The Third Arm: Why Some Programmes Are Worth the Extra Engineering Complexity
How powerful AI platforms are enabling a new generation of therapeutic bi- and tri-specifics.
This year, 2026, marks 40 years since the first monoclonal antibody approval and since then the field of immunotherapy has gone from strength to strength. In the 12 years since the first bispecific approval, multi-specific biologics have already changed how the industry thinks about drug design.
Here we’ll explore the potential of trispecifics, identify the pros and cons versus bispecifics and discuss how AI such as Etcembly’s AI immune engineering platform, EMLy™, is set to transform their development. Bispecific Immunotherapy: a proven modality
Bispecific antibodies and T cell engagers have proven hugely successful; they are now an established modality, with 15 approved therapies in 12 years (Table 1) and hundreds more in the clinic.
Name | Active Ingredient | Targets | Year Approved | Indication |
blinatumomab | CD19 x CD3 | 2014 | R/R Philadelphia chromosome-negative B cell precursor ALL | |
emicizumab | FIXa x FX | 2017 | Prevention/ reduction in the frequency of bleeding episodes in hemophilia A with factor VIII inhibitors | |
amivantamab | EGFR x c-MEt | 2021 | Locally advanced or metastatic NSCLC with certain mutations | |
tebentafusp | gp100 x CD3 | 2022 | Unresectable or metastatic uveal melanoma | |
faricimab | VEGF-A x Ang-2 | 2022 | Neovascular (wet) age-related macular degeneration and diabetic macular oedema | |
teclistamab | BCMA x CD3 | 2022 | R/R multiple myeloma | |
mosunetuzumab | CD20 x CD3 | 2022 | R/R follicular lymphoma | |
epcoritamab | CD20 x CD3 | 2023 | R/R diffuse large B cell lymphoma | |
glofitamab | CD20 x CD3 | 2023 | R/R diffuse or large B cell lymphoma | |
talquetamab | GPRC5D x CD3 | 2023 | R/R multiple myeloma | |
elranatamab | BCMA x CD3 | 2023 | R/R multiple myeloma | |
tarlatamab | DLL3 x CD3 | 2024 | R/R SCLC | |
zanidatamab | HER2 x HER2 | 2024 | HER2 positive biliary tract cancer | |
zenocutuzumab | HER2 x HER3 | 2024 | NRG1 fusion positive NSCLC | |
linvoseltamab | BCMA x CD3 | 2025 | R/R multiple myeloma |
Table 1: FDA approved bispecific therapeutics (September 2026)
But what are bispecifics - and what challenges do they face?
Instead of relying on a single target to do all the therapeutic work, bispecific antibodies and T cell engagers enable one molecule to do two jobs at once; one arm of the bispecific binds a disease target, typically tumour cells, and the other arm, recruits an immune effector cell, usually a T cell. This process guides the patient’s immune system to the right place, enabling T cell activation and tumour cell killing.
However, bispecific antibodies and T cell engagers face their own biological challenges.
Tumour targeting affinity: Bispecific antibodies typically have low affinity (tightness of binding) for their targets. Kimmtrak is the first, and so far only, approved bispecific T cell engager with ultra-high affinity for its target.
Intracellular targets: This is particularly pertinent for bispecific antibodies as they can only target cell surface antigens - a limitation addressed by TCRs.
T cell exhaustion and tumour escape: Key limitations for long-term efficacy and a problem that has not yet been solved (Figure 1).
Figure 1: Drivers of T cell exhaustion following immunotherapy. Repeated or chronic antigen engagement drives T cells toward a progressively dysfunctional state, marked by upregulation of inhibitory receptors (e.g. PD-1, LAG-3, TIM-3), loss of cytotoxic and proliferative capacity, and metabolic and epigenetic reprogramming. Tumours further compound this through antigen loss/heterogeneity and an immunosuppressive microenvironment, together enabling immune escape and disease relapse.
What are trispecifics - and why are they so exciting?
While a bispecific typically either bridges two cell types or co-engages two targets on the same cell, as the name suggests, trispecifics are highly engineered proteins that can simultaneously bind to three different antigens.
The Trispecific Trio
|
Engaging multiple elements of the immune response has a synergistic effect, bringing in more components of the immune system and activating a more powerful therapeutic response.
Deeper immune engagement
The immunological synapse (IS) is a natural phenomenon where immune cells recognise tumour cells and form a specialised cell-cell junction. This structure enables highly focused signalling and directed delivery of cytotoxic molecules, allowing immune cells to kill tumour cells while minimising damage to surrounding healthy tissue. The signalling molecules that comprise the IS include a target-recognition signal and a co-stimulatory signal; these are the elements that a trispecific can deliver. Using both a tumour target arm and a co-stimulatory arm alongside a CD3 effector arm to drive a stronger, more durable T cell response than target-plus-CD3 engagement alone.
Reduce toxicity
In another scenario where creating a trispecific that targets two tumour markers in addition to the immune engager, there is the added advantage of improved specificity and selectivity. Many attractive tumour antigens are expressed on healthy tissue leading to a major source of on-target, off-tumour toxicity, particularly for bispecifics. A trispecific can be engineered so that potent effector engagement only occurs when two tumour markers are co-expressed on the same cell, reducing unwanted toxicity.
Taken together, the multi-functional nature of trispecific immune engagers – targeting cancer cells, modulating the tumour microenvironment, and activating the immune response -, may reduce the need for multiple drugs and associated toxicities, a meaningful benefit for patients.
Trispecifics on trial: Early data validates the three-way approach
Although no trispecific therapeutic has been approved yet there is an impressive crop of early stage clinical results validating the idea behind trispecifics.
