TCR-T cell therapy: current development approaches, preclinical evaluation, and perspectives on regulatory challenges

Abstract

TCR-T cell therapy represents a promising advancement in adoptive immunotherapy for cancer treatment. Despite its potential, the development and preclinical testing of TCR-T cells face significant challenges. This review provides a structured overview of the key stages in preclinical testing, including in silico, in vitro, and in vivo methods, within the context of the sequential development of novel therapies. This review aimed to systematically outline the processes for evaluating TCR-T cells at each stage: from in silico approaches used to predict target antigens, assess cross-reactivity, and minimize off-target effects, to in vitro assays designed to measure cell functionality, cytotoxicity, and activation. Additionally, the review discusses the limitations of in vivo testing in animal models, particularly in accurately reflecting the human tumor microenvironment and immune responses. Performed analysis emphasizes the importance of these preclinical stages in the safe and effective development of TCR-T cell therapies. While current models provide valuable insights, we identify critical gaps, particularly in in vivo biodistribution and toxicity assessments, and propose the need for enhanced standardization and the development of more representative models. This structured approach aims to improve the predictability and safety of TCR-T cell therapy as it advances towards clinical application.

Introduction

Currently, innovative personalized approaches to immunotherapy are undergoing significant development. One such approach, adoptive T-cell therapy, which involves the infusion of engineered T cells into patients, has led to breakthroughs in the treatment of certain cancers [1, 2]. This therapy entails the genetic modification of primary human T cells to express tumor-specific receptors, either T-cell receptors (TCR) or chimeric antigen receptors (CAR). These modifications enable T cells to recognize and destroy malignant cells [3]. While several CAR T-cell therapies have been approved by the Food and Drug Administration (FDA) for hematologic cancers, their effectiveness against solid tumors remains limited [4]. TCR-T cell therapy, although not yet approved for clinical use, is currently being evaluated in numerous preclinical and pilot trials targeting various antigens [4]. TCR-T cells are considered a more promising therapeutic approach for the treatment of solid tumors due to several key advantages. TCR-T cells are more sensitive to antigen levels in tumor cells compared to CAR-T cells and have the ability to recognize not only surface antigens but also intracellular antigens, making them applicable to a broader spectrum of tumors [5]. Moreover, TCR-T cells may demonstrate greater efficacy in the complex microenvironment of solid tumors, as they can interact with MHC-presented antigens, facilitating deeper tumor infiltration and enhanced activity within the tumor tissue. In addition, TCR-T cell therapy is associated with a potentially lower incidence of adverse effects, such as cytokine release syndrome, when compared to CAR-T therapy [6].

The main limitation of TCR-T cell therapy, when compared to CAR-T therapy, is the requirement for antigen recognition through major histocompatibility complex (MHC) molecules. Consequently, the effectiveness of TCR-T cell therapy is dependent on the patient’s specific human leukocyte antigen (HLA) profile, necessitating individualized cell modifications based on HLA compatibility. This poses a challenge for the widespread application of TCR-T cells, unlike CAR-T cells, for which standardized “off-the-shelf” options are being developed [6]. However, recent efforts have begun to develop pluripotent TCR-T cells capable of interacting with multiple HLA alleles [7].

Tumors can also develop resistance to TCR-T cell therapy by downregulating or losing HLA expression, which impairs antigen presentation and makes the tumor invisible to TCR-T cells. This phenomenon increases the risk of relapse and diminishes the therapeutic efficacy [8]. Additionally, successful antigen recognition by TCR-T cells requires proper antigen processing and presentation on the surface of tumor cells in complex with HLA. If this processing mechanism is disrupted, the effectiveness of the therapy may be significantly reduced [9, 10].

Before clinical trials can be approved and marketing authorization granted for T-cell-based medicinal products, preclinical trials must be conducted [11]. The goal of these preclinical trials is to evaluate the pharmacological, kinetic, and toxicological properties of the novel drug to predict its efficacy and safety in clinical practice. Unlike pharmaceuticals, there is currently no standardized approach for testing T-cell therapies. Therefore, preclinical study planning must consider potential risks identified in clinical trials of similar cell-based products [12].

Past safety issues associated with TCR-T cell therapy, particularly those related to unexpected "on-target, off-tumor toxicity," have resulted in severe adverse effects and fatalities in clinical trials. This phenomenon occurs because TCR-T cells may attack not only tumor cells expressing the target antigen but also healthy tissues that express the same antigen, even in minimal amounts. One of the most well-known examples involves TCR-T cells targeting the MAGE-A3 antigen (melanoma-associated antigen). In clinical trials, patients developed cardiotoxicity, which led to fatal outcomes [13, 14]. The cause of this toxicity was the cross-reactivity of TCR-T cells with titin, a protein expressed in cardiac muscle.

Another example of unexpected toxicity involves TCR-T cells targeting the MART-1 antigen, a melanoma antigen. In some clinical trials, patients developed autoimmune reactions against melanocytes, leading to vitiligo and other skin-related adverse effects [15, 16]. These reactions were caused by the cross-reactivity of TCR-T cells with normal tissues expressing MART-1.

In another clinical trial, patients receiving TCR-T cells engineered to target CEA (carcinoembryonic antigen) developed inflammatory colitis, which required systemic immunosuppression with high-dose steroids and anti-cytokine antibodies [17].

These examples underscore the importance of more stringent preclinical testing and thorough evaluation of potential cross-reactivity between tumor and normal tissues before initiating clinical trials.

To facilitate the planning of preclinical trials, we have consolidated and systematized data from existing protocols of T-cell-based drug studies available in the literature. Our analysis includes information from current international guidelines applicable to adoptive cell therapy, such as Fumagalli et al. [12]; the Guideline on the Non-Clinical Studies Required Before First Clinical Use of Gene Therapy Medicinal Products, 2019 [18]; the Guideline on Quality, Non-Clinical and Clinical Aspects of Medicinal Products Containing Genetically Modified Cells, 2019 [19]; and the Guideline on Quality, Non-Clinical, and Clinical Requirements for Investigational Advanced Therapy Medicinal Products in Clinical Trials, 2019 [20]; as well as the Preclinical Assessment of Investigational Cellular and Gene Therapy Products: Guidance for Industry, 2019 [21]. Additionally, we considered specific features of Russian legislation, including the Order of the Ministry of Health of the Russian Federation dated August 8, 2018, No. 512n “On the Approval of the Rules of Good Practice for Working with Biomedical Cellular Products” [22], and Federal Law dated June 23, 2016, No. 180-FZ “On Biomedical Cellular Products” [23].

This study aims to provide a summarized universal protocol for the key stages of TCR-T cell development and approval, while also addressing the existing challenges and issues in the field.

The key stages of the development and approval of TCR-T cell therapy

Figure 1 presents a generalized flowchart outlining the development of TCR-T cells and the conduct of their preclinical trials. This flowchart considers both the complex nature of these therapies and the limitations of currently available experimental models.

Fig. 1

Flowchart of TCR-T cell manufacturing and preclinical trials. Stages that can be conducted using various in silico methods are marked with asterisks. These methods are typically employed during the “Development of Production Technology” stage. AG: antigen, TCR: T-cell receptor

This review focuses on the critical stages of TCR-T cell preclinical trials: “Development of Production Technology,” “Efficiency Assessment,” and “Safety Assessment,” each of which consists of multiple steps. While these stages align with the general phases of preclinical trials in drug development, TCR-T cell therapy presents unique challenges due to the technological and functional characteristics of autologous bioengineered cell-based products. A key challenge is the difficulty in selecting suitable animal models due to the infusion of human T cells, which necessitates a much broader use of in silico approaches compared to traditional pharmaceuticals. These approaches are employed not only at the initial stages of design and development but also in later phases of research to predict mechanisms of action and potential side effects. Previous studies have extensively described the efficacy tests used [2] and the safety tests [11] along with their advantages and limitations. However, our work is the first to organize these tests according to the stages of the technological production chain.

Development of production technology

The development of production technology encompasses all stages, from selecting a relevant antigen (AG) for TCR design and the corresponding genetic construct, to evaluating the efficiency of transducing selected cells with the synthesized construct (Fig. 2).

Fig. 2

Flowchart of the “Development of Production Technology” stage. AG: antigen; TCR: T-cell receptor

Choosing an appropriate tumor antigen

Antigen selection is a critical factor in developing safe and effective adoptive cell therapy. TCRs recognize target antigens that are bound by major histocompatibility complex (MHC) molecules, which is a key advantage over CAR-T cells, as it allows for the detection of both surface and intracellular antigens [1]. Consequently, a wide range of antigens can be targeted in TCR-T cell therapy, though there are certain limitations related to how these antigens are presented on the cell surface. First, antigens with peptide epitopes identical to those recognized by TCR-T cells may also be expressed in healthy human tissues. Second, peptide epitopes in healthy tissues may structurally resemble target peptides sufficiently to be recognized by engineered TCR-T cells. Third, different MHC molecules may present off-target antigen peptides on healthy cells, leading to cross-reactivity with the engineered receptor [24]. Therefore, while an ideal target for TCR-T cells would be an MHC-peptide complex that is highly expressed and specific to all tumor cells, finding such an antigen is often challenging in practice.

Current clinical trials focus on two major classes of tumor antigens: tumor-associated antigens (TAAs) and tumor-specific antigens (TSAs) [25].

Tumor-associated antigens (TAAs) are common antigens expressed across various healthy tissues but are overexpressed in tumors. Examples include HER2 [26] and mesothelin [27]. TAAs can be further classified into differentiation antigens and cancer germline antigens. Differentiation antigens are shared by tumors and their tissue of origin, such as MART-1 in melanoma [28]. Cancer germline antigens, whose expression is restricted to immune-privileged organs like the testes and placenta, include antigens such as MAGEs [29] and NY-ESO-1 [30]. Since these organs do not express MHC molecules, these antigens are not presented during spermatogenesis and are absent in healthy tissues [31].

In CAR-T cell therapy, the CD19 antigen, which is expressed on B cells, has become one of the primary targets for the treatment of B-cell lymphomas and leukemias [32, 33]. The depletion of B cells leads to B-cell aplasia, but this effect can be managed through immunoglobulin replacement therapy. Another example is BCMA (B-cell maturation antigen), which is expressed on mature B cells and plasma cells. BCMA is a target for CAR-T cells in the treatment of multiple myeloma [34]. Since BCMA is primarily expressed on plasma cells, targeting it does not result in severe side effects, similar to CD19.

TAAs are commonly found across different patients, making them promising targets for antitumor therapy [35]. However, because they are also expressed in normal tissues, albeit at low levels, the use of engineered receptors targeting these antigens may lead to on-target, off-tumor toxicity. Additionally, high-affinity T cells targeting these antigens might be eliminated during negative selection in the thymus, making it difficult to obtain highly active TCRs [1].

Tumor-specific antigens, or neoantigens, are expressed exclusively by tumor cells and are absent in healthy tissues. For instance, mutation-derived antigens represent unique antigens that arise from specific point mutations and are expressed exclusively in tumor tissue [36]. Such “shared neoantigens,” like KRASG12D neoantigens [37] or PIK3CA [38], may be common among different patients with the same HLA allele [1]. TCR-T cells targeting these shared neoantigens are currently undergoing clinical trials [1]. Viral-derived neoantigens, like E6 and E7, which are expressed in human papillomavirus-associated cancers, are also noteworthy [39]. While the application of such antigens is restricted to certain types of cancer, viral antigens exhibit tumor specificity and are non-toxic to normal tissues [1]. Moreover, a single viral antigen can only be used to treat patients with a specific combination of virus-induced cancer and HLA type [5]. Alternative TSAs are highly specific tumor antigens derived from genomic sources other than nonsynonymous single nucleotide polymorphisms, such as frameshift mutation-derived antigens, splice variants, gene fusions, endogenous retroelements, etc. [40] (e.g., HERV-E in renal cell cancer [41]). These antigens are fairly common in multiple tumor types.

Neoantigen-based therapy, however, has several limitations. High-affinity T cells specific to TSAs are not eliminated during negative selection in the thymus and can be isolated from the patient’s tumor or from the peripheral blood of a healthy donor [43]. However, the number of neoantigens in a tumor is relatively limited and can vary among patients, making it difficult to identify appropriate targets for each individual and potentially reducing the efficacy of TCR-T cell therapy. Another critical limitation in the use of TSAs for TCR-T cell construction is the risk of cross-reactivity. To prevent this type of toxicity, in silico studies are required to exclude cross-reactivity between TCRs and extratumoral epitopes, including screening for potential amino acid and single nucleotide substitutions [1, 35].

One of the key approaches to overcoming the challenges of targeting neoantigens is the development of TCRs directed at unique antigens that are exclusively present in the tumor cells of a specific patient [44]. Currently, research teams are developing personalized TCRs that account for specific mutations in a patient’s tumor, increasing the likelihood of successfully eradicating the tumor without damaging normal tissues [45].

Accurately predicting target antigens is essential for developing TCR-based immunotherapy [35]. Today, in addition to selecting suitable antigens based on literature and clinical data, researchers can also utilize a variety of in silico methods. These tools help identify potential antigens or epitopes that could be effective targets for TCRs [42]. This process considers several factors, including the MHC's ability to present the antigen, its interaction with MHC molecules, and its specificity for tumor cells. Antigen prediction tools use algorithms to analyze the molecular characteristics of antigens and their association with MHC molecules [46]. This allows for the identification of antigens that can effectively present themselves on the cell surface and bind to TCRs. At the same time, identifying antigens with high specificity for tumor cells is crucial for minimizing adverse events and maximizing the efficacy of immunotherapy [1].

In summary, the key factors for successfully engineering TCRs for therapeutic applications include the choice of target antigen, the efficiency of antigen presentation by target tissue cells, the relative levels of antigen expression in tumor cells compared to healthy cells, and the appropriate HLA allele restriction.

Designing tcrs

The design of T-cell receptors (TCRs) in biotechnology relies heavily on various algorithms for sequence prediction, which play a crucial role in the development of biomedical engineering products. These algorithms help identify potential TCRs capable of recognizing specific antigens or epitopes, as well as analyze similarities among them [47].

At the early stages of TCR technology development, various methods were widely employed to identify specific receptors capable of recognizing tumor antigens. The main approaches used in this process included: the incubation of peripheral blood mononuclear cells (PBMC) or tumor-infiltrating lymphocytes (TIL) from cancer patients with tumor antigens [48], the use of MHC multimers incubated with T cells to select those that interact with target antigens [49], vaccinating mice with human tumor antigens to isolate high-avidity T cells capable of reacting to human tumors [50], and the use of hybridomas created by fusing T cells with tumor cells to generate stable T cell lines expressing specific receptors [51]. Additionally, PBMCs from healthy donors were incubated with tumor antigens to identify TCRs capable of recognizing tumors [52]. Each of these methods has played a crucial role in identifying specific TCRs that can be used in cancer immunotherapy. The advancement of these technologies has improved the process of identifying and selecting TCRs, enhancing their precision and effectiveness.

The widespread use of protein crystallization techniques has led to the accumulation of extensive structural data on atomic positions, bond lengths, dihedral angles, and more, including for T-cell receptors [53]. This structural information about TCRs, MHC–peptide complexes, and antigens facilitates the design of TCRs with improved binding and specificity, often utilizing methods such as molecular modeling and molecular dynamics [54]. Molecular modeling techniques, including molecular docking, can predict interactions between TCRs and MHC–peptide complexes, enabling predictions of TCR binding affinity and specificity to target antigens, as well as simulations of MHC–antigen complex structures [55, 56].

Various methods are employed to forecast the immunogenicity of engineered TCRs and antigens, helping to predict possible immune reactions and assess the safety of therapies [57]. Computer modeling is also used to predict the pharmacokinetic and pharmacodynamic characteristics of TCR-T cell therapy, optimizing dosage regimens and treatment strategies [58]. In silico tools can identify potential external targets of engineered TCRs, thereby minimizing adverse reactions. Machine learning and predictive modeling are further applied to anticipate and mitigate potential adverse events related to TCR-T therapy [11]. Additionally, various models assist in patient selection and the personalization of TCR-T therapy by predicting which patients are most likely to respond positively to treatment [59].

Several -omics technologies are used to obtain TCR sequences. One such method, repertoire sequencing (AIRR-seq) [60], is employed for systematic analysis of TCRs, allowing for the determination of their sequences and diversity. Single-cell sequencing technology is another method being actively developed, providing more precise data on receptor distribution in immune cells through both DNA and RNA analysis. This technology allows for the identification of specific alpha and beta chains that form TCRs, significantly enhancing analysis resolution [61].

Other modern tools and methods, such as RNA sequencing (e.g., Nanostring) and secretome analysis (e.g., Polyfunctionality with Isoplexis), are also utilized to study T-cell receptors and related processes in research and clinical practice. RNA sequencing provides insights into the T-cell transcriptome, helping to identify the diversity of TCRs and study their expression [62]. This method allows for a deeper understanding of T-cell dynamics and functionality in response to various stimuli and pathological conditions. The Nanostring technology, an innovative approach for analyzing gene transcriptional activity and protein translation, helps identify signatures and molecular markers specific to different T-cell subpopulations, and characterizes their functional activity [63]. The Polyfunctionality method, particularly using Isoplexis, enables the analysis of the secretome of individual cells, allowing simultaneous assessment of multiple T-cell functions, including their effector functions and cytokine production capabilities [64]. This is essential for better understanding the functional characteristics of T cells and their potential in immunotherapy. Additionally, 3D RNA sequencing is a relatively new technique that allows for the analysis of gene expression within the context of the cellular architecture of tissues [65], making it valuable for studying TCR interactions with surrounding cells and their microenvironment.

Advancements in these molecular genetic technologies provide researchers and medical professionals with a deeper and more detailed understanding of T-cell functionality and their roles in the immune system and various pathologies. These tools are contributing to improvements in diagnosis and the development of personalized TCR-based therapies, thereby opening new avenues in immunotherapy and biomedical research.

Tcrs production

The design and synthesis of genetic constructs are not always performed within a single institution. Design is primarily an R&D task and is often carried out by research departments, while the synthesis of constructs is frequently outsourced to commercial companies [66,67,68]. This process is followed by T-cell transfection, a technique used to insert genetic information into T cells to enhance their antitumor potential.

Numerous technological platforms have been developed for the genetic modification of primary T cells, including non-viral platforms such as transposon-based systems (PiggyBac, Sleeping Beauty) [69], TALEN [70], CRISPR-Cas9 [71], electroporation [72], and gene delivery technologies utilizing chemical compounds like polymers and lipids [73]. The production of viral vectors encoding TCRs remains the most commonly used method in both preclinical and clinical settings [74]. Despite the existing risk of random gene integration into critical regions of the genome, viral vectors provide significantly higher gene delivery efficiency compared to non-viral transduction methods [5]. They are capable of integrating genetic material into the host cell genome, ensuring long-term expression [75].

Viral vectors, such as retroviruses and lentiviruses, integrate genetic material into the genome of T cells to deliver the TCR gene [76,77,78]. Given that transduction using viral vectors is currently the most popular method for genetically modifying primary T cells, this review will focus on assessing the efficiency of this specific method.

Assessing the efficiency of TCR-T cell transduction

The efficiency of TCR-T cell transduction is a crucial parameter, as it directly impacts the number of genetically modified cells capable of recognizing and destroying tumor cells. High transduction efficiency ensures a sufficient number of functional TCR-T cells to achieve a therapeutic effect. Conversely, low efficiency may result in an insufficient number of active cells, reducing the overall efficacy of the therapy and the likelihood of achieving clinical remission. Moreover, evaluating transduction efficiency helps determine how well the technology is performing and what improvements can be made in the cell modification process [79].

The efficiency of genetic transduction can be assessed using various methods, including fluorescent labeling of marker genes, real-time PCR, monitoring of transgene expression levels, and analysis of transgene integration into the T-cell genome.

Fluorescent labeling of marker genes is an effective tool for evaluating the efficiency of TCR-T cell transduction [80, 81]. This method allows researchers to track and quantify successful transgene insertion and monitor the expression of markers in cells. Fluorescent labeling involves inserting a marker gene that codes for a fluorescent protein into engineered T cells. After successful transduction, these T cells can be examined under a microscope or by flow cytometry to determine the presence and intensity of fluorescence.

