Macrophages have two distinct ontogenies. The first of these is monocyte-derived macrophages (MDMs) which originate from progenitors in the bone marrow and other hematopoietic niches (2), progress through several stages of differentiation, and enter the systemic circulation as monocytes. Circulating monocytes are recruited to tissues in response to locally released chemo-attractants where they differentiate into macrophages (3). Once inside the tissue, MDMs may be short- or long-lived, and their population is maintained through recruitment of new circulating monocytes as well as proliferation of pre-existing MDMs (4, 5). The second group, known as tissue-resident macrophages, arise early in embryonic development, migrating from the yolk sac or fetal liver into developing organs where they differentiate into tissue-specific macrophages, including Kupffer cells (liver), osteoclasts (bone), and microglia (brain) (5). In adults, these macrophages self-renew largely independently of the bone marrow (6, 7). Macrophages in the TME are referred to as tumor-associated macrophages (TAMs). While most TAMs are monocyte-derived, tissue-resident macrophages make up a considerable percentage of TAMs in some tumor types (8–10).

Once inside the tumor, macrophages take on various phenotypes and functions in response to stimuli in the microenvironment. These phenotypes are referred to as “polarization states.” There is a wide spectrum of macrophage polarization states ranging from pro-inflammatory (M1) to anti-inflammatory (M2).

M1 macrophages (historically called “classically activated”), are pro-inflammatory cells that participate in the host immune response against pathogens and can have anti-tumor activity. As such, environmental factors associated with infection and inflammation (including interferon (IFN)-γ, bacterial lipopolysaccharide (LPS), and granulocyte-macrophage colony stimulating factor (GM-CSF)) promote M1 polarization (11). These signals cause macrophages to express surface proteins related to antigen presentation and T cell activation (including HLA-DR, CD80, and CD86) (11–14), and secrete inflammatory cytokines such as tumor necrosis factor (TNF)-α and interleukin (IL)-1β to further enhance the immune response (11). M1 macrophages promote tumor cell killing through strong antigen presentation and effective activation of the innate and adaptive immune responses (15). Indeed, high M1 macrophage infiltration is correlated with positive outcome in cancer patients (16, 17).

M2 (or “alternatively activated”) macrophages, are anti-inflammatory cells that are involved in tissue repair and immune suppression. While these cells are essential in maintaining homeostasis in healthy tissues, they can also promote tumor growth and metastasis in the TME. M2 macrophages are induced by anti-inflammatory cytokines in the microenvironment, including IL-4 and IL-10 (11). These signals cause macrophages to express surface proteins such as CD163 and CD206, which are involved in tissue “clean-up” and homeostasis, and to secrete additional anti-inflammatory factors, such as IL-10 and transforming growth factor (TGF)-β, which further suppress the immune response (11, 18, 19). M2 macrophages also express high levels of vascular endothelial growth factor (VEGF), which promotes tumor vascularization, enhancing the delivery of oxygen and nutrients to the tumor (11). M2 macrophages are poor antigen presenters and suppress both innate and adaptive anti-tumor immunity (15). Furthermore, M2 macrophages are implicated in chemoresistance and metastasis, and high M2 infiltration is associated with poor prognosis in cancer patients (20–26).

Though macrophage polarization is a spectrum with M1 and M2 on opposing ends, it is common in the literature to oversimplify this state and treat macrophage polarization as a dichotomy (M1 or M2). Given that many “anti-inflammatory” macrophages also participate in inflammatory signaling and vice versa, the terms “M1” and “M2” should merely give a sense of how a macrophage is predominantly functioning in a particular environment. It is also important to note that macrophage phenotype is highly plastic. Similar to other components of the innate immune system, macrophage phenotype can quickly change in response to environmental cues (27). Indeed, in vitro and in vivo studies confirm that macrophages may repolarize in response to particular stimuli (28–30), an effect that has been leveraged in several immunotherapy clinical trials (31). Promising macrophage-targeting therapies and the challenges associated with their development are reviewed elsewhere (32–37). Given their significant, plastic, and diverse roles in cancer progression, understanding the mechanisms behind macrophage-mediated cancer progression and the effects of chemotherapy and host factors is crucial to refining cancer treatment strategies.

