A large number of animal experiments and clinical studies have proved that all stages of tumorigenesis are affected by tumor-related inflammation. TAMs release a large number of reactive oxygen species and reactive nitrogen mediators, which lead to DNA damage and genome instability and increase the frequency of gene mutation. Accumulatively, the DNA repair function becomes abnormal, thus forming cells with the potential to develop into tumors. If the damage continues, it will further expand the cloning of cells with tumor potential, increase genomic instability and cell atypia, and finally form tumor cells, thus starting tumor occurrence. It was reported that the pro-inflammatory effect of M1-type TAMs increased the genomic instability of malignant tumor cells and became the driving force of tumorigenesis [19].

TAMs are closely related to tumor growth. TAMs stimulate tumor cell proliferation by interacting with them and secreting cytokines such as epidermal growth factor (EGF), platelet-derived growth factor (PDGF), TGF-$\beta$, hepatocyte growth factor (HGF), epidermal growth factor receptor (EGFR) and basic fibroblast growth factor (bFGF) (Fig. 2). In the mouse model of ovarian cancer, TAMs secreted a large amount of EGF to activate EGFR in surrounding tumor cells, up-regulate the VEGF/vascular endothelial growth factor (VEGFR) signaling pathway, and promote the proliferation and migration of tumor cells [20]. In hepatocellular carcinoma, TAMs promoted the stem cell-like characteristics of tumor cells through high TGF-$\beta$ expression [21]. In the mouse model of bladder cancer, type I collagen protein produced by TAMs activated the PI3K/AKT signaling pathway through integrin 21, which promoted tumor proliferation. In addition, co-culture of TAMs and esophageal squamous cell carcinoma promoted the high expression of annexin A10 (ANXA) 10 in tumor cells and then phosphorylated the AKT/ERK pathway to promote tumor cell proliferation [22].

Tumor cells usually leave the primary site and migrate to other sites. The ability of tumor cells to spread distantly is dependent on the TME. As the main component, TAMs play an important role in tumor metastasis. TAMs promote the invasion and migration of tumor cells by secreting TNF-$\alpha$, TGF-$\beta$, IL1, CSF-1, and metalloproteinases (MMPs) (e.g., MMP-2 and MMP-9) to change the connection between cells and destroy the basement membrane (Fig. 2). For instance, TAMs triggered local and systemic expression of MMP-9, VEGF, chitinase-3-like protein 1 (CHI3L1), and Lipocalin-2 (LCN2) to promote tumor metastasis in a mouse model of triple-negative breast cancer [23]. In another study, TAMs secreted large amounts of CCL18 and activated the NF-$\kappa$B/miR-21/PTEN/AKT axis in myofibroblasts through its receptor PIPTNM3, further promoting the invasion and metastasis of malignant breast tumors [24]. In one mouse lung cancer model, hypoxia increased the generation of M2 macrophages and simultaneously the expression of IL-10, VEGF, and hypoxia-inducible factor-1$\alpha$ (HIF-1$\alpha$), which further promoted cancer cell metastasis and recruited more macrophages to infiltrate the primary tumor tissue [25]. Moreover, MMP-2 and MMP-9 secreted by TAMs degraded the matrix and promoted gastric and colon cancer cell invasion [26].

TAMs induce epithelial-mesenchymal transition (EMT) in the TME. Breast cancer cells activated macrophages to transform into TAMs through GM-CSF, and TAMs secreted CCL18 to induce EMT and promote tumor metastasis [27]. Experiments in mice in vitro and in vivo showed that TAMs induced EMT in colon cancers by regulating STAT3/miR-506-3p/FoxQ1 axis, and then promoted the production of CCL2 to recruit more TAMs, which significantly increased the rate of tumor invasion and metastasis [28]. Ovarian cancer cells induced macrophages to polarize into M2-type TAMs by secreting macrophage colony-stimulating factor (M-CSF). M2 TAMs induced EMT by releasing CCL18, which further increased M-CSF transcription in ovarian cancer cells through ZEB1 protein. Thus, the CCL18-ZEB1-M-CSF interaction loop between ovarian cancer cells and TAMs increased tumor metastasis [29].

