The in vivo revolution in CAR-T therapy medicinal products: challenges and regulatory prospects

Abstract

Chimeric antigen receptor T-cell (CAR-T) therapy, a groundbreaking advancement in cancer immunotherapy, has demonstrated remarkable efficacy in treating hematological malignancies and autoimmune diseases. However, conventional ex vivo CAR-T therapy medicinal products face multiple limitations, including complex manufacturing processes, high production costs, and challenges in quality control and risk management. In recent years, the emergence of in vivo CAR-T therapy medicinal products using viral vector or lipid nanoparticle platforms has provided a promising new direction by simplifying manufacturing, enhancing scalability, lowering costs, and increasing accessibility, though it may introduce elevated risks such as off-target effects and immunogenicity. From a regulatory perspective, this article reviews the progress, technological strengths, and regulatory landscape of in vivo CAR-T therapy medicinal products. We analyze their potential benefits in manufacturability, scalability, and discuss challenges including quality control, safety risks, and mitigation strategies in drug development. We propose adaptive regulatory strategies together with early regulatory engagement and international coordination to accelerate the clinical translation and standardized use of this emerging modality.

Introduction

Chimeric antigen receptor (CAR)-T cell therapy medicinal products constitute the paradigm shift in cancer immunotherapy, employing genetic engineering to endow T cells with the capacity for precise recognition and eradication of malignant cells. Since 2017, the U.S. Food and Drug Administration (FDA) has approved seven CAR-T cell products, a number mirrored by regulatory approvals in China for managing B-cell malignancies, including acute B-cell lymphoblastic leukemia and B-cell lymphoma.^1–3 These advanced therapies have exhibited profound efficacy in patients with refractory or relapsed disease.

Nevertheless, conventional autologous CAR-T therapy medicinal products rely on an ex vivo engineering process. This necessitates apheresis, isolation of T cells, activation, introduction of the CAR transgene via viral or non-viral vectors, in vitro expansion of the genetically modified cells, and subsequent reinfusion into the patient. This bespoke, patient-specific manufacturing framework is intrinsically complex, time-consuming and costly, thereby presenting a significant barrier to broad clinical dissemination and precluding the scalable production achievable with conventional pharmaceuticals.⁴ The allogeneic CAR-T approach represents a rapidly advancing area as this approach sources from healthy donors or induced pluripotent stem cells (iPSCs) and provides a scalable, off-the-shelf treatment option with better accessibility. However, it introduces distinct challenges such as graft-versus-host disease (GvHD) and immune rejection, and long-term safety and efficacy remain to be validated.⁵

In recent years, in vivo CAR-T therapy medicinal products have emerged as a transformative alternative that allows modified immune cells to directly target cells in patients, both locally within the engineering microenvironment and systemically after trafficking to distant tissues.⁶ This innovative strategy leverages targeted delivery platforms such as engineered viral vectors or lipid nanoparticles (LNPs), thereby obviating the need for the intricate ex vivo workflow.^7,8 When using viral vectors, the displayed antibody may activate T cells and the resulting CAR⁺ populations in vivo may self-regulate their expansion and persistence according to antigen levels. On the other hand, mRNA delivered via ligand-functionalized LNPs operates transiently, allowing for the control of CAR⁺ cell numbers and persistence relatively independently of antigen burden through dose modulation.⁶ By streamlining the production process and treatment regimen, in vivo CAR-T therapy medicinal products are positioned to rapidly expand the potential applicability of CAR-T therapy beyond hematological malignancies to include autoimmune disorders, and at the proof-of-concept stage, conditions such as fibrosis and other chronic diseases, heralding a new era of accessible therapeutic interventions.^9,10