Macrophages also promote tumor cell migration and invasion through ECM remodeling. The ECM of non-cancerous soft tissue is characterized by “curly,” non-dense collagen fibers that lay parallel to the epithelium (67). This soft ECM is involved in maintaining an epithelial phenotype, and matrix stiffening has been shown to play a direct role in promoting EMT (68, 69). Indeed, clinical conditions characterized by a stiff ECM – including cirrhosis of the liver (70), pulmonary fibrosis (71), and mammographically dense breast tissue (72) – are associated with a higher incidence of cancer in the respective tissues. In tumors, collagen deposition increases, and fibers become stiff, cross-linked and linearized – a process known as desmoplasia that has been associated with immune evasion and metastasis (67, 73–76). Indeed, tumors have been shown to be stiffer than healthy tissue in breast (77), pancreas (78), bladder (79), and ovarian (80) cancers. The stiffened matrix of tumors promotes malignant transformation, proliferation, and invasion of tumor cells, and acts as a “highway,” guiding tumor cells towards the vasculature, where they further invade and intravasate (67, 69, 76, 77, 81–85). Recent work sheds light on the mechanistic role of TAMs in pro-tumoral matrix stiffening.

Macrophages promote matrix deposition and stiffening in both cancerous and healthy tissue (86, 87). In pancreatic cancer, macrophages foster desmoplasia indirectly by activating pancreatic stellate cells. Mechanistically, macrophages internalize and degrade surrounding collagen, which leads to an increase in inducible nitric oxide synthase (iNOS) and the production of reactive nitrogen species (RNS). RNS then activate pancreatic stellate cells leading to increased collagen deposition and desmoplasia (87).

In the desmoplastic reaction, excessive ECM deposition is followed by cross-linking, which confers increased stiffness to the TME. ECM crosslinking is mediated primarily by lysyl oxidase (LOX) and lysyl oxidase-like (LOXL) proteins, which are expressed by a variety of cells in the TME (88). In pancreatic cancer, LOXL2 expression is positively associated with tumor burden and metastasis (89). Macrophages both express their own LOXL2 and promote its expression in tumor cells (89, 90). Alonso-Nocelo et al. recently demonstrated that macrophage depletion leads to a significant decrease in LOXL2, collagen fibril orientation, and metastasis in mice, indicating that macrophages promote matrix stiffness (89). In a positive feedback loop, the stiffened matrix then promotes macrophage infiltration and M2 polarization (89). Mechanistically, macrophages increase matrix stiffness by secreting oncostatin M (OSM), which upregulates LOXL2 in tumor cells (89). In turn, the stiffened matrix activates integrin β5 in macrophages, leading to FAK/MEK/ERK activation and LOXL2 upregulation, further supporting ECM crosslinking in the TME (90). In addition to promoting tumor cell migration, stiffened matrices cause macrophages to take on a more immunosuppressive phenotype (89, 91). Indeed, macrophages cultured on stiff matrices recruit cytotoxic T cells less efficiently than those cultured on softer matrices (91).

Together, these studies indicate that macrophages support the development of a stiff ECM through direct and indirect mechanisms. In turn, the stiff ECM promotes macrophage recruitment, M2 polarization, tumor cell migration, and metastasis.

Equally as important as matrix stiffness for cancer progression is matrix degradation, which is mediated by matrix metalloproteinases (MMPs). MMPs are a group of zinc-containing proteolytic enzymes responsible for degrading the extracellular matrix (92). MMPs are upregulated in nearly every type of cancer, and their activity has been shown to facilitate angiogenesis, tumor cell immune evasion, migration, and metastasis (92, 93). MMPs are expressed by a variety of stromal, immune, and tumor cells, and a growing body of evidence reveals the role of MMPs in the dynamic pro-metastatic interplay between macrophages and tumor cells.

