The antitumor activity of our immune system is a highly sophisticated process with several regulatory and negative feedback pathways. When malignant cells are identified and attacked by macrophages and natural killer (NK) cells -components of the innate immunity-, aberrant proteins derived from the cumulative occurrence of mutations are released and phagocytosed by dendritic and other antigen-presenting cells (APC) (1). In the peripheral lymph nodes, these tumor-associated antigens (TAA) are exposed by APC through major histocompatibility complex type I (MHC-I) molecules to the T cell receptor (TCR) of naïve CD8+ T cells, leading to their activation. For this ‘immune synapsis’ to be successful, other co-stimulating receptors on the T cell membrane (such as B7) should be activated (2). At the same time, the interaction between MHC type II (MHC-II) molecules and the TCR of CD4+ T helper lymphocytes leads to the activation of B cells and subsequent production of antitumor antibodies (3), and unleashes additional mechanisms that elicit CD8+ T cells function and differentiation, including dendritic cell licensing and cytokine production (4). Once CD8+ T cells are activated, they travel to the tumor site and recognize TAA presented by MHC-II molecules on the surface of malignant cells, unleashing the effector phase of adaptive immunity, and ultimately leading to tumor cell death (4). The quantity and phenotype of these tumor-infiltrating lymphocytes (TILs) have been widely associated with the biological behavior, prognosis, and response to anti-cancer therapies in virtually all subtypes of solid cancers (5–7). The success of the effector phase is compromised by the inhibition of T-cell response by immunosuppressive cells from the tumor microenvironment (TME), including myeloid-derived suppressor cells (MDSCs), tumor-associated macrophages (TAMs), and regulatory T lymphocytes (T-regs) (8). Tumor cells are able to modulate their function and differentiation through the activation of NF-κB and STAT3 signaling pathways, inducing the release of immunosuppressive cytokines (IL-6, IL-10, TGF-β) that inhibit TILs antitumor activity (9).

The antitumor response is controlled by negative feedback mechanisms performed by molecules known as ‘immune checkpoints’, both in the priming phase -including CTLA-4, LAG-3 and TIM-3- and in the effector phase -mainly programmed cell death protein 1 (PD-1), activated by ligands (PD-L1) expressed both by cells from the tumor and TME- (10). The mechanisms of the antitumor response and potential immune biomarkers are shown in Figure 1 .

The ability to avoid the immune system is a hallmark of malignant cells (11). Some of the most relevant mechanisms of immune evasion in solid tumors are the upregulation of immune checkpoints (12) -which sets the rationale for the use of immune checkpoint inhibitors (ICI)-, the loss of MHC-I or other molecules with a key role in antigen presentation (13), the production of cytokines (IL-6, IL-10, TGF-β) that lead to an immunosuppressive TME (14), and the activation of oncogenic routes that promote T-regs infiltration -such as the indoleamine 2,3-dioxygenase (IDO) pathway (15)- or inhibit CD8+ T-cell trafficking to the tumor site -such as the Wnt/β-catenin pathway (16)-. Recent data suggest that remote transference of biomolecular cargoes from malignant to healthy cells, mediated by exosomes, may also play a crucial role in misleading the mechanisms of antigenic recognition, thus contributing to immune tumor evasion (17).

Since the FDA approval of anti-CTLA4 and anti-PD1 therapy for advanced melanoma in 2011 and 2015, respectively, ICI alone or in combination with other therapies have transformed the therapeutic landscape of nearly all solid tumors (18). However, there are several other emerging strategies to enhance the immune antitumor response, some of them with promising results in ICI-refractory tumors, which will surely increase the relevance of immunotherapy in cancer treatment throughout the following years.

Adoptive cell therapy (ACT) encompasses several techniques that use the T lymphocytes themselves, after a process of artificial modification or genetic engineering, to improve their antitumor activity (19). This implies the extraction of autologous T lymphocytes from the patient, and their manipulation and amplification in vitro. Meanwhile, the patient undergoes treatment with lymphodepleting chemotherapy (CT) -usually fludarabine plus cyclophosphamide- to annihilate ineffective and immune-suppressing lymphocytes. After this process, the improved cell product is reinfused into the patient.

