T Cell Engagers in Solid Tumours: Targets, Toxicity & Patient Selection

T Cell Engagers in Solid Tumours: Targets, Toxicity & Patient Selection

For a decade, the T cell engager was a haematology story: blinatumomab in acute lymphoblastic leukaemia, then CD20, BCMA and GPRC5D molecules in lymphoma and myeloma. Solid tumours were harder, with fewer clean targets, more normal-tissue toxicity and tumours that keep T cells out.

That has changed. Two T cell engagers now hold full approvals in solid tumours on overall survival data, tebentafusp in uveal melanoma and tarlatamab in small cell lung cancer, and a pipeline of DLL3, STEAP1, PSMA, KLK2 and CLDN18.2 programmes sits behind them. For IO, translational and biomarker teams, the question has shifted from “can this work?” to “in whom, at what expression level, and at what safety cost?”

Key takeaways

• Tebentafusp (gp100 x CD3, uveal melanoma) and tarlatamab (DLL3 x CD3, extensive-stage SCLC) have full FDA approval; both won on overall survival rather than response rate.
• CRS is common but mostly low grade with step-up dosing; ICANS and on-target, off-tumour toxicity are the harder problems in solid tumours.
• Patient selection is inconsistent: HLA typing for TCR-based molecules, no expression test for tarlatamab, and unstandardised IHC cut-offs in newer trials.
• Masked (protease-activated) and CD28 co-stimulatory designs are the main strategies to widen the therapeutic window.

Why T cell engagers worked first in haematology

A T cell engager links CD3 on a T cell to an antigen on a tumour cell, forcing a cytolytic synapse regardless of the T cell’s native specificity. Blood cancers suited this for three reasons:

• Tolerable on-target toxicity. CD19, CD20 and BCMA are shared with normal B cells or plasma cells, and B-cell aplasia is manageable. Few solid tumour antigens allow that kind of acceptable normal-tissue loss.
• Access. Malignant cells in blood, marrow and lymph nodes share compartments with T cells. Solid tumours have stroma, abnormal vasculature and a suppressive microenvironment.
• Homogeneous expression. Lineage antigens are on nearly every malignant cell; solid tumour antigens are often heterogeneous, and antigen-low cells escape.

In solid tumours, CD3 engagement can cause dose-limiting toxicity in normal epithelium before an effective dose is reached. Roche’s CEA x CD3 cibisatamab showed responses but was discontinued for gastrointestinal toxicity, as summarised in a 2026 review in The Oncologist. The successes have come from targets with sharply tumour-restricted expression.

Approved T cell engagers in solid tumours

Tebentafusp

Tebentafusp’s targeting arm is an engineered T cell receptor, so it reaches an intracellular antigen presented on HLA that antibodies cannot. In the phase 3 IMCgp100-202 trial (378 patients, randomised 2:1 against pembrolizumab, ipilimumab or dacarbazine), median overall survival was 21.7 versus 16.0 months (HR 0.51), per the New England Journal of Medicine report, while progression-free survival improved only modestly (HR 0.73). The FDA approved it on 25 January 2022 and the European Commission in April 2022.

The gap between survival and radiological response matters for biomarker teams. In a phase 2 study in previously treated patients, falls in circulating tumour DNA tracked with better outcomes even where scans did not change. The authors called this exploratory, but it makes ctDNA a candidate early readout for the class (see our piece on ctDNA and MRD in immunotherapy trials).

Tarlatamab

Tarlatamab targets DLL3, a Notch ligand on the surface of most SCLC cells and minimally expressed in normal tissue. After accelerated approval on phase 2 DeLLphi-301 data, the phase 3 DeLLphi-304 trial randomised 509 patients against topotecan, lurbinectedin or amrubicin. Median overall survival was 13.6 versus 8.3 months (HR 0.60) in the New England Journal of Medicine, and the FDA granted full approval on 19 November 2025. The European Commission followed in June 2026.

Key targets in development

The pipeline concentrates on antigens with restricted normal-tissue expression. Most of the data below are early-phase and from small cohorts.