Accumulating preclinical and early-phase clinical evidence suggests in vivo CAR-T therapy medicinal products capable of achieving functional cures. Illustrating this potential, a lentiviral vector platform engineered with cocal glycoprotein and anti-CD3 single-chain variable fragments (scFv) has demonstrated T cell activation, CAR transduction and B cell depletion in mice and non-human primate (NHP) models, and secondary CAR-T cell expansion was observed in 1 animal to show the persistence of CAR-T cells in vivo.¹¹ A lentiviral vector with detargeted viral fusogen and CD7 binder has produced detectable CAR-T and CAR-NK cells with subsequent depletion of CD20⁺ B cells in mice and NHP models.¹² The first-in-human clinical trial of ESO-T01, the anti-T-cell receptor nanobody-targeted, immune-shielded lentiviral vector further substantiated this promise, reporting responses in 4 patients with relapsed/refractory multiple myeloma.¹³ As for LNP platform, Hunter et al. presented a targeted LNP platform for in vivo mRNA delivery to specific T cell subsets, demonstrating reprogramming of CD8⁺ T cells in both healthy and autoimmune human samples and achieving B cell depletion in cynomolgus monkeys.¹⁰ The first global clinical study employing LNP-based in vivo CAR-T therapy medicinal products for systemic lupus erythematosus (SLE) delivered crucial proof-of-concept, underscoring the platform’s potential to redefine treatment paradigms.¹⁴

Typically, these investigator-initiated clinical trials are designed as open-label, single-arm studies. Their primary objectives include evaluating safety, tolerability, pharmacokinetic (PK) and pharmacodynamic (PD) properties, and preliminary efficacy. Intravenous administration of the lentiviral vector was associated with acute inflammatory reactions, cytokine release syndrome (CRS), hematological toxicities, and isolated neurotoxicity, while the LNP vector elicited low-grade CRS and transient lymphocyte decrease, with fewer high-grade toxicities reported. CAR-positive cells peak around 10-17 days post-vector infusion, and within 6 h after LNP dosing, confirming effective in vivo uptake, with less than 10% off-target transduction reported in the LNP cohort.^13,14 In terms of efficacy, patients with multiple myeloma showed a significant reduction in tumor burden.¹³ Patients with SLE achieved B-cell depletion, decreased antibody levels, and normalization of complement.¹⁴ Currently, clinical research remains limited by small patient cohorts and a lack of long-term follow-up data, underscoring the need for larger, rigorously designed studies to further evaluate the safety and efficacy.

From a regulatory standpoint, it is imperative to recognize that in vivo CAR-T products, as a novel class of therapy products integrating advanced platforms including viral vector engineering, RNA medicine, nano-technology, CAR-T therapy, and antibody conjugate technologies, present a unique set of challenges for their development and oversight.

in vivo vs ex vivo CAR-T therapy medicinal product

Manufacturing

Currently, all globally approved CAR-T therapy medicinal products are autologous ex vivo CAR-T therapies. These rely on the collection, selection of patient-derived T cells and manufacturing under good manufacturing practice (GMP) condition (Fig. 1). The CAR-T cells are expanded ex vivo, formulated with excipients into a cell-based product. The whole manufacturing usually involves complex raw materials and reagents. Quality and safety assessments should be implemented for all raw materials, particularly those of human or animal origin and genetic modification reagents.

Fig. 1.

Schematic comparison of manufacturing workflow of ex vivo and in vivo CAR-T therapy medicinal products. Left panel shows the ex vivo CAR-T manufacturing process. This multi-step procedure requires leukapheresis, T cell isolation. T cells are activated, genetically modified typically via lentiviral transduction to express a chimeric antigen receptor (CAR), and extensively expanded over 2–4 weeks. The final CAR-T cell product is then cryopreserved, shipped, and reinfused into the patients. Right panel shows the in vivo CAR-T manufacturing strategy. This streamlined approach involves the direct administration of a targeted delivery vector (e.g., lentiviral vector or non-viral nanoparticle) encoding the CAR gene. The vector systemically and specifically transduces T cells within the patient’s body in vivo. These modified T cells then undergo activation, CAR expression, and functional expansion entirely in vivo. Key differences include the elimination of ex vivo cell processing, reduction in manufacturing time, and the potential for a truly “off-the-shelf” modality with the in vivo approach

In contrast, in vivo CAR-T therapy medicinal products eliminate the need for harvesting patient T cells (Fig. 1). Instead, vectors such as lentiviral vector, adeno-associated viral vector (AAV), LNP or exosome carrying the genetic cargo are manufactured in vitro under GMP condition and administered directly to the patient, where the generation of functional CAR-T cells occurs entirely within the body.^15,16 The design of vectors requires targeted delivery of CAR construct, enhanced tissue selectivity of the delivery system and cell-specificity of the expression promoters, and elimination of interactions with natural cell surface receptors.^17–19 In addition to the production of gene delivery systems, the manufacturing workflow may also involve the synthesis of small-molecule linkers and antibody conjugation steps. As CAR-T cell culture is not involved in the manufacturing process, the requirement for human- or animal-derived materials is significantly reduced, consequently lowering the risk of adventitious agent contamination.

