Putting a cost model behind Treg manufacturing

Esmond Lee, Ph.D., senior scientist at the A*STAR Bioprocessing Technology Institute, a Singapore-based research institute, set out the work in an interview at Sartorius's Process Intensification Forum in Hyderabad, India held in June, where he presented on design of experiments as a decision tool for cell therapy manufacturing.

One patient, one batch

Regulatory T-cells (Tregs) are the immune system's brakes, stopping the body attacking material it should tolerate. As an example, Lee pointed to pregnancy, where a fetus carrying the father's human leukocyte antigen (HLA) markers – the surface proteins the immune system reads to tell self from foreign – goes nine months without being rejected. Redirecting a patient's own Tregs to a site of inflammation, a rheumatoid joint for instance, would dampen the response there. “We're really using the cells as a living therapeutic,” he said.

Such a therapeutic is autologous, and thus “in terms of production, we're thinking less of scale up at this point, but more scale out,” Lee said. The harder problem is that the starting material changes with every patient. A cell line behaves the same way each run, “but when you are manufacturing for [individual] patients, there's biological variability,” he said, with some batches failing outright. Instead, Lee is looking to develop a process flexible enough to absorb that variation while still running the same way every time, whatever the donor material does, to deliver something that works in vivo.

The design-of-experiments work sits on a CRISPR-Cas9 process that introduces a corrected copy of FOXP3 into patient T cells and stem cells to fix the mutation behind IPEX syndrome, a multisystem autoimmune disease, with the aim of reconstituting a functioning tolerance mechanism once the cells are engrafted. The FOXP3 work took the 2025 Nobel Prize in Physiology or Medicine, shared by Mary Brunkow, Fred Ramsdell, and Shimon Sakaguchi for their discoveries on peripheral immune tolerance.

Fitting a cost model to the data

Several inputs drive whether that gene editing produces a usable product, including the quantity of virus and the concentration of guide RNA, the sequence that directs the Cas9 enzyme to the right point in the genome. “We use design of experiments as a tool to help us to find a sweet spot for manufacturing,” Lee said, with the experimental data fitted to a cost model to establish what a clinical dose will actually cost to make. “We mainly included inputs, raw material inputs. We actually didn't include CAPEX or labor costs and all the other things associated with marketing and delivering a product from vein to vein.”

There is little public data focused on costs, especially for Tregs. The standard academic model for autologous CAR-T was built from public information rather than company data, as manufacturers rarely disclose their cost of goods. However, a point-of-care trial in India is one of the few to put a figure on it, at $35,107 per patient to manufacture.

Cost, of course, governs access, and India (where the forum took place) has moved faster than most. Lee named ImmunoACT and Immuneel as companies working on the problem, with the former’s NexCAR19 cleared by India's Central Drugs Standard Control Organization in October 2023. The blood cancer treatement is priced at roughly ₹30 to 40 lakh, or about $50,000, against CAR-T pricing approaching $500,000 in the U.S. and Europe.

Manufacturing is a major, but not the only, factor in the price. Singapore subsidizes CAR-T through the public system for patients who could not otherwise reach it, Lee said, and outcome-linked reimbursement is gaining traction. “You pay only if the therapy is working for you.”

Off-the-shelf

Process intensification, the subject of the forum Lee was speaking at, is largely about extracting more product from a smaller footprint, and is typically spoken about monoclonal antibodies, where one batch serves an entire market.

For personalized medicines like Tregs, intensified processes could be adopted if the immune rejection problem could be engineered out, which Lee suggested might come through modifying receptors on the cell surface. That would make an off-the-shelf – or allogeneic – product, “producing higher cell numbers at a cheaper price, and that one batch can then be used for multiple patients.”

His group is, meanwhile, working on what happens to Tregs once they leave the body. Cells change state in culture, so the team is using transcriptomics and metabolomics to work out which pathways matter and what the cells need to be fed, then building an expansion process around that.

Much of the current method is inherited, borrowed from the CAR-T space, which has a decade’s head start. “The regulatory T cell space is relatively nascent,” Lee said, and the stimulation regimens, media, and IL-2 that Tregs are currently grown on were worked out for a cell type with different metabolic requirements.