Engineered T-cell Receptor T Cells for Cancer Immunotherapy

Adoptive cell therapy is a rapidly advancing field of medicine that is now focused mostly on cancer immunotherapy but already extending to immune manipulation against infections and autoimmune diseases. One of the first FDA-approved cellular therapy products was sipuleucel-T, which consists of autologous antigen-presenting cells (APC) cultured ex vivo and administered to patients with prostate cancer. Newer classes of cellular therapy include tumor-infiltrating lymphocytes (TIL), which recognize tumor-associated antigens (TAA), but are often suppressed in the tumor microenvironment (TME), precluding efficient killing of cancer cells. Nevertheless, TILs, which are isolated from tumor samples, show clinical efficacy after expansion in culture, especially if the patient was not previously exposed to checkpoint inhibitors (1). TIL therapy was recently combined with checkpoint inhibition or vaccination with tumor lysate–loaded dendritic cells (DC), and this elicited prolonged antitumor responses in some patients (2). Indeed, TILs have shown promising preliminary results against various tumor types, mainly melanoma, and research is ongoing to develop ways to more carefully select and expand TILs aimed at specific TAAs, and to enhance TIL antitumor efficacy.

Advances in genetic engineering have also enabled production of cells with receptors directed at antigens presented on cancer cells. The chimeric-antigen receptor (CAR) T-cell therapies tisagenlecleucel, axicabtagene ciloleucel, idecabtagene vicleucel, and lisocabtagene maraleucel became the first commercially approved engineered cellular products after showing remarkable efficacy against hematologic malignancies. Unfortunately, CAR T cells have been less effective against solid tumors (3). CAR T cells can only detect surface TAAs, yet suitable antigens are challenging to identify in many tumors, especially since most aberrant cancer proteins are intracellular and are only expressed in the context of MHC complexes after proteasomal degradation.

Constructs encoding a T-cell receptor (TCR) that recognizes MHC–antigen complexes can be generated and introduced into T cells (known as TCR T cells), and the ensuing TCR–peptide-MHC interaction can be exploited to trigger an immune response (Fig. 1). Importantly, TCR T cells elicit lower maximal cytokine levels at high antigen density, whereas some CAR T cells can prompt significant immune toxicities that may even be life threatening (4). In this review, we summarize preliminary TCR T-cell clinical trials, and survey preclinical studies seeking to enhance the safety and efficacy of this modality through better construct design, combination with other treatments, better TAA identification, and improved platform selection.

MHC class I molecules are expressed on all nucleated cells, and present peptides derived from intracellular proteins to CD8⁺ T cells (Fig. 1). Conversely, MHC class II molecules are expressed predominantly on APCs, and they usually present extracellular peptides, internalized through endocytosis, to CD4⁺ T cells. In both cases, the TCR–MHC interaction is HLA subtype–restricted, which necessitates a specific match between the TCR T cellular therapy product and the patient's MHC haplotype. Furthermore, TCR–MHC interactions can only take place in the presence of an MHC-bound peptide, and each TCR is specific to an MHC–peptide complex, although some cross-reactivity with other MHC–peptide complexes can occur. In addition, T-cell activation requires adjacent molecules to interact with target cells. For example, the CD8αβ coreceptor stabilizes the TCR–peptide–MHC complex, and triggers the intracellular T-cell activation cascade (5). The CD3 complex participates in this process as well, as it comprises several motifs that when activated lead to T-cell proliferation, activation, and survival.