Scalability and patient access

The inherently personalized nature of autologous ex vivo CAR-T therapy medicinal products presents unique challenges for capacity expansion. Key constraints including elevated raw material costs, complex quality control, and a multi-step workflow. Manufacturing lead times typically extend from several weeks to months, while costs vary considerably based on factors such as labor intensity, equipment utilization, cleanroom requirements, and processing scale.²⁰ Furthermore, expanding capacity through a “scale-out” model introduces additional complexities, including the need for robust cold-chain logistics for cellular materials and final products. Even under centralized oversight, ensuring product consistency across distributed or point-of-care manufacturing sites remains challenging, necessitating harmonized processes, standardized quality control, GMP compliance, staff training, and coordinated hospital activities.

In contrast, in vivo CAR-T therapy medicinal products streamline production through industrialized pharmaceutical processes, reducing the total manufacturing timeline to an estimated 1.5-3 weeks and lowering projected costs by ≥50% per treatment course.^21,22 Although the process scaling and comparability studies of viral vectors and LNPs present challenges, this approach supports the “scale-up” model in scalable manufacturing, where a single batch can treat numerous patients. By eliminating the need for apheresis and CAR-T cell processing, in vivo CAR-T therapy medicinal products simplify clinical administration, allow for flexible and repeat dosing, and improve stability, thus expanding access to remote regions. The avoidance of lymphodepletion and long waiting periods may further reduce hospitalization costs and enhance patient compliance.

Quality control strategy

Traditional ex vivo CAR-T therapy medicinal products, given that partial CAR activation and T-cell expansion occur in vitro, enable comprehensive quality control through multidimensional attribute monitoring of CAR-T cells, including viability, CAR expression and potency, etc. The current quality management strategy addresses critical quality attributes from the viral vector to the final cellular product. Additionally, the management of out-of-specification (OOS) batches of CAR-T cells requires particular attention for releasing autologous products, given the inherent donor-to-donor variability.

For in vivo CAR-T therapy medicinal products, quality assurance focuses primarily on the delivery vector. There is a pressing need to develop novel analytical methods to evaluate the targeting efficiency and specificity of delivery vectors. For viral vector-based products, critical quality attributes may encompass viral titer, impurities, antibody density (if applicable) and potency, etc.²³ For LNP-based products, critical quality attributes may include the physicochemical properties, size and potency of the LNPs, the ratio and site-specific distribution of conjugated antibodies, as well as the purity, integrity of the mRNA payload, etc.

The comparison of chemistry, manufacturing, and control (CMC) information between in vivo and ex vivo CAR-T therapy medicinal products is presented in Table 1.

Table 1.

Comparison of CMC information for ex vivo versus in vivo CAR-T therapy medicinal products

CMC information		Ex vivo CAR-T (Lentiviral vector)	In vivo CAR-T (Lentiviral vector)	In vivo CAR-T (LNP-mRNA)
Design of delivery system		• Self-inactivation • Non-replicating • Pseudotyping	• Self-inactivation • Non-replicating • Envelope engineering or antibody conjugation	• Lipid composition optimization • Surface modifications • Antibody conjugation
Scalability		• Scale out strategy • Decentralized or point-of-care manufacturing • Validation of production capacity	• Scale up strategy • Centralized manufacturing • Standardized and suitable for scaled production	• Scale up strategy • Centralized manufacturing • Standardized and suitable for scaled production
Manufacturing of active substance	Starting materials and raw materials	• Donor-derived live cells • Human serum (if applicable) • Magnetic beads • Cytokines and antibodies	• Plasmids and packaging cell banks • Fetal bovine serum (if applicable)	• Plasmids, transcriptase and nucleotides • Lipids • Serum free
	Manufacturing process	• Cell processing procedures • Extended, and labor-intensive • Complex supply chain with process-is-the-product model	• Efficient processes and shorter production timelines • Higher batch-to-batch consistency • Rapid production with optional antibody conjugation	• Efficient mRNA transcription and LNP production • Rapid production with optional targeted LNP conjugation
Quality Control		• Viral vector: physical titer, infectious titer, purity, impurities, potency, sterility, mycoplasma, RCL • CAR-T cell product: T-cell viability, quantity, CAR+ cell percentage, immunophenotype, VCN, RCL, sterility and mycoplasma	• Viral vector: physical titer, infectious titer, purity, impurities, potency, sterility, mycoplasma, RCL • Conjugated product: Antibody density and conjugation sites (if applicable)	• mRNA: purity, integrity, capping efficiency, polyA tail, impurities, potency • LNP: size, PDI, EE, potency, impurities • Conjugated LNP: antibody density and sites; physicochemical properties, purities and impurities (if applicable)

CMC chemistry, manufacturing and control; EE encapsulation efficiency; LNP lipid nanoparticle; PDI polydispersity index; VCN vector copy number; RCL replication-competent lentivirus

Safety and persistency

For ex vivo CAR-T therapy medicinal products, safety profiles include known risks like CRS, neurotoxicity, off-target effects, and insertional mutagenesis. Despite inherent limitations from interspecies biological constraints, murine xenograft models can still inform the trafficking and proliferation profiles of human CAR-T cell samples.²⁴ Greater experience has been accumulated in managing side effects in clinical trials, though more data is needed to identify and mitigate unforeseen risks.

For in vivo CAR-T products, since CAR-T cell generation occurs entirely in vivo, precise control over transduction efficiency, activation, expansion, and persistence remains challenging, while concurrently presenting elevated risks such as off-target effects and immunogenicity. It has been reported that viral vectors present higher potential to trigger immune responses that may cause unwanted inflammatory responses,²⁵ while lipid formulation-dependent tropism for hepatic and myeloid cells may lead to acute infusion reactions, hepatotoxicity, and immunogenicity-mediated hypersensitivity responses.²⁶

Available preclinical studies of in vivo CAR-T therapy medicinal products in mice and NHPs (surrogate administration) demonstrated targeted delivery of the viral vector to T cells in peripheral blood, spleen, and lymphoid organs. Off-target transduction was not significantly observed in in vitro cell studies, murine models and NHPs, supporting the success of the targeting viral engineering strategy; however, long-term clinical monitoring remains warranted.¹² The reported preclinical data supported a low risk profile for insertional oncogenesis, which did not appear to exceed that of approved lentiviral ex vivo CAR-T therapy medicinal products.²⁷ Preliminary clinical trial results indicate potential post-infusion immune reactions, including acute inflammatory responses, elevated liver enzymes, and isolated cases of neurotoxicity.¹⁴ These preclinical and clinical observations remain preliminary and highlight the need for an investigational new drug (IND) study for further validation.

Given the distinct safety profile of in vivo CAR-T therapy medicinal products, comprehensive non-clinical studies using humanized or NHP models are essential to evaluate pharmacology, PK, and toxicity. General toxicity should be assessed via good laboratory practice studies (e.g., mortality, clinical pathology, cytokines) for an IND application, with first-in-human doses selected considering safety margins from the animal maximum tolerated dose (MTD). Biodistribution in non-lymphoid tissues can clarify targeting specificity, while continuous monitoring of vector copy number and integration sites is needed throughout development.

Unlike ex vivo CAR-T, dosing for in vivo CAR-T therapy medicinal products is based on vector quantity rather than cell number. New dose-exploration strategies are needed to define relationships between vector dose, regimen, and in vivo CAR-T cell generation. It is noted that variability in lymphocyte counts and function due to prior therapies can affect in vivo transduction efficiency, expansion, and persistence. Early clinical data confirm detectable CAR-positive cells across patients, supporting feasibility, although variability impacts expansion kinetics and potential clinical durability, and it remains unclear whether consistent conclusions can be drawn in later clinical trials. For indications requiring repeat dosing, immune responses against the vector or CAR transgene may pose a safety and efficacy concern. Therefore, defining lymphocyte thresholds and dose-response relationships requires further preclinical and clinical data.

Hurdles to overcome

The development of current in vivo CAR-T therapy medicinal products integrates design concepts from novel delivery systems, antibody-drug conjugates, and cell therapy products. Compared to traditional biologics and cell therapy products, these products exhibit distinct characteristics in manufacturing processes, quality assessment, and in vivo behavior, presenting significant challenges for regulatory oversight.

Structural design, manufacturing, and control

Currently, the common approaches including genetically engineered viral vectors and targeted non-viral vectors face limitations in targeting specificity and cell selectivity.^11,28 Although current clinical studies have not demonstrated a clear correlation between lentiviral vectors and carcinogenic mutations, and third-generation lentiviral systems have reduced oncogenic risks, these concerns persist.^29,30 In addition, the scalable production of highly purified viral vectors necessitates the rigorous optimization and validation of the manufacturing process, and batch-to-batch consistency poses considerable technical difficulties. For the utilization of LNPs to deliver mRNA, significant challenges remain, including endosomal/lysosomal barrier, low transduction efficiency, and short duration of action.^31,32 The selection of an antibody conjugation method like biotin-avidin interaction or covalent conjugation is a critical issue, as it must simultaneously preserve antigen-binding ability and ensure nanoparticle stability.³³

The development of novel analytical approaches poses substantial regulatory issues, highlighting the critical need to establish alternative methods to assess transduction efficiency, T-cell targeting specificity, and potency. And substantial challenges remain regarding how to conduct comprehensive comparability studies when manufacturing change happens, and to determine the necessity and scope of non-clinical and/or clinical bridging studies based on risk assessment.^20,34

Safety concerns

Viral reactivation is a recognized clinical risk associated with ex vivo autologous CAR-T therapy medicinal products. Current management strategies for this risk are well-established and comprise comprehensive protocols, including long-term monitoring and prophylactic medication. While reactivation of HHV-6 and CMV has been reported with ex vivo products, data for in vivo CAR-T therapy medicinal products remain limited.^35,36 Although viral vectors are designed as self-inactivated and non-replicating, their direct infusion and subsequent induction of T-cell activation and expansion could still pose a theoretical risk of viral reactivation and require careful evaluation. Continuous monitoring throughout treatment is recommended to mitigate adverse events and improve patient outcomes.

Currently, there is no CAR-T product approved on market for non-oncological indications. For patients with non-malignant diseases, the acceptable threshold for safety risks is substantially lower than for those with cancer. In refractory end-line cases lacking effective treatment options, such therapies may be considered when clinical benefit is judged to outweigh potential risks. This approach necessitates careful assessment of the acceptable safety margin and whether the treatment aligns with ethical standards for patient care. While acute toxicities like cytokine release syndrome are managed consistently, the long-term safety is critical for patients with chronic conditions and near-normal life expectancy. Decades of expected survival amplify theoretical risks such as insertional mutagenesis, and persistent CAR-T activity could elevate lifetime susceptibility to infections and secondary malignancies. Therefore, developing in vivo CAR-T therapy medicinal products for non-oncological diseases may require rigorous target validation, integrated safety switches, extended follow-up and precise control over cellular persistence.³⁷

Ethical and biosafety

In vivo CAR-T therapy medicinal products represent a promising frontier in the treatment of cancer and autoimmune diseases, yet they introduce complex ethical considerations. A central challenge arises from their potential use of gene transfer and permanent genomic editing, which demands rigorous ethical oversight and thoroughly informed consent processes to ensure patients fully comprehend the associated risks and benefits.^38,39 Furthermore, the production and quality control of viral vector-based products involve replication-competent viruses, heightening biosafety concerns, particularly when HIV-derived positive controls are utilized in vector release testing. Consequently, a critical imperative for both developers and regulators is to establish clinically relevant benefit-risk assessment frameworks that effectively evaluate the ethical justifiability of human genetic modification, and multidisciplinary expert team is recommended to conduct comprehensive biosafety evaluations for various viral vector types.