The design of fluorescently labeled tetrameric MHC–peptide complexes allows for direct visualization, quantification, and phenotypic characterization of antigen-specific T cells via flow cytometry [82]. MHC–peptide tetramers are complexes composed of four fluorophore-labeled MHC molecules, each of which binds to a specific peptide. Due to their tetrameric structure, these complexes can bind to up to three of the four monomeric MHC units, enhancing the avidity of their binding to targets [83]. Multimer binding is analyzed to verify the proper assembly and correct conformation of TCRs [1]. For example, Ishihara M. et al. used a retroviral vector encoding NY-ESO-1 157-165/HLA-A*02:01-specific TCR-α and TCR-β chains with increased affinity for interfering RNA constructs that specifically inhibit endogenous TCRs [52]. The efficiency of retroviral transduction was assessed using CD8 + and CD4 + T cells from healthy volunteers, demonstrating that 87.6% of CD8 + T cells and 89.0% of CD4 + T cells were tetramer positive.

The efficiency of TCR-T cell transduction can also be assessed through real-time polymerase chain reaction (PCR) [84]. This method allows for the quantification of the presence and expression levels of marker genes, which is crucial for monitoring and improving transduction and therapy effectiveness. Real-time PCR relies on DNA amplification using fluorophores, enabling real-time monitoring of the amount of PCR products [85]. When evaluating TCR-T cell transduction efficiency, the marker gene inserted alongside the TCR gene expresses a fluorescent protein, and its expression level is measured throughout the amplification process. Real-time PCR quantifies the level of transduced TCR-T cells and the expression level of the marker gene, which is essential for optimizing experimental conditions. Additionally, this method is used to monitor the stability of the marker in transduced T cells over extended periods.

Efficiency assessment

The “Efficiency Assessment” stage consists of two key steps: in vitro and in vivo proof of concept, aimed at comprehensively verifying the effectiveness of the T cell-based therapy under investigation (Fig. 3).

Fig. 3

Flowchart of the “Effectiveness Assessment” stage. AG: antigen; TCR: T-cell receptor

In vitro proof of concept

Regardless of their origin, selected candidate TCR sequences must be tested to confirm their specificity and effectiveness [86]. Cell proliferation and cytotoxicity assays are used to determine whether the tested molecules affect cell proliferation and are capable of exhibiting direct cytotoxic effects.

The effectiveness of TCR-T cells largely depends on their ability to destroy target tumor cells and induce their lysis [87]. Cytotoxicity assays, where T cells are co-cultured with tumor cells or human cell lines, are the most common methods for assessing the antitumor activity of TCR-T cells in vitro [2].

There are numerous assays based on various functional characteristics of cells, including enzyme activity, cell membrane permeability, adhesion, ATP production, coenzyme production, nucleotide uptake activity, and cell proliferation [88]. To choose the optimal viability assay, one must carefully consider the cell type, culture conditions, and specific study objectives. Quantification of specific target cell lysis induced by modified T cells is typically performed using methods such as the chromium (⁵¹Cr) release assay, bioluminescence imaging (BLI) using luciferase [89], impedance-based cell analysis and flow cytometry [87]. These methods are often combined with approaches that measure the quantitative release of effector cytokines and/or degranulation markers, which are indicators of T-cell activation in response to specific signals from target cells, and they allow for the assessment of modified T cell functionality [2].

Cr-release assay

The ⁵¹Cr release assay, based on radioactive chromium labeling of cells and quantitative measurement of target cells with compromised plasma membranes, was first described over fifty years ago [90] and has become a standard for measuring cytotoxicity induced by T cells and NK cells [87]. In this method, target cells are labeled with radioactive chromium (⁵¹Cr) and incubated with effector cells at varying effector-to-target ratios. During effector-mediated killing, the target cells lose plasma membrane integrity, releasing ⁵¹Cr into the culture medium. The radioactivity measured in the supernatant using a gamma-ray scintillation counter is proportional to the number of target cells destroyed during a given period [87]. In the work by Ishihara M. and colleagues, cytotoxicity assays using the chromium release assay (Cr-release assay) demonstrated that NY-ESO-1 TCR-T cells effectively eliminated NY-ESO-1-positive melanoma cell lines in vitro [52].

Measurement of lactate dehydrogenase (LDH) activity

Another common method for assessing cytotoxicity is based on measuring the activity of cytoplasmic enzymes released by damaged cells. Lactate dehydrogenase (LDH), a stable cytoplasmic enzyme present in all cells, is rapidly released into the cell culture supernatant when the plasma membrane is damaged, and its activity can be easily quantified [91].

Bioluminescent imaging

Bioluminescence imaging (BLI) is another widely used technology for cytotoxicity assessment, which measures the activity of cytoplasmic enzymes released by damaged cells. BLI-based cytotoxicity assays utilize the luciferase reporter gene to measure the light signal (photons) emitted by target cells. This signal is produced exclusively by living cells expressing the luciferase transgene, so the cytotoxicity of effector cells can be assessed based on the decline in BLI signal intensity [92]. Mensali et al. demonstrated the cytotoxicity of electroporated TCR-T cells specific to the colorectal cancer cell line MSI + HLA-A2 + HCT 116 using a BLI-based assay [93].

Impedance analysis

Impedance-based cell analysis is an alternative to the ⁵¹Cr and BLI-based assays. This method uses microelectrodes embedded in the bottom of each well of a microtitration plate to measure the impedance of electric current between the electrodes during cell adhesion [87]. In the absence of cells, impedance (a measure of electrical resistance between electrodes) is near-zero; however, cell adhesion alters the impedance depending on the number, morphology, viability, and adhesion strength of each cell type [94]. Suspension cells or non-adherent cells typically do not significantly contribute to electrical impedance or cause much smaller changes compared to adherent cells. This principle is used to measure the cytolysis of adherent target cells by non-adherent effector cells such as T cells [95]. In the study by Jin and colleagues, the effector functions of E7 TCR T cells, including IFN-γ production and tumor cell lysis, were demonstrated in vitro. The ability of E7 TCR T cells to kill tumor cells was evaluated using a real-time impedance-based cytolysis assay [96].

