Immune Determinants of the Pre-Metastatic Niche

Abstract

Primary tumors actively and specifically prime pre-metastatic niches (PMNs), the future sites of organotropic metastasis, preparing these distant microenvironments for disseminated tumor cell arrival. While initial studies of the PMN focused on extracellular matrix alterations and stromal reprogramming, it is increasingly clear that the far-reaching effects of tumors are in great part achieved through systemic and local PMN immunosuppression. Here, we discuss recent advances in our understanding of the tumor immune microenvironment and provide a comprehensive overview of the immune determinants of the PMN’s spatiotemporal evolution. Moreover, we depict the PMN immune landscape, based on functional pre-clinical studies as well as mounting clinical evidence, and the dynamic, reciprocal crosstalk with systemic changes imposed by cancer progression. Finally, we outline emerging therapeutic approaches that alter the dynamics of the interactions driving PMN formation and reverse immunosuppression programs in the PMN ensuring early anti-tumor immune responses.

Introduction

The systemic effects of cancer are crucial determinants of patient outcome. Tumor-secreted factors, including tumor-derived extracellular vesicles and particles (EVPs), circulate and alter distant organs, governing metastatic progression, which is responsible for the majority of cancer-related deaths. The concept that tumors have a predilection to metastasize to specific organs was first introduced in Stephen Paget’s ‘seed and soil’ theory of cancer metastasis in 1889. However, only in the 1970s did Isaiah Fidler provide experimental evidence of metastatic organotropism, demonstrating that cancer cell-intrinsic properties dictate preferential homing to specific target sites, independently of their nonspecific distribution following injection¹.

Decades later, our lab and others demonstrated that primary tumors induced early changes in the microenvironment of secondary distant organs devoid of cancer cells, thereby providing a milieu favorable for cancer cell settlement and metastasis initiation, creating niches termed ‘pre-metastatic niches’ (PMNs)^2,3. The PMN is shaped by interactions between cancer-secreted factors and resident stromal cells at these distant sites as well as bone marrow-derived cells (BMDCs)². The priming of a distant organ consists of stepwise processes that skew tissue homeostasis towards a dysfunctional environment receptive to circulating tumor cell (CTC) colonization: vascular leakiness^4,5, lymphangiogenesis³, extracellular matrix (ECM) remodeling, and the generation of an immunosuppressive microenvironment (Fig. 1).

Figure 1. Pre-metastatic niche formation.

Primary tumors secrete soluble factors, such as cytokines, chemokines, hormones, and metabolites as well as extracellular vesicles and particles (EVPs) which disseminate systemically, reprogramming distant organ microenvironments. Primary tumor-derived factors exert both direct effects, via tumor-derived secreted factors (TDSFs) and EVPs circulating through the hematogenous and lymphatic systems to distant future sites of metastasis and indirect effects, mediated through their impact on non-metastatic organs that then influence host physiology. The indirect factors promote the mobilization of bone-marrow-derived cells (BMDCs), secretion of metabolites by the gut microbiome, and of stress hormones following hypothalamic-pituitary-adrenal (HPA) axis activation. The direct factors alter distant microenvironments by affecting the tissue resident cells, such as endothelial cells and fibroblasts, leading to vascular leakiness and extracellular matrix (ECM) remodeling while the indirect effects induce the recruitment of non-resident immune cells. Together, these changes generate an immunosuppressive and inflammatory environment which includes myeloid-derived suppressor cells (MDSC) BMDCs, such as VEGFR1⁺ HPCs and CD11b⁺ myeloid cells, tumor-associated macrophages (TAMs), tumor-associated neutrophils (TANs), T regulatory (Treg) and B regulatory (Breg) cells, γδT cells, dysfunctional dendritic cells (DCs) and natural killer (NK) cells, suppressed CD4⁺/CD8⁺ T cells, and cancer-associated fibroblasts (CAFs). The reprogrammed microenvironment then attracts disseminated tumor cells to home and colonize this permissive soil. The most common PMN sites studied to date are the LN, lung, liver, omentum, bone, and brain.

This review offers a holistic overview of the immune determinants driving the PMN’s spatiotemporal evolution (Fig.2), clinical evidence of PMN and methods for clinical detection of PMN in patients, exploring the impact of current therapies on PMN formation, and potential therapeutic strategies to promote anti-tumor immunity throughout PMN evolution (Fig. 3).

Figure 2. Pre-metastatic niche evolution.

Tumor-derived factors such as TDSF, pro-angiogenic, pro-inflammatory factors, and EVPs trigger the formation of an early PMN characterized by ECM remodeling and inflammatory changes that reprogram the tissue-resident immune and stromal cells in distant organs that orchestrate the PMN initiation. The mid PMN, is characterized by increased fibronectin deposition and upregulation of tenascin C, enhance ECM deposition and remodeling and vascular permeability that facilitates BMDCs recruitment. In the late PMN, local secretion of inflammatory and chemotactic molecules reinforce the continued recruitment of hematopoietic progenitor cells and immature myeloid cells, neutrophils, macrophages, Tregs, and dysfunctional DCs and NK cells that crosstalk with the PMN resident cells to maintain an inflammatory and immunosuppressive circuitry. The resulting ECM modifications accompanied by enhanced immune cell recruitment, persistent inflammation, and immunosuppression support CTC adhesion and extravasation, and CTC outgrowth in the Metastatic niche.

Figure 3. Therapeutic strategies to target the PMN and associated immunosuppression.

Direct interference with innate immunity-mediated PMN formation (purple) includes platelet targeting, blocking of myeloid cell recruitment, trafficking or docking at the PMN site, TAM re-education, MDSC depletion, and NET formation. Direct interference with adaptive immunity-mediated PMN formation (green) employs immune checkpoint blockade molecules on T and NK cells, as well as specific blockade of Tregs. Reversal of PMN immunosuppression by co-opting the anti-tumor immunity (blue) include the use of engineered MDSCs, normalization of the NK cell pool, and the use of cancer vaccines to enhance long-term CD8⁺ T cell immune responses. Therapies indirectly targeting immunity (black) that result in reduced immune evasion and boost immunotherapies include the normalization of vascular permeability, targeting pro-metastatic effects of CAFs in the PMN, targeting cancer innervation, co-opting the gut microbiome to exert anti-metastatic immunomodulation, and EV-based therapies.

Pre-metastatic niche development and evolution

Organotropism: Paget’s hypothesis solved

Organotropism represents the propensity of cancer cells to metastasize to PMNs in specific organs (Fig.1D)^1,2 and is driven by selective interactions between cancer cells, their secretome, and the cellular and extracellular microenvironmental components of the distant future metastatic organ. While some organs, such as the lymph nodes (LNs), bones, lungs, omentum, brain, and liver, are more susceptible to metastasis, other organs, such as the muscle⁶ and kidney⁷ are rarely targeted by metastasis (Box 1). Furthermore, both cancer type (e.g., pancreatic vs breast) and subtype (e.g. luminal vs. basal breast cancer) dictate the particular organ tumor cells invade^8,9. Metastatic site specificity was evident in the earliest PMN studies, as tumor-secreted VEGF selectively induced MMP-9 expression in lung endothelial cells and macrophages¹⁰. Tumor-secreted factors also simultaneously mobilized VEGFR1⁺ very late antigen–4⁺ (VLA-4/integrin α4β1) bone marrow-derived hematopoietic progenitor cells (HPCs) and upregulated fibronectin, an integrin α4β1 ligand, in selective PMNs, such as the lung, which cooperated to recruit BMDCs to the altered distant sites². Thus, cancer cell integrin expression is a critical mediator of organotropism: β1 and β3 integrins mediate bone metastasis^11,12 while integrin α3β1 correlates with lung colonization¹³. Integrin α5 overexpression in either early-stage primary breast cancer (BC) or bone metastases, but not lung, liver, or brain metastases, correlated with osteoclast-mediated bone resorption and colonization by CTCs¹⁴. The key role of integrins in mediating lymphocyte homing highlights PMN dependence on these adhesion molecules. Lymphocyte function-associated antigen 1 (LFA-1/integrin αLβ2) and integrin α4β1 are crucial for leukocyte rolling and arrest in blood vessels^15,16. Myeloid cells use integrins LFA-1, α4, and β2 to interact with the vasculature while neutrophil extravasation requires LFA-1 for attachment and integrin β2 for crawling to an endothelial junction^17,18.

Box 1: Anti-metastatic niches

Certain tissue microenvironments, collectively referred to as “anti-metastatic niches,” are inhospitable to DTCs²⁷⁹. Clinical meta-analysis of autopsies revealed that, paradoxically, arterial output of the heart in tissues does not correlate with their colonization by DTCs^280–282, thereby suggesting, yet again, that the “soil” of the target organ and not hematogenous access, determines metastatic outcome. However, few studies have explored the specific features of these “soils” that reject metastatic colonization. Proposed mechanisms for DTC outgrowth suppression in the four anti-metastatic niches (i.e., skeletal, spleen, thyroid, and yellow bone marrow) are reviewed elsewhere²⁷⁹.

Despite seeding by DTCs, tumors do not grow in skeletal muscle. Mechanistically, the intrinsic sustained oxidative stress in muscle imposes a profound “metabolic bottleneck” on DTCs, which are not molecularly equipped to overcome such environmental challenges. Functionally, disruption of redox homeostatic balance toward ROS production can impair DTC outgrowth in the lung, whereas the enhancement of anti-oxidant enzymatic tumor cell defense mechanisms (e.g., catalase ectopic expression) allowed DTCs to colonize the skeletal muscle⁶.

Similar mechanisms driving DTC death via metabolic restraints may also exist in the spleen and thyroid: blood filtration in the spleen is a source of redox-active iron that could promote ROS-mediated lipid peroxidation and DNA damage or ferroptosis^283,284, while the thyroid gland stores the majority of body iodine, a highly oxidizing element leading to ROS production²⁸⁵. Moreover, the spleen and thyroid are also immune-rich niches populated by mature immune cells, which may allow early anti-tumor surveillance and hinder DTC outgrowth. Thus, the outstanding question is: given our knowledge of immunometabolism and the crosstalk between exercise and cancer prevention, do specific immunosurveillance mechanisms contribute to the skeletal muscle and other, yet to be characterized, anti-metastatic niches? Understanding the underlying mechanisms responsible for supporting inhospitable, anti-metastatic niches could lead to new therapeutic strategies to prevent PMN formation and metastasis.