Strategies for addressing challenges

Given the innovative nature of in vivo CAR-T therapy medicinal products and the current limited understanding, development and regulatory strategies for this modality may not be fully adapted from existing frameworks. Addressing these challenges requires adopting a quality by design (QbD) approach, integrating regulatory considerations from the earliest stages of product development. By aligning technological advances in the field with continuous improvement of regulatory tools, it is possible to formulate fit-for-purpose regulatory strategies that are appropriate to the current stage of product development (Table 2).

Table 2.

Challenges for in vivo CAR-T therapy medicinal products, examples of mitigation strategies and regulatory considerations

Challenges	Examples of Mitigation Strategies	Regulatory Considerations
Targeting specificity	• Viral vector engineering • Improve targeting ligand affinity and stability • Develop site-specific, stable conjugation methods • Optimize cationic lipids	• Quality control of viral vector and targeting ligand • Characterization of conjugation sites and density • Process validation • Evaluation of off-target effects
CAR expression durability	• Sustained-release systems • Self- and trans-amplifying RNA • Circular RNA	• Assessment of mRNA sequence design, translational fidelity and immunogenicity • Risk mitigation strategies
Scalable manufacturing process	• Quality by design • Definition of critical process parameters • Establish comparability study protocols	• Standardize comparability studies • Risk-based approach • Non-clinical or clinical bridging studies when necessary
New testing methods	• Method development and validation • Compliance with ICH Q14 and Q2 (R1) guidelines	• Multi-dimensional quality characterization • Standardize sample handling and analytical procedures
Off-target effects, insertional mutagenesis, immunogenicity, and biodistribution issues	• Engineer viral capsid • Develop sensitive detection methods with bioinformatics tools • Conduct long-term follow-up studies	• End-to-end safety optimization • Prioritize rational animal models • Assay validation
Ethical	• Compliance with applicable regulations • Thorough risk-benefit assessments	• Assembly of expert panels • Evaluation of gene therapy-related ethical issues
Biosafety	• Compliance with applicable biosafety regulations • Enhance personnel protection during viral handling	• Conduct comprehensive biosafety assessments • Differentiated biosafety standards based on viral vector classification

Development strategies

Developers may address current challenges by refining target selection, engineered cell types, delivery systems, and CAR design.

Expanding the target repertoire is a promising strategy. Dual-targeting and multi-targeting strategies may potentially enhance efficacy. In autoimmune patients, targeting CD19 can broadly deplete B cell populations, and most anti DNA/nucleosomal autoantibodies are cleared, though some (e.g., anti Ro) persist.⁴⁰ The studies in refractory SLE show autoreactive antibodies may also originate from CD19^-BCMA⁺ long-lived plasma cells.⁴¹ The phase 2 trial of targeting BCMA reported promising results in autoantibody-associated autoimmune disorders such as refractory generalized myasthenia gravis (MG), and the phase 3 clinical trial is ongoing.⁴² Dual targeting of CD19 and BCMA reported encouraging clinical outcomes in cases where autoantibodies are primarily plasma cell driven.⁴¹ For the exploration of new targets, CRISPR screening platform and artificial intelligence can be employed to select safer targets, optimize the function of CAR-T cells and mitigate on-target off-tumor toxicity.^43,44

In addition, multiple immune cell types can be rationally engineered. For example, CAR technology offers a non-MHC-restricted strategy for redirecting Treg specificity. The preliminary safety, PK/PD and efficacy data in the Phase 1 clinical trial of CAR-Treg reveal the potential for antigen-directed Tregs to treat rheumatic arthritis, though manufacturing challenges such as limited expansion capability and long-term stability remain to be solved.^45,46 The simultaneous engineering of multiple cell types is also being explored, as exemplified by a circular RNA-based panCAR platform that can generate CAR-T, CAR-NK, and CAR-macrophage cells, demonstrating tumor suppression and microenvironment reprogramming in mice.⁴⁷