Flow cytometry

Unlike the three methods mentioned above, flow cytometry analysis allows for the study and quantification of cytotoxicity within heterogeneous cell populations, offering a unique advantage when investigating the sensitivity of different cell types to target cells within a mixed population [97]. Flow cytometry distinguishes between target and effector cells based on properties such as cell size and granularity (as indicated by forward and side scatter, respectively), as well as specific staining with fluorescently labeled antibodies [87]. Cell death is typically assessed using fluorescent DNA intercalating agents such as propidium iodide or 7-aminoactinomycin D, which are preferentially absorbed by dead cells and exhibit spectral shifts when interacting with DNA [98].

Apoptosis stages can be identified by staining cells with annexin V, which specifically binds to phosphatidylserine expressed on the cell surface during apoptosis [99]. Standard methods involve assessing the cytotoxicity of engineered T cells against cell lines stained with cell-penetrating dyes, such as CFSE (5-(and-6)-carboxyfluorescein diacetate succinimidyl ester), which interact with intracellular free amines, followed by flow cytometry analysis [97, 100].

Therefore, flow cytometry enables simultaneous analysis of both target cell elimination and effector cell phenotyping, as well as the separation of heterogeneous cell populations [87].

In vivo proof of concept

The application of TCR-T cells in animal models is essential for verifying their viability, functionality, and persistence in vivo [1]. A significant challenge in these studies is the requirement to infuse human cells into animals.

Various mouse models are employed in experiments with engineered T cells, including humanized mouse models, syngeneic mouse models, and HLA class I transgenic mice [11]. However, most preclinical in vivo studies to date have utilized xenograft models derived from human cell lines, where tumor cells are grafted onto immunodeficient mice prior to the infusion of TCR-T cells [11].

For instance, in a study using a mouse sarcoma model resistant to immune checkpoint inhibition, Ishihara M. et al. investigated the preclinical efficacy of TCR-T cells in combination with a vaccine based on long peptide antigens (LPA) loaded into pullulan nanogel nanoparticles [52]. The LPA consisted of 38 amino acid residues containing three murine CD8 + T-cell epitopes, including 9 m. CMS5a murine fibrosarcoma cells were subcutaneously implanted into wild-type BALB/c mice. DUC18 CD8 + T cells were isolated from DUC18 transgenic mice expressing a murine TCR that recognized the 9 m epitope of the neoantigen, resulting in TCR-T cells capable of recognizing CMS5a tumors. The combination of the LPA-containing vaccine and DUC18 TCR-transgenic T cells successfully destroyed CMS5a sarcoma, whereas T cells alone were ineffective [52].

The in vivo ability of TCR-engineered T cells targeting E7 to kill cancer cells was also investigated using xenograft mouse models. The administration of TCR-engineered T cells targeting E7 led to complete regression of subcutaneous 4050 tumors and suppressed the growth of CaSki tumors in NSG mice [96].

Despite their utility, xenograft models have several limitations. These models do not accurately replicate components of the human tumor microenvironment, such as the presence of immunosuppressive cells. Additionally, engineered TCRs interact with peptide-HLA complexes, but major histocompatibility complex molecules are specific to humans. This species difference means that experiments using HLA-transgenic mice may not fully capture the complexity of human immune responses.

Given the similarities between the immune systems of humans and primates, primates are considered potential large animal models for better toxicity assessment of engineered T cells. However, challenges arise from the need to obtain autologous engineered T cells, and the natural occurrence of tumors in primates is relatively rare [11]. Nonetheless, primate models closely replicate the development of cytokine release syndrome and immune effector cell-associated neurotoxicity syndrome, which are among the most frequent and potentially fatal immune-related adverse events during engineered T cell therapy [101,102,103].

Spontaneous cancers in dogs share many similarities with human cancers in terms of clinical presentation, histological features, molecular profiles, and treatment response [104]. The biological similarities between cancer in dogs and humans also offer the possibility of testing engineered T cells in dogs with naturally occurring tumors [11]. However, there are limitations due to ethical considerations, such as the need to select a group of animals with relevant spontaneously occurring tumors, and the potential risk of anaphylactic reactions when injecting human biological products into dogs [105]. Nonetheless, studies are being conducted on CAR-T and TCR T-cell therapies for dogs, the results of which may serve as a model for the potential application of these T-cell therapies in humans [106, 107]. Conversely, the results of clinical trials in humans may contribute to the development of T-cell therapy in veterinary oncology [108].