Early signals from the clinic 100% response rate - J&J's JNJ-5322 (BCMA x GPRC5D x CD3), first-in-human data (2025) in a small group of patients with relapsed or refractory multiple myeloma. 79% sustained overall response rate - Ichnos' ISB 2001 (BCMA x CD38 x CD3), phase 1 TRIgnite-1 study in heavily pretreated relapsed or refractory multiple myeloma patients. |
These aren’t incremental gains. They’re potential game-changers.
The case is clear: using trispecific approaches to bring more components of the immune system into play is set to drive a quantum leap in efficacy, especially if the recent advances in blood cancers can be expanded into solid tumours.
But there are a number of technical hurdles still standing in the way.
Challenges in designing and engineering trispecific therapeutics
Designing a drug capable of simultaneously binding three different targets without interference is far from simple.
Five hurdles standing between concept and clinic Target selection: The three chosen antigens should have a clear biological rationale, both from the perspective of tumour targeting and immune activation. Binding affinity: The relative binding affinities of the component antibodies need careful balancing to ensure optimal efficacy. Optimising geometry: The three antibodies must be arranged to allow effective target engagement and immune activation without interference. Linker selection: The structure and flexibility of the linkers connecting different binding domains affects protein folding, function, and stability. Manufacturability: Trispecifics are larger and more complex than bispecifics, increasing the risk of problems with expression, purification and stability. |
Addressing all these issues at the same time within one molecule poses a significant problem for protein engineers. Fortunately, solving such complex, multidimensional challenges is exactly where machine learning and AI tools excel. Modelling binding geometry, arm orientation, and interface compatibility computationally, the same principle behind Etcembly's approach EMLy™, is what makes trispecific programmes tractable rather than a search through an unmanageable design space.
Using EMLy™ AI to solve the challenge of trispecific antibody and TCR development
By using predictive structural modeling and generative capabilities to simultaneously optimise multiple parameters, AI-driven design can rapidly accelerate the development of novel trispecifics for a wide range of diseases.
Our powerful immune engineering platform, EMLy™, is demonstrating what’s possible at the cutting edge where AI meets biology. Bringing together billions of unique T cell receptor (TCR) and antibody sequences from public and proprietary sources – including long term cancer survivors – with cutting-edge bioinformatics, generative AI and structural modeling, EMLy™ allows us to discover, design and optimise highly effective novel trispecific therapeutics with unprecedented speed.
Proof-of-concept: Etcembly’s PRAME trispecific immune engager
Combined with the computing power of EMLy™, our plug-and-play modular trispecific ETCer (Etcembly’s T cell Engager) platform offers superior efficacy and engineerability within a three-way molecule.
Our first proof-of-concept trispecific molecule builds upon our ETC-101 PRAME programme and brings together a picomolar affinity TCR against the validated cancer antigen PRAME with an optimised CD3 receptor and third ligand that activates a co-stimulatory pathway.
In head-to-head tests we’ve shown that this trispecific ETCer is more potent against PRAME-positive cancer cells than our original bispecific version, while retaining high specificity for the target antigen and minimal off-target cross-reactivity. |
We also proved that our trispecific ETCer overcomes T cell exhaustion and dysfunction and maintains potency over repeated rounds of dosing compared with our original bispecific (Figure 2).
Figure 2: Etcembly's trispecific vs. bispecific ETC-101. Comparative in vitro potency and durability of the ETC-101 PRAME-targeting trispecific ETCer against the parent bispecific across repeated rounds of tumour cell challenge. The trispecific format maintains cytotoxic potency and T cell functionality over successive dosing rounds, while the bispecific shows progressive loss of activity consistent with T cell exhaustion.
This data shows how our trispecific platform can deliver improved efficacy and safety, and address the durability challenges that limit current immunotherapies.
Trispecific vs Bispecific: A conceptual comparison
Bispecific | Trispecific | |
Binding domains | 2 | 3 |
Typical use case | Bridge two cell types, or co-engage two targets on one cell | Add selectivity logic, redundancy, or a third signal (e.g., co-stimulation) |
Design space | Manageable; established formats exist | Substantially larger; fewer validated formats to build on |
Manufacturability risk | Moderate; well-characterised chain-pairing solutions available | Higher; more pairing combinations and developability failure modes |
Best suited to | Targets with a clear, uniform disease marker | Heterogeneous or escape-prone disease biology; on-target/off-tumour toxicity risk; multi-signal potency needs |
Computational design value | Useful for optimising affinity and specificity | Close to essential - geometry and format screening in silico is what keeps the programme feasible |
Table 2: Trispecific vs bispecific overall comparison
Forging the way ahead for trispecific therapeutics
We're entering an era where we are witnessing the convergence of multiple breakthroughs as the intersection of AI and immunology creates innovations that seemed impossible just a few years ago. EMLy™ brings a significant leap forward in our ability to discover and optimise the trispecifics of the future, opening up new possibilities for future cures for a wide range of different diseases, from cancer to autoimmune conditions to chronic infections.
We can look forward to a time when:
Multi-target engagement becomes the standard therapeutic approach.
Rapid individual immune profiling enables the creation of personalised immunotherapy.
Advances in manufacturing scalability deliver cures for as many people as possible, not just a lucky few.
If you're weighing up whether your programme's biology calls for a bispecific or a trispecific approach, we'd be glad to talk with you and hear your thoughts on the technical hurdles and collaboration opportunities that could accelerate clinical translation. Get in touch at hello@etcembly.io, or book a consultation to learn more about how EMLy can support multi-specific engineering programmes.
Comments