Recent studies have reported CD68+ macrophages in PAs at levels three times higher than those in the normal pituitary (32, 33), and these cells are the most highly infiltrating immune cells in PAs (34). Detected mostly by immunohistochemistry, infiltrating macrophages vary greatly among different PA subtypes. In a study that included 20 nonfunctioning pituitary adenomas (NFPAs) and 50 functional PAs, Zhang et al. reported that the CD68+ macrophage infiltration of growth hormone (GH)-secreting adenomas and prolactin (PRL)-secreting adenomas was significantly higher than that of NFPAs and adrenocorticotropic hormone (ACTH)-secreting adenomas (35). A study by Marques et al., which included 16 NFPAs and 8 GH-secreting adenomas, showed no significant difference in macrophage infiltration between these two adenoma types (34). Principe et al. found that gonadotropin-secreting adenomas (n=12) had higher CD68 expression than other functional PAs (n=16) (33). Another specimen analysis including 44 NFPAs and 28 functional PAs revealed no significant difference in CD68+ cell infiltration between NFPAs and functional PAs, while among functional PAs, PRL-secreting adenomas had higher CD68+ cell infiltration than ACTH- and GH-secreting adenomas (32). Lu’s team, however, performed a more pathologically detailed analysis with 9 densely granulated GH-secreting adenomas, 9 sparsely granulated GH-secreting adenomas, 9 null cell adenomas and 8 ACTH-secreting adenomas. They found that null cell adenomas and sparsely granulated GH-secreting adenomas had higher CD68+ cell infiltration than the other two subtypes (36). For CD163, an M2-TAM marker, few relevant statistical results on its expression in PAs have been published. Using bioinformatics analysis, Yeung et al. reported that NFPAs had higher CD163 expression than functional PAs (37). According to Principe et al., CD163+ cell infiltration of gonadotropin-secreting adenomas was higher than that of other PAs (33).

Based on the current results, no obvious pattern of macrophage infiltration was observed in different subtypes of PAs. A study with a larger sample size supports the conclusion that PRL-secreting adenomas are more highly infiltrated by macrophages, while ACTH-secreting adenomas, gonadotrophin cell adenomas or NFPAs are infiltrated to a lesser extent (35). However, Principe et al. demonstrated that gonadotrophin cell adenomas had higher CD68+ cell infiltration than other functional PAs. Clinically, gonadotrophin cell adenomas are the main pathological subtype of NFPAs. Moreover, they found higher F4/80+ cell infiltration in tumors originating from gonadotrophin cell lines than in tumors originating from GH cell lines in a tumor-bearing animal model. This difference was no longer significant when tumorigenic cells were replaced by human PA primary cells (33). According to the results of this study, macrophage infiltration varied greatly among samples, which indicated that gonadotropin-secreting adenoma has substantial heterogeneity in macrophage infiltration. Several reasons are considered possible. First, Yagnik et al. found that the infiltration of CD11b+ myeloid macrophages in NFPAs differed greatly between samples (38). For null cell adenomas, a subtype of NFPA, it was observed that CD68+ macrophage infiltration was higher than that in ACTH-secreting adenomas and densely granulated GH-secreting adenomas (36), which indicates that null cell adenomas may have affected the statistical results of NFPAs in other studies. Although null cell adenomas only account for a small proportion of NFPAs, they should be studied separately from the other pathological types of NFPAs in future studies. Second, chemokines might be expressed differently among subtypes of PAs. CCL5 was found to be expressed higher in GH adenomas than in gonadotrophin adenomas, while CSF1 was found to be expressed more in gonadotrophin adenomas. Meanwhile, gonadotrophin-secreting cell lines could increase the expression of CSF1R on monocyte cell line THP1, compared to GH cell lines (33). Marques et al. have similar results: they found that NFPAs could secret more IL-8, CCL2 and CCL4 than GH adenomas (34). Due to the differential results shown above, macrophages infiltration must be the consequence of the secretion of multiple chemokines. Thus, the marker of macrophage subgroup in different pituitary adenomas should be further studied. Third, differential tumor development stage with differential chemokines expression could lead to differential macrophage infiltration. Yagnik et al. inferred a dynamic change of macrophages infiltration in the development of NFPAs, caused by dynamic chemokines expression, such as GM-CSF and MCP-1, supporting the opinion mentioned above (38).