MMP production can be induced through several major signal transduction pathways including STAT3, ERK, and NF-κB (94–102). Evidence shows that macrophages provide multiple ligands for these pathways that cooperate to promote MMP expression. For example, macrophages secrete AEG-1, TGF-β, and IL-6, which all increase MMP-9 expression in tumor cells by activating STAT3 (94, 95, 97). Indeed, inhibiting STAT3 in tumor cells, or its activators in macrophages, causes a significant decrease in MMP expression and migration in tumor cells (95, 97). Macrophages also secrete TNF-α and IL-1β, which activate the NF-κB pathway. Yamanaka et al. found that IL-1β activates NF-κB in gastric cancer cells, and this leads to increased MMP-9 expression and tumor cell invasion (103). Furthermore, tumor cells cultured in M2 macrophage-conditioned media express significantly increased levels of MMP-9 (98). This effect can be seen to a lesser (but still significant) extent when tumor cells are stimulated with TNF-α alone, suggesting that macrophages provide multiple ligands that stimulate MMP production (98).

MMPs expressed by macrophages also play a significant role in tumor cell invasion and metastasis. Macrophage – but not tumor cell – expression of MMP-11 is a negative prognostic marker in breast cancer (104). MMP-11-overexpressing macrophages secrete increased amounts of CCL2. CCL2 then activates MAPK signaling in tumor cells and increases tumor cell migration and MMP-9 expression (104). In Wilms’ tumor and gastric cancer, MMP-9 is upregulated in M2 macrophages, and MMP-9 initiates EMT and increases tumor cell invasion (105, 106). Mechanistically, macrophage-derived MMP-9 activates the PI3K/AKT pathway in tumor cells leading to the upregulation of SNAIL and subsequent EMT (105, 106). These studies identify MMP-9 as a promising therapeutic target. Indeed, MMP-9 inhibition increased the efficacy of chemotherapy and decreased metastasis to the lungs in a mouse model of gastric cancer (106). Together, these studies identify macrophages as important regulators of tumor cell MMP production.

Paclitaxel – a microtubule stabilizing agent in the taxane group – has been shown to increase the immune response and tumoricidal activity of murine macrophages. In mice, paclitaxel treatment leads to a robust increase in macrophage expression of TNFα and IL-1β, pro-inflammatory cytokines associated with the M1 phenotype (147–149). Paclitaxel also increases macrophage expression of IL-12, a Th1-type cytokine involved in activating the innate and adaptive immune response (148, 150). As a LPS mimetic, paclitaxel activates toll-like receptor (TLR) 4 on murine macrophages leading to NF-κB activation and increased production of pro-inflammatory signals (148). Though some studies show that paclitaxel also activates TLR4 in human cells (151–153), others show that species-specific differences in the TLR4 accessory protein, myeloid differentiation factor 2 (MD-2), do not allow this activation (154–159). Interestingly, some studies indicate that docetaxel – another taxane – has more potent effects on human macrophages than paclitaxel. Millrud et al. found that docetaxel, but not paclitaxel, promoted an M1 phenotype in human monocytes (160). Furthermore, a clinical study assessing immune responses to taxanes in breast cancer patients showed that, while both docetaxel and paclitaxel lead to an increase in serum M1-associated markers (including IL-6, GM-CSF, and IFN-γ), the effects were significantly more pronounced in patients who received docetaxel (161). The effects of taxanes on macrophages are also highly context dependent. For instance, IFN-γ has been shown to “prime” macrophages for tumoricidal activity, and paclitaxel affords macrophages increased cytotoxicity after macrophage exposure to IFN-γ (162). While the exact mechanisms have yet to be elucidated, these studies indicate that taxanes can promote anti-tumor M1 polarization in a context-dependent manner.

In addition to taxanes, platinum-based drugs, antimetabolites, and alkylating agents have also been shown to promote M1 macrophage polarization. The combination of platinum-based agents with antimetabolites is a common first-line treatment for gastric cancer (163). In studies analyzing the TME of matched pre- and post-treatment biopsies from gastric cancer patients, post-treatment samples harbored significantly more M1-polarized macrophages, and this increase was associated with a favorable response to treatment (164, 165). Furthermore, when given at high doses, the alkylating agent cyclophosphamide is highly immunosuppressive. However, lower doses of the drug strikingly improve anti-tumor immunity (166). This has led to the development and use of metronomic schedules of administration, in which low doses of the drug are administered more frequently (166). In line with observations that low-dose cyclophosphamide improves anti-tumor immunity, several studies show that low-dose, metronomic cyclophosphamide increases macrophage M1 polarization and decreases tumor burden (167–169).

