Introduction

T cell malignancies encompass a heterogeneous group of diseases, each reflecting a clonal evolution of dysfunctional T cells at various stages of development. T cell acute lymphoblastic leukemia (T-ALL) accounts for 15% and 25% of childhood and adult ALL cases respectively, and is the most common form of T cell cancer seen in children [1, 2]. T-lymphoblastic lymphoma (T-LLy) is a non-Hodgkin lymphoma with similar biology to T-ALL. Adult T cell leukemia/lymphoma (ATLL) is an extremely aggressive form of blood cancer driven by the human T cell lymphocytic virus type 1 (HTLV1) [3,4,5]. Other rare forms of T cell leukemia include T cell large granular lymphocytic leukemia (T-LGL) and T-prolymphocytic leukemia (T-PLL) [6]. T cell lymphomas are broadly divided into two categories, cutaneous T cell lymphoma (CTCL) and peripheral T cell lymphoma (PTCL) [7]. Mycosis fungoides (MF) and Sezary syndrome (SS) represent the two most common subtypes of CTCL, accounting for the majority of cases [8]. PTCL can be classified into several different subtypes, among which include anaplastic large cell lymphoma (ALCL), angioimmunoblastic T cell lymphoma (AITL), extranodal natural killer (NK)-T cell lymphoma (ENKTL), enteropathy-associated T cell lymphoma (EATL), hepatosplenic T cell lymphoma (HSTCL), and PTCL-not otherwise specified (PTCL-NOS) which is the most common of the group [9, 10].

The overall prognosis for T cell malignancies varies depending on the type of disease, but in general is much poorer when compared to B cell malignancies. While the survival in T-ALL/LLy has significantly improved with the intensification of chemotherapy, there still remain very limited options for patients with relapsed/refractory disease [11,12,13]. ATLL remains a very challenging disease to treat, with a median survival of less than 12 months for the acute form of this disease [3,4,5]. Advanced stage CTCL has a median overall survival of 5 years [14, 15], whereas outcomes of PTCL vary depending upon the subtype, with ENKTL, EATL, and HSTCL having the poorest prognosis [9, 10]. While immunotherapy has revolutionized the treatment landscape of various cancers with the use of monoclonal antibodies, checkpoint inhibitors, bispecific T cell engagers, and chimeric antigen receptor (CAR) T cell therapy, only limited responses have been seen in T cell disease [15]. Some promising results have been seen with use of brentuximab vedotin, a CD30-directed immunotoxin, in CD30-positive PTCL and CTCL [16, 17] and the use of pembroluzimab, a programmed cell death receptor 1 (PD-1) inhibitor, in the treatment of ENKTL [18]; however, these positive results have been limited to very specific subsets of T cell disease. One form of immunotherapy that has not yet been successfully translated to T cell malignancies is that of chimeric antigen receptor (CAR)-based immunotherapy. CAR T cell therapy has been extremely successful in relapsed/refractory B cell malignancies as evidenced by the recent Food and Drug Administration (FDA) approval of two CAR T cell therapeutics for this disease [19,20,21,22,23]. However, implementing this technology to treat T cell malignancies has been difficult, primarily due to the lack of a tumor-specific surface antigen in cancerous T cells. In this review, we will discuss the challenges involved in translating this novel technology to T cell disease, review all the preclinical and clinical progress made in adapting this therapy for this challenging disease, and examine potential solutions for the future development of this innovative therapy.

CAR T cell therapy

Genetic engineering of primary T cells was first presented in the late 1980s [24]. Since then, chimeric antigen receptor T cells have emerged as a promising technique for the treatment of relapsed/refractory malignancies. CAR therapy brings together numerous fields including immunology, tumor biology, genetic engineering, synthetic biology, and pharmacology. CARs are comprised of the intracellular signaling domain from the natural T cell receptor (TCR), CD3ζ, linked to a single-chain variable fragment (scFv) which serves as the antigen recognition domain. The scFv sequence is derived from a monoclonal antibody by combining the variable heavy (VH) and light (VL) domains using a small peptide linker. Commonly used CARs also include one or two costimulatory domains, such as CD28, 4-1BB, ICOS, and/or OX40. Although the kinetics have yet to be fully elucidated, it is essential that CAR T cells have mechanisms of trafficking to the tumor site where they can recognize their cognate antigen. This results in CAR T cell activation and expansion, and ultimately cytolytic activity against cells expressing the target antigen. CAR-based ligand recognition is advantageous compared to TCR-based ligand recognition because CAR-targeting is not restricted by major histocompatibility complex (MHC) interactions. Therefore, CARs can recognize cell surface proteins that have not been processed and presented by antigen presenting cells (APCs). Importantly, the interactions between scFvs and ligands have much higher affinity and avidity compared to that of TCR-ligand interactions [25]. Furthermore, the immune synapse formed from the interaction between a CAR and its ligand likely results in a much greater functional avidity than is observed using a targeted antibody approach with the same antibody (25).