Prostate cancer is now the most crowded solid tumour indication, with STEAP1 in phase 3. Programmes against broadly expressed epithelial targets (CEA, HER2, EGFR) have struggled, which is why many are being rebuilt as masked molecules.

Toxicity: CRS, ICANS and on-target, off-tumour effects

Cytokine release syndrome

In the tarlatamab pooled safety population (n=473), CRS occurred in 57% of patients, but grade 3 was 1.7% and grade 4 was 0.2%, mostly after the first two doses. Tebentafusp’s US label carries a boxed warning for CRS, with monitoring for at least 16 hours after each of the first three infusions. Management is now fairly standard:

• Step-up dosing: tarlatamab goes 1 mg on cycle 1 day 1, then 10 mg on days 8 and 15 and every two weeks.
• Monitoring: the tarlatamab label requires 22 to 24 hours of monitoring after the first two doses, and patients stay within one hour of a suitable facility for 48 hours.
• Treatment: corticosteroids and tocilizumab, graded by ASTCT criteria. Steroids were given to 31% of tarlatamab patients with CRS, tocilizumab to 10%.

ICANS

Tarlatamab’s boxed warning also covers neurologic toxicity including ICANS, seen in 10% of the pooled population with a median onset of 16 days. That is later than CRS, which matters for outpatient follow-up.

On-target, off-tumour toxicity

This is where solid tumour TCEs differ most from haematology. Tebentafusp’s common skin effects (rash, pruritus, dry skin) are consistent with gp100 on normal melanocytes; CEA engagers caused gut toxicity. Mapping normal-tissue expression with quantitative IHC and spatial methods is a core preclinical biomarker task.

The patient-selection question

HLA restriction for TCR-based engagers

Tebentafusp is indicated only for HLA-A*02:01-positive patients, so HLA genotyping is a mandatory pre-treatment test, as it is in brenetafusp’s phase 3. As TCR-based engagers extend to other alleles, typing method, turnaround and regulatory status become real questions. Loss of antigen presentation (HLA loss, B2M mutations) is a plausible resistance mechanism worth measuring.

Target expression and IHC thresholds

Tarlatamab is approved without a DLL3 test, since DLL3 is expressed in most SCLC. Outside SCLC the picture differs: in neuroendocrine prostate cancer, a phase 1b study of tarlatamab reported a 22% response rate in DLL3-positive patients against 10.5% overall. There is no agreed cut-off; DLL3 positivity has been defined at more than 1% and more than 50% of tumour cells in different studies, and DAREON-5 part 2 requires centrally assessed high expression. CLDN18.2 engagers also enrol on IHC expression.

The practical issues will be familiar from ADCs: antibody clone, scoring algorithm, heterogeneity between lesions, and archival versus fresh tissue. Our article on ADC and immunotherapy patient selection covers that debate, and the FDA companion diagnostic approval guide explains when an expression assay must become a CDx.

Tumour immune context

Baseline T cell infiltration, exhaustion and the myeloid microenvironment are logical predictors, though none is validated. On-treatment readouts such as ctDNA kinetics may prove more useful, especially where scans underestimate benefit.

Masked and co-stimulatory designs

Masked engagers block a binding arm with a peptide or polymer that tumour-enriched proteases remove, aiming to limit systemic CD3 activation:

• VIR-5500 (Vir Biotechnology), a dual-masked PSMA x CD3: in February 2026 Vir reported phase 1 data in 58 mCRPC patients, with no dose-limiting toxicities to date and CRS mostly grade 1 (company-reported).
• JANX007 (Janux Therapeutics), a masked PSMA x CD3 in phase 1b; Janux expects further data in the first half of 2027.

Masking changes the biomarker brief: programmes need evidence that cleavage happens in tumour rather than normal tissue, which means protease activity assays and measurement of cleaved versus intact drug in plasma and biopsies.

Co-stimulatory designs add a CD28 or 4-1BB signal through a separate tumour-targeted molecule to deepen T cell activity. Regeneron’s CD28 bispecifics (such as MUC16 x CD28) are built to pair with CD3 engagers or PD-1 blockade, and Janux plans a masked PSMA-targeted CD28 molecule, JANX013, to combine with JANX007. The risk is amplified cytokine release. The wider bispecific antibody field, including non-TCE formats such as PD-1 x VEGF, is a topic we will return to separately.