The difference between different ACT modalities lies in the characteristics of T cell modification in vitro: CAR-T and TCR gene-modified T cells are genetically engineered to incorporate modified membrane receptors with a high affinity for selected tumor antigens. However, in contrast to hematologic malignancies, solid tumors are composed of polyclonal cell populations with huge antigenic heterogeneity, which hampers the efficacy of antigen-targeted therapies. TIL therapy is based on the activation and expansion of infiltrating T cells extracted from the tumor itself, which are intrinsically reactive against tumor antigens, setting an interesting rationale for its use against the changing and heterogeneous cell population of solid cancers.

CAR-T cells are genetically modified lymphocytes that incorporate a chimeric antigen receptor (CAR) composed by three parts: an extracellular domain with a single-chain fragment variable (scFv) that allows antigen recognition, a transmembrane domain -linked to the extracellular part through a spacer-, and an intracellular domain. This includes a CD3 complex -which activates the downstream signaling pathways- and several costimulatory domains (usually CD28 and/or 4-1BB) that intensify the cytoplasmatic activity of T cells unchained by antigenic recognition (20). In recent years, innovations in the structure and manufacturing of CAR-T cells have led to significant improvements in their clinical efficacy, especially with the development of fourth-generation CAR-T cells (21). Fifth generation CARs equipped with three costimulatory domains and able to secrete anti-PDL1 scFv blockade molecules, targeted against B cell maturation antigen (BCMA) have shown heightened antitumor efficacy and decrease of T cell exhaustion in patients with multiple myeloma (22).

Nanobodies or single domain antibodies (VHH) have recently been exploited as an alternative to scFvs for antigen-targeting domains on T cell surface, based on numerous advantages including their small size, high affinity, specificity and stability (23). VHH-based CD19-redirected CAR-T cells have shown similar expansion rate, cytotoxicity and anti-tumor reactions when compared with their scFv-based counterparts (24).

Due to their molecular structure, CAR-T cells only recognize extracellular antigens, and are particularly efficient when their scFv has a high affinity for the targeted protein -and it is homogeneously expressed by tumor cells-. This explains the efficacy of CAR-T cells in patients with leukemia and lymphoma, which comprise a clonal population of cells that uniformly express certain antigens -such as CD19- on their membrane (25). Since 2018, tisagenlecleucel and axicabtagene-citoleucel have EMA approval for the treatment of B-cell LLA and refractory non-Hodgkin lymphoma, and in 2020, brexucabtagene-autoleucel was approved by the EMA for mantle lymphoma, achieving complete response rates of over 50% in heavily pre-treated patients (26).

The outstanding results of CAR-T cells in hematologic malignancies have led to their investigation in solid tumors, mainly using overexpressed epithelial antigens as targets. The epithelial growth factor receptor (EGFR), HER2, carcinoembryonic antigen (CEA), mesothelin and soluble antigen GD2 have been frequent targets of CAR-T therapies, although many other antigens have been the object of preclinical studies (27).

Nearly 500 clinical trials evaluating CAR-T cells in solid tumors have been registered, most of them in Asian population, and many still ongoing (28). Most completed studies are phase I/II trials that have reported modest results, with only occasional and generally brief clinical responses. Clinical research on CAR-T cells has mainly focused on glioblastoma (GBM), sarcoma, neuroblastoma, and gastrointestinal cancer.

In 2016, Brown et al. (29) reported the case of a GBM patient with an 8-month complete response (CR) after IL13-targeted CAR-T therapy, although further research has failed to confirm these results (30). Her2 may be another interesting target for CAR-T cells in GBM; in a clinical trial with 17 patients, 1 partial response (PR) (lasting 9 months) and 7 cases of stable disease (SD) (ranging from 2 to 29 months) were reported (31). EGFR-targeted CAR-T cells have been evaluated in GBM in two clinical trials, with negative results (32, 33).

Her2-targeted CAR-T cells have also been tested in sarcoma patients. In a phase I/II study including 19 patients with Her2+ recurrent or refractory sarcoma of several histological subtypes, 4 SD were observed [3 osteosarcoma and 1 small round cell desmoplastic tumor) (34). In neuroblastoma, at least three clinical trials have evaluated the efficacy of GD2-targeted CAR-T cells -based on the efficacy of anti-GD2 monoclonal antibodies such as dinutuximab- (35), with promising results (3 CR among 19 patients (36), 4 PR among 8 patients (37), and 5 SD among 11 patients (38)].