TRMs are also involved in the progression of different tumors and have organ specificity. At the early stages of NSCLCs, TRMs gathered nearby tumor cells and promoted tumor invasion and Treg cell reaction. TRMs promoted Treg cell differentiation by up-regulating the expression of CD73 and cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) in Treg cells, and protected tumor cells from being killed by CD8${}^{+}$ T cells. These results show that TRMs provide a perfect microenvironment for tumor progression [8]. In one developmental model of rat liver cancer induced by diethylmitrosamine, TNF-$\alpha$ secreted by Kupffer cells (KCs) in the liver triggered the chromosomal instability of liver prognosticator cells (LPCs) through the disorder of ubiquitin D and checkpoint kinase 2. The self-renewal of LPCs was enhanced, which promoted the transformation of LPCs into liver cancer stem cells, thus facilitating the development of liver cancer [30]. The anti-inflammatory factors macrophage galactose-type C-type lectin 2 (Mgl2), IL-10, Mgl1 and Arg1 in KCs were found increased significantly in the liver cancer model constructed by injecting AKT/Ras plasmid. KCs cells were polarized to M2-type and led to the occurrence of liver cancer. Interestingly, miR-206 in KCs cells reversed the polarization of KCs from M2-type to M1-type, and activated CCL2/CCR2 signal to recruit CD8${}^{+}$ T cells for an anti-tumor effect [31].

Microglia, the resident macrophages in the central nervous system, played a similar role in promoting the progress of glioblastoma. When microglia and glioma cells were co-cultured in vitro, microglia lost phagocytosis activity and simultaneously secreted factors such as MMP-9, EGF, IL-6 and VEGF to enhance the invasion of glioblastoma [32].

In one mouse pancreatic cancer model, TRMs were amplified in the tumor tissue. Depletion of pancreatic resident macrophages led to a significant reduction in the tumor load and showed a transcription profile of cancer fibrosis [33].

The foundation of tumor angiogenesis is endothelial cells in the TME, where new blood vessels serve as the primary route for tumor invasion and metastasis as well as supplying nutrients and oxygen for tumor growth. The density of TAMs and tumor blood vascularity are highly correlated in malignancies. TAMs promote the growth of tumor microvessels and lymphatic vessels by secreting VEGF and EGF to accelerate the proliferation of tumor cells (Fig. 2). IL-17 expressed by TAMs promoted angiogenesis in laryngeal squamous cell carcinoma [34]. In the co-culture system of TAMs and melanoma cells, TAMs promoted angiogenesis and melanoma growth by secreting adrenomedullin (ADM), which stimulated the endothelial nitric oxide synthase (eNOS) signal pathway [35]. Under hypoxic conditions, TAMs expressed the transcription factor HIF-1, which induced the transcription of VEGF, PDGF, and EGF and promoted angiogenesis. Furthermore, TAMs secreted MAPK-activated protein kinase-2 in a mice colon cancer model to increase the expression of chemokine (C-X-C motif) ligand 12(CXCL-12) in the tumor, thus promoting tumor progression and angiogenesis [36].

TAMs promote tumor progression through immunosuppression. TAMs are the main immunoregulatory cells in tumors and participate in the suppression of cytotoxic T lymphocytes (CTLs) responses in the TME (Fig. 2). Effective antitumor immunity mainly depends on the activation of CD8${}^{+}$ T cells. However, TAMs employ various mechanisms to directly or indirectly inactivate CTLs, contributing to tumor immune evasion and development [37]. TAMs with high expression of B7-H1 and B7-H4 inhibited the proliferation of CD4${}^{+}$ T cells and the secretion of IFN-$\gamma$, and promoted the immune escape and metastasis of gastric cancer cells [38]. The increased TGF-$\beta$ secreted by TAMs in lung cancer caused CD4${}^{+}$ and CD8${}^{+}$ T cell dysfunction, allowing the tumor to evade immune surveillance [39]. Moreover, TNF-$\alpha$ and IL-10 secreted by TAMs increased PD-L1 expression in monocytes. The interaction of PD-L1 expressed by monocytes and PD-1 receptor expressed on activated T cells led to tumor immune escape. Monocytes expressing PD-L1 significantly inhibited tumor-specific T cell proliferation, cytokine production, and cytotoxic potential in vitro. Moreover, PD-L1${}^{+}$ monocytes inhibited tumor-specific immunity in vivo and fostered tumor growth in NOD/SCID mice bearing human tumors [40].

TAMs promote tumor growth by regulating tumor cell metabolism and increasing the aerobic glycolysis (Fig. 2). During tumor growth, the TME presents different degrees of hypoxia. Hypoxia inhibited the uptake of glucose by TAMs, leading to an increase in the content of glucose in the TME, which further increased the utilization of glucose by tumor cells, and ultimately promoted tumor angiogenesis and metastasis [41]. TAMs with tumor-inducing metabolic features, including purine metabolism, amino acid (AA) metabolism, and glycolysis, expressed high levels of pro-angiogenic genes. During the transition from lipid metabolism to purine metabolism, macrophages gradually lost the expression of genes related to antigen presentation, while gained the expression of genes related to angiogenesis and immunosuppression, including Trem2, v-domain Ig suppressor of T cell activation (Vista) and paired immunoglobulin like receptor b (Pirb) [13]. In patients with medullary thyroid carcinoma, lactic acid produced by tumor cells caused the glucose metabolism mode of TAMs to change from oxidative phosphorylation to glycolysis, which led TAMs to secrete more lactic acid, TNF, and IL-6 to further promote tumor development [42].

In TME, the metabolism of TAMs is always in a dynamic process, which plays an important role in tumor development and anti-tumor immune response. Nutrients such as glucose, lipids and amino acids are absorbed by TAMs and catabolized or converted into biosynthetic intermediates or signal metabolites to regulate TAMs function. M1-type TAMs had a high level of aerobic glycolysis activity, which produced reactive oxygen species to kill pathogens. M2-type TAMs relied on oxidative phosphorylation and produced IL-10 and VEGF to promote the growth of malignant tumor cells [43]. In another study, M2-type TAMs expressed high-level glucose transporter and had stronger glucose-uptake capacity than M1-like TAMs. Glucose uptake promoted O-GlcN protein acylation in M2-type TAMs, which further promoted tumor metastasis [44].

In TME, various chemokines such as CCL2, CCL5, and CSF-1 secreted by tumor cells and stromal cells recruit monocytes circulating in the blood to the tumor area to differentiate into TAMs. Blocking the relevant signaling pathways with small molecule inhibitors or specific antibodies can inhibit the recruitment of TAMs.

In the liver cancer xenotransplantation mice model, the CCR2 antagonist RDC018 blocked the CCL2/CCR2 axis, and inhibited the recruitment of inflammatory monocytes and the infiltration and M2-type polarization of TAMs. The blockade resulted in the reversal of the immunosuppressive status of the TME and the activation of an antitumor CD8${}^{+}$ T cell response [45]. A phase II clinical trial of pancreatic ductal adenocarcinoma (PDAC) showed that the CCR2 inhibitor PF-04136309 effectively blocked the CCL2-CCR2 pathway, and its combined treatment with FOLFIRINOX (leucovorin + fluorouracil + irinotecan + oxaliplatin) reduced the number of CCR2${}^{+}$ monocytes in the tumor [46]. The monoclonal antibody Leronlimab, and the small-molecule inhibitors maraviroc and vicriviroc, blocked the CCL5/CCR5 signal axis and inhibited the recruitment of TAMs and tumor growth [47, 48]. A single-arm IIa trial in patients with metastatic PDAC was conducted with the CXCR4 inhibitor BL-8040, the PD-1 antagonist pembrolizumab, and NAPOLI-1. It was found that the disease control rate was 77% and the average effective duration was 7.8 months, which brought hope for PDAC treatment [49].

TAMs depletion can reduce the density of TAMs in the tumor tissue and the immunosuppressive TME. CSF-1 is necessary for the maturation, differentiation, and survival of the mononuclear phagocytes, which express the CSF-1 receptor (CSF-1R) exclusively. TAMs were massively apoptotic when the CSF-1/CSF-1R axis was blocked [50]. Given its importance in macrophages, many clinical drugs targeting CSF-1R have been developed, such as BLZ945, PLX3397, PLX7486 and PLX7486. In a mouse model of PDAC, CSF-1 neutralizing antibodies not only reduced the number of T cells but also reprogramed the remaining T cells to enhance antigen presentation and anti-tumor responses. The combination of CSF-1 neutralizing antibodies and CTLA-4 antagonists resulted in tumor regression [51]. The monoclonal antibody (RG-7155), which inhibits CSF-1R activation, induced macrophage apoptosis in vitro, and significantly reduced TAMs density and increased the ratio of CD8${}^{+}$/CD4${}^{+}$ T cells in the mouse model to activate anti-tumor immunity [52].

