Progress and pitfalls of chimeric antigen receptor T cell immunotherapy against T cell malignancies

Abstract

Chimeric antigen receptor T cell (CAR-T) immunotherapy has revolutionized the treatment of relapsed and refractory B cell-derived hematological malignancies. Currently, there are six FDA-approved commercial CAR-T products that target antigens exclusively expressed on malignant B cells or plasma cells. However, concurrent advancement for patients with rarer and more aggressive T cell-derived hematological malignancies have not yet been achieved. CAR-T immunotherapies are uniquely limited by challenges related to CAR-T manufacturing and intrinsic tumor biology. In this review tailored for practicing clinician-scientists, we will discuss the major barriers of CAR-T implementation against T cell-derived neoplasms and highlight specific scientific advancements poised to circumvent these obstacles. We will summarize salient early-stage clinical trials implementing novel CAR-T immunotherapies specifically for patients with relapsed and/or refractory T cell neoplasms. Lastly, we will highlight novel manufacturing and treatment strategies that are poised to have a meaningful future clinical impact.

1. Introduction

Chimeric antigen receptor T cell (CAR-T) immunotherapy has demonstrated impressive responses in treating patients with hematologic malignancies^1–6. In 2017, tisagenlecleucel (Kymriah^®), a CD19-directed CAR-T product, was approved by the U.S. Food and Drug Administration (FDA) for individuals younger than 26 years old with CD19-expressing relapsed and/or refractory (R/R) B cell acute lymphoblastic leukemia (B-ALL). Over the following five years, three additional CAR-T products—axicabtagene ciloleucel (Yescarta^®), brexucabtagene autoleucel (Tecartus^®), and lisocabtagene maraleucel (Breyanzi^®)—have also received FDA approval for adults with R/R large B cell and follicular lymphomas, mantle cell lymphoma and B-ALL, and large B cell and follicular lymphomas, respectively. The FDA has additionally awarded recent approval to two B cell maturation antigen (BCMA)-directed CAR-T products—idecabtagene vicleucel (Abecma^®) and ciltacabtagene autoleucel (Carvykti^®) for adults with multiple myeloma following four or more prior lines of anti-myeloma therapy. These commercially available CAR-T products, in addition to dozens of experimental cellular immunotherapeutics targeting B cell antigens in early-phase clinical trials, have generated clinically significant and durable anti-tumor responses for many patients. However, CAR-T immunotherapies have had limited utility to date in both children and adults with T cell-derived leukemias and lymphomas. This is, in part, driven by the inherent aggressivity and refractoriness of most T cell neoplasms to classical chemotherapy, but also challenges intrinsic to CAR-T design, manufacturability, and expansion⁷.

T cell leukemias and lymphomas represent a broad spectrum of hematological neoplasms with overall poor prognosis that are heterogenous by several classifiers. Specific to peripheral T cell lymphomas (PTCL), the current World Health Organization categorizes 27 distinct phenotypes of T and NK-cell neoplasms⁸. Diagnosis of these entities is dependent on morphological, immunophenotypic, and molecular features that often overlap between subsets, making diagnosis and clinical management challenging. Consequently, unlike CD19 and BCMA that are uniformly expressed on specific subsets of B cell derived neoplasms, T cell-derived diseases lack a reliably expressed surface antigen that is unique to neoplastic cells^9,10. Absence of a uniquely identified surface antigen has posed a barrier to translate CAR-T products to patients with T cell-derived malignancies.

In this review, we will consider the major hurdles of CAR-T implementation against T cell neoplasms and highlight specific scientific advancements poised to circumvent these obstacles. We will emphasize salient early-stage clinical trials that have translated CAR-T immunotherapy to patients with T cell neoplasms. Lastly, we will discuss novel potential strategies that are poised to have a meaningful clinical impact.

2. Brief overview of clinical CAR-T engineering

CAR-T utilize a synthetic CAR that is engineered to be expressed on either patient- or donor-derived T cells. This CAR targets tumor-associated antigens in a major histocompatibility complex (MHC) independent manner to trigger T cell activation and generate anti-tumor effects. A standard CAR derives its unique immunophenotype from 1) an extracellular single-chain fragment variable (scFv) region and 2) an intracellular signaling domain. The scFv region dictates specificity as it is derived from the variable heavy (V_H) and light (V_L) chains of a monoclonal antibody that binds to an antigen of interest. Upon scFv engagement, the intracellular signaling domain becomes stimulated to trigger T cell activation¹¹.

First generation CAR-T were developed in the 1990s and were composed of a single CD3ζ intracellular domain. These CAR-T did not generate strong T cell activation and ultimately lacked durability and persistence with no substantiative clinical efficacy^12,13. Second generation CARs were born by conjugating a co-stimulatory domain, such as CD28 or 4–1BB, to the CD3ζ intracellular signaling domain^14–16. These second-generation CARs have demonstrated potent responses in vitro and in vivo, as well as in the clinical space where they represent all six currently FDA-approved CAR-T products. Advanced generations of CAR-T are more recently being tested. Third generation CARs consist of multiple co-stimulatory domains: typically, but not restricted to, a combination of both CD28 and 4–1BB, that have exhibited preclinical efficacy and now undergoing clinical examination¹⁴. Additional T cell costimulatory molecules from both the immunoglobulin and tumor necrosis factor receptor superfamilies, such as ICOS (CD278), OX40 (CD134), CD27, CD40, and MyD88 have also been used with success in preclinical studies, as reviewed elsewhere^17,18. Newer generations of CAR-T in pre-clinical development possess more complex molecular machinery that foster cytokine secretion and/or additional intracellular activity upon CAR binding^19,20.