As for gastrointestinal (GI) cancer, several antigens have been evaluated as potential targets of CAR-T therapy. In a phase I trial including 23 patients with several GI tumors treated with CD133-targeted CAR-T cells, 3 PR (2 pancreatic and 1 hepatocellular carcinoma [HCC]) and 14 SD were observed (39). Zhan et al. (40) evaluated CAR-T therapy targeting Claudin 18.2 (CLDN 18.2) in 11 patients with CLDN 18.2-positive gastric or pancreatic carcinoma, reporting 1 CR, 3 PR and 5 SD. EGFR-CAR-T therapy has mainly been evaluated in biliopancreatic tumors, with promising results. Liu et al. (41) conducted a phase I study including 14 patients with refractory advanced pancreatic carcinoma, reporting 4 PR and 8 SD; in a phase I study by Guo et al. (42) with 19 patients (14 cholangiocarcinoma, 5 gallbladder carcinoma), 1 CR and 10 SD were observed. Glypican-3 (GPC3)-targeted (43) and CEA-targeted (44) CAR-T cells have shown modest activity in HCC and colorectal cancer, respectively.

Mesothelin-targeted intrapleural CAR-T cells have been evaluated -in combination with ICI- in 14 patients with malignant mesothelioma and non-small cell lung cancer (NSCLC), with promising results (2 CR, 5 PR, and 4 SD) (45). In NSCLC, a phase I study showed clinical activity of EGFR-CAR-T cells, with 2 PR and 5 SD among 11 patients (46), and ROR1-directed CAR-T cells showed preliminary positive results in ROR1+ tumors (4 PR among 6 patients) (47). CAR-T therapy has obtained modest results for metastatic castration-resistant prostate cancer (mCRPC) -targeting prostate specific membrane antigen (PSMA)- (48). Evidence for CAR-T therapy in other solid tumors is even scarcer and mainly comes from preclinical studies (49).

The outcomes of the most relevant clinical trials that evaluated CAR-T cells in solid tumors are summarized in Figure 2 .

T cells expressing an engineered T-cell receptor (TCR-T-cells) represent an alternative modality of ACT, with several advantages compared to CAR T therapy (50). By using engineered TCR -instead of CAR-, T cells can recognize not only membrane proteins but also intracellular antigens presented by MHC molecules, covering a wider repertoire of tumor neoantigens and therapeutic targets. Intrinsic features of T cells, such as high antigen sensitivity and the use of physiological signaling pathways, improve TCR-T-cells antitumor functions and reduce the risk of on-target off-tumor (OTOT) toxicity.

However, TCR-T-cells also have some disadvantages compared to CAR-T cells, such as their weaker avidity for target antigens and their limitation to a certain human leukocyte antigen (HLA) haplotype, which restricts the number of patients that potentially benefit from each modified TCR. In addition, as antigen-targeted modified T lymphocytes, all the limitations of CAR-T therapy in solid tumors -antigenic heterogeneity, difficulty of T cell trafficking, and detrimental effects of immunosuppressive TME- are also applicable to TCR-T-cells (51).

In 2006, Morgan et al. (52) published the results of the first trial with TCR-T-cells in solid tumors, targeting MART-1 in 17 patients with metastatic melanoma (with 2 PR). Since then, at least three other phase I/II clinical trials have evaluated MART-1-targeted TCR-T-cells in advanced melanoma -all using retrovirus as vectors-, and have reported variable results, with ORR ranging from 0% to 30% (53–55). In one of these trials (53), protein gp100 was also evaluated as a target in 16 patients, observing 1 sustained CR (>14 months) and 2 PR (ORR: 18.8%). TCR-T-cell-related toxicity was similar in all these studies, with a predominance of skin toxicity (23-94%) and a low incidence of CRS (<15%).

The New York esophageal squamous cell carcinoma (NY-ESO)-1 antigen is a promising target both in melanoma and in some sarcoma immunogenic subtypes -particularly synovial sarcoma, with NY-ESO-1 overexpression in 80% of patients (56)-. In a phase II trial with 38 melanoma and synovial sarcoma patients treated with NY-ESO-1 TCR-T-cells, there was an ORR of 57.9%, including 5 maintained CR and several long-term PR, with no severe toxicity related to TCR-T-cells (57). Similar results were obtained in two other phase I studies, with an incidence of CRS of 10% in one of them (58, 59). Two other phase I/II trials evaluated NY-ESO-1 TCR-T-cells in patients with synovial sarcoma -using lentivirus as a vector instead of retrovirus-, obtaining similar results (ORR 30% to 50%), with an incidence of CRS of 41.7% in one of them (60, 61). NY-ESO-1 was also used as TCR-T-cells target in a small phase I trial including 3 patients with different solid tumors, with negative results (62).

Other cancer-tesis antigens, such as MAGE family proteins, have been evaluated as TCR-T-cell targets in multi-tumor phase I/II trials. A phase I trial evaluated MAGE-A3-targeted TCR-T-cells in 17 patients with solid tumors and observed an ORR of 23.5% (including 1 sustained CR) (63). Morgan et al. (64) treated 9 solid tumors patients (including 7 synovial sarcomas) with MAGE-A3-targeted cells, obtaining an ORR of 56% (including 1 sustained CR), but with severe toxicity (ICANS in 3 patients and 2 treatment-related deaths). In a phase I trial with 38 patients treated with MAGE-A4-TCR-T-cells, 50% of CRS was observed (65), and another trial with MAGE-A3-TCR-T-cells was suspended after 2 treatment-related deaths (66). Cross-reactions against proteins normally expressed by nervous system cells -such as EPS8L2- have been hypothesized as the source of severe neurotoxicity and toxic deaths reported by MAGE-TCR-T-cells trials (67).

MAGE-targeted TCR-T-cells have obtained negative or modest results in NSCLC (68) and esophageal cancer (69). Similarly, significant toxicity has been observed in studies targeting other proteins that are not exclusively expressed by tumor cells. For example, a phase I trial evaluating CEA-targeted TCR-T-cells in patients with colorectal cancer was suspended due to severe colitis in 100% of the patients (70).

The treatment of virus-related tumors is an interesting approach for TCR-T-cells, since targeting specific viral antigens, which are expressed by infected tumor cells but not by normal tissues, should avoid the problem of cross-reactions and OTOT toxicity. In fact, two trials have evaluated TCR-T-cells targeted against human papillomavirus (HPV) carcinogenesis-related proteins (E6 and E7) in HPV+ tumors, without any relevant toxicity and interesting clinical outcomes (ORR of 16.7% and 50%, respectively) (71, 72). Meng et al. evaluated TCR-T-cells against hepatitis B virus (HBV) in 8 patients with HBV+ HCC, observing 1 prolonged PR and only one case of liver toxicity (73). Veatch et al. (74) tested TCR-T-cells against Merkel carcinoma polyomavirus (MCPyV) in 5 patients with refractory Merkel cell carcinoma, reporting 1 PR and no significant adverse events. However, given that virus-related tumors represent a small proportion of advanced solid cancers, other strategies are needed to overcome TCR-T-cells cross-reactions and OTOT toxicity.

In relation to this, an exciting hypothesis is the use of neoantigens -byproduct of tumor somatic mutations, not present in non-malignant tissues- as tumor-restricted and immunogenic targets of TCR-T-cells. In fact, some studies suggest that long-term responses to immune checkpoint inhibitors are mediated by neoantigen-reactive effector T cells (75), providing a rationale for combining TCR-T-cells and ICI. Unfortunately, a phase I trial exploring the effectiveness of personalized neoantigen-targeted TCR-T-cells in 16 patients with advanced solid tumors has obtained disappointing results (51).

Occasional responses to TCR-T-cells have been observed in other tumors. Leidner et al. (76) reported a partial response with KRAS G12D-targeted TCR-T-cells (lasting >6 months) in a patient with advanced pancreatic adenocarcinoma. Kim et al. (77) reported a partial response to TP53-targeted TCR-T-cells in a patient with metastatic breast cancer.

