Introduction

In recent years, immunotherapy has attracted considerable research attention as a new modality of cancer treatment. Among the newly developed cancer immunotherapy technologies, those using chimeric antigen receptor-modified T cells (CAR-T) have been of particular interest. Substantial progress has been made in the CAR-T-based cancer immunotherapy field following the initial generation of CAR-T in 1989 (Kershaw et al. 2013; 2014; Wang and Riviere 2015). Currently, dozens of CAR-T clinical trials are ongoing worldwide (Fig. 1). To construct CAR-T, T cells are transduced with genes encoding fusion proteins for cancer antigen-recognizing single chain Fv (scFv) linked to the intracellular signaling domain of T cell receptors.

Fig. 1
figure 1

Clinical trials of CAR-T worldwide

Full size image

CAR-Ts are classified as first-, second- and third-generation, depending on the intracellular signaling domain numbers of T cell receptors (Fig. 2). First-generation CAR-T cells were designed to express scFv in the extracellular region and the signaling sequence of the T cell receptor intracellular domain with no co-stimulatory molecules. However, first-generation CAR-Ts were limited in their tumor cell-killing efficacy after specific recognition of tumor cells by antigens. To overcome these limitations, second- and third-generation CAR-Ts were designed to express co-stimulatory molecules in the intracellular domain (Fig. 2, Casucci and Bondanza 2011; Maus et al. 2014). Chimeric antigen receptor (CAR) gene cassettes for second-generation CAR-Ts encompass one co-stimulatory molecule, such as CD28 or 4-1BB. Third-generation CAR-Ts have been further developed to include two co-stimulatory molecules among CD27, CD28, 4-1BBand OX40.

Fig. 2
figure 2

Construction of CAR in each CAR-T generation

Full size image

Manufacture and administration of CAR-T

CAR-Ts are manufactured using three consecutive steps (Lee et al. 2012; Wang and Rivière 2015; Levine 2015). The first step is to generate genetic constructs of CAR to encode tumor antigen-specific Fv linked to signaling sequences of T cell receptors. Next, T cells are transduced with CAR using viral, nonviral or physical methods. Retroviral or lentiviral vectors have been successfully employed as viral vectors for transduction of T cells with CAR. In other studies, T cells have been transduced with plasmid DNA (Huang et al. 2012; Kumaresan et al. 2014; Wang et al. 2014a, b) or RNA (Zhao et al. 2010) encoding CAR via electroporation. The third step is cultivation of CAR-T cells.

Several protocols have attempted to activate T cells for transduction, one of which is to use anti-CD3-antibodies and cytokines, such as interleukin-2. Various other T cell activation methods are currently under investigation (Fig. 3). Antigen-presenting cells expressing 4-1BBL and Fc receptor have additionally been employed to activate T cells. Another procedure involves the use of beads modified with anti-CD3 and anti-CD28 antibodies for T cell activation. The resulting CAR-T cells are derived from CD4 or CD8 T cells, expanded using cytokines, and administered to patients via intravenous infusion (Lee et al. 2012).

Fig. 3
figure 3

Activation of T cells and expansion of CAR-T

Full size image

Advantages of CAR-T over existing cancer immunotherapy

CAR-Ts have attracted considerable research attention due to their potent efficacy against terminal cancers, relative to monoclonal antibodies and antibody–drug conjugates. Moreover, CAR-Ts act through different mechanisms from T cell receptor-engineered T cells (TCR-T). In TCR-T, TCRs recognize complexes of tumor antigens that are processed in APC cells and presented on APC cell surfaces with MHC class molecules. Unlike TCR-T, CAR-Ts do not require processing and presentation of tumor antigen-recognizing moieties with MHC molecules (Kershaw et al. 2014; Fig. 4). The differences between CAR-T and TCR-T are summarized in Table 1 (Gill and June 2015). Notably, the lack of MHC restrictions means that eligible patient groups for CAR-T are wider compared to those for TCR-T that requires the identification of patient MHC types.

Fig. 4
figure 4

Different cell surface structures between CAR-T and TCR-T

Full size image
Table 1 Differences between CAR-T and TCR-T
Full size table

Current status of CAR-T studies

The utility of CAR-T for treatment of lymphoma and solid tumors has been examined in several studies, grouped as cell-level (Table 2; Fig. 5), animal-level (Table 2; Fig. 5), preclinical (Table 3; Fig. 5), and clinical trials (Table 4; Fig. 6). Current status summary of CAR-T under basic investigational stages shows that retroviral vectors are most actively used for introducing CAR genes into T cells. Although the use of viral vectors are dominating, nonviral approaches using cationic polymers or electroporations have been attempted (Fig. 5a). Hematological cancers such as lymphoma have been the major target of CAR-T in basic research stage (Fig. 5b). Second-generation CAR-Ts have been most widely investigated in basic cell-level and animal studies, with focus on hematological cancers and solid tumors (Fig. 5c). The application of CAR-T has been extended from cancer immunotherapy to treatment of autoimmune diseases, such as multiple sclerosis in Europe (Fransson et al. 2012). Another recent study reported on the effectiveness of CAR-T in treating fungal infection, suggesting a new field of CAR-T application (Kumaresan et al. 2014). Similar to the investigational stage research, preclinical trial studies used retroviral vectors and second generation CAR constructions in higher frequency than other viral vectors (Fig. 5d), and generations, respectively (Fig. 5f). Notably, the number of preclinical trials for hematological tumors were lower than brain cancers (Fig. 5e).

Table 2 In vitro and in vivo CAR-T studies
Full size table
Fig. 5
figure 5

Current status of CAR-T therapeutics under investigational stages and preclinical trials. CAR-T therapeutics under investigational stages were analyzed by delivery vectors (a), target diseases (b), and generations (c). CAR-T therapeutics in preclinical trials were analyzed by delivery vectors (d), target diseases (e), and generations (f)

Full size image
Table 3 CAR-T in preclinical trials
Full size table
Table 4 CAR-T in clinical trials
Full size table
Fig. 6
figure 6

Current status of CAR-T therapeutics in clinical trials. CAR-T therapeutics in preclinical stages were analyzed by tumor antigens (a), target diseases (b), and generations (c)

Full size image

For introduction of CAR-T genes into T cells, viral vectors have been predominantly used. The duration of CAR-T survival in vivo is reported to be longer than several months. Since month-long sustainable expression of CAR can evoke undesirable side-effects, the optimal length of expression time requires further investigation. To shorten the duration of CAR expression and minimize safety concerns, physical electroporation studies are underway. Introduction of plasmid DNA encoding CAR directly into the cytoplasm of T cells via electroporation may lead to an expected duration of CAR of several days, while avoiding side-effects of delivery vectors.

Currently, more than 80 CAR-T cases are in clinical trials worldwide (www.clinicaltrials.gov). No clinical trials are in phase 3 as yet, but the potential is high, given the number in phase 2 (Table 4). The majority of CAR-T clinical trials is being held in USA, and has also been initiated in Asia, China and Japan (Fig. 1). CD19, widely studied as a tumor antigen target of CAR-T, is overexpressed on the surfaces of leukemia cells of acute lymphocytic leukemia (ALL) patients. CD19 CAR-T therapy is reported to be effective in children with recurrent ALL after bone marrow transplantation (Lee et al. 2015). Among the products in clinical trials, CTL019 (Novartis, USA), CD19-targeted CAR-T (CD19-CAR-T) against ALL has been recognized as a “breakthrough therapy” by the Food and Drug Administration of USA and is in phase 2 development. Several other global pharmaceutical companies are developing CAR-T products in the pipeline. In clinical trials, CD19 has been most extensively used as target tumor antigens of hematological cancers (Fig. 6a). Other antigens in clinical trials include carcinoembryonic antigen (CEA), human epidermal growth factor receptor 2 (HER2), GD2, CD30, and CD20 (Fig. 6a). Hematological cancers have been predominantly studied in clinical phase (Fig. 6b). Until now, phase 2 trial are the most advanced stage for CAR-Ts (Fig. 6c).

Challenges

Limitations of each generation of CAR-T

Although CAR-T takes advantage of the immune response of T cell killing abnormal cells, tumor cell-killing effects are decreased in solid tumors. This reduced antitumor activity is attributable to the immunosuppressive microenvironment of tumor tissues, resulting in low penetration efficiencies of CAR-T into solid tumor tissues. Moreover, leukocytes in tumor tissues are known to secrete factors that lower T cell activity.

Second-generation CAR-Ts containing a co-stimulatory signaling domain can induce the expression of immunosuppressive receptors, such as T cell membrane protein-3, cytotoxic T lymphocyte associated antigen 4 (CTLA-4), and programmed death-1 (PD-1). To overcome suppression of CAR-T activity by PD-1, the effects of co-administration of anti-PD-1 antibody with CAR-T were recently examined (John et al. 2013). An ongoing clinical trial (NCT00586391) is exploring the effects of co-administration of CAR-T and ipilimumab, an anti-CTLA-4 antibody (Maher 2014).

Third-generation CAR-Ts with two co-stimulatory molecules, such as OX40 and 4-1BB, display enhanced activity in vivo. However, excessive stimulation of T cell activity by two co-stimulatory molecules may induce an abrupt increase in cytokine secretion, known as ‘cytokine release syndrome’ (CRS). The onset of this severe side- effects one of the biggest safety concerns that require addressing for further successful development of CAR-T.

