T-cell–engaging Therapy for Solid Tumors

The recent development of immunotherapies has entailed a change of paradigm in hematologic and solid tumors. Although checkpoint inhibitors have been granted the major part of recent approvals in many solid tumor types, other immunotherapies are under development. One promising approach is the redirection and recruitment of T cells against tumors through the concurrent engagement of the tumoral target cell antigen and a T-cell surface molecule, leading to activation of polyclonal cytotoxic T cells and subsequent tumor lysis. This novel concept of T- cell engagers (TCE) include genetically engineered adoptive cell therapies (GE-ACT) and bispecific antibodies (BiAb; Fig. 1). GE-ACTs involve the administration of a patient's own immune cells that have been previously adapted in vitro with modified T-cell receptors (TCR) or chimeric antigen receptors (CAR). BiAbs are recombinant proteins composed of two linked antibody binding regions capable of recognizing two antigens simultaneously (1), one on the surface of a tumor cell and the other antigen, which is a shared component of the TCR, acting as a bridging agent for tumoral and T cells and engaging the TCR to trigger a redirected T-cell attack on the targeted tumor cell.

The first TCE was a fragment crystallizable (Fc) containing BiAb, described by Bevan and colleagues in the 1980s (2). Since then, TCEs have shown excellent results in some hematologic cancers and, to date, there are three CD19-targeting TCEs that have been approved, two CAR T cells and one BiAb, by the FDA and European Medicines Agency (3–5). However, solid tumors have different molecular, physical, and environmental barriers implicated in less efficient response to these agents, especially the lack of genuinely specific surface tumor antigens (6, 7).

The activity of TCEs depends mainly on the different strategies of TCR activation. The natural TCR/cluster of differentiation 3 (CD3) is an octameric protein complex, formed by hetero- and homodimers at a fixed stoichiometry: CD3γ/CD3ϵ, CD3δ/CDϵ, CD3ζ/CD3ζ, and CD3α/CD3β, the last one is responsible of recognizing polypeptide fragments presented by the MHC, whereas the rest of the units promote a cascade activation through the immune receptor, tyrosine-based activation motifs, that results in release of cytotoxic granzyme and perforin, as well as different cytokines and chemokines, leading to additional T-cell recruitment and proliferation (8, 9).

A majority of BiAbs simultaneously link the tumor cell antigen and the CD3ϵ unit of the TCR by antibody binding regions, therefore, bypassing MHC restriction and facilitating cytotoxicity. BiAbs can be divided into two subtypes; small fragment–based forms with two or more linked antigen-binding moieties with no Fc-domain, and IgG-like BiAbs. Most of the IgG-like BiAbs contain an engineered Fc domain designed to maintain neonatal Fc receptor binding to protect the molecule from catabolism and improve its half-life and stability, while avoiding FcγR and C1q linking to prevent severe infusion reactions from occurring (10–12).

Genetically engineered TCRs recognize tumor-associated antigens (TAA) as the natural biological pathway, that is, through CD3α/CD3β activation, and, therefore, require MHC presentation. Consequently, both intracellular and membrane surface antigens can be recognized, which allows for a broader range of targets (13). These GE-ACTs also encompass the emerging novel field of TCRs recognizing truly specific tumor antigens, namely personalized mutated neoantigens displayed on MHC type I molecules, which currently involves cloning of those TCRs and introducing them into autologous T cells (14).

CAR T cells are MHC independent; their chimeric structure contains a single-chain variable fragment (scFv) as antigen-binding extracellular domain with a hinge region, a transmembrane domain, and an intracellular CD3ζ chain responsible for T-cell activation. Because antigen recognition is based on scFv, just like BiAbs, binding to intact surface antigens occurs. Different generations of CARs have appeared in time; the latter include not only activation domains, but also costimulatory molecules and cytokine expression (7, 13, 15). Thus, BiAbs' and CAR T cells' CD3 activation depends on antibody binding region–mediated antigen recognition, whereas TCRs require MHC presentation of the tumor antigen to become activated (Table 1).

Technical challenges with BiAbs are mainly related to their structure. Fragment-based BiAbs lacking an Fc domain have a short half-life because of faster renal elimination, and in some cases also have drug stability and aggregation issues (6). Their small size may improve tumor penetration if sufficient exposure is achieved depending on dosing approach, and the lack of an Fc region limits bystander immune system activation via antibody-dependent cell-mediated cytotoxicity, antibody-dependent cell-mediated phagocytosis, and complement-dependent cytotoxicity. On the contrary, IgG-like BiAbs are more stable molecules, and the introduction of silenced Fc domains allowed better tolerability (11).

TCR T-cell activity is restricted to the mandatory presence of their specific human leukocyte antigen (HLA) match, that is, subtype *0201 is limited to only 27%–47% of patients depending upon ethnicity (14). Antigen specificity represents another selection problem; TCRs targeting widely shared antigens lead to unfavorable safety profiles, but targeting cancer-specific antigens, such as New York esophageal squamous cell carcinoma 1 (NY-ESO-1), a cancer-testis antigen, might confine this treatment to a small minority of patients with cancer with the double antigen–HLA match. In this context, there are certain tumor types that are more suitable for TCR T-cell therapy as they have frequent expression of specific antigens, such as MAGE-A1–4 in 70% melanoma, NY-ESO in 46% breast cancer (16), 52% in melanomas (17), and 80%–100% in synovial sarcomas/myxoid and round cell liposarcomas (18), or SSX in 23% metastatic prostate cancer (19).

On the other hand, both CAR T cells and BiAbs just act on membrane surface molecules, which represent fewer than 10% of cellular proteins, therefore, limiting their potential targets (14). CAR T cells have high specificity to the target and, in addition, can recognize it without MHC restriction, so high target antigen levels are not required for developing optimal activity. Indeed, this might translate into significant dose variations between different GE-ACTs; while the standard CD19-specific CAR T cells recommended dose is below 10⁸ infused cells (3, 4), the NY-ESO TCR needed 10⁹ cells to achieve tumoral responses (20). TCR cells may penetrate tumors better than CAR T cells, which often stay on the outer layer of solid tumors rather than trespassing it. One potential explanation for this is that membrane surface antigens for CARs can be present in high concentrations and, therefore, restrict T-cell capacity to penetrate the tumor. However, TCRs can miss their antigens, which are often present in fewer than 50 copies per cell and will penetrate better ensuring further distribution (14).

TCEs may lead to a variety of different toxicities. Lymphodepletion chemotherapy with fludarabine and cyclophosphamide, administered before T cells infusion, subsequently causes pancytopenia and potentially febrile neutropenia with reports of some fatal cases (21).

Cytokine release syndrome (CRS) is a TCE-related adverse event (AE), most times antigen dependent, as seen for hematologic malignancies, depicted by massive release of proinflammatory cytokines (IL6, TNFα, IFNγ, and IL1) and reactive increasing of inflammatory markers (ferritin and C-reactive protein), which resembles a sepsis-like picture. Patients develop a heterogeneous presentation that may include fever, myalgia, tachycardia, hypotension, respiratory insufficiency, and can result in death. Circulating cytokines and chemokines also lead to monocyte and macrophage activation and are able to cross the blood–brain barrier. GE-ACTs have also been associated with immune effector cell–associated neurotoxicity syndrome (ICANS) defined by decreased consciousness, confusion, seizures, and brain edema, not always presented along with systemic CRS (9, 22, 23). Severe toxicity is also described with BiAbs; blinatumomab, a CD3-CD19 bispecific TCE, is associated with neurologic AEs clinically indistinguishable from ICANS, and with rapid and sometimes fatal development of CRS (24). With cibisatamab, the CEA-CD3 BiAb for colorectal cancer, up to 24% of patients had severe infusion-related reactions (IRR) consistent with CRS-like symptoms (25).

In general, the severity of the CRS might be associated to tumor burden, number of cells infused, or with the starting dose of BiAbs (26), and, in fact, step-up dosing approaches have been used to ameliorate this toxicity. The treatment to reverse these symptoms includes antidotes to neutralize IL6 activity, either by blocking the receptor, such as by tocilizumab, or directly binding IL6 itself, such as by siltuximab. Tocilizumab has been approved by the FDA for the treatment of CRS, whereas siltuximab has not been sufficiently studied and its use remains investigational. In addition, these treatments are normally used in combination with corticosteroids and sometimes with other immunosuppressants, such as anti-TNFα (infliximab) or anti-IL1 (anakinra). Other drugs have been proposed for CRS management; the JAK/STAT inhibitor, ruxolitinib, prevented inflammatory cytokine release without affecting efficacy in xenograft models, while the recent approval of anti-IFNγ (emapalumab) for the treatment of primary hemophagocytic lymphohistiocytosis also opens the intriguing possibility to test this drug for CRS (27, 28). Isolated severe ICANS is usually managed with steroids alone rather than tocilizumab, as it has shown no significant activity in this setting, probably due to the lack of blood–brain barrier penetrance (9). Preclinical models have found that blocking CRS does not reduce TCE function, and so tocilizumab is being studied for prophylaxis of CRS with CAR T-cell therapies. A proper management of TCE administration would improve safety without limitation on efficacy, as it was found with the ADP-A2M4 SPEAR TCR T cells for MAGE-A4, where just 5.3% of patients reported severe CRS (23, 29).

Dismally, TAAs are also expressed on healthy tissues. When T cells bind nontumoral cells “on-target off-tumor” toxicities may appear. This phenomenon led to severe skin rash, uveitis, and ototoxicity with TCR cells studied in melanoma targeting MART-1 and gp100 (30). Subsequent GE-ACTs had similar issues; the carcinoembryonic antigen cancer (CEA)-directed TCR cells caused severe inflammatory colitis (31), whereas several patients died with MAGE-A3 targeting TCR cells, either by cross-reaction with brain epitopes (32) or the cardiac protein, titin (33). Also, “off-tumor” fatal toxicity has been described with CAR T cells in 1 patient because of respiratory distress related to ERBB2 lung expression (34). Although “on-target off-tumor” toxicities have been more extensively described with TCRs, it is a partially unpredictable TCE class problem. With cibisatamab monotherapy, 46% of patients had diarrhea, 7% of which was reported to be grade ≥3 (25), and in a study with an autologous CAR-CLDN18.2 T cell, 2 patients had gastric injuries due to claudin gastric expression (35).

TCEs clinical development is becoming a challenging journey for solid tumors (Fig. 2). In 2006, MART-1 TCR T cells for patients with melanoma demonstrated good tolerance and two bona fide responses in patients that had sustained engineered cells in blood 1 year after their infusion, so expectations for GE-ACTs rose (36). Some years later, catumaxomab was approved in the European Union (EU) for the treatment of malignant ascites. This CD3 anti-EpCAM trifunctional antibody demonstrated in a phase II/III trial to reduce the need for paracentesis. However, the drug was withdrawn in 2017 for commercial reasons (37). In 2010, the first hematologic successes with CD19-directed CAR T cells arrived for patients with acute lymphocytic leukemia (ALL) and non-Hodgkin lymphomas (NHL), and although commercial approval was not available until 2018, it triggered the research for the solid tumor scenario (38, 39). Unfortunately, in this field a clinical dark side was observed soon and different trials had to stop recruitment due to life-threatening or fatal toxicities, as discussed before, and lack of enough antitumor activity. Thus, redirecting clinical research to better and more specific targets led to a second wave of TCEs with better outcomes.

Three complete responses were seen with the next generation of CAR T cells directed to the GD2 ganglioside in neuroblastoma (40); similarly, a phase I/II trial for HER2-expressing sarcomas found clinical benefit with HER2-CAR T cells along with a good tolerability profile (41). Two other CAR T cells have demonstrated encouraging results as well; intrapleural administered CAR T cells directed to mesothelin for patients with mesothelioma reported a 41% objective response rate. Interestingly, CAR T cells were also detected in peripheral blood of these patients (42). The second ones were autologous CAR-CLDN18.2 T cells for CLDN 18.2–positive gastric or pancreatic tumors that showed a 36% response rate (35).

