2.1.1 MSCs as Precursors of CAFs

MSCs serve as a primary source of CAFs within the TME. These multipotent stromal cells, which originate from bone marrow, adipose tissue, and other mesenchymal tissues, can be actively recruited to tumor sites by cytokines and chemokines, including TGF-β, PDGF, and stromal cell-derived factor-1 [36]. Upon exposure to these tumor-derived signals, MSCs differentiate into CAFs, acquiring specific characteristics, such as elevated expression levels of α-SMA, FAP, and vimentin [37].

The differentiation of MSCs into CAFs is predominantly driven by the TGF-β/SMAD, NF-κB, and JAK/STAT signaling pathways. Among these, TGF-β plays a particularly critical role by initiating a transcriptional program that enhances the expression of CAF-associated markers and reinforces their profibrotic [38]. The activation of the NF-κB and STAT3 pathways further amplifies the protumorigenic functions of MSC-derived CAFs, leading to increased cytokine production and therapeutic resistance [39, 40]. Additionally, MSCs can undergo functional differentiation regulated by factors such as myocardin-related transcription factor A, which stabilizes their tumor-promoting CAF phenotype [37]. Notably, MSC-derived CAFs exhibit a high degree of plasticity, contributing to their heterogeneity within tumors.

MSC-derived CAFs have been identified in various cancer types, including breast, lung, and pancreatic cancers, highlighting their significant role in tumor progression [37, 41, 42]. However, the precise molecular mechanisms governing the recruitment and differentiation of MSCs into CAFs remain incompletely understood, particularly regarding the heterogeneity of CAF subsets and their context-dependent interactions with immune cells. Elucidating these pathways is critical for a more comprehensive understanding of the contributions of MSCs to CAF heterogeneity and their functional diversity across different tumor contexts [43].

2.1.2 EndMT in CAF Formation

In addition to MSCs, endothelial cells represent another significant source of CAFs through a process known as EndMT [44]. EndMT involves a phenotypic conversion in which endothelial cells lose their characteristic markers, such as CD31 and VE-cadherin, while acquiring mesenchymal traits, including the expression of α-SMA and FAP [45, 46]. This transition facilitates the detachment of endothelial cells from the vascular endothelium, allowing them to integrate into the tumor stroma as functionally active CAFs.

EndMT in the formation of CAFs is primarily driven by key tumor-derived factors, particularly TGF-β and IL-6 [47]. Among these factors, TGF-β plays a central role by activating the SