REvIEWS

Macrophages as regulators of tumour

immunity and immunotherapy
David G. DeNardo   1* and Brian Ruffell2*
Abstract | Macrophages are critical mediators of tissue homeostasis, with tumours distorting this
proclivity to stimulate proliferation, angiogenesis and metastasis. This had led to an interest in
targeting macrophages in cancer, and preclinical studies have demonstrated efficacy across
therapeutic modalities and tumour types. Much of the observed efficacy can be traced to the
suppressive capacity of macrophages, driven by microenvironmental cues such as hypoxia and
fibrosis. As a result, tumour macrophages display an ability to suppress T cell recruitment and
function as well as to regulate other aspects of tumour immunity. With the increasing impact
of cancer immunotherapy , macrophage targeting is now being evaluated in this context.
Here, we discuss the results of clinical trials and the future of combinatorial immunotherapy.

Tumour microenvironment
The presence of tumour-associated macrophages by input from three distinct elements: developmental
(TME). The cellular and (TAMs) is generally associated with a poor prognosis origin, tissue of residence and acute microenvironmen-
acellular components in which in solid tumours. This has been shown in studies per- tal cues (Fig. 1). The diversity of macrophage functions
malignant cells reside. These formed on individual tumour types using traditional is regulated in turn by the integration of the epigenetic
include surrounding blood
immunohistochemistry techniques to quantify cellu- memory of these cells and their plasticity to respond to
vessels, immune cells,
fibroblasts, extracellular matrix lar density1,2 and in more recent analyses that infer the new cues13–16. The extent to which macrophages regulate
components and extracellular presence of macrophages across malignancies using gene tumour growth is therefore critically linked to properties
signalling molecules such as expression profiles3. These findings are consistent with of the tumour itself. This includes a role for malignant
chemokines, cytokines and
the established role of macrophages in promoting mul- cell-derived factors such as colony-stimulating factor 1
growth factors, as well as
metabolic regulators such
tiple aspects of tumorigenesis in experimental models, (CSF1; also known as M-CSF) and CC-chemokine ligand
as oxygen. from initiation through to angiogenesis and systemic 2 (CCL2) in promoting macrophage recruitment; how-
dissemination4,5. Most relevant for patients, TAMs ever, the elements within the tumour microenvironment
Tumour immune are known to suppress responses to standard-of-care (TME) and tumour immune microenvironment (TIME),
microenvironment
therapeutics, including chemotherapy, irradiation and such as fibrosis, hypoxia, nutrient availability and
(TIME). The components of the
tumour microenvironment angiogenic inhibitors6–9. Although this includes direct lymphocyte-derived factors, appear to most dramati-
represented by leukocytes or regulation of survival and cell death pathways in tumour cally shift macrophage phenotypes (Fig. 2). Before dis-
their derived factors. cells10,11, in vivo modelling indicates that improved effi- cussing these factors, it is important to note that most
cacy following macrophage depletion is often depend- of the available data are contextualized within the
ent upon enhanced recruitment or function of cytotoxic binary M1–M2 macrophage polarization system. Thus,
CD8+ T cells6. Perhaps not surprisingly, macrophage macrophages have traditionally been considered anti-
antagonists demonstrate combinatorial efficacy when tumorigenic when they express high levels of tumour
1
Department of Medicine, combined with immunotherapy, including checkpoint necrosis factor (TNF), inducible nitric oxide synthase
ICCE Institute, Department of blockade12. Clinical trials examining these combinations (iNOS; also known as NOS2) or MHC class II mole-
Pathology and Immunology, are now ongoing. In this Review, we discuss how macro­ cules and pro-tumorigenic when they express high
Siteman Cancer Center,
Washington University in
phages are induced into becoming immunosuppressive, levels of arginase 1 (ARG1), IL-10, CD163, CD204 or
St. Louis, School of Medicine, the mechanisms by which they suppress antitumour CD206 (ref.17). Changes to any of these markers were
St. Louis, MO, USA. immunity and how this information is being utilized to then used to conclude that macrophage repolarization
2
Department of Immunology, develop therapeutics and design clinical trials. has occurred. However, it is now clear that macrophage
Department of Breast activation states consist of a continuum of phenotypes,
Oncology, H. Lee Moffitt Factors regulating macrophage function and the use of markers to delineate their functional role
Cancer Center, Tampa,
FL, USA.
Macrophages are not a single cell population with a within the tumour is circumspect18. In the following
defined phenotype and biological activity but rather are sections, we therefore highlight studies that demon-
*e-mail: [email protected];
[email protected] a diverse collection of cell types with a wide range of strate a change in macrophage phenotype and function
https://doi.org/10.1038/ functional roles in homeostatic and pathological con- in vivo. Although not discussed here, it should also be
s41577-019-0127-6 ditions. This diversity of cellular activities is regulated acknowledged that factors such as anatomical location,

Nature Reviews | Immunology

Reviews

Embryonic progenitor Adult HSC

progenitor

Origin

Embryonic
macrophage Monocyte
pool

Tissue-resident macrophage pool

Homeostatic
tissue

Neoplastic
tissue

Phenotypic
responses

Tissue remodelling Immune regulation

Fig. 1 | macrophage origin and polarization state. Tissue macrophages are derived from embryonic or adult
haematopoietic stem cell (HSC) progenitor cells under homeostatic conditions, with the relative contribution of these
populations varying by tissue. Monocyte-derived cells also contribute to the macrophage population in some tissues but
are mostly associated with a response to inflammatory conditions, including cancer. The combination of their
developmental origin and tissue of residence is thought to fine-tune the eventual response of a macrophage to polarizing
stimuli. These distinctions in both macrophage phenotypes in response to the same tumour microenvironment and
underlying origin-based epigenetics patterns are depicted by the coloured gradients.

pathological or molecular cancer subtype and even the although it is known that tissue macrophages of both
specific microenvironmental niche occupied by the cell embryonic and HSC origin assume epigenetically regu-
likely contribute to intertumoural and intratumoural lated programmes indicative of their residence in these
M1–M2 macrophage macrophage heterogeneity (Box 1). Thus, the role of tissues (for example, brain, liver and lung) to drive spe-
polarization macrophages in cancer types can differ considerably6, cific phenotypes, especially those related to metabolism
M1 and M2 are classifications and there may even be an underappreciated level of and interferon responsiveness28–30.
historically used to define
macrophages activated in vitro
­variability between individual patients. In tumours, however, several recent studies have sug-
as pro-inflammatory (when gested that embryonic-derived macrophages may have
classically activated with IFNγ Macrophage origin. Macrophages arise from three dis- distinct phenotypes and functions compared with their
and lipopolysaccharide) or tinct developmental pathways (Fig. 1). A large proportion monocyte-derived counterparts28,31,32. TAMs are usu-
anti-inflammatory (when
of tissue-resident macrophages are now recognized as ally thought to predominantly derive from circulating
alternatively activated with IL-4
or IL-10), respectively. originating from embryonic precursors that seed tissues monocytes33,34, but up to 50% of the macrophages in
However, in vivo macrophages in the prenatal and perinatal periods. This occurs in at murine models of brain, lung and pancreatic cancer were
are highly specialized, least two functional waves, with macrophage precursor found to derive from tissue-resident populations28,31,32.
transcriptomically dynamic cells from fetal yolk sac or fetal liver progenitors16,19. Within these tumours, the TAMs of HSC origin have
and extremely heterogeneous
with regards to their
These precursors seed distant tissue and give rise to elevated expression of genes involved in immunosup-
phenotypes and functions, locally proliferating, self-maintained tissue-resident pressive networks and antigen presentation31,32,35. By con-
which are continuously macrophages that can persist into adulthood16,19–22. trast, embryonic-derived TAM gene sets are enriched
shaped by their tissue For some tissues, such as the colon, these embry- for tissue remodelling and wound healing. These data
microenvironment. Therefore,
onic macrophages are rapidly replaced by monocytes suggest that tissue-specific and origin-specific pro-
the M1 or M2 classification is
too simplistic to explain the derived from haematopoietic stem cells (HSCs) after grammes can fine-tune macrophage responses in ways
true nature of in vivo birth. However, for other macrophage subsets, such as that may have a substantial impact on tumour immu-
macrophages, although these microglia, their sole origin appears to be embryonic, nity; however, the rules for how these programmes
terms are still often used to with little contribution from HSCs under homeostatic might integrate with macrophage polarizing cues are
indicate whether the
macrophages in question are
conditions23,24. Still other tissues contain macrophages largely unknown. Additionally, while the origins of mac-
more pro-inflammatory or with a mixed origin, including the pancreas, breast and rophages have been mapped in multiple animal models,
anti-inflammatory. lung25–28. The relevance of this remains to be determined, our ability to interrogate these populations in human

www.nature.com/nri
 Reviews

a Direct
int M
kpo et
ec ab
Ch ol
ic
ROS
CTL

PDL2
PDL1
L-arginine
B7-H4

TAM ARG1

IL-10 TGFβ

Cytokine

b Indirect
Treg cell recruitment Proteases ECM density Immune exclusion
and remodelling

ECM CAF
Treg cell
CCL22 IL-6
IL-10 CCL17 TGFβ

Nitrosylation
CCL2-N
and CCL5-N
CTL
PIGF
Pericyte
IL-12 TIE2

VEGFA
IL-10
Antigen
DC uptake
WNT7B

DC dysfunction ANG2 Vascular dysfunction

Fig. 2 | Direct and indirect regulation of tumour immunity by TAms. a | Tumour-associated macrophages (TAMs) can
directly inhibit cytotoxic T lymphocyte (CTL) responses through three distinct mechanisms. These include immune
checkpoint engagement via their expression of molecules such as programmed cell death 1 ligand 1 (PDL1), production
of inhibitory cytokines (such as IL-10 and transforming growth factor-β (TGFβ)) and their metabolic activities, including the
depletion of metabolites (such as l-arginine) and production of reactive oxygen species (ROS). Thus, TAMs can regulate
T cell function directly by multiple mechanisms. b | TAMs also inhibit T cell responses indirectly by controlling the immune
microenvironment. This includes control through TAM-mediated recruitment of immunosuppressive populations (such as
regulatory T (Treg) cells) or inhibition of stimulatory populations (such as dendritic cells (DCs)) by TAMs. TAMs also blunt
T cell recruitment via regulation of vascular structure and through their ability to exclude T cells from intratumoural
regions via regulation of the extracellular matrix (ECM) and the chemokine milieu. Thus, TAMs can regulate tumour
immunity by multiple indirect mechanisms that ultimately impact T cells function. ANG2, angiopoietin 2; ARG1,
arginase 1; CAF, cancer-associated fibroblast; CCL , CC-chemokine ligand; VEGFA , vascular endothelial growth factor A.

Nature Reviews | Immunology

Reviews

Box 1 | Tissue localization and macrophage heterogeneity IL-4-stimulated macrophages, which show preferential
oxidative phosphorylation and fatty acid oxidation42
Most of the drivers of macrophage phenotype occur within distinct and heterogeneous and are therefore not impacted by impaired glycolysis44.
microscopic areas, or niches. as reviewed elsewhere205,206, these niches exist both in Presumably, low glucose availability will thus favour
primary tumours and at disseminated sites and can include areas defined by their TAMs to adopt a pro-tumour polarization state, but this
relative proximity to cancer cell invasive fronts, tumour cell nests, fibrotic stroma
has not been formally demonstrated. However, because
and functional vasculature or even the presence of tertiary lymphoid structures.
these parameters lead to the classification of distinct macrophage populations a metabolic shift does control macrophage function,
(for example, perivascular or hypoxic macrophages) that are better defined by their interfering with the genetic regulators of this process can
functionality within these niches than by their expression of surface markers or blunt the pro-tumour bioactivity of TAMs and reduce
activation state. as a consequence, it is likely that macrophages within these unique tumour growth in animal models45–47.
niches differentially regulate T cell function. For example, perivascular macrophages As mentioned above, lactic acid promotes Vegf and
are expected to control vascular structure and T cell infiltration into the tumour Arg1 expression by macrophages in a Hif1a-dependent
parenchyma182, while hypoxic macrophages might have a more specialized role in manner41. This is not true for lactate salt41, indicative
governing T cell infiltration into tumour beds39. additionally, these niches may be of a role for acidic pH in either promoting the activity of
shaped by cancer therapy, for example, by immunotherapy increasing the frequency monocarboxylate transporters48 and/or directly regulat-
of tertiary lymphoid structures or by vascular directed-therapy reshaping perivascular
ing macrophage polarization. In support of the second
and hypoxic niches. Certainly multiple cytotoxic and targeted therapies are known to
alter the phenotype of tumour macrophages as a whole7. thus, while there are consistent scenario, a recent study found that a pH of 6.1 was suf-
drivers of macrophage function, the cumulative impact of these is highly dependent ficient to promote expression of Arg1, Vegfa and Hif1a
upon the microenvironmental niche occupied by the cells within the malignant tissue, by unstimulated macrophages in vitro49, and similar
in addition to the macroenviroment of the tumour and tissue type of residence. observations have been made at pH 6.8 during stimula-
tion with IL-4 (ref.50). Increasing the pH within tumours
similarly reduces expression of ARG1 by TAMs50.
tissues is limited. Thus, one challenge for the field going How macrophages sense pH at a molecular level is some-
forward will be to determine the impact of these findings what vague, but activation appears to be mediated by
in human cancers, perhaps with an approach such as G protein-coupled receptors and production of cAMP51
single cell RNA sequencing to permit an evaluation of leading to expression of the transcription factor induc-
macrophage heterogeneity and origin. ible cyclic AMP early repressor (ICER)49. Importantly,
mice with myeloid-specific deficiency of ICER resist the
Metabolism. Tissue hypoxia impacts macrophages in growth of highly glycolytic tumours49.
two ways. First, hypoxia can induce the production
of key monocyte recruitment factors including CCL2, Fibrosis. Desmoplasia is a hallmark of many solid
CCL5, CXC-chemokine ligand 12 (CXCL12), CSF1 and tumours, with pancreatic cancer representing one
vascular endothelial growth factor (VEGF) by tumour extreme end of the spectrum. Fibrotic stroma has the
cells and the stroma. Once recruited into hypoxic potential to shape the TAM phenotype through direct
regions, the receptors for several of these factors are effects of its components, like activated fibroblasts or
downregulated, effectively locking TAMs in hypoxic changes in the extracellular matrix (ECM), or indirect
microenvironments36. Second, macrophages directly effects on factors such as oxygen and nutrient avail-
sense hypoxic conditions via hypoxia-inducible factors ability. Cancer-associated fibroblasts (CAFs) are per-
(HIFs): the absence of HIF1α leads to reduced ARG1 haps the most relevant component of fibrosis because
expression and immunosuppressive activity in vitro37, these cells overexpress numerous pro-inflammatory
and the absence of HIF2α reduces macrophage infil- cytokines (for example, CCL2, CCL3, CCL5, IL-6,
tration and cytokine production 38. In both cases, granulocyte–macrophage colony-stimulating fac-
myeloid-specific loss of these factors significantly delays tor (GM-CSF; also known as CSF2), CSF1, VEGF
tumour progression in autochthonous tumour mod- and CXCL8) with the potential to regulate recruit-
els37,38. Notably, the localization of TAMs in hypoxic ment, differentiation and activation of TAMs 52–56.
regions is critical to the generation of an immuno­ In particular, CAFs have been reported to impair
suppressive phenotype in vivo because myeloid-specific the maturation of macrophages, locking recruited
loss of neuropilin 1 excludes macrophages from hypoxic monocytes in an immature, suppressive state. This
areas, and this promotes antitumour immunity 39. is possibly due to high levels of IL-6 production,
However, it should be noted that the roles of neuropilin especially in pancreatic CAFs, which can induce
1 and its ligand, semaphorin 3A (SEMA3A), remain signal transducer and activator of transcription 3
Desmoplasia controversial40 and that HIF1α can also be stabilized by (STAT3) phosphorylation and prevent macrophage
When associated with cancer, the presence of lactic acid41. differentiation57–59. In addition, IL-6 production by
the growth and expansion of Irrespective of tissue hypoxia, aerobic glycolysis endothelial cells has been shown to promote M2-like
fibrous and/or connective
within tumours can limit glucose availability and pro- macrophage polarization and tumour growth in a glio-
tissue surrounding the
malignant cells. Desmoplasia mote the accumulation of organic acids. Do these fac- blastoma model60, and TAMs themselves produce IL-6
may occur around a growing tors also impact macrophage function? Macrophages in multiple other model systems33,61,62. The source of
neoplasm and consists of stimulated with lipopolysaccharide and/or IFNγ display these polarizing cytokines may therefore vary consid-
expansion of the non-­malignant enhanced glucose uptake and aerobic glycolysis42, and erably across tumour types or even within microen-
cellular components, such as
activated fibroblasts, beyond
consistent with the important role for this metabolic vironments of the tumour. Adding to this complexity
the norms of the homeostatic shift, increasing glucose transport promotes expres- is the diversity of CAF subsets and their differential
tissue levels. sion of reactive oxygen species43. This is in contrast to potential to alter immune function63,64. Thus, although

www.nature.com/nri
 Reviews

CAFs are assumed to be important regulators of TAM stromal compartment leads to macrophage repolariza-
function, their role remains poorly defined in vivo. tion and T cell-dependent growth restraint85, as well as
The extensive ECM associated with fibrosis can also a prolonged response to radiation therapy86. Similarly,
impact macrophages in several ways. Periostin has been the deletion of protein S in tumour cells, which medi-
shown to promote TAM recruitment via integrin bind- ates the recognition of phosphatidylserine-presenting
ing65, while collagen promotes an M2-like phenotype66–68, membranes by MERTK and the tyrosine protein kinase
with the increased presence of collagen I acting as a res- receptor TYRO3, increases iNOS expression and leuko­
ervoir for secreted factors, such as transforming growth cyte infiltration87. It remains possible that activation
factor-β (TGFβ), or directly activating inhibitory recep- of select pattern recognition receptors may shift macro­
tors such as leukocyte-associated immunoglobulin-like phages towards an antitumour phenotype, especially
receptor 1 (LAIR1)69. In addition to the concentration when using agonists therapeutically. However, it gen-
of ECM components, their macromolecular structure, erally appears that recognition of cellular and extra­
including biophysical and mechanical properties, can cellular debris by macrophages is detrimental in terms
direct macrophage function within tumours. Collagen of tumour growth.
crosslinking and matrix stiffness are properties of solid
tumours that have direct effects on macrophage differ- Lymphocytes. Most lymphocyte subpopulations can be
entiation, motility and phenotype70,71. This is largely identified within the tumour stroma, with the compo-
mediated by β1 integrin clustering and hyperactiva- sition and density varying by tissue, molecular subtype
tion of focal adhesion kinases (FAKs), and genetic or and individual patient. In addition to directly impacting
pharmacological loss of FAK signalling reduces tumour tumour growth, many of these populations have been
fibrosis72 and the number of M2-like TAMs73,74, lead- shown to shape the phenotype of macrophages in some
ing to increased antitumour immunity and improved fashion88 (Fig. 2). IFNγ can be produced by CD8+ cyto-
responses to immune checkpoint blockade73. toxic T lymphocytes, T helper 1 (TH1) cells and natural
Other ECM components known to regulate mac- killer (NK) cells and primes macrophages towards a
rophage activation include hyaluronan, versican and classically activated phenotype, increasing macrophage
tenascin, and this mostly occurs through altered forms of antigen presentation, pro-inflammatory cytokine pro-
these molecules being recognized as damage-associated duction and direct tumour cell killing 89. Similarly,
molecular patterns (DAMPs) by Toll-like receptor 2 expression of CD40 ligand (CD40L) by T cells can
(TLR2) or TLR4 (ref. 75) . Versican acts via TLR2 to activate monocytes via CD40 to increase their expres-
increase expression of receptors for IL-6 and IL-10 and sion of MHC class II, iNOS and TNF. This induction
sensitize cells to these cytokines76, and in tumour mod- of an antitumour phenotype is needed during acute
els, versican has been suggested to promote metastasis T cell responses following therapeutic intervention90,91
via macrophage activation77. Meanwhile, hyaluronan and could presumably promote a cycle of antitumour
production by fibroblasts or keratinocytes promotes ­immunity if maintained.
macrophage recruitment78,79, possibly through its role in Unfortunately, many lymphocyte-derived factors
providing a scaffold for ECM proteoglycans and enhanc- engage the tumour-promoting activities of TAMs. This
ing chemokine retention. Hyaluronan has also been includes lymphocyte production of IL-4 and IL-13 by
shown to alter macrophage activation in vitro through TH2 cells, which enhance epidermal growth factor
CD44 or TLR2 and/or TLR4, depending on the state of expression by TAMs to foster cancer cell metastasis92, as
the cells and the molecular mass of the hyaluronan80. well as the suppressive activity of TAMs to blunt CD8+
T cell responses to chemotherapy and radiation therapy93.
Cellular debris. Cell death is prevalent within tumours, In addition, there is a role for forkhead box P3 (FOXP3)+
particularly regions of hypoxia, and is substantially regulatory T (Treg) cells in driving TAMs towards an
induced by anticancer therapies. Whereas the release of immunosuppressive phenotype marked by production of
intracellular DAMPs can promote tumour immunity IL-10 and expression of B7-H4 (also known as B7S1)94,95.
through activation of dendritic cells (DCs)81, the chronic Although the available data suggest that neutrophils are
stimulation of macrophages induces negative regulatory the major effector cells associated with IL-17-driven
mechanisms to dampen inflammation. Thus, although responses96, IL-17 produced by TH17 cells or γδ T cells
the release of high mobility group protein B1 (HMGB1) can increase monocyte recruitment and macrophage
in response to chemotherapy promotes immunity via activation, and this could possibly lead to immune tol-
TLR4, it can also drive IL-10 expression in TAMs via the erance when macrophages engulf recruited neutrophils
receptor for advanced glycation end products (RAGE; undergoing apoptosis in the tissue97–99. Finally, the loss
also known as AGER)82. Whether this is true for other of B cells in murine models inhibits tumour progression,
DAMPs in tumours is unknown, but even the expres- and in squamous cell carcinoma, this has been traced to
sion of pro-inflammatory cytokines such as IL-1β, IL-6 the pathogenic production of autoantibodies, which act
and TNF by macrophages can promote tumour progres- via crystallizable fragment-γ (Fcγ) receptor signalling
sion6. Macrophage recognition of apoptotic cells is also to promote macrophage-dependent angiogenesis and
well known to suppress their activation potential83, with tumour progression100,101. A similar phenotype has been
activation of the Mer tyrosine protein kinase (MERTK) observed in pancreatic tumours, and, in both cases, B cell
receptor elevating expression of immunosuppres- depletion or inhibition of Fcγ receptor signalling
sive factors such as TGFβ, IL-10 and ARG1 (refs84,85). (via BTK, SYK or phosphoinositide 3-kinase-γ (PI3Kγ))
Consistent with this observation, loss of MERTK in the relieves macrophage-mediated T  cell dysfunction,