Box 2: Anti-PMN innate immunity

While most studies report PMN-promoting functions of innate immunity, some evidence of PMN-suppressing effects have been suggested. For example, EVPs from non-metastatic cell lines induce the development of anti-PMN patrolling monocytes. These “non-metastatic” exosomes stimulated the expansion of Ly6C^lo patrolling monocytes in the bone marrow, which upon PMN arrival caused cancer cell clearance via the recruitment of NK cells and TRAIL-dependent killing of melanoma cells by macrophages²⁸⁶. These findings unveil a novel view of metastatic organotropism, in which the non-metastatic organs may in fact have an active anti-PMN function. In another study, B220⁺CD11b⁺NK1.1⁺ liver-entrained leukocytes travelled to the pre-metastatic site and efficiently blocked deposition of FN, an essential ECM component of the PMN, through coagulation factor X and thrombospondin expression²⁸⁷. Lastly, tumor-entrained neutrophils could also inhibit metastatic seeding in the lungs by generating H₂O₂⁴³. These findings highlight the dynamic nature of the PMN and the potential for anti-PMN treatments by tipping the balance in favor of anti-tumor reponses.

Box 3: PMN modulation by Diet, Exercise, and Aging

An intriguing but understudied area is how systemic changes in homeostasis influence the formation of PMN-like tissue microenvironments. Aging is the most significant risk factor for cancer overall, as a result of various molecular and cellular processes that progress over time. Interestingly, “aged” tissues resemble PMNs, and may serve as pre-existing “fertile soils”, explaining the higher cancer incidence in the elderly population. inflammaging, hampered DNA damage repair mechanisms, and accumulation of epigenetic modifications and somatic mutations. “Aged” tissues are characterized by an inflammatory⁴⁹ and immunosuppressive^288,289 state which may be a critical contributor to PMN evolution. These similarities imply that lessons from studying the elderly population may shed light on parallel PMN mechanisms which are often less accessible and less prevalent.

Exercise, diet, and other lifestyle factors, such as smoking and alcohol consumption, may also modulate PMN progression. PMN formation have been associated with high cholesterol, obesity, and high fat diet^{179,188,193,201,290}. Several metabolic changes associated with obesity correlate with cancer ^290,193). Conversely, dietary restriction enhances T cell fitness and protects against tumors ^291,292. Smoking and alcohol consumption induce chronic inflammation and can exert detrimental effects on the immune system and recent studies showed that chronic nicotine exposure supports lung PMNs²²⁵. Extended lifespans and low cancer incidence in Blue Zones support the idea that healthy diets and lifestyles can reduce cancer incidence.

Exercise boosts the immune system and prevents cancer recurrence and metastasis^293,294. Exercise-induced metabolic changes, as well as muscle cell-derived EVPs are likely to play a role in reprogramming distal PMNs²⁹⁵. Exercise facilitates immune cell infiltration/homing at the primary tumor or PMN, thereby augmenting the efficacy of immunotherapies^293,296,297. Given that in terms of cancer progression, metastasis prevention has the highest potential to impact outcome, nutrition and exercise as tools to minimize inflammation associated with aging are promising interventions to affect PMN development that should be further studied²⁹⁴.

Integrins on EVPs also direct organotropism by their specific interactions with cells and ECM molecules in PMNs^19,20. For example, laminin-binding integrins α6β4 and α6β1 are enriched in lung-tropic EVPs, RGD-motif binding integrin αvβ5 in liver-tropic EVPs, and integrin β3 in brain-tropic EVPs²¹, and α5 integrin, partnering with cadherin 11, in bone-tropic EVPs²². Importantly, the integrin profiles of plasma-derived EVPs from cancer patients were consistent with pre-clinical model findings and, indeed, predicted future metastatic patterns^11,19. As tumor-derived EVPs release precedes CTC arrival during PMN formation, differential EVP integrin expression essentially dictates the site prepared for metastasis, and could be targeted to control metastasis. PMN-promoting alterations supporting metastatic organotropism include cellular, extracellular, and structural features of the affected sites^20,23 and these functions are organ-specific. At the cellular level, platelet depletion reduced lung metastasis, but enhanced LN metastases²⁴. ECM components are critical for organotropism to PMN²⁰. Proteomic analysis of ECM composition of brain, lung, liver, and bone marrow BC metastatic niches, as well as loss of function studies of ECM modulators, revealed that distinct molecules comprise and regulate spread to each metastatic site²⁵. Tissue architecture may also contribute to PMN organotropism: lungs, for example, have dense capillary beds and serve as a lymphatic drainage reservoir, properties which enhance the probability of CTC arrest and extravasation²³.

To enable the systematic investigation of key biological features of PMN and organotropism, 3D models that mimic tissue complexity were bioengineered^26–28. Hydrogel scaffolds seeded with human peripheral blood mononuclear cells and subdermally implanted in mice induced vascularized and pro-inflammatory microenvironments that recruited disseminated tumor cells (DTCs) from an orthotopic prostate tumor xenograft²⁸. These bioengineered PMNs advance our understanding of the biological processes underlying PMN formation and organotropism and allow high-throughput testing of anti-tumor therapies.

Tumor-derived soluble factors (TDSFs)

PMN establishment requires the combined systemic effects of a wide variety of soluble factors: cytokines, growth factors, enzymes, and EVPs secreted by the primary tumor and its milieu. Primary tumors secrete key pro-angiogenic factors VEGF-A and placental growth factor ² that mobilize and recruit VEGFR1⁺VLA-4⁺ BM-derived HPCs to fibronectin-rich pre-metastatic organs. VEGFR1⁺ HPC clusters are a common denominator for several pre-metastatic organs in numerous murine and human cancers, highlighting their clinical relevance². Several TDSFs, including VEGF-A, TNF-α, and TGF-β, induce the expression of inflammatory chemoattractant proteins S100A8 and S100A9, which promote Mac1⁺ myeloid cell lung PMN infiltration²⁹. Functionally, VEGF secretion by primary BC promotes endothelial cell-focal adhesion kinase (FAK) activation, inducing E-selectin-mediated vascular hyperpermeability foci in the pre-metastatic lung³⁰. Moreover, VEGF-induced PGE2 production in the pre-metastatic lung potentiates tumor cell adhesion and ensures late PMN to metastatic niche transition³¹.

Other pro-inflammatory growth factors, angiopoietin-like protein 2 (ANGPTL2), G-CSF, IL-6, and CCL2/MCP-1, are critical for PMN formation through their effects on myelopoiesis. ANGPTL2 binds integrin α5β1 and mediates osteosarcoma cell intravasation³², whereas CXCL chemokine production by lung epithelial cells and recruitment of pro-metastatic neutrophils to the pre-metastatic lung support metastasis³³. Increased circulating levels of tumor-derived G-CSF by early-stage BC, reprogram BM hematopoiesis toward a myeloid lineage expansion and aberrant differentiation towards immunosuppressive, ROS-producing neutrophils, which then egress into the pre-metastatic lung³⁴.

Secreted by both primary tumor cells and associated stroma, IL-6 promotes PMN formation via STAT3 activation in both tumor cells and tumor-infiltrating myeloid cells, facilitating intravasation and trafficking to the lung PMN³⁵. Tumor-secreted IL-6 induces expression of HIF-1α, VEGF, and CCL5 in lymphatic endothelial cells, thereby promoting angiogenesis and CTC lung and LN colonization³⁶. IL-6 secreted by pancreatic cancer α-SMA⁺ cancer-associated fibroblasts (CAFs) also induced liver PMN via STAT3 signaling in hepatocytes through production of serum amyloid A (SAA) 1, SAA2, FN, and collagen I leading to Ly6G⁺ myeloid cell recruitment³⁷.

Tumor-derived CCL2 is a potent chemoattractant for monocytes, macrophages, memory T lymphocytes, and natural killer (NK) cells³⁸. Elevated systemic CCL2 levels were associated with poor prognosis in BC patients³⁹, potentially due to preferential recruitment of CCR2⁺ inflammatory monocytes to pulmonary metastases⁴⁰. Moreover, CCL2 recruited mature macrophages that favored homing and outgrowth of tumor cells in the pre-metastatic lung and enhanced osteoclast differentiation in the pre-metastatic bone⁴¹. CCL2 derived from hypoxic primary tumors also promoted lung PMN infiltration by dysfunctional myeloid and NK cells with decreased capacity to eliminate incoming CTCs⁴². Interestingly, CCL2 can also promote anti-tumor responses, by recruiting to the PMN tumor-entrained neutrophils that exert cytotoxic effects and interfere with tumor cell seeding⁴³.

Tumor-derived EV-driven immunosuppression in the PMN

EVPs mediate intercellular communication by transporting proteins, lipids, nucleic acids, and metabolites, orchestrating a complex crosstalk between innate and adaptive immune cell types in the PMN and systemically, ultimately supporting the formation of an inflammatory and immunosuppressive microenvironment conducive to tumor growth^44–49. Of note, EVPs are extremely heterogeneous in biogenesis, cargo, size, and function, including newly discovered amembraneous extracellular particles, such as exomeres^50–52 and supermeres⁵³, both of which have yet unexplored, distinct contributions to PMN formation and immunosuppression.

Seminal studies revealed that melanoma-derived EVPs reprogram BM progenitor cells towards pro-tumorigenic phenotypes that migrate to the PMN⁴⁷. In melanoma patients, primary tumor-derived EVPs package multiple immune-modulatory proteins, including S100A9, which travel via lymphatics to sub-capsular macrophages in the pre-metastatic sentinel LNs where they inhibit DC dendritic cell (DC) maturation^46,54. Melanoma EVPs transfer MHC class I molecules to antigen-presenting cells (APCs), downregulating CD86 and CD40 co-stimulatory molecules and suppressing DC function⁵⁵. Smilarly, EVP uptake by monocytic cells and DCs in ovarian ascites induced inflammatory mediators, i.e., IL-6, IL-1β, and TNF-α, thereby generating an immunosuppressive environment⁵⁶.