Several TAAs presented on MHC molecules have been identified, enabling the engineering of TCR T cells that recognize such TAAs (Table 1). Many trials of these cells are in progress or have reported preliminary results. One of the first trials to report results evaluated engineered TCR T cells that recognize a cancer-testis antigen, NY-ESO-1, normally expressed in spermatogonia and placenta, but lost during spermatid differentiation and not expressed on healthy tissues. It is overexpressed on epithelial tumors, as well as in synovial cell sarcoma and melanoma. In a trial using TCR T cells targeting NY-ESO-1, 4 of 6 patients with synovial cell sarcoma and 5 of 11 patients with melanoma had objective clinical responses (6). The responses seen in patients with synovial sarcoma may be at least partially attributable to the conditioning regimen administered prior to the TCR T cells, as alkylating agents demonstrated responses in this etiology (7). However, 2 of the patients with melanoma who had a response demonstrated complete responses (CR) lasting over a year. In fact, NY-ESO-1 has been a preferred TAA for TCR T-cell trials due to its relatively abundant expression on various tumors and lack of on-target, off-tumor effects (8–10). On the other hand, a dose-escalation study of TCR T cells targeting MAGE-A10, a TAA presented on cancers like melanoma, head and neck, urothelial, and non–small cell lung cancers (NSCLC), showed TCR T-cell persistence in peripheral blood, but no antitumor effects (11). Remarkably, 1 patient receiving TCR T cells targeting MAGE-A3 in another dose-escalation trial achieved CR lasting 29 months, and 3 of 9 patients receiving the full dose of the TCR T cells achieved partial response (PR), one of which lasted more than 19 months (12). Researchers are focusing on ways to detect TAAs presented on MHC molecules, as well as on ways to enhance the ability of TCR T cells to recognize, bind to, and attack cancer cells, and to recruit other immune components that can reinforce the immune response. These techniques, and clinical trials using the resultant enhanced TCR T cells, are discussed in the following sections.

For effective and safe administration of cellular therapy, an ideal target antigen must be uniformly expressed on tumor cells but absent from healthy tissue. Such antigens include tumor-specific neo-antigens produced as a consequence of cancer-associated mutations, e.g., KRAS mutations. The proteins encoded by viral oncogenes can also be targets for TCR T cells, e.g., E6 and E7 in human papillomavirus (HPV)–associated tumors. One trial of TCR T cells targeting HPV16 E6 elicited objective tumor responses at higher doses in 2 of 12 patients (13). No off-tumor toxicity was observed. Similarly, a trial targeting HPV16 E7 elicited objective clinical responses in 6 of 12 patients (14). TAAs can also consist of proteins produced by abnormal transcription of silenced genes, such as some members of the MAGE antigen family, α-fetoprotein (AFP), and carcinoembryonic antigen, as well as cancer-testis antigens such as PRAME and synovial sarcoma X breakpoint 2 (SSX2), neither of which are expressed on healthy tissue except testis and placenta (15, 16). Other possible TAAs are antigens overexpressed on tumors, such as WT1 and mesothelin, which are observed on healthy tissues at concentrations low enough not to trigger immune activation.

One of the main challenges in target selection for TCR T cells is cancer-cell plasticity. Due to the high division rate, often with chromosomal instability, cancer cells mutate at a high rate, and at times, different areas of a tumor display different antigens, or different densities of a given antigen (17). Furthermore, cells have been shown to downregulate cell surface proteins, as well as downregulate TAA presentation on MHC molecules (18). Alternative splicing and proteasome degradation can also cause the presentation of different peptides on MHC molecules, even without actual genetic mutations. These genetic and proteomic mechanisms can explain tumor resistance to TCR T cells, as well as loss of response or relapse of previously responding cancers. Developing TCR T cells that target multiple domains is a possible strategy to mitigate differential antigen presentation (19), and to improve cell expansion and persistence to prevent mutational escape over time.

TCR T-cell therapy carries an inherent risk that an engineered TCR will recognize healthy tissues that express the same antigen, triggering immune attack, and that TCR T cells will cross-react with similar antigens. Although extensive cell line and animal testing is always performed prior to patient administration, it is sometimes hard to predict the toxicities that may occur. For instance, marked reduction in circulating carcinoembryonic antigen was achieved in a Phase I trial of TCR T cells targeting this TAA, yet inflammatory colitis developed in all 3 patients in whom this reduction was observed (20). In another study of TCR T cells expressing an affinity-enhanced TCR against MAGE-A3, the first 2 patients to receive therapy developed severe cardiotoxicity, leading to cardiogenic shock and death within a few days of product administration (21), although no MAGE-A3 expression was found on myocardial tissue on autopsy, despite evident myocardial damage. Subsequent studies demonstrated that the engineered TCR T cells also targeted titin, a protein presented on HLA molecules on cardiac myocytes that is unrelated to MAGE-A3. Titin was not detected in preclinical studies, as it is only presented on beating cardiomyocytes. In a different trial of TCR T cells targeting MAGE A3, 3 of 9 patients had severe neurotoxicity that led to death in 2 (22). This was caused by cross-reactivity of TCR T cells with MAGE-A12 in brain tissue.

These dramatic off-tumor effects highlight the need for careful target selection and safety testing that takes into account not only TCR affinity and specificity toward the TAA, but also TAA expression on malignant and healthy tissues (Fig. 2). Complex TCR interactions may preclude complete assurance of product safety prior to clinical trials since both the target and the TCR must be carefully tuned for efficacy and safety. Some adoptive cell therapies have used a “kill switch” as a conditional way to stop T-cell activity through activation of an apoptosis pathway in case of severe toxicities. However, as with other introduced transgenes, this strategy has been shown to induce immune-mediated elimination of the engineered cells (23). Less immunogenic transgenes are being developed (24, 25), and this important way to overcome unexpected toxicities is now being explored preclinically (15).

One suggested method for generating safe and effective TCR T cells is to use T cells from patients responding to treatment with vaccines designed to induce cancer-specific T-cell responses (26). T cells from such patients have been generated by expanding peripheral blood mononuclear cells (PBMC) using a panel of overlapping peptides that collectively encompass the full NY-ESO-1 sequence, with selected cells further expanded into clonal cell lines. This method has also been used for T-cell selection from healthy donors (27), who typically have low-affinity TCRs against WT1, which is expressed at very low levels in normal adult tissues. However, some healthy individuals express higher-affinity TCRs that may be more active against WT1-presenting tumor cells but are unreactive against normal tissues. Accordingly, one higher-affinity, optimal TCR was identified after testing of more than 1,000 T-cell clones from more than 50 healthy subjects with the HLA-A2⁺ haplotype. This TCR was engineered into T cells derived from donors of allogeneic stem cells to patients with acute myeloid leukemia, and the cells transfused to the latter following the initial stem-cell transplant. Twelve patients at high risk for relapse received these TCR T cells, and at a median follow-up of 44 months, all patients had no evidence of disease relapse and did not require any further treatment. When compared with a group of patients with similar disease characteristics who did not receive the TCR T cells, the study cohort had significantly higher overall survival and relapse-free survival.

In another trial (28), tumor-specific neoantigens were found by whole-exome sequencing of cells from tumor samples, and used to identify and expand T cells, the TCR of which was then cloned and engineered into TCR T cells from healthy donors with the same HLA subtype. These cells reacted to cells bearing the neoantigenic peptides. As above, the risk of off-tumor effects was minimal, as these TCRs were found in patient T cells. In the future, this technique may enable the design of patient-specific TCR T cells based on the mutational profile of the patient's own tumor. Similarly, patient DCs pulsed with cancer-related peptides such as Hormad1, a cancer-testis antigen found in lung, esophageal, and other head and neck tumors, allowed for selection and cloning of TCR T cells for therapeutic use in the same patients (29). Others have tried to augment TCR T-cell activity with direct injection of their target protein, e.g., MAGE-A or WT1, but no clinically significant benefit has been observed (30, 31).

Cellular therapy products are extensively screened for activity against healthy tissue. However, some of this potential activity may not be apparent in vitro, as described above and as observed with the HLA-A*01–restricted MAGE-A3 TCR T cells that elicited cardiogenic shock following cell infusion due to cross-reactivation with a protein expressed in striated muscle, titin (21). To enhance screening for possible off-tumor effects, some investigators have tested each possible mutation in tumor-associated proteins for T-cell activation against primary cell lines. This systematic screening is possible due to the small number of possible mutations in small proteins, e.g., in a cell line targeting MAGE-A10, all 172 possible amino-acid substitutions were tested (32). Moving forward, computational methods may also be used to analyze immune-cell repertoires for use as possible targets for adoptive cell therapy (33).

The avidity and affinity of the TCR–MHC peptide complex defines the strength of the interaction between a single TCR and the peptide–MHC complex (Fig. 2). Avidity relates to a broader phenomenon: the interaction of many TCR molecules, as well as their coreceptors, with target cells (34). The specific affinity of the TCR for the MHC–peptide complex, as part of the general avidity between interacting cells, can significantly affect cytotoxic activity. In a trial of TCR T cells targeting MART1 in patients with MART1-positive melanoma, only 2 of 15 patients had objective responses (35). A higher-affinity TCR boosted the response rate to 30%, although nonmalignant melanocytes were also affected, resulting in uveitis, hearing loss, vitiligo, and other off-tumor effects (36). These trials show that affinity and avidity are crucial in achieving efficacy while controlling off-tumor effects, which may worsen with high-affinity TCRs.

The delicate balance of affinity and avidity sufficient to induce an effective antitumor response, sparing healthy tissue presenting the same TAAs at a lower density, can be achieved by tuning TCR affinity to target only cells abundantly expressing the TAA, while sparing healthy cells with low TAA expression. AFP, a known TAA also used for monitoring hepatocellular carcinoma, is expressed at low concentrations on nonmalignant liver cells, and usually only during inflammation and stress conditions. Some researchers are developing specific peptide-enhanced affinity receptor (SPEAR) T cells in which the TCRs of a patient's T cells are engineered in silico and ex vivo to target cells with higher antigen levels, such as malignant liver cells expressing AFP, while sparing nonmalignant cells (37). These are now being used in preliminary clinical trials, and initial safety data show no on-target, off-tumor toxicity from AFP-targeted SPEAR T cells in hepatocellular carcinoma (38), MAGE-A4–targeted SPEAR T cells in sarcoma (39, 40), and MAGE-A10–targeted SPEAR T cells in melanoma, head and neck, urothelial, and NSCLCs (11). Another report of SPEAR T cells targeting NY-ESO-1 showed promising results in 42 patients with synovial sarcoma, with 14 experiencing PR and one experiencing CR (41). The response was associated with peak TCR T-cell levels, which in turn was associated with lymphodepletion doses. It is important to note that higher-affinity constructs may lose specificity, and thus may have reduced efficacy (42). A fine balance between affinity and specificity is therefore necessary when designing and selecting TCR constructs (Fig. 2).

Many of the first TCR T-cell clinical trials did not demonstrate clinically significant efficacy (10, 11, 31, 43). Although the use of MHC–bound peptides as targets for engineered TCR T cells exploits T-cell cytotoxicity to induce tumor-cell killing, T cells are under a complex system of checks and balances to control cytotoxic activity and direct it only at specific targets, preventing autoimmune phenomena and damage to healthy tissue due to uncontrolled immune responses. This intricate regulatory network is important at steady state yet may pose substantial limitations on immune reactions against cancer cells, which may be identified as bearing self-antigens, or even deploy heightened immune-suppressing mechanisms that provide tumors with survival advantage. Thus, researchers are studying ways to overcome these obstacles and thereby improve TCR T-cell efficacy.

Some investigators have used a vector that encodes both the therapeutic TCR and siRNAs that silence the T cell's endogenous TCR (31, 44) to improve efficacy and to avoid “combined” TCRs that may trigger autoimmune toxicity (45). In a study of 9 patients with endometrial cancer (n = 1), ovarian cancer (n = 1), melanoma (n = 3), and synovial sarcoma (n = 4), who received cells engineered to express a TCR targeting NY-ESO-1 and siRNA to silence the native TCR, 2 patients achieved PR and 5 achieved stable disease (SD; ref. 46). Other investigators have used a novel CRISPR/CAS9 mechanism that does not require transduction with a viral vector to knock out native TCRs and replace them with engineered cancer-specific TCR (47). In another study that used a viral vector following CRISPR/CAS9 mediated TCR knockout, the transgene was expressed at a higher level due to the lack of competitive native TCR expression (48). A different group generated engineered cells with the native TCR knocked out and demonstrated that these cells yielded a stronger response to cancer cell lines, with higher antigen sensitivity, than transduced cells that did not undergo native TCR deletion (49). In a subsequent Phase I study of such cells targeting NY-ESO-1, all 3 treated patients had durable cell engraftment, but NY-ESO-1 expression on tumor cells declined over time, signifying tumor evasion (18).