Nature Reviews | Immunology

Reviews

leading to delayed progression or improved responses due to iNOS-expressing macrophages inducing vascu-
to chemotherapy and programmed cell death 1 (PD1) lar cell adhesion molecule 1 (VCAM1) expression by
blockade102–105. the tumour endothelium, leading to enhanced recruit-
ment of adoptively transferred CD8+ T cells91. Thus,
Regulation of T cell function by TAMs the impact of iNOS expression by macrophages may
Macrophages are widely acknowledged as one of the cen- be highly context-dependent, promoting or hindering
tral suppressive populations within tumours, and deplet- T cell responses under different therapeutic conditions.
ing these cells can unleash T cell responses under several Despite the results in murine models, l-arginine
therapeutic conditions6–8. Underlying this functional metabolism by macrophages has not been implicated
role are molecular mechanisms that range from nutrient in regulating T cell responses in humans; ARG1 is
depletion to recruitment of Treg cells, although the extent expressed by human granulocytes instead of mac-
to which these mechanisms are involved in any particu- rophages94,117,118, and even combining ARG1 and iNOS
lar tumour is less well defined. Here, we describe what inhibitors fails to blunt macrophage suppressive activ-
is known about the ability of macrophages to directly or ity94. Instead, the available data point to expression of
indirectly suppress T cell responses within tumours, as immune checkpoint ligands as being necessary for
well as discuss several theoretical concepts that may be suppression. Programmed cell death 1 ligand 1 (PDL1)
applicable to cancer (Fig. 3). expression by CD14+ or CD68+ macrophages is observed
in multiple cancer tissues, including hepatocellular carci-
Direct regulation of T cells by TAMs. Numerous studies noma, melanoma and ovarian cancer119,120, and positive
have shown that TAMs suppress naive T cell prolifera- correlations have been noted specifically between PDL1
tion ex vivo, indicating that macrophages can directly expression by macrophages and response to PD1 block-
suppress T cell function33,37,106–108. This is often thought ade119. Induced expression of PDL1 on monocytes also
to relate to the ability of murine tumour macrophages to suppresses activation of tumour-specific T cells in vitro
metabolize l-arginine via expression of ARG1 (ref.109), and in vivo following adoptive transfer120. Similarly,
a common marker of M2-like polarization in murine PDL1 blockade depends upon target expression by
macrophages, and l-arginine is necessary for T cell fit- CD11b+ myeloid cells in murine models119,121. However,
ness and antitumour activity110. However, neither ARG1 it should be noted that PDL1 expression by macrophages
inhibition nor Arg1 deficiency impact macrophage sup- has not been established as an independent predictor of
pressive capacity in vitro90,108. Instead, a secondary role response, and the relative contribution of PDL1 expres-
for ARG1 is observed only when iNOS is inhibited or sion by tumour cells or the various stromal populations
absent33,90,108. These observations may be a by-product is likely to be highly variable between patients.
of the supraphysiological concentration of l-arginine The other immune checkpoint ligand that has been
in cell culture medium because tumour macrophages identified as functionally important in human mac-
can deplete l-arginine and prevent recovery of CD3ε rophages is B7-H4. Although the receptor for B7-H4 is
expression by T cells following stimulation111. Inhibition unknown, B7-H4-expressing cells and B7-H4–immuno­
of ARG1 in vivo also reduces the growth of tumours only globulin fusion proteins suppress IL-2 production and
in immune competent mice111, and Arg1-deficiency in T cell proliferation122. The fusion protein has also been
macrophages improves responses to adoptive cell used to identify expression of the receptor in T cells
transfer (ACT) therapy90. However, it remains unclear within healthy livers and hepatocellular carcinomas123.
whether these in vivo observations are the result of Importantly, B7-H4 is preferentially expressed by CD14+
direct immune suppression, the effects of polyamines macrophages within ovarian and liver tumours, and
on tumour cell proliferation112 or enhanced production inducing B7-H4 expression on human monocytes and/or
of nitric oxide90. macrophages confers suppressive capacity in vitro94,123.
The importance of iNOS expression by tumour Blocking B7-H4 also suppresses the growth of subcuta-
macrophages is less opaque, at least ex vivo 33,90,108. neously implanted tumours in mice by reducing CD8+
Nos2-deficient tumour myeloid cells lose substantial T cell exhaustion123; however, other groups have found
suppressive capacity in co-culture assays90, and iNOS that B7-H4 promotes or has no impact on tumour
inhibition restores T cell proliferation in the presence immunity in murine models of breast and prostate can-
of macrophages33. In addition to potential direct effects of cer, respectively124,125. These differences may relate to the
nitric oxide on T cells113, this may be due to secondary specifics of the tumour model examined, especially as
production of peroxynitrites, which can prevent the B7-H4 expression is regulated by cytokines common in
interaction of the T cell receptor with MHC through the TME such as IL-6, IL-10 and IFNγ94,125.
nitration of either protein114–116. In vivo, scavenging Several other molecules expressed by macro­
peroxynitrites with uric acid improves T cell activation phages potentially have direct suppressive effects on
and enhances response to a tumour vaccine116, although tumour-infiltrating T cells. For example, tumour macro­
whether this is attributed to expression by macrophages, phages can be an important source of IL-10 (ref.107),
other myeloid subsets or the combination may vary by and IL-10 is known to suppress CD8+ T cell stimulation
the tumour model. On the basis of these observations, it by increasing N-glycan branching, thereby reducing
is surprising that iNOS expression by myeloid cells has colocalization of CD8 protein with the T cell receptor126.
also been implicated in promoting a T cell response90,91. This process requires the presence of galectin 3, and
Nos2 deficiency diminishes the efficacy of ACT90, and, interfering with this association restores IFNγ expres-
at least in the context of low-dose irradiation, this is sion by CD8+ T cells from human ovarian ascites127.

www.nature.com/nri
 Reviews

a Regulator of TAM phenotype

TIME drivers TME drivers
B cell ECM

Hypoxia
Immunoglobulins
TH2 cell NKT
cell IL-4
IL-13

IL-6 CAFs
IL-10
TGFβ TAM GM-CSF
Treg cell CSF1

CCL2
MDSC IFNγ
TNF TLR
Tumour cell
HMGB1
NK CTL TH1 cell
cell
Cell death

b Integrated view

Immune function Immune dysfunction

• Functional vasculature • Dysfunctional vasculature
• Normoxic • High hypoxia
• Lower ECM density • High ECM density
• TH1 cells > TH2 cells • TH2 cells >TH1 cells
• CTL infiltration • CTL exclusion

Immune-suppressive and
Antitumour TAMs
tissue-remodelling TAMs

• Hypoxia
• Fibrosis
• Immune suppression

Fig. 3 | Cell type versus integrated views on drivers of TAm phenotype. a | Tumour-associated macrophage (TAM)
phenotype is driven by a combination of the tumour microenvironment (TME) and the tumour immune microenvironment
(TIME). On the left, responses by adaptive and innate immune cells provide cytokines and other factors that regulate
macrophage bioactivities. On the right, properties of the TME like hypoxia, fibrosis and cellular stress also tailor the
phenotype of TAMs. b | Both immune-related and non-immune-related factors integrate to drive functional or
dysfunctional antitumour immunity. On the left, the presence of a robust adaptive immune response is concomitant with
limited tissue pathology and macrophages programmed to drive inflammation. On the right, tumour hypoxia and fibrosis
are integrated with high numbers of cancer-associated fibroblasts (CAFs) and immunosuppressive cell infiltration, and
macrophages are programmed to drive immune suppression and tissue remodelling, leading to cytotoxic T lymphocyte
(CTL) exclusion and/or suppression. CCL2, CC-chemokine ligand 2; CSF1, colony-stimulating factor 1; ECM, extracellular
matrix; GM-CSF, granulocyte–macrophage colony-stimulating factor ; HMGB1, high mobility group protein B1;
MDSC, myeloid-derived suppressor cell; NK cell, natural killer cell; NKT cell, natural killer T cell; TGFβ, transforming
growth factor-β; TH cell, T helper cell; TLR , Toll-like receptor ; TNF, tumour necrosis factor ; Treg cell, regulatory T cell.

As macrophages can be an important source of galec- of regulation via protein–carbohydrate interactions.

tin 3 in inflamed tissues128, tumour macrophages may This has not been extensively evaluated, but it is worth
even regulate both aspects of this suppressive pathway. noting that the mannose receptor, CD206, which is
Macrophages actually express a variety of C-type and highly expressed by tumour macrophages, can impair
I-type lectins, which could represent an additional layer the cytotoxicity of CD8+ T cells by suppressing CD45

Nature Reviews | Immunology

Reviews

phosphatase activity129. Given the importance of carbo- are chemotactic for CD8+ T cells ex vivo102 and have
hydrate modifications in T cell signalling, lectin expres- been described to highly express the T cell-attracting
sion by macrophages may have an unappreciated role in chemokine CXCL10 (refs33,143,144), it is conceivable that
controlling antitumour immunity. they could retain T cells within the stroma or perivascu-
lar regions of the tumour. Macrophages and T cells are
Indirect regulation of T cells by TAMs. Extending predominantly found within these regions, and at least
from their function in maintaining tissue homeostasis, one study has shown that CSF1R inhibition increases
macrophages are important regulators of multiple T cell motility and localization in tumour islets145.
aspects of the TME, in particular, the structure of the Finally, macrophages have been shown to promote
vasculature and ECM6. Do tumour macrophages reg- Treg cell recruitment to human ovarian carcinomas
ulate T cell recruitment? Several studies have reported via CCL22 production146. In addition to, presumably,
that inhibiting macrophage recruitment by targeting suppressing cytotoxic T cell responses, Treg cells from
the CSF1–CSF1 receptor (CSF1R) pathway improves ovarian ascites stimulate IL-6 and IL-10 expression by
T cell infiltration, including during chemotherapy or macrophages and increase B7-H4 expression in an auto-
high-dose irradiation12,61,93,106,107. Similar results are crine fashion in vitro94,95. This ability of macrophages to
observed when blocking macrophage recruitment suppress T cell responses via regulation of intermediate
through CC-chemokine receptor 2 (CCR2) inhibition61. cell types has also been described in mammary carci-
Assuming the increase in T cell numbers is not due to noma models. In this case, macrophage production of
changes in cell death or proliferation, the mechanisms IL-10 suppresses IL-12 expression by CD103+ conven-
underlying these observations are unclear. tional dendritic cells (cDCs) in tumours, resulting in a
One possibility could be through regulation of diminished CD8+ T cell response during chemother-
vascular adhesion molecules, but iNOS-expressing apy107. Together, these data highlight macrophages as
macrophages actually promote VCAM1 expression central drivers of the immunosuppressive TME through
during low-dose irradiation91, and while perivascu- their ability to regulate the recruitment and the function
lar macrophages regulate vascular structure through of multiple immune subtypes.
expression of VEGFA130–132, this would be expected to
exacerbate vessel leakiness and increase the expression Phagocytosis and antigen presentation. Consistent
of adhesion molecules. Another possible mechanism by with their name, several studies have confirmed that
which macrophages could suppress T cell recruitment macrophages are the main phagocytic population within
is through production of peroxynitrites and the subse- tumours143,144,147. However, tumour macrophages do not
quent nitration of CCL2 or CCL5 (refs133,134). Nitration express CCR7 and are unable to migrate into the draining
of CCL2 specifically has been shown to reduce accumu- lymph nodes147,148. They also display somewhat limited
lation of T cells within subcutaneous tumours135. This ability to activate or restimulate CD8+ T cells144. What
appears to be due to low CCL2 responsiveness by T cells, impact could phagocytosis by macrophages have on the
resulting in nitration preventing chemotaxis of human induction of antitumour immunity? First, it is possible
and mouse T cells without impacting monocyte migra- that macrophages redirect antigens away from cDCs, an
tion135. Although this does not explain the increased observation that might not be apparent in experiments
T cell recruitment following CCR2 inhibition61, it is conducted using highly expressed model antigens (for
possible that nitration of CCL5 has a similar effect, espe- example, ovalbumin). Second, clearance of apoptotic
cially given its more important role in promoting T cell cells and debris might be expected to reduce the presence
recruitment into tumours136. of alarmins or DAMPs. Third, tissue-resident macro­
Rather than recruitment, it may be that macrophages phages are known to suppress their own activation in
are involved in restricting the intratumoural localiza- response to apoptotic cell phagocytosis83. This includes
tion of T cells. For example, inhibiting reactive nitrogen phagocytic pathways designed to avoid activation of
species can increase the number of CD8+ T cells within cytosolic sensors such as stimulator of interferon genes
tumours without impacting their frequency in the sur- (STING)149,150, as well as the secretion of factors that
rounding stroma135. Increasing the degree of fibrosis suppress the activation of neighbouring cells151. One of
would be another mechanism by which macrophages these secreted factors, insulin-like growth factor 1, has
could shield tumours from T cell infiltration. This has already been shown to promote the regrowth of gliomas
been shown in pancreatic cancer with macrophage following CSF1R inhibition152. Any of these mechanisms,
production of granulin promoting the accumulation or a combination of them, could theoretically reduce the
of myofibroblasts137,138. Two recent studies have also capacity of the CD103+ cDC subset to transport tumour
reported that TGFβ signalling acts to exclude T cells antigens to the draining lymph node and initiate a
from human and murine tumours139,140. Tumour macro­ de novo CD8+ T cell response147,148.
phages are one of many cells known to produce TGFβ1 Alternatively, it has recently been described that
(refs33,107) and are not a major source of TGFβ2 or TGFβ3 embryonically derived tissue-resident macrophages
(ref.139); however, it is possible that macrophages can reg- are present within tumours and that these cells retain a
ulate desmoplasia through expression of matrix metallo­ distinct phenotype from their monocyte-derived macro­
proteinases and activation of latent TGFβ141. Human phage counterparts31,32. A unique property of many
monocytes and macrophages even display an enhanced tissue-resident macrophages is expression of CD169
ability to activate TGFβ through their expression of αvβ8 (also known as SIGLEC1)153, which has been shown to
integrin142. Alternatively, because tumour macrophages selectively bind to CD8α+ cDCs and promote antigen

