The tumor microenvironment (TME) is a dynamic and heterogeneous ecosystem in which malignant cells interact directly or indirectly with stromal, immune, endothelial cells, and proteins of extracellular matrix (ECM). Far from being a passive scaffold, the TME actively shapes tumor initiation, progression, immune evasion, and therapeutic resistance through an intricate network of intercellular communications and molecular signaling.

Over the past two decades, high-dimensional profiling techniques, including single-cell transcriptomics, proteomics, and spatially resolved mapping, have provided unprecedented insights into the cellular and molecular complexity of the TME. These studies have revealed that cell–cell communication within tumors involves soluble mediators such as cytokines, chemokines, and growth factors (ligands) that bind to plasma membrane-bound receptors, in addition to extracellular vesicles, and direct membrane-membrane physical contacts (1–3). Collectively, these exchanges regulate the crosstalk among stromal, immune, and cancer cells themselves.

A defining feature of TME is its dual capacity to modulate the immune response. While specific subsets of immune cells retain cytotoxic functions, others acquire regulatory or exhausted phenotypes that suppress immunity and facilitate tumor growth. Stromal components, including cancer-associated fibroblasts (CAFs), tumor-associated macrophages (TAMs), and endothelial cells, further contribute to this shift by secreting pro-tumor mediators, promoting cell fusions, remodeling the ECM architecture, stimulating angiogenesis, and supporting metastatic dissemination.

This review synthesizes and integrates current insights into the complex interactions among heterogeneous cell populations in the tumor microenvironment, with a particular focus on how these networks support immune defense yet paradoxically promote tumor progression. We examine the roles of key immune and stromal cell populations, the impact of metabolic and oxidative stress, and the role of extracellular vesicle-mediated signaling. Additionally, we emphasize emerging therapeutic strategies that aim to disrupt pro-tumoral communication networks while enhancing antitumor immunity. By integrating recent findings, we seek to provide a comprehensive framework for understanding and targeting the cellular crosstalk that underlies cancer progression and therapeutic resistance.

Within the TME, a complex and constantly evolving network of interactions occurs among malignant cells, non-malignant stromal and immune cells, and the acellular compounds of ECM. In healthy tissues, exposure to physical, chemical, or biological genotoxic agents continuously induce genomic and epigenomic alterations. Such alterations lead to DNA damage and genomic instability, disrupting key cellular functions. These include sustained proliferative signaling, loss of cell death mechanisms, and changes in genes encoding molecules essential for cell death induction, among others. In the premalignant phase, genomic and epigenetic changes generate structural alterations in proteins (neoantigens) or induce overexpression of self-proteins. Natural killer (NK) cells, innate lymphoid cells specialized in immune surveillance, are critical for detecting and eliminating such aberrant cells (4).

The cytotoxic–regulatory balance of NK cells, classically defined by inhibitory interactions with major histocompatibility complex (MHC) class I molecules and activating signals through stress ligands such as MICA/B, acquires new meaning within the tumor context (5). Stromal- and tumor-derived cytokines remodel this equilibrium, progressively converting immune surveillance into tolerance. Even the perforin–granzyme axis, once the hallmark of NK-mediated cytotoxicity, becomes blunted or repurposed under chronic exposure to transforming growth factor-β (TGF-β), interleukin-10 (IL-10), and hypoxic stress (6). Rather than a fixed killing program, the NK–tumor interface acts as a dynamic rheostat integrating metabolic cues and cytokine availability (7).

NK cell activity is amplified by invariant natural killer T (iNKT) cells, which express a semi-invariant T cell receptor (TCR) recognizing glycolipid antigens presented by CD1d molecules on transformed or stressed cells. Upon antigen recognition, iNKT cells release cytokines such as interferon-γ (IFN-γ), which stimulates NK proliferation and cytotoxicity, establishing a positive feedback loop that strengthens innate immune surveillance. Given the assumption that the human body is in constant exposure to mutagenic insults, the NK–iNKT axis plays a central role in early tumor immunosurveillance.

In injured and dead cells, intracellular molecules are released or translocated to the plasma membrane, serving as alarm signals for immune recruitment. These damage-associated molecular patterns (DAMPs) and, in cancer, lifestyle-associated molecular patterns (LAMPs) (8) or xenobiotic-associated molecular patterns (XAMPs) (9), act as “find me” and “eat me” cues recognized by pattern recognition receptors (PRRs) on phagocytes such as macrophages and dendritic cells. PRRs include membrane-bound Toll-like receptors (TLRs) and C-type lectin receptors (CLRs), as well as cytosolic sensors such as RIG-I-like receptors (RLRs), AIM2-like receptors (ALRs), cyclic GMP–AMP synthase (cGAS), and NOD-like receptors (NLRs) (10). These systems enable rapid detection of cellular stress and initiate both innate and adaptive immune responses (9, 11).

Activation of PRRs on tissue-resident macrophages induces the secretion of growth factors, chemokines (CXCL4, CXCL8, CXCL10, CXCL12, etc.) and cytokines, including CCL2/MCP-1, CXCL13, IL-1α, IL-1β, IL-6, IL-8, IL-12, IL-23, and tumor necrosis factor-α (TNF-α). DAMPS and chemokines that reach the endothelium induce the expression of an array of adhesion molecules to promote the recruitment, rolling, adhesion, and diapedesis of neutrophils. Neutrophils and phagocytes at the site of injury generate reactive oxygen species (ROS) via NADPH oxidase and reactive nitrogen species (RNS) through inducible nitric oxide synthase (iNOS), inflicting oxidative and nitrosative damage on lipids, proteins, and DNA of early transformed cells (12).

For complete tumor eradication, adaptive immunity must be engaged. DCs, the most potent antigen-presenting cells (APCs), phagocytose tumor material, process antigens, and migrate to lymph nodes, where they present peptides via MHC class I and II molecules to naïve CD8⁺ and CD4⁺ T cells, respectively (13). Antigen recognition through the TCR–MHC interaction delivers the first activation signal, costimulatory interactions such as CD80/86–CD28 provide the second, and cytokines like IL-12 or IFN-α supply the third, directing T cell differentiation.

During the activation of the immune response, innate cells and the clonal expansion of CD4⁺ and CD8⁺ T lymphocytes require a large amount of available bioenergy products. Glucose consumption and the synthesis of macromolecules such as nucleotides, lipids, and proteins are necessary for their optimal functioning (14). This process, known as the Warburg effect, is similar to that described in malignant cells, which switch to aerobic glycolysis while maintaining fatty acid oxidation and amino acid uptake to meet biosynthetic demands. This metabolic reprogramming has recently been proposed as a fourth signal (15–18). After activation of CD4⁺ T cells, they differentiate into Th1 cells producing IL-2 and IFN-γ, whereas CD8⁺ cytotoxic T lymphocytes (CTLs) synthesize granzyme and perforin to eliminate tumor cells presenting cognate antigens (19–21).

DAMPs, LAMPs chemokines, and cytokines act as a secondary communication layer, translating immune-mediated injury into stromal activation. Endothelial and fibroblast responses, vasodilation, adhesion molecule expression, and matrix remodeling, facilitate immune infiltration but also seed the architecture of future tumor niches. Proteases, reactive oxygen and nitrogen species, while initially cytotoxic, simultaneously erode extracellular matrix integrity and generate bioactive fragments that signal repair (22). As inflammation resolves, neutrophils and macrophages transition toward pro-resolving phenotypes, initiating tissue regeneration yet preserving the molecular memory of stress that can later be co-opted for tumor growth (23).

When the coordinated activity of innate and adaptive immune cells achieves complete elimination of incipient transformed tumor cells, the acute inflammatory phase subsides. At this stage, a network of regulatory mechanisms, mediated by soluble factors and cellular interactions, actively limits inflammation and initiates tissue repair, restoring homeostasis (24).

As mentioned above, different immune cells consume nutrients to perform their effector function, releasing waste compounds. Interestingly, these waste products can modulate the immune response, reduce local inflammation and maintain immune homeostasis for successful tissue remodeling. A number of waste products act as immunomodulators, promoting events ranging from the inhibition of the cytotoxic capacity of NK cells and CD8⁺ effector T cells, to the alteration of immune cell differentiation patterns, to the promotion of regulatory T cells (Tregs), among others (25). At high concentrations of lactate or other local waste components, these Tregs are generated from naive, memory, or tissue-resident T cells. They act through two processes: one mediated by cell–cell contact, through the expression of molecules designed as immune checkpoints, or through the release of soluble factors with inhibitory activity.

Neutrophils are key initiators of the resolution phase. Beyond their pro-inflammatory functions, they produce specialized pro-resolving mediators (SPMs) including resolvins and lipoxins, a superfamily of endogenous bioactive lipids that play central roles in this transition regulating inflammation and restore tissue homeostasis (26). During resolution, neutrophils undergo apoptosis, releasing enzyme-rich microparticles that contribute to SPM biosynthesis (27).

SPMs, derived from omega-3 and omega-6 polyunsaturated fatty acids (PUFAs) through the enzymatic activity of lipoxygenases (LOX) and cyclooxygenases (COX), include lipoxins, resolvins, protectins, and maresins. Unlike classical pro-inflammatory eicosanoids (e.g., prostaglandins and leukotrienes), SPMs actively terminate inflammation and promote tissue repair without causing immunosuppression (28). Lipoxins, generated from arachidonic acid (AA; C20:4, ω-6) via transcellular LOX pathway, were the first pro-resolving mediators identified. LXA₄ and LXB₄ counteract excessive inflammation and signal the onset of resolution (29). Resolvins, named for their role in the “resolution phase,” are subdivided into E-series (EPA-derived) and D-series (DHA-derived), which include RvE1, RvE2, RvD1, and RvD2. Protectins (PD), also known as neuroprotectins in neural tissues, are DHA-derived and attenuate inflammatory damage while promoting cell survival. Maresins (MaR), primarily synthesized by macrophages from DHA, facilitate clearance of death cells and stimulate tissue regeneration; examples include MaR1 and MaR2. Additionally, cysteine-conjugated SPMs (cys-SPMs), such as maresin conjugates in tissue regeneration (MCTRs) and protectin conjugates in tissue regeneration (PCTRs), arise from glutathione-dependent pathways. Together, these lipid mediators form a coordinated network of “immunoresolvents” that exert potent effects at picomolar to nanomolar concentrations (29).

In addition, neutrophil–platelet interactions also promote lipoxin formation: neutrophils supply leukotriene A₄ (LTA₄) via the 5-LOX pathway, which platelets convert to LXA₄ through 12-LOX activity. This process exemplifies the “lipid mediator class switch,” whereby pro-inflammatory mediators give way to pro-resolving mediators (30). Other proteins, such as annexin A1, galectin-1, and chimerin, secreted by fibroblasts and endothelial cells cooperate to halt leukocyte infiltration, counter-regulate pro-inflammatory signals, and enhance efferocytosis of apoptotic neutrophils (31).

SPMs also orchestrate macrophage polarization. Classically activated M1 macrophages, which produce pro-inflammatory eicosanoids and cytokines (IL-1β, TNF-α, IL-6, IL-8, IL-23), gradually shift to a resolving M2 phenotype under the influence of SPMs. M2 macrophages secrete anti-inflammatory mediators such as IL-10, PDGF, and TGF-β, and they actively produce maresins, resolvin D, and protectins (32, 33). This shift is tightly regulated by metabolic reprogramming in the microenvironment, characterized by increased glutamine utilization, reduced glucose availability, acidic pH, and hypoxia, which favors oxidative phosphorylation over glycolysis and enhances efferocytosis (34).

Anti-inflammatory cytokines from M2 macrophages (IL-4, IL-10, IL-6, TGF-β) stimulate fibroblasts to produce proangiogenic and ECM components, including VEGF, FGF, PDGFD, fibronectin, hyaluronic acid, proteoglycans, and collagen, critical for re-epithelialization. Fibroblasts may also transdifferentiate into contractile myofibroblasts, which contribute to wound closure through production of matrix metalloproteinases (MMPs) and structural ECM proteins (35, 36). This remodeling restores tensile strength and tissue integrity.

Collectively, changes in the cellular processes of immune cells, induced by Tregs, other immunomodulatory cells like M2, or high concentration of metabolic by-products, are critical mechanisms for regulating the activity of immune cells (37). In summary, coordinated crosstalk among immune cells, stromal populations, and resident tissue cells not only ensures the elimination of transformed cells but also establishes a tightly controlled resolution phase. The in situ biosynthesis of SPMs limits inflammation locally, prevents chronic inflammatory states, and promotes tissue regeneration through synchronized and highly regulated repair mechanisms that preserve the structural and functional integrity of the affected tissue (See Figure 1). These complex and redundant mechanisms involve the participation of various cells that generate a metabolic change, promoting regulatory functions in immune cells through the release of soluble factors with antiproliferative activity or the expression of a series of molecules to suppress immune activity. In addition, cellular restructuring occurs through the induction of changes in cytoskeletal adhesion molecules. Fibroblasts and transdifferentiated myofibroblasts involved in the deposition of extracellular matrix components for tissue repair must develop cytoprotective mechanisms to prevent cell death, all in an environment of factors that promote endothelial activity, initiating the process of angiogenesis (38). This physiological process can trigger a pathophysiological environment when these mechanisms are dysregulated by the persistence of some residual transformed cells, creating persistent oxidative stress that induces greater genomic instability and generates microenvironmental conditions that favor the development of cancer.

The host immune response may fail to eradicate some clones of nascent tumor cells. In such scenarios, oxidative stress generated by macrophages and neutrophils plays a dual role. While they can destroy transformed cells, they may also induce further genomic alterations in the resistant tumor subsets.

Recently, genomic instability can also arise through cell–cell fusion. Hybrid cells represent a pathophysiological process promoted by certain molecules acting as fusogens that is classified as homotypic when cancer cells with slight genetic differences between them fuse, increasing intratumoral heterogeneity; whereas heterotypic fusion between tumor and normal cells generates hybrids with mixed phenotypes (39–41). During early stages of tumor development, when inflammation and cytokines predominates, some signaling pathways appear to be involved in inducing fusion between tumor cells and immune cells like tissue macrophages has been documented (42).

The overlapping expression of tumor and non-tumoral tissue markers has been used to identify hybrid tumor cells. Such cells have also been detected in cancer patients following bone marrow transplantation, via donor–recipient allele analysis (42, 43). In any condition, residual tumor cells continue to accumulate genomic and epigenetic alterations over time, some lethal, others promoting tumor growth or resistance to immune-mediated cytolysis.

Progressive and stochastic genomic instability produces diverse tumor cell clones with specific metabolic adaptations, collectively forming a heterogeneous tumor mass. Sustained tumor expansion, driven by intrinsic mutations and potentially by the formation of homotypic or heterotypic hybrids, imposes high nutrient demands for macromolecule synthesis. In this aspect, tumor cells reprogram their metabolism state toward anabolic pathways that maintains the demand for energy and metabolites necessary for their continuous proliferation rate, increasing aerobic glycolysis and glutaminolysis for production of cellular building blocks needed to generate novel biomass (44). This metabolic switch results in high lactate production, extracellular acidification linked to hypoxia, and alterations in amino acid, creatine, and lipid metabolism (45–47). Hypoxia, in addition to inflammation, is considered to be another trigger for cell-cell fusion events, and the resulting hybrid cells have been associated with autophagy and increased survival in hostile environments (48–50).

Importantly, these metabolic shifts not only reflect adaptation to environmental stress but also drive clonal selection under immune pressure, favoring the expansion of immune-evasive phenotypes (51). The various metabolites produced by this environment not only influence the activity of multiple immune cells, but also affect stromal cells located within the tumor itself or far from its growth (See Figure 2).

Ongoing tumor growth perpetuates genomic instability and cellular heterogeneity. Tumors develop in close association with stroma composed of inflammatory and immune cells, endothelial cells, pericytes, cancer-associated fibroblasts (CAFs), and the ECM. Tumor niches can arise from the gain of intrinsic mutations in tumor cells, from fusion between genotypically distinct tumor cells, between tumor cells and different immune cells, or from interactions with stromal cells, which encourages excessive tumor cell growth and cellular heterogeneity (52).

In any event, within these niches, continuous intercellular communication occurs through soluble mediators binding to specific receptors, triggering autocrine, paracrine, or juxtacrine signaling. These interactions promote tumor growth and progression, amplified by increased synthesis of growth factors, cytokines, and chemokines, as well as overexpression or enhanced affinity of membrane receptors.

Below is a brief summary of some of the key signaling pathways that are frequently overexpressed in cancer and their biologic effects.

Tumor development and progression are also driven by the intricate communication networks that regulate metabolic fluxes, immune responses, and tissue remodeling. Multiple models highlight the pivotal role of intercellular communication in shaping the process of immunoediting and facilitating the transition to immune escape.

The immunoediting concept posits that immune responses continuously sculpt the tumor cell phenotype. Under immune pressure, sensitive tumor cells are eliminated, while those capable of resisting destruction or evading detection survive and expand (64). This selection process is strongly influenced by the spatial and temporal dynamics of intercellular signaling with dysregulated growth factors and cytokine production and tumor cell metabolic reprogramming. Recently, the Galluzzi group has integrated into the “three Cs” model the mechanisms that tumor cells develop to evade immune recognition, proposing events such as camouflage, coercion, and cytoprotection as distinctive hallmarks of cancer immune evasion (65).