Cell persistence and expansion has been a concern in adoptive cell therapy. This is especially true for TCR T cells, which bear a structure similar to the native TCR, and thus may require high antigen exposure to activate and persist. Accordingly, many clinical protocols incorporate cytokine injection, mainly IL2 administration for up to 14 days, to enhance cell expansion, engraftment, and cytotoxicity (50). To overcome low antigen presentation by some tumors, DCs pulsed with viral peptides have been given to patients with melanoma together with MART1-directed TCR T cells. In this manner, transgenic T cells are exposed to more presented antigens, and, importantly, to antigens presented by professional APCs. A clinical trial using this method had promising results, with 9 of 13 (69%) patients experiencing tumor regression (43). Another study of TCR T cells against NY-ESO-1 included, in one arm, three subcutaneous shots of DCs pulsed with NY-ESO-1 peptide, once every 2 weeks, following TCR T-cell infusion (51). Another arm in that trial included the anti-CTLA4 immune checkpoint inhibitor ipilimumab in addition to DCs to bypass checkpoint blockade. Of the first 10 patients with sarcoma or melanoma treated, all with progressive stage 3 or 4 disease, 4 of 6 patients treated with DCs showed evidence of tumor regression and 2 of 4 patients treated with DCs and ipilimumab showed similar outcomes. Interestingly, 1 patient has an ongoing CR lasting more than 4 years without additional therapy.

The use of checkpoint inhibition is also being explored in a clinical trial, now recruiting, that will compare two cohorts of patients with multiple myeloma receiving TCR T cells targeting NY-ESO-1/LAGE-1a with or without the anti-PD1 immune checkpoint inhibitor pembrolizumab (52). Other small molecules have been examined as adjuncts to TCR T-cell therapy, either to enhance the immune response or overcome intrinsic tumor-related mechanisms that impede it.

The TME, especially in solid tumors, is inhibitory to immune cells. This is both due to harsh conditions, including hypoxia, low pH, and electrolyte disturbances as a result of accelerated and uncontrolled tumor growth (53), and to mitochondrial dysfunction and upregulation of coinhibitory molecules that prevent sustained T-cell antitumor activity and lead to T-cell exhaustion (54). Likewise, tumor size and decreased blood supply relative to healthy tissue can impair T-cell migration and trafficking. For example, DCs lacking CD103 in the TME impede immune-cell infiltration, thereby promoting tumor immune escape (55). Immune inhibitory pathways enhanced by tumor cells and causing immune evasion may also impair T-cell cytotoxicity, as is the case with the TGF-β pathway and its activation of T regulatory cells (Treg) and several other suppressive cell types (56, 57).

To overcome these inhibitory pathways, novel TCR T and CAR T cells are being developed. Coexpression of a dominant-negative TGF-β receptor and a TCR T cell has been shown in a mouse model to reduce the number of Tregs in the TME (58). Other costimulatory receptors have been coexpressed to mitigate the immune-suppressive TME. For example, a receptor for IL4, an immune inhibitory cytokine, was fused with the intracellular domain of the IL7 receptor, an activating cytokine, to overcome IL4-mediated immune suppression in pancreatic cancer (59). Cytokine-producing motifs are also being incorporated to improve cell persistence via autocrine signaling (60, 61). For example, TCR T cells producing IL18, in comparison with the same cells producing IL12, shifted the immune milieu in the TME toward CD8⁺ cytotoxic T cells, and resulted in reduction of melanoma tumor burden and prolonged survival in mice (62).

Several groups are working on constructs incorporating the advantages of the native TCR and its cell-activating machinery with a CAR-like antibody-derived binding domain (Fig. 3). For example, an antibody-based CD19-binding domain fused to a TCR and termed a "T-cell receptor fusion construct" (TRuC) showed better in vivo activity when compared with CAR T cells (63). One advantage of this novel receptor is that binding is mediated by a single-chain variable fragment, and it is therefore not HLA-dependent like “standard” TCR T cells, paving the way for future allogeneic applications. Another study found that transgenic coexpression of both CD8αβ and a tumor antigen–specific TCR enhanced tumor-cell killing through upregulation of transcriptional pathways that led to antigen-dependent cytotoxicity with less T-cell exhaustion (5). Furthermore, transgenic CD8αβ expression can maintain previous cell specificity to other antigens, such as viral peptides, and still allow for activity against tumor cells via the engineered TCR (64). The SURPASS trial, based on an affinity-enhanced MAGE A4-directed TCR and a CD8α coreceptor, showed 2 of 5 patients had PR, and the 3 others SD (65). Interestingly, CD4⁺ T cells engineered with these receptors exhibited direct cytotoxic activity in vitro. These improvements may prove instrumental in enhancing TCR T-cell antitumor activity through better target recognition, cell persistence, and decreased exhaustion.

DNA methylation is known to suppress expression of various tumor-related antigens, and the addition of an epigenetic modulator to NY-ESO-1–targeted therapy in a mouse model of breast cancer elicited expression of NY-ESO-1 in 5 of 7 cell lines that did not previously express it (58). This, in turn, caused significantly enhanced tumor infiltration and control by TCR T cells in comparison with administration of TCR T cells alone. Interestingly, TCR transgene expression may also be controlled through epigenetic mechanisms. Although insertion into DNA occurs during TCR T-cell manufacturing, leading to transgene persistence, TCR transgene expression, as can be measured by transgene RNA levels and TCR cell-surface expression, can fluctuate over time. Several studies have shown rapidly declining TCR T-cell persistence over time (43). Furthermore, retroviral or lentiviral promoter regions can be silenced through epigenetic methylation or histone modification (66, 67), hampering transgene expression (68). In a follow-up study of the aforementioned MART-1– and NY-ESO-1– TCR T cells, comparison of samples collected from 16 patients on day of and day 70 after infusion showed large differences in RNA and cell-surface expression between patients that were not due to loss of transgenic TCR DNA in circulating cells (69). Patients with reduced TCR surface expression had increased levels of methylation at the transgenic viral promoter region at day 70. Another study showed that exhausted CAR T cells could regain functionality with drug-induced “rest periods” (i.e., intermittent deactivation) through epigenetic and other mechanisms (70). Epigenetic control of transgene expression as well as other cellular stemness states are subjects of ongoing research and may provide further avenues toward enhancing TCR T-cell therapy.

Another important consideration for successful cellular therapy is the choice of platform to be used for adoptive cell therapy in general, and for TCR T cells specifically. T cells have different maturation and differentiations subtypes, and although most T cells actively fighting infection are cytotoxic, these cells are short lived. T cells with memory phenotypes are usually home to the lymph nodes and may be better candidates than cytotoxic T cells for adoptive cell therapy due to persistence and ability to mount an immune response upon rechallenge with the cognate antigen (71–73). Some CD8⁺ T cells, termed “veto cells,” can cause apoptosis of T cells targeting their HLA (74). Thus, veto T cells have been proposed as a platform for off-the-shelf allogeneic CAR or TCR cell therapy, with prolonged persistence due to suppression of host rejection (75). Researchers have also proposed inhibition of the AKT pathway as a method of inducing T-cell expansion without terminal differentiation and exhaustion (76). SPEAR T cells produced with AKT inhibition during the expansion phase showed enhanced responses, better proliferation even in the presence of prolonged antigen exposure, and improved cytotoxic activity after restimulation (77). Other immune cells have been used as adoptive cell platforms, and natural killer cells bearing CARs have demonstrated clinical responses in CD19 malignancies (78). Hematopoietic stem cells derived from umbilical cord blood and induced pluripotent stem cells from healthy donors have also been used as sources for cellular therapy platforms in preliminary clinical trials and may provide an “off-the-shelf” alternative to autologous cells (79–82). For example, CD8αβ T cells derived from induced pluripotent stem cells were transduced with a TCR targeting GPC3 and tested in NOD/SCID gamma (NSG) mice against cell lines expressing GPC3, eliciting a marked reduction in tumor growth (83).

Preliminary trials using TCR T-cell therapy usually recruited small numbers of patients (Table 1). These were mostly Phase I or limited Phase II trials, focusing mostly on safety or initial signals of efficacy. As TCR T-cell therapy design advances and more promising results are demonstrated, the idea of using off-the-shelf products is enticing, and this is being advanced as a way to improve availability and widespread implementation of the treatment. Options being explored include mass-produced allogeneic cells that evade immune rejection (78) and soluble TCR-like molecules named "immune mobilizing monoclonal TCRs against cancer" (ImmTAC) that can bind to native T cells and to the TAAs they are targeting (84, 85).

Terminal differentiation or exhaustion is an important factor influencing both ex vivo production and in vivo persistence of T cells. It often arises when cells are exposed to their cognate antigen over an extended duration, as in the setting of high tumor burden or lengthy production. To this end, researchers are looking into different T-cell subtypes as substrates of transgenic T-cell production, and initial signals hint at memory stem-cell or central memory phenotypes as potential ways to enhance persistence of the transgenic clone without reaching phenotypic exhaustion of a large proportion of the cells (86).

Initial clinical trials, especially in cancer therapies with low therapeutic index, have usually been designed with small subcohorts, reflecting the need for careful examination of the balance between efficacy and toxicity. However, the pharmacokinetics of cellular therapies are very different from those of other drugs, as they proliferate in vivo and the initial dose does not always reflect the toxicity that manifests after cell expansion. Dose–response curves are often nonlinear, necessitating a different early-phase trial design that can assess toxicity more closely, also considering other factors such as lymphodepletion, patient and disease status, and product dosing. This is especially true since lymphodepletion regimens vary greatly between trials, and while higher doses of chemotherapy may enable more robust expansion and persistence of TCR T cells (87), organ toxicity, and bone marrow suppression that ensue may outweigh benefits. For instance, prolonged cytopenia in a patient resulted in an amendment to a study to lower the fludarabine dose (43), and a death was attributed to a plastic marrow following lymphodepletion in another trial (39). Furthermore, when direct local or regional administration of cells is employed to improve trafficking of infused cells to the tumor, special interventional expertise is needed. Since cellular therapy has novel toxicities, namely cytokine release syndrome, neurotoxicity, and other immune manifestations, only specialized centers with experience in immune toxicity management can participate in clinical trials, especially as there is no way of predicting which patients will have these adverse effects. To address these issues, the FDA is developing a separate regulatory process for cellular therapy than the usual approval pathway for pharmacologic agents.

Great strides are being made in adoptive cell therapy, with commercial CAR T-cell therapies now available, and many more in various stages of preclinical and clinical development. TCR T cells are a promising approach to immune-mediated cancer treatment, as well as to immune modulation and therapy for nonmalignant diseases such as infections and autoimmune diseases. However, use of TCR T cells necessitates fine-tuning of the delicate balance between affinity and specificity to the cognate antigen and the ability of the T cells to expand, persist, and mount an immune response.

Novel strategies may enhance the usability of TCR T cells, and adjuncts such as small molecules, immune modulators, and epigenetic modulators may further improve efficacy. Further exploration of cellular therapy platforms, including off-the-shelf cellular therapies, may increase availability and accessibility, and with further improvements in efficacy may significantly widen the spectrum of diseases treatable with this promising therapeutic modality.

C.L. Haymaker reports personal fees from Briacell Therapeutics; other support from Mesothelioma Applied Research Foundation; and personal fees from Nanobiotix outside the submitted work. D.S. Hong reports grants from AbbVie, Adaptimmune, Aldi-Norte, Amgen, AstraZeneca, Bayer, Bristol Myer Squibb, Daiichi Sankyo, Deciphera, Eisai, Erasca, Fate Therapeutics, Genentech, Genmab, Infinity, Kite, Kyowa, Lilly, LOXO, Merck, MedImmune, Mirati, Mologen, Navier, National Cancer Institute-Cancer Therapy Evaluation Program (NCI-CTEP), Novartis, Numab, Pfizer, Pyramid Bio, SeraGen, Takeda, Turning Point Therapeutics, Vernstam, and VM Oncology during the conduct of the study; other support from Bayer, Genmab, American Association for Cancer Research (AACR), Society for Immunotherapy of Cancer (SITC), and Telperian and personal fees from Adaptimmune, Alpha Insights, Acuta, Alkermes, Amgen, Aumbiosciences, Antheneum, Axiom, Barclays, Bayer, Baxter, Boxer Capital, BridgeBio, CDR-Life AG, COR2ed, COG, ECOR1, Genentech, Gilead, GLG, Group H, Guidepoint, HCW Precision, Immunogen, Infinity, Janssen, Liberium, Medscape, Numab, Oncologia Brasil, Pfizer, Pharma Intelligence, POET Congress, Prime Oncology, Seattle Genetics, ST Cube, Takeda, Tavistock, Trieza Therapeutics, Turning Point, WebMD, and ZioPharma outside the submitted work; and other ownership interests in OncoResponse (founder) and Telperian (advisor). No disclosures were reported by the other authors.

U. Greenbaum is a recipient of a fellowship grant from the American Physicians' Fellowship for Medicine in Israel (APF). Some elements of the graphics are adapted from Servier Medical Art, licensed under a Creative Commons Attribution 3.0 Unported License.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.