www.nature.com/nri
 Reviews

Table 1 | Combination therapies in clinical trials for cancer Therapeutic targeting of TAMs
Given the importance of TAMs in regulating tumour
Therapy type TAm-targeted agent Immune Clinical trials immunity, there has been considerable interest in thera-
modulator
peutic strategies that target macrophages, which can be
Immuno-oncology CSF1 and/or CSF1R PD1 or PDL1 • NCT02554812a roughly divided into those that deplete TAMs and those
combination agonists antagonists • NCT02452424b
therapy • NCT02777710a that alter pro-tumoural TAM activities. As preclinical
• NCT02880371a evidence largely supports combinatorial approaches
• NCT03238027a being necessary to observe efficacy6, these strategies are
• NCT02526017b currently in evaluation either to augment tumour immu-
• NCT02829723a nity during standard chemotherapy or radiation therapy
• NCT02323191a
• NCT03158272a or in combination with T cell-directed immunotherapy
• NCT02713529b (Table 1) . Herein, we discuss the potential strengths
CCR2 and CCR5 PD1 or PDL1 • NCT02723006b and weaknesses of these approaches and suggest future
antagonists antagonists • NCT03184870a ­directions for clinical trials.
CXCR4 antagonist PDL1 antagonists • NCT02737072b
• NCT03193190a Targeting TAM recruitment and survival. One strat-
• NCT02907099a egy to deplete TAMs is to cut off their replenishment
• NCT03154827a by circulating inflammatory monocytes. Circulating
• NCT03337698a
• NCT02823405a monocytes are highly dependent on CCL2–CCR2 sig-
nalling for their mobilization from the bone marrow and
ANG2 and/or TIE2 PD1 antagonists NCT03239145a
recruitment into inflammatory sites; thus, CCR2 inhibi-
CD40 agonists PD1 or PDL1 • NCT03123783a tion keeps monocytes in the bone marrow, resulting in
antagonists • NCT02304393a
depleted pools of circulating cells and reduced numbers
CD47 (SIRPα–Fc) PD1 or PDL1 • NCT03530683a of TAMs in primary and metastatic sites157–162. In pre-
antagonists • NCT02890368a
• NCT03013218a clinical models, CCL2 or CCR2 blockade can improve
the efficacy of chemotherapy, radiation therapy and
PI3Kγ and/or PI3Kδ PD1 or PDL1 • NCT03471351a
inhibitors antagonists • NCT02637531a immunotherapy61,157,158,163–165. Several CCR2 blockade
combination clinical trials are therefore ongoing. One
Multimodality CSF1R Chemotherapy • NCT02323191a
immuno-oncology and PD1 and/or • NCT03336216a early study in pancreatic cancer showed a greater than
combinations PDL1 40% increase in responsiveness to chemotherapy when
CSF1R PD1 antagonist NCT03431948a CCR2 inhibitors were combined with the FOLFIRINOX
and SBRT regimen157. Biomarker analysis in this study also showed
CSF1R PD1 antagonist NCT03502330a that combination therapy was associated with increased
and CD40 agonist T cell infiltration. Similar observations have been made
CSF1R GVAX and PD1 NCT03153410a in some of the preclinical models, facilitating further
antagonist combinations with checkpoint immunotherapy. Clinical
CCR2 and/or CCR5 Chemotherapy and • NCT03184870a testing of this triple combination — that is, CCR2 inhi-
inhibition PD1 antagonist • NCT03496662a bition, chemotherapy and checkpoint blockade — is
ANG2, angiopoietin 2; CCR , CC-chemokine receptor ; CSF1, colony-stimulating factor 1; CSF1R , now ongoing. Although CCR2 plays a dominant role
colony-stimulating factor 1 receptor ; CXCR4, CXC-chemokine receptor 4; Fc, crystallizable in macrophage recruitment, important considerations
fragment; PD1, programmed cell death 1; PDL1, programmed cell death 1 ligand 1; PI3K ,
phosphoinositide 3-kinase; SBRT, stereotatic body radiotherapy; TAM, tumour-associated
for combined efficacy include rapid compensation by
macrophage. aTrial currently open and active. bTrial closed. granulocytes and a lack of effect on resident TAM pop-
ulations31,166. It has also been observed that cessation
transfer within secondary lymphoid organs154. Whether of CCL2 and/or CCR2 blockade leads to release of the
CD169 + macrophages are present in tumours and monocytes previously trapped within the bone marrow,
transfer antigen to the equivalent CD103+ cDC sub- and this has been shown to exacerbate metastasis in a
set is unknown. This pathway appears irrelevant for murine model of breast cancer167. These parameters will
soluble or Fc receptor-mediated uptake of antigens by be critical to consider in the design of future clinical tri-
cDCs154 but could be involved in the transfer of cellular als, and alternative targets that address these limitations
antigens, particularly during early tumour progression may be required for durable responses.
when tissue-resident macrophages might be expected The CSF1–CSF1R axis has also been heavily investi-
to dominate the microenvironment. Interestingly, it has gated in preclinical models. In most tumours, inhibition
been demonstrated that tissue-resident CD169+ macro­ of CSF1–CSF1R signalling leads to the apoptotic death of
phages in the spleen and lymph node capture extra­ a substantial portion of TAMs, the outlier being glioma
cellular vesicles155, a process that restricts their entrance models in which compensatory activity by GM-CSF
into the lymph node cortex and prevents their inter­ leads to TAM survival and repolarization10. Independent
action with B cells155,156. In tumour models, the absence of of the mechanism of action, in a substantial array of ani-
CD169+ macrophages increases production of immuno­ mal models, CSF1R inhibition improves T cell responses
globulin156, which subsequently drives the polarization in combination with radiation or chemotherapeutic
of tumour macrophages via activating Fcγ receptors to agents61,106,107,168–170. Additionally, CSF1–CSF1R blockade
increase neoplastic progression, tumour growth and improves the efficacy of a diversity of immunotherapeu-
resistance to chemotherapy100–102. tic modalities, including CD40 agonists, PD1 or cytotoxic

Nature Reviews | Immunology

Reviews

T lymphocyte antigen 4 (CTLA4) antagonists and adop- Intriguingly, efficacy has also been described when
tive T cell therapy12,145,171–173. The sum of these studies has combining CD40 agonists with various CSF1R-targeted
led to a number of clinical trials combining CSF1 and/or agents. This effect occurs in systems where the number
CSF1R inhibitors with immune checkpoint blockade. of TAMs is not impacted by CSF1R inhibition173,190 but
In a promising study in patients with pancreatic cancer, also when TAM depletion is effective191. In either case, it
who do not traditionally respond to immunotherapy, appears that blocking CSF1R sensitizes residual TAMs to
it has been reported that some patient responses were reprogramming by anti-CD40 and promotes enhanced
observed when CSF1R and PD1 antagonists were com- T cell-dependent immunity.
bined174, and these studies are now moving forward to a Epigenetic reprogramming of macrophages through
multi-arm phase II clinical trial. Despite the breadth and inhibition of histone deacetylases (HDACs) can also
consistency of these findings, it is perhaps not surpris- elicit a T cell supportive role. Specifically, in mammary
ing that compensatory resistance pathways have already tumour models, a selective class IIa HDAC inhibi-
been defined that ultimately limit the durability of the tor induces antitumour macrophage phenotypes that
response to CSF1R inhibitors152,166,175. The other major support T cell responses and increase responses to
barrier to translation may be that TAM depletion coin- chemotherapy and immune checkpoint blockade192.
cides with loss of tissue-resident populations important Pan-HDAC inhibitors are already being tested with
for maintaining homeostasis176. Although elevated liver PD1 antagonists, and it is possible that macrophage
enzymes appear to be a by-product of reduced clear- reprogramming could have an important role in medi-
ance by Kupffer cells, and not evidence of liver toxicity, ating therapeutic efficacy. An alternative approach to
grade 3 adverse events may limit the utility of CSF1R reprogramming, which has been evaluated extensively
­antagonists in combination studies177. in preclinical models, involves targeting the major
Other pathways involved in macrophage recruit- pathways that drive the immunosuppressive function
ment include the CXCL12–CXCR4 and angiopoietin of TAMs, including IL-4, IL-13 and immunoglobulins6.
2 (ANG2)–TIE2 axes. In glioma and breast models, However, although clinical agents for these pathways
radiation, chemotherapy and vascular disruption have exist, the interest in using them in solid malignancies
been shown to increase CXCL12 expression and pro- to promote immunity has been minimal. An exception to
mote CXCR4-dependent macrophage repopulation and this has been the targeting of PI3Kγ, which is activated in
treatment resistance131,178–180. Intriguingly, it appears that TAMs downstream of multiple pathways, including the
CXCL12 acts by preferentially recruiting TIE2+ macro­ Fcγ receptor. Activation of PI3Kγ signalling in macro­
phages, a population that is strongly associated with phages has been shown to drive TAM immunosup-
the vasculature and is important for tumour angiogen- pressive activities in models of lung cancer, pancreatic
esis181. Thus, depleting TIE2+ macrophages improves cancer and melanoma105,193–196. In animal models, pharma­
vascular disruption180, neutralizing ANG2 improves cological inhibition of PI3Kγ results in macrophage
responses to VEGFA blockade182 and inhibiting TIE2 reprogramming and augmentation of T cell responses
blocks chemotherapy-induced TIE2+ TAM recruitment both as a single agent104 and in combination with T cell
in breast models and leads to decreased metastasis183,184. checkpoint blockade105,194,196.
Though the impact of TIE2+ TAMs on tumour immu-
nity and the potential for immunotherapy combinations Future directions
are unclear, it has been found that dual neutralization of Given the myriad of roles for TAMs, therapeutically tar-
ANG2 and VEGFA promotes T cell infiltration and that geting one specific function of these cells may prove dif-
efficacy is completely dependent upon CD8+ T cells182. ficult. Success has been observed in multiple xenograft
CXCR4 inhibition also renders pancreatic cancer mod- models by blocking CD47, a membrane-bound protein
els more responsive to checkpoint blockade185, although that interacts with SIRPα (also known as SHPS1) on
whether this is related to macrophages is unclear. TAMs to inhibit phagocytosis197. However, it should be
noted that SIRPα is also expressed on CD11b+ cDCs,
Targeting TAM activation. An intrinsic downside and in syngeneic models, the ability of CD47 block-
to depleting TAMs is the loss of their latent immune ade to promote an immune response is dependent
stimulatory role as the primary phagocyte and profes- upon cytosolic sensing of tumour DNA by cDCs198,199.
sional antigen-presenting cell type within tumours. Although this does not negate the approach, the degree
Reprogramming or repolarizing TAMs towards an anti­ to which efficacy is macrophage-dependent is unclear.
tumour phenotype could therefore prove a more efficacious Intriguingly, recent data have suggested that PDL1–PD1
— if potentially more toxic — approach to augmenting signalling in TAMs impairs their phagocytic capacity
other forms of immunotherapy. One of the most pro- and antitumour phenotype200,201, suggesting a second
ductive approaches to date has been the use of an agonist point of synergy for CD47–PD1 combinations. In addi-
CD40 antibody186, which shows combinatorial efficacy in tion, macrophages have been shown to uptake anti-PD1
pancreatic cancer with gemcitabine187 and gemcitabine antibodies through their Fcγ receptors, thereby limiting
and/or nab-paclitaxel188. As CD40 is expressed by cDCs, efficacy in animal models202. The binding of antibodies
the relative contribution of TAM versus cDC activation to macrophage Fc receptors has been shown to regulate
is unclear, but, critically, enhanced responses to PD1 and a number of therapeutic responses in vivo and is a criti-
CTLA4 antagonists have been observed189. Clinical trials cal aspect of antibody drug design203. The importance of
combining a CD40 agonist with chemotherapy, immuno- macrophages in regulating these responses also needs to
therapy and angiogenic inhibitors are currently ongoing. be considered when designing clinical studies, especially

