Although many improvements in molecularly directed therapies have been achieved, the prognosis of GI cancers remains poor (12). Breakthroughs in the immune checkpoint blockade offer potential therapeutic avenues, particularly with tools to overcome the mechanisms of immunosuppression in the TME. Currently, extensive publications highlight critical roles of the TIME in GI cancers, including CRC and GC (16, 17). As an important determinant of tumor progression and outcome in GI cancers, the TIME can shape cancer cell phenotypes and therapy responses through interplay with cancerous cells via chemokine and cytokine signaling or direct contact (18–20). Several reports have revealed that tumor progression and metastasis are subjected to not only genetic alterations within tumor cells but also the TIME elements. In brief, CD4+ T helper cells, CD8+ CTLs, NK cells, M1 macrophages, and DCs have been shown to be associated with a good prognosis (21). Conversely, CD4+ FOXP3+ Th2 cells, M2 macrophages, and myeloid-derived suppressor cells (MDSCs) have been attributed to a poor outcome (18).

Tumor-infiltrating lymphocytes (TILs) are immune cells that have migrated to tumor tissue and the local microenvironment. This population is indicative of an immune response generated by the patient against the malignancy. TIL populations across GI tumors generally contain T lymphocytes, particularly CD8+ cytotoxic T lymphocytes (CTLs) (12). In EC cells, blocking the programmed cell death 1 (PD-1)/programmed cell death ligand 1 (PD-L1) and TGF-β signaling pathways can synergistically restore the function of antigen-specific CD8+ T cells and the capacity of antitumor T cells (22). In addition, functional MAGE-A3-specific CD8+ T cells have an independent prognostic effect on the survival of patients with ESCC (22). Recent studies have demonstrated that higher numbers of CD3+, CD8+, or CD45RO+ T cells in tumor tissue are significantly correlated with a superior disease outcome in patients with GC, and an imbalance in Th1 and Th2 cells can lead to an immunosuppressive state dominated by Th2-type cells (23). The Th1/Th2 cell ratio in peripheral blood in GC can be used to predict postoperative prognosis (24). Similarly, the type, density, and location of immune cells in CRC also have prognostic value that is superior to and independent of those of the tumor node metastasis (TNM) classification (25). In addition to T cells, there are many other immune cell types that infiltrate GI cancers.

Tregs, as a subtype of CD4+ T cells, can inhibit effector T cells via a series of chemokine signaling (26). FOXP3+ Tregs, a subtype of Tregs, their roles are ambiguous. Some studies have shown that a high density of FOXP3+ Tregs is beneficial to the prognosis of CRC after undergoing chemo or chemoimmunotherapy (27). On the other hand, it has been shown that Tregs in the esophageal mucosa and peripheral blood of patients with esophageal cancer increase significantly (28).

DCs, on the one hand, express MHC Class II and can present their antigenic peptides to CD4+ T cells. They activate effector T cells to attack tumors and play a crucial role in shaping the host response to cancerous cells. GC patients with good DC infiltration had lower lymph node metastases and lymphatic invasion and better 5-year survival rates (78%) than patients with less DC infiltration (29). On the other hand, activated DCs help in the expansion of Tregs, consequently leading to regulation of immune responses and thereby tumor immune escape (30). Meanwhile, DCs also stimulate the formation of M2 macrophages, thereby increasing the secretion of IL-10 and TGF-β (31), which reduces the expression of IL-12 expressed by DCs and inhibits the activation of adaptive responses (32).

Tissue-resident macrophages are present prior to the development of any malignancy (33, 34). Tumor-associated macrophages (TAMs) can differentiate into two distinct subtypes, M1 and M2. M1 macrophages secrete IL-6 and IL-12 to mitigate resistance during tumor development; they can also be activated by IFN-γ to secrete TNF to kill cancer cells, while M2 macrophages secrete growth factors that promote neoangiogenesis and tumor proliferation (35). In various types of cancers, increased numbers of TAMs are often related to a poor prognosis. However, the roles of TAMs in CRC remain controversial. According to some reports, on the one hand, a high density of TAMs predicts a better postoperative outcome (36), and on the other hand, TAMs also secrete cytokines that favor tumor development (37), which indicates that the impact of TAMs on CRC needs to be further explored. Additionally, in accordance with some studies, TGF-β and other growth factors secreted by cancer-associated fibroblasts (CAFs) promoted the proliferation of CRC cells through the Smad2/Smad4 pathway (38) and MAPK/PI3K/AKT pathway (39).

