Following hundreds of billions of dollars in bets on the in vivo CAR-T track by multinational pharmaceutical companies, 2026 has brought the first concentrated wave of early clinical validation for this field.

Multiple companies have successively disclosed preliminary clinical data: from lentiviral vectors to LNP, from single-target to dual-target, companies including Eli Lilly's Kelonia, Legend Biotech, WestGene Biopharma, and Weitao Biotech have previously presented clinical data on their respective in vivo CAR-T therapies at academic conferences.

However, it is worth noting that these advances have mostly concentrated in hematologic malignancy indications. For example, Kelonia's KLN-1010 targets relapsed/refractory multiple myeloma, and Legend Biotech's LB2501 targets non-Hodgkin lymphoma.

On September 2, a study newly published in the Correspondence section of the New England Journal of Medicine (NEJM) opened a new door for the imaginative possibilities of in vivo CAR-T.

This is a study conducted through collaboration between the team at Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, and Shenzhen Jiyin Biotechnology Co., Ltd., on a lentiviral vector-based in vivo CD19 CAR-T cell therapy (JY231) for refractory neurological autoimmune diseases — the first validation of the feasibility of the in vivo CAR-T approach in neuroimmunological diseases.

The study enrolled 16 adult patients with refractory neurological autoimmune diseases. All patients showed in vivo CAR-T cell expansion and achieved rapid, complete B-cell depletion; at a median follow-up of 6 months, clinical and biological improvement signals were observed across all disease cohorts.

This result quickly drew attention from the cell therapy industry. A founder of a cell therapy company commented, "These data are of great significance in the neuroscience field. For neurological diseases, especially CNS-involved diseases like MS, existing drugs are very limited. Antibody drugs require repeated administration and may lose efficacy, whereas JY231 is a one-time administration, relatively simple."

TXY immediately contacted and interviewed the corresponding author — Professor Tashiro Minoru of Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology — to interpret the study results and data, and to conduct an in-depth discussion on topics including the IIT pathway, safety monitoring, ethical review, risk-benefit ratio, and regulatory changes.

In the interview, Professor Tashiro pointed out that this study demonstrates the technical feasibility, controllable safety signals, and preliminary biological activity of in vivo CAR-T, but that the insertion sites, clonal expansion, and long-term oncogenic risks associated with lentiviral integration still require years of monitoring.

Furthermore, in his view, IIT is an important "engine" for original clinical discovery, but it is by no means a channel for lowering standards or bypassing regulation, and should be connected sequentially with IND; high-risk research must rely on dynamic safety monitoring, continuous ethical review, and rigorous risk-benefit ratios.

Professor Tashiro Minoru
Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology

Chapter 1: Reading In Vivo CAR-T Through the JY231 Data

TXY: You and your team have just published in the New England Journal of Medicine the IIT results of a lentiviral vector-based in vivo CD19 CAR-T cell therapy (JY231) for refractory neurological autoimmune diseases. The patient cohort was 16 cases. Could you first walk us through this dataset specifically?

Tashiro Minoru: This study was an investigator-initiated, single-arm, Phase I clinical study that enrolled 16 patients with refractory neurological autoimmune diseases, including 7 with progressive multiple sclerosis, 3 with MOG antibody-associated disease, 3 with generalized myasthenia gravis, and 3 with idiopathic inflammatory myopathy.

Unlike traditional CAR-T therapy, JY231 does not require collecting the patient's T cells, nor does it require weeks of ex vivo cell manufacturing. We administered a single intravenous infusion of 2.0×10⁹ transducing units of the targeted lentiviral vector to generate CD19 CAR-T cells directly in the patient's body, and without using traditional lymphodepleting chemotherapy.

From the pharmacokinetic and pharmacodynamic results, CAR-T cell expansion was detected in all 16 patients, with a median peak occurring on day 11 after infusion and a median peak vector copy number of 53,962.5 copies/μg DNA. By day 4 after infusion, the viral capsid protein p24 in peripheral blood had dropped to undetectable levels; more than 99% of CAR-positive cells were confirmed to be T cells, indicating high cellular selectivity of the in vivo targeting.

