Advanced Therapies Meet Biomarkers: What IO Teams Should Watch This Winter

Advanced Therapies Meet Biomarkers: What IO Teams Should Watch This Winter

Advanced therapies have changed what a biomarker has to do. For a checkpoint inhibitor, a biomarker mostly helps decide who is likely to respond. For an engineered T cell, a bispecific, a personalised vaccine or an injected oncolytic virus, the biomarker can also decide who is eligible, shape what gets manufactured and determine how response is measured.

The ESMO Immuno-Oncology & Advanced Therapies Congress meets in London on 14–16 December 2026, and it closes a year in which several advanced therapies crossed important regulatory and clinical lines.

This article looks at six areas where advanced therapies and biomarker science now overlap: cell therapy in solid tumours, multispecifics and T cell engagers, cancer vaccines, intratumoural therapy, spatial and single-cell biology, and AI. For each one, we set out what has changed and which biomarker questions are still open. For the broader picture across immuno-oncology, see our companion piece on immuno-oncology biomarker trends heading into 2027.

Key takeaways

• In advanced therapies, the biomarker is increasingly part of the product: an eligibility gate, a manufacturing input and a response read-out.
• The first engineered TCR-T approval needed two companion diagnostics (HLA typing and MAGE-A4 IHC), a pattern other HLA-restricted products are likely to follow.
• Tarlatamab’s conversion to full approval shows a T cell engager can succeed without a selection biomarker, which moves the biomarker question to safety and resistance.
• Positive phase 3 topline data for an individualised neoantigen vaccine (INTerpath-001) make the neoantigen pipeline itself a regulated, biomarker-driven process.
• The accelerated approval of the first intratumoural oncolytic virus for anti-PD-1-refractory melanoma puts response assessment of injected and non-injected lesions under scrutiny.
• Spatial, single-cell and AI tools are most useful when they answer one of these modality-specific questions, not as stand-alone discovery engines.

Why advanced therapies change the biomarker brief

The table below sums up how the biomarker job differs across the main advanced therapy classes. The common thread is that patient selection, product design and pharmacodynamic monitoring are converging into one integrated biomarker strategy.

1. Cell therapy in solid tumours: the antigen is the eligibility test

TCR-T: HLA and antigen, both required

The first engineered cell therapy approved for a solid tumour set the template. In August 2024 the FDA granted accelerated approval to afamitresgene autoleucel (Tecelra) for previously treated synovial sarcoma in patients who carry specific HLA-A*02 alleles and whose tumours express MAGE-A4. Two companion diagnostics support that label: an HLA sequencing assay and a MAGE-A4 immunohistochemistry assay. Every patient must clear both gates before manufacturing can begin.

The next NY-ESO-1 product follows the same logic. Letetresgene autoleucel (lete-cel) received FDA Breakthrough Therapy designation in January 2025 for NY-ESO-1-expressing myxoid/round cell liposarcoma, based on the IGNYTE-ESO trial, in which 42% of patients with synovial sarcoma or myxoid/round cell liposarcoma (27 of 64) responded by independent review. Its cell therapy assets moved to US WorldMeds in 2025.

TIL therapy: the product is the biomarker

Lifileucel (Amtagvi) has held US accelerated approval in post-PD-1 melanoma since February 2024. Its European application was withdrawn in July 2025, with the company noting it may refile. For TIL therapy, the most interesting biomarker work sits on the product side: which clonal and phenotypic features of the infused cells predict durable benefit.

CAR-T: first randomised evidence in a solid tumour

In CT041-ST-01, published in The Lancet in 2025, the Claudin18.2-directed CAR-T satricabtagene autoleucel (satri-cel) was compared with physician’s choice in 156 patients with heavily pretreated CLDN18.2-positive gastric or gastro-oesophageal junction cancer. Median progression-free survival was 3.25 months versus 1.77 months (HR 0.366). The overall survival difference was smaller and less certain. For biomarker teams, the questions are where to set the CLDN18.2 expression cut-off and how heterogeneous expression and antigen escape limit durability.