Neutrophils are similar to TAMs in classification. Neutrophils differentiate into N1 and N2 according to their polarization state. N1 has antitumor activity, which directly kill tumor cells by releasing reactive oxygen species (ROS) and reactive nitrogen species (RNS) (40, 41) and can also recruit M1 macrophages and promote T cell activation. In contrast, N2 has tumor-promoting activity, promoting angiogenesis and inhibiting the function of NK cells by releasing matrix metalloproteinase 9 (MMP9) (42). In addition, N2 recruits M2 macrophages and Treg cells and suppresses the function of CD8+ T cells, as well as other native neutrophils (43). Neutrophil extracellular traps (NETs) activate Toll-like receptor 9 on CRC cells through the MAPK pathway, which leads to the growth, migration and invasion of CRC cells (44).

B lymphocytes play an important role in tumor immunity and cancer biology. However, several studies have revealed that B lymphocytes can take part in carcinogenesis and tumor progression by producing antibodies to facilitate chronic inflammation in the TIME (45, 46). Moreover, regulatory B cells can also restrain antitumor responses mediated by T cells in cancer (47). Regulatory B cells (Breg cells), a novel subset of B lymphocytes, appear to facilitate tumor growth and progression via the production of IL-10 to suppress the activity of CD8+ T cells in squamous carcinoma (48). In CRC, the presence of B cells seems to be detrimental to prognostic outcome (49).

NK cells are unique in that they have both innate and adaptive immune properties. NK cells participate in the antitumor immune response through the production of proinflammatory cytokines, which recruit and induce the proliferation of other immune cells (12). Activated NK cells can directly kill some tumor cells and virus-infected cells. In EC, expanded NK cells had high cytotoxicity against ESCC cells, especially those with the epithelial-mesenchymal transition phenotype (50). In another investigation, a higher NK cell density was shown to be significantly related to a higher survival rate in GC, especially for advanced patients (51).

Cytokines are important players in the tumor microenvironment, and cytokines/chemokines are used to establish connections between various immune cells, such as IL-6 (52), IL-10 (53), IL-12 (54), IL-35 (55), epidermal growth factor (EGF), vascular endothelial growth factor (VEGF), tumor necrosis factor alpha (TNF-α) (56), interferon γ (IFN-γ) (57), indoleamine-2,3-dioxygenase (IDO) (58), and transforming growth factor beta (TGF-β) (59). For example, IL-10 and TGF-β switch macrophages from a M1-like (proinflammatory or classically activated) state to a M2-like (anti-inflammatory or alternatively activated) state. M1 macrophages induce antitumor immune signaling and correlate with tumor killing capacity. Conversely, M2 macrophages exhibit protumor effects and contribute to fibrosis and the production of matrix proteins (60) as well as angiogenesis, metastasis, and the suppression of adaptive immunity (61, 62). Human CRC cell lines cultured in vitro are able to polarize macrophages toward the M2 phenotype (63). In GC patient samples, M2 macrophages in the stroma may be correlated with the presence of a lesion (64). The CCL2-CCR2 axis in esophageal carcinogenesis contributes to immune evasion and tumor promotion through the PD-1 signaling pathway (65). At the same time, the activity of NK cells and effector T cells was suppressed due to the increased secretion of cytokines such as IL-10, TGF-β, EGF and VEGF (50).

Unlike strictly defined immune cells, fibroblasts are present within the stromal microenvironment and serve to produce extracellular matrix (ECM) proteins in particular collagen. They actively manufacture and respond to cytokines in cooperation with immune cells within the stromal microenvironment. Fibroblasts are also associated with epithelial cell polarity, proliferation, and, to some extent, tumorigenic potential. CAFs have been shown to drive increased tumor growth compared to normal fibroblasts (66). They contribute to cancer cell survival and progression via a series of nutrient-rich ECM proteins or ECM-degrading proteases, resulting in persistent chronic inflammation within the tumor microenvironment and enhanced epithelial mesenchymal transition (EMT) of tumor cells (67, 68).