Leukemia and lymphoma are often treated with monoclonal antibodies. While these treatments are largely effective at targeting cancer cells in many niches, cancer cells often become resistant to such antibodies in the bone marrow (170–172). Several studies found that combining antibody therapy with cyclophosphamide prevented antibody therapy resistance in the bone marrow in part by promoting macrophage phagocytosis of antibody-targeted cancer cells (170–172).

In summary, these studies show that in some circumstances, chemotherapy reprograms macrophages to increase anti-tumor activity.

The macrophage response to chemotherapy is a double-edged sword. While some studies show that chemotherapy promotes the anti-tumor activity of macrophages, others show that chemotherapy causes a macrophage-mediated pro-tumoral response.

Chemotherapy causes tumor cell death and tissue damage followed by a cytokine storm that promotes the release of endothelial and immune progenitor cells from the bone marrow (173–175). In response to this tissue damage, cells in the TME initiate a wound healing response by increasing their secretion of CSF1, CXCL12, and other chemokines that recruit these circulating progenitor cells to the tumor (176, 177). One result of this response is that perivascular TAMs increase following chemotherapy (133, 177). These newly recruited perivascular TAMs express high levels of VEGFA and the angiopoietin receptor Tie2, which have been shown to promote relapse and metastasis following chemotherapy (132, 133, 177–180). Several studies show how chemotherapy induces a macrophage-mediated pro-tumoral effect that can be abrogated by targeting macrophages.

Hughes et al. used mouse models of breast cancer to show that treatment with cyclophosphamide causes an increase in CXCR4-expressing perivascular macrophages, which promote tumor revascularization and regrowth via VEGFA signaling (177). Blocking CXCR4 signaling prevents the accumulation of perivascular macrophages and delays tumor regrowth (177).

In neuroblastoma, chemotherapy leads to the selective expansion of CCL2-expressing mesenchymal-like tumor cells and macrophage infiltration in patients, which promotes relapse and chemo-resistance (181). In mouse models, combining chemotherapy with CSF1R inhibition prevents macrophage infiltration and tumor regrowth (181).

Furthermore, while chemotherapy increases the infiltration of Tie2+ macrophages, Tie2 inhibitors have been shown to work synergistically with chemotherapy to delay tumor growth (145) and relapse (182).

In addition to promoting tumor relapse, macrophages can also increase tumor cell dissemination following chemotherapy. We have previously shown that treatment with paclitaxel causes a robust, macrophage-dependent increase in MenaINV expression, which promotes tumor cell migration, intravasation, and metastasis (130, 133). This indicates that paclitaxel causes a macrophage-mediated increase in metastasis-competent tumor cells, though the exact mechanism behind this effect remains unknown.

Furthermore, chemotherapy significantly increases the assembly and function of TMEM doorways, which are portals for tumor cell intravasation. Indeed, circulating tumor cells and lung metastases are more prevalent in mice treated with paclitaxel compared to vehicle control (133). Concerningly, TMEM doorway assembly is also increased in patients with ER+/HER2- breast cancer following neoadjuvant chemotherapy (133), thus increasing their risk of distant metastasis (136). As mentioned in Section 5, targeting TMEM doorway-associated macrophages with the Tie2 inhibitor rebastinib dramatically decreases the pro-metastatic effects of chemotherapy in pre-clinical studies, indicating that Tie2 inhibition in combination with a cytotoxic agent may improve patient outcomes (133, 145).

Another mechanism by which TAMs promote metastasis in response to chemotherapy is by upregulating the enzyme heparanase. Heparanase cleaves heparan sulfate, which is an important structural component of the ECM (183). Similarly to MMPs, this matrix-degrading enzyme is upregulated in many cancers and correlates with increased metastasis and poor prognosis (184). Unfortunately, heparanase has been shown to increase in some patients following chemotherapy (185). Mechanistically, treatment with chemotherapy leads to an increase in VEGFR3-expressing TAMs which secrete cathepsins that activate heparanase and promote ECM remodeling, lymphangiogenesis, and metastasis (186). Notably, blocking VEGFC/VEGFR3 signaling inhibits chemotherapy-induced lymphangiogenesis and metastasis (186).