CARs targeting the B cell antigen CD19 have been studied extensively for the treatment of B cell malignancies. In 2017, the FDA approved the first CAR T cell therapy, Kymriah, a CD19-directed CAR therapy for the treatment of relapsed/refractory B cell acute lymphoblastic leukemia (B-ALL) and in 2018, Yescarta was approved to treat relapsed diffuse large B cell lymphoma (DLBCL). These therapies, including others in clinical trials, have been widely successful in eliminating malignant cells and re-inducing remission in patients who were otherwise treatment-refractory [19,20,21, 26, 27]. Patients receiving CAR therapy undergo leukapheresis resulting in the collection of T cells, which are subsequently modified using a lentiviral or retroviral vector to express the CAR. These cells are expanded ex vivo while the patient undergoes lymphodepletion, a process involving chemotherapeutic agents. Finally, the CAR T cells are re-infused into the patient [28]. Lymphodepletion prior to re-infusion of the autologous T cells has been shown to augment both CAR T cell proliferation as well as persistence [29,30,31]. The administered dose of CAR T cells and the pre-existing tumor burden do not appear to be the sole determinants of the degree of T cell expansion, engraftment, and overall response. Other factors may be involved, such as the density of cognate antigen expression on the cancer cells [32]. However, the optimal degree of persistence of CAR T cells required to prevent leukemic relapse has not been determined [25, 33].

One of the mechanisms of relapse post-CD19 CAR T cell therapy is due to surface antigen escape with relapsed leukemia cells being CD19-negative. The mechanism may be due to the expansion of a small subset of CD19-negative cancer cells or alternatively, the cells may downregulate CD19 from the cell surface in order to evade detection by CAR T cells, rendering them resistant [19, 21, 34,35,36,37]. Additionally, it was recently shown that a phenomenon referred to as trogocytosis is a mechanism of antigen escape whereby the antigen is transferred to the CAR T cell [38]. It has also been shown that transduction of a single leukemic blast with an anti-CD19 CAR that was re-infused into a B-ALL patient, ultimately resulted in relapse and death of the patient [39]. Transduction of the leukemic cell resulted in masking of the target antigen through interactions between the CAR and the cognate antigen on the same cell. Clonal expansion of this population resulted in resistance to CAR therapy. This report emphasized the importance of strict and perfect isolation of normal, healthy T cells for modification with the CAR construct. As we discuss below, this is particularly challenging in T cell leukemia patients who are more likely to have circulating cancerous T cells, and therefore have a higher probability of these cells being inadvertently isolated, transduced, and re-infused.

Of note, there are severe toxicities that have been associated with CAR therapy. Cytokine release syndrome (CRS) is a systemic inflammatory response directly resulting from robust T cell activation following infusion. IL-6 is one pro-inflammatory cytokine that is secreted at high levels during CRS. During a particularly severe CRS condition, tocilizumab, an IL-6R antagonist monoclonal antibody, was used to rapidly and effectively reverse the symptoms of a pediatric patient [27]. Tocilizumab has since been FDA approved for treatment of CAR T cell-induced life-threatening CRS [40]. Neurological toxicities have been reported following CAR T cell infusion as well; however, preventative approaches remain elusive [36, 41,42,43,44]. Compared to CRS and neurotoxicity, a much more manageable consequence of CAR T cell therapy targeting B cell malignancies is the resulting B cell aplasia. This is a potentially lifelong outcome due to memory cell formation against a B cell antigen; but currently is managed by periodic infusions of intravenous immunoglobulins. Unfortunately, this is an extremely problematic outcome for T cell malignancies, as persistent T cell aplasia would be life threatening. There are currently > 200 clinical trials using CAR T cells registered at clinicaltrials.gov being carried out in the USA. However, the majority of these trials are enrolling patients with B cell malignancies. Advances are being made to expand CAR T cell therapy to the treatment of other cancers, and to minimize toxicities associated with treatment while reducing difficulty and cost of production.