The development of regulatable CAR expression systems includes the design of in vivo safety switch systems that enable selective elimination of CAR-T cells in the event of severe toxicity, as well as the use of small-molecule drugs or antibodies to precisely control CAR-T cell activity and persistence.^48,49 And optimizing genetic components like promoters enables tissue- and time-specific CAR expression, improves expression efficiency and minimizes off-target.⁵⁰ Combined with strategies such as immune checkpoint inhibitors or small-molecule drugs, these innovations may further overcome T cell exhaustion and treatment resistance.^51–53

For the optimization of delivery systems, efforts are focused on developing more targeted, efficient, and durable platforms. Current approaches include novel LNP formulations, engineered viral capsids, and polymeric nanoparticles, for which platform processes have been preliminarily established.^54–60 The implementation of automation, high-throughput screening, and machine learning-guided design may advance next-generation LNP therapeutics.⁶¹ With ongoing technological advances, emerging delivery technologies such as adeno-associated viral vectors, protein-based delivery, and exosome-mediated delivery are also being explored for in vivo CAR-T applications, necessitating thorough risk assessment and control measures tailored to the characteristics of each vector system.^62–65

In terms of safety and efficacy control, novel regulatory strategies are required to address the unique challenges posed by these innovative products. Safety optimization should be integrated from the initial product design phase and maintained throughout the entire development process.^66,67 For clinical trials involving multiple dosing, it is recommended that the safety and efficacy of a single administration first be thoroughly investigated. Once sufficient data have been accumulated, the necessity and safety of repeated dosing should then be rigorously evaluated. Dosing regimens should be rationally designed according to the in vivo distribution and persistence profile of the product to balance therapeutic efficacy and safety. Risk management should integrate non-clinical and clinical safety data to control known risks while continuously monitoring for unknown ones. Preventive measures could be informed by safety data from both the investigational product and relevant analogues, supported by long-term subject follow-up to ensure ongoing safety. The comprehensive patient monitoring plan including tracking of cytokine levels and dynamic changes in CAR-T cell populations should be established, accompanied by well-defined procedures for adverse event grading and management, as well as detailed criteria for dose adjustment and study suspension.

Regulatory support

Adaptive regulatory strategies

Alternative regulatory pathways could be designed to accommodate the specific characteristics of in vivo CAR-T therapy medicinal products. In the current stage, a comprehensive, risk-proportionate approach is required to address uncertainties and unique challenges through integrated evidence assessment. As product maturity and clinical experience accumulate, a weight-of-evidence (WOE) approach integrating non-clinical, clinical, and CMC data should be applied to support regulatory decisions. And regulatory agencies should actively participate in early-stage product development and enhance multi-level communication mechanisms.

International collaboration

Regulatory frameworks for advanced therapies differ across major jurisdictions (Table 3). The U.S. FDA classifies in vivo CAR-T therapy medicinal products as gene therapy products and has implemented accelerated pathways, including the Regenerative Medicine Advanced Therapy (RMAT) designation. In the EU, the Committee for Advanced Therapies (CAT) evaluates Advanced Therapy Medicinal Products (ATMPs) and offers expedited routes such as the PRIME scheme.¹ Center for Drug Evaluation, NMPA categorizes these products as Gene Therapy Medicinal Products (GTMPs) within its ATMP framework. Globally, the ICH Cell and Gene Therapy Discussion Group (CGTDG) is prioritizing harmonization efforts for established modalities including ex vivo CAR-T therapy medicinal products and in vivo viral vector products.⁶⁸ And many collaboration mechanisms such as FDA-EMA ATMP Cluster are established to develop common understandings and align regulatory approaches.

Table 3.