Overproduction of growth factors and cytokines, or gain-of-function mutations in membrane receptors or intracellular trafficking molecules, can sustain the uncontrolled proliferation of tumor cells. These events may also compromise immune function by limiting the availability of soluble mediators at concentrations sufficient to support robust anti-tumor responses. Furthermore, the high proliferative rate of tumor cells depletes essential nutrients and increases metabolic by-products, predominantly through aerobic glycolysis, to sustain their growth.

Metabolic by-products such as ROS, lactate, and oncometabolites function as signaling mediators that modulate immune cell activation, polarization, and survival, thereby contributing to immunosuppression. Within the TME, gradients of lactate, glutamine, acidity, and hypoxia create spatially distinct niches with unique cellular and noncellular compositions (66, 67). These conditions are largely orchestrated by hypoxia-inducible factor-1 (HIF-1), which regulates key enzymes in glycolysis and lipid metabolism (67–69).

Metabolic reprogramming also profoundly influences phagocyte behavior. During acute inflammation phase, macrophages adopt a pro-inflammatory M1 phenotype, secreting IL-1, TNF-α, IL-6, IL-12, IL-23, and other cytokines that sustain innate immune activity and promote Th1 adaptive responses. Th1-derived IFN-γ reinforces M1 activity. Upon resolution of the insult, macrophages typically transition to an M2 (pro-resolving) phenotype, engaging in efferocytosis of apoptotic cells and debris, and secreting IL-10, TGF-β, and growth factors to coordinate tissue repair. SPMs, such as maresins (MaRs) produced by macrophages, accelerate the M1-to-M2 shift, enhancing debris clearance and fostering an anti-inflammatory milieu (70).

In the TME, however, M1-mediated antitumor activity is progressively suppressed, facilitating the predominance of M2-like TAMs. Sustained tumor growth alters the local environment through persistent release of soluble mediators, CSF-1, G-CSF, IL-6, IL-10, CCL2, CXCL12, PDGF, VEGF, among others, produced by both tumor cells and reparative M2 macrophages (71). These signals promote TAM polarization and differentiation into a hybrid phenotype with features of both M1 and M2 macrophages.

Environmental stressors, hypoxia, acidity, nutrient deprivation, and tumor-derived metabolites, in combination with soluble factor signaling, further drive TAM differentiation. TAMs then secrete growth factors, cytokines, and chemokines that fuel tumor proliferation via autocrine and paracrine loops, or through exosome-mediated communication. This promotes endothelial activation, recruitment of additional monocytes, and maintenance of TAM heterogeneity. In advanced cancer stages, these subsets preferentially infiltrate specific tumor niches (72, 73).

Recent high-resolution profiling techniques, including single-cell RNA sequencing (scRNA-seq), CyTOF, and spatial transcriptomics, have revealed that TAMs exist along a continuous spectrum of transcriptional states, rather than fitting the classical M1/M2 dichotomy. scRNA-seq studies consistently identify functionally distinct TAM clusters, such as MHC-II^high antigen-presenting TAMs, MHC-II^low immunosuppressive TAMs, and SPP1⁺ TAMs, the latter preferentially enriched in hypoxic niches and associated with angiogenesis, extracellular matrix remodeling, and tumor progression (74).

Within a single tumor, TAMs display remarkable phenotypic diversity and occupy specialized microanatomical regions, reflecting adaptation to gradients of hypoxia, nutrient availability, cytokines, and lipid metabolites. Their high plasticity allows rapid transcriptional and metabolic reprogramming in response to local cues. Intriguingly, emerging evidence suggests that specific TAM subsets may even arise through heterotypic fusion events between macrophages and tumor cells, potentially generating hybrid populations with enhanced protumoral capabilities, an attractive but still evolving hypothesis that warrants further investigation (75).

Competition for nutrients and the profound metabolic rewiring of the TME serve as critical determinants of immune suppression. Tumor cells, cancer-associated fibroblasts (CAFs), endothelial cells, tumor-associated macrophages (TAMs), myeloid-derived suppressor cells, and infiltrating lymphocytes coexist in a landscape of restricted glucose, amino acids, oxygen, and lipids. These constraints do not simply reflect metabolic stress; rather, they function as potent regulatory cues that bias immune cells toward exhaustion, tolerance, or protumoral phenotypes (76).

A paradigmatic example is the Warburg effect, where tumor cells preferentially utilize aerobic glycolysis even in oxygen-rich conditions. This shift results in massive glucose consumption and lactate secretion, lowering extracellular pH and generating an environment that promotes M2-like TAM polarization, reduces NK and CD8⁺ T cell cytotoxicity, and impairs dendritic cell (DC) activation (77).

Glutamine dependency is also significant, tumor cells upregulate transporters such as SLC1A5 and GLS, allowing proliferation even in nutrient-poor environments, while limiting glutamine availability for effector T cells (78). Arginine is an essential amino acid for T cell proliferation, survival, and effector function. Tumor cells and TAMs both depend on arginine, but TAMs express high levels of arginase-1, which depletes extracellular arginine and profoundly impairs T cell activation and expansion following antigen presentation by APCs (79–81). Arginine depletion restricts mitochondrial respiration and blocks AKT/mTOR activation, reinforcing T cell exhaustion in the TME (82).

Tryptophan metabolism constitutes another crucial immunoregulatory pathway. Mature DCs, macrophages, endothelial cells, and specific epithelial cells upregulate the indoleamine-2,3-dioxygenase (IDO) in response to inflammatory stimuli such as LPS, Toll-like receptor ligands, CpG DNA, IL-1β, TNF-α, IL-6, and IFN-α/β. Physiologically, IDO limits tissue damage. It preserves homeostasis during tissue repair by catalyzing the degradation of tryptophan (Trp) into immunomodulatory kynurenines (Kyn) and downstream metabolites, including nicotinamide adenine dinucleotide (NAD⁺), an essential cofactor for glycolysis and oxidative phosphorylation.