Tumor-derived EVPs also induce suppression of T cell function in the PMN. Ex vivo, they reduced proliferation and enhanced apoptosis of CD8⁺ T cells by downregulating CD69, suppressing CD4⁺ T cell activation, and inducing regulatory T cell (Treg) expansion^57–59. Ovarian cancer-derived EVPs suppressed T cell function both directly by GD3 ganglioside production^48,60 and indirectly through production of immune-suppressive adenosine by Treg cells⁵⁷. Disease activity directly correlated with tumor-derived EVP-dependent CD8⁺ T cell apoptosis, decreased CD4⁺ T cell proliferation, elevated Treg function, and NK suppression⁶¹. Importantly, immune checkpoint molecules packaged in EVPs contribute to PMN immune suppression, a finding with critical implications for immunotherapy outcomes. EVPs derived from plasma of head and neck cancer patients carried active PD-L1, which inhibited activated T cells and correlated with LN metastasis⁶². Consistent with these findings, administration of PD-L1⁺FasL⁺ tumor-derived EVPs at a pre-malignant stage promoted disease progression and reduced immune infiltration in oral squamous cell carcinoma⁶³. The role of PD-L1-expressing EVPs in PMN formation remains to be evaluated. Taken together these data reveal that tumor-derived EVPs induce PMN immunosuppression through multiple mechanisms impacting both innate and adaptive immunity, as well as their crosstalk, and can serve as biomarkers of disease activity and adaptive immune dysfunction.

PMN hallmarks: through the lens of immunity

Vascular permeability and lymphangiogenesis

Vascular permeability is a hallmark of PMN formation⁶⁴, where prostaglandins, nitric oxide (NO), matrix metalloproteinases (MMPs), VEGF, and cytokines disrupt endothelial cell junctions, leading to vascular permeabilization^5,65, enabling extravasation of immune and cancer cells into the tissue. In turn, tumor-secreted VEGF further supports vascular permeability by recruiting tumor-entrained VEGFR1⁺ BMDCs. Several MMPs, such as MMP−1, −2, −3, and −10, upregulated in the PMN, interact with angiopoietin-2 and cyclooxygenase-2 (COX-2) to promote vascular permeability^5,66. Immune cells also contribute to vascular permeability, through activation of innate immune cell CCL2 signaling that upregulates secretion of SAA3 and S100A8 permeability factors⁶⁷. Perivascular macrophages induce permeabilization and metastasis in the brain⁶⁸ and the lung via tenascin C, NO, and TNF secretion⁶⁹. Neutrophils also contribute to vascular instability through the production of neutrophil extracellular traps (NETs), networks of extracellular DNA containing histones and enzymes designed to entrap and destroy pathogens. NETs anchor to the vessel wall and release neutrophil elastase (NE) that proteolytically disrupts vascular integrity^70,71. Moreover, activation of platelets via tumor-derived CD97 leads to granule secretion and ATP release, disrupting endothelial junctions⁷². Lastly, CD4⁺ T cell inhibition prevents vessel normalization⁷³ while secretion of IL-17A by γδ T cells promotes tumor-endothelial transmigration⁷⁴, supporting a role for circulating T cells in shaping vascular leakiness.

EVPs are also potent promoters of vascular leakiness in the lung and brain^19,47,75. Brain metastatic cell-derived EVPs carrying cell migration-inducing and hyaluronan-binding protein (CEMIP) induced endothelial cell branching and inflammation in the perivascular niche by upregulating the prostaglandin-endoperoxide synthase 2 (Ptgs2), Tnf, and Cxcl genes encoding pro-inflammatory, pro-metastatic cytokines⁷⁵. Breast cancer EVPs also promoted brain metastases by transferring the microRNA miR-181c to endothelial cells, altering actin localization via PDPK1 downregulation, leading to blood-brain barrier disruption⁷⁶.

Lymphatic alterations have also been implicated in PMN formation. In triple-negative breast cancer (TNBC), tumor-secreted IL-6-induced CCL5 expression on lymphatic endothelial cells, promoting cancer cell migration to the lung and LN PMN³⁶. Furthermore, lymphatic ablation increased the frequency of immunosuppressive cells in the PMN, such as regulatory T cells (Tregs) and myeloid-derived suppressor cells (MDSCs)^77,78. Thus, both blood and lymphatic vasculature normalization would be required to inhibit PMN formation.

PMN-associated thrombosis

Thrombosis is an early PMN feature orchestrated by platelets that promote ECM remodeling and vascular dysfunction, leading to inflammation and immunosuppression. The coagulation cascade, its interactions with tumor cells and TDSFs are critical for PMN formation and seeding by CTCs. Tumor cells directly promote coagulation via secretion of tissue-factor (TF) and EVPs⁷⁹ that induce platelet activation and aggregation through coagulation factor VII, triggering a proteolytic cascade that generates thrombin⁸⁰. Platelet-derived thrombin catalyzes fibrinogen to fibrin conversion and crosslinking to fibronectin, resulting in platelet aggregation and thrombus formation. This initiates a feed-forward loop, where PMN fibronectin deposition acts as a scaffold for microthrombi that then recruit macrophages, creating a platform for CTC arrest and survival in the lung PMN⁸¹, protecting CTCs from hemodynamic shear stress and patrolling NK cells. Importantly, platelet activation in the PMN is indirectly regulated by the activation of the adenosine diphosphate (ADP) receptor, as P2Y12 loss attenuates fibronectin deposition, VEGFR1⁺ BMDC recruitment, and lung metastasis⁸². Conversely, platelet-derived cytokines, such as thrombospondin-1 and TGF-β1, mediate osteoclast differentiation and BMDC mobilization to the PMN in bone-tropic cancers⁸³.

ECM remodeling in the PMN

ECM remodeling is the defining feature of PMN development. ECM components, such as fibronectin, versican, collagens, and periostin, are modified by resident fibroblast and endothelial cells, as well as by resident and recruited immune cells⁸⁴. Hypoxia-induced matrix-degrading enzymes of the lysyl oxidase (LOX) family play pivotal roles in PMN collagen remodeling and ECM preconditioning to allow immune cell recruitment^42,85. Importantly, systemic release of LOX, LOX-like (LOXL), and FN1 into the circulation by primary BC favors CD11b⁺ myeloid cell accumulation in the pre-metastatic lung⁸⁶ and bone, ensuring adhesion and formation of osteolytic lesions, respectively⁸⁷. Alternatively, HIF-1-mediated upregulation of carbonic anhydrase IX (CAIX) induced G-CSF production downstream of NF-kB activation and subsequent mobilization of granulocytic MDSCs to the lung PMN⁸⁸. Another ECM-remodeling molecule, TIMP-1, induces specific proteolytic cleavage of MMPs and controls their local activities⁸⁹. High systemic levels of TIMP-1 mediated SDF-1/CXCR3-dependent recruitment of neutrophils to the pre-metastatic liver and TIMP-1 upregulation was sufficient to promote liver metastasis in various cancers⁹⁰.

Among stromal cells, fibroblasts are central to ECM deposition and remodeling in the PMN. In the primary tumor microenvironment^91,92 fibroblasts are pivotal mediators of immunosuppression, and accumulating evidence suggests that similar interactions occur in the PMN. Following activation by TDSFs, PMN fibroblasts deposit fibronectin, to which incoming BMDCs adhere via integrins^2,47,93,94 (Fig. 1, 2). Upon arrival, CD11b⁺ BMDCs produce MMP-2, which cleaves collagen fibers, thus promoting BMDC recruitment and the colonization and proliferation of CTCs ⁹⁵. Lung PMN fibroblasts secrete periostin, which also supports MDSC accumulation⁹⁶ and CTC colonization⁹⁷. CAFs are also the main source of CXCL12, a CXCR4 ligand, an immunosuppressive signaling axis⁹⁸ that recruits Tregs to the bone marrow, supporting bone metastasis in prostate cancer patients⁹⁹. High CXCL12 levels were evident in pre-metastatic lungs of patients with colorectal cancer compared to controls with benign lung disease¹⁰⁰. These pleiotropic functions of fibroblasts in supporting an immunosuppressive PMN environment highlight their potential as targets for therapies to hinder PMN formation.

The PMN Immune Landscape and Inflammation

Modulation of immune cell function at pre-metastatic organ sites, (e.g. lung, liver, bone, and LNs) is a critical step in PMN formation. These cells interact with resident stromal cells, such as fibroblasts and endothelial cells, as well as each other, to reprogram the distal microenvironment. A variety of recruited and resident innate and adaptive immune cells shape PMN evolution, immunosuppression, and inflammation (Fig.1). Early changes in the PMN initiate an inflammatory niche that facilitates recruitment of innate immune cells (Fig.1). Subsequent cooperation between recruited cells and factors secreted by remote primary tumors induces changes in resident immune and stromal cells, enhances BMDC chemotaxis to the pre-metastatic organ and establishes a persistent inflammatory yet immune-suppressive PMN circuitry (Fig.2).

Innate cell recruitment to the PMN

Among innate immune populations, recruited myeloid cells are critical drivers of PMN formation^2,101 and inflammation. PMN inflammation is shaped by production of damage-associated molecular pattern (DAMP) molecules (e.g., S100 proteins, SAA 3, high-mobility group box1 [HMGB1]¹⁰²) and expression of their pattern recognition receptors (PRRs) (e.g., receptor for advanced glycation end-products (RAGE) and TLR4^103,104) that induce potent STAT3³⁵ and NF-kB-mediated^103,105 inflammatory signaling, regulating PMN myeloid cell composition and function.

Myeloid cells recruited to the PMN by S100 chemotactic signals subsequently produce S100 proteins in a pro-inflammatory autocrine loop, thereby reinforcing S100 signaling^105,106. Through their diverse autocrine and paracrine functions^107,108, S100 proteins, small molecular Ca²⁺ sensors, collectively orchestrate PMN inflammation, mediating the crosstalk between resident cells of the distant organ, recruited immune cells, and CTCs^{29,103,107,109}. S100 proteins are also involved in EV-mediated PMN reprogramming. For example, S100A4 is a tumor EVP-specific DAMP that induces PMN inflammation^19,110, while S100A8 and S100A9 are upregulated in the pre-metastatic lung by tumor EVPs^19,47 and tumor-secreted VEGF-A, TNF-α, and TGF-β²⁹. Indeed, calprotectin, the S100A8/A9 heterodimer, a ligand/agonist for PRRs, amplifies the production of pro-inflammatory mediators (e.g., SAA3) in the lung PMN to further attract CD11b⁺ MDSCs^103,104. TNF-α accumulated in the PMN then hinders MDSC maturation by regulating S100A8/A9, their receptor RAGE and thus NF-kB signaling^105,111. Recently, additional S100A8/A9 receptors were identified, such as basigin (BSG/CD147)^112,113, although the role of these interactions in PMN evolution remains unexplored.

Once in the PMN, CD11b⁺Gr1⁺ immature myeloid cells downregulate IFN-γ and upregulate pro-inflammatory cytokine and VEGF-A production, facilitating tumor cell extravasation and cancer cell-macrophage interactions^114,115. Gr1⁺CCR2⁺ inflammatory monocytes recruited to the PMN lung via tumor/stroma-derived CCL2 differentiate into tumor-associated macrophages (TAMs) and further recruit metastasis-promoting macrophages (MAMs) that secrete CSF-1 in a VEGF-A-dependent manner (Table 1)^40,116. Meanwhile, pro-tumorigenic M2 TAMs with poor antigen presenting capacity that secrete low IL-12 and high IL-10, IL-4, and IL-13¹¹⁷ differentiated from recruited monocytes reinforce PMN immunosuppression through PD-L1 expression^118,119. Moreover, monocytic MDSCs (mo-MDSCs) accumulate in the lung PMN in response to microenvironmental CCL12 chemotactic signals, stimulating endothelial activation, E-selectin expression, CTC arrest, and metastasis via IL-1β secretion¹²⁰.

Table 1.

The immune landscape of the PMN

Cell type	PMN site	Function	References
INNATE IMMUNE CELLS
BMDCs and BM HPCs	LU, LI, K, SPL, LN, BO	Recruitment of BMDCs to FN-rich areas, enhancing further BMDC recruitmment	2,298
Myeloid cells
Recruited mo-MDSC	LU	Inflammation, immunosuppression, endothelial activation (IL-1β/E-selectin) and CTC arrest. Recruitment of MAM.	40,120
Patrolling mo	LU	Immune surveillance	286
Recruited/res mo, Φ, TAM, Perivascular Φ	LU, BO	Pro-metastatic vascular niche induction, CTC arrest and survival, extravasation, and outgrowth. Secrete MMP-9, S100A8, and FN. Polarized towards an M2 phenotype.	41,69,81,141,143,148
MAM	LU	VEGFA-dependent CTC extravasation, seeding, proliferation., and regulation of inflammation via secreting CSF-1	40,116
Recruited g-MDSC	LU, BR	PMN inflammation and vascular remodeling. Reduced IFN-γ and production of S100A8, S100A9, SAA3, CCL2, MMP-2.	42,103,114,299
Neutrophils	LU, LI, Peripheral blood	Dysregulate the cytokine mileu via a pro-inflammatory signature: Bv8, MMP-9, S100A8, S100A9, IL-1β, TNF-α, IL-6, and COX-2. Upregulate PD-L1 and impair NK and T cell function. TAN release NETs that promote PMN formation.	33,90,125,127,136,300
TENs	LU	Inhibit metastatic seeding by exerting cytotoxic effects	43
Platelets	LU, BO	Promote ECM remodeling, thrombosis and vascular dysfunction, osteoclast differentiation and BMDC mobilization via TGF-β	72,82, 83
DCs	LU, LN	Inhibit NK cell cytotoxicity, DC dysfunction via COX-2	150,151
Tissue-resident macrophages
Lung-res (AM, IM)	LU	Promote inflammation (TNF-α), ECM degradation (MMP-9), and immunosuppression (PD-L1)	10,119,137
Kupffer cells	LI	ECM remodeling (FN) mediating BMDC recruitment Promote immunosuppression via reduction of serum IL-2, TNF-α levels and CD4+ T cell levels in the PMN	44,145,147
Microglia	BR	Polarized towards M2 pro-tumorigenic phenotype Initiate neuroinflammation and regulate cell adhesion	75,140,141,148
Osteoclasts	BO	Formation of osteolytic lesions and bone resorption, critical for tumor cell seeding and colonization	14,83,217
Omental-res	PER	Omental CD163+Tim4+ macrophages are critical for PER PMN	142
ILCs : NK cells	LU, LI	Reduced infiltration and anti-tumor immunity	42,178
ADAPTIVE IMMUNE CELLS
CD4+ T cells	LI, BO	Marked reduction of CD4+ T cell numbers, dysfunctional CD4+ T cells and conversion to Tregs with immunosuppressive role	147,168
T helper 17 (Th17)	LI	Reduced frequency. Production of IL-17. Produce RANKL and induce osteolytic disease before CTC homing	170,178
Regulatory T cells (Tregs)	LU, BO, LN, LI	Inhibit proliferation and function of T and NK cells Enriched and maintained via CXCL12 and CXCR4 signaling Protect infiltrating CTCs from immune responses	99,163–165,178
CD8+ T cells	LI, LN	Constrain myeloid cell infiltration via immunosurveillance. Reduced frequency and cytotoxic function. Increased exhaustion phenotype.	168,178,227
γδ T cells (res)	-	Prevalent at common future metastatic sites	172
NKT-like	Peripheral blood	Associated with shorter disease-free survival	301
NKT type I	LI	Immune surveillance by CXCR6+ NKT cell. Reduced frequency.	176,178,188
Regulatory B cells (Bregs)	LU	Promote CD4+ T cells conversion to Tregs and mediate TGF-βdependent crosstalk with MDSCs and stroma	163
Liver-res leukocytes	LU	Anti-metastatic effects via IFN-γ induction and elimination of fibrinogen depositions	287

PMN sites were abbreviated as follows: BO = bone, BR = brain, LU = lung, K = kidney, LI = liver, LN = lymph node, PER = peritoneum, SPL = spleen; mo = monocyte; Φ = macrophage; res = resident; TAN = tumor-associated neutrophils; TEN = tumor-entrained neutrophils; AM = alveolar macrophages; IM = interstitial macrophages; ILC = innate lymphoid cells.

Neutrophils have recently garnered attention for their roles in cancer metastasis¹²¹. Chemotactic signals for neutrophil recruitment to the lung PMN include CXCL1/2/5/12, S100A8/A9, and Lin28B^{33,122,123,124,125}, while neutrophil recruitment to the pre-metastatic liver is stromal cell-derived factor-1 (SDF-1)-dependent⁹⁰.

Upon PMN arrival, neutrophils initiate inflammatory responses via leukotriene signaling and ROS generation, the latter strongly impairing NK and T cell function. Neutrophil leukotriene signaling supports metastasis-initiating cell proliferation in BC pre-metastatic lungs¹²⁶. In melanoma, UV-damaged epidermal keratinocytes release HMGB1 to activate TLR4-MYD88 inflammatory signaling in neutrophils, stimulating angiogenesis as well as lung and LN metastasis¹⁰², while Bv8, MMP-9, and S100A8/A9 secreted by lung PMN-infiltrating neutrophils promote cancer cell extravasation and proliferation¹²⁷. In the liver, neutrophils induce PMN formation downstream of SDF-1/CXCR4 signaling⁹⁰. Notably, systemic expansion and polarization of PMN neutrophils in response to IL-17 production by IL-1β-activated γδT cells underscores the cooperation between innate and adaptive immune cells during PMN induction¹²⁸.

Neutrophils support each stage of the PNM evolution via multiple mechanisms¹²⁹. First, upon exposure to tumor-derived EVPs¹³⁰, PMN neutrophils form NETs that sequester CTCs via integrin β1¹³¹, thus promoting colonization. NETs also induce resistance to chemo-, immuno-, and radiation therapies through IL-17 mediated NETosis, driving T cell exhaustion via CEACAM1/TIM-3 interactions¹³² and the PD-1/PD-L1 axis¹³³, as well as by mediating drug detoxification¹³⁴ and NET-dependent epithelial to mesenchymal transition (EMT)¹³⁵. GM-CSF secreted by the primary tumor activates a positive feedback loop, where neutrophil JAK-STAT5β signaling induces synthesis of transferrin, an iron-transporting protein and mitogenic factor for incoming CTCs. Lastly, PMN neutrophils release elastase and cathepsin G that degrade thrombospondin-1, enhancing metastasis¹³⁶. Thus, monocytic and granulocytic cells recruited to the PMN cooperate to generate an inflammatory and immunosuppressive PMN through a feed-forward loop that sustains chemoattractive recruitment.

PMN-promoting Innate Resident Cells

Resident cells in future metastatic organs are subject to active reprogramming that facilitates the acquisition of a pro-metastatic phenotype by recruited immune cells and can be diverse and organ-specific, of epithelial, fibroblast, endothelial, or hematopoietic origin. For example, lung resident club cells are non-ciliated epithelial cells that maintain a chronic homeostatic inflammatory state in the lung via S100A8–SAA3–TLR4 signaling. In a metastatic lung model, TLR4⁺ club cells auto-amplified SAA3, inducing TNF-α production by lung alveolar type II cells and macrophages. Meanwhile, club cell ablation halted the recruitment of CD11b⁺TLR4⁺ cells and spontaneous lung metastasis¹³⁷. Thus, club cell-mediated homeostatic inflammatory imbalance due to constant exposure to airborne microbial endotoxins may contribute to lung metastatic susceptibility.

Critically, tissue resident macrophages, such as lung alveolar and interstitial macrophages¹³⁸, liver Kupffer cells¹³⁹ or brain microglia¹⁴⁰, are not only rheostats regulating inflammation in their home organs, but also central effectors of PMN formation^{44,75,119,141}. In the omentum PMN, CD163⁺Tim4⁺ resident macrophages were pivotal for ovarian cancer metastatic spread¹⁴². Both lung resident macrophages and endothelial cells upregulate MMP-9 secretion in response to primary tumor-secreted VEGF-A, thereby promoting lung metastasis by ECM remodeling¹⁰. CYP4A⁺ lung macrophages also regulate PMN formation as depletion of this population significantly decreased the number of circulating VEGFR1⁺ myeloid cells and the levels of MMP-9, S100A8, and FN in pre-metastatic lungs¹⁴³. In addition to soluble factors, tumor-derived EVPs also reprogram PMN tissue resident interstitial macrophages towards an immunosuppressive function via de novo PD-L1 expression¹¹⁹.

Liver-resident Kupffer cells are first-line defense macrophages/phagocytes surveilling the early PMN, but can be corrupted to exert pro-metastatic functions^144,145. Human liver-tropic pancreatic cancer EVPs containing integrin αvβ5 localize to fibronectin-rich areas and upregulate S100P and S100A8 in Kupffer cells via Src activation, thereby promoting S100-dependent inflammation¹⁹. Pancreatic cancer-derived EVPs preferentially uptaken by Kupffer cells trigger TGF-β secretion and subsequent fibronectin upregulation by hepatic stellate cells, generating a fibrotic liver microenvironment supportive of BMDC recruitment and PMN establishment⁴⁴. EVP-packaged ANGPTL1 in turn reprogrammed the Kupffer cell secretome by suppressing MMP-9 production via JAK2-STAT3 signaling in colorectal cancer (CRC) models¹⁴⁶. Phagocytosis of hypoxia-induced EVP miR-135a-5p in the liver PMN promoted MMP-7 production and CD30-TNF receptor–associated factor 2-p65 signaling in Kupffer cells, exerting immunosuppression by reducing TNF-α and IL-2 serum levels and intrahepatic CD4⁺ T cell numbers¹⁴⁷. These studies suggest that Kupffer cells, due to their intrinsic function, are highly susceptible to reprogramming towards tumor-supporting phenotypes.

Microglia are brain-resident innate immune cells that respond to TDSF and EVPs^75,141,148 by adopting an M2-like phenotype and releasing pro-tumor cytokines (e.g., IL-4, IL-10, IL-13, TGF-β) typically involved in neurovascular network repair¹⁴⁹. M2-like microglia impair T cell proliferation in the brain PMN by upregulating PD-L1 expression. Loss of X-inactive–specific transcript, which correlates clinically with brain metastasis in BC patients, led to the production of EVP miRNA-503 responsible for M2 microglia polarization in brain metastatic BC models¹⁴¹. In lung cancer models, EVP uptake by brain microvascular endothelial cells induced endogenous DKK-1 secretion, which activated and skewed M1-like microglia dominant in the PMN towards a pro-tumorigenic M2-like one, concurrently upregulating both M1 (IL-1β, iNOS, TNFα, and TLR4) and M2 (ARG1 and CD206)¹⁴⁸ markers. Importantly, the uptake of brain tropic human BC-derived EVPs containing CEMIP by brain endothelial cells and microglia induced perivascular inflammation and vascular remodeling^19,75. Thus, PMN-resident macrophages establish a sustained pro-tumor inflammatory milieu at a secondary site and reprogram recruited immune cells, sealing the fate of tumor cells seeding the PMN.

Dendritic cell antigen-presenting functions are also altered in the PMN. Lung DC function is suppressed by PGE2 secreted by resident fibroblasts at steady state but is reinforced by tumor-secreted IL-1β, supporting PMN formation. Fibroblast-specific deletion of COX-2, the enzyme that produces PGE2, reverses DC suppression, leading to enhanced immune activation and reduced lung metastasis in multiple BC models¹⁵⁰. Consistent with these findings, COX-2 expression in DCs promoted LN metastasis, which was markedly suppressed by COX-2 inhibition¹⁵¹.

These diverse myeloid cell functions highlight their key roles in PMN formation. Data on involvement of other innate immune populations, such as basophils, eosinophils, and mast cells, in the PMN are limited, but warrant further exploration, especially in barrier tissues, such as the lung, in the context of chronic inflammation induced by environmental exposures, infections, and aging.

The PMN complement system

The complement system functionally bridges innate and adaptive immunity, clearing pathogens and modulating host immune responses through antibody-mediated activation or anaphylatoxins^152,153. Anaphylatoxins (C3a, C4a, C5a) recruit myeloid cells that express their cognate 7-transmembrane G-protein-coupled receptors, thereby regulating myeloid cell chemotaxis, migration, and phagocytosis¹⁵². Recent studies highlighted the role of C3a-C3aR crosstalk between activated CAF and macrophage/neutrophil chemotaxis during metastatic progression^154–156. In the pre-metastatic lung of BC models, pro-tumor inflammatory Th2 cytokines induced sustained C3 expression in mesenchymal stromal cells in an IL-4/13-STAT6-dependent manner, enhancing metastasis through recruitment of C3aR⁺ neutrophils and induction of NETosis¹⁵⁷. Importantly, the complement system may also account for acquired resistance to chemotherapy early during metastasis through C3a and C5a crosstalk between resident fibroblasts and recruited myeloid cells that restricts T cell function¹⁵⁸. These studies highlight the involvement of the complement system in PMN formation and stress the need to investigate the therapeutic potential of targeting specific components of the complement cascade to hinder PMN establishment.

Adaptive immune landscape

The systemic immune surveillance exerted by adaptive immune cells, such as cytotoxic T lymphocytes (CTLs; i.e., CD8⁺ T cells) and NK T (NKT) cells restricts the survival of DTCs and regulates metastatic colonization and outgrowth^159–161. By recruiting and promoting the differentiation and expansion of tumor-infiltrating lymphocytes (TILs) into Treg cells, T helper 17 (T_H17) cells, γδT cells, and regulatory B (Breg) cells in the primary tumor and potentially at future metastatic sites, cancer cells avert anti-tumor immunity, thus initiating the metastatic cascade¹⁶² (Table 1).

The most prevalent adaptive immune cells in the PMN are Tregs, which have regulatory and suppressive functions. Treg recruitment to the PMN relies on a complex interplay between primary tumor-derived signals, systemic Breg-mediated secretion of TGF-β, crosstalk with immunosuppressive myeloid populations, and PMN stroma-derived signals^162–164. In turn, Treg enrichment maintains an immunosuppressive environment, supporting PMN progression. Although the complex Treg-mediated PMN interactions remain to be explored, they shield DTCs from immune attack by inhibiting T cell proliferation, cytokine production, and cytotoxic function in several cancers^162,165,166. Moreover, local and systemic TGF-β and MDSCs cooperate to expand Tregs in distant organs, such as draining LNs, liver, lung, and bone marrow¹⁶⁷. In the bone marrow PMN, Treg enrichment and maintenance depends on CXCL12 and CXCR4 signaling⁹⁹. A recent study exploring immune cell dynamics in the rhabdomyosarcoma lung PMN¹⁶⁸ reported a significant increase in exhausted, dysfunctional PD1^hiCD44^int T cells at the expense of CD4⁺ helper T cells¹⁶⁹. Interestingly, the PMN-promoting role of dysfunctional Th17 cells was described in BC murine models, where production of RANK ligand, a major regulator of osteoclastogenesis, and IL-17F by these cells induced osteolysis in a manner proportional with the metastatic potential of the tumor model ¹⁷⁰.

Immune surveillance is achieved by the three main lymphocyte lineages: αβT cells (CD4⁺ and CD8⁺), γδT cells, and B cells¹⁷¹. While αβT cells and B cells exert both systemic and local effects, γδT cells (mostly CD4⁻CD8⁻) are present at lower frequencies in the circulation but are highly abundant in tissues, especially at sites prone to metastases¹⁷². Since these cells are not MHC-restricted, they contribute to an unconventional form of immunosurveillance and have different roles depending on their molecular make-up and residence. As suggested by pre-clinical and clinical data, γδT cells may hold the key to understanding mechanisms of immunosenescence, anti-tumor immunity, and tumor regression. Future discoveries will unveil the role of these cells in PMN initiation and will assist the development of therapeutic strategies for preventing PMN evolution (e.g.,through γδT cell-mediated tissue protection from fibrosis).

Although humoral immunity is critical in modulating tumor progression, few studies have investigated the role of B cells in the PMN. A novel subset of tumor-evoked Bregs, characterized by constitutively active STAT3, induced TGF-β-dependent conversion of resting CD4⁺ T cells into Tregs, which protected metastatic cells from host immune responses¹⁶³ (Table 1). Moreover, the primary tumor educated pre-metastatic lungs to express TARC/CCL17 and MDC/CCL22 chemokines, which attracted C-C motif chemokine receptor 4 (CCR4)-positive tumor and immune cells ¹⁶⁴. PMN infiltration by CCR4⁺ Tregs induced NK cell apoptosis via secretion of β-galactoside-binding protein, supporting metastatic cell seeding¹⁶⁴. The role of B cells and immunoglobulins (Ig) in LN metastasis was recently investigated in spontaneous BC models. Primary tumors induced B cell accumulation in pre-metastatic tumor-draining LNs and production of pathogenic IgG that targeted HSPA4, a tumor cell glycosylated membrane protein, which led to integrin β5 activation. This in turn triggered Src-NF-κB activation in cancer cells, which further promoted CXCR4/SDF1α-mediated metastasis¹⁷³.

Local and systemic factors contributing to PMN formation

Gut and tumor microbiome contributions to PMN formation

Groundbreaking discoveries in the past decade highlighted that the microbiome shapes host immune responses, modulates inflammation, and orchestrates anti-tumor immunity¹⁷⁴. During CRC metastasis, the combined effects of gut dysbiosis and intestinal inflammation extend to the liver, generating a pro-inflammatory and immunosuppressive hepatic microenvironment^175–178. Environmental factors, such as a high-fat diet, also promoted lung PMN formation in murine models of BC and melanoma, an effect mediated by gut dysbiosis via M1 polarization, TNF-α and CCL2 secretion by gut macrophages, and subsequent recruitment of S100A8/A9-producing MDSCs to future metastatic sites ¹⁷⁹. Similar to PMN vascular leakiness, gut vascular barrier disruption is an essential step for CRC resident Escherichia coli bacteria translocation to the liver, PMN formation, and metastasis ¹⁸⁰. These observations spurred further studies that showed that each tumor type harbors its own microbiome, dwelling intracellularly in both tumor and immune cells, although systematic comparisons with pre-metastatic and metastatic niches have yet to be made ¹⁸¹.

Among commensals turned onco-pathobionts, Fusobacterium nucleatum, a biomarker of inflammatory bowel disease, associated with poor prognosis in CRC patients^182,183. F. nucleatum, found to reside not only in primary patient-derived xenografts (PDXs) but also within liver metastatic cells ¹⁸⁴, could evade the immune system during CRC progression by shedding its wall and invading cancer cells, thus modulating host E-cadherin-β-catenin signaling¹⁸⁵. Studies in immunocompetent mice suggest that F. nucleatum may drive liver PMN formation by promoting MDSC and Treg infiltration, decreasing NK and CD8⁺ and Th17 T cell frequencies and elevating CXCL1 levels, all defining features of CRC liver PMN^186,187,178.

Importantly, gut dysbiosis and inflammation alter microbial metabolites, resulting in local and systemic metabolic changes. Liver metastasis was inhibited by secondary bile acid conversion by gut commensal bacteria, through upregulation of CXCL16 in liver sinusoidal ECs, and immunosurveillance by recruited CXCR6⁺ NKT¹⁷⁷. However, capsaicin-induced disruption of gut vascular barrier and altered bile acid metabolism reduced NKT cell infiltration in the liver and increased inflammation (IL-12, TNF-α, IFN-γ) that facilitated PMN formation¹⁸⁸. Moreover, sodium butyrate rescued gut dysbiosis, significantly reducing liver metastasis in pre-clinical models of CRC^176,177. These findings suggest that remodeling of the liver immune microenvironment by gut-derived metabolites may occur prior to tumor dissemination and shape the liver PMN in gastrointestinal cancers.

The role of local and systemic metabolism in PMN induction

The role of glucose metabolism in PMN formation

During metastatic progression in cancer patients, the differential availability of circulating nutrients and metabolites is marked by accumulation of glucose metabolism byproducts, such as lipid, lactate, pyruvate, and glutamate, with a concomitant decrease in glutamine, a precursor of gluconeogenesis¹⁸⁹, which together contribute to systemic immune suppression. For instance, lactate accumulation impairs NK cell function, leading to MDSC and Treg expansion, as well as M2 macrophage polarization¹⁹⁰. Proteomic characterization of early lung PMN changes in BC identified metabolic changes in glucose metabolism, while late PMN changes revealed an increase in Ca²⁺ signaling/sensing, likely due to extensive lung tissue remodeling, ECM reorganization, and inflammatory signaling¹⁹¹. Despite increasing interest in metabolic rewiring during cancer progression and metastasis, few studies have investigated PMN immunometabolic rewiring, with those that did identifying EVPs as the main drivers of systemic and local metabolic reprogramming.

EV-mediated immunometabolic reprogramming of resident macrophages in the lung PMN may promote immunosuppression via several mechanisms¹¹⁹. On one hand, tumor-derived EVPs triggered HMGB-1-TLR2-NF-kB signaling in lung PMN interstitial macrophages leading to immunometabolic reprogramming toward an immunosuppressive phenotype, while NOS2 induction shifted metabolism from mitochondrial oxidative phosphorylation towards glycolysis. On the other hand, increasing lactate levels further fueled HIF-1α activation downstream of NF-kB, increasing PD-L1 expression on the surface of macrophages¹¹⁹.

Notably, BC can alter systemic glucose metabolism and modulate the metabolic environment of distant organs favoring PMN formation. BC-derived EVPs limit glucose availability and utilization by PMN resident cells (e.g., lung fibroblasts, brain astrocytes, and neurons) via miR-122-mediated downregulation of pyruvate kinase and glucose transporters (i.e., GLUT-1) to instead provide this nutrient for the outgrowth of infiltrating tumor cells¹⁹². Enrichment in EVPs, and particularly exomeres in metabolic enzymes and metabolites⁵⁰ indicates they may play a role in local and systemic metabolic reprogramming of target cells in distant organs.

Lipid metabolism and metabolites involved in PMN formation

Recent studies suggest that abnormal cancer cell and cancer-associated stroma lipid metabolism fuel tumor progression^193–195. Increased fatty acid accumulation and subsequent oxidation represent major ATP sources and one of the key metabolic adaptations associated with metastasis. For example, acetyl-coenzyme A carboxylase alpha stabilizes fatty acid synthesis in BC lung PMN¹⁹⁶, while fatty acid oxidation (FAO) is involved in sequential steps of the pre-metastatic cascade and LN metastasis¹⁹⁷. A recent study documented palmitate enrichment in pre-metastatic lungs and lung interstitial fluid of murine models and patients with BC without lung metastasis, and that this fatty acid was used for acetyl-CoA synthesis¹⁹⁸. Thus, FAO and related lipid metabolic processes might be involved in reprogramming PMN-associated immune cells, contributing to immunosuppression.

Lipid metabolism also favors eicosanoid synthesis, converting arachidonic acid (AA) to PGE2, which is central to PMN formation. PGE2 metabolism promoted immunosuppression and immune escape in a murine bladder carcinoma model, upregulating PD-L1 on BMDCs¹⁹⁹. Notably, lung PMN pro-inflammatory S100A8/A9 chemokines complex with AA to facilitate its uptake by endothelial cells via CD36²⁰⁰. Thus, PGE2 and S100A8/A9 cooperate, linking inflammation and metabolism in the pre-metastatic lung, suggesting that inhibiting BMDCs PGE2 metabolism may effectively target their immunosuppressive function. Expression of CYP4A, another enzyme involved in AA metabolism, in macrophages was also associated with M2 polarization, pro-tumor cytokine production, and recruitment of VEGFR1⁺ myeloid cells to the lung PMN¹⁴³. Given that high cholesterol, obesity, and high fat diets are major risk factors for BC recurrence and metastasis²⁰¹, it is not surprising that cholesterol and sterol hydroxylation in BC models resulting in high levels of pro-metastatic 27-hydroxycholesterol directly altered myeloid cell function, increased γδT cell frequency, and neutrophil infiltration, and suppressed cytotoxic CD8⁺ T cells in the lung niche²⁰¹.

Nutrient, oxygen, and mineral sensing in immune cell function

Cancer cells are often exposed to metabolically challenging environments and their metabolic plasticity enabled by nutrient-, oxygen-, and cation-sensing molecules (e.g., Ca²⁺, Fe²⁺) ensures their adaptation. The lack of studies on the metabolic crosstalk in the PMN raises the question regarding the roles of Fe²⁺ and Ca²⁺ sensing in PMN immunometabolism, and whether they may drive inflammatory and immunosuppressive processes, likely through the S100 Ca²⁺-sensing proteins and ferroptotic cell death. Recent studies suggest a link between ferroptosis and PMN formation, as CD36-mediated AA uptake by CD8⁺ T cells resulted in lipid peroxidation and ferroptosis, followed by impairment of the anti-tumor immunity²⁰². Among immune cells, T cells are exquisitely dependent on Ca²⁺ signaling downstream of TCR activation for effector functions, as well as self-tolerance programs and homeostasis and Ca²⁺ sensing dysregulation results in autoimmunity, inflammation, and immunodeficiency^203,204. Intriguingly, the late lung PMN is dominated by upregulation of Ca²⁺ signaling via autocrine and paracrine chemotactic loops orchestrated by the S100 Ca²⁺ sensor proteins¹⁹¹. Thus, whether competition for Ca²⁺ in the PMN and Ca²⁺ sequestering by PMN stromal cells limit Ca²⁺ availability to effector PMN immune cells remains an important question.

Innervation in cancer and its implications for PMN formation

The role of cancer innervation and stress in PMN evolution

The sympathetic nervous system (SNS) mediates acute stress responses following hypothalamic-pituitary-adrenal (HPA) axis activation²⁰⁵ by modulating adrenergic and glucocorticoid signaling²⁰⁶ and supports cancer progression in response to chronic stress²⁰⁷. The immunosuppressive effects of stress^205,208,209, including suppression of NK²¹⁰ and CD8⁺ T cells²¹¹, and MDSC accumulation²¹² extend to the lymphatic system, as chronic stress induces remodeling of tumor and peritumor lymphatic networks via tumor-derived VEGF-C and TAM-mediated COX-2-PGE2 inflammatory signaling to provide lymphatic dissemination routes for tumor cells²¹³.

Recent studies identified stress-related catecholamine as a primary factor in regulating pre-metastatic to metastatic evolution in the lung niche²¹⁴ and favoring colonization by BC CTC. Pharmacological inhibition of sympathetic nerve function or blocking the β-adrenergic signaling markedly suppressed stress-induced lung metastasis along with the reduction of circulating monocytes and lung resident macrophages²¹⁴. Critically, a clinical trial in patients with early BC assessing the efficacy of combined blockade of β-adrenergic receptors and PGE2-reducing COX-2 inhibitors during the perioperative period, when stress flares due to both the surgical procedure and psychological stressors, revealed reduced primary tumor STAT3 levels, lower circulating IL6, C-reactive protein (CRP) and monocyte levels and increased NK cell activation²¹⁵. Given these data, an outstanding question is what are the potential contributions of stress and cancer innervation to PMN generation, and more specifically, to modulating immune function in the PMN?

Neuro-immune interactions in the PMN

The effects of neuro-immune interactions on the PMN were first explored in the bone marrow (BM), a highly innervated tissue where the SNS regulates the physiological HSC niche²¹⁶. Indeed, stress-induced or pharmacological sympathetic activation in bone metastatic BC increased RANKL levels in the bone microenvironment, induced bone lesions, altered the cytokine profile of the bone PMN and rendered the BM microenvironment more receptive for CTC colonization²¹⁷. Direct neuro-immune interaction of SNS fibers with BMDCs also favor CTC colonization by disrupting Sema3A-mediated neuron-macrophage interactions in healthy lungs. Chemical sympathectomy inhibited tumor angiogenesis, prevented bone loss, and decreased lung mo-MDSC recruitment and metastatic burden, whereas specific adrenoreceptor beta 2 (ADRB2) inhibition reduced S100A8 levels and CTC recruitment to the lung²¹⁸. These findings ascribe cancer innervation and sympathetic activation an active role in PMN development, thereby providing a rationale for blocking SNS signaling in cancer patients by repurposing FDA-approved drugs (e.g., βAR or β-blockers) as anti-PMN therapies.

Clinical implications of the PMN for cancer detection and management

Systemic detection of PMN formation in cancer patients

While pre-clinical cancer models furthered our understanding of PMN biology, overcoming challenges associated with PMN detection, characterization, and therapeutic targeting in patients is essential to translating these advances into improved patient outcomes. Several mediators of PMN formation, such as ECM-remodeling molecules, detected in patient liquid biopsies, correlated with metastatic outcome. High MMP serum levels correlated with metastasis and poor survival in gastric²¹⁹, breast²²⁰ and ovarian cancer²²¹, while elevated systemic levels of TIMP-1 associated with liver metastasis and poor prognosis in colon and pancreatic cancers^90,222. Similarly, ANGPTL2 was elevated in osteosarcoma patients and promoted lung PMN formation³³.

Blood-based liquid biopsies could reveal distant PMN formation and inform on metastatic progression through EVP cargo. In head and neck cancer patients, PD-L1 on EVPs, but not free-circulating PD-L1, associated with disease progression⁶². In pancreatic cancer, levels of MIF, a cytokine associated with liver PMN formation, were markedly higher in EVPs isolated from the plasma of stage I PDAC patients who later developed liver metastasis compared to patients without metastasis⁴⁴. Comparison of EVP-packaged pyruvate kinase M2 in localized versus metastatic prostate cancer patients identified it as a potential marker of bone pre-metastatic niche formation ²²³. Unbiased proteomic profiling of EVPs from tumor explants and plasma of patients identified pan-cancer EVp markers (e.g., VCAN, tenascin C, and THBS2), as well as tumor-type-specific EV proteins that, in stage I patients, may allow PMN biomarker identification¹¹⁰ to improve early detection.

Urine EVP biomarkers may also indicate PMN status not only in the urinary system but also in other cancers such as BC^224,225. Given the accessibility of urine samples, this biofluid is a convenient liquid biopsy source. Human lymphatic fluid may offer unique insight into systemic immune responses: in early-stage melanoma, immune-modulating S100A9 EVP cargo correlated with IHC staining of non-metastatic LNs⁴⁶. Large studies are however required to evaluate correlations between lymphatic fluid biomarkers and metastasis.

Profiling PMN-associated immune cells

Profiling the frequency, phenotype, and signaling pathways activated in BMDCs, MDSCs, neutrophils, and T cells, can offer additional insight into PMN evolution in cancer patients. MDSC frequencies correlated with cancer stage and metastatic burden in BC patients²²⁶. In melanoma patients, STAT3 activation in myeloid cells inversely correlated with the number of CD8⁺ T cells in pre-metastatic LNs²²⁷. CD68⁺ myeloid cells exhibiting elevated S1PR1 and p-STAT3 levels accumulated in tumor-free LNs of high-risk prostate cancer and melanoma patients³⁵. Similarly, CYP4A⁺ TAM accumulation in primary tumors correlated with reduced overall and recurrence-free survival in BC patients¹⁴³. Lastly, circulating neutrophils in HCC patients bearing lung or liver metastasis showed enhanced NET formation compared to non-metastatic patients or controls²²⁸. Interestingly, while NET deposition occurred in non-metastatic and metastatic LNs, high LN NETs correlated with reduced survival, even in the absence of apparent metastasis²²⁹, suggesting NETosis is a LN PMN biomarker predictive of patient outcome. Thus, patient validation of PMN immune hallmarks identified in pre-clinical studies highlights their conserved role and warrants clinical trials to establish their prognostic power.

PMN tissue biopsies during the clinical workup

Most clinical evidence of PMN reprogramming is based on sentinel LN (SLN) samples from cancer patients as these tissues are routinely resected for staging. LNs are a common initial site of metastasis in most solid tumors and reliable indicators of patient outcomes. A PMN hallmark, VEGFR1⁺ myeloid clusters were detected in cancer-free pre-metastatic LNs of patients^2,230 and their abundance predicted prostate cancer recurrence, marked by the rise of prostate-specific antigen (PSA) blood levels, and metastatic progression²³⁰. ECM-remodeling enzyme MT1-MMP and LOX expression in LN macrophages correlated with lymphatic metastasis in oral cancer²³¹. Comparison of non-metastatic, metastatic SLNs, and non-metastatic distal LNs in early-stage cervical and oral squamous cell carcinoma revealed that non-metastatic and metastatic SLNs shared similar features absent from distal LNs. Specifically, high lymphatic vessel density was observed in both SLN groups, suggesting this characteristic as a biomarker of LN metastasis^232,233. Similarly, high endothelial vessel remodeling in BC pre-metastatic LNs associated with accumulation of CCL21⁺ lymphocytes²³⁴. Moreover, primary BC tumor CYP4A⁺ TAM levels correlated with several PMN markers, including VEGFR1⁺ cell infiltration, S100A8 expression, and fibronectin deposition in uninvolved LNs¹⁴³. Nevertheless, in patients, LN PMN-associated immunosuppression is mainly uncharted²³⁵. Thus, despite routine collection and analysis of LN, it is still unknown which PMN LN signatures can risk-stratify patients or predict their potential involvement in LN and distant metastases, highlighting the need of extensive profiling of these patient tissues.

Clinical evidence of bone and lung PMN formation also exists. High primary tumor LOX expression correlated with metastatic disease and poor survival²³⁶,as well as with bone tropism and osteolytic lesions in ER-negative BC patients⁸⁷. A comprehensive analysis of lung autopsy samples suggested that MMP-9 initiates lung PMN formation in patients, as high MMP-9 levels were detected in lung endothelial cells of uninvolved lungs from patients with nine distinct cancers ¹⁰.

Another PMN site that has garnered attention is the omentum, a common site of ovarian cancer metastasis. Adhesion molecules, such as integrin α2, mediate cancer cell adherence to collagen-rich omentum, initiating ovarian cancer metastasis²³⁷. A correlation between integrin α2 expression and altered collagen expression in pre-metastatic omentum predicts poor outcome in patients²³⁷. Moreover, neutrophil recruitment promotes omental PMN formation in early-stage, high-grade serous carcinoma patients. Neutrophils accumulated around blood vessels and high levels of neutrophil elastase were detected in omental tissues from cancer patients but not healthy subjects²³⁸. EVP proteomic analysis revealed an omental PMN signature characterized by IL-6 and IL-8 expression²³⁹.

While liver, bone, and especially brain PMN patient tissues are challenging to obtain, their prognostic potential is immense. Our multi-omics characterization of pre-metastatic livers collected during primary tumor resection from pancreatic cancer patients with localized disease revealed the presence of various PMN hallmarks, such as inflammation, metabolic dysfunction, ECM changes, and immunosuppression. Importantly, liver neutrophil infiltration at the pre-metastatic stage correlated with metastatic progression in these patients at 8-year follow-up²⁴⁰, a finding that could inform treatment course. Such biomarkers may predict not only the presence of a PMN, but also the specific site where these processes occur, thus allowing tailored interventions. The development of novel technologies that require less material may enable clinicians to perform analyze fine needle biopsies, which will facilitate future studies.

Imaging tools for clinical PMN detection

Monitoring PMN progression non-invasively can also be achieved via imaging, either by repurposing current methods or by developing novel, more sensitive imaging techniques to detect PMN-related features. A few pre-clinical and clinical studies reported such techniques; however, they have yet to be adopted for routine patient screening.

Imaging the PMN in the clinic

Several pre-clinical imaging techniques were developed to detect PMN-associated changes (Table 2(. These studies underscore the potential such applications have in monitoring PMN formation while highlighting the need for additional imaging techniques, focusing on key high-translational determinants, such as integrins. Multiple imaging methods, including fluorescent antibodies²⁴¹ and positron-emission-tomography (PET) scans^242,243, enable visualization of integrin localization, activation, and function. In patients, noninvasive imaging of αvβ3²⁴⁴ and αvβ6²⁴⁵ integrins was well tolerated and revealed significant uptake in primary tumors and multiple metastatic foci. Such methodologies, with improved sensitivity, may hold great promise in PMN detection.

Table 2.

Pre-clinical imaging of the PMN

Target molecule	Imaging technique	Reference
S100A8/A9	Single-photon emission computed tomography (SPECT)	302
VLA-4 (α4β1 integrin)	High-affinity positron emission tomography (PET)	298
VEGFR3	Bioluminescence of Vegfr3Luc	303
ECM	Raman spectroscopy	304

In the clinical setting, several studies employed radiomic methods to identify unexplored PMN-related features in lungs of BC patients^246,247 or in liver of pancreatic²⁴⁸ and rectal²⁴⁹ cancer patients. By comparing computed tomography (CT) scans of metastatic and non-metastatic regions at two time points, prior to and after detection of evident metastasis, five features that distinguish the regions prior to metastasis formation were identified, suggesting this imaging method could clinically detect lung PMN formation²⁴⁶. Another study analyzed lung scans of BC patients with or without lung metastasis and healthy controls. Increased lung density, indicative of inflammation, was evident in metastatic and non-metastatic, but not control lungs. Notably, enhanced lung density was detected prior to the clinical diagnosis of pulmonary metastasis and correlated with disease score, demonstrating the potential of CT scans in clinically diagnosing the lung PMN²⁴⁷.

Similarly, a retrospective study in pancreatic cancer patients aimed to predict early liver metastasis development (within 6 months from primary tumor excision) based on contrast-enhanced CT before primary tumor resection. Analysis of pre-operative enhanced CT scans of livers from a total of 688 resected PDAC patients found a predictive ability of a multifactored model using radiomic data combined with clinicopathological variables. Six radiomic texture features that reflect hepatic parenchymal heterogeneity and fibrosis²⁵⁰ were significantly elevated in patients that developed early liver metastasis. Combined with tumor size and differentiation, this analysis identified patients at high risk for early liver metastases²⁴⁸. Retrospective analysis of contrast-enhanced CT radiomics predicted liver metastasis within two years from primary tumor resection in rectal cancer patients with no pre-operative evidence of liver metastases, based on a mixed model including both radiomic data and clinical parameters²⁴⁹. Another imaging method assessed in an ongoing clinical trial is the FibroScan, an ultrasound-based transient elastography. This study aims to identify pre-metastatic liver biomarkers in stage I-IV colon cancer patients, correlating the EV cargo of liver biopsies with FibroScan imaging (NCT03432806), as measures of liver fibrosis (tissue stiffness) and steatosis (fatty liver), indicative of pre-metastatic modifications.

Therapeutic strategies targeting the PMN

Risk of promoting PMN with current therapeutic approaches

Epidemiological studies have shown that current cancer therapies, including surgery, radiotherapy, and chemotherapy may promote distant metastases by triggering BMDC-mediated repair resembling wound healing and systemic inflammation²⁵¹. Several chemotherapies support PMN formation systemically by eliciting release of tumor-derived EVPs and locally by activating CAFs and endothelial cells as well as by recruiting MDSCs, ultimately impairing T cell function^{158,251–253}. Similarly, radiotherapy promotes the recruitment of pro-tumor TAMs and MDSCs in various cancers²⁵⁴, while NETosis is a common consequence of both chemotherapy and radiotherapy^134,255. Notably, primary tumor resection releases tumor cells in the circulation, increasing the frequency of CTCs^256–259 thus promoting metastasis, should these cells encounter a favorable niche generated by adjuvant chemotherapy, radiation, and pre-existing inflammation.

Changing the status quo: targeting PMN formation

To date, tailored therapies targeting PMN formation are nonexistent and metastasis-directed therapies are scarce (Fig. 3). Among existing therapies, cytoreductive surgery may also hinder PMN formation and enhance patients survival by blocking the systemic effects of the primary tumor^260,261, including the recruitment of tumor-supporting innate immune subsets.