What to watch: antigen expression thresholds tied to outcome, early expansion and persistence markers, and evidence of antigen loss at progression. Our review of CAR-T cell therapy biomarkers covers the haematology evidence these programmes build on.

2. Multispecifics and T cell engagers: from selection to safety

T cell engagers sit on a spectrum. At one end, TCR-based molecules such as tebentafusp (Kimmtrak), approved in 2022 for uveal melanoma, depend on peptide presentation by HLA and are restricted to HLA-A*02:01-positive patients. At the other end, tarlatamab (Imdelltra), a DLL3 x CD3 bispecific, moved from accelerated to traditional FDA approval in November 2025 for extensive-stage small cell lung cancer after platinum chemotherapy. In the phase 3 DeLLphi-304 trial (509 patients), median overall survival was 13.6 months with tarlatamab versus 8.3 months with standard chemotherapy (HR 0.60).

Tarlatamab reached that point without a DLL3 selection test in its label. That shifts the biomarker effort away from enrolment and towards three other questions: who will not respond despite target expression, who is at risk of cytokine release syndrome or neurotoxicity, and what drives resistance (target loss, T cell exhaustion or an immune-excluded microenvironment).

What to watch: target expression in tumour versus normal tissue for new solid tumour engagers, early cytokine and T cell activation markers as dose-finding tools, and multispecific formats that add co-stimulation or checkpoint blockade, which make the pharmacodynamic picture more complex. Where a selection marker is needed, plan the companion diagnostic co-development from the first in-human study.

3. Cancer vaccines: the neoantigen pipeline becomes the product

On 19 August 2026, Merck and Moderna announced that the phase 3 INTerpath-001 trial of intismeran autogene plus pembrolizumab met its primary endpoint of recurrence-free survival, and the key secondary endpoint of distant metastasis-free survival, against pembrolizumab alone in 1,137 patients with resected high-risk melanoma. Detailed efficacy figures have not yet been presented, and overall survival is still being assessed. The phase 2b KEYNOTE-942 study had shown a 49% reduction in the risk of recurrence or death.

If the full data hold up, individualised vaccines will force a set of biomarker questions into the regulatory spotlight. Tumour and normal sequencing quality, neoantigen prediction algorithms and ranking rules become part of how the product is made. Tissue sufficiency becomes an eligibility issue. And immune monitoring, measuring whether the vaccine produced a meaningful T cell response, becomes central to proving mechanism.

In England, the NHS Cancer Vaccine Launch Pad already refers patients into personalised mRNA vaccine trials, under a government agreement with BioNTech to provide precision cancer immunotherapies to up to 10,000 patients by 2030. That makes UK sites a meaningful source of the translational data these questions need.

What to watch: the full INTerpath-001 data, whether ctDNA is used to select or monitor patients in the adjuvant setting, and standardisation of T cell response assays across programmes.

4. Intratumoural therapy: rethinking how response is measured

On 6 August 2026, the FDA granted accelerated approval to vusolimogene oderparepvec (Tudriqev, formerly RP1), an engineered herpes simplex virus given by intratumoural injection, in combination with nivolumab for advanced melanoma that has progressed on anti-PD-1 therapy. The approval followed two complete response letters and rests on the single-arm IGNYTE study, in which about a quarter of patients responded, with a median duration of response of just over 14 months. A randomised confirmatory trial is under way.

The regulatory debate around this product was partly a biomarker debate. With an injected therapy, the questions are how to separate the contribution of the virus from that of the checkpoint inhibitor, and how to read responses in injected versus distant, non-injected lesions. Pseudoprogression and inflammatory flare complicate standard imaging criteria.

What to watch: paired biopsies showing immune changes in non-injected lesions, the use of ctDNA as a systemic read-out, and proposals for response criteria adapted to intratumoural therapy.

5. Spatial biology and single-cell biomarkers: answering the mechanism questions

Each modality above raises a spatial or single-cell question. For CAR-T and TCR-T, it is how antigen expression is distributed across the tumour and whether antigen-negative regions explain relapse. For T cell engagers and oncolytic viruses, it is whether effector T cells actually reach tumour nests or stay in the stroma. For TIL and vaccines, single-cell TCR sequencing can track which clones expand, persist and correlate with response.