Multiple studies have reported a correlation between TAMs and the biological behavior of PAs, such as tumor growth and invasion. Regarding PA growth indicators, after studying PA specimens, Lu et al. and Zhang et al. found that CD68+ macrophage infiltration was related to the size of PAs (35, 36). Principe et al. found that the percentage of CD68+ cells was positively correlated with tumor size in gonadotropin cell adenomas (33). In terms of PA invasion indicators, Lu et al. team found that CD68+ macrophage infiltration was positively correlated with the Knosp classification grade, and sparsely granulated GH adenomas, which are considered to be more aggressive, had more CD68+ macrophage infiltration than densely granulated GH-secreting adenomas (36). An analysis by Zhang et al. showed that CD68+ macrophage infiltration was significantly increased in invasive PAs compared with noninvasive PAs (35). Principe et al. reported that CD68+ macrophage infiltration was related to invasion in both functional PAs and NFPAs (33). Yagnik et al.’s found that the M2/M1 gene expression ratios of 88% NFPA samples with cavernous sinus invasion was higher than one (38). These results show that macrophage infiltration plays an important role in tumor growth and invasion. And among the two subtypes of TAMs, M2-TAMs seem to have a positive relation with PA invasion compared to M1-TAMs.

In addition to tumor growth and invasion, the PA-related biological behaviors of clinical concern include drug resistance, recurrence, and even metastasis. However, no convincing data support the relationship of these behaviors with macrophage infiltration. Different opinions also exist about the relationship between PAs and macrophage infiltration. As mentioned earlier, a recent analysis of a larger sample size of pathological specimens found that macrophage infiltration is related to the growth and invasion of PAs (33, 35, 36). However, according to the study by Marques et al., CD68+ macrophage infiltration was not involved with cavernous sinus invasion and was not related to the Ki67 index, which might be attributed to the study’s relatively small sample size (34).

Several studies have shown that M2-TAMs are the primary infiltrative macrophage subtype (33, 34, 37). However, Yagnik et al. also observed that a few specimens were predominantly infiltrated with M1-TAMs rather than M2-TAMs (38). M1-TAMs, which are traditionally considered an anti-tumoral cell type, may promote immunity and suppress tumor growth in PAs. In 1993, TtT/M-87 was separated from PAs. TtT/M-87 is a macrophage cell line derived from thyroid stimulating hormone (TSH)-secreting adenomas that express TNF-α, IL-1α, and MHC-II; these factors can assist spleen-derived lymphocytes in inhibiting the growth of tumor cell lines in vitro (39). Due to the lack of relevant research, further in vivo and in vitro studies are needed to verify the role of M1-TAMs in PAs.

M2-TAMs promote PA invasion by immunity suppression, PA cell epithelial-mesenchymal transition and proliferation, vascularization and ECM remodeling (40) ( Figure 1 ), as described below.

Studies have found that metabolites, cytokines, chemokines and other factors can act on macrophages, affect their phenotypes, and promote their polarization to M2-TAMs (52). In PAs, little is known about factors that affect TAM polarization. Through in vivo and in vitro models, Zhang et al. demonstrated that lactic acid could activate the mTORC2 pathway to cause M2-type polarization and promote the expression of multiple factors in macrophages, such as CCL17, CCL22, IL-1α, IL-10, and TGF-β (35). Marques et al. found that the IL-4 level was 5 times higher than that of IFN-γ in the PA-TME (34). Both IL-4 and IFN-γ are well-known cytokines that polarize macrophages to M1 and M2 types, respectively. Lactic acid is due to the anaerobic environment inside the solid tumor. The source of IL-4 and IFN-γ could be tumor cells, the tumor-infiltrating lymphocytes, mesenchymal cells or macrophages. A recent study found that in lung cancer, IL-4 secreted by M2-myeloid cells and tumor cells activate the STAT6 pathway and promote M2 polarization and tumor progress (53). Similarly, PA cells and TAMs might be the potential source of IL-4 and worth exploring. During the development of PAs, the balance of M1 and M2 macrophages changes based on factors in the TME. Yagnik et al. found that as the number of CD11b+ myeloid cells increased, the expression of the M1-TAM marker CD64 increased. Compared with CM from M1-TAMs, CM from M2 macrophages reduced the expression of MCP-1 in NFPA cells. They also observed that GM-CSF inhibition abrogated the M1 polarization of THP-1 cells induced by CM from NFPA cells. These results suggested dynamic changes that affect the balance of M1- and M2-TAM infiltration in NFPA. Therefore, they divided NFPA into M1-NFPA and M2-NFPA according to the percentage of M1- and M2-TAM infiltration. Assuming the tumors behaved as M2-NFPA during early formation, infiltrating monocytes differentiated into M2-TAMs due to MCP-1 produced by tumor cells. However, over time, the aggregated M2-TAMs downregulated MCP-1 and upregulated GM-CSF expression in PA cells. As suggested above, GM-CSF can induce the differentiation of monocytes into M1-TAMs, while the MCP-1 downregulation can also slow monocyte recruitment. Therefore, as time passed, the proportion of infiltrating M1-TAMs gradually increased (38). Fujawara et al. found that the number of infiltrating macrophages increased even before any tumors formed and that they were mainly M2 macrophages, with few M1 macrophages (54). Such results supported Yagnik’s theory, which at least partially explains the heterogeneity of macrophage infiltration in NFPAs. NFPAs intrinsically have different statuses of macrophage infiltration, which has caused substantial differences among the results in the PA studies mentioned above. This theory has provided a potential transition for PAs from M2-dominant to M1-dominant status. Inferred from their theory, most PAs must end up in an M1-dominant status. However, according to Yeung et al.’s results of 134 patients, infiltration of M2-TAMs is far more than that of M1-TAMs in all subtypes of PAs. Therefore, uncovered molecular feedback could exist to shift the balance to M2-type and needs exploring.

Although most studies support the relationship between macrophages and PAs, the relationship may be complicated. A recent study analyzed 140 PAs and 20 normal pituitary tissues and found that PAs can be divided into three subtypes according to immune cell infiltration, of which only one subtype is characterized by relatively higher levels of infiltrating macrophages (55). This indicates that the role of macrophages in PA may not be simply summarized as purely “relevant” or “irrelevant”.

As mentioned above, the pattern of macrophage infiltration among different subtypes of PAs has not been established. The infiltration of CD68+ cells and CD163+ cells in NFPAs and functional PAs varies among different studies (32, 33, 35). NFPAs include tumors with a variety of pathological classifications. Clinically, NFPAs are mainly composed of gonadotropin cell adenomas. In addition, this tumor subtype also encompasses null cell adenomas and hormone-silent adenomas (56). Marques et al. reported on different subtypes of NFPAs and found no difference in macrophage infiltration among gonadotropin cell adenomas, silent ACTH adenomas and null cell adenomas. However, only 1–2 cases of the latter two were observed, which was not enough for a convincing conclusion (34). Therefore, it is not advisable to study relevant mechanisms and therapies without knowing the pattern of macrophage infiltration of different PA subtypes. Large multicenter studies are needed to clarify the pattern mentioned above and to further clarify the relationship between macrophages and PA invasion, drug resistance and recurrence.

Several directions are considered promising. Barry et al. found that AIP-mutated tumors showed an increased infiltration of CD68+ macrophages comparing to sporadic GH tumors (41). Thus, AIP-related pathway should be paid attention to. A study had found a correlation between AIP and cAMP signaling (57). Thus, the activation of PKA/cAMP pathway might explain the heterogeneity mentioned above. In addition to that, Principe et al.’s study found that the infiltration of CD4+T and CD8+T rather than NK cells and B cells was higher in functional PAs than gonadotropin-secreting adenomas, which was the main component of NFPAs (33). Zhou et al. ran RNA-seq on 115 human PAs and found B cells, CD8+ T cells and CD4+ T cells were negatively correlated, while NK cells were positively correlated with macrophage infiltration. The infiltration of T cells and B cells in functional PA was significantly higher than that of NFPA, especially in GH-secreting adenoma. Meanwhile, they found that higher CD8+T and CD4+T infiltration corresponded to stronger tumor invasiveness and worse survival (58). Mei et al.’s results also suggested functional PAs had a higher CD4+T and CD8+T infiltration than NFPAs (32). From these results, functional PAs seem to attract more CD4+T and CD8+T cells than NFPAs. Considering the negative relation between these lymphocytes and macrophages, T cells might affect the recruitment of macrophages. The molecular and genetic mechanism underneath needs to be further studied.