www.nature.com/nri
 Reviews

when incorporating CSF1R or CCR2 inhibitors that In addition, if TAM antagonists are being used to over-
could limit therapeutic efficacy. come resistance to immunotherapy, then more clinical
The sum of preclinical animal and correlative human data correlating macrophage infiltration or phenotype
studies suggests that targeting TAMs could significantly with patient outcomes are needed to guide patient
improve the efficacy of conventional therapies and selection. Finally, it will be critical to determine how
immunotherapeutics. However, in spite of the clinical exactly to use these therapeutic combinations, includ-
interest and a few suggestive early trial outcomes157,174, ing data-driven considerations of dosing strategies and
the optimum therapeutic approach has yet to be iden- sequencing to minimize potential toxicities and/or
tified. This may be partially due to a lack of data on maximize immune stimulatory properties. For example,
key clinical parameters that may determine the suc- anti-CD40 antibody demonstrated lethal toxicity in mice
cess or failure of combinatorial therapeutic approaches when given before gemcitabine in a pancreatic tumour
in humans. For example, while several cancer types, model204. The challenge will be to empirically design tri-
such as ovarian and pancreatic cancers and mesothe- als rather than base them upon historical doses, clinical
lioma, have relatively high macrophage infiltration, it practicality or financial considerations. In spite of these
is unknown whether these reflect patient populations challenges, there remains substantial potential to harness
that will see maximal benefit. Other questions include macrophage biology to improve outcomes for patients
what type of approach (that is, cell depletion or repro- with cancer.
gramming) should be used and whether the best thera­
peutic modalities might be cancer type-dependent. Published online xx xx xxxx

1. Zhang, Q. W. et al. Prognostic significance of 18. Murray, P. J. et al. Macrophage activation and subsets of macrophages derived from Ly6C(high)
tumor-associated macrophages in solid tumor: a polarization: nomenclature and experimental monocytes. Cancer Res. 70, 5728–5739 (2010).
meta-analysis of the literature. PLOS ONE 7, e50946 guidelines. Immunity 41, 14–20 (2014). 34. Strachan, D. C. et al. CSF1R inhibition delays cervical
(2012). 19. Mass, E. et al. Specification of tissue-resident and mammary tumor growth in murine models by
2. Komohara, Y., Jinushi, M. & Takeya, M. Clinical macrophages during organogenesis. Science 353, attenuating the turnover of tumor-associated
significance of macrophage heterogeneity in human aaf4238 (2016). macrophages and enhancing infiltration by CD8+
malignant tumors. Cancer Sci. 105, 1–8 (2014). 20. Hashimoto, D. et al. Tissue-resident macrophages T cells. Oncoimmunology 2, e26968 (2013).
3. Gentles, A. J. et al. The prognostic landscape of genes self-maintain locally throughout adult life with minimal 35. Chen, Z. et al. Cellular and molecular identity of
and infiltrating immune cells across human cancers. contribution from circulating monocytes. Immunity tumor-associated macrophages in glioblastoma.
Nat. Med. 21, 938–945 (2015). 38, 792–804 (2013). Cancer Res. 77, 2266–2278 (2017).
4. Canli, O. et al. Myeloid cell-derived reactive oxygen 21. Schulz, C. et al. A lineage of myeloid cells independent 36. Henze, A. T. & Mazzone, M. The impact of hypoxia on
species induce epithelial mutagenesis. Cancer Cell 32, of Myb and hematopoietic stem cells. Science 336, tumor-associated macrophages. J. Clin. Invest. 126,
869–883 (2017). 86–90 (2012). 3672–3679 (2016).
5. Qian, B. Z. & Pollard, J. W. Macrophage diversity This study identifies tissue-resident macrophages 37. Doedens, A. L. et al. Macrophage expression of
enhances tumor progression and metastasis. Cell 141, as having HSC-independent origins in mice. hypoxia-inducible factor-1 alpha suppresses T cell
39–51 (2010). 22. Yona, S. et al. Fate mapping reveals origins and function and promotes tumor progression.
6. Ruffell, B. & Coussens, L. M. Macrophages and dynamics of monocytes and tissue macrophages Cancer Res. 70, 7465–7475 (2010).
therapeutic resistance in cancer. Cancer Cell 27, under homeostasis. Immunity 38, 79–91 (2013). 38. Imtiyaz, H. Z. et al. Hypoxia-inducible factor 2alpha
462–472 (2015). 23. Hoeffel, G. et al. C-Myb+ erythro-myeloid regulates macrophage function in mouse models of
7. Coffelt, S. B. & de Visser, K. E. Immune-mediated progenitor-derived fetal monocytes give rise to adult acute and tumor inflammation. J. Clin. Invest. 120,
mechanisms influencing the efficacy of anticancer tissue-resident macrophages. Immunity 42, 665–678 2699–2714 (2010).
therapies. Trends Immunol. 36, 198–216 (2015). (2015). 39. Casazza, A. et al. Impeding macrophage entry into
8. Engblom, C., Pfirschke, C. & Pittet, M. J. The role of 24. Ginhoux, F. et al. Fate mapping analysis reveals that hypoxic tumor areas by Sema3A/Nrp1 signaling
myeloid cells in cancer therapies. Nat. Rev. Cancer 16, adult microglia derive from primitive macrophages. blockade inhibits angiogenesis and restores antitumor
447–462 (2016). Science 330, 841–845 (2010). immunity. Cancer Cell 24, 695–709 (2013).
9. Noy, R. & Pollard, J. W. Tumor-associated This study identifies microglia as having an This study demonstrates the importance of tumour
macrophages: from mechanisms to therapy. Immunity embryonic origin in mice. hypoxia as a regulator of the immunosuppressive
41, 49–61 (2014). 25. Calderon, B. et al. The pancreas anatomy conditions nature of macrophages.
This study describes a role for CSF1R in regulating the origin and properties of resident macrophages. 40. Wallerius, M. et al. Guidance molecule SEMA3A
the polarization of tumour macrophages rather J. Exp. Med. 212, 1497–1512 (2015). restricts tumor growth by differentially regulating the
than simply recruiting them into tumours. 26. Epelman, S. et al. Embryonic and adult-derived proliferation of tumor-associated macrophages.
10. Pyonteck, S. M. et al. CSF-1R inhibition alters resident cardiac macrophages are maintained through Cancer Res. 76, 3166–3178 (2016).
macrophage polarization and blocks glioma distinct mechanisms at steady state and during 41. Colegio, O. R. et al. Functional polarization of
progression. Nat. Med. 19, 1264–1272 (2013). inflammation. Immunity 40, 91–104 (2014). tumour-associated macrophages by tumour-derived
11. Olson, O. C., Kim, H., Quail, D. F., Foley, E. A. 27. Gibbings, S. L. et al. Transcriptome analysis highlights lactic acid. Nature 513, 559–563 (2014).
& Joyce, J. A. Tumor-associated macrophages the conserved difference between embryonic and This study finds that altered tumour metabolism
suppress the cytotoxic activity of antimitotic agents. postnatal-derived alveolar macrophages. Blood 126, contributes to the pro-tumour phenotype of
Cell Rep. 19, 101–113 (2017). 1357–1366 (2015). macrophages.
12. Zhu, Y. et al. CSF1/CSF1R blockade reprograms 28. Loyher, P. L. et al. Macrophages of distinct origins 42. Geeraerts, X., Bolli, E., Fendt, S. M.
tumor-infiltrating macrophages and improves contribute to tumor development in the lung. J. Exp. & Van Ginderachter, J. A. Macrophage metabolism
response to T cell checkpoint immunotherapy in Med. 215, 2536–2553 (2018). as therapeutic target for cancer, atherosclerosis, and
pancreatic cancer models. Cancer Res. 74, 29. Lavin, Y. et al. Tissue-resident macrophage enhancer obesity. Front. Immunol. 8, 289 (2017).
5057–5069 (2014). landscapes are shaped by the local microenvironment. 43. Freemerman, A. J. et al. Metabolic reprogramming
13. Bonnardel, J. & Guilliams, M. Developmental control Cell 159, 1312–1326 (2014). of macrophages: glucose transporter 1 (GLUT1)-
of macrophage function. Curr. Opin. Immunol. 50, 30. Gosselin, D. et al. Environment drives selection and mediated glucose metabolism drives a
64–74 (2018). function of enhancers controlling tissue-specific proinflammatory phenotype. J. Biol. Chem. 289,
14. Epelman, S., Lavine, K. J. & Randolph, G. J. Origin macrophage identities. Cell 159, 1327–1340 7884–7896 (2014).
and functions of tissue macrophages. Immunity 41, (2014). 44. Wang, F. et al. Glycolytic stimulation is not a
21–35 (2014). 31. Zhu, Y. et al. Tissue-resident macrophages in requirement for M2 macrophage differentiation.
15. Lavin, Y., Mortha, A., Rahman, A. & Merad, M. pancreatic ductal adenocarcinoma originate from Cell Metab. 28, 463–475 (2018).
Regulation of macrophage development and function embryonic hematopoiesis and promote tumor 45. Huang, S. C. et al. Metabolic reprogramming
in peripheral tissues. Nat. Rev. Immunol. 15, progression. Immunity 47, 323–338 (2017). mediated by the mTORC2-IRF4 signaling axis is
731–744 (2015). This study describes a functional role for essential for macrophage alternative activation.
16. Ginhoux, F. & Guilliams, M. Tissue-resident tissue-resident macrophages in tumour growth, Immunity 45, 817–830 (2016).
macrophage ontogeny and homeostasis. Immunity 44, an unexpected result given the prominence of the 46. Wenes, M. et al. Macrophage metabolism controls
439–449 (2016). recruited population. tumor blood vessel morphogenesis and metastasis.
17. Mantovani, A., Sozzani, S., Locati, M., Allavena, P. 32. Bowman, R. L. et al. Macrophage ontogeny underlies Cell Metab. 24, 701–715 (2016).
& Sica, A. Macrophage polarization: tumor-associated differences in tumor-specific education in brain 47. Penny, H. L. et al. Warburg metabolism in
macrophages as a paradigm for polarized M2 malignancies. Cell Rep. 17, 2445–2459 (2016). tumor-conditioned macrophages promotes metastasis
mononuclear phagocytes. Trends Immunol. 23, 33. Movahedi, K. et al. Different tumor in human pancreatic ductal adenocarcinoma.
549–555 (2002). microenvironments contain functionally distinct Oncoimmunology 5, e1191731 (2016).