Side-effects

The biggest hurdle in CAR-T clinical trials is severe side-effects, the most acute being CRS. The mortality list of patients undergoing clinical trials of CAR-T highlights the need for design of improved clinical protocols and regulatory decisions of investigational new drug development applications. Symptoms of CRS include high fever, joint pain, muscle pain, low blood pressure, and dyspnea, with death in a few cases. In 2014, clinical trials were temporarily held in Memorial Sloan Kettering Cancer Center after deaths of two patients within two weeks after infusion with CD19-CAR-T from Juno Therapeutics (USA). CRS has been determined as the main cause of death of patients in clinical trials to date.

Although CRS is the most common side-effect related to CAR-T therapy, the development of CRS has been considered to be correlated to the response to therapy. In previous studies, it has been observed that CAR-T responsive patients developed CRS, whereas non-responsive patients did not develop CRS (Maude et al. 2015). The severity of CRS has been reported to be rather correlated with tumor burden at injection time of CAR-T. Given the importance of CRS in clinical monitoring, the markers which can predict the severity of CRS need to be identified. Peak levels of cytokines such as interferon-γ have been found to be more elevated in severe CRS than mild CRS (Davila et al. 2014). Other studies proposed C-reactive protein as an indicator of severe CRS (Maude et al. 2014). However, the decisive biomarkers for CRS still remains to be studied.

To minimize these side-effects, it is crucial to select the appropriate group among enrolled patients and optimize the CAR-T dose in clinical protocol design. Currently, anti-interleukin 6-antibody or steroid drugs are co-administered with CAR-T to reduce CRS (Davila et al. 2014), although further studies are required in this respect. In addition, safety studies assessing whether CAR-T can induce autoimmunity or graft-versus-host disease are warranted. Several studies have examined the efficacy of co-administration of cytokine inhibitors with CAR-T or a suicide gene, with the aim of reducing CRS. A recent study (Grupp et al. 2013) reported that the co-administration of CAR-T with tocilizumab, an anti-interleukin-6 antibody, alleviates CRS.

The majority of CAR-Ts for clinical trials have been constructed using viral vectors for CAR gene transfection. Although these vectors are designed to be non-replicating, a long-term study (over a number of years) should be performed to monitor potential replicative ability. Profiling and standardization of cytokines after CAR-T administration is necessary. The deaths of two patients in 2014 have highlighted CRS as the most severe limitation in clinical trials of CAR-T. This excessive increase in cytokine secretion results from amplification of CAR-T cells in vivo and may be sufficiently fatal to cause death. To monitor and control CRS, identification of the specific roles of individual cytokines and types of relevant cytokines is required to predict the risk of CRS in clinical trials. Moreover, profiling of cytokines and CRS marker cytokines during clinical trials should be standardized.

In the case of CD19-CAR-T, anticancer effects are accompanied by a certain level of side effects. The possibility of segregating efficacy from side-effects of CAR-T products should be explored, with the aim of establishing an optimal regimen with minimization of CRS. Autoimmunity of CAR-T is another potential concern. Low doses of CAR-T and serial tumor cell-killing effects of single CAR-T cells may lower the possibility of autoimmunity. However, the artificial nature of CAR-T may increase the risk of autoimmunity. Thus, to guarantee safety, regulatory concerns regarding the autoimmunity issue should not be overlooked.

In addition to CRS, another concern is “on-target/off-tumor” side effects. The side effect is due to that majority of target antigens for CAR-Ts are existing on both tumor and normal tissues, showing overexpression on tumor cells (Kakarla and Gottschalk 2014). To minimize the “on-target/off-tumor” side effects, the discovery of new target molecule exclusively expressing on tumor tissues would be necessary. Another approach is to remove inappropriately activated CAR-Ts. A recent study reported that small molecule drug AP1903 could induce caspase 9 and apoptosis in transduced cells, killing only activated cells expressing high levels of CAR (Gargett and Brown 2014).

Regulatory perspective

CTL019, a CD19-CAR-T designed by Novartis (USA), is the first known compound in the CAR-T class that has entered phase 2 clinical trials. From the regulatory perspective, manufacturing and clinical trials are major concerns (Puri 2014). Production issues include consistency of CAR-T products, patient-dependent variations in T cell transfection efficiency, optimal T cell types for CAR transfection, and labeling of CAR-T. The major clinical trial concerns include potency and safety.

Pharmacokinetics and biodistribution of CAR-T

Pharmacokinetics and biodistribution experiments provide essential information for predicting the possible side-effects of CAR-T. A recent study reported that CAR-T is distributed to the bone marrow after intravenous administration and circulates in the blood up to 10 months post-injection (Ritchie et al. 2013). Further quantitative analyses of pharmacokinetics, tissue distribution and retention in the body are essential. The tumor antigen, HER-2, has been shown to be expressed in brain tissues and the mammary gland (Wang et al. 2010). Carcinoembryonic antigen, another tumor antigen, has been identified in colon or normal tissues (Schölzel et al. 2000).The nonexclusive expression patterns of tumor antigens in normal tissues increase the possibility of normal tissue damage upon administration of CAR-T targeting these tumor antigens. From this viewpoint, regulatory considerations associated with the distribution and pharmacokinetics of CAR-T should be addressed.