Recent studies have demonstrated that CAFs have the capacity to produce proinflammatory cytokines, which disrupt the normal cytokine balance to stimulate tumor growth by initiating angiogenesis and inhibiting CTLs (69, 70). Additionally, CAFs have been shown to secrete high levels of the proinflammatory cytokines IL-1β, IL-8, IL-10, tumor necrosis factor-alpha (TNF-α), monocyte chemoattractant protein-1 (CCL2), stromal derived factor-1 (CXCL12) and interferon-beta (IFN-β) (71). The crucial role of CAFs in tumorigenesis has been addressed by genetic analyses showing that their gene expression profiles are very different from those of normal breast fibroblasts. Moreover, the expression profiles of CAFs obtained from tumors with poor versus good prognosis are also very different (72). Good-outcome fibroblasts were associated with immune modulators involved in the Th1 immune response. This includes the expression of T cell receptor complexes (CD8a, CD247, and CD3D), MHC class I protein binding and granzyme A/B activity. The poor-outcome stroma had increased levels of hypoxia and angiogenesis and decreased chemokines that stimulate NK migration and T cell survival (73).

As the first inhibitory receptor identified, CTLA-4 is mainly located on activated T lymphocytes and is also highly expressed in several types of cancer (83). There is a competitive interaction between CD28 and CTLA-4 in binding with CD80/CD86 molecules. CTLA-4 transmits an inhibitory signal to T cells, whereas CD28 transmits a stimulatory signal. A recent study revealed that the combined consideration of CTLA-4 expression and the platelet-lymphocyte ratio (PLR) has potential prognostic value for people suffering from ESCC (84). Therefore, it is not surprising that CTLA-4-targeted agents have shown great potential in the treatment of many cancers. In clinical trials, the tumor response rates of two anti-CTLA-4 agents, ipilimumab and tremelimumab, were approximately 10% (85). A large phase II clinical study explored the efficacy and safety of nivolumab ( ± ipilimumab) in the treatment of advanced CRC. The results revealed that most patients with microsatellite instability-high (MSI-H) metastatic CRC (mCRC) benefited from the nivolumab + ipilimumab regimen, and the regimen was well tolerated, but whether the combination therapy is superior to either drug as monotherapy therapy is still uncertain (86).

PD-1 is one of the crucial immune checkpoint receptor proteins on T cells, B cells and NK cells and can bind to PD-L1 and/or PD-L2, which are located on the surface of multifarious tumor cells as well as hematopoietic cells. The combination of PD-1 and PD-L1/L2 expressed on tumor cells can dramatically inhibit the apoptosis of tumor cells, promoting the depletion of peripheral effector T cells and catalyzing the transformation of effector T cells into Treg cells (87). In 2017, the US Food and Drug Administration (FDA) approved pembrolizumab for the treatment of unresectable or metastatic MSI-H or mismatch repair deficient (dMMR) CRC that progressed following treatment with fluoropyrimidine, oxaliplatin, and irinotecan. This approval was based on results from the phase II CheckMate-142 trial (88), in which the objective response rate (ORR) was 28% in mCRC patients who received prior fluoropyrimidine, oxaliplatin, and irinotecan; in the trial, 1 patient had a complete response (CR), and 14 patients had partial responses (PRs). Furthermore, the expression of PD-L1 in cancer cells and the TME may contribute to the development of EBV-associated GC, and PD-L1 overexpression predicts large tumor size, lymph node metastasis, and poor prognosis in GC (89). A recent study presented at the American Society of Clinical Oncology (ASCO) meeting revealed that an anti-PD-1 monoclonal antibody could significantly prolong the OS of PD-L1-positive patients with advanced EC regardless of whether it was used alone or in combination with chemotherapy; furthermore, the antibody was highly safe (90, 91).

TIM-3 is a cell surface molecule expressed on DCs, monocytes, CD8 T cells and Th1 cells. The ligands of TIM-3 are frequently overexpressed in tumor cells. TIM-3 is a negative regulator of the antitumor immune response and has been proven to be associated with the occurrence and development of several malignant tumors. At both the mRNA and protein levels, the expression of TIM-3 in ESCC is remarkably higher than that in matched adjacent normal tissues; furthermore, researchers have found a close connection between the overexpression of TIM−3 and poor survival in patients with ESCC. Similarly, TIM-3 knockdown significantly inhibited the propagation, migration and invasion of ESCC cell lines. Further research explored whether the depletion of TIM-3 could restrain several signal transduction pathways, including the snail, p-GSK-3β and p-AKT pathways (92). In addition, some studies have suggested that the increased expression of TIM-3 on T cells may be partially responsible for the development of GC by inducing the secretion of IFN-γ and TNF-α (93). Another study has shown that reduced TIM-3 expression induced by genetic polymorphisms of TIM-3 may promote CRC invasion and metastasis (94). Taken together, these findings show that TIM-3 is a critical mediator in the progression of various cancers and may serve as a potential therapeutic target.