Translating CAR T cell therapy for treatment of T cell malignancies

Harnessing and redirecting the cytotoxicity of T cells to malignant B cells has been established, but reprogramming T cells to kill malignant T cells, while sparing normal T cells, is much more complex and challenging. This requires aberrant expression of an antigen on malignant T cells that is absent or expressed at very low levels on normal T cells. CAR therapy requires isolation of healthy T cells from malignant T cells, a complicated procedure that can result in product contamination and subsequent CAR-modification of tumor cells. Additionally, expression of the targeted antigen on CAR T cells results in fratricide and limited expansion of the CAR T cells. Furthermore, targeting of an antigen regularly expressed on normal T cells would result in T cell aplasia, leading to profound immunosuppression, likely to be associated with high rates of morbidity and mortality (Fig. 1).

Fig. 1
figure 1

Potential outcomes of CAR T cell therapy in a patient with T cell disease. Upon re-infusion into a patient, CAR T cells recognize their cognate antigen, expanding upon this recognition, and initiating an attack. However, due to shared antigen expression on CAR T cells, normal T cells, and tumor cells, numerous outcomes can be observed. CAR T cells target tumor cells as intended, reducing tumor burden. However, without further engineering, the CAR-modified T cells are likely to express the targeted antigen as well, resulting in fratricide. CAR T cells would also target healthy T cells, resulting in unintended T cell aplasia. Lastly, CAR T cell therapy involves isolating normal T cells from malignant T cells for CAR-modification. A single malignant cell contaminating this population can result in masking of the antigen, leading to antigen-positive relapse. *Figure was created using BioRender

Full size image

Various approaches have been used to overcome these challenges, including CRISPR-Cas9 genome editing to remove the antigen from the CAR T cells [45,46,47], Tet-OFF expression system to limit fratricide during ex vivo expansion [48], protein expression blocker (PEBL) to retain the antigen in the ER/Golgi to prevent cell surface expression [49, 50], or using CAR-modified natural killer cells instead of T cells [47, 51,52,53,54]. Additionally, to date, four targets have been investigated as targets for CAR T cell therapy for the treatment of T cell malignancies with limited to no expression in the normal population of T cells, CD30, CD37, TRBC1, and CD1a [55,56,57,58]. Table 1 provides a summary of potential solutions to the three main challenges seen in adapting CAR technology for T cell malignancies—fratricide, T cell aplasia, and product contamination. A list of all current CAR-based clinical trials targeting T cell disease is presented in Table 2. Below, we review all preclinical and clinical CAR studies targeting T cell malignancies categorized according to the target antigen of interest.

Table 1 Strategies to overcome challenges in translating CAR therapy to treat T cell malignancies
Full size table
Table 2 Clinical CAR trials targeting T cell malignancies
Full size table

CD5

CD5 expression is limited to normal T cells and a small subpopulation of B cells, called B-1a cells [65,66,67,68,69]. CD5 acts as a negative regulator of TCR signaling and has a role in protecting against autoimmunity [70, 71]. CD5 is highly expressed on many T cell malignancies, particularly T-ALL and PTCLs, rendering it a good target for CAR T cell therapy [72,73,74]. Since CD5 expression on T cells is approximately ten times that on B cells [75], a low-affinity, high-avidity CAR targeting CD5 may steer clear of CD5-positive B cells while selectively killing T cells [76, 77]. Furthermore, CD8+ tumor-infiltrating lymphocytes (TILs) express lower levels of CD5 compared to that of peripheral blood T cells, and one study showed downregulation of CD5 improves the ability of T cells to lyse malignant cells [78]. CD5 was previously targeted as a tumor antigen in clinical trials using immunotoxin-conjugated CD5 monoclonal antibodies, with responses seen in patients with cutaneous T cell lymphoma and T-ALL [79, 80].

A preclinical study showed that expression of a CD5-CAR with a CD28 costimulatory domain resulted in surface downregulation of CD5 in CAR T cells. As a result, fratricide was observed only transiently, allowing the CD5-CAR T cells to expand. These cells had significant in vitro cytotoxicity against two T-ALL cell lines and primary T-ALL cells and delayed leukemia progression in two different CD5-positive T-ALL models [59]. Based on these results, CD5-CAR T cells with a CD28 costimulatory domain are being tested in patients with relapsed or refractory T cell disease (MAGENTA trial, NCT03081910). Our group used CRISPR-Cas9 to knockout CD5 expression in primary T cells prior to transduction with the CD5-CAR. We showed that gene editing of CD5 in effector CAR T cells increased CAR surface expression and decreased self-activation [47]. The increased CAR surface expression is predicted to enhance CAR T cell anti-tumor efficacy. We also showed antagonism of vasoactive intestinal peptide (VIP) signaling in conjunction with inhibition of the PI3Kδ pathway increased expansion of CD5-CAR-modified T cells as well as their cytotoxicity against CD5-specific tumor cell lines. This combination of compounds was also demonstrated to prolong in vivo persistence of treated T cells in NOD scid IL2Rγ-chain knockout (NSG) mice [81].

Interestingly, use of 4-1BB as the costimulatory domain in a CD5-CAR resulted in a significant fratricidal effect [48]. It was shown that tumor necrosis factor (TNF) receptor-associated factor (TRAF) signaling from the 4-1BB endodomain upregulated the intercellular adhesion molecule 1 (ICAM1), which subsequently stabilized the fratricidal immunological synapse between CD5-CAR T cells containing the 4-1BB costimulatory domain. To limit and control the effects of fratricide, a Tet-OFF expression system was used, which allowed for controlled transgene expression using the small molecule inhibitor, doxycycline. In the presence of doxycycline, CD5-41BB-CAR T cells expanded ex vivo without evidence of fratricide, while maintaining a more naïve genotype. Doxycycline was removed from the culture prior to injecting the CD5-41BB-CAR T cells into mice, resulting in CD5-CAR expression and improved survival outcomes in a T-ALL mouse model. Furthermore, there was a survival advantage in mice treated with Tet-OFF CD5-41BB-CAR T cells compared to survival of mice treated with CD5-CD28-CAR T cells without the Tet-OFF expression system [48].

Alternatively, we expressed the CD5-CAR in NK-92 cells, an interleukin-2 (IL-2) dependent natural killer cell line, which are inherently CD5-negative. Our data demonstrates that CD5-CAR-modified NK-92 cells have increased cytotoxicity against T cell leukemia cell lines compared to the cytotoxicity of naïve NK-92 cells [47, 51], and there is a significant improvement in survival of T-ALL xenograft mouse models compared to survival of mice treated with naïve NK-92 cells [47]. This data confirms previously published data illustrating significantly improved survival and enhanced tumor reduction in irradiated T-ALL mouse models treated with CD5-CAR-modified NK-92 cells compared to that of mice treated with control NK-92 cells [53]. Recently, another group tested CD5-CAR-modified NK-92 cells, using a NK-specific costimulatory domain 2B4 in their CAR constructs [82]. Interestingly, the CD5-2B4-CAR NK-92 cells displayed superiority to CD5-41BB-CAR NK-92 cells, in both in vitro and in vivo experiments [82].

CD7

CD7 is a transmembrane glycoprotein with expression on T cells and NK cells [83]. The majority of T-ALLs are CD7-positive, despite some populations lacking expression of other common markers, such as the TCR [74, 84]. Additionally, early T cell precursor acute lymphoblastic leukemia (ETP-ALL), a high-risk subset of T-ALL, highly express CD7 [84,85,86]. Two clinical trials have been initiated in China studying CD7-CAR-modified T cells for the treatment of CD7-positive malignancies (NCT04033302 and NCT04004637). However, preclinical studies showed significantly reduced expansion of CD7-CAR T cells compared to control T cells, as a result of fratricide [45, 49]. Fratricide appears to be observed to a greater extent in CD7-CAR T cells compared to CD5-CAR T cells [45]. It is hypothesized that this is due to a more incomplete internalization mechanism of CD7 from the cell surface following ligation of the antigen with an anti-CD7 scFv. CRISPR-Cas9 editing of CD7 from the cell surface of T cells prior to CAR expression demonstrated a superior method of develo** CD7-CAR T cells. These cells exhibited limited fratricide, expanded in vitro, and showed no evidence of impaired cytotoxicity in vitro nor in vivo. Investigations in a T-ALL mouse xenograft model revealed a statistically significant prolonged survival of CD7-edited CD7-CAR-treated mice compared to survival of control mice [45]. Based on these results, a phase I clinical trial has been initiated testing CD7-CD28-CAR T cells in T-ALL patients (NCT03690011). Additionally, a UCART7 was generated using CRISPR-Cas9 genome editing to disrupt the CD7 and TCRα constant (TRAC) loci. This study demonstrated that NSG mice engrafted with primary T-ALL blasts and treated with UCART7 donor cells exhibited tumor clearance from the peripheral blood, and, did not develop graft versus host disease (GvHD) or other severe side effects [46].

A new technique using protein expression blockers (PEBLs) has been established as an alternative to genome editing. This strategy couples an scFv with a retention peptide to maintain the protein of interest in the ER/Golgi preventing cell surface expression of the antigen. PEBL-CD7-CAR T cells exhibited superior cytotoxicity against primary T-ALL cells in vitro compared to non-PEBL CD7-CAR T cells. Using a patient-derived xenograft (PDX) model of ETP-ALL, upon detection of leukemic cell expansion in peripheral blood, PEBL-CD7-CAR T cells were injected. PEBL-CD7-CAR T cell-treated mice had a significant survival advantage over control mice. However, CD7-positive relapse did occur in all PEBL-CD7-CAR T cell-treated mice [49].

Despite CD7 expression on NK-92MI cells (IL-2 producing NK-92 cells), they have been used for CD7-CAR therapy demonstrating only a small percentage of cells are CD7-positive, and upon CD7-CAR expression, fewer than 1% CD7-positive NK-92MI cells remain [60]. Two CD7-CAR constructs, a monovalent and bivalent construct, were generated using a humanized CD7 nanobody sequence that had been previously developed in the laboratory. Both CAR constructs demonstrated enhanced CD7-specific cytotoxicity against T-ALL cell lines and primary patient cells ex vivo when expressed in NK-92MI cells. The bivalent CD7-CAR-modified-NK-92MI cells exhibited slightly greater cytotoxicity compared to that of the monovalent CAR-modified cells, and significantly inhibited disease progression in a T-ALL PDX model when compared to naïve unmodified NK-MI cells.

CD4

Most cancers derived from lineage-differentiated T cells are likely to be of CD4-positive origin, making CD4 a potential target for CAR therapy. A preclinical study was performed to consider the cytotoxicity of CD4-CAR-modified T cells against T-ALL tumors in NSG mice. This study also included the use of alemtuzumab to clear the CAR T cells as a safety mechanism. NSG mice were injected with luciferase-expressing Jurkat T cells and subsequently treated with naïve T cells or CD4-CAR-modified T cells. CAR-treated mice displayed a survival advantage and an ~ 80% reduction in tumor burden compared to mice treated with naïve T cells. CD4-CAR-modified T cells were also injected into mice to evaluate the ability of alemtuzumab to effectively eliminate CAR-modified T cells. Alemtuzumab was administered 24 h post-CAR T cell injection. A > 95% depletion of CD4-CAR-modified T cells was observed within 6 h following injection signifying the use of alemtuzumab as a safety mechanism for CAR T cell therapy [87]. Additionally, a phase I clinical trial to assess the safety and feasibility of CD4-CAR T cell infusions in patients with relapsed/refractory T cell lymphoma and T cell leukemia has been initiated (NCT03829540).

However, expression of CD4 on T cells can complicate CD4-CAR T cell therapy as previously described. NK-92 cells are inherently CD4-negative, and therefore the use of NK-92 cells as opposed to T cells reduces risk of fratricide and avoids the need for further modifications. Additionally, it abrogates the risk of aplasia of CD4-positive cells that can occur with long-term engraftment of CAR T cells. Anti-CD4-CAR NK-92 cells have shown in vitro success eliminating PTCL cell lines and both adult and pediatric primary cells. Using a xenograft model in NSG mice, CD4-CAR NK-92 cell-treated mice demonstrate significantly prolonged survival compared to control-modified NK-92 cell-treated mice [54].

CD37

CD37 is a member of the tetraspanin superfamily with expression limited to lymphoid tissues, particularly B cells [88, 89]. CD37 expression in cancer cells is typically characteristic of B cell malignancies; however, its expression can be found in some cases CTCL and PTCL [90, 91]. Since CD37 is not expressed in T cells, there is no evidence of fratricide occurring in anti-CD37 CAR T cells. However, in the presence of CD37-positive PTCL cell lines, CD37-CAR T cells exhibit increased activation and degranulation as well as specific cytolytic activity in vitro [56]. The restricted expression of CD37 makes it a safer target for CAR T cell therapy, given there would be no concern of T cell aplasia. Additionally, CD37 is not expressed in NK cells, providing an opportunity to utilize NK cells as effector cells in place of T cells. The versatility of CD37-CARs to treat B cell and T cell lymphomas suggests that this may be an important target for further investigations. While CD37 is predominantly being examined for dual targeting for B cell malignancies, the target has potential for CAR therapy against T cell malignancies.

CD30

CD30, a member of the tumor necrosis factor receptor (TNFR) superfamily, promotes T cell proliferation and cytokine production following TCR stimulation, while also having an opposing role in promoting apoptosis [92]. Expression is limited to a subset of activated lymphocytes found around the follicular regions of lymphoid tissues [93,94,95]. While CD30 is well known for its strong expression in virtually all classical Hodgkin lymphoma, expression of CD30 can also be found on a subset of PTCLs, including ALCL [92,93,94, 96]. One study demonstrated that CD30 expression is upregulated during chemotherapy regimens in T-ALL patients. Of 34 T-ALL patients, approximately 38% had CD30-positive T-ALL [96]. Therefore, some T-ALL patients who relapse following chemotherapy may still respond to CD30-directed CAR therapy.

Preclinical studies have previously demonstrated CD30-CAR T cell capacity for lysing tumor cells [97, 98] and numerous clinical investigations into CD30-CAR T cell therapy have been launched with encouraging results. Eleven phase I/II trials treating patients with CD30-positive malignancies are currently active (NCT01316146 [55], NCT01192464, NCT03049449, NCT02690545 [62], NCT02958410, NCT02663297, NCT03383965, NCT02917083 [64], NCT04008394, NCT02259556 [63], and NCT03602157). To date, no toxicities related to CAR T cell infusion nor impaired immunity against common viruses has been reported from these trials. However, one trial reported that the in vivo CAR T cell expansion and persistence was reduced following subsequent infusions compared to those following initial doses [55]. The decreased persistence of the CAR T cells may have prevented the development of severe adverse events such as CRS and neurotoxicity that are commonly observed following CAR T cell infusion. Of the two ALCL patients in this trial, one patient was non-responsive to the therapy, while the other entered complete remission lasting 9 months [55]. Results from another phase I trial in China for patients with relapsed/refractory CD30-positive lymphomas (NCT02259556) corroborate the limited toxicity and anti-tumor activity of CD30-CAR T cells [63].

TRBC1

T cells express the αβ TCR; the β-chain can either be encoded by the T cell receptor beta constant 1 (TRBC1) gene or TRBC2 gene [99, 100]. Therefore, expression of TRBC1 and TRBC2 is mutually exclusive. Additionally, CD4- and CD8-positive T cell populations express both subsets and CD8-positive T cell populations specific for common viruses also contain both TRBC1 and TRBC2 cells [58]. However, as malignant T cells develop from a single cell, the entire population of cancerous cells will be either TRBC1- or TRBC2-positive. Numerous T cell malignancy cell lines and primary samples have been analyzed by flow cytometry to validate the homogeneity of β-chain expression in a malignant cell population [58]. Many cancer cells downregulate the αβ TCR; however, it is expressed on > 95% of PTCLs [101] and > 30% of T-ALLs [102].