The recent emergence of human ex vivo models, particularly organoids and organotypic systems, has the potential to elevate preclinical TCR-T cell research to a new level in the future, as these models can mimic human biology and provide personalized predictions of certain types of toxicity [11]. For example, van Amerongen et al. proposed using human-induced pluripotent stem cells, which can be differentiated into specific organoids of vital tissues, for toxicity studies [109]. Another study suggested using organoids derived from a patient's tumor tissue as a novel preclinical model [110]. This approach preserves key biological characteristics of the original tumor. Li et al. demonstrated that compared to lung cancer xenograft models, the success rate of culturing increases with patient-derived organoids, while the time and cost of creating the model are significantly reduced. These organoid models are also expected to be more personalized, offering better predictions of in vitro anticancer therapy efficacy. A recent review highlighted the potential of three-dimensional (3D) tumor models, such as spheroids from cervical cancer cell lines and patient-derived organoids, for evaluating new treatment modalities, particularly immunotherapy targeting tumor cells and modulating the tumor microenvironment [111]. Studies have already successfully used organoids to assess the cytotoxic effects of CAR-modified cells [112, 113].

In conclusion, the in vivo toxicity of adoptive T-cell therapy can currently be reliably assessed only in human clinical trials. Therefore, toxicity assessment primarily involves determining whether human tissue contains the target antigen and conducting in vitro tests to check TCRs for cross-reactivity against human peptides [96], as will be discussed in the following sections.

Safety assessment

The safety assessment includes stages such as verifying the biodistribution of the tested T cells and evaluating their toxicity. This process involves determining virus-specific toxicity, off-tumor toxicity, and cross-reactivity (Fig. 4).

Fig. 4

Flow diagram of the “Safety Assessment” stage. AG: antigen, TCR: T-cell receptor

In vivo biodistribution

In preclinical trials, the biodistribution stage using animal models is typically conducted alongside the in vivo proof-of-concept stage described earlier.

Studies in mice have shown that adoptively transferred T cells can localize and kill target cells expressing a cognate antigen, regardless of their anatomical location. For example, in the experiments by Hinrichs C.S. et al., mice received T cells targeting gp100, a melanocyte differentiation antigen expressed by both normal melanocytes and the B16 melanoma line. The T cells migrated indiscriminately to all tissues [24]. However, the engineered T cells exhibited effector functions only in gp100-expressing tissues, indicating an antigen-specific response during their ubiquitous trafficking [114]. In this mouse model, tumor regression was correlated with autoimmunity in the skin and eyes [115].

Nevertheless, animal models used for assessing T-cell biodistribution have several limitations. These models do not fully replicate the components of the human tumor microenvironment, such as the presence of human immunosuppressive cells. Moreover, they are not suitable for predicting the migration of T cells to metastatic tumor sites throughout the patient's body [96]. These challenges in biodistribution analysis remain unresolved, and new approaches are needed to address them.

Toxicity assessment

Virus-specific toxicity

Genotoxicity is a significant concern in adoptive cell therapy. Various tests are conducted to address the potential problems associated with this type of toxicity. Since engineered T-cell receptors are primarily delivered via viral vectors, safety concerns arise due to the risk of insertional mutagenesis and cellular transformation [116]. Clinical experience shows that these risks vary depending on the cell type. For example, the risk of transformation is higher for hematopoietic stem cells than for T cells [117].

Severe adverse events caused by insertional mutagenesis have been observed when using genetically modified hematopoietic stem cells, but no such adverse effects were observed during long-term follow-up of patients who received engineered T cells modified with retroviral constructs [11].

The use of lentiviral and retroviral vectors for generating TCR-T cells carries a risk of insertional mutagenesis. Random vector integration may result in uncontrolled genotoxic effects, cell proliferation, and clonal expansion, potentially leading to multiple primary malignancies [118].

Clonal expansion of engineered T cells due to vector insertion into the TET2 gene (a tumor suppressor gene and a master regulator of blood cell formation) has been documented in several studies. Insertion disrupted the gene’s structure and function, leading to the expansion of a therapeutically effective and significant cell clone [119]. Researchers expanded the study cohort to 39 patients, and Nobles et al. developed a model predicting the response to chronic lymphocytic leukemia treatment with CD19-specific CAR-T cells based on vector integration sites [120]. Another case of clonal expansion involved vector insertion into the gene encoding E3 ubiquitin–protein ligase CBL during the production of CAR-T cells targeting CD22 [121].

Adoptive cellular products have also been developed using transposon-based delivery methods. Early-phase clinical trials of CAR-T cells produced via the Sleeping Beauty system showed no severe toxicity [122,123,124,125]. The PiggyBac (PB) system was also successfully used to generate CAR-T cells [126], but later trials reported cases of CAR-T cell lymphoma [127, 128].

Recent studies suggest a potential link between genetic modification of T cells and their possible oncogenic transformation following CAR-T cell therapy [129, 130]. Several possible mechanisms underlying the development of secondary malignancies, such as lymphomas or leukemia, after CAR-T therapy are currently under discussion. One key factor is the genetic alterations that occur during T cell modification, such as the use of retroviral vectors for CAR delivery. These alterations may accidentally activate oncogenes or inactivate tumor suppressor genes, promoting cancer development [131].

A recent systematic review examines non-relapse mortality (NRM) following CAR-T cell therapy, including rare cases of secondary malignancies [132]. According to the meta-analysis, approximately 7.8% of NRM cases were caused by secondary malignancies. These findings highlight the need for further research and long-term monitoring of patients following CAR-T therapy.

Potential insertional mutagenesis can be evaluated by studying the vector integration patterns used in TCR-T cells [133]. Different variations of Integration Site Analysis are applied to identify integration sites [134]. While the original analysis was PCR-based, bioinformatic tools for processing integration site sequencing data are continuously being developed and improved [135, 136].