The outcomes of published TCR-T-cells clinical trials for solid tumors have been summarized in Table 1 .

Both CAR-T and TCR-T cells harbor inherent limitations for the treatment of patients with solid tumors, which may explain the significantly worse clinical outcomes than those obtained with CAR-T therapy in hematologic malignancies.

Ex vivo expanded tumor-infiltrating lymphocytes (TILs) from different solid cancers share a composition of oligoclonal effector T cells that are reactive against a heterogeneous repertoire of tumor-associated antigens (94). This establishes the rationale for TILs artificial expansion and activation in vitro (out of the TME detrimental influence) and their subsequent reinfusion -together with stimulating cytokine IL-2- into a more favorable environment, after chemical depletion of immunosuppressive cells.

Not being a specific antigen-targeted therapy and using naturally ‘selected’ tumor-reactive lymphocytes, TIL therapy may theoretically overcome the problems of antigenic heterogeneity, tumor trafficking, and on target off tumor toxicity that limit the effectiveness of CAR-T and TCR-T therapies against advanced solid tumors.

TIL therapy has mainly been evaluated in advanced melanoma. Since the first positive studies conducted by Rosenberg (95–97), several phase I/II trials have obtained clinical responses with expanded TILs -alone or in combination with total body irradiation- with a significant variability in the IL2 dosage, the number of infused cells, and the intensity of lymphodepletion (98–101). All these studies were performed before the large-scale expansion of ICI and targeted therapy as the standard of care for advanced melanoma.

In ASCO 2020, Sarnaik et al. (102) communicated the results of a phase II trial evaluating cryopreserved autologous TIL therapy lifileucel (LN-144) in patients with metastatic melanoma in progression to anti-PD1 +/- anti-CTLA4 therapy (and BRAF/MEK inhibitors in BRAF-mutant tumors). Among the 66 evaluable patients, there were 2 CR and 22 PR (ORR 36.4%), with the median duration of response not reached at 18.7 months. A reduction in tumor burden was observed in 81% of patients. Responses were demonstrated regardless of the BRAF mutational status, PD-L1 expression, and tumor location. Objective responses were observed in patients with brain and liver metastases, baseline bulky disease, elevated lactate dehydrogenase (LDH) levels, and prior anti-PD1 treatment. The safety profile was consistent with the known toxicities of the lymphodepletion and IL-2 regimens.

Two years later, Rohaan et al. (103) published the first phase III trial of ACT for solid tumors, comparing TIL therapy with ipilimumab in patients with stage IIIC/IV melanoma. A total of 168 patients were randomly assigned in a 1:1 ratio to receive TILs (at least 5 x 10⁹ cells, preceded by lymphodepleting CT, followed by IL2 at high-doses of 600.000 IU/kg) or ipilimumab (84 patients in each group). 89% of the patients had received previous systemic therapy, most of them adjuvant or first-line anti-PD1 antibodies. Median progression-free survival (mPFS) was significantly higher in the TIL group (7.2 vs 3.1 months; hazard ratio [HR] 0.50; p <0.001), as well as the ORR (49% vs 21%). The median overall survival was 25.8 months (vs 18.9 months in the ipilimumab group).

All the above-mentioned studies have been conducted on patients with cutaneous melanoma. A phase II trial evaluated TILs in 21 patients with metastatic uveal melanoma, with 1 sustained CR and 6 PR (ORR of 35%) (104). Interestingly, 3 of these responders (43%) had previously received immunotherapy with anti-PD1 and/or anti-CTLA4 agents, without any clinical benefit. These results demonstrate that TIL therapy merits further research in non-cutaneous melanoma, especially considering its refractory nature to ICI and other systemic treatments.

The outcomes of the published clinical trials of TIL therapy have been summarized in Table 2 .

Although TIL therapy has not yet achieved robust results in non-melanoma tumors, some phase I/II clinical trials have shown promising data in other immunogenic malignancies, such as NSCLC, cervical cancer, and HNSCC.