LAG-3 is another membrane protein expressed on B cells, some T cells, NK cells and TILs. LAG-3 can enhance the activation of Treg cells and prevent T cells from proliferating and differentiating into effector cells by binding with MHC II. In addition, some basic research has shown that dual blockade of LAG-3 and PD-1 can induce strong antitumor action (95). Furthermore, high expression of LAG-3 is associated with an improved survival rate in patients with ESCC or CRC and is being evaluated as a biomarker to predict response to antitumor treatment. These results indicate that LAG-3 can be used as a marker of the immune response for patients with ESCC or CRC (96, 97).

Signal regulatory protein-α (SIRPα) acts as an inhibitory receptor and is expressed on myeloid cells such as macrophages, DCs, mast cells and neutrophils. It interacts with the broadly expressed transmembrane protein CD47, also called the “don’t eat me” signal. In addition, the interaction between CD47 and SIRPα can help tumor cells avoid phagocytosis in the TME (98). The biological and preclinical relevance of the interaction of SIRPα and CD47 has been extensively investigated. A recent study showed that blocking SIRPα/CD47 signaling in ESCC cell lines could enhance the phagocytosis of ESCC cells in a dose-dependent manner (99). In addition, with the evolution of cancer immunotherapy, SIRPα/CD47 immunotherapy to stimulate the innate immune system as a cancer treatment has drawn increasing interest. Other reports have confirmed that the high expression of CD47 is an independent prognostic factor in GC, and the OS rate of patients with CD47-positive tumors was significantly lower than that of patients with CD47-negative tumors (100). Aberrant expression of CD47 has also been reported to strongly promote the proliferation of tumor cells (101).

NK cells are innate lymphocytes and play a pivotal role in host immunity by killing virally infected and/or cancerous cells. Several clinical studies have suggested NK cell-based immunotherapy as a potential therapy for cancer. Inhibitory receptors, such as killer-cell immunoglobulin-like receptors (KIRs) and CD94/NK group 2 member A (NKG2A), recognize human leukocyte antigen (HLA) class I molecules on normal host cells. By lacking inhibitory receptor ligands, tumor cells that have downregulated surface MHC-I expression become susceptible to attack by NK cells. Current NK cell-based cancer immunotherapy aims to reverse tumor-induced NK cell dysfunction and sustain NK cell effector functions (111). To fight against tumors, NK cells can also be modulated to target the cell types responsible for maintaining the immunosuppressive TME, including M2-polarized macrophages, MDSCs, Treg cells, and fibroblasts (112). According to recent reports, adoptive cell therapy (ACT) has become a powerful tool to improve host antitumor activity (113), mainly including autologous NK cell transfer, allogeneic NK cell transfer and CAR-engineered NK cells (114). The three methods all artificially activate and expand NK cells in vitro and then transfer the modified NK cells to patients. Under the stimulation of some cytokines and costimulators, the antitumor ability of NK cells has been significantly improved (113, 115), and NK cells are expected to achieve staged success in the course of immunotherapy.

Natural killer T cells (NKTs) are unique T lymphocytes that have the characteristics of conventional T cells and natural killer cells (116). They play an important role in connecting the natural immune system with the adaptive immune system and are also an important intermediary for mediating the immune response and tumor monitoring (117). It is usually composed of two parts, namely, the invariant TCRα chain and the semivariant TCRβ chain, recognizing the lipid antigen presented by the nonclassical MHC class I molecule CD1d (118). More importantly, the activation of CD1d-dependent invariant natural killer T cells (iNKT cells) can promote the release of cytokines such as IL-5, IL-6, IL-10, IL-17, IL-21, TNF-α, TGF-β and granulocyte-macrophage colony stimulating factor (GM-CSF) (116, 119, 120) and indirectly activate the antitumor mechanism. At present, several cancer immunotherapies based on iNKT cells have been developed, mainly including antigen-presenting cells (APCs) pulsed with α-GalCer, the transfer of ex vivo–expanded and/or activated iNKT cells, and iNKT cell–activating ligands (117). First, a clinical experimental study has not indicated any major toxicity or severe side effects in patients who have received repeated injection of the tumor immunity agonist α-GalCer (121), implying its clinical value in improving tumor immunity. In addition, patients with head and neck squamous cell carcinoma, who were treated with the combination of iNKT cells and α-GalCer-pulsed DCs, the antitumor immunity was greatly improved (122, 123)