In 2017, the first genetically modified TCR NY-ESO T cells showed a significant 50% response rate in patients with synovial sarcoma, with manageable CRS (43). NY-ESO TCR cells have also been explored in patients with liposarcoma with encouraging 50% responses (4/8 patients; 20). In 2019, similar response rates were reported with the MAGE-A4 TCR cells in patients with advanced synovial sarcoma (15).

More recently, two other TCRs have reported promising results in early-phase trials. One targets the E7 oncoprotein of human papillomavirus-positive–associated epithelial cancers. In this study, 6 of 12 heavily pretreated patients achieved durable partial responses across the three dose levels studied (44). The other was the ADP-A2M4 SPEAR TCR T cells for MAGE-A4–positive tumors; among the 38 patients treated, an approximate 90% disease control rate and 44% confirmed partial responses were reported in a broad range of tumors including synovial sarcoma, lung, and head and neck cancers. Most of the toxicities were related to lymphodepletion, but 2 patients developed fatal consequences (29).

Regarding BiAbs, in 2014, blinatumomab became the first BiAb to be approved by the FDA that is currently available in the market for patients with ALL (24). At the same time, the results of the cibisatamab phase I studies in monotherapy or in combination with atezolizumab in patients with microsatellite stable colorectal cancer showed 6% and 18% response rates, respectively (25). Two half-life extended BiAbs are currently under development by Amgen; the AMG 199 for MUC17-positive gastric and gastroesophageal junction tumors and the AMG 757 for DLL3-positive small-cell lung cancer (45, 46). Moreover, there is variety of clinical trials running in different settings (Fig. 3).

Some major differences as compared with hematologic malignancies have determined the lesser efficacy of TCEs in solid tumors so far; universal tumor marker expression, such as CD19, is rare, treatments need to get into the tumor and not just act on the bloodstream, and have to work on a complex and inhibitory tumor microenvironment (15). These, and other factors, lead to TCE resistance in solid tumors.

First, antidrug antibodies (ADA) may appear after TCE infusion, mediated by CD4 T lymphocytes, memory B cells, and plasma cells (47, 48). The structure of the drug itself, presence of foreign sequences, dose, route of administration, and patients' immune system have been related to ADA generation after treatment with BiAbs, leading to IRR, potential decrease of treatment efficacy, and affecting clearance with subsequent changes on pharmacokinetics and pharmacodynamics properties (47, 49). One interesting possibility to decrease ADA formation is to administer anti-CD20 antibodies, such as obinutuzumab, prior to the BiAb administration to abrogate B-cell activity. This strategy is currently being explored in studies with TCEs such as cibisatamab (NTC03866239; Fig. 4, 1).

Second, TCEs need to be distributed appropriately. Some trials are considering their local administration, as for brain (NCT04045847) or mesothelioma tumors (42), while other strategies include T-cell trafficking improvements, for example, CAR T cells expressing chemokine receptors for CCL2 or VEFGR2 to allow better distribution (refs. 7, 50; Fig. 4, 2).

In the third place, when the drug reaches the tumor, heterogeneity is present in all phenotypic features; cellular morphology, metabolism, proliferative expression, immunogenic properties, and gene expression, including cell surface markers (51), and, precisely, the lack of tumor cell surface antigens that are truly specific and can be targeted with these engineered T cells is the major limitation for GE-ACT redirection approaches for solid tumors (52). This is different in B-cell or plasma-cell malignancies, where specific lineage antigens, like CD19, CD20, or BCMA, can be targeted and depletion of normal B or plasma cells during treatment can be tolerated. Actually, those normal cells will be regenerated from target-negative precursors once treatment is completed and the drug is cleared. Intratumoral heterogeneous antigen density and variable antigen distribution among solid tumor tissues make it difficult to identify appropriate targets for this precision therapy. Combinatorial antigen recognition approaches, such as bi- or trispecific CAR T cells, or the other way around, promoting epitope spreading or increasing antigen expression, might overcome tumor heterogeneity (refs. 15, 53; Fig. 4, 3).

The inhibitory tumor microenvironment is another main factor for TCE failure. Thus, fourth, the TCE must resist the tumor microenvironment, composed not only of immunosuppressive cells, such as regulatory T cells, tumor-associated macrophages, and myeloid-derived suppressor cells, but also of hypoxia-related factors, necrosis, nutrient deficiency, and immunosuppressive molecular factors, like PD-L1, IL10, or TGFβ (15, 54). CAR T cells directed to fibroblast activation proteins, or CAR T cells expressing heparanase, could make a more favorable environment and improve tumor penetrance, respectively (55, 56). For BiAbs, activity will depend on the density and class of T cells surrounding the tumor, in theory, but they are quite unpredictable and highly variable. Preexisting T cells with an exhausted phenotype, or regulatory T cells infiltrating the tumor may hamper the action of BiAbs (ref. 50; Fig. 4, 4).

Fifth, fully competent T cells need to remain active and alive. Arginine depletion is associated with reduced proliferation, survival, and T-cell dysfunction. Restoring arginine levels by CAR gene constructs could improve these functionalities (54). T-cell exhaustion is also related to inhibitory signal receptor upregulation, that is, PD-1, TIM-3, or LAG-3, and to PD-L1 expression in tumor cells (15). Combining TCEs with checkpoint inhibitors, or T cell PD-1 gene editing, may reestablish functionalities (57). Finally, CAR T-cell dysfunction has been associated with tonic signaling, a noncoordinated and sustained activation of T cells ending in apoptosis, exhaustion, and impaired antitumor effects, related mainly to the structure and kind of target expressed by the CAR (58). Finally, T-cell survival may improve with shortening ex vivo expansion protocols, or enriching the infusion product in T cells with stem cell properties (refs. 59, 60; Fig. 4, 5).

The final point is to overcome the mechanisms of evasion. Loss of antigen expression is an adaptive resistance mechanism to CAR T cells commonly observed in hematologic malignancies, and in glioblastoma multiforme (53, 61, 62). Whether the absence of TAA expression on certain tumoral cells is present before treatment and overgrow due to selective advantage under TCE exposure or if nonexpressing TAA cells are the result of later mutations, is unclear. TAA downregulation has been described with BiAb as well, and resulted in decreased drug activity (10). For TCR T cells, the main mechanism of evasion is loss or downregulation of MHC, as a result of different processes, such as LOH, transcriptional repression of individual HLA genes, epigenetic silencing of key genes involved in antigen processing or presentation, and the IFN response pathway (refs. 63, 64; Fig. 4, 6).

Many different strategies are being researched to further improve TCE efficacy. Checkpoint inhibition in combination with TCEs has shown to be an interesting approach; CAR T-cell and PD-L1 inhibitors have been studied successfully in different clinical trials (40, 65). Alternatively, preclinical studies have shown that PD-1 inhibition can also be achieved by gene deletion with CRISPR-Cas9 technology (66), incorporating switch receptor constructs comprising PD-1, adding CD28 domains that are able to transform a negative signal into a stimulatory one (67), or through expression of PD-1 dominant negative receptors lacking a signaling domain (68). Moreover, combinations with other immune checkpoints could have an interesting role, including CTLA-4, LAG-3, TIM-3, and IDO, or costimulatory receptors, such as 4-1BB, OX40, CD40, and CD27 (69).

The new generations of armored or TRUCK-CAR T cells incorporate cytokine modulation approaches, such as IL12, IL15, IL18, that could overcome the immunosuppressive tumor microenvironment (50, 70, 71). Personalized TCR therapy based on patients' own tumor neoantigens is another highly complex approach in research. The process includes neoantigen identification, TCR cloning, viral vector, or other techniques, assembly, and validation. However, the process should be repeated for different neoantigens, as single mutations are not expressed in all tumor cells, which is a significant hinderance to this approach (14).

New technologies are being applied to combine GE-ACTs and BiAbs, which are the so-called BiAb-armed activated T cells (BAT). These BATs have the ability to redirect bystander T cells to tumor cells, and have potent antitumor activity in vivo thus, obviating the need for repeated BiAbs infusion dosing (72). Novel BiAb engagers, other than CD3, are also in research; the γ-δ TCEs are designed to bind to an homogeneous effector T-cell subpopulation with low inhibitory checkpoint molecule expression and preference to trigger killing of cancer cells (73).

Meanwhile, other strategies are trying to improve TCE safety. Suicide genes have the ability to control apoptosis of engineered cells. Herpes simplex virus thymidine kinase (HSV-TK) is being studied in CAR T cells and has shown preliminarily safe elimination under ganciclovir exposure (74). However, it remains unknown whether HSV-TK has potential immunogenicity that could compromise T-cell survival, and whether the time for inducing T-cell apoptosis is adequate for patients' toxicity control (75, 76). Another less immunogenic approach is the inducible safety switch caspase 9 (iCasp9) that contains a modified human caspase 9 fused to the FK506 binding protein. Induction of iCasp9 depends on the administration of AP1903, a chemical inducer of dimerization which activates caspase molecules resulting in apoptosis (77, 78).

Inhibitory CARs (iCAR) can be incorporated in T cells to control healthy tissue toxicity. iCAR consists of an scFv specific to antigens expressed only in normal cells, with potent acute inhibitory signaling to restrict T-cell activation despite concurrent engagement of the activating receptor (79).

Combination of CARs and BiAbs can also improve tolerability. In this case, as an example, the CAR is designed to bind FITC, which is not a cell target, but a universal binder instead, while the BiAbs would conjugate FITC and the target. Therefore, in the absence of a BiAb, CAR T cells will not be activated, and can be modulated by the dose of BiAbs administered (79, 80).

Finally, combinatorial target–antigen recognition is another approach to improve tolerability. Activation of the T cell would, in this context, depend on coactivation of two different CARs; one with CD3 transduction and the other one with the costimulatory receptors CD28 or 4-1BB (81). Regarding BiAbs, a novel generation of BiAbs directed to non-CD3 TCRs in addition to tumor-engaging antigens is currently being explored, as those targeting tumoral PD-L1 (NCT03917381) or FAP (EudraCT 2017-000292-83) to 4-1BB, LAG-3 (NCT04140500), or TIM-3 (NCT03752177).

As off-the-shelf recombinant proteins, BiAbs are manufactured using standard and scalable processes of fermentation, chromatographic purification, and soluble protein formulation. Currently, a significant proportion of BiAbs molecular formats carry an Fc region engineered to minimize FcγR binding and can contain up to four protein chains. Manufacturing of BiAbs often requires the use of different expression and purification technologies to favor the generation of a bispecific molecule with the correct pairing of different heavy and light chains in good production yields. Such technologies, for example, include: (i) engineered Fc regions to favor heterodimerization, for example, the so-called “knob-into-hole” approach; (ii) crossing of variable and constant region domains to minimize light chain mispairing, known as the “CrossMab” approach; (iii) use of a common light chain between the two Fab regions of different specificity; and (iv) use of a single chain construct, for example, a single chain Fv, an autonomous V_H domain, or an alternative protein scaffold, for at least one of the target binding domains (1, 82).

GE-ACT production must balance adequate timing, efficiency, safety, and costs. Viral vector–based protocols are most frequently used for T-cell transduction, and both retroviral and lentiviral vectors are able to deliver enough payloads to integrate into host genomes and consistently express the construct. Lentiviral vectors are often preferred as they may confer less risk of insertional oncogenesis, and, therefore, decrease the risk of cell transformation (7). However, viral vectors are associated with high production costs and complicated quality controls. Nonviral integrative vectors, such as transposons, integrate stably into the target cell genome, enabling expression of therapeutic genes. Currently, the most promising transposons for gene therapy integration are derived from Sleeping Beauty or PiggyBac systems (83). Electroporation also allows mRNA cell incorporation by electrical disruption of the membrane and does not need genomic integration, so it could be theoretically safer, avoiding transgene integration. However, it is associated with transient expression of the construct and may require repeated dosing (84). A new approach is using electroporation together with CRISPR/Cas9 for replacement of the endogenous TCR by the recombinant construct, which is being developed successfully (85).

Obtaining the high number of cells required for achieving antitumor effect may suppose a hurdle, especially with TCR T cells. Moreover, during the manufacturing procedures, to generate large number of cells, microbial contamination is the main risk. Using automated closed large-scale modular systems and semi-automated devices may decrease contamination (86). Contamination of the transfected T cells also occurs when the modified TCR is incorporated to the lymphocyte, as the cell has its own natural TCRs that may interfere with activity and decrease potency. The use of siRNAs to block endogenous TCR expression has been tested successfully (87). Finally, manufacturing timelines for ACT need to be refined; currently, the time for the patient to be infused is taking approximately 2 months considering prescreening and culture processes, which is certainly inefficient (88). The use of “universal CAR” would overcome many of these limitations as with the “SUPRA CAR system,” which is composed of a universal receptor expressed on T cells and a tumor-targeting scFv adaptor molecule, easily interchangeable to target different antigens (89).

More and better TCEs are currently under research and will presumably have an important role in the future perspective of solid tumors therapeutics. GE-ACTs and BiAbs have similarities among their mechanism of action and challenging tolerability, where there is still room for improvement, as well as already some encouraging bona fide tumor responses in refractory cancers. Some interesting clinical trial results are rising, but because of the complexity of solid tumors, many resistance mechanisms, such as the inhibitory tumor microenvironment, tumor antigen heterogeneity, or evasion mechanisms, still have to be dealt with. Personalized neoantigen-directed TCRs, bi- or trispecific CAR T cells, BATs, or combinations with checkpoint inhibitors are already on the horizon as new promising therapeutics. From the safety perspective, the introduction of suicide genes, iCARs, combinations of CARs and BiAbs, or combinatorial target–antigen recognition are some new approaches under research that will probably change the future landscape for patients with solid tumors.

M. de Miguel reports other from MSD (research funding), PharmaMar (research funding), Roche (research funding), Novartis (research funding), AbbVie (research funding), Array (research funding), Eisai (research funding), and Sanofi (research funding), and personal fees from MSD, Janssen, and Roche outside the submitted work. P. Umana reports personal fees from Roche outside the submitted work. V. Moreno reports personal fees from Bristol-Myers Squibb, Bayer, Janssen, and Pieris outside the submitted work. E. Calvo reports grants and personal fees from Astellas, Novartis, Nanobiotix, Pfizer, Janssen-Cilag, PsiOxus Therapeutics, Merck, Bristol-Myers Squibb, Seattle Genetics, Boehringer Ingelheim, AstraZeneca, Roche/Genentech, Servier, Celgene, AbbVie, Amcure, Alkermes, PharmaMar, and BeiGene, personal fees from GLG, Medscape, Gilead, Pierre Fabre, Cerulean Pharma, EUSA, Gehrmann Consulting, Guidepoint, and OncoDNA, and grants from ACEO, Adaptimmune, AMGEN, CytomX, GlaxoSmithKline, H3, Incyte, Kura, Lilly, Nektar, Loxo, MacroGenics, Menarini, Merus, Principia, PUMA, Sanofi, Taiho, Tesaro, Transgene, Takeda, Inovio, MSD, Mersana Therapeutics, Daiichi Sankyo, ORCA, Boston Therapeutics, Dynavax Technologies, Debiopharm, Regeneron, Millenium, Synthon, Spectrum, and Rigontec outside the submitted work, as well as is a scientific board member at PsiOxus Therapeutics, HM Hospitals Group, and START Program of Early Phase Clinical Drug Development in Oncology (employee, medical oncologist; director, clinical research), founder and president, non-for-profit Foundation INTHEOS (Investigational Therapeutics in Oncological Sciences), and codirector, Methods in Clinical Cancer Research (MCCR) Workshop, Zeist, Netherlands (Joint ECCO-AACR-EORTC-ESMO Workshop). No disclosures were reported by the other author.

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.