More notably, CAR-T cells not only appeared in peripheral blood and bone marrow but were also able to enter the cerebrospinal fluid. After treatment, peripheral blood B cells were rapidly and deeply depleted; B cells that reappeared after more than two months were predominantly naive B cells, and baseline B-cell receptor clones were replaced by new clones. This provides molecular-level evidence for "systemic immune reset."

At a median follow-up of 6 months, varying degrees of clinical and biological improvement were observed across all four disease cohorts. In the 7 patients with progressive multiple sclerosis, no new gadolinium-enhancing lesions or T2 lesions appeared at the 3-month or 6-month follow-up, while some patients showed improvements in motor, cognitive, and fatigue measures, along with reductions in cerebrospinal fluid neuroaxonal injury and intrathecal immune activation markers. Patients with myasthenia gravis, inflammatory myopathy, and MOG antibody-associated disease also showed improvements in disease scores, muscle strength, antibody, or imaging measures.

But it must be emphasized that this is an early, single-arm study of 16 patients. What it first demonstrates is technical feasibility, controllable safety signals, and preliminary biological activity — it cannot replace larger-sample, long-term follow-up, and randomized controlled studies.

TXY: Previously, the industry generally believed that lentiviral vectors (LVVs) were less safe than the LNP (lipid nanoparticle) platform — the data from your study form a direct contrast to this. JY231 demonstrated controllable safety without special preconditioning, and some CGT industry professionals have called this "quite astonishing and very much worth watching." How do you view the pros and cons of these two different pathways?

Tashiro Minoru: I believe we cannot simply say that lentiviral vectors are necessarily "less safe" than LNP, nor can we use a single small-sample study to prove in reverse that lentiviral vectors are necessarily safer. The two platforms have different risk-benefit characteristics.

The mRNA-LNP platform is mainly characterized by transient expression and typically no genomic integration, thus offering a theoretical advantage in terms of long-term genetic risk; however, CAR expression duration is shorter, some applications may require repeated dosing, and there is also a need for continuous evaluation of LNP tissue distribution, repeated-dose immune responses, and non-target cell uptake.

Lentiviral vectors are characterized by the ability to achieve relatively stable CAR expression through a single administration, which is conducive to producing sustained B-cell depletion effects. But because of genomic integration, special attention must be paid to insertion sites, clonal expansion, and potential oncogenic risk. Therefore, it is not "risk-free" — the risk types differ, and so do the monitoring methods.

Within the limited follow-up period of this study, JY231 demonstrated relatively controllable safety: no ICANS was observed, nor was local immune effector cell-associated toxicity syndrome, or LICATS, observed; 11 patients developed grade 1 cytokine release syndrome, all transient reactions that resolved within two weeks; 3 patients developed grade 3–4 neutropenia, all of which fully recovered, with no persistent iatrogenic myelosuppression.

Integration site analysis showed that CAR-T cells are currently polyclonal, with integration sites mainly located in intronic regions, and no oncogene-related dominant clonal expansion was found. These results can reduce our concerns about near-term insertional mutagenesis, but cannot eliminate long-term risk. For integrating vectors, continuous multi-year monitoring of vector copy number, integration sites, clonal evolution, hematologic abnormalities, and secondary malignancies is an indispensable safety responsibility.

Chapter 2: The Value and Boundaries of IIT

TXY: This study was conducted through the IIT pathway. A large amount of early clinical exploration in China's CGT field relies on IIT. How do you view the value of IIT itself? As an investigator-led early exploration mechanism, where does its irreplaceability lie? What should its relationship with the IND pathway be?