However, tumors constitutively express IDO and tryptophan-2,3-dioxygenase (TDO/TDO2) accelerating Trp catabolism and amplifying immune suppression (83, 84). Trp depletion activates the GCN2/eIF2α pathway, which in turn inhibits AKT/mTOR signaling, inhibiting T cell proliferation and promoting apoptosis. Meanwhile, Kyn accumulation promotes nuclear translocation of the aryl hydrocarbon receptor (AhR), which complexes with HIF-1β to induce expression of IL-10, IL-17, and IL-22, reinforcing immunosuppression (83, 84). AhR/HIF-1β complexes also upregulate the ectonucleotidases CD39 and CD73, which hydrolyze extracellular ATP (a pro-inflammatory DAMP) into adenosine that promotes immunosuppression. Adenosine not only attenuates cytotoxicity but also promotes angiogenesis and impairs leukocyte extravasation (85). This adenosinergic loop is further reinforced under hypoxia via stabilization of HIF-1α, which enhances CD39/CD73 expression (86).

As was previously mentioned, TGF-β1 represents a master regulator of tissue repair and immune tolerance and is one of the most potent immunosuppressive cytokines in the TME. It is secreted by tumor cells, fibroblasts, and TAMs, and stored as latent form (LTGF-β) in the ECM. Activation involves proteases such as plasmin and thrombospondin-1, matrix metalloproteinases (MMPs), glycoprotein A repetitions predominant protein (GARP), or integrins, and other cofactors. Tregs cells also present membrane-bound LTGF-β, which upon receptor binding triggers SMAD-dependent transcriptional programs (87).

In naïve CD4⁺ T cells, TGF-β induces the transcription factor FOXP3, promoting differentiation into Tregs (88, 89). Recent evidence shows that Treg cells exist as a broad spectrum of heterogeneous subsets, identified by distinct phenotypic markers. In addition, the markers described are nonexclusive of Tregs, as some of them can be found in some populations of effector Th cells. The wide phenotypic heterogeneity is mainly based on their functional capacity to inhibit the immune response, more than specific CD markers, cytokines production, chemokines, transcription factors expression, tissue localization, and homeostatic or pathologic process (88, 89).

Asimnasab-Sorkhabi et al. suggest that Treg cells that share expression of two master transcription factors with other T helper subsets, like Th1 (Tbet+), Th2 (GATA-3+), Th17 (ROR-γt+), and Tfh (Bcl6+) should be named “hybrid Treg cells”. This novel proposal is based on the protumor and antitumor activities demonstrated by these subsets of Tregs (90). Whether coexpression of markers and functionality of the varied Tregs rely on distinct stages of their development promoted by the wide plasticity of T cells previously demonstrated, or this proposal may derive from membrane-membrane cell fusions among Treg cells and the varied phenotypes of T effector cells, event promoted by the tissue environment, requires a deeper investigation, as well as defined the role that the diverse subsets of Tregs plays in TME. Table 1 summarizes the major nutrient-dependent pathways in the TME, the cellular sources involved, and their functional impact on innate and adaptive immune responses.

Advances in non-targeted metabolomics have recently unveiled critical regulators of endothelial and tumor metabolism, such as MFSD8, which orchestrate the balance of lipid and amino acid availability in the TME, further influencing immune cell exhaustion (91).

Despite their heterogeneity, all Tregs subsets potently suppress antitumor immunity through secretion of TGF-β, IL-10, and IL-35, supporting tissue homeostasis, and in the TME tumor reinforcing the immunosuppressive tone (95). This regulatory axis promotes macrophage polarization from M1 to M2 phenotypes, induces ectoenzyme expression in macrophages and APCs that increase local adenosine production, and suppresses T cell proliferation and Th1/Th17 differentiation, reducing secretion of effector cytokines such as IFN-γ, IL-17, and IL-22. These effects extend to cytotoxic lymphocytes, in NK cells and CD8⁺ T cells, TGF-β signaling downregulates granzyme and perforin synthesis, impairing cytolytic activity (94). Collectively, nutrient competition (glucose, glutamine, arginine, methionine, tryptophan), secretion of suppressive metabolites (lactate, Kyn, adenosine), cytokine signaling (TGF-β, IL-10), and metabolic symbiosis between tumor cells and TAMs create a resilient immunosuppressive niche (96, 97). These intertwined networks blunt both innate and adaptive immunity, establishing an immune-privileged environment that supports tumor progression (See Figure 2).

The distribution and density of immune cells within tumor nests and stromal compartments is a defining feature of tumor immunophenotypes. Most cancers are heavily infiltrated by M2-polarized macrophages, accompanied by variable proportions of polyclonal and phenotypically distinct CD3⁺ T cell subsets, including memory CD4⁺/CD8⁺ T cells, Tregs, and follicular helper T cells (98, 99). In contrast, CD56⁺ or CD57⁺ NK cells appear at variable-to-low frequencies, while neutrophils and mast cells are generally scarce (100–102). The composition of tumor-infiltrating lymphocytes (TILs) varies considerably even among tumors of the same histological type (103, 104).

Based on immune cell infiltration, tumors can be broadly categorized into three phenotypes. Immunologically inflamed, or “hot” tumors, exhibit abundant immune infiltrates often associated with highly immunogenic antigens, robust and chronic immune reactivity, and better prognosis, as well as improved responses to therapy. However, immune infiltration alone does not ensure therapeutic success, highlighting the importance of characterizing the specific immune subpopulations present. Immunologically excluded tumors lack T cells in direct contact with tumor cells, with immune cells confined to stromal or peritumoral fibroblast zones, patterns that may arise from MHC-I downregulation, impaired chemotactic factor production, or physical stromal barriers. Immunologically desert, or “cold” tumors, display no detectable T cell infiltration, frequently due to MHC-I loss, defects in the antigen presentation mechanisms, or high levels of inhibitory cytokines from tumor cells, fibroblasts, and myofibroblasts.