Anti-TRBC1 CAR T cells exhibited specific and efficient cytotoxicity against the JKO T cell line transduced with TRBC1, but not against non-transduced cells or cells transduced with TRBC2, even in a mixed population. Furthermore, in primary samples from patients with T cell malignancies, the anti-TRBC1 CAR T cells preserved a significant fraction of healthy T cells (TRBC2 cells), thereby circumventing a limitation of CAR T cell therapy for the treatment of T cell malignancies [58]. In an NSG mouse model using TRBC1-positive Jurkat T cells to establish cancer, mice treated with the anti-TRBC1 CAR T cells exhibited reduced tumor burden and elongated survival. In additional preclinical studies, NSG mice were injected with both TRBC1 and TRBC2 cancer cells, and then treated with either naïve T cells or anti-TRBC1 CAR T cells. TRBC1-positive Jurkat T cells could not be detected in mice treated with anti-TRBC1 CAR T cells; however, TRBC2-positive cells were identified. This is in contrast to mice treated with naïve T cells, whose bone marrow confirmed the presence of both TRBC1-positive and TRBC2-positive cells [58]. Thus, targeting TRBC1-positive malignant cells offers a unique approach to avoiding T cell aplasia, a consequence of many proposed CAR T cell therapies for the treatment of T cell malignancies.

CD3

CD3 is a pan T cell marker comprised of four distinct polypeptide chains, epsilon, gamma, delta, and zeta, which form pairs of dimers, transmitting T cell activation signals. As CD3 is exclusively expressed on T cells, it has been a popular target in preclinical CAR T cell therapies for the treatment of T cell malignancies. As expected, due to fratricidal issues, manufacturing of anti-CD3 CAR T cells does not yield a viable cellular product [61]. Various approaches using an anti-CD3 CAR have been investigated including the use of transcription activator-like effector nuclease (TALEN) mRNA to disrupt the TRAC locus and using NK-92 cells in place of T cells as the effector cell type. Disruption of the TRAC locus prevents assembly of the TCRαβ/CD3 complex, allowing for anti-CD3-CAR expression without compromising cellular proliferation and viability. Enrichment of the CAR-positive, CD3-negative population was observed. In patient T-ALL samples, anti-CD3 CAR T cells demonstrated specific cytotoxicity against CD3-positive cells. In a T-ALL NSG model, anti-CD3 CAR T cells were shown to clear luciferase-expressing CD3-positive Jurkat cells, but showed no effect in NSG mice engrafted with CD3-negative Jurkat cells [61]. To circumvent the need for additional modifications, NK-92 cells can also be used to express the anti-CD3-CAR, since they are CD3-negative cells. CD3-CAR NK-92 cells demonstrated efficient ex vivo lysis of PTCL primary samples, resulting in less than 0.5% lymphoma cells remaining at 5:1 effector to target ratios. Furthermore, CD3-CAR NK-92-treated T-ALL NSG mice exhibited prolonged survival with ~ 87% reduced tumor burden through day 23 [52].

CD1a

CD1a is a lipid-presenting molecule whose expression is restricted to develo** cortical thymocytes, skin Langerhans cells, and some circulating myeloid dendritic cells [145]. However, recently developed methods are being used to enhance NK-cell expansion, such as through K562-feeder cell expression of OX40 ligand [146]. As mentioned above, a limitation to CAR-NK therapy is the extreme sensitivity of NK cells to cryopreservation. They have demonstrated poor viability and diminished cytotoxicity after cryopreservation. While cytotoxicity can be restored to normal levels after a few days in culture with exogenous IL-2, the low viability post-cryopreservation remains a concern [141].

Gamma delta T cells

While αβ T cells function as a part of the adaptive immune system, γδ T cells play roles in both the innate and the adaptive immune systems. γδ T cells and αβ T cells originate from two distinct T cell lineages [224]. γδ T cells are the only innate immune cells expressing a TCR [147]; however, their target recognition is independent of MHC recognition [148, 149]. Lack of MHCI- and MHCII-restriction make γδ T cells optimal candidates for allogeneic cell therapy. The peripheral blood subset of γδ T cells known as Vγ9Vδ2 T cells represents the most commonly studied subset in this context. Studies by our group have demonstrated that similar transduction efficiencies can be achieved in Vγ9Vδ2 T cells grown under cGMP serum-free conditions as are achieved in αβ T cells using lentiviral vectors. Additional studies were performed revealing peak low-density lipoprotein receptor (LDL-R) expression on days 6–8 of γδ T cell expansion [150]. As LDL-R is the major receptor for VSV-G-pseudotyped lentiviral vectors, this data suggests that greater transduction efficiency can be achieved on these days using lentiviral vectors compared to earlier or later in the expansion [151].