The most promising therapy for metastatic cancers is immune checkpoint blockade (ICB) targeting T cell checkpoints (e.g., PD-1, PD-L1, CTLA-4); however, therapeutic success has been limited to cancers with high mutational burden and hindered by development of therapy resistance. Recent therapeutic advances aim to overcome ICB failure by targeting new inhibitory checkpoint molecules on T cells or NK cells as well as co-stimulatory molecules to enhance endogenous anti-tumor immunity²⁶². Moreover, potential targets for interfering with pro-metastatic immune cell function and PMN formation have emerged. For instance, blocking immune cell recruitment to the PMN (e.g., antibodies targeting CCL2 for BM monocytes, CCR4 for Tregs, CXCR2 for neutrophils, CSF-R1 for overall myeloid cell infiltration) or reverting macrophages to M1, anti-tumor phenotypes, could impair PMN formation. Selective targeting of IL-2 receptors on Tregs could overcome immune escape by compromising their survival while antibody blocking of CD47 “don’t eat me” signals upregulated by metastatic cancer cells may enhance phagocytosis²⁶². Using these drugs as adjuvants or as monotherapy could provide new therapeutic approaches for treating PMN and metastases.

Reversing PMN immunosuppression

Therapeutic strategies targeting PMN formation would represent a paradigm shift in treating cancer, by preventing or eradicating metastases before they manifest clinically. This could be achieved by interfering with PMN immunosuppression by targeting cells with crucial roles in PMN formation or by activating anti-tumor immunity (Box 2), thus preventing seeding and outgrowth of CTCs at these distal sites (Fig.3).

Co-opting anti-tumor immunity

Targeting S100A4 that perpetuates MDSC-driven PMN inflammation with the 6B12 antibody reduced metastatic burden by restoring Th1 anti-tumorigenic phenotype in the lung PMN²⁶³. Future studies should investigate the translational immunomodulatory potential of the 6B12-based therapy in cancer patients. Kaplan and colleagues pioneered a promising proof-of-principle pre-clinical study that could rebalance the core immunosuppressive PMN environment and reduce metastatic burden. Genetically engineered myeloid cells (GEMys) were used for local delivery of immunomodulatory cargo (IL-12) to the lung PMN in a rhabdomyosarcoma model. This approach activated APCs, and T and NK cells¹⁶⁸, extending survival of 90% of mice for over one year. This study provides a rationale for IL12-GEMy as adjuvant therapy after surgery and in combination with chemotherapy for unresectable tumors, highlighting the potential of engineered myeloid cells as promising reconfigurable cell-based therapies that boost anti-tumor immune responses and reverse immunosuppression in the PMN.

Normalizing NK cell function

NK cells are promising cellular targets for immunotherapies given their critical role in controlling metastatic dissemination. Dormant DTCs that remain undetectable after primary tumor resection offer a therapeutic window to prevent their reactivation. In BC models, NK cells enforce DTC dormancy in the liver, whereas hepatic stellate cells suppress NK cell proliferation allowing cancer cell outgrowth. Adjuvant IL-15-based immunotherapy normalized the NK cell pool sustaining DTC dormancy via IFN-γ signaling, preventing hepatic metastases and significantly prolonging survival without affecting dormant cells²⁶⁴. Thus, IL-15 or IFN-γ therapy, or CXCR4 inhibition in early-stage cancer could prevent PMN formation by normalizing the NK cell pool and rescuing their anti-tumor function.

Normalizing vascular permeability and anti-tumor immune cell trafficking in the PMN

Recent attempts to reverse PMN-promoting vascular permeability employing a protein fusion between LIGHT (a TNF family cytokine) and a vascular targeting peptide (LIGHT-VTP) that homes to angiogenic blood vessels in primary tumors normalized intra-metastatic blood vessels and increased infiltration by GZMB⁺ effector T cells that further sensitized lung metastases to antiPD-1 checkpoint inhibitors²⁶⁵.

Targeting innate cell-mediated immunosuppression in the PMN

Epigenetic therapy targeting MDSC trafficking

Following surgical resection of primary lung, breast, and esophageal cancers, low-dose adjuvant epigenetic therapy (LD-AET) selectively disrupted MDSC PMN recruitment, reprogramming them towards an interstitial macrophage-like phenotype²⁶⁶, thus reversing immunosuppression and improving disease-free and overall survival²⁶⁷. A recent randomized, phase-II adjuvant LD-AET clinical trial in patients with stage-I NSCLC undergoing tumor resection, reduced tumor recurrence twofold²⁶⁷.

Obstructing myeloid cell recruitment by the PMN ECM

Disrupting MDSC recruitment to the PMN may represent another viable strategy to reverse the effects of PMN immunosuppression. As ECM regulators and docking sites for BMDCs, PMN-targeting LOX proteins hold promise, especially in preventing osteolytic bone PMN lesions⁸⁶ and could benefit BC and renal cell carcinoma patients who exhibit LOXL2 overexpression²⁶⁸. In clinical trials, achieving high serum concentrations of vitamin derivatives, such as all-trans retinoic acid, decreased peripheral MDSCs and improved T cell immune responses in cancer patients and thus could block PMN formation²⁶⁹.

Targeting NET formation

Neutrophils, the most abundant cell type in the human blood, have remarkably plasticity and represent our first defense against blood pathogens through NETosis²⁷⁰. Although cancers employ NETs to protect themselves by inducing cytotoxic T cell exhaustion in an IL-17-dependent manner^132,133,271, indirect NET targeting via IL-17 blockade potentiated responses to ICB²⁷¹. Moreover, pharmacological inhibition of IL-17/IL-17RA signaling using monoclonal antibodies in combination with ICB, proved highly effective in enhancing CD8⁺ T cell recruitment and cytotoxicity in pancreatic cancer, normally refractory to ICB²⁷¹. A humanized mAb against IL17A is already FDA-approved, suggesting that anti–PD-1+anti-IL-17 combination therapy could be easily translated to the clinic²⁷¹. Additional strategies blocking NET formation include targeting PAD4, the enzyme responsible for NET formation, inducing NET degradation via DNAse I treatment and neutrophil elastase inhibition (NEi) could also be combined with ICB in patients otherwise resistant to immunotherapy^238,134.

Anti-thrombotic therapies that prevent PMN formation

Anti-thrombotic agents could also prevent PMN formation and metastatic progression. A calcium phosphate nanoparticle nanotherapeutic containing an anti-inflammatory agent and decorated with an anti-thrombotic agent, low molecular weight heparin could target systemic metastases²⁷². In a BC model, administration of the nanotherapeutic as an adjuvant to paclitaxel chemotherapy or surgery reduced tumor foci by 10–15-fold and increased survival. Moreover, the nanotherapeutic agent targeted both the primary tumor and the lung pre-metastatic site where it markedly reduced expression of PMN proteins²⁷².

Vaccine-based anti-tumor immunotherapies

The implementation of therapeutic strategies based on cancer vaccines in patients who are at risk for developing metastasis, could boost systemic immunosurveillance programs and prevent PMN progression. Liposomal RNA-based cancer vaccines targeting tumor antigens to APC²⁷³ could overcome systemic as well as local PMN immunosuppression and T cell tolerance and improve therapeutic outcome. The ongoing, first phase I trial in patients with advanced melanoma (NCT02410733) has yielded encouraging responses so far, with shrinkage or complete regression of multiple LN and lung metastases, while also hindering the development of novel metastases in responsive patients with lower mutational burden that previously received ICB therapy. These effects were achieved through expansion and endurance of tumor-specific memory CD8⁺ T cells and enhanced cytotoxic responses.

Co-opting gut microbiota to boost immunotherapies

Strategies that reshape microbiota to improve anti-cancer treatment are being evaluated. High-fat diet (HFD) are linked to increased risk of metastatic cancer^274,275 (Box 3), an effect attributed to the induction of microbial dysbiosis and PMN formation^179,275. Glycyrrhizic acid (GA) prevented HFD-induced PMN formation and metastasis by reversing microbiota dysbiosis¹⁷⁹. Treatment with GA obstructed myeloid cell migration and S100A8/A9 expression in the lung PMN. Antibiotic treatment or recolonization of the gut with beneficial bacteria exerted effects similar to GA, highlighting the gut microbiome role in systemic crosstalk with PMNs in distal organs¹⁷⁹. Similarly, sodium butyrate, another major gut microbiome metabolite, rescued gut dysbiosis and significantly reduced liver metastasis in CRC pre-clinical models by increasing intrahepatic NKT and Th17 cell infiltration, along with decreasing Treg and IL-10 accumulation^176,177.

EVP-based therapies

EVPs are promising reconfigurable therapeutic tools. Chimeric EVPs produced by macrophage-tumor hybrid cells accumulated in the LN where they activated T cells, inducing tumor regression and reducing metastases in lymphoma, breast, and melanoma models²⁷⁶. In the metastatic lung of breast cancer models, these EVPs increased CD8⁺ T cell/Treg ratio and in combination with PD-1 ICB, lung metastasis was significantly suppressed²⁷⁶. This study warrants further exploration of using personalized EVPs for cancer therapy. Meanwhile, EVP-targeted therapeutic approaches focused on disrupting circulating EVPs could interfere with the capacity of primary tumor-derived EVPs to prepare a metastasis-supportive microenvironment at early stages, thereby preventing PMN formation. For example, targeting key EVP proteins in systemic circulation using monoclonal antibodies against S100A4 or integrins could interfere with EVP homing hindering PMN development ^110,277,278.

Discussion

Indisputable evidence underscores PMN’s prominent role in determining cancer progression and outcome. However, translation of this knowledge to the clinic has been slow. To better characterize pre-metastatic alterations in patients, we propose that biopsies from metastasis-prone organs, such as LN, bone marrow, liver, and when possible, lung, and omentum, should be regularly collected and studied in parallel to liquid biopsies. Systematically correlating the signatures of these samples collected at diagnosis/during primary tumor resection with the subsequent metastatic site and patient survival data will accelerate biomarker discovery and early cancer detection, prior to the metastatic colonization. Such findings will enable a paradigm shift in cancer treatment, where personalized prediction of metastatic sites will allow early intervention and tailoring treatments to specific PMN sites/characteristics. Such approaches may reduce or even prevent metastasis by reverting the “soil” of these distant sites to one unfavorable for CTC growth. Moreover, as current therapies may promote PMN formation, sensitive biomarkers or imaging methods could improve patient stratification for current standard of care therapies, sparing patients that show no evidence of systemic disease or metastases aggressive treatments. As metastasis remains the most critical factor determining cancer patient survival, such approaches bear unprecedented potential in changing patient care and outcomes.