The cellular and molecular bases of these immunophenotypes, particularly those that mimic wound-healing environments, remain incompletely understood (105). Advances in spatial transcriptomics and high-resolution tumor mapping have improved immune infiltrate characterization, in hepatocellular carcinoma, these technologies have mapped tumor evolution and revealed highly variable genes associated with TME heterogeneity (106–110). The marked intratumoral heterogeneity requires rigorous analysis using multiple sections of the same tumor to obtain a more accurate representation.

Currently, it is recognized that a subset of tumors develops ectopic lymphoid aggregates known as tertiary lymphoid structures (TLS), which resemble secondary lymphoid organs. Mature TLS contains dendritic cells, macrophages, subsets of T and B lymphocytes, fibroblasts, and endothelial cells, forming organized immune hubs capable of initiating potent adaptive antitumor responses. Their functional relevance is particularly evident in tumors treated with immune checkpoint inhibitors (ICIs). However, tryptophan (Trp) metabolites and stromal-derived factors can impair TLS maturation, limiting their immunological potency. Despite these challenges, TLS presence correlates with improved clinical outcomes, especially in patients receiving ICIs (111–114).

Fibroblasts are ubiquitous mesenchymal cells essential for cell differentiation, tissue morphogenesis, and tissue repair, bone marrow progenitors, mesenchymal stem cells, epithelial or endothelial cells, or malignant epithelial cells. Their plasticity allows differentiation into heterogeneous populations, including stromal fibroblasts, activated fibroblasts, myofibroblasts, and cancer-associated fibroblasts (CAFs). CAFs secrete a diverse repertoire of signaling molecules such as TGF-β, bFGF, IL-6, IL-8, PDGF, VEGF, and HGF, as well as ECM-remodeling enzymes including matrix metalloproteinases (MMPs) and tissue inhibitors of metalloproteinases (TIMPs) (115, 116). Through autocrine and paracrine signaling, as well as extracellular vesicle-mediated exchange, CAFs engage in extensive crosstalk with tumor cells, immune populations, and other stromal elements (116, 117). Functionally, they promote tumor progression by stimulating proliferation via FGF, IL-6, and CXCL12. In addition, the secretory proteases MMPs and urokinase-type plasminogen activator (uPA) that cleave key ECM proteins, including collagen, elastin, and fibronectin, thereby facilitating immune cell infiltration, driving EMT through PDGF, HGF, and IL-1, and enhancing migration and metastasis, effects reinforced by tumor-associated macrophage-derived factors in a pro-tumorigenic feedback loop.

Macrophages, when responding to tissue insults, initiate a respiratory burst that generates ROS capable of damaging lipids, proteins, and nucleic acids, while also activating diverse signaling pathways (118). Cells counter ROS through enzymatic and non-enzymatic antioxidant defenses (119). The Nrf2/HO-1 pathway is a central regulator of these cytoprotective mechanisms, mitigating oxidative stress, suppressing inflammation, and preserving tissue integrity. Under oxidative stress, kinases such as ERK, PI3K, AMPK, and PKC mediate dissociation of Nrf2 from its cytoplasmic repressor Keap1, enabling nuclear translocation and binding to antioxidant response elements (AREs; 5′-RTGAYnnnGCR-3′), which drives transcription of antioxidant genes including superoxide dismutase (SOD), glutathione-S-transferase (GST), catalase (CAT), and γ-glutamylcysteine synthase (γGCS) (120, 121).

The free heme group itself can be pro-oxidant, but heme oxygenase-1 (HO-1), a stress-inducible enzyme, catabolizes it into carbon monoxide, ferrous iron, and biliverdin, subsequently converted to bilirubin, all with antioxidant and anti-inflammatory properties (122, 123). Nrf2 recruitment to the ARE within the HO-1 gene amplifies these cytoprotective programs (124). During wound-healing resolution, hypoxia, acidosis, metabolic shifts, and the transition to anti-inflammatory mediators can activate Nrf2/HO-1 in fibroblasts, myofibroblasts, and TAMs, granting resistance to cell death and enabling coordination of inflammation resolution, angiogenesis, and tissue regeneration (125–128).

In the TME, oxidative stress intensifies, imposing selective pressure on malignant and non-malignant cells to adopt antioxidant and tissue-protective strategies. Hybrid cells from fusion events between tumor cells and fibroblasts, myofibroblasts, TAMs, or bone marrow cells (129) may combine phenotypes with or without prior Nrf2/HO-1 activation, potentially allowing transient quiescence or stem-like states that enhance survival under stress. Although this remains speculative, it complements the view that individual tumor cells can autonomously upregulate Nrf2/HO-1 to evade death. Constitutive HO-1 overexpression, reported in multiple cancers, may or may not indicate hybrid-cell origins.

Beyond survival, tumor–stromal cell fusion may aid immune evasion by incorporating self-antigens from autologous cells into tumor membranes, masking neoantigen recognition and reducing immune detection. Tumor cells may further resist NK and cytotoxic T lymphocyte-mediated killing by suppressing granzyme and perforin activity or altering pro-/anti-apoptotic gene balance, processes potentially modulated by Nrf2/HO-1. Resistance is reinforced by MHC loss through allele deletion or genetic alterations and reduced co-stimulatory molecule expression, impairing T cell activation (130).

Quantitative analyses further illustrate the magnitude and functional consequences of these alterations within the TME. TAMs frequently dominate the myeloid infiltrate, representing 30–50% of all tumor-infiltrating immune cells in pancreatic, breast, lung, and glioma tissues (71). In highly fibrotic tumors, TAM density can surpass that of dendritic cells by more than an order of magnitude, reinforcing an immunosuppressive milieu through IL-10, TGF-β, and ARG1 secretion (131). Similarly, Tregs accumulate in stromal-rich or immune-excluded regions, often outnumbering effector CD8⁺ T cells with Treg: Teff ratios ranging from 2:1 to 5:1 (132). This imbalance correlates with impaired cytotoxic function, reduced granzyme B expression, and diminished responsiveness to checkpoint blockade (133).