The practical challenge is moving from description to decision. Spatial and multiplex imaging data are rich but expensive, and most current studies are small and retrospective. The multiplex platform chosen early in a programme affects whether findings can later become a robust assay.

What to watch: spatial features measured before treatment that predict benefit in prospective cohorts, and single-cell signatures of the infused product or vaccine-induced T cells that could become release or monitoring criteria.

6. AI tools for biomarker discovery: credibility over novelty

AI already sits inside several advanced therapy workflows. Neoantigen selection depends on machine learning models of peptide processing and HLA binding. Digital pathology algorithms can score antigen expression and immune infiltration more consistently than manual reads, which matters when a cut-off decides eligibility for a cell therapy.

Regulators are setting expectations. In January 2025, the FDA issued draft guidance on the use of AI to support regulatory decision-making for drugs and biologics. The guidance, which had not been finalised as of October 2026, proposes a risk-based credibility framework that starts with the model’s context of use. For biomarker teams, the message is that an AI model that informs eligibility or efficacy claims needs the same validation discipline as any other assay. Our guide on validating an IO biomarker in a clinical trial sets out that discipline, and our FDA companion diagnostic guide covers the regulatory path.

What to watch: AI models validated on independent, multi-site cohorts, and AI-assisted scoring locked down before pivotal trials begin.

Reading this winter’s data

As results emerge at the ESMO congress in London in December and through early 2027, apply three tests to each advanced therapy dataset. Is the gating biomarker defined with a cut-off that links to outcome? Can the pharmacodynamic read-out be measured outside the originating lab? And is there a credible route to a validated, scalable assay? Data that pass all three are the data most likely to change programmes.

Continue the conversation in London in March

Large congresses are where advanced therapy data first appears. The follow-up work of turning that data into biomarker strategy, trial design and diagnostics partnerships happens in smaller, more focused rooms.

That is what our immuno-oncology conference in London is built for. ImmunoMark Summit London returns for its second year on 3–4 March 2027, with a target of 400+ attendees from pharma, biotech and diagnostics. Two parallel streams map onto the themes in this article: Next-Gen Immuno-Oncology and Biomarkers & Companion Diagnostics. The format combines keynotes, panels and case studies with pre-arranged 1:1 meetings and a Day 1 drinks reception. The 2026 London edition brought together 300+ attendees from 100+ companies, 70% of them senior decision-makers.

Register for ImmunoMark London 2027

If your company wants to meet this audience of advanced therapy and biomarker decision-makers directly, see the London 2027 sponsorship opportunities.

FAQs

Which biomarkers decide eligibility for TCR-T cell therapy?

TCR-T therapies recognise a tumour antigen presented by a specific HLA molecule, so patients usually need both a matching HLA type and tumour expression of the target antigen. Afamitresgene autoleucel, for example, requires specific HLA-A*02 alleles and MAGE-A4 expression, each confirmed with a companion diagnostic.

Do T cell engagers need a companion diagnostic?

Not always. TCR-based engagers such as tebentafusp are restricted to HLA-A*02:01-positive patients, while tarlatamab was approved for small cell lung cancer without a DLL3 selection test. For engagers without a selection test, biomarker work focuses on safety, pharmacodynamics and resistance.

When and where is the ESMO Immuno-Oncology & Advanced Therapies Congress 2026?

It takes place in London, United Kingdom, from 14 to 16 December 2026.

Why does response assessment matter for intratumoural therapies?

Injected therapies act locally and systemically, so teams need to show responses in non-injected lesions as well as injected ones, and to separate the effect of the injected agent from any combination partner. Pseudoprogression and inflammation can also confuse standard imaging criteria.

Where can advanced therapy and biomarker teams continue these discussions in 2027?

ImmunoMark Summit London on 3–4 March 2027 brings together senior pharma, biotech and diagnostics leaders in two parallel streams, Next-Gen Immuno-Oncology and Biomarkers & Companion Diagnostics, with pre-arranged 1:1 meetings.