The vector is the medicine.

Banyan Therapeutics is a preclinical biotechnology company engineering lentiviral vectors for in vivo genetic medicine. Oncology, autoimmune disease and rare disease.

The case for in vivo generation is made. The engineering is the unsolved part. Reaching the right cell, making it behave predictably, surviving contact with the immune system without provoking it, and what the dose actually contains are four separate problems, and a product is only as good as the weakest of them. We work on all four as one system.

Enquiries Our approach

The constraint

Autologous manufacturing is the ceiling, not the biology.

Ex vivo CAR-T is among the most effective interventions in modern oncology, and its reach is limited by its supply chain rather than its mechanism. Apheresis, a slot in a qualified manufacturing facility, release testing, a vein-to-vein interval measured in weeks, bridging therapy for patients who cannot wait, and lymphodepleting conditioning before infusion. Each step is a point of attrition. A substantial share of patients who are eligible on paper are never dosed.

That structure has held the modality inside oncology, and mostly inside academic and specialist centres. It is also what has kept it out of indications where the benefit-risk calculation cannot absorb conditioning and complexity.

Generating CAR-T in the patient removes the manufacturing step rather than optimising it. It replaces a per-patient biological product with an administered one. The field broadly agrees on this. What separates approaches is whether the delivery vehicle is precise enough to make it a therapeutic product.

The same constraint applies wherever a genetic medicine is made per patient. CAR-T is where the clinical argument is furthest advanced. It is not where it ends.

One consequence is easy to miss. When a CAR-T is manufactured outside the body, the vector is a starting material — an input to someone else’s process, held to the standards that apply to reagents. When the CAR is generated inside the patient, the vector is the drug product: the entity administered, the entity that carries a specification and a stability profile, the entity whose distribution in the body is pharmacokinetics, the entity a regulator assesses as the active substance.

That changes what it has to be. It is a pharmaceutical product, and has to be engineered as one.

Ex vivo:weeks to dose, per-patient manufacture, conditioning required
In vivo:administered dose, no apheresis, conditioning-sparing intent
The hard part:delivering to the intended cell, and only that cell

Approach

Four problems that have to be solved together.

Our work begins where the biology leaves off. The receptors and transgenes that drive clinical responses are, in the areas we work in, well characterised and widely understood. What is unsolved is the engineering: getting a vector to a defined cell population in a living patient, making the payload behave once it is there, surviving an immune system built to clear it, and controlling what else the patient receives with it. Each has a literature of its own. None of them is sufficient alone.

Receptor-directed entry

VSV-G pseudotypes transduce almost anything, which is an advantage in a manufacturing suite and a liability in a patient. Administered systemically, a promiscuous vector distributes across tissues, and the dose required to reach the intended cell population carries everything that follows from transducing the rest.

We work on entry that depends on a defined marker present on the target cell rather than on a ubiquitous entry receptor. The aim is a therapeutic effect at a lower total administered dose. That is the lever that makes the safety arithmetic work in a systemically dosed genetic medicine: fewer particles administered, fewer cells transduced outside the intended compartment, and a smaller modification footprint overall.

Dose is also a supply constraint. A vector that is effective at a lower dose is a vector that can be made in the quantities a real patient population needs.

Controlled expression

Ex vivo, the drug product is the cell, and its critical quality attributes are established before it is returned to the patient: vector copy number, transduction efficiency, potency, identity, all tested against a specification before release. In vivo, that product is generated after dosing. Some of those attributes migrate onto the vector specification. The rest cease to be release attributes at all, and have to be controlled by design.

What we design for is expression that holds, confined to the intended cells, at the lowest vector copy number that delivers function. Payloads differ in all three, and the differences are often what separates a construct that expresses from one that treats.

Immune modulation

Human serum complement inactivates commonly used pseudotypes. Pre-existing antibodies against vector surface proteins restrict the population that can be dosed at all. Opsonisation accelerates clearance before a particle reaches its target. These are properties of the vector particle, and they are addressable in how it is designed.

The requirement runs in two directions, and both are bounded. Survival is a matter of time rather than invisibility — a vector has to persist long enough to reach the intended cells. Provocation is a matter of threshold — it has to get there without triggering a response that becomes the safety problem in its own right. The early clinical signal in systemically dosed in vivo programmes has been inflammatory, which makes the innate response a design parameter rather than something managed clinically after the event.

Product quality

A dose is everything the process puts in the vial: functional and non-functional particles, aggregates, residual nucleic acid and host-cell protein, and whatever the formulation carries. The patient receives all of it.

That makes product quality a determinant of the other three problems rather than a step after them. Non-functional particles raise the immune burden at the same functional dose. The ratio between what is administered and what works is set in the process, not recovered downstream.

We treat it as a design parameter from the start, because for a vector that is the drug product there is no later step at which quality gets added.

Each of the four is tractable alone. Solving one can cost another. A targeting modification can alter complement sensitivity; narrowing tropism can cost titre; designing for predictable expression can cost payload capacity. We treat the combination as the unit of work.

Treating the combination as one unit only works if its parts can move independently. A vector that has to be rebuilt whenever one element changes is a result, not a platform. We build for the ability to vary one element with the rest held constant — because the first version of a complex system is never the good one, and what matters is whether the second version is a design exercise or a new start.

Pipeline

Three areas, selected on delivery risk.

Indication selection is a platform decision before it is a clinical one. Where the target biology is already clinically validated, the variable under test is delivery rather than mechanism, which is the fastest way to establish that the platform works. Across these three areas the vector does two different jobs — engineering an effector cell that acts on disease, and delivering to the affected tissue directly — and the delivery problem is common to both, which is what makes them one platform rather than three unrelated bets.

Area	Rationale
Oncology	CD19 and BCMA are where the market and the clinical comparators sit, which makes them the cleanest place to establish that delivery works. They are not where the opportunity ends. The vector’s problem is the same whichever malignancy the resulting cell is directed at, because the vector targets the effector cell and not the tumour, and in vivo oncology is an open field rather than a settled one.
Autoimmune disease	Published case series in refractory lupus, inflammatory myopathy and systemic sclerosis report durable drug-free remission after B cell-directed CAR-T. The barrier has been conditioning and complexity, which is exactly what in vivo administration addresses.
Rare disease	Defined genetic cause, no disease-modifying option, and populations too small to support per-patient manufacture. Durable, predictable expression matters most here, where a single administration has to hold for years.

Specific targets and development candidates are undisclosed. Programmes are preclinical; we will report data when it is worth reporting.

Company

Who we are.

Banyan Therapeutics is a privately held biotechnology company incorporated in Singapore in 2025. We are a therapeutics developer. The platform is the vector and the engineering that makes it a medicine. It exists to make our own programmes possible, and is built to the standard a drug product requires rather than the standard a research reagent requires. Platform and pipeline are therefore inseparable here.

The founding team combines decades of work across viral vector engineering, advanced therapy development, regulatory practice in cell and gene therapy, and building and running listed and private biotechnology companies in Europe, the United States and Asia.

Singapore is a deliberate base. It offers clinical trial infrastructure, translational research capacity and public research partnership, under a regulator that has engaged seriously and early with cell and gene therapy products.

Founded 2025
Singapore
Preclinical
Privately held

Enquiries.

We are open to conversations with investors, potential collaborators and partners, and scientists who want to work on delivery.

Banyan Therapeutics Pte. Ltd.
540 Sims Avenue, Sims Avenue Centre, #03-05
Singapore 387603