The field of genetic engineering has reached new heights with the use of site-specific genomic modifications, particularly with the CRISPR-Cas system in adoptive cell therapy. However, poorly predictable off-target genomic cleavage may occur. These off-target effects include insertions or deletions at unintended sites, which could result in chromosomal translocations between the target and off-target sites or even between different off-target sites [11]. Several genome-wide techniques for assessing off-target cleavage are now in use, including methods for detecting chromosomal translocations in engineered cells [137]. Various bioinformatic tools are available for predicting off-target effects when using CRISPR-Cas or other nucleases, including MIT CRISPR Design Tool [138], E-CRISP [139], GUIDE-seq [140], Digenome-seq [141], CIRCLE-seq [142], SITE-seq [143], and CAST-seq [137].

It’s important to note that the number and efficiency of off-target sites depend heavily on the specific gRNA. In silico methods may not predict experimentally identified off-target sites (leading to false negatives), while in vitro methods may result in false positives, reducing specificity [11, 144].

Off-tumor toxicity and cross-reactivity

Ideally, engineered T cells should selectively target malignant cells. However, target antigens are often expressed on both tumor and healthy tissues, raising concerns about on-target, off-tumor toxicity [145]. The severity of toxic effects depends on how accessible, widespread, and vital the target tissue is [11].

Cross-reactivity of TCRs represents a major safety hazard in TCR gene therapy. T-cell receptors recognize peptides presented by human leukocyte antigen (HLA) class I and II molecules. TCR binding involves interactions between loops of the complementarity-determining region and amino acid residues on HLA molecules and their presented peptides [146]. Thus, TCR binding includes both a peptide-independent HLA-binding component and a peptide-specific component [146]. Both components are crucial for achieving the binding affinity required for T-cell activation. Cross-reactivity occurs when the engineered TCR recognizes one of the autoantigens presented by the HLA allele used to recognize the cognate ligand, leading to off-tumor toxicity. This is especially critical if antigens targeted by the engineered T cells are expressed by essential normal tissues, even at very low levels [24].

Cross-reactivity can also arise when TCRs interact with different HLA alleles that present a variety of peptides, to which the TCR is intolerant. When engineering TCR-T cells, it is important to consider the HLA alleles present in patients. In the case of MHC allele mismatch, the designed TCR may not exhibit the required tolerance to HLA molecules [11].

The risk of autoimmune toxicity increases when using TCRs that have not undergone thymic negative selection, which is designed to remove autoreactive T cells. These include TCRs derived from immunized mice or artificially enhanced TCRs with mutations in the complementarity-determining region that increase affinity [24]. Although effective, this approach increases the risk of cross-reactivity and the recognition of non-specific peptides [11].

Both cross-reactivity scenarios must be evaluated in preclinical trials. Engineered T cells can strongly react with HLA alleles absent in the TCR donor, and allogeneic HLA molecules are among the most immunogenic antigens. Alloreactivity of therapeutic TCRs can be avoided by extensive in vitro screening using cell lines expressing different HLA alleles and by testing TCR-expressing engineered T cells on patient cells before treatment [11].

Additionally, off-target, off-tumor toxicity can result from cross-reactive binding to a mimotope—an epitope that is similar, but not identical, to the target, and is expressed on normal tissues. This off-target binding to cell surface proteins is difficult to predict in preclinical animal trials [11, 13].

To evaluate this risk, the fine specificity profile of the therapeutic TCR must be identified by varying individual residues in the cognate peptide and observing which changes result in the loss of T-cell activation [11].

Several methods for identifying critical amino acid residues for TCR-pMHC interactions are available. Alanine scanning is the most common method, involving the systematic replacement of each peptide residue with alanine. This identifies the combination of amino acids necessary for TCR binding to the pMHC complex [147, 148]. However, alanine scanning does not account for the potential substitution of an amino acid with one having similar physicochemical properties. Mutational position scanning (X-scanning), which uses a peptide library where each epitope residue is sequentially substituted with all possible amino acids, overcomes this limitation [149]. Higher-throughput methods include positional scanning with synthetic combinatorial libraries, where each amino acid is fixed at each position while the remaining peptide consists of random combinations of other amino acids [150,151,152]. The resulting sequences can be compared to the human proteome to predict potential cross-reactive epitopes.

Furthermore, 3D modeling and structure-based analysis strategies using X-ray crystallography data of TCR-pMHC complexes are also available [153, 154]. The extensive structural data collected on TCR-pMHC complexes allow machine learning algorithms to predict TCR-pMHC binding and potential cross-reactivity [155].

Due to the lack of adequate animal models, the on-target toxicity of engineered TCR-T cells cannot be effectively assessed in preclinical studies. This remains a technically challenging task [116].

Conclusions

In this review, we have summarized the requirements outlined in current guidelines for testing genetically modified TCR-T cells and proposed a flow diagram for conducting preclinical trials. Despite significant progress in this field, there remain knowledge gaps in certain stages of preclinical testing of modified T cells, as no existing in vivo models can fully demonstrate the biodistribution and potential toxic effects of these cells. Therefore, further research and clinical experience are necessary. The findings from ongoing studies should be carefully analyzed to improve and standardize the technologies associated with preclinical trials of TCR-T cells.