Nature Reviews | Immunology

Reviews

48. Jones, R. S. & Morris, M. E. Monocarboxylate modes of human macrophages: differential 96. Coffelt, S. B. et al. IL-17-producing gammadelta T cells
transporters: therapeutic targets and prognostic involvement of proteases and podosome-like and neutrophils conspire to promote breast cancer
factors in disease. Clin. Pharmacol. Ther. 100, structures. J. Immunol. 184, 1049–1061 (2010). metastasis. Nature 522, 345–348 (2015).
454–463 (2016). 72. Sulzmaier, F. J., Jean, C. & Schlaepfer, D. D. FAK in 97. Shahrara, S., Pickens, S. R., Dorfleutner, A. & Pope, R. M.
49. Bohn, T. et al. Tumor immunoevasion via cancer: mechanistic findings and clinical applications. IL-17 induces monocyte migration in rheumatoid
acidosis-dependent induction of regulatory Nat. Rev. Cancer 14, 598–610 (2014). arthritis. J. Immunol. 182, 3884–3891 (2009).
tumor-associated macrophages. Nat. Immunol. 19, 73. Jiang, H. et al. Targeting focal adhesion kinase renders 98. Jovanovic, D. V. et al. IL-17 stimulates the production
1319–1329 (2018). pancreatic cancers responsive to checkpoint and expression of proinflammatory cytokines, IL-β and
50. El-Kenawi, A. et al. Acidity promotes tumor immunotherapy. Nat. Med. 22, 851–860 (2016). TNF-α, by human macrophages. J. Immunol. 160,
progression by altering macrophage phenotype in 74. Laklai, H. et al. Genotype tunes pancreatic ductal 3513–3521 (1998).
prostate cancer. Preprint at bioRxiv https://doi. adenocarcinoma tissue tension to induce matricellular 99. Greenlee-Wacker, M. C. Clearance of apoptotic
org/10.1101/478420 (2018). fibrosis and tumor progression. Nat. Med. 22, neutrophils and resolution of inflammation.
51. Radu, C. G., Nijagal, A., McLaughlin, J., Wang, L. 497–505 (2016). Immunol. Rev. 273, 357–370 (2016).
& Witte, O. N. Differential proton sensitivity of related 75. Sorokin, L. The impact of the extracellular matrix on 100. Andreu, P. et al. FcRγ activation regulates
G protein-coupled receptors T cell death-associated inflammation. Nat. Rev. Immunol. 10, 712–723 (2010). inflammation-associated squamous carcinogenesis.
gene 8 and G2A expressed in immune cells. Proc. Natl 76. Tang, M. et al. Toll-like receptor 2 activation promotes Cancer Cell 17, 121–134 (2010).
Acad. Sci. USA 102, 1632–1637 (2005). tumor dendritic cell dysfunction by regulating IL-6 and 101. de Visser, K. E., Korets, L. V. & Coussens, L. M.
52. Kalluri, R. The biology and function of fibroblasts in IL-10 receptor signaling. Cell Rep. 13, 2851–2864 De novo carcinogenesis promoted by chronic
cancer. Nat. Rev. Cancer 16, 582–598 (2016). (2015). inflammation is B lymphocyte dependent. Cancer Cell
53. Chomarat, P., Banchereau, J., Davoust, J. 77. Kim, S. et al. Carcinoma-produced factors activate 7, 411–423 (2005).
& Palucka, A. K. IL-6 switches the differentiation myeloid cells through TLR2 to stimulate metastasis. 102. Affara, N. I. et al. B cells regulate macrophage
of monocytes from dendritic cells to macrophages. Nature 457, 102–106 (2009). phenotype and response to chemotherapy in
Nat. Immunol. 1, 510–514 (2000). 78. Kobayashi, N. et al. Hyaluronan deficiency in tumor squamous carcinomas. Cancer Cell 25, 809–821
54. Wu, M. H. et al. Targeting galectin-1 in stroma impairs macrophage trafficking and tumor (2014).
carcinoma-associated fibroblasts inhibits oral neovascularization. Cancer Res. 70, 7073–7083 103. Gunderson, A. J. et al. Bruton tyrosine
squamous cell carcinoma metastasis by (2010). kinase-dependent immune cell cross-talk drives
downregulating MCP-1/CCL2 expression. Clin. Cancer 79. Jameson, J. M., Cauvi, G., Sharp, L. L., Witherden, D. A. pancreas cancer. Cancer Discov. 6, 270–285 (2016).
Res. 17, 1306–1316 (2011). & Havran, W. L. Gammadelta T cell-induced hyaluronan 104. Kaneda, M. M. et al. Macrophage PI3Kγ drives
55. Torres, S. et al. Proteome profiling of production by epithelial cells regulates inflammation. J. pancreatic ductal adenocarcinoma progression.
cancer-associated fibroblasts identifies novel Exp. Med. 201, 1269–1279 (2005). Cancer Discov. 6, 870–885 (2016).
proinflammatory signatures and prognostic markers 80. Lee-Sayer, S. S. et al. The where, when, how, and why 105. Kaneda, M. M. et al. PI3Kγ is a molecular switch that
for colorectal cancer. Clin. Cancer Res. 19, of hyaluronan binding by immune cells. controls immune suppression. Nature 539, 437–442
6006–6019 (2013). Front. Immunol. 6, 150 (2015). (2016).
56. Mathew, E. et al. Mesenchymal stem cells promote 81. Kroemer, G., Galluzzi, L., Kepp, O. & Zitvogel, L. This is one of several studies showing that
pancreatic tumor growth by inducing alternative Immunogenic cell death in cancer therapy. Annu. Rev. targeting downstream kinases can alter
polarization of macrophages. Neoplasia 18, 142–151 Immunol. 31, 51–72 (2013). macrophage phenotype and improve the T cell
(2016). 82. Huber, R. et al. Tumour hypoxia promotes melanoma response against tumours.
57. Kim, J. H. et al. The role of myofibroblasts in growth and metastasis via high mobility group Box-1 106. DeNardo, D. G. et al. Leukocyte complexity predicts
upregulation of S100A8 and S100A9 and the and M2-like macrophages. Sci. Rep. 6, 29914 (2016). breast cancer survival and functionally regulates
differentiation of myeloid cells in the colorectal cancer 83. Roberts, A. W. et al. Tissue-resident macrophages are response to chemotherapy. Cancer Discov. 1, 54–67
microenvironment. Biochem. Biophys. Res. Commun. locally programmed for silent clearance of apoptotic (2011).
423, 60–66 (2012). cells. Immunity 47, 913–927 (2017). 107. Ruffell, B. et al. Macrophage IL-10 blocks CD8+
58. Mace, T. A. et al. Pancreatic cancer-associated stellate 84. Graham, D. K., DeRyckere, D., Davies, K. D. T cell-dependent responses to chemotherapy by
cells promote differentiation of myeloid-derived & Earp, H. S. The TAM family: phosphatidylserine suppressing IL-12 expression in intratumoral dendritic
suppressor cells in a STAT3-dependent manner. sensing receptor tyrosine kinases gone awry in cancer. cells. Cancer Cell 26, 623–637 (2014).
Cancer Res. 73, 3007–3018 (2013). Nat. Rev. Cancer 14, 769–785 (2014). This study describes how macrophages can
59. Kumar, V. et al. CD45 phosphatase inhibits STAT3 85. Cook, R. S. et al. MerTK inhibition in tumor leukocytes indirectly suppress a T cell response by reducing
transcription factor activity in myeloid cells and decreases tumor growth and metastasis. J. Clin. IL-12 expression in tumour-associated dendritic
promotes tumor-associated macrophage Invest. 123, 3231–3242 (2013). cells.
differentiation. Immunity 44, 303–315 (2016). This study describes how recognition of dying 108. Kusmartsev, S. & Gabrilovich, D. I. STAT1 signaling
60. Wang, Q. et al. Vascular niche IL-6 induces alternative tumour cells by macrophages acts to suppress regulates tumor-associated macrophage-mediated
macrophage activation in glioblastoma through HIF- antitumour immunity. T cell deletion. J. Immunol. 174, 4880–4891
2α. Nat. Commun. 9, 559 (2018). 86. Crittenden, M. R. et al. Mertk on tumor macrophages (2005).
61. Mitchem, J. B. et al. Targeting tumor-infiltrating is a therapeutic target to prevent tumor recurrence 109. Gabrilovich, D. I. & Nagaraj, S. Myeloid-derived
macrophages decreases tumor-initiating cells, relieves following radiation therapy. Oncotarget 7, suppressor cells as regulators of the immune system.
immunosuppression, and improves chemotherapeutic 78653–78666 (2016). Nat. Rev. Immunol. 9, 162–174 (2009).
responses. Cancer Res. 73, 1128–1141 (2013). 87. Ubil, E. et al. Tumor-secreted Pros1 inhibits 110. Geiger, R. et al. L-arginine modulates T cell
62. Song, L. et al. Valpha24-invariant NKT cells mediate macrophage M1 polarization to reduce antitumor metabolism and enhances survival and anti-tumor
antitumor activity via killing of tumor-associated immune response. J. Clin. Invest. 128, 2356–2369 activity. Cell 167, 829–842 (2016).
macrophages. J. Clin. Invest. 119, 1524–1536 (2018). 111. Rodriguez, P. C. et al. Arginase I production in the
(2009). 88. Ruffell, B., Affara, N. I. & Coussens, L. M. Differential tumor microenvironment by mature myeloid cells
63. Costa, A. et al. Fibroblast heterogeneity and macrophage programming in the tumor inhibits T cell receptor expression and antigen-specific
immunosuppressive environment in human breast microenvironment. Trends Immunol. 33, 119–126 T cell responses. Cancer Res. 64, 5839–5849
cancer. Cancer Cell 33, 463–479 (2018). (2012). (2004).
64. Givel, A. M. et al. miR200-regulated CXCL12β 89. Biswas, S. K. & Mantovani, A. Macrophage plasticity 112. Chang, C. I., Liao, J. C. & Kuo, L. Macrophage arginase
promotes fibroblast heterogeneity and and interaction with lymphocyte subsets: cancer as a promotes tumor cell growth and suppresses nitric
immunosuppression in ovarian cancers. Nat. Commun. paradigm. Nat. Immunol. 11, 889–896 (2010). oxide-mediated tumor cytotoxicity. Cancer Res. 61,
9, 1056 (2018). 90. Marigo, I. et al. T cell cancer therapy requires CD40- 1100–1106 (2001).
65. Zhou, W. et al. Periostin secreted by glioblastoma CD40L activation of tumor necrosis factor and 113. Bronte, V. & Zanovello, P. Regulation of immune
stem cells recruits M2 tumour-associated inducible nitric-oxide-synthase-producing dendritic responses by L-arginine metabolism. Nat. Rev.
macrophages and promotes malignant growth. cells. Cancer Cell 30, 377–390 (2016). Immunol. 5, 641–654 (2005).
Nat. Cell Biol. 17, 170–182 (2015). 91. Klug, F. et al. Low-dose irradiation programs 114. Lu, T. et al. Tumor-infiltrating myeloid cells induce
66. Pickup, M. W., Mouw, J. K. & Weaver, V. M. The macrophage differentiation to an iNOS+/M1 tumor cell resistance to cytotoxic T cells in mice.
extracellular matrix modulates the hallmarks of phenotype that orchestrates effective T cell J. Clin. Invest. 121, 4015–4029 (2011).
cancer. EMBO Rep. 15, 1243–1253 (2014). immunotherapy. Cancer Cell 24, 589–602 (2013). 115. Lu, T. & Gabrilovich, D. I. Molecular pathways:
67. Stahl, M. et al. Lung collagens perpetuate pulmonary This study is one of the few to describe tumor-infiltrating myeloid cells and reactive oxygen
fibrosis via CD204 and M2 macrophage activation. macrophages as promoting antitumour immunity, species in regulation of tumor microenvironment.
PLOS ONE 8, e81382 (2013). in this case, during low-dose irradiation. Clin. Cancer Res. 18, 4877–4882 (2012).
68. Wesley, R. B. 2nd, Meng, X., Godin, D. & Galis, Z. S. 92. DeNardo, D. G. et al. CD4+ T cells regulate pulmonary 116. Nagaraj, S. et al. Altered recognition of antigen is a
Extracellular matrix modulates macrophage functions metastasis of mammary carcinomas by enhancing mechanism of CD8+ T cell tolerance in cancer.
characteristic to atheroma: collagen type I enhances protumor properties of macrophages. Cancer Cell 16, Nat. Med. 13, 828–835 (2007).
acquisition of resident macrophage traits by human 91–102 (2009). This study identifies nitration of the T cell receptor
peripheral blood monocytes in vitro. Arterioscler 93. Shiao, S. L. et al. TH2-polarized CD4+ T cells and as a mechanism by which reactive oxygen and
Thromb. Vasc. Biol. 18, 432–440 (1998). macrophages limit efficacy of radiotherapy. Cancer nitrogen species can suppress the T cell response
69. Meyaard, L. The inhibitory collagen receptor LAIR-1 Immunol. Res. 3, 518–525 (2015). against cancer.
(CD305). J. Leukoc. Biol. 83, 799–803 (2008). 94. Kryczek, I. et al. B7-H4 expression identifies a novel 117. Zea, A. H. et al. Arginase-producing myeloid
70. McWhorter, F. Y., Davis, C. T. & Liu, W. F. Physical and suppressive macrophage population in human ovarian suppressor cells in renal cell carcinoma patients: a
mechanical regulation of macrophage phenotype and carcinoma. J. Exp. Med. 203, 871–881 (2006). mechanism of tumor evasion. Cancer Res. 65,
function. Cell. Mol. Life Sci. 72, 1303–1316 (2015). 95. Kryczek, I. et al. Relationship between B7-H4, 3044–3048 (2005).
71. Van Goethem, E., Poincloux, R., Gauffre, F., regulatory T cells, and patient outcome in human 118. Munder, M. et al. Arginase I is constitutively
Maridonneau-Parini, I. & Le Cabec, V. Matrix ovarian carcinoma. Cancer Res. 67, 8900–8905 expressed in human granulocytes and participates in
architecture dictates three-dimensional migration (2007). fungicidal activity. Blood 105, 2549–2556 (2005).