Commercial CAR-T manufacturing begins with collection of peripheral blood mononuclear cells (PBMCs), commonly achieved through apheresis. The treating physician must importantly time apheresis relative to systemic treatment to coincide with sufficient numbers of circulating T lymphocytes available for adequate CAR-T production²¹. White blood cell (WBC) and absolute lymphocyte count (ALC) parameters have not been systematically established²², however, functional CAR-T have been produced from patients with pre-apheresis ALC less than 100 cells per microliters (μL) of blood^21,23. CAR-T yield is linearly proportional to the pre-apheresis ALC and CD3⁺ T cell count, with some analyses suggesting inferior antitumor responses when CAR-T are manufactured using lymphocyte-depleted apheresis material^21,24. Repeated and recent administration of cytotoxic chemoimmunotherapy or systemic corticosteroids adjacent to apheresis, either as definitive or bridging treatment, has also been hypothesized to have an adverse effect on CAR-T yield and functionality^25,26. Washout periods typically of four to five drug half-lives are recommended per expert opinion prior to apheresis, if possible^27,28. However, monoclonal antibodies now integrated with T cell leukemia/lymphoma management (i.e. alemtuzumab or brentuximab vedotin) possess single half-lives on the order of weeks, making completion of such a washout period difficult while maintaining low tumor burden suitable for CAR T-cell administration²⁹.

Following T cell enrichment, T cells are then expanded by co-culture with artificial antigen presenting cells or CD3/CD28 antibody-coated beads, followed with viral transduction of a CAR-possessing plasmid that permanently integrates into the T cell genome. While mechanisms of T cell genetic modification are outside the scope of this review, all FDA-approved commercial CAR-T products to date currently utilize either gammaretroviral or lentiviral gene editing vectors that are replication-incompetent. Non-viral methods of CAR T manufacturing utilizing transposon systems, such as Sleeping Beauty or piggyBac, mRNA, and liposomal nanocarriers are of high interest and are of active pre-clinical and early-phase clinical trial investigation^30–35.

3. Challenges of CAR-T immunotherapy for T cell neoplasms

Implementation of CAR-T against T cell-derived neoplasms not only encounters similar obstacles that are seen as in the treatment of B-cell derived neoplasms, but simultaneously new biological and clinical challenges, such as fratricide, production contamination, T cell aplasia, and clonal escape, that require innovative solutions (FIGURE 1). Recent advancements in CAR-T engineering and manufacturing have been made to systematically address these issues.

FIGURE 1. Schematic of challenges in using chimeric antigen receptor T cell (CAR-T) immunotherapy against T cell derived hematological neoplasms.

CAR-T cells (green) engineered with specificity against pan T-cell antigens (blue receptor) can mediate cytotoxicity against healthy, non-malignant T cells (blue) resulting in T cell aplasia. Production contamination yields an admixture of manufactured CAR-T products with malignant T cells or the inadvertent development of a “malignant CAR T cell” (red with CAR) with potential for resistance to cellular and chemoimmunotherapeutics. Fratricide, where an anti-T cell receptor CAR T cell is activated against another anti-T cell receptor CAR-T, may result in an inability to successfully manufacture and expand CAR-T to clinically significant dose levels. Lastly, malignant T cells may downregulate CAR-T target antigens or they may undergo epitope masking such that malignant T cell antigens are not accessible, resulting in clonal escape and CAR-T resistance. Illustration generated using Biorender.

3.1. Fratricide

Fratricide, derived from the Latin term for the “killing of one’s brother,” describes the phenomenon when CAR-T not only targets malignant T cells, but also other CAR-T that express the target antigen. This inadvertent “on-target, off-tumor” cross-reactivity prohibits adequate CAR-T expansion to clinically relevant dose levels that are necessary for therapeutic benefit with standard manufacturing methods. The relative fratricide magnitude is dependent upon 1) the absolute number of CAR and target antigen expression on manufactured CAR-T^36,37 and 2) the scFv conformation and proximity to an accessible target antigen epitope^38,39.

Unlike B-cell derived neoplasms with ubiquitous and sustainable expression of unique B-cell specific surface antigens, there are a paucity of T-cell tumor-specific antigens without appreciable expression on non-malignant and manufactured CAR-T. Consequently, most CAR design to date has focused on pan-T cell antigens with high quantitative expression on neoplastic T cells, such as CD3, CD5, and CD7. While CD3 and CD7 targeting lead to complete fratricide, CD5, which is robustly expressed in 85% of T-cell acute lymphoblastic leukemia (T-ALL) and PTCL cases, partially circumvents fratricide, possibily due to its rapid internalization upon binding to tumor ligand, specifically in activated effector T cells, and due to epitope masking of the CD5 extracellular domain^40–43. These phenomena, unique to CD5, mitigated fratricide in anti-CD5 CAR-T and permitted cytotoxicity against primary T-ALL cells in vitro and in vivo⁴⁴, however, their clinical grade manufacturing using conventional methods remains limited.

Targeting other pan-T cell antigens necessitates genome editing techniques to delete the target protein of interest or to modulate target protein recruitment towards the cell membrane concurrent with CAR viral transduction. While the former has proven successful for several targets, targets that are important for T cell growth and survival, such as CD3, CD8, or CD30, may be unable to be constitutively deleted for optimal cytotoxic CAR-T function. CRISPR-Cas9 editing of unessential T cell antigens can be utilized with standard CAR transduction methods to generate clinical grade CAR-T that are fratricide-resistant^45,46. Analogously, sequestration of T cell antigens intracellularally via protein expression blocker systems also permits expansion of cytotoxic CAR-T and presents a suitable non-gene editing strategy to bypass fratricide^47,48.

Temporospatial CAR expression on the extracellular membrane conceptually temporarily bypass fratricide using drug-induced genetic switches⁴⁹. For example, tetracycline-induced genetic elements can “turn-off” CAR transcription during T cell manufacturing, with removal of drug selective pressure to “turn-on” CAR expression timed for desired anti-tumor activity ^50–52. While titratable and reversible control of CAR-T activity is appealing, fratricide-dependent cytotoxicity is likely to predominate as compared to anti-tumor cytotoxicity, although it is to be seen whether the amount of anti-tumor cytotoxicity is sufficient to generate clinical responses.