In summary, chemotherapy can act on macrophages to promote relapse and metastasis in a variety of ways. Recent pre-clinical studies show that targeting macrophage recruitment or function is a promising approach to optimize cancer treatment. Indeed, there has been a 3-fold increase in clinical trials on macrophage-targeted therapies in the past 10 years (31). However, due to the diversity of patients, chemotherapies, and macrophage phenotypes, more research is needed to clarify the exact mechanisms of chemotherapy-induced cancer progression to refine treatment strategies and determine biomarkers that can identify good – and bad – candidates for different treatments.

Evidence supporting the role of a healthy diet in cancer prevention and treatment is ever-increasing (187, 188). While natural compounds are known to have a profound role in regulating EMT in cancer cells (189, 190), numerous recent studies have also shed light on how dietary agents and natural compounds target tumor-associated macrophages.

Obesity is a fast-growing global health crisis that is associated with an increased risk of cancer, as well as general morbidity and mortality (209, 210). In fact, women with obesity who are diagnosed with breast cancer have an increased risk of distant metastasis and are less likely to respond to some cancer treatments (211–214). Unsurprisingly, adipose tissue macrophages are thought to play a key role in creating a pro-tumorigenic microenvironment (86). CCL2 expression is significantly increased in the adipose tissue of obese compared to lean mice (215) which leads to the accumulation of macrophages. Indeed, it is estimated that macrophages make up <10% of cells in the adipose tissue of lean individuals and nearly 40% of cells in the adipose tissue of obese individuals (216). These recruited macrophages surround dead and dying adipocytes, forming crown-like structures (CLS) that are characteristic of adipose tissue inflammation (217). Overweight and obese patients (BMI ≥ 25 kg/m²) with breast cancer are more likely to have CLS compared to patients at a healthy weight (BMI < 25 kg/m²), and BMI ≥ 25 kg/m² is associated with a shift in CLS macrophage phenotype that may be indicative of metabolic dysfunction and poor treatment outcomes under some conditions (218). These macrophages also interact with pre-adipocytes and prevent their differentiation, instead causing them to take on a fibroblastic phenotype and enhance the synthesis and deposition of ECM components (219, 220). This indicates that macrophages cause increased ECM density in obese adipose tissue, including in the breast where ECM density is a significant risk factor for cancer (72, 221). Further implicating adipose tissue macrophages in breast tumor development, a study on transgenic mice that overexpress CCL2 in the mammary epithelium showed that CCL2 overexpression causes increased macrophage density, stromal density, and ECM crosslinking enzyme LOX compared to non-transgenic controls (222). CCL2-overexpressing mice also had an increased susceptibility to DMBA-induced mammary tumors, demonstrating a relationship between macrophages, ECM density, and cancer risk (222).

Though much is still unknown about how obesity shapes cancer development, these studies suggest that obesity promotes macrophage-mediated ECM stiffening that is known to support tumorigenesis (see section 3.1).

There are widespread racial disparities in the diagnosis, treatment, and outcome of cancer patients (223–226). While some of these disparities are due to systemic racism in the medical field, some studies identify biological differences that could be contributing to this effect (227). Indeed, Black men and women with prostate and breast cancer, respectively, have significantly worse outcomes than their white counterparts even after socioeconomic and other mediating factors are accounted for (228, 229). We previously evaluated distant recurrence-free survival (DRFS) in breast cancer patients following neoadjuvant (NAC) versus adjuvant (AC) chemotherapy and found that treatment type had no impact on DRFS for white women (230). However, Black women had significantly worse DRFS when treated with NAC (230). Differences in TAMs may contribute to this. Black breast cancer patients have significantly increased macrophage infiltration and M2 polarization compared to white patients, and this is prognostic of progression-free survival (231, 232). Black, compared to white women treated with neoadjuvant chemotherapy for ER+/HER2- breast cancer also have a higher TMEM doorway score and macrophage density in the residual tumor tissue, which may contribute to poorer outcomes (223). In summary, racial disparities in cancer development and outcome may be mediated in part by macrophage infiltration and activity.