Spatial metabolite profiling underscores the metabolic pressures that shape immune fate. Lactate concentrations reach 10–20 mM in hypoxic tumor cores—nearly an order of magnitude higher than in adjacent nonmalignant tissues (<2 mM) (134). This sharp metabolic gradient drives intracellular acidification of T cells, inhibits mTOR signaling, and suppresses cytokine production, while simultaneously supporting M2-like polarization of macrophages and promoting Treg stability (135).

The structural remodeling of the ECM further constrains immune access. Tumors commonly exhibit 200–300% increases in collagen fiber density, accompanied by elevated crosslinking mediated by lysyl oxidase (LOX). These changes raise tissue stiffness into the 1–10 kPa range, comparable to fibrotic organs, creating physical corridors that redirect or fully prevent lymphocyte infiltration (136). Dense collagen bundles can reduce T cell migration speed by 50–70%, while disorganized fibrillar networks alter chemokine diffusion, generating “immune deserts” even in tumors with substantial neoantigen burden (137).

Spatial and temporal heterogeneity within the TME critically shapes treatment responses. ECM remodeling by CAFs and TAMs generates irregular stromal densities, heterogeneous collagen fiber alignment, and hypoxic pockets that restrict T cell infiltration and reduce drug penetration (73). These spatial gradients also influence the distribution of immune phenotypes, with immunosuppressive M2 macrophages enriched in hypoxic cores, while cytotoxic T cells tend to localize toward better-perfused peripheral regions (138). Temporally, dynamic shifts in stromal stiffness, angiogenic activity, and ECM turnover alter the accessibility of tumor nests throughout disease progression and under therapeutic pressure. As a result, both spatial compartmentalization and time-dependent ECM remodeling directly impact the efficacy of immunotherapies, metabolic inhibitors, and cytotoxic regimens, contributing to heterogeneous and often unpredictable treatment outcomes (139).

Together, these quantitative features—immune composition imbalances, metabolite gradients, and ECM stiffening—demonstrate how the TME evolves into a compartmentalized ecosystem that systematically restricts antitumor immunity. They provide a measurable framework for understanding why certain tumors remain refractory to immunotherapy despite robust antigenicity.

Nowadays, it is accepted that all cell types express an evolutionarily conserved communication system though the cell-derived membrane-surrounded vesicles designated extracellular vesicles (EVs) and contribute to a distinct type of cell-cell fusion. EVs carry molecules of distinct nature and transport information to distant sites via body fluids. The EVs are heterogeneous and comprises microvesicles (MVs) and exosomes that are classified with respect to density, subcellular origin, function and bioactive cargo, like a wide-variety of components of cytosolic and cell surface, diverse nuclear proteins, transcripts, and even fragments of DNA (140–143). EVs can be categorized according their size in exosomes which are mainly derived from endosomal compartment (30–100 nm diameter), microvesicles also called ectosomes, they derive from plasma membrane budding (100–1000 nm diameter), and recently the large oncosomes that seem to be specific for transporting oncogenic material, as they have been detected in malignant process (1-10 μm in diameter) (144). The EV cargo can be transferred to normal or malignant cells, inducing genotypic, phenotypic and cellular functional changes. Various groups have produced excellent reviews on this topic (143, 145, 146) that summarize the information regarding the role of different EVs in various immune cells, both activating antitumor actions and promoting their induction or polarization towards immune cells involved in protumor actions that encourage tumor evasion mechanisms and tumor resistance to treatments.

Within the TME, the production and release of MVs by the various types of cells present, in addition to hypoxia and acidic conditions, metabolic stress, oxidative imbalance, and signalization induced by membrane-membrane and soluble interactions, etc., is an additional factor involved in the remodeling of the phenotype of malignant and non-malignant cells (147). These signals reinforce the loss of cell adhesion, reorganization of the cytoskeleton, and increased migration and invasion in cancer cells (148). This miscellaneous environment and stochastic events that induce in the varied cellular components of the TME, stimulate tumor cells, CAFs, and M2 macrophages to produce and secrete matrix metalloproteinases (MMPs) and other proteases. These enzymes degrade basement membrane components and weaken endothelial junctions, facilitating local invasion and vascular escape by cancer cells (149).

Intercellular membrane fusion and exchange of tumor-derived microvesicles with fibroblasts have been proposed to facilitate EMT and promote drug resistance by transferring oncogenic proteins, RNAs, and signaling lipids (150–153).

Platelets, another key stromal component, have long been recognized for their role in shielding tumor cells from mechanical shear forces in circulation and protecting them from immune-mediated lysis (154). Changes in platelet count or function correlate with disease stage and prognosis (155–157). Increasing evidence indicates that platelets can infiltrate tumors by binding collagen through glycoprotein VI (GPVI) and forming integrin-mediated contacts with tumor cells. Upon degranulation, they release potent proangiogenic factors, including vascular endothelial growth factor (VEGF), fibroblast growth factor (FGF), epidermal growth factor (EGF), and platelet-derived growth factor (PDGF), which collectively drive neovascularization and sustain tumor growth. In turn, tumor cells release exosomes enriched with tumor-specific proteins, growth factors, and mRNA, which are taken up by platelets. This bidirectional exchange not only amplifies local tumor growth but also promotes systemic dissemination, immune evasion, and metastatic colonization (158, 159). While the pro-tumoral roles of platelets are well established, some studies suggest they can exert antitumor effects, a duality likely dictated by tumor type and the diverse bioactive cargo they carry (160).