www.nature.com/nri
 Reviews

119. Lin, H. et al. Host expression of PD-L1 determines 142. Kelly, A. et al. Human monocytes and macrophages macrophages disrupts myeloid recruitment and
efficacy of PD-L1 pathway blockade-mediated tumor regulate immune tolerance via integrin alphavbeta8- improves chemotherapeutic responses in pancreatic
regression. J. Clin. Invest. 128, 805–815 (2018). mediated TGFβ activation. J. Exp. Med. 215, ductal adenocarcinoma. Gut 67, 1112–1123 (2018).
120. Kuang, D. M. et al. Activated monocytes in 2725–2736 (2018). 167. Bonapace, L. et al. Cessation of CCL2 inhibition
peritumoral stroma of hepatocellular carcinoma foster 143. de Mingo Pulido, A. et al. TIM-3 regulates CD103+ accelerates breast cancer metastasis by promoting
immune privilege and disease progression through dendritic cell function and response to chemotherapy angiogenesis. Nature 515, 130–133 (2014).
PD-L1. J. Exp. Med. 206, 1327–1337 (2009). in breast cancer. Cancer Cell 33, 60–74 (2018). This study highlights that compensatory pathways
This study finds that PDL1 expression by myeloid 144. Broz, M. L. et al. Dissecting the tumor myeloid are a potential risk and limitation of targeting
cells negatively correlates with patient survival, compartment reveals rare activating myeloid cells.
suggesting that host expression of PDL1 is an antigen-presenting cells critical for T cell immunity. 168. Xu, J. et al. CSF1R signaling blockade stanches
important contributor to immune suppression. Cancer Cell 26, 638–652 (2014). tumor-infiltrating myeloid cells and improves the
121. Tang, H. et al. PD-L1 on host cells is essential for 145. Peranzoni, E. et al. Macrophages impede CD8 T cells efficacy of radiotherapy in prostate cancer.
PD-L1 blockade-mediated tumor regression. J. Clin. from reaching tumor cells and limit the efficacy of Cancer Res. 73, 2782–2794 (2013).
Invest. 128, 580–588 (2018). anti-PD-1 treatment. Proc. Natl Acad. Sci. USA 115, 169. Seifert, L. et al. Radiation therapy induces
122. Ceeraz, S., Nowak, E. C. & Noelle, R. J. B7 family E4041–E4050 (2018). macrophages to suppress T-cell responses against
checkpoint regulators in immune regulation and 146. Curiel, T. J. et al. Specific recruitment of regulatory pancreatic tumors in mice. Gastroenterology 150,
disease. Trends Immunol. 34, 556–563 (2013). T cells in ovarian carcinoma fosters immune privilege 1659–1672 (2016).
123. Li, J. et al. Co-inhibitory molecule B7 superfamily and predicts reduced survival. Nat. Med. 10, 170. Wang, Q. et al. Therapeutic effects of CSF1R-blocking
member 1 expressed by tumor-infiltrating myeloid 942–949 (2004). antibodies in multiple myeloma. Leukemia 32,
cells induces dysfunction of anti-tumor CD8+ T cells. 147. Salmon, H. et al. Expansion and activation of CD103+ 176–183 (2018).
Immunity 48, 773–786 (2018). dendritic cell progenitors at the tumor site enhances 171. Neubert, N. J. et al. T cell-induced CSF1 promotes
124. Kreymborg, K. et al. Ablation of B7-H3 but not B7-H4 tumor responses to therapeutic PD-L1 and BRAF melanoma resistance to PD1 blockade. Sci. Transl
Results in highly increased tumor burden in a murine inhibition. Immunity 44, 924–938 (2016). Med. 10, eaan3311 (2018).
model of spontaneous prostate cancer. 148. Roberts, E. W. et al. Critical role for CD103+/CD141+ 172. Mok, S. et al. Inhibition of CSF-1 receptor improves
Cancer Immunol. Res. 3, 849–854 (2015). dendritic cells bearing CCR7 for tumor antigen the antitumor efficacy of adoptive cell transfer
125. Rahbar, R. et al. B7-H4 expression by trafficking and priming of T cell immunity in immunotherapy. Cancer Res. 74, 153–161 (2014).
nonhematopoietic cells in the tumor melanoma. Cancer Cell 30, 324–336 (2016). This study shows that targeting macrophages can
microenvironment promotes antitumor immunity. 149. Cunha, L. D. et al. LC3-associated phagocytosis in improve the efficacy of immunotherapy.
Cancer Immunol. Res. 3, 184–195 (2015). myeloid cells promotes tumor immune tolerance. Cell 173. Wiehagen, K. R. et al. Combination of CD40 agonism
126. Smith, L. K. et al. Interleukin-10 directly inhibits CD8+ 175, 429–441 (2018). and CSF-1R blockade reconditions tumor-associated
T cell function by enhancing N-glycan branching to 150. Ahn, J., Xia, T., Rabasa Capote, A., Betancourt, D. macrophages and drives potent antitumor immunity.
decrease antigen sensitivity. Immunity 48, 299–312 & Barber, G. N. Extrinsic phagocyte-dependent STING Cancer Immunol. Res. 5, 1109–1121 (2017).
(2018). signaling dictates the immunogenicity of dying cells. This study finds that blocking CSF1R improves the
127. Demotte, N. et al. Restoring the association of the Cancer Cell 33, 862–873 (2018). efficacy of a CD40 agonist, an unexpected finding
T cell receptor with CD8 reverses anergy in human 151. Han, C. Z. et al. Macrophages redirect phagocytosis given that targeting CSF1R is usually associated
tumor-infiltrating lymphocytes. Immunity 28, by non-professional phagocytes and influence with reduced macrophage infiltration.
414–424 (2008). inflammation. Nature 539, 570–574 (2016). 174. Wainberg, Z. A. et al. First-in-human phase 1 dose
128. Henderson, N. C. et al. Galectin-3 expression and 152. Quail, D. F. et al. The tumor microenvironment escalation and expansion of a novel combination,
secretion links macrophages to the promotion of underlies acquired resistance to CSF-1 R inhibition in anti–CSF-1 receptor (cabiralizumab) plus anti–PD-1
renal fibrosis. Am. J. Pathol. 172, 288–298 (2008). gliomas. Science 352, aad3018 (2016). (nivolumab), in patients with advanced solid tumors
129. Schuette, V. et al. Mannose receptor induces T cell 153. Gupta, P. et al. Tissue-resident CD169+ macrophages [abstract O42]. J. Immunother. Cancer 5 (Suppl. 3),
tolerance via inhibition of CD45 and up-regulation form a crucial front line against plasmodium infection. 89 (2017).
of CTLA-4. Proc. Natl Acad. Sci. USA 113, Cell Rep. 16, 1749–1761 (2016). 175. Kumar, V. et al. Cancer-associated fibroblasts
10649–10654 (2016). 154. van Dinther, D. et al. Functional CD169 on neutralize the anti-tumor effect of CSF1 receptor
130. Stockmann, C. et al. Deletion of vascular endothelial macrophages mediates interaction with dendritic blockade by inducing PMN-MDSC infiltration of
growth factor in myeloid cells accelerates cells for CD8+ T Cell cross-priming. Cell Rep. 22, tumors. Cancer Cell 32, 654–668 (2017).
tumorigenesis. Nature 456, 814–818 (2008). 1484–1495 (2018). 176. Ries, C. H. et al. Targeting tumor-associated
131. Hughes, R. et al. Perivascular M2 macrophages 155. Saunderson, S. C., Dunn, A. C., Crocker, P. R. macrophages with anti-CSF-1R antibody reveals a
stimulate tumor relapse after chemotherapy. & McLellan, A. D. CD169 mediates the capture of strategy for cancer therapy. Cancer Cell 25, 846–859
Cancer Res. 75, 3479–3491 (2015). exosomes in spleen and lymph node. Blood 123, (2014).
132. Harney, A. S. et al. Real-time imaging reveals local, 208–216 (2014). This study demonstrates that targeting CSF1R can
transient vascular permeability, and tumor cell 156. Pucci, F. et al. SCS macrophages suppress melanoma deplete macrophages in the tumours and
intravasation stimulated by TIE2hi by restricting tumor-derived vesicle-B cell interactions. metastatic lesions of patients.
macrophage-derived VEGFA. Cancer Discov. 5, Science 352, 242–246 (2016). 177. Cannarile, M. A. et al. Colony-stimulating factor 1
932–943 (2015). 157. Nywening, T. M. et al. Targeting tumour-associated receptor (CSF1R) inhibitors in cancer therapy.
133. Sato, E., Simpson, K. L., Grisham, M. B., Koyama, S. macrophages with CCR2 inhibition in combination J. Immunother. Cancer 5, 53 (2017).
& Robbins, R. A. Effects of reactive oxygen and with FOLFIRINOX in patients with borderline 178. Sanchez-Martin, L. et al. The chemokine CXCL12
nitrogen metabolites on RANTES- and IL-5-induced resectable and locally advanced pancreatic cancer: regulates monocyte-macrophage differentiation and
eosinophil chemotactic activity in vitro. Am. J. Pathol. a single-centre, open-label, dose-finding, RUNX3 expression. Blood 117, 88–97 (2011).
155, 591–598 (1999). non-randomised, phase 1b trial. Lancet Oncol. 17, 179. Wang, S. C., Yu, C. F., Hong, J. H., Tsai, C. S.
134. Sato, E., Simpson, K. L., Grisham, M. B., Koyama, S. 651–662 (2016). & Chiang, C. S. Radiation therapy-induced tumor
& Robbins, R. A. Effects of reactive oxygen and 158. Sanford, D. E. et al. Inflammatory monocyte invasiveness is associated with SDF-1-regulated
nitrogen metabolites on MCP-1-induced monocyte mobilization decreases patient survival in pancreatic macrophage mobilization and vasculogenesis.
chemotactic activity in vitro. Am. J. Physiol. 277, cancer: a role for targeting the CCL2/CCR2 axis. PLOS ONE 8, e69182 (2013).
L543–L549 (1999). Clin. Cancer Res. 19, 3404–3415 (2013). 180. Welford, A. F. et al. TIE2-expressing macrophages limit
135. Molon, B. et al. Chemokine nitration prevents 159. Zhao, L. et al. Recruitment of a myeloid cell subset the therapeutic efficacy of the vascular-disrupting
intratumoral infiltration of antigen-specific T cells. (CD11b/Gr1 mid) via CCL2/CCR2 promotes the agent combretastatin A4 phosphate in mice. J. Clin.
J. Exp. Med. 208, 1949–1962 (2011). development of colorectal cancer liver metastasis. Invest. 121, 1969–1973 (2011).
Expanding upon the impact of reactive nitrogen Hepatology 57, 829–839 (2013). 181. Mazzieri, R. et al. Targeting the ANG2/TIE2 axis
species on the TME, this study finds that CCL2 160. Qian, B. Z. et al. CCL2 recruits inflammatory inhibits tumor growth and metastasis by impairing
nitration negatively impacts T cell infiltration. monocytes to facilitate breast-tumour metastasis. angiogenesis and disabling rebounds of proangiogenic
136. Franciszkiewicz, K., Boissonnas, A., Boutet, M., Nature 475, 222–225 (2011). myeloid cells. Cancer Cell 19, 512–526 (2011).
Combadiere, C. & Mami-Chouaib, F. Role of 161. Lim, S. Y., Yuzhalin, A. E., Gordon-Weeks, A. N. 182. Schmittnaegel, M. et al. Dual angiopoietin-2 and
chemokines and chemokine receptors in shaping the & Muschel, R. J. Targeting the CCL2-CCR2 signaling VEGFA inhibition elicits antitumor immunity that is
effector phase of the antitumor immune response. axis in cancer metastasis. Oncotarget 7, enhanced by PD-1 checkpoint blockade. Sci. Transl
Cancer Res. 72, 6325–6332 (2012). 28697–28710 (2016). Med. 9, eaak9670 (2017).
137. Nielsen, S. R. et al. Macrophage-secreted granulin 162. Zhang, J., Patel, L. & Pienta, K. J. CC chemokine ligand 183. Harney, A. S. et al. The selective Tie2 inhibitor
supports pancreatic cancer metastasis by inducing 2 (CCL2) promotes prostate cancer tumorigenesis and rebastinib blocks recruitment and function of Tie2(Hi)
liver fibrosis. Nat. Cell Biol. 18, 549–560 (2016). metastasis. Cytokine Growth Factor Rev. 21, 41–48 macrophages in breast cancer and pancreatic
138. Quaranta, V. et al. Macrophage-derived granulin (2010). neuroendocrine tumors. Mol. Cancer Ther. 16,
drives resistance to immune checkpoint inhibition in 163. Connolly, K. A. et al. Increasing the efficacy of 2486–2501 (2017).
metastatic pancreatic cancer. Cancer Res. 78, radiotherapy by modulating the CCR2/CCR5 184. Karagiannis, G. S. et al. Neoadjuvant chemotherapy
4253–4269 (2018). chemokine axes. Oncotarget 7, 86522–86535 induces breast cancer metastasis through a
139. Tauriello, D. V. F. et al. TGFβ drives immune evasion (2016). TMEM-mediated mechanism. Sci. Transl Med. 9,
in genetically reconstituted colon cancer metastasis. 164. Kalbasi, A. et al. Tumor-derived CCL2 mediates eaan0026 (2017).
Nature 554, 538–543 (2018). resistance to radiotherapy in pancreatic ductal 185. Feig, C. et al. Targeting CXCL12 from FAP-expressing
140. Mariathasan, S. et al. TGFβ attenuates tumour adenocarcinoma. Clin. Cancer Res. 23, 137–148 carcinoma-associated fibroblasts synergizes with
response to PD-L1 blockade by contributing to (2017). anti-PD-L1 immunotherapy in pancreatic cancer.
exclusion of T cells. Nature 554, 544–548 (2018). 165. Fridlender, Z. G. et al. CCL2 blockade augments cancer Proc. Natl Acad. Sci. USA 110, 20212–20217 (2013).
141. Kessenbrock, K., Plaks, V. & Werb, Z. Matrix immunotherapy. Cancer Res. 70, 109–118 (2010). 186. Vonderheide, R. H. The immune revolution: a case for
metalloproteinases: regulators of the tumor 166. Nywening, T. M. et al. Targeting both priming, not checkpoint. Cancer Cell 33, 563–569
microenvironment. Cell 141, 52–67 (2010). tumour-associated CXCR2+ neutrophils and CCR2+ (2018).

