Cell therapy in a drip. No weeks in a factory

A drip instead of weeks in a factory

Classic CAR-T therapy (T cells with a chimeric antigen receptor) starts in a factory. The patient donates blood. For days or weeks a laboratory genetically rewires their T cells, expands them, freezes them, and ships them back. Before the infusion the patient receives chemotherapy that clears space in the immune system. The cost in Poland: according to a 2025 report on CAR-T therapies, the National Health Fund accounts a single administration at about PLN 1.3–1.4 million. The country has 15 accredited hematology centers — 13 for adults and 2 pediatric — and in 2021–2024 they treated a total of 265 patients, mostly with blood cancers.

On 3 September 2026 the New England Journal of Medicine published a correspondence that flips that scale. A team at Tongji Hospital (Huazhong University of Science and Technology) in Wuhan described 16 patients with severe neurologic autoimmune diseases. Each received a single intravenous drip of JY231: a lentiviral vector that, inside the patient’s body, teaches their own T cells to hunt the CD19 protein on B cells. No weeks in a clean room. No lymphodepleting chemotherapy in the reported protocol. The factory fits in a vein.

First contrast: weeks outside the body versus hours in circulation

Traditional CAR-T is airline logistics at the scale of one human. Leukapheresis, transport, a viral vector in a bioreactor, quality control, freezing, return to the hospital, preparatory chemotherapy, infusion, intensive monitoring. Every step needs certified staff and a bed. That is why, even as oncology indications grow, access is limited by infrastructure, not by the idea of the receptor itself.

JY231 shortens the chain to one infusion. A lentiviral vector delivers the genetic instruction to the patient’s T cells in vivo (directly inside the body). According to the clinical report and independent discussions of the paper, CAR-T cells appear in the bloodstream and peak around day 11 after infusion. Within 24 hours after dosing there is transient lymphopenia and a surge in neutrophils. These swings resolve on their own. Three patients had reversible grade 3 or 4 neutropenia; prolonged marrow suppression of the kind typical of aggressive chemical preconditioning was not seen.

The scale is tangible: instead of weeks on a laboratory calendar, one infusion and a day-counter inside the patient’s body are enough. The metaphor is physical. The lab is not behind a wall in a tech park. It circulates with the blood.

Second contrast: B cells vanish, then return “naive”

In all 16 cases JY231 rapidly and completely depleted circulating B cells. Immunoglobulins also fell transiently. After more than two months B cells began to return, but they were mostly naive, freshly produced cells. The B-cell receptor clones present before treatment were replaced. That is molecular evidence of an immune-system rebuild: wiping out old clones, not merely a temporary cut in cell counts.

Seven patients with progressive multiple sclerosis had lower EDSS scores, better motor and cognitive function, and less fatigue. At 3 and 6 months none of that group had new gadolinium-enhancing lesions or new T2 lesions. Cerebrospinal fluid biomarkers of axonal injury (neurofilament light chain) and intrathecal IgG synthesis fell; PET–MRI pointed to less microglial activation. Patients with generalized myasthenia gravis, inflammatory myopathies, and MOG antibody–associated disease also showed clinical and biological improvement. Median observation: about six months.

Here the scale jumps from one drip to the whole patient: fewer “old” B-cell clones, less inflammation in the central nervous system, measurable changes on MRI and in CSF. This is still a case series, not a randomized phase III trial. But the micro-to-macro contrast is already visible in the data, not only in presentations.

Third contrast: safety without factory-grade neurotoxicity

Classic oncology CAR-T carries the risk of cytokine release syndrome (CRS) and ICANS neurotoxicity. In the JY231 series there were no cases of ICANS or LICATS. Mild grade 1 CRS occurred in 11 of 16 patients (69%), appeared between day 4 and day 28, and resolved spontaneously within two weeks. There was no grade 2 or higher CRS. Vector capsid protein p24 fell quickly and was undetectable from day 4. Integration-site analysis showed a polyclonal vector distribution, mainly in intronic regions, without oncogene-linked clonal expansion. That is reassuring, but with a small sample the risk of secondary malignancy remains open and needs years of monitoring.

A separate caution: in the broader Chinese gene-therapy field, severe events including deaths have been reported in other indications and platforms. The neurologic JY231 series does not include them, yet that does not turn a lentiviral vector into a “gentle drip for everyone.” Safety scale is not marketing — it is the CRS grade counts, the absence of ICANS, and the duty of long observation.

Fourth contrast: 16 bodies versus the map of B-cell diseases

Sixteen patients are too few to rewrite the global standard of care for MS. They are enough to rewrite the business question. Until now, expanding CAR-T beyond blood cancers hit two walls at once: the cost of ex vivo manufacturing (outside the body) and the lack of centers ready for neurotoxicity. In vivo knocks down the first wall. The second remains and is sharper than it looks in an investor presentation.

Neuromyelitis optica in the United States affects on the order of a little over twenty thousand people; MS in Poland almost 59,000; the JY231 series spans several neurologic indications at once. Each has its own endpoints, its own background MRI noise, and its own risk of mistaking disease progression for a treatment reaction. Regulators in rare and chronic diseases demand durability over years, not only a response at month six. That is why JY231’s clinical scale is both promising and dangerously small: it shows the vector works in humans, but not yet how to build a trial EMA or FDA will accept as registrational.