In a phase I/II trial, Creelan et al. (105) evaluated the efficacy of autologous TILs in combination with anti-PD1 nivolumab in 20 patients with advanced NSCLC following progression to nivolumab monotherapy. Among the 13 evaluable patients, there were 3 objective responses (2 CR -both ongoing >18 months-, 1 PR, and 8 SD with tumor reduction). Interestingly, one of the patients with sustained CR was EGFR-mutant (exon 19 deletion), which is a known predictor of anti-PD1 failure (113), suggesting that NSCLC subtypes that are commonly refractory to ICI might not be resistant to TIL therapy. Further research is required to assess whether TIL therapy could have a re-sensitizing effect for ICI use.

A phase II trial evaluated TIL monotherapy (LN-145) in 28 patients with advanced pretreated NSCLC, observing 1 CR (>21 months) and 5 PR (4 of them >8 months) among 24 evaluable patients (ORR 25%) (106).

TIL therapy also has promising results in a phase II trial with patients with advanced cervical cancer, observing 1 CR and 11 PR (2 of them unconfirmed by study criteria) (ORR: 44%) (107). Another study evaluated TILs in 29 patients with HPV+ epithelial tumors (HNSCC, cervical and anal cancer), with an ORR of 24% (including 2 CR, ongoing after 53 and 67 months) (109).

O’Malley et al. (112) conducted a multi-tumor trial with TIL therapy plus pembrolizumab in 31 patients (13 HNSCC, 10 cervical cancers, and 8 melanoma), with positive results (ORR 46.2%, 50%, and 87.5%, respectively). This TIL+ICI combination has also been evaluated in small phase II trials in HNSCC (108) and metastatic breast cancer (110), with promising results (ORR 44% and 50%, respectively). Interestingly, the second study included a patient with hormone-positive breast cancer who achieved a sustained CR (>5 years). However, a limitation of these trials is that some patients, especially those with ICI-responding tumors such as melanoma and HNSCC, might have responded to anti-PD1 blockade itself, making the role of TIL therapy difficult to assess. The sequential use of ipilimumab, TIL therapy, and nivolumab in different solid tumors was evaluated in a phase I/II trial of 25 patients, with modest results (2 PR ranging from 3 to 7 months) (111).

Despite the promising results of TIL therapy in clinical trials, many practical and economic challenges have limited its large-scale implementation (114).

TIL manufacturing begins with a surgical resection of tumor tissue, preferably from metastases accessible with minimally invasive surgery. To date, the generation of TILs has been equally successful regardless of the resected lesion site (115). Enzymatic digestion and/or mechanical fragmentation of the surgical sample is followed by culture of the fragments in IL-2 containing media, a process that might take 2-6 weeks. This minimally expanded or “young” TILs are massively expanded using high doses of IL-2, anti-CD3, and irradiated peripheral blood mononuclear cells (PBMCs) or ‘feeder cells’, within a 2-weeks rapid expansion protocol (REP). The expanded TILs are eventually reinfused into the lymphodepleted patients, followed by high-dose bolus IL2 (114).

This is an expensive and logistically complex process that requires highly specialized facilities, protocolized procedures, and qualified technical staff. These are important limitations for the widespread application of TIL therapy, particularly in developing countries. Grade 3-4 adverse events, which appear in virtually all patients treated with TIL therapy, mainly following lymphodepletion and IL2 administration, also require highly specialized management. The significant treatment toxicity, together with the long duration of the manufacturing process, are substantial obstacles to the successful application of TIL therapy in patients with refractory metastatic tumors that often lead to rapid clinical deterioration.

To date, TIL therapy has been evaluated in young, fit patients with an ECOG performance status (PS) of 0-1. Even in this selected population from clinical trials, it is often found that only a small proportion of the screened patients can finally receive treatment. For example, in a phase II trial of TIL therapy in advanced breast cancer, only 6 of the 46 screened patients (13%) received TILs infusion (110). No residual or evaluable disease after resection, inadequate material for screening, negative or weak isolation of lymphocytes in the resected samples, clinical progression, PS deterioration, and lymphodepletion-related severe toxicity were common causes of treatment failure.

Lastly, not only the economic expenses but also the individualized nature of TIL therapy, which differs from conventional commercial drugs, are important challenges for TIL therapy regulatory approval. Previous experience with CAR-T cells in hematologic malignancies will surely smooth the way for dealing with the health-economic aspects of TIL therapy and its implementation in clinical practice.