Alternatively, target-antigen negative T cells could be chosen as the platform for CAR-T manufacturing. Specific to CD7, upwards of 20% of the total T cell population consists of naturally circulating CD7-negative T cells in healthy donors⁵³. CD7-negative cells could be transduced with an anti-CD7 CAR, with resulting CAR-T products able to elicit anti-tumor responses both in in vitro and in vivo T-ALL tumor models. It remains to be determined if this approach can be used with other pan T-cell antigens.

3.2. Production contamination

Product contamination describes the admixture of manufactured CAR-T products with malignant T cells. Previous analyses have established that a low CAR-T expansion to tumor burden ratio is associated with a higher probability of disease relapse and overall poor prognosis^54–56. These inferior outcomes in patients with high circulating disease burden correlate with poor normal T cell quality at the time of CAR-T manufacturing, driven in part by the underlying clonal T cell malignancy and cumulative effect from cytotoxic chemoimmunotherapy treatments.

As compared to CAR-T engineered against B cell neoplasms, the risk of product contamination in T cell malignancies is substantially magnified, driven by the need to purify malignant versus non-malignant T cells in the absence of a reliably distinguishing cell surface antigens in the autologous setting. Compounding this risk further is a poor understanding on what constitutes an acceptable “risk” of product contamination, or in other words, the lowest boundary of measurable residual disease (MRD) that statistically minimizes product contamination. Clinically validated methods to detect MRD in a manner to segregate non-malignant vs. malignant cells based on immunophenotypic or genetic profiling (i.e., multiparameter flow cytometry and digital droplet PCR) are only sensitive to one out of every 10³-10⁴ cells⁵⁷. Consequently, utilization of autologous CAR-T products in patients with peripheral blood MRD positivity, in our opinion, is currently not advisable for patients with circulating T cell neoplasms.

One strategy to eliminate product contamination and an additional risk of malignant cell transduction is to employ allogeneic CAR-T (alloCAR-T) from healthy patient donors. AlloCAR-T allows for an “off-the-shelf” cellular therapy that is appealing for patients with underlying aggressive disease who may be unable to wait for autologous CAR-T manufacturing. Additionally, alloCAR-T avoids manufacturing limitations related to prior chemoimmunotherapy exposure, and may be further enriched with highly active cytotoxic CAR-T. However, analogous to allogeneic hematopoietic cell transplant (alloHCT), graft-versus-host disease (GVHD) potentially may develop due to the T cell receptor (TCR) and MHC incompatibility. The TCR of naïve and mature T cells consists of both an α and β constant peptide chain (TRAC and TRBC, respectively) that forms a complex with CD3. As CAR-T cells function in an MHC-independent manner, deletion of either TRAC or TRBC (TRBC1 or TRBC2) with or without perturbation of class I HLA antigens can eliminate alloreactivity while preserving anti-tumor CAR-T responses^58–61. These “universal” CAR-T have further been multiplexed with perturbation of T cell lymphodepletion targets, such as CD52, the ligand of alemtuzumab. Such a strategy renders CAR-T resistant to cytotoxic lymphodepleting agents and is hypothesized to enhance CAR-T efficacy and persistence⁶².

AlloCAR-T are of compelling interest for T cell malignancy treatment. UCART7, an allogeneic, fratricide-resistant, anti-CD7 CAR-T product, incorporates dual TRAC and CD7 multiplexed gene deletion. While additional alloCAR-T are currently in early clinical study, it remains to be seen if GVHD is abrogated and whether anti-tumor responses are comparable to autologous CAR-T products. One predictable limitation of alloCAR-T in humans that limits some appeal is that they may not induce complete and/or durable remissions due to rejection by recipient natural killer (NK) cells. Potential strategies to mitigate NK cell rejection include pre-treatment with NK cell depleting monoclonal antibodies, or CAR-T further engineered with HLA-E, HLA-G, or CD47 overexpression to prevent NK cell detection^63–66.

Moreover, CAR therapies derived from less common immune cell subsets that lack allogenicity, such as gamma-delta (gd) T cells, natural killer (NK) cells, invariant natural killer cells (iNKT), and macrophages, have also been reported and studied in early phase clinical trials, as reviewed elsewhere⁶⁷. CAR-NK cells comprise the most studied non-T cell cohort against a variety of cancer models with preclinical efficacy both in vitro and in vivo⁶⁸. While initially limited by ex vivo expansion, incorporation of proliferative transgenes, such as IL-2 and IL-15, have mitigated this pitfall. However, limitations with clinical feasibility and efficacy remain, inclusive of their short in vivo persistence (approximately 1–2 weeks) and suboptimal trafficking and penetration of solid tumors. Limited clinical data is currently available using CAR-iNKT and CAR-gd T cells, both of which are suspectible to immunosuppressive effects that potentially hinder their efficacy. Further studies are needed to explore the clinical integration of these rarer immune cell platforms with CAR therapy.

3.3. T cell aplasia

One effect of CAR-T immunotherapy for B-cell malignancies is chronic B-cell aplasia and hypogammaglobulinemia due to on-target, off-tumor effects against non-malignant B cells. While these are generally accepted, recent analyses have shown patients treated with anti-CD19 CAR-T (CART19) are at a significantly higher risk of developing severe COVID19, regardless of supplemental intravenous immunoglobulin administration⁶⁹. Similar effects on non-malignant T cells have the potential to substantially enhance the risk of severe infection due to attenuation of MHC-mediated adaptive immune responses. Encouragingly, in the limited clinical experience with T cell-directed CAR-T immunotherapy, persistent T cell aplasia has not been described and Grade 3 or 4 adverse events directly attributed to infection have not been appreciated in the majority to enrolled patients. Nonetheless, because of these potentially life-threatening infectious risks, prolonged antimicrobial prophylaxis with anti-bacterial, -viral, and -fungal agents, as well as preemptive monitoring for indolent viral reactivation syndromes, such as cytomegalovirus (CMV) and Epstein-Barr Virus (EBV), should be integrated with pan-T cell antigen specific CAR-T regimens.

Potential upfront strategies to minimize T cell aplasia are like those described to mitigate CAR-T fratricide. Drug-dependent safety or suicide switches inducing caspase overexpression to facilitate CAR-T apoptosis after a suitable clinical response has been attained could also be considered^70,71, however, these systems are of lower utility in T cell-derived malignancies given the poorer likelihood of achieving durable responses due to intrinsic tumor heterogeneity and aggressivity.

A more feasible, and curative, solution to transient T cell aplasia is to utilize CAR-T therapy as a bridge to alloHCT for complete adaptive immune cell reconstitution following tumor debulking. The utility of a consolidative alloHCT following CART19 is currently a crticial question under investigation, specifically for patients with B-ALL, as upwards of 50% patients fail to respond or ultimately relapse within one year of CAR-T administration^1–3,5,6,72. Standard alloSCT conditioning regimens functionally deplete previously administered CAR-T for which it has been shown that the durability of remission is strongly correlated with CAR-T persistence. However, in the registrational trials investigating CART19 against B-ALL, CAR-T persistence was dependent on the CAR costimulatory domain and that CAR-T were no longer detectable in the majority of patients within 6 months of treatment^73–77. While there are no randomized studies directly investigating the role of alloHCT following CAR-T therapy, patients who achieved MRD-negative complete responses with CART19 and proceeded with alloSCT had improved event- and relapse-free survival^5,73,78. Thus, for patients with T-cell derived neoplasms, for which consolidative alloSCTs are particularly considered in the relapsed and refractory setting, a time-limited duration of T cell aplasia is likely to be acceptable prior to definitive immune reconstitution.

Another mitigating strategy hinges upon incomplete CAR transduction of infused T cells during CAR-T manufacturing. Current commercial CAR-T administration is dosed proportionally to the number of CAR-expressing (“CAR-positive”) viable cells, with current gammaretroviral and lentiviral manufacturing methods yielding approximately 60–90% and 20–60% CAR-positive cells, respectively. Additionally, with CRISPR/Cas9 gene deletion, pan-T cell antigen expression can be reliably reduced upwards of 90% relative to wild-type controls ^79–82. Thus, a proportion of the manufactured and infused CAR-T consists of target antigen-negative, CAR-negative T cells that may engraft and maintain normal immune effector functions mediated by their endogenous TCRs, so long as their target antigen is dispensable. Analogous to a donor lymphocyte infusion (DLI) following alloHCT, these target antigen-negative, CAR-negative T cells may confer ongoing immunological protection⁸³. Currently, such a strategy is most likely to be successful using autologous CAR T cell or non-alpha/beta allogeneic CAR T cells sources in which the endogenous TCRs are not genetically manipulated to mitigate graft-versus-host phenotypic effects as is the case using allogeneic products.

3.4. Clonal Escape

Clonal escape is the inability of a CAR-T to recognize its cognate surface antigen on neoplastic cells. Clonal escape is a shared pitfall among B- and T cell-targeting CAR-T therapies and is responsible for CAR-T failure at any point following CAR-T infusion. Several mechanisms of clonal escape have been identified, including antigen escape, antigen downregulation, and lineage switching.

Antigen-escape occurs when tumor cells no longer express the endogenous extracellular protein for which a CAR-T is specific. This can either occur due to a specific point mutation with the antigen scFv target⁸⁴ or, more commonly, target-antigen negative malignant subclone expansion that confers a competitive growth advantage following CAR-T treatment⁸⁵. Analysis of relapsed CART19 treated patients showed that CD19-negative relapse occurs in approximately 30% of patients with B cell lymphoma⁸⁶ and 50% of patients with B-ALL⁵. Antigen downregulation may manifest as either a reduction in the total target-antigen density on tumor surface, as is seen with pan T cell antigens upon T cell activation, or as epitope masking⁸⁷. Lastly, lineage switching, a rare phenomenon in which neoplastic lymphoma cells can acquire immunophenotypic properties of other hematopoietic lineages, can also result in loss of the desired CAR-T target and lead to disease relapse^88,89.

4. Clinical experiences with T cell antigen specific CAR-T against T cell neoplasms

4.1. CD5

CD5 is a transmembrane protein of the scavenger receptor cysteine-rich family. In normal T cells, CD5 becomes upregulated upon activation and functions as a negative regulator of T cell signaling to reduce autoreactive T cell functions. We, and others, have found elimination of CD5 transcription using CRISPR/Cas9 had no functionally significant effect on T cell proliferation and enhanced the production of T cells with improved anti-tumor cytotoxicity^43,81,82,90. Along with the unique ability to be rapidly internalized upon T cell activation, CD5 became the first pan-T cell antigen CAR-T target translated into the clinic. This transition was further supported by a favorable safety profile of H65-RTA, an anti-CD5 murine monoclonal antibody used to eliminate cutaneous T cell lymphoma (CTCL)⁹¹.

The MAGENTA study is a phase I, first-in-human, dose-escalation study of autologous anti-CD5 CAR-T (CD5 CAR) for patients with R/R T cell leukemia and lymphoma (NCT03081910). CD5 CAR is a second-generation CAR-T transduced with a gammaretroviral CD5-specific CAR possessing a CD28 costimulatory domain and without further genetic alteration of endogenous CD5 antigen. To mitigate T cell aplasia, CD5 CAR served as a bridge to alloHCT in a heavily pretreated patient population^92–94. Two distinct patient and CD5 CAR manufacturing cohorts have been reported at three distinct data cuts since 2019. An early description of conventionally manufactured CD5 CAR reported outcomes of 5 patients with relapsed and/or refractory non-Hodgkin T cell lymphoma (T-NHL). CD5 CAR dosing included 1 × 10⁷ CAR T cells/m² (n=2 patients) and 5 × 10⁷ CAR T cells/m² (n=5 patients), which is roughly 10-fold less than the original dose-finding studies of anti-CD19 CAR-T for B cell malignancies. Cytokine release syndrome (CRS) occurred in 3 of 5 reported patients (all at dose level 2) and Grade 2 neurotoxicity occurred in 1 patient that resolved with tocilizumab administration and supportive care. While most patients experienced prolonged lymphopenia, specifically within the peripheral blood CD3⁺ compartment, no patient had complete T cell aplasia with 3 patients undergoing CD3⁺ count recovery at 3 months following CD5 CAR infusion. Only one patient was documented to have viral reactivation (CMV and BK virus) requiring therapeutic antiviral therapy. At disease assessment at 4–8 weeks post CD5 CAR infusion, 3 patients achieved a best response of complete response (CR) with one patient proceeding to alloHCT.

Patients with T cell leukemias were also enrolled and stratified to receive conventionally manufactured CD5 CAR or CD5 CAR manufactured under selective pressure with the tyroskine kinase inhibitors dasatinib and ibrutinib to disrupt tonic T cell signaling and improve anti-leukemic activity by mitigating T cell exhaustion (n=7 patients)⁹⁴. An additional dose level of 5 × 10⁷ CAR T cells/m² was also enrolled within this phase, as compared to the T-NHL cohort. In the conventionally manufactured cohort (n = 8 patients), responses were poor, with an overall response rate (ORR) of 12.5%, although notably enrolled patients had significantly aggressive underlying disease with 62.5% possessing > 25% bone marrow blasts, 50% possessing extramedually disease, and 25% having received a prior alloSCT. No patients in the conventionally manufactured cohort developed a Grade 3 or greater CRS or ICANS adverse event and only one patient developed a Grade 3 or greater infection. In the TKI-manufactured cohort (n=7 patients) 4 of 7 patients were able to achieve a best overall response of an MRD negative CR. Two of these patients developed EBV viremia and ultimately passed as a result of treatment complications from post-transplant lymphoproliferative disease (PTLD). These unexpected fatalities were hypothesized to be driven by inadequate viral prophylaxis within EBV seropositive patients and possible suppression of CD5+ EBV-specific T cells secondary to CD5 CAR therapy. Collectively, these preliminary results serve as proof-of-principle that while CAR-T immunotherapy is potentially efficacious for T cell neoplasms and safe in the short-term immediately proceeding infusion, mitigation of long-term infectious predisposition must still be addressed either in protocol prophylaxis design and/or CAR T manufacturing. These findings have paved the way for alternative CD5-directed therapeutic strategies. Our group has developed Senza5^™ CART5, an autologous CD5 CRISPR-Cas9 knockout anti-CD5 CAR-T product (CD5KO CD5 CAR) with high on-target, low off-target specificity capable of maintaining reactivity against opportunistic pathogens⁸². Phase I clinical trial testing of Senza5^™ CART5 for patients with R/R CD5-expressing nodal T cell lymphomas is currently in preparation.

4.2. CD7

CD7 is a transmembrane protein of the immunoglobulin superfamily expressed on mature human T and NK cells, as well as cells in early stages of T-, B-, and myeloid cell differentiation where it plays an important role in T- and B-cell interactions during early lymphoid development^95,96. CD7 is found on >95% of lymphoblastic T cell leukemias and and a subset of PTCL⁹⁷, making it an attractive candidate target. Similar to CD5, CRISPR-Cas9 editing to eliminate CD7 expression in CAR-T limits fratricide, improves expansion efficiency, and demonstrates in vivo anti-tumor activity^46,98. Anti-CD7 CAR-T have been an attractive platform for testing alloCAR-T in T cell-derived malignancies, with phase I clinical trial data being recently reported.

The first robust, first-in-human, clinical trial dataset utilizing anti-CD7 CAR-T (CAR7) was reported in 2021 (ChiCTR2000034762)⁹⁹. A lentiviral construct possessing an anti-CD7 CAR with 4–1BB costimulatory domain additionally possessed CD7 and intracellular organelle anchoring domains to prevent CD7 protein release from the endoplasmic reticulum (termed IntraBlock technology). 20 pediatric and adult patients with R/R T-ALL enrolled with 16 patients receiving a target CAR-T dose (1 × 10⁶ CAR-positive cells/kg) and 4 patients receiving 0.5 × 10⁶ CAR-positive cells/kg due to expansion failure. CAR7 were manufactured from a previous alloHCT donor (12 patients) or new alloHCT compatible donor (8 patients). Grade 3 or higher CRS occurred in 2 patients and no Grade 2 or higher neurotoxicity events occurred. Endogenous CD7-positive T cells were depleted at 2 weeks following CAR7 infusion, however, the number of CD7-negative T cells increased in most patients (median: 379.80 T cells/uL of blood). Importantly, predominantly CD7-negative T cells isolated from peripheral were able to generate immune effector responses to both viral and fungal antigen challenges in vivo. Like CD5 CAR treated patients where complete T cell aplasia was not appreciated, notably, four patients had viral reactivation (EBV and CMV) requiring antiviral therapy and one patient with a pre-enrollment history of fungal pneumonia died of a fungal pneumonia-associated pulmonary hemorrhage. Overall, CAR7 were generally safe with adverse effects manageable at the data cut approximately 250 days post CAR-T infusion. 17 of 20 patients attained an MRD-negative CR at 30 days following CAR-T infusion, with 7 of 8 of those receiving donor-derived CAR-T receiving an alloHCT. A phase II multicenter study is now currently enrolling (NCT04689659).

A second institution reported their experience using autologous and donor-dervied “naturally selected” anti-CD7 CAR (NS7CAR) in patients with R/R T-ALL or T cell lymphoblastic leukemia (T-LBL) (NCT04572308)¹⁰⁰. Instead of using gene editing tools, NS7CAR, an anti-CD7 lentiviral CAR with 4–1BB costimulation and undectable surface CD7 expression, were allowed to naturally emerge in vivo. While the mechanism of CD7 negativity is unclear, it is hypothesized to develop from CD7 internalization or epitope masking, similar to CD5 CAR. Again, 20 adult and pediatric patients were enrolled to either autologous (18 patients) or donor-derived (2 patients) NS7CAR at one of three dose levels (16 patients received the second dose level of 1–1.5 × 10⁶ CAR-positive cells/kg) with 14 patients proceeding to alloHCT. Grade 3 or higher CRS occurred in only 1 patient. Like in the study by Pan et al., total peripheral blood T cell counts declined immediately following NSCAR7 infusion (median time 7 days) with reconstitution of CD7-negative T cells to pre-CAR-T baseline levels; no sustained T cell aplasia was observed. Two patients developed sepsis related to gram-positive cocci, and only one patient (a prior alloHCT patient) developed CMV reactivation. All but one patient treated with NSCAR7 developed a bone marrow MRD-negative complete response at Day 28 post-CAR T infusion. These preliminary positive results provide complementary rationale for ongoing anti-CD7 CAR-T phase I studies in the United States¹⁰¹ (NCT03690011)

As previously described, UCART7 is the first pan-T cell antigen targeting alloCAR-T and is currently under phase I clinical investigation (NCT04984356). In preclinical models, UCART7 was cytotoxic to both primary T-ALL cell lines and primary patient-derived T-ALL in vitro and in vivo, without developing significant xenogeneic GVHD. It remains to be seen whether UCART7 will yield sustainable, clinical responses, particularly in the absence of immune evasion by host innate immune cells. Potentially, UCART7 may fill a niche as a bridge to alloHCT or DLI, in particular with CD7-depleted stem cell donor pools to repopulate normal immune cell compartments⁷⁹.

4.3. CD4

CD4 mediates critical signaling mechanisms required for adaptive immunity by supporting both cytotoxic T and humoral cell responses, as well as mediating helper T cell function. Although considered to be a mature T cell antigen identifier, T-ALL blast subsets possess aberrant CD4 expression. With the intention as a bridge to alloHCT, third-generation anti-CD4 CAR-T (CD4CAR) were successfully generated using CD8⁺ T cells as a platform^41,102. Preclinical studies supported CD4CAR cytotoxicity against CD4-expressing tumor targets and a predominantly central memory-like phenotype following CAR-T expansion. CD4CAR is currently in an active and recruiting Phase I clinical trial with results pending (NCT03829540).

CD4CAR were further modified with an IL15/IL15sushi domain to exploit autocrine IL15 signaling to improve CAR-T memory and persistence (termed CD4-IL15/IL15sushi CAR-T)¹⁰³. To date, 3 patients with R/R T cell lymphoma treated with CD4-IL15/IL15sushi have been reported (NCT04162340) with dosing ranging from 2.8 × 10⁶ - 3.5.× 10⁶ CAR-positive cells/kg. No Grade 3 or higher CRS or neurotoxicity occurred, and no patient had severe opportunistic infections, despite two not treated with prophylactic antimicrobial therapy. In all patients, CD3 and CD8 dual-positive T cells expanded, which compensated for CD4⁺ T cell aplasia seen in the peripheral blood within 30 days of CD4-IL15/IL15sushi CAR-T infusion. One patient in long-term follow-up after achieving a CR had increased dual CD3⁺ and CD4⁺ T cells after 3 months, suggesting loss of CD4-IL15/IL15sushi CAR-T persistence. It remains to be seen what long-term adverse events are seen because of prolonged CD4⁺ T cell aplasia.

4.4. CD30

CD30 is a member of the tumor necrosis factor receptor (TNFR) family whose function is important for T cell proliferation and cytokine production. CD30 expression is limited to a subset of activated T cells, allowing anti-CD30 CAR-T (CART30) to be generated with minimal fratricide. Due to its association with classical Hodgkin Lymphoma (cHL), CART30 are largely being shown to be efficacious against B cell malignancies ^104–106. However, CD30 is also found in a significant proportion of PTCL, including near 100% expression of anaplastic large cell lymphoma (ALCL)¹⁰⁷ for which CD30-CAR-T could prove particularly effective. Although there are no dedicated early phase clinical trials that are adequately statistically powered to assess clinical endpoints of CART30 against T cell neoplasms, several ongoing trials have included rare patients with R/R CD30-expressing PTCLs concurrently with R/R cHL patients (NCT04083495, NCT04526834, NCT02917083).

4.5. CD70

CD70 is a type 2 transmembrane glycoprotein that is a member of the tumor necrosis factor (TNF) ligand family¹⁰⁸. CD70 was initially shown to display high expression on malignant hematopoietic stem cells and solid tumors, but more recently, constitutive CD70 expression has also been robustly decribed in activated T cell leukemias and lymphomas^109–112. Anti-CD70 targeting CAR-T are currently being evaluated; three of which are specific for patients with solid tumors, one specific for patients with B-cell derived malignancies, and one which includes patients with T cell lymphoma (NCT04502446). The latter study (COBALT-LYM) is a phase I, dose-escalation study of an allogeneic anti-CD70 targeting CAR-T (CTX130) with multiplexed CRISPR/Cas9 deletion of TRAC, β2 microglobulin, and CD70 (to mitigate fratricide of activated CAR-T)¹¹³. Preliminary results of 15 patients with R/R T cell lymphoma treated with CTX130 showed a 71% overall response rate (ORR) and 29% CR rate without any dose-limiting toxicities. Notably, one patient experienced at least a Grade 3 infection at the first dose level, and one patient with an underlying genetic syndrome had sudden death secondary to a lung infection unrelated to CTX130. These responses and generally acceptable safety profile provide encouraging data for the utility of alloCAR-T products against T cell neoplasms.

5. The future of rational CAR-T engineering for T cell neoplasms

Lessons learned from early phase cellular therapy trials, in conjunction with advancements in gene-editing precision and single-cell omics, have spearheaded more rational CAR-T design and manufacturing (FIGURE 2). One potential powerful avenue to circumvent several issues of CAR-T development for T cell neoplasms is base editing. Base editing employs highly specific enzymes, such as the nickase version of Cas9 enzyme, together with a gRNA to introduce single-nucleotide variants (SNVs)¹¹⁴. Because base edited genes are dependent on single-strand DNA breakage and repaired with user-specified, single-nucleotide point mutations, alteration of specific functional motifs while retaining the endogenous protein structure and function is possible. Base editing machinery can be delivered to cells through conventional viral delivery approaches, however, lipid-based ribonucleoprotein complexes and nanoparticles can also facilitate in vivo gene editing without viral machinery^115–117.

FIGURE 2. Schematic of current and proposed strategies to circumvent chimeric antigen receptor T cell (CAR-T) challenges for implementation against T cell derived malignancies.

CAR T cells (blue) with specificity against T cell antigens (red) may be artificially or naturally modified using several techniques to bypass manufacturing challenges that are unique to T cell neoplasms. 1) Protein expression blocker systems (PEBL) prevent trafficking of endogenous T cell antigens to the cellular membrane and prevent fratricide. 2) Direct gene editing of endogenous T cell antigens using either CRISPR/Cas9 or base editing techniques prevent endogenous T cell antigen protein production and prevent fratricide. 3) Drug-inducible switches can be used to either force expression of CAR T cell pro-apoptic machinery to prevent T cell aplasia, or, can force downregulation of transcribed CAR transgene to mitigrate CAR T fratricide. 4) Naturally selected CAR T cells, either through the direct manufacturing of concomitant T cell antigen negative/CAR-negative T cells, or through the clonal expansion of T cell antigen-deficient subpopulations, can be used to prevent both CAR T cell fratricide and T cell aplasia while maintaining effector T cell activity mediated by the endogenous T cell receptor (TCR). 5) Antigen restriction of CAR specificity to only a subset of T cells (i.e. choosing CD4- or CD8- single expressing T cells) can be used as a method to prevent complete T cell aplasia.

Base editing can be applied in CAR-T engineering by introducing SNVs within an scFv-binding peptide sequence of endogenous T cell antigens to prevent CAR engagement. Such an approach would permit fratricide-resistant CAR-T while retaining normal endogenous protein function possibly required for optimal T cell activity. Indeed, this has been recently demonstrated in both anti-CD3 and anti-CD7 CAR-T, the latter of which in tandem with multiplexed base edited targets generates an allogeneic anti-CD7 CAR-T product ^118,119. In September 2023, the first three patients with pediatric T-ALL were treated with BE-CAR7, an off-the-shelf, allogeneic, base-edited anti-CD7 CAR-T in the United Kingdom (ISRCTN15323014)¹²⁰. BE-CAR7 utilizes ablation of TRBC, CD7, and CD52 to minimize GVHD, decrease CAR T cell fratricide, and enhance CAR T resistance to anti-CD52 lymphodepleting chemotherapy, respectively. Two of the three patients were able to develop an MRD negative remission in the bone marrow at Day 27 or 28 following BE-CAR7 infusion, with the other patient developing an MRD positive morphological and flow-based remission at Day 25. One patient who also had CNS involvement also showed an MRD negative remission within the CSF with intravenous infused BE-CART7. These two patients achieving MRD negative remissions were advanced to alloSCT. Similar to the previously described clinical trials with alternative CAR T products, one patient developed Candida fungemia and respiratory infections with Aspergillus fungal species, the latter of which was fatal. While promising from a preliminary efficacy standpoint, these BE-CAR7 as currently design still carry significant immunosuppressive and cytopenic risks that must still be addressed in future product and protocol iterations.

Improvements in multiomic techniques have allowed for a finer scrutiny of immunophenotypic differences between malignant and non-malignant T cells. These technologies are likely to play key roles in identifying novel, single-antigen CAR T targets unique to neoplastic T cells that can be employed to minimize T cell aplasia. One novel single agent target, CCR9, was recently shown to be expressed in >70% of pediatric T-ALL cases and <5% of CD3⁺ healthy donor T cells¹²¹. Anti-CCR9 CAR-T possessed anti-tumor activity in patient-derived xenograft models of T-ALL, including tumor cells with CCR9^low antigen density. Another opportunity takes advantage of clonal TCR expression with regards to TRBC1 or TRBC2 expression in PTCL. While normal, healthy T cells heterogeneously express either TCR beta chain constant region genes, clonal PTCL are homogeneously restricted to TRBC1 or TRBC2; an anti-TRBC1 CAR-T could preferentially target malignant clones and only half the normal T cell compartment⁵⁸. Indeed, anti-TRBC1 CAR-T cleared both malignant and non-malignant TRBC1-positive T cells, without off-targeted destruction of TRBC2-positive T cells. However, the recently presented phase I AUTO4 clinical trial, using an autologous anti-TRBC1 CAR-T for patients with TRBC1-expressing PTCL, resulted in 3 of 10 patients with Grade 3 or higher serious treatment-emergent adverse events despite a 40% CR rate¹²². Furthermore, unlike other autologous CAR-T products, no peripheral CAR T expansion was detected, which may correlate to short duration of response in the future. This study is continuing to enroll with modified CAR T manufacturing to better support infusion of a naïve CAR-T product (NCT03590574).

Single CAR antigen targets, particularly those with low antigen density, increases the probability of antigen escape and leads to disease relapse. Like recent efforts with B cell leukemias and lymphomas^123–125, dual- and tandem-ScFv expressing CARs can be used to simultaneously target multiple T cell antigens. However, careful consideration must be made regarding 1) the knockout efficiency of multiplexed T cell antigen genes and 2) the size of the CAR constructs such that a clinically acceptable T cell transduction efficiency is maintained. Tandem anti-CD5 + anti-CD7 and dual anti-CD3 + anti-CD7 have demonstrated preclinical efficacy and are primed for testing in phase I clinical trials^118,126.

Lastly, innovative treatment platforms using multipurposed CAR-T are likely to be pursued clinically. An off-the-shelf, allogeneic anti-CD45 CAR-T product could conceivably be used as a comprehensive cell therapy for all hematological malignancies, as CD45 is universally expressed on all mature non-erythroid hematopoietic cells and is required to sustain hematopoiesis. Theoretically, treatment with an anti-CD45 CAR-T could be combined with autologous or alloHCT in which CD34⁺ hematopoietic stem and progenitor cells (HSPCs) could present a base-edited CAR-resistant CD45 molecule. A preliminary preclinical study intriguingly demonstrates base-edited anti-CD45 CAR-T are fratricide resistant and that base-edited CD45 HSPCs engrafted and differentiated in immunodeficient mice¹²⁷.

6. Conclusion

T cell-derived hematological malignancies confer an extremely poor prognosis due to tumor heterogeneity, chemotherapy-refractoriness, and intrinsic aggressivity. Lessons learned in treating B cell-derived hematological malignancies with anti-CD19 and anti-BMCA CAR-T have spearheaded innovative engineering and manufacturing strategies to overcome unique barriers associated with T cell leukemia and lymphoma treatment with cellular immunotherapy. These innovations have preliminarily demonstrated efficacy and are further destined to improve clinical outcomes in the future.

TABLE 1.

Summary of key clinical trials of FDA-regulated cell therapy products against T cell neoplasms as registered on clinicaltrials.gov.

Trial Number	Sponsor	Target	Phase	Product	Disease	Status	Ref.*
NCT04712864	Legend Biotech USA, Inc.	CD4	I	Autologous	R/R CD4+ PTCL R/R CTCL and MF	Terminated
NCT03829540	iCell Gene Therapeutics	CD4	I	Autologous	R/R T-cell leukemia, T-NHL	Recruiting
NCT03081910	Baylor College of Medicine	CD5	I	Autologous	R/R T-ALL, T-NHL	Recruiting	92,94
NCT03690011	Baylor College of Medicine	CD7	I	Autologous	R/R T-ALL, T-NHL	Recruiting
NCT04984356	Wugen, Inc.	CD7	I/II	Allogeneic	R/R T-ALL, T-LBL	Recruiting	128
NCT05377827	University of Washington	CD7	I	Allogenic	T-NHL, AML	Recruiting
NCT04083495	University of North Carolina – Chapel Hill	CD30	II	Autologous	R/R PTCL	Recruiting	129
NCT02690545	University of North Carolina – Chapel Hill	CD30	I	Autologous	R/R HL, NHL	Recruiting	105,130
NCT02663297	University of North Carolina – Chapel Hill	CD30	I	Autologous	HL, NHL in remission	Active, NR	131
NCT04526834	Tessa Therapeutics	CD30	I	Autologous	R/R PTCL, DLBCL, PMBCL	Active, NR
NCT02917083	Baylor College of Medicine	CD30	I	Autologous	R/R HL, NHL	Recruiting	105
NCT01192464	Baylor College of Medicine	CD30	I	Autologous cytotoxic T lymphocytes	R/R HL, NHL	Active, NR
NCT04288726	Baylor College of Medicine	CD30	I	Allogeneic	R/R NK/T-cell lymphoma, HL	Recruiting	132
NCT03049449	National Cancer Institute	CD30	I	Autologous	R/R NHL	Completed	133
NCT03602157	University of North Carolina – Chapel Hill	CD30 + CCR4	I	Autologous	NHL	Recruiting	134
NCT04136275	Massachusetts General Hospital	CD37	I	Autologous	R/R CD37+ hematological malignancy	Recruiting	135
NCT04502446	CRISPR Therapeutics	CD70	I	Allogeneic	R/R T-NHL	Recruiting	113
NCT03590574	Autolus Limited	TRBC1	I/II	Autologous	R/R T-NHL	Recruiting	122

PCTL = peripheral T-cell lymphoma; CTCL = cutaneous T cell lymphoma; MF = mycosis fungoides; T-ALL = T cell acute lymphoblastic leukemia; T-NHL = T cell Non-Hodgkin lymphoma; T-LBL = T cell lymphoblastic lymphoma; AML = acute myelogenous leukemia; DLBCL = diffuse large B cell lymphoma; PMBCL = primary mediastinal B cell lymphoma; HL = Hodgkin lymphoma; NHL = Non-Hodgkin lymphoma; R/R = Relapsed and/or refractory, TRBC1 = T-cell receptor p-chain constant region 1.

^*Ref. = Most recent reference is provided for reported and publicly distributed clinical trial results as of October 2023.

Highlights:

• CAR-T efficacy for T cell antigens necessitates innovative engineering strategies

• Preliminary clinical trial data suggests CAR-T are safe against T cell neoplasms

• Base edited and multi-targeted CAR-T hold clinical promise against T cell neoplasms

Data-sharing statement:

There are no data to share.