Efficacy of CAR-T

One concern with regard to the efficacy of CAR-T is the consideration of optimal T cell subsets. The majority of CAR-Ts in clinical trials introduce CAR genes after isolation of total T cells from patients. The issue of whether the T cell subtype affects the transfection efficiency and potency of CAR-Ts should be examined.

Efficacy evaluation methods for CAR-T should be further explored. Currently, tumor cell lysis induction and cytokine secretion capabilities of CAR-Ts are under examination. Establishment of a relevant method reflecting the efficacy of CAR-T may facilitate standardized evaluation of CAR-T products. Moreover, owing to in vivo amplification, it is necessary to assess the relationships among the ratio of CAR-T within total administered T cells, CAR copy numbers and efficacy.

Unlike chemical or protein drugs, CAR-T is a living drug that amplifies in the body after administration. This in vivo amplification property leads to an increase in the actual effective dose of CAR-T. The severity of symptoms and ages of patients may serve as factors affecting the in vivo amplification efficiency of CAR-T. The discrepancy between administration and working doses is a unique feature of CAR-T. Further analysis of the relationships between administered and working doses and therapeutic effects is warranted.

The specific T cell subtypes contributing to anticancer activity need to be identified. Currently, CAR-T is administered as mixtures of various T cell subsets. A recent study reported that increasing the frequency of CD8(+)CD45RA(+)CCR7(+) CAR-T cells, a subset closest to T cell memory stem cells, within total CAR-T enhances anticancer activity in an animal model (Xu et al. 2014).

Clinical trials of CAR-T

Several issues require clarification with regard to the optimal dose of CAR-T. As a result of in vivo amplification of intravenously infused CAR-T, the initial doses infused are not the same as the actual working cell number. Moreover, adjustment criteria of doses should be fixed between the weight and body surface of patients. Considering the amplification of CAR-Ts in bone marrow after infusion, we need to ascertain whether measurement of CAR-T numbers in the blood reflects the amplification extent in bone marrow.

Manufacture and quality control of CAR-T products

To validate consistency in CAR-T quality among batches, regulatory studies on chemistry, manufacturing and control (CMC) are essential. Analysis of quality control for each step of CAR-T production is important (Fig. 7). Quality control should be performed to maintain transfection efficiency of CAR among different batches, with parameters including the acceptable ranges of gene-modified CAR-T ratios among total T cells and quantification of copy numbers of the CAR gene per cell. For consistent production of CAR-T among batches, standardization of the stock of viral vectors is required to provide constant multiplicity of infection. Moreover, effects of patient age and medical treatment history on transfection of T cells with CAR-encoding vectors should be assessed.

Fig. 7
figure 7

Quality control flow chart for production of CAR-T

Full size image

Viral vectors, such as retroviral or lentiviral vectors, are frequently used for introduction of CAR into T cells (Wang and Riviere 2015). Quality control of viral vectors for CAR gene delivery should be performed in terms of purity, safety, T cell transfection efficiency, and physicochemical characterization.

Moreover, for quality control, validation of CAR-T sterility is essential. Since CAR-T is manufactured ex vivo by isolation of T cells, introduction of CAR genes, amplification, microbial assays and the scope of microorganisms for CAR-T need to be established.

Production and distribution of CAR-T need to be standardized. After conceptual design and proof-of-concept studies, production techniques have mostly been transferred from the laboratory benches of academia to industry. The manufacturing processes of CAR-carrying viral vectors and CAR-T are complex and differ among developers. CAR-T products under clinical trials are generated and distributed to patients using different protocols. The production and distribution processes of gene-modified cells are sufficiently crucial to affect quality. Quality control and standardization of manufacturing and distribution processes should thus be performed under good practice principles.

Labeling of CAR-T products should be carefully assessed. Given the autologous nature of CAR-T in which patient T cells are transfected with CAR genes and infused back into the same patient, it is important to clarify labeling of CAR-T to minimize the fatal risk of potential administration to the wrong patients.

Conclusions

CAR-Ts have attracted considerable research attention as a novel and potent modality of cancer immunotherapy. Global pharmaceutical companies have started investing in CAR-Ts, with several products in the pipeline for approval. However, substantial regulatory issues for CAR-T need to be addressed. The living nature of CAR-T necessitates careful assessment of safety and efficacy issues. The ex vivo manufacturing process of CAR-T highlights the significance of validating sterility and extensive product quality control. Further focus on regulatory studies and establishment of regulatory science-based guidelines may expedite the development of effective CAR-T products for patient use.