Tashiro Minoru: The most important value of IIT is answering the earliest, most critical questions in clinical practice that company-sponsored registration trials typically have not yet had time to answer: Can this new technology be realized in humans? Is the initial dose reasonable? What risks may arise? Which patients may benefit? What clinical endpoints and biomarkers should be used? In addition, possible mechanism exploration can also be completed during the IIT stage.

Neurological autoimmune diseases often have relatively limited patient numbers, significant disease heterogeneity, and different evaluation metrics across different diseases. Investigators who have long faced these patients have direct knowledge of the natural history of the disease, the limitations of existing treatments, and unmet needs. Therefore, IIT can rapidly connect clinical problems, mechanistic research, and technological innovation — it is an important "engine" for original clinical discovery.

But IIT is by no means a trial that lowers standards, and even less can it become a channel to bypass drug regulation. For technologies with clear product attributes that are preparing for scaled development, entry into the IND pathway is still required, through standardized pharmaceutical and manufacturing quality control, dose exploration, multi-center studies, and confirmatory trials, to ultimately evaluate whether they can be approved for marketing.

I prefer to understand the two as a sequentially connected relationship: IIT is responsible for discovering direction, identifying risks, and establishing proof of concept; IND is responsible for product finalization, systematic validation, and registration translation. High-quality IIT can improve IND development efficiency, but cannot replace IND.

TXY: Could you discuss, from a researcher's perspective, how IIT ensures safety monitoring and subject rights?

Tashiro Minoru: Whether IIT or IND, as long as CAR-T is involved, it must be managed as high-risk research.

The safety system should include at least the following elements: strict inclusion/exclusion criteria and baseline infection screening; sequential enrollment starting from low doses or a limited number of participants; pre-specified monitoring indicators for CRS, ICANS, IEC-HS, infection, and hematologic toxicity; clear criteria for pausing dosing, pausing enrollment, and terminating the study; establishment of a rescue mechanism involving neurology, hematology, critical care medicine, infectious disease, and laboratory teams; timely reporting of serious adverse events; and long-term follow-up and mechanisms for subject injury treatment and compensation.

The Regulations on the Administration of Clinical Research and Clinical Translation and Application of Biomedical New Technologies further clarify: once a serious adverse reaction occurs, the research institution should suspend the study, and the ethics committee should reassess whether it can continue; when uncontrollable risks or major safety or efficacy issues arise, the study should be terminated and reported.

Truly effective safety monitoring is not "completing a one-time approval," but a dynamic process. Once new adverse events emerge, the risk-benefit ratio should be re-evaluated, and if necessary, the protocol modified, doses reduced, observation intervals extended, enrollment populations adjusted, or the study directly suspended.

TXY: The patient population in IIT often consists of those with refractory diseases with limited existing treatment options — there is both urgent need and higher risk. At the trial design stage, how is the risk-benefit ratio typically assessed and managed? What are the key considerations?

Tashiro Minoru: "Refractory" does not mean safety standards can be relaxed. Precisely because these patients have limited treatment options and a strong desire for treatment, it is even more necessary to prevent research hope from being mistaken for confirmed efficacy.

At the trial design stage, we first assess the severity, progression rate, and risk of irreversible damage of the disease itself, while determining whether the patient has already received adequate, standardized treatment. Only when the expected benefit of continuing conventional treatment is limited, and the investigational therapy has sufficient preclinical rationale and clear biological plausibility, can an acceptable risk-benefit ratio potentially be formed.

Second, we assess the patient's own risk tolerance, including age, infection status, cardiopulmonary function, liver and kidney function, bone marrow reserve, prior immunosuppressive therapy, and disease target cell burden. For technologies that may cause strong immune activation, we also comprehensively assess CAR-T expansion kinetics, B-cell burden, inflammatory baseline, and organ reserve.

Third, we must turn risks into risks that are "identifiable, monitorable, and intervenable" as much as possible. The protocol must clearly specify the dose rationale, sequential enrollment, observation windows, discontinuation criteria, and emergency management pathways. For risks that cannot be effectively monitored, cannot be promptly treated, or lack sufficient prior evidence, even if patients have a strong treatment desire, the study should not be hastily initiated.