Comparison of regulatory frameworks across agencies for in vivo CAR-T-related products

Regulatory frameworks		NMPA	FDA	EMA
Classification		Gene Therapy Medicinal Product	Gene Therapy Product	Gene Therapy Medicinal Product
Accelerated Pathways		• Pre-Pre-IND meeting • Breakthrough therapy • Priority review • Conditional approval	• INTERACT meeting • Expedited programs • Platform technology designation	• Innovation Task Force meeting • ATMP Classification • PRIME scheme
Covered modalities		• Gene therapy products for human use • In vivo gene therapy medicinal products • Lentiviral vector • ɤ-Retroviral vector • rAAV • LNP-mRNA • Gene-editing tools • Immune cell therapy products	• Human gene therapy products • Microbial vectors • Human gene therapy products incorporating human genome editing • CAR-T	• ATMP in clinical trials • Gene therapy medicinal products • Lentiviral vector • rAAV • Genetically modified cells
Considerations/topics covered in guidelines	CMC	• GMP of Cell Therapy Products • Structural Design • Raw materials requirements • RCL • Quality control • Bioassays • Changes of autologous CAR-T cell therapy	• CGMP • CMC requirement for IND application • Human- and animal-derived materials • RCL • Potency Assurance • Manufacturing changes and comparability	• GMP specific to ATMPs • Design modifications • The use of tumorigenic cells of human origin • Potency testing • Comparability considerations for ATMP
	Non-clinical	• Product-specific considerations • Insertional mutagenesis • Viral shedding	• First-in-human requirements • Disease and product-specific considerations • Viral shedding	• First-in-human requirements • Inadvertent germline transmission • Insertional mutagenesis • Viral shedding
	Clinical	• Product-specific considerations • Risk management plan of CAR-T products for marketing application • Long-term follow-up	• Early-phase clinical trials • Disease and product-specific considerations • Long-term follow-up	• Risk management of ATMP • Clinical risks deriving from insertional mutagenesis • Safety and efficacy follow-up of ATMP

ATMP advanced therapy medicinal product; rAAV recombinant adeno-associated viral vector; CMC chemistry, manufacturing and control; CGMP current good manufacturing practice; GMP good manufacturing practice; INTERACT initial targeted engagement for regulatory advice on CBER/CDER products; LNP lipid nanoparticle; PRIME priority medicines; RCL replication-competent lentivirus

Regulatory agencies should strengthen international coordination to avoid redundant development and evaluation efforts. A concerted framework should proactively address key translational gaps, including criteria for species relevance, rationale for toxicology models, and the adoption of new approach methodologies. Furthermore, it should establish standardized assessments of vector biosafety encompassing replication competence and shedding studies to evaluate patient and environmental risks and inform mitigation strategies throughout the treatment lifecycle. Building on existing frameworks for gene therapy products, clear guidance specific to in vivo CAR-T therapy medicinal products should be established to assist developers in evaluating and managing risks.

Outlook

Currently, most ex vivo CAR-T therapy medicinal products remain restricted to high-income regions and specialized centers that possess the technology background and specific expertise. In vivo CAR-T therapy medicinal products could democratize advanced therapies by eliminating complex manufacturing, reducing costs and enabling treatment in local clinical settings worldwide. Driven by continuous technological advancements and evolving regulatory science, this progress has the potential to realize substantial economic and accessibility benefits through centralized bioprocessing, simplified administration, and versatile platform technologies. These features collectively support scalability, rapid adaptation to new targets, and broader implementation across diverse healthcare systems, particularly in resource-limited regions, making transformative cell and gene therapies more feasible across diverse healthcare systems.⁶⁹ However, this nascent technology also introduces a distinct set of developmental and regulatory challenges, necessitating innovative quality control strategies, novel safety assessment methodologies, and adapted clinical development pathways.

Regulatory agencies should engage in close collaboration with developers to construct tailored frameworks that effectively balance the acceleration of innovation with the imperative of risk mitigation. Concurrently, sustained attention must be paid to the concomitant ethical and accessibility considerations to ensure these groundbreaking therapies can reach the patients most in need. Adaptive regulatory frameworks are expected to enhance patient access to in vivo CAR-T therapy medicinal products in the coming years, and help shape the future landscape of advanced therapy regulation.

Author contributions

Jiaqi Lu, Jing Cui, Jing Dong, Liping Yan wrote the manuscript, and prepared the figures and tables. Xue Wang and Di Zhou prepared and edited tables. Shuang Lu edited the article. Wei Wei conceptualized and wrote the article. All authors have read and approved the article.

Competing interests

The authors declare no competing interests.