Another strategy to promote the depletion of TAMs is the application of bisphosphonates, including zoledronic acid and clodronate. Both in vitro and in vivo studies have shown that bisphosphonate treatment induced apoptosis in macrophages. In mouse PDAC model, clodronate liposomes significantly reduced the density of TAMs and enhanced the anti-tumor effect of CD8${}^{+}$ T lymphocytes to inhibit tumor growth [53]. In addition, trabectedin exerted an anti-tumor immune effect by mediating the depletion of TAMs in an ovarian cancer model [54].

Given the plasticity of macrophages, reprogramming TAMs to an antitumor phenotype is a highly desirable cancer treatment strategy. In addition to inducing TAMs to differentiate into M1-type, M2-type TAMs also have the potential to be repolarized into M1-type. Recent research has shown that the cyclodextrin nanoparticles (CDNP-R848) loaded with R848, an agonist of Toll-like receptors TLR7 and TLR8, delivered drugs to TAMs efficiently in vivo. Application of CDNP-R848 in multiple mouse tumor models polarized TAMs towards the M1-phenotype and inhibited tumor growth. Furthermore, the combination of CDNP-R848 and anti-PD-1 antibodies had a synergistic effect and led to tumor shrinkage [55]. In a xenograft mice model of breast cancer, the nanomedicine of STAT6 inhibitor AS1517499 and inhibitor of kappa B kinase (IKK) siRNA effectively induced the polarization of TAMs from M2-type to M1-type, reshaped the TME and significantly inhibited tumor growth [56]. Inhibition of PI3k-$\gamma$ pathway and targeting CSF-1R in M2-type TAMs upregulated the secretion of pro-inflammatory cytokines IFN-$\gamma$ and IL-12, while reduced the inhibitory cytokines IL-10 and IL-6, thereby promoting TAMs transformation from M2-type to M1-type and inhibiting tumor growth [57]. Another strategy is to induce the polarization of TAMs at the epigenetic level. The histone deacetylase inhibitor TMP195 changed the epigenetic characteristics of TAMs and made them differentiate into M1-type, which induced TAMs accumulation in breast cancer TME, enhanced their phagocytic activity, and reduced tumor burden and lung metastases [58]. Agonists for stimulator of interferon genes (STING) reprogrammed M2-type TAMs into M1-type, and overcame the immunosuppression and resistance to PARP inhibition in BRCA1 deficient breast cancer [59]. Knock-down of STING and STING activator promoted both human PBMC-derived macrophages and mouse BMDMs into M1-like subtype, and induced apoptosis of gastric cancer cells through activation of IL-6R-JAK-STAT pathway and its downstream target IL24 [60].

Checkpoint molecules, including PD-1, PD-L1, Tim-3 and Tim-4 are expressed in TAMs, which inhibit TAMs phagocytosis and promote tumor immune escape. Recent studies showed that PD-L1 antibody therapy increased the proliferation, survival and size of TAMs from mouse bone marrow. Moreover, PD-L1 antibody therapy increased the phosphorylation level of Akt and mTOR, thus inducing TAMs activation by up-regulating the expression of MHC-II and CD86. TAMs produced more TNF$\alpha$ and IL-12, which induced the polarization of TAMs to M1-type. The in vivo study results showed that the tumor growth was significantly inhibited, and the number and activation of TAMs, as well as the tumor-infiltrating T cells, were increased. Moreover, the proliferation of TAMs and its MHC-II expression were increased in PD-L1${}^{-/-}$ mice compared with wild-type mice [61].

Another study showed that almost all PD-1${}^{+}$ TAMs were M2-type. Upregulation of PD-1 expression in M2-type TAMs decreased its phagocytic ability. After transplantation of PD-L1-overexpressed and PD-L1-knockedout colon cancer CT26 cells in mice, it was found that PD-L1 knockout significantly increased the phagocytosis of PD-1${}^{+}$ TAMs and reduced the tumor growth. Human colon cancer DLD-1 cells expressing PD-L1 but unable to express mouse PD-1 were xenografted into NSG mice. The treatment with anti-mouse PD-1 or PD-L1 antibody significantly reduced mouse tumor growth, while it had no effect on tumor growth after TAMs depletion. The results suggested that the anti-mouse PD-1 antibody triggered the anti-tumor effect by binding with PD-1 on macrophages rather than on tumors [62].

Tim-3 knocked-down macrophages and hepatocellular carcinoma H22 cells were injected subcutaneously into mice. Compared with control macrophages, Tim-3 knocked-down macrophages significantly inhibited H22 tumor growth. Mechanism study found that the inhibition of Tim-3 significantly reduced the phosphorylation level of STAT-6 in macrophages, and impaired the polarization of macrophages to M2-type, thus hindering tumor growth [63]. Tim-4${}^{+}$ macrophages in pleural cavity or abdominal cavity damaged the proliferation of CD8${}^{+}$ T cells. Antibody blockade and gene elimination of Tim-4 improved the therapeutic effect of PD-1 antibody on mice bearing MC38 colon cancer [64].

Increasing the phagocytic ability of TAMs is another important anti-tumor strategy, which can directly kill tumor cells. At present, three main pathways inhibit the phagocytosis of TAMs: the signal regulatory protein $\alpha$ (SIRP$\alpha$)/CD47 pathway, the major histocompatibility complex class I/leukocyte immunoglobulin (Ig)-like receptor subfamily B member 1 (MHC-1/LILRB1) pathway, and the CD24/sialic acid-binding Ig-like lectin 10 (Siglec-10) pathways. CD47 is a membrane protein that suppresses anti-tumor immunity by inhibiting phagocytosis and participates in cell proliferation, migration, apoptosis, and immune homeostasis. The CD47 major ligand SIRP, is a membrane protein mainly expressed on macrophages and myeloid cells. The N-terminus of its extracellular domain binds to CD47, resulting in tyrosine phosphorylation on immune receptor tyrosine inhibitory motifs (ITIMs) and the release of the “don’t eat me” signal, thereby inhibiting the phagocytosis of macrophages and protecting normal cells from the destruction by the immune system [65]. Anti-CD47 antibodies (e.g., Hu5F9-G4, SRF231, and IBI188) and anti-SIRP antibodies blocked the CD47/SIRP signaling pathway and enhanced the phagocytosis of TAMs to tumor cells. A phase 1b study in patients with Non-Hodgkin Lymphoma showed that Hu5F9-G4 and rituximab synergistically enhanced the phagocytosis of TAMs to inhibit Non-Hodgkin Lymphoma [66]. Anti-SIRP monoclonal antibodies blocked the recognition of the ligand CD47 by SIRP, thus enhancing the phagocytosis of TAMs [67].

The high expression of MHC-I in tumor cells inhibited the phagocytosis of TAMs. MHC-I is coupled with two members of the leukocyte immunoglobulin-like receptor (LILR) family, LILRB1 and LILRB2, and contains ITIMs to help inhibit intracellular signal transduction. Blockade of the MHC-1/LILRB1 pathway in the tumor that exhibited drug resistance after blocking the SIRP/CD47 axis enhanced the phagocytosis of MHC-I${}^{+}$ tumor cells by TAMs in vitro and in vivo [68].

CD24 is a highly glycosylated surface protein that interacts with Siglec-10. Tumor cells highly express CD24, while TAMs highly express Siglec-10. After binding to CD24, the ITIM of Siglec-10 recruited and activated the SH2-domain-containing tyrosine phosphatase SHP-1 or SHP-2, thereby inhibiting the phagocytosis of TAMs [69]. Knocking out the CD24/Siglec-10 gene or blocking the CD24/Siglec-10 axis with monoclonal antibodies increased the ability of TAMs to phagocytize tumor cells and reduced tumor growth [70]. Recent studies showed that both mouse and human TAMs expressed PD-1, which was related to the decrease of phagocytosis. The expression level of PD-1 in M2-type TAMs was significantly higher than that in M1-type TAMs. In vivo blocking of PD-1/PD-L1 signaling increased macrophage phagocytosis, reduced tumor growth, and prolonged mouse survival time in a macrophage-dependent manner [63] (Fig. 3).