What assay and diagnostic vendors can offer

• Validated IHC for DLL3, CLDN18.2, STEAP1, PSMA and B7-H3, with agreed scoring and digital pathology for heterogeneous expression.
• Fast HLA typing, with allele coverage beyond HLA-A*02:01.
• Normal-tissue expression mapping with quantitative IHC and spatial proteomics to predict off-tumour risk.
• Cytokine, immune monitoring and anti-drug antibody assays for step-up dose optimisation.
• ctDNA assays as early response indicators, and cleaved-drug assays for masked molecules.

Many of these monitoring problems also arise in cell therapy; see our article on in vivo CAR-T.

Discuss T cell engagers at ImmunoMark San Diego

T cell engagers sit where immuno-oncology and biomarker science meet, the gap the immuno-oncology conference in San Diego is built to address. On 10–11 February 2027, 300+ senior pharma, biotech and diagnostics leaders meet across two parallel streams. Stream 1: Next-Gen Immuno-Oncology highlights next-generation cell therapies, multispecific antibodies and cancer vaccines; Stream 2 covers biomarkers and companion diagnostics.

Relevant sessions include “Overcoming Resistance Mechanisms to Improve Long-Term Responses to Immunotherapy” and “Novel Combination Immunotherapy Strategies for Hard-to-Treat Cancers” in Stream 1, and “Clinical Utility of Predictive Biomarkers in Personalised Oncology” in Stream 2. Speakers include Gregory J Opiteck, Head of Translational Medicine – Vice President, Janux Therapeutics, and Jitendra Kanodia, Vice President Head of Translational, Xencor. The format combines keynotes, panels, case studies, pre-arranged 1:1 meetings and a Day 1 drinks reception.

Register for ImmunoMark San Diego 2027

If your company wants to meet this audience directly, whether you offer IHC and companion diagnostic development, HLA typing, immune monitoring or spatial biology, see the San Diego 2027 sponsorship opportunities or our sponsor overview of the cancer immunotherapy conference in San Diego.

FAQs

What is a T cell engager?

A T cell engager is a bispecific molecule that binds CD3 on T cells with one arm and a tumour antigen with the other, so the T cell kills the tumour cell regardless of its native specificity. Most are antibody-based, such as tarlatamab (DLL3 x CD3). Some, such as tebentafusp, use an engineered T cell receptor to recognise a peptide presented on HLA.

Which T cell engagers are approved for solid tumours?

Tebentafusp is approved by the FDA (January 2022) and in the EU (April 2022) for HLA-A*02:01-positive unresectable or metastatic uveal melanoma. Tarlatamab received FDA accelerated approval in May 2024 and full approval in November 2025 for extensive-stage small cell lung cancer after platinum chemotherapy, with EU approval in June 2026. In the EU, catumaxomab (Korjuny) is also authorised for intraperitoneal treatment of malignant ascites in EpCAM-positive carcinomas.

How is cytokine release syndrome managed with T cell engagers?

Through step-up dosing, premedication, hydration and close monitoring after the first doses, with corticosteroids and tocilizumab for established CRS. With tarlatamab, CRS occurred in 57% of the pooled safety population but grade 3 or higher events were around 2%, mostly after the first two doses.

Do patients need a biomarker test before receiving a T cell engager?

It depends on the molecule. Tebentafusp requires HLA-A*02:01 genotyping. Tarlatamab is approved without a DLL3 test. Many newer programmes, including DLL3 engagers outside SCLC and CLDN18.2 engagers, select patients by IHC expression, but cut-offs are not yet standardised.

What are masked T cell engagers?

Masked or conditional T cell engagers carry a mask over a binding domain that is removed by proteases enriched in tumours. The aim is to limit CD3 activation in normal tissue and widen the therapeutic window. Examples in the clinic include VIR-5500 and JANX007, both targeting PSMA in prostate cancer.