
• Toxicity tends to combine the class effects of both arms (for PD-1 x VEGF, immune-related events plus hypertension, proteinuria and bleeding), which makes baseline and on-treatment monitoring part of the biomarker plan.

Two families of non-TCE bispecific antibody

A bispecific antibody binds two different epitopes. In T cell engagers, one arm binds CD3 to recruit T cells. In the molecules discussed here, both arms do therapeutic work on their own targets, and the design aims for an effect that two separate antibodies would not give: higher avidity on cells that express both targets, receptor degradation, or concentrating an immune checkpoint blocker where its partner target is most abundant.

Receptor-targeting bispecifics: the established approvals

Amivantamab (EGFR x MET)

Amivantamab, from Johnson & Johnson, was the first bispecific antibody approved in a solid tumour. The FDA granted accelerated approval in May 2021 for EGFR exon 20 insertion NSCLC after platinum chemotherapy, with the Guardant360 CDx liquid biopsy approved alongside it as a companion diagnostic. The FDA then approved first-line use with carboplatin and pemetrexed on 1 March 2024 (PAPILLON) and approved amivantamab plus lazertinib on 19 August 2024 for first-line NSCLC with EGFR exon 19 deletions or L858R mutations (MARIPOSA). The final MARIPOSA overall survival analysis reported a hazard ratio of 0.75 against osimertinib.

A subcutaneous formulation, Rybrevant Faspro, was approved by the FDA in December 2025 across the existing indications. Beyond lung cancer, the FDA granted Breakthrough Therapy Designation and then, in July 2026, Priority Review to subcutaneous amivantamab in recurrent or metastatic head and neck squamous cell carcinoma after platinum chemotherapy and a PD-(L)1 inhibitor, based on the OrigAMI-4 study. That application is under review.

The biomarker lesson is striking. Amivantamab targets MET, yet none of its approved indications requires MET testing. Patients are selected on EGFR mutation status alone. The MET arm contributes to activity, particularly against MET-driven resistance, but it is not the selection marker.

Zanidatamab (biparatopic HER2)

Zanidatamab binds two different domains of HER2 (the ECD2 and ECD4 epitopes targeted separately by pertuzumab and trastuzumab), which drives receptor clustering and internalisation. The FDA granted it accelerated approval as Ziihera on 20 November 2024 for previously treated, unresectable or metastatic HER2-positive (IHC 3+) biliary tract cancer, based on HERIZON-BTC-01. Roche’s PATHWAY anti-HER2/neu (4B5) assay was approved as the companion diagnostic, the first for HER2 in biliary tract cancer. The confirmatory phase 3 HERIZON-BTC-302 is ongoing.

On 25 August 2026 the FDA approved two first-line regimens in HER2-positive gastroesophageal adenocarcinoma based on the phase 3 HERIZON-GEA-01 trial: zanidatamab plus tislelizumab and chemotherapy for IHC 3+ or IHC 2+/ISH-positive disease, and zanidatamab plus chemotherapy for IHC 3+ disease only. Jazz Pharmaceuticals reported median progression-free survival of 12.4 months with the triplet against 8.1 months with trastuzumab plus chemotherapy, and noted that the triplet is indicated regardless of PD-L1 status.

That label shows how bispecific biomarker definitions can split by regimen: the same drug, in the same disease, has a different HER2 cut-off depending on whether a PD-1 inhibitor is added. HER2 scoring in gastric and biliary tumours is already heterogeneous, so the IHC 2+/ISH boundary will matter in practice. Our article on claudin and other ADC biomarker strategies covers similar IHC threshold problems in the same tumour types.

IO bispecifics: PD-1 x VEGF and beyond

Ivonescimab (PD-1 x VEGF)

Ivonescimab was developed by Akeso and is licensed to Summit Therapeutics in the US, Europe and other territories. It is a tetravalent antibody that binds PD-1 and VEGF-A; the rationale is that VEGF is abundant in the tumour microenvironment, so the molecule should accumulate where PD-1 blockade is most needed.

In China, the NMPA has approved three indications:

• May 2024: with chemotherapy in EGFR-mutated non-squamous NSCLC after EGFR TKI therapy (HARMONi-A).
• April 2025: as monotherapy in first-line PD-L1-positive NSCLC, based on HARMONi-2, a head-to-head trial against pembrolizumab with a progression-free survival hazard ratio of 0.51. In September 2026, a pre-specified interim analysis reported median overall survival of 30.8 months versus 22.6 months (HR 0.73), according to Summit.
• August 2026: with chemotherapy in first-line squamous NSCLC, based on HARMONi-6 against tislelizumab plus chemotherapy, regardless of PD-L1 expression.