Recent evidence highlights that specific cargo within tumor-derived EVs actively reshapes the immune landscape. For instance, exosomal PD-L1 has emerged as a critical mediator of immune evasion, capable of inhibiting T cell activation and promoting senescence through lipid metabolism reprogramming, effects that are mechanistically distinct from membrane-bound PD-L1 (161). Furthermore, a mutual regulation exists where tumor-derived exosomes induce PD-L1 expression in TAMs, while TAM-derived exosomes can reciprocally enhance tumor cell immune resistance (162).

Additionally, non-coding RNAs such as miR-21 shuttled by exosomes play a dual role in remodeling the TME. Exosomal miR-21 has been shown to induce M2 macrophage polarization via STAT3 signaling and suppress CD8+ T cell function (163). Notably, recent studies indicate that exosomal miR-21-5p also targets PDHA1 to promote glycolysis in glioblastoma, directly linking vesicle-mediated communication with metabolic reprogramming and the Warburg effect (164). Other molecules like TGF-β carried by EVs further amplify this immunosuppressive network by driving fibroblast activation.

Deciphering the functional behavior of immune cells within the complex, dynamic TME is essential for optimizing strategies aimed at sustaining durable antitumor immunity. Advances in the characterization of TME composition have provided a clearer understanding of the mechanisms that allow tumors to evade antitumor immune activities (178). Notably, high macrophage infiltration often coincides with the presence of tumor-infiltrating lymphocytes (TILs), yet these T cells frequently display elevated expression of multiple inhibitory receptors that dampen their effector capacity.

This immunosuppressive landscape is shaped by the overexpression of immune checkpoint molecules that inhibit T cell activation and drive them toward functional exhaustion. Immune checkpoints encompass a diverse set of molecules with inhibitory or regulatory activity on T cells (179). Functionally, only glycosylated receptors are capable of delivering inhibitory signals, which they transmit through immunoreceptor tyrosine-based inhibitory motifs (ITIM) and immunoreceptor tyrosine-based switch motifs (ITSM) (180). While the intricate molecular mechanisms, receptor–ligand interactions and downstream signaling networks involved in their inhibitory effects are well documented elsewhere (179), they fall outside the scope of this section.

The acquired immunity, particularly that of tumor-associated T cells due to their high capacity for clonal expansion in response to an antigen, is the predominant and novel immunotherapy against cancer. Nevertheless, other forms of cancer treatment are being applied with great success.

In this review, we depart from the evolutionary oncology approach and its close relationship with the different immune and stromal cells that characterize the tumor microenvironment. Based on excellent reviews detailing new technologies and procedures, we highlight the different mechanisms of cellular communication with the aim of presenting a multifaceted and comprehensive point of view. This review has outlined how multidirectional crosstalk between tumor, immune, and stromal cells, via soluble mediators, membrane ligands, extracellular vesicles, and direct physical interactions, reprograms tissue physiology. This network modulates immunoediting, drives metabolic reconfiguration, involves tissue repair cells and molecules, promotes invasion and angiogenesis, enables immune evasion, and ultimately leads to tumor resistance (See Figure 3).

We highlight the role that tissue repair plays in promoting tissue conditions that may act as potential triggers for tumor development. In addition, central regulatory nodes included canonical signaling cascades (JAK/STAT, PI3K/AKT/mTOR, TGF-β, cadherin-catenin/WNT), metabolic and redox programs (HIF-1, lactate, ROS), immunoregulatory circuits (IDO/TDO-kynurenine-AhR), adenosine via CD39/CD73), and key microenvironmental players such as TAMs, CAFs, platelets, and TLS, EMT, and cell-cell fusion add greater tumor plasticity by increasing heterogeneity and resistance to treatment.

Progression from immune defense to tumor promotion is not a single binary switch but a network phenomenon. Nutrient depletion (e.g., arginine), acid–hypoxia gradients, and vesicle trafficking act in concert with checkpoint signaling to blunt cytotoxic responses and polarize myeloid and fibroblast phenotypes toward tumor support. This interconnectedness might explain why monotherapies often fail in solid tumors: disabling one pathway leaves much of the circuit intact (81).

Advancing the field will require mapping tumor–stroma–immune communication with spatial, temporal, and causal precision. Spatial multi-omics and intravital imaging must be integrated with functional perturbation screens to decipher intercellular crosstalk—revealing precisely which cells interact, when, and with what consequence. Longitudinal sampling, utilizing liquid biopsies to analyze vesicle cargo and soluble mediators, offers a means to track TME dynamics throughout the course of therapy. Furthermore, functional preclinical models, such as patient-derived organoids and ex vivo immune–stroma co-cultures, are essential for testing circuit-level interventions under physiologically relevant conditions of hypoxia, acidity, and matrix complexity. The biology of cell fusion and hybrid states warrants dedicated investigation, as does the standardization of TLS quantification and maturation scoring. Consequently, communication biomarkers should be developed to guide combination therapies, ideally incorporating pharmacodynamic readouts that capture TME rewiring in real time.

Finally, the immense complexity of tumor heterogeneity and antigen escape necessitates precision that transcends conventional methodologies. The integration of artificial intelligence (AI) and machine learning has become indispensable for the next generation of immunotherapies. Emerging tools, such as AI-assisted antigen screening and computational modeling, are now being leveraged to predict optimal targets and design combinatorial regimens capable of effectively rewiring TME communication networks, thereby tailoring treatments to the unique molecular landscape of each patient.

In summary, decoding and deliberately reprogramming the languages of cellular crosstalk may transform oncology from targeting isolated nodes to rewiring entire tumor ecosystems. Done effectively, the TME can be converted from cancer’s accomplice into its Achilles’ heel, turning hostile microenvironments into therapeutic opportunities. This review is not intended as an exhaustive mechanistic analysis but rather as a comprehensive reference framework that synthesizes key concepts and literature for researchers investigating tumor–stroma–immune interactions.