To date, numerous preclinical studies have evaluated CAR-modified γδ T cells targeting neuroblastoma [152, 153], melanoma [154], B cell malignancies [153, 155], and epithelial cell adhesion molecule (epCAM)-positive adenocarcinomas [156]. GD2-CAR-modified γδ T cells expressing the RQR8 suicide gene were shown to expand 2.5-fold upon antigen exposure [152]. Furthermore, both GD2-CAR- and CD19-CAR-modified γδ T cells were demonstrated to secrete pro-inflammatory cytokines in the presence of GD2- or CD19-expressing tumor cells, respectively [153]. While these studies utilized viral vectors to express the CAR, electroporation of a Slee** Beauty transposon has also been shown to result in CD19-CAR expression in γδ T cells, resulting in anti-tumor cytotoxicity in both the in vitro and in vivo settings [155]. Additionally, expression of a CAR targeting melanoma-associated chondroitin sulfate proteoglycan (MCSP) was established in γδ T cells using mRNA transfection. Despite comparable anti-tumor cytotoxicity, lower cytokine secretion was observed in MCSP-CAR-modified γδ T cells compared to that from conventional CAR-modified αβ T cells [154]. Reduced pro-inflammatory cytokine secretion is favorable due to anticipated reduced severity of CRS. Lastly, epCAM CAR-modified γδ T cells demonstrated high levels of in vitro cytotoxicity of tumor cell lines when γδ T cells were both fresh and cryopreserved [156]. These studies pave the way for additional trials using CAR-modified γδ T cells targeting T cell malignancies. They demonstrate that engineering of γδ T cells is feasible and results in enhanced in vitro and in vivo cytotoxicity upon CAR expression.

CAR-modified γδ T cells may be able to overcome the obstacle of antigen escape seen in some treatment-resistant cases by relying on their innate ability to recognize tumor cells through other means. Naïve γδ T cells have been shown to have anti-tumorigenic activity against leukemia, neuroblastoma, and colon cancer cell lines as well as primary cancer cells in vitro [2). NK cells (i) are non-alloreactive and can be obtained from healthy donors, eliminating risk of product contamination; (ii) do not form memory responses, preventing T cell aplasia; and (iii) do not express the same antigen repertoire as T cells, avoiding fratricidal concerns. CD7 is an exception as it is expressed on NK cells and therefore fratricide could occur. While several groups have published studies with CAR NK-92 cells targeting T cell malignancies, more effort needs to be put into using primary NK cells for targeting this disease, especially given the limitations of NK-92 cells. Other, equally promising approaches, such as utilizing γδ T cells as the cellular vehicle for CAR therapy represents an alternative, less studied approach. Similar to NK cells, γδ T cells are non-alloreactive and are unlikely to form a memory response against a T cell antigen. γδ T cells are likely to succumb to fratricide in certain circumstances; however, targeting an antigen such as CD5 that results in only transient and limited fratricide may be especially advantageous. Furthermore, γδ T cells exhibit innate MHC-independent mechanisms of cytotoxicity by which they can recognize tumor cells. Thus, CAR therapy using γδ T cells represents an understudied avenue with the potential of develo** into a superior cellular product.

Fig. 2
figure 2

Venn diagram representing challenges and solutions in targeting T cell antigens with CAR therapy. Each circle represents a hurdle associated with translation of CAR therapy to T cell disease—fratricide, T cell aplasia, and product contamination. As seen in the figure, only the use of NK cells or NK-92 cells as the CAR-effector cell can potentially address all three issues concurrently. However, using NK cells or NK-92 cells comes with its own limitations as previously described. All other approaches require multiple modifications to generate a translatable CAR product to target T cell disease. Potential alternative solutions such as use of γδ T cells as the CAR-effector cell, transient CAR expression with mRNA electroporation or AAV viral delivery, as well as incorporating suicide genes and safety switches, remain largely unexplored. A greater focus on implementing such strategies is required to enable successful translation of this therapy for T cell malignancies

Full size image

Many advances have been made toward translating CAR therapy for the treatment of T cell malignancies. Both academia and industry are focused on the identification of tumor-specific antigens to enhance the safety and efficacy of CAR T cell products as well as on the development of superior cellular products. Unfortunately, due to vast variability in the design and execution of preclinical studies, it is often difficult to compare the different strategies. However, the numerous preclinical and clinical studies currently underway provide optimism for successful translation of this therapy to treat this aggressive and challenging group of diseases.