In the US, the FDA accepted Summit’s BLA for ivonescimab plus chemotherapy in EGFR-mutated non-squamous NSCLC after TKI therapy, with a PDUFA goal date of 14 November 2026. The application rests on the global HARMONi trial, which met its progression-free survival endpoint; overall survival favoured ivonescimab but did not reach statistical significance at the primary analysis, and Summit’s later update reported a Western-patient hazard ratio of 0.76. The ongoing global HARMONi-3 trial compares ivonescimab plus chemotherapy with pembrolizumab plus chemotherapy in first-line NSCLC regardless of PD-L1. No Western head-to-head data against pembrolizumab have been reported.

Dual checkpoint bispecifics

Cadonilimab (Akeso, PD-1 x CTLA-4) was approved by the NMPA in June 2022 for recurrent or metastatic cervical cancer after platinum chemotherapy, later for first-line cervical cancer with chemotherapy with or without bevacizumab (COMPASSION-16), and in September 2024 for first-line gastric or GEJ adenocarcinoma with chemotherapy (COMPASSION-15). Akeso began enrolling the global phase 3 COMPASSION-37 trial against chemotherapy with or without nivolumab in HER2-negative gastric cancer in September 2026.

Volrustomig (AstraZeneca, PD-1 x CTLA-4) is a cautionary case. In August 2026 AstraZeneca stopped the phase 3 eVOLVE-Lung02 trial in first-line PD-L1-negative NSCLC after an independent data monitoring committee concluded it was unlikely to meet its PFS or OS endpoints against pembrolizumab plus chemotherapy. Other volrustomig trials continue.

Rilvegostomig (AstraZeneca, PD-1 x TIGIT) is in a large ARTEMIDE phase 3 programme across lung, gastric and biliary cancers, at a time when several monoclonal anti-TIGIT programmes have failed. PD-1 x LAG-3 bispecifics remain earlier stage.

The PD-(L)1 x VEGF dealmaking wave

Ivonescimab’s head-to-head win over pembrolizumab triggered a run of licensing deals for Chinese-originated PD-(L)1 x VEGF bispecifics. The main ones are confirmed by company announcements:

The strategic logic is the PD-1 franchise: these molecules are positioned as potential successors to pembrolizumab and nivolumab, which is why so many of the pivotal trials are head-to-head against pembrolizumab rather than placebo. Pumitamig’s ROSETTA Breast-01 trial is notable from a biomarker angle because it enrols PD-L1-negative TNBC, a population where PD-1 monotherapy has little to offer.

Comparison of key non-TCE bispecific programmes

Biomarker questions specific to bispecifics

Which target do you select on?

The approved examples give a consistent answer: you select on the target that defines the disease biology, not necessarily on both. Amivantamab is selected on EGFR mutation, not MET; zanidatamab on HER2, which is both its targets. For PD-1 x VEGF molecules, neither VEGF expression nor circulating VEGF has been used as a selection marker. Angiogenesis biomarkers have a long history of failing to predict benefit from bevacizumab, and there is no evidence yet that the bispecific format changes that.

Does PD-L1 still matter for PD-1 x VEGF?

Partly. HARMONi-2 enrolled PD-L1-positive patients only, and the Chinese monotherapy label follows that. But HARMONi-6 and HARMONi-3 enrol regardless of PD-L1, and the theory of the class is that VEGF blockade normalises tumour vasculature and reduces immunosuppression, which could extend benefit into PD-L1-low tumours. The eVOLVE-Lung02 failure in PD-L1-negative NSCLC is a reminder that this has to be shown trial by trial. Expect PD-L1 to remain a stratification factor and a key subgroup analysis, and expect regulators to look closely at efficacy by PD-L1 band. Our PD-L1 predictive biomarker guide covers the assay and scoring issues that will carry over.

Target co-expression

Bispecific avidity effects depend on both targets being present on the same cell or in the same tissue compartment. That makes co-expression, rather than the level of either marker alone, the more mechanistically relevant measure, and it is something single-plex IHC cannot capture. Multiplex immunofluorescence and spatial methods can, and they are the natural tools for exploratory analyses of PD-1 x VEGF activity (PD-1-positive T cells near VEGF-rich vasculature) or EGFR and MET co-localisation.

Histology and mutation as de facto biomarkers

For ivonescimab, histology already functions as a safety biomarker: bleeding risk with anti-angiogenics has historically been concentrated in squamous NSCLC, and HARMONi-6 was run specifically to establish the squamous population. EGFR mutation status defines the HARMONi and HARMONi-A populations. In practice, most patient selection for these drugs will be driven by existing NGS and IHC workflows rather than new bispecific-specific tests.

Toxicity: the sum of two mechanisms

The toxicity of non-TCE bispecific antibodies generally looks like the combined class effects of both arms, without the cytokine release syndrome seen with T cell engagers.