Tumor-infiltrating immune cells play an ambiguous role in tumor development and progression. It has been extensively reported that TAMs promote or inhibit the expansion and dissemination of cancer cells depending on their functional states. Tumor-infiltrating macrophages are mainly recruited by CCL2 or CSF-1, and TAM depletion therapy (e.g., CCR2 or CSF1R antagonists) has already been tested in clinical trials for malignant solid tumors (124, 125). Some studies have also shown that CSF1R inhibitors represent a promising combination partner for T cell-enhancing immunotherapies (126, 127). Further investigation of the synergistic effects of these agents with immunotherapies will lead to the improvement of ongoing immunotherapeutic strategies. At the same time, some cancer immunotherapies targeting TAMs are on the rise, and most of them follow the principle of directly reducing the formation of TAMs or polarizing TAMs from the pro-tumor phenotype (M2-like), which helps tumor development to an antitumor phenotype (M1-like) (128). By preventing the recruitment of TAMs or blocking the CSF1/CSF1R axis, the survival rate of TAMs can be significantly reduced (129, 130), as mentioned above. In addition, some studies have shown that the polarization of TAMs to M1 macrophages can be promoted by targeting antibody-dependent cellular phagocytosis (ADCP) induced by the SIRPα/CD47 axis, thus enhancing the tumor immune response (131). Finally, a possible strategy for cancer treatment is the use of natural living macrophages as drug delivery vehicles (128).

As a subtype of T lymphocytes, γδT cells have a wide range of antigen recognition abilities and can be used to recognize a series of antigens, including mitochondrial ATPase, human mutS homolog 2 and nonpeptide phosphoantigens (132–134), without using traditional APC to mediate antigen recognition (135, 136). The mechanism of antigen recognition depends on its peptide sequence. Since the sequence of complementarity determining region 3 (CDR3) of the delta chain is longer and more varied, it is more important in the process of antigen recognition (137). In addition to expressing NK cell receptors such as NGG2D, it may be a bridge between the innate immune system and adaptive immunity (138). According to related reports, under certain conditions, γδT lymphocytes may be activated in the late stage of the immune response (139, 140), which shows cytotoxicity by secreting perforin and granzyme and can also regulate the secretion of IFN-γ and TNF-α (141). Conversely, some cytokines may drive γδT lymphocytes to remodel tumor-immunosuppressive functions; thus, there is an urgent need to optimize γδT lymphocytes to obtain new antitumor targets (142). A recent study demonstrated for the first time that CAR-γδT cells obtained by modification using CAR technology are able to migrate to tumor cells and cross-express antigens (143), which opens a new avenue for cancer immunotherapy research.

IDO1 is the rate-limiting enzyme that participates in tryptophan metabolism by converting the amino acid L-tryptophan (Trp) into L-kynurenine (Kyn). This enzyme assists tumor cells in maximally utilizing essential amino acids such as tryptophan in the TME to support tumor growth. However, as immune cells lose tryptophan, they lose their ability to fight cancer cells. Overexpression of IDO1 in tumor cells inhibits T cell function and impairs immune surveillance, leading to immune escape (144). In addition, it has been verified that the expression of IDO1 is associated with prognosis in EC and could be used as a prognostic biomarker (145). In a multivariate analysis, IDO1 expression was proven to be an independent prognostic factor for developing recurrence (146). Chen et al. reported that the enzyme activity of IDO1 directly affected the radiosensitivity of CRC, and IDO1 inhibition made CRC cells more sensitive to radiation-induced cell death and protected the normal small intestinal epithelium from radiation toxicity, indicating that IDO1 inhibition enhances the radiotherapy effect in CRC (147). According to relevant studies, IDO1 can also cooperate with COL12A1 to promote GC metastasis. This new discovery suggests that IDO1 and COL12A1 may be promising targets for anticancer therapy in GC (148).