For an investor and a payer the difference is simple. An ex vivo product sells as a hospital procedure worth a million. An in vivo product sells as a biologic with logistics closer to an antibody infusion, provided safety holds at hundreds of patients, not sixteen. That is a two-order-of-magnitude scale jump. Either the vector factory holds, or we return to the freezer.

Who stands behind the product and where the market goes

JY231 was developed by Shenzhen Genocury Biotech. The neurologic study was run by the clinical team at Tongji Hospital; NEJM correspondence authors include Yu-Hang Cheng and colleagues, and registries list Dai-Shi Tian among investigators. ClinicalTrials.gov lists several early JY231 protocols: in B-cell cancers (including NCT06678282, Genocury as sponsor) and in relapsed/refractory neurologic immune disorders (NCT06797024, Tongji Hospital) with dose escalation expressed in transducing units.

Commercially the stake is not another expensive, personalized product in a freezer. The stake is a ready-made vector: the same production lot and the same preparation for many patients, dosing closer to neurology than only hematology-oncology, and a potentially lower unit cost once regulation and safety allow it. Competition in in vivo CAR-T is rising on parallel platforms (including CAR instructions delivered by RNA in lipid nanoparticles): a race to combine B-cell clone rebuilding with an acceptable risk profile and manufacturing scale. JY231 is a NEJM signal, not an EMA or FDA verdict.

Poland: almost 59,000 people with MS and 15 centers built for a factory

According to the Ministry of Health algorithm, at the end of 2023 Poland had 58,878 people living with multiple sclerosis. In drug program B.29 in 2025 the payer settled benefits on the order of PLN 848–901 million; about 28,000 patients are actively treated with disease-modifying therapies. That is a European penetration level — and a system built around medicines, not around rebuilding B cells with a vector.

Poland’s CAR-T footprint is real but calibrated to oncology: 15 accredited centers, a cost of PLN 1.3–1.4 million per administration, hundreds of hematology-oncology patients. For MS, Gdańsk-based PolTREG is building a separate path: in February 2026 its GMP facility released the first transfer batch of CAR-TREG product PTG-021, and a phase I/IIa trial start in primary progressive MS and ALS is planned for late 2026. That is still an ex vivo model (Treg cells modified outside the body), but it shows that Polish advanced technologies (deep tech) are already betting on cell therapies in neuroimmunology.

On the global market a race of platforms, not only companies, is already visible. Beside Genocury’s lentiviral approach, strategies based on RNA delivered in lipid nanoparticles appear, with different CAR expression durability and a different genomic-integration risk. For Poland that means licensing and clinical-partnership windows will open in parallel. Someone will look for neurologic centers experienced in MS and myasthenia; someone else for vector manufacturers and CTIS pathways. A country with only 15 CAR-T centers for hematology-oncology can either become a target market for big players or build an early-phase research node around PolTREG, medical universities, and existing B.29 drug programs.

Here the scale returns. If in vivo CAR-T keeps its safety and effect in larger trials, Poland’s question is not “do we have laboratories.” It is whether 15 centers and URPL/CTIS pathways can move from freezer logistics to vector logistics, and whether neurology will get the same safety protocols hematology-oncology spent a decade building. The decision window is the next 24–36 months of phase II protocols and endpoint talks with regulators, before large firms lock Central European partnerships.

In practice that also means a different competence map. A hematologist-oncologist knows how to manage CRS and ICANS after classic CAR-T. An MS neurologist knows how to read MRI and the EDSS scale, but rarely has an ASCO-style protocol for cell-therapy neurotoxicity on the ward. In vivo CAR-T forces those two worlds onto the same shift: the same drip needs both languages at once. Countries that build such teams earlier collect registrational data. The rest become importers of a finished vector — or stay outside the indication altogether.

The stake

One drip does not erase the factory overnight. It erases the assumption that CAR-T must be a product tailored to one lymphocyte in a clean room. The series of 16 patients from Tongji / Genocury sets TRL around 5: proof of effect in humans, early clinical stage, without broad approval yet. The contrasts are physical: a week in the lab versus day 11 in the bloodstream, a million zlotys versus a ready-made vector for many patients, 15 centers versus a potentially wider neurologic network. Poland has patients, money in program B.29, and its own CAR-TREG track. The question is who first connects the drip to a system that so far only knew how to freeze cells.

Sources

1. Cheng Y.-H. et al., Lentiviral In Vivo CD19 CAR T-Cell Therapy in Neurologic Autoimmune Disorders, N Engl J Med. 2026 Sep 3;395(9):926-929. DOI: 10.1056/nejmc2603114
2. ClinicalTrials.gov — JY231 Injection for Relapsed/Refractory Neurologic Immune Disorders (NCT06797024); Genocury oncology protocols (including NCT06678282).
3. Gavin Giovannoni — discussion of the JY231 series (6 Sep 2026): EDSS data, CRS in 11/16, no ICANS, CAR-T peak around day 11. gavingiovannoni.substack.com
4. Polish Ministry of Health — MS prevalence in Poland (end of 2023: 58,878). analizy.mz.gov.pl / ZPA SM
5. Report CAR-T Therapies in Poland (2025) — 15 centers, 265 patients 2021–2024, NFZ cost about PLN 1.3–1.4 million per administration.
6. Alert Medyczny / PolTREG — first GMP batch of PTG-021 (12 Feb 2026), planned I/IIa start late 2026.
7. Shenzhen Genocury Biotech — originator of JY231 (ClinicalTrials.gov registries).