• Amivantamab: infusion-related reactions (much less frequent with the subcutaneous form), EGFR-type rash and paronychia, MET-related oedema and hypoalbuminaemia, and, with lazertinib, venous thromboembolism, for which the US label recommends prophylactic anticoagulation in the first months of treatment.
• Zanidatamab: diarrhoea, infusion reactions and HER2-class effects including left ventricular dysfunction, plus a boxed warning for embryo-fetal toxicity.
• PD-1 x VEGF: in HARMONi-2, grade 3 or higher treatment-related adverse events occurred in 29.4% of ivonescimab patients against 15.6% with pembrolizumab, driven largely by VEGF-type effects (hypertension, proteinuria), while grade 3 or higher immune-related events were similar (7.1% vs 8.0%). Merck reported hypertension and proteinuria with MK-2010 too.
• PD-1 x CTLA-4: the design goal of cadonilimab and volrustomig is to deliver CTLA-4 blockade with fewer immune-related adverse events than the ipilimumab-nivolumab combination. eVOLVE-Lung02 reported no new safety signals, but its failure on efficacy shows that safety alone does not make the case.

For trial teams this means the monitoring plan covers two toxicity profiles at once: blood pressure, urine protein and bleeding risk alongside immune-related adverse event surveillance for PD-1 x VEGF molecules, or dermatological and thrombotic monitoring for amivantamab. Predictive markers of immune-related toxicity remain unvalidated, and that gap applies to bispecifics as much as to monotherapy.

What diagnostics vendors should do now

• Follow the selection marker, not the second target. The near-term CDx demand for approved bispecifics is in EGFR mutation testing (tissue and liquid biopsy) and HER2 IHC/ISH in GI cancers, including the IHC 2+/ISH-positive boundary written into the zanidatamab GEA label.
• Prepare for PD-L1 in new contexts. If PD-1 x VEGF bispecifics replace pembrolizumab in some settings, PD-L1 assay concordance, scoring algorithm and cut-off questions will be reopened for a new drug class.
• Offer co-expression and spatial readouts. Multiplex and spatial assays that measure both targets in context are the most likely route to a genuinely bispecific biomarker, and pharma sponsors need them in phase 2.
• Build safety monitoring into the offer. Proteinuria, blood pressure and thrombosis risk markers are routine laboratory tests, but standardised capture across global trials is not.
• Talk to the licensees now. With Pfizer, BMS, Merck and AbbVie all running or planning global trials, the companion and complementary diagnostic decisions for the PD-(L)1 x VEGF class are being made in the next two years. Our FDA companion diagnostic approval guide sets out the regulatory path.

Continue the conversation in Boston

Bispecific antibodies sit where next-generation immuno-oncology and biomarker strategy meet. ImmunoMark Summit Boston, our East Coast Edition in June 2027, brings together 300+ senior pharma, biotech and diagnostics leaders across two parallel streams. Stream 1: Next-Gen Immuno-Oncology covers novel IO modalities and combinations, and Stream 2 covers biomarkers and companion diagnostics. The format is keynotes, panels, case studies, pre-arranged 1:1 meetings and a Day 1 drinks reception.

For a wider view of the meeting and why it matters for this field, see our cancer immunology conference page.

Register for ImmunoMark Boston 2027

If your company wants to meet this audience directly, whether you supply PD-L1 or HER2 assays, multiplex and spatial platforms, or clinical trial laboratory services for IO programmes, see the Boston 2027 sponsorship opportunities.

FAQs

What is a bispecific antibody?

A bispecific antibody is an engineered antibody that binds two different targets, or two different epitopes on the same target. In cancer, some bispecifics redirect T cells to tumour cells (T cell engagers), while others block two tumour receptors, such as amivantamab (EGFR and MET), or two immune or angiogenic pathways, such as ivonescimab (PD-1 and VEGF).

Which bispecific antibodies are approved for solid tumours?

Outside the T cell engager class, the FDA has approved amivantamab for EGFR-mutated non-small cell lung cancer and zanidatamab for HER2-positive biliary tract cancer and first-line HER2-positive gastroesophageal adenocarcinoma. In China, the NMPA has approved ivonescimab (PD-1 x VEGF) for three NSCLC indications and cadonilimab (PD-1 x CTLA-4) for cervical and gastric cancer. An FDA decision on ivonescimab is due on 14 November 2026.

Is PD-L1 testing needed for PD-1 x VEGF bispecifics?

It depends on the indication. Ivonescimab monotherapy in China is approved for PD-L1-positive NSCLC based on HARMONi-2, but its squamous NSCLC approval with chemotherapy and the global HARMONi-3 trial do not require PD-L1 positivity. PD-L1 is likely to remain an important stratification factor and subgroup analysis even where it is not a selection requirement.

How does bispecific antibody toxicity differ from monotherapy?

Non-T cell engager bispecifics usually combine the class effects of both targets. PD-1 x VEGF bispecifics show immune-related events similar to PD-1 inhibitors plus VEGF-type effects such as hypertension and proteinuria; in HARMONi-2, grade 3 or higher treatment-related events were 29.4% with ivonescimab against 15.6% with pembrolizumab. Amivantamab combines EGFR-type skin toxicity with infusion reactions and, with lazertinib, a venous thromboembolism risk.