Nature Reviews | Immunology

Reviews

187. Beatty, G. L. et al. CD40 agonists alter tumor stroma suppressive myeloid cell polarization during tumor 204. Byrne, K. T., Leisenring, N. H., Bajor, D. L.
and show efficacy against pancreatic carcinoma in progression. Cancer Immunol. Res. 5, 957–968 & Vonderheide, R. H. CSF-1R-dependent lethal
mice and humans. Science 331, 1612–1616 (2011). (2017). hepatotoxicity when agonistic CD40 antibody is given
This is the first clinical study of a CD40 agonist in 196. De Henau, O. et al. Overcoming resistance to before but not after chemotherapy. J. Immunol. 197,
pancreatic cancer, with complementary murine checkpoint blockade therapy by targeting 179–187 (2016).
studies highlighting macrophages as the relevant PI3Kγ in myeloid cells. Nature 539, 443–447 205. Kitamura, T., Qian, B. Z. & Pollard, J. W. Immune cell
cellular target. (2016). promotion of metastasis. Nat. Rev. Immunol. 15,
188. Byrne, K. T. & Vonderheide, R. H. CD40 stimulation 197. Willingham, S. B. et al. The CD47-signal regulatory 73–86 (2015).
obviates innate sensors and drives T cell immunity in protein alpha (SIRPa) interaction is a therapeutic 206. Yang, M., McKay, D., Pollard, J. W. & Lewis, C. E. Diverse
cancer. Cell Rep. 15, 2719–2732 (2016). target for human solid tumors. Proc. Natl Acad. Sci. functions of macrophages in different tumor
189. Winograd, R. et al. Induction of T cell immunity USA 109, 6662–6667 (2012). microenvironments. Cancer Res. 78, 5492–5503
overcomes complete resistance to PD-1 and CTLA-4 198. Liu, X. et al. CD47 blockade triggers T cell-mediated (2018).
blockade and improves survival in pancreatic destruction of immunogenic tumors. Nature Med. 21,
carcinoma. Cancer Immunol. Res. 3, 399–411 1209–1215 (2015). Acknowledgements
(2015). 199. Xu, M. M. et al. Dendritic cells but not macrophages The laboratory of D.G.D. is supported by funding from the
190. Perry, C. J. et al. Myeloid-targeted immunotherapies sense tumor mitochondrial DNA for cross-priming National Cancer Institute, including P50CA196510,
act in synergy to induce inflammation and antitumor through signal regulatory protein alpha signaling. R01CA177670, R01CA203890 and P30CA091842
immunity. J. Exp. Med. 215, 877–893 (2018). Immunity 47, 363–373 (2017). Supplement-15S3, as well as The Mary Kay Foundation. The
191. Hoves, S. et al. Rapid activation of tumor-associated 200. Gordon, S. R. et al. PD-1 expression by laboratory of B.R. is supported by funding from the National
macrophages boosts preexisting tumor immunity. tumour-associated macrophages inhibits Institutes of Health (R00CA185325), the Florida Department
J. Exp. Med. 215, 859–876 (2018). phagocytosis and tumour immunity. Nature 545, of Health Bankhead-Coley Cancer Research Program (8BC02)
192. Guerriero, J. L. et al. Class IIa HDAC inhibition reduces 495–499 (2017). and the Florida Breast Cancer Foundation. The authors thank
breast tumours and metastases through anti-tumour 201. Hartley, G. P., Chow, L., Ammons, D. T., Wheat, W. H. members of their laboratories for helpful discussion.
macrophages. Nature 543, 428–432 (2017). & Dow, S. W. Programmed cell death ligand 1 (PD-L1)
This study describes how macrophages can be signaling regulates macrophage proliferation and Author contributions
reprogrammed by targeting their epigenetic state activation. Cancer Immunol. Res. 6, 1260–1273 The authors contributed equally to all aspects of the article.
as opposed to utilizing receptor agonists or (2018).
antagonists. 202. Arlauckas, S. P. et al. In vivo imaging reveals a Competing interests
193. Schmid, M. C. et al. PI3-kinase gamma promotes tumor-associated macrophage-mediated resistance The authors declare no competing interests.
Rap1a-mediated activation of myeloid cell integrin pathway in anti-PD-1 therapy. Sci. Transl Med. 9,
alpha4beta1, leading to tumor inflammation and eaal3604 (2017). Publisher’s note
growth. PLOS ONE 8, e60226 (2013). This study describes how macrophages reduce the Springer Nature remains neutral with regard to jurisdictional
194. Sai, J. et al. PI3K inhibition reduces mammary tumor efficacy of immune checkpoint blockade through Fc claims in published maps and institutional affiliations.
growth and facilitates antitumor immunity and receptor-mediated uptake of the antibody.
anti-PD1 responses. Clin. Cancer Res. 23, 203. DiLillo, D. J. & Ravetch, J. V. Fc-receptor interactions Reviewer information
3371–3384 (2017). regulate both cytotoxic and immunomodulatory Nature Reviews Immunology thanks M. De Palma,
195. Foubert, P., Kaneda, M. M. & Varner, J. A. PI3Kγ therapeutic antibody effector functions. J. Guerriero and S. Patel for their contribution to the peer
activates integrin α4 and promotes immune Cancer Immunol. Res. 3, 704–713 (2015). review of this work.

www.nature.com/nri