Finally, informed consent must clearly state that this is an exploratory study, that potential benefits are uncertain, and that unknown or even life-threatening adverse reactions may occur. Whether patients participate should be based on full understanding and autonomous decision-making.

TXY: Ethical review is a key component of the IIT system. Could you introduce the current ethical review procedures? In practice, how is the independence and professionalism of the review ensured?

Tashiro Minoru: Specific operational procedures may differ slightly across institutions, but the basic principles are consistent. Before study initiation, complete research protocols, investigator brochures or technical materials, preclinical safety and efficacy evidence, risk management plans, recruitment materials, informed consent forms, and subject compensation and injury treatment arrangements must be submitted, and the study must undergo scientific review and ethical review in sequence. Projects involving biomedical new technologies also require filing in accordance with regulations.

The ethics committee focuses not on whether the technology is "advanced," but on whether the research is worth asking subjects to bear the corresponding risks, including whether the research question has scientific value, whether prior evidence is sufficient, whether inclusion/exclusion criteria are fair, whether risks are minimized, whether informed consent is genuine and adequate, whether vulnerable patients receive additional protection, and how research-related injuries are treated and compensated.

Independence is primarily ensured through institutional arrangements: ethics committees should have members with medical, pharmaceutical, ethical, legal, and non-medical backgrounds; members with direct conflicts of interest in the project should recuse themselves; review opinions should be formally recorded; and ethics committees should not only have the power to approve but also the power to require modification, suspend, and terminate research.

Ethical review is also not "a one-time decision that lasts forever." Protocol modifications, newly discovered risks, serious adverse events, annual progress, and long-term follow-up all require ongoing review. Investigators cannot unilaterally change doses, expand populations, or weaken monitoring simply because the project has already started.

Chapter 3: The New Regulatory Landscape Under Order No. 818

TXY: In May of this year, State Council Order No. 818 officially came into effect, establishing a dedicated regulatory framework for clinical research and translational application of biomedical new technologies. Clinical research operates under a filing system, translational application under an approval system, and the entry threshold is limited to top-tier grade-A tertiary hospitals. From your perspective as a researcher, what impact will this system have on the long-term development of IIT?

Tashiro Minoru: I believe the core value of Order No. 818 is that, for the first time at the level of administrative regulations, it establishes a complete institutional framework covering "nonclinical research — clinical research — translational application" for biomedical new technologies, giving innovation a clear entry point, safety a rigid bottom line, and translation a legal pathway.

First, it raises the institutional entry threshold. The regulation requires that institutions conducting such clinical research must be grade-A tertiary medical institutions with qualified academic committees, ethics committees, quality management systems, professional personnel, rescue capabilities, and stable funding. This will push high-risk projects toward centers that genuinely possess comprehensive capabilities, rather than simply pursuing the number of projects.

Second, it strengthens whole-process responsibility. Before research, there must be sufficient nonclinical evidence and completion of academic and ethical review and filing; during research, complete original records must be preserved, regular reports submitted, and risks monitored; when serious adverse reactions occur, the study must be suspended and re-evaluated. This helps transform IIT from an "investigator's personal project" into a systematic endeavor in which the hospital bears primary responsibility.

Third, it opens a compliant translation pathway, but does not mean hospitals can directly sell products from IIT to patients. The regulation explicitly states that during the clinical research stage, subjects must not be charged fees related to the research; for research results to enter clinical translation and application, dedicated approval is still required. If a technology is defined as a drug or medical device, it should enter the IND and registration pathway in accordance with the drug or medical device regulatory system.

From a long-term perspective, this system may reduce low-level, repetitive, and quality-control-deficient projects, but will give genuine IIT with originality, clinical value, and a complete risk control system more stable room for development. For researchers, future competition is not just about "who does it first," but about "who can provide verifiable, traceable, reproducible evidence and bear long-term responsibility."