Traditionally, TAMs are believed to enter tumors as a result of chemotaxis of peripheral blood mononuclear cells. However, studies have reported embryonic macrophages in the TME of lung cancer and glioma (59, 60). At present, studies on the origin of PA-TAMs are still in the initial stages. Yagnik et al. found that CD11b+ myeloid cells in NFPAs expressed CCR5 rather than CX3CR1. CCR5 is expressed by recruited macrophages, while CX3CR1 is expressed by microglia, which are tissue-resident macrophages in the brain (38). However, Yagnik et al.’s research was limited to one PA subtype, and whether this was a common phenomenon in other subtypes remains to be determined.

In recent years, novel pro-tumoral mechanisms of TAMs have been consistently discovered in brain tumors. In gliomas, M2-TAMs can induce CECR1, activate the MAPK pathway, and stimulate the proliferation and migration of tumor cells (61). Another study showed that microglia can induce the expression of PDGFRB in glioma cells, thereby enhancing their migration ability (62). In PAs, little is known about the pro-tumoral mechanism of TAMs. Except for Zhang et al.’s work (35), no other relevant studies have been published. The prospects of mechanistic studies of PA-TAM are described next.

Studies on macrophage-targeted therapy have been widely performed in other tumors (70). In solid tumors, the mechanisms of macrophage-targeted therapy mainly include macrophage depletion, macrophage recruitment blockade and macrophage reprogramming. The classification according to the mechanism and the relevant information of the corresponding representative drugs are shown in Table 1 . The CSF1-CSF1R axis is necessary for macrophage differentiation and survival (78). Targeting CSF1R with a CSF1R monoclonal antibody or small molecule inhibitor can effectively deplete macrophages (79). Principe et al. found that blocking CSF1 can inhibit the migration of THP1 monocytes toward PA cells in vitro (33). This indicates that targeting the CSF1-CSF1R axis in PA may not only deplete macrophages but also reduce their recruitment. Targeting macrophage recruitment is another strategy for macrophage-targeted therapy. This strategy primarily targets chemokine pathways using CCL2 and CCR2 monoclonal antibodies to reduce macrophage recruitment. A study found that in esophageal cancer, blocking the CCL2-CCR2 axis could inhibit the recruitment of TAMs, thereby enhancing the anti-tumor immunity of CD8+ T cells in the TME (80). In PAs, it was found that CCL2 expression was higher than in the normal pituitary (34). Barry et al. found that blocking CCR5 could inhibit the migration of macrophages induced by CM of the GH3 cell line (41). Zhang et al. reported that the CCR4 antagonist AZD2098 significantly inhibited the promotive effect of TAM-derived CCL17 on the proliferation, invasion and migration of the PA cell line GH3 in vitro and reduced tumor burden in GH3 tumor-bearing mice in vivo (35). Therefore, targeting CCL2, CCR4 or CCR5 may also reduce the recruitment of macrophages to the PA-TME. Using multiple methods, M2-TAMs can be reprogrammed into M1-TAMs with anti-tumor properties. In other solid tumors, macrophage reprogramming therapy includes the use of anti-CD47 antibodies, anti-CD40 antibodies, Toll-like receptor (TLR) agonists, histone deacetylase (HDAC) inhibitors and PI3K inhibitors (76, 77, 81, 82). In PAs, Zhang et al. found that lactic acid induced M2 polarization of TAMs through the mTORC2 pathway. Additionally, CM from GH3 cells treated with an LDHA inhibitor significantly reduced M2 markers in THP-1 cells compared with the control group (35). Therefore, targeting LDHA or mTORC2 may repolarize M2-TAMs to M1-TAMs. A few reviews have recently summarized the drugs that target macrophages in cancer (83, 84). These drugs will allow for PA macrophage-targeted therapy in the future. Related drugs and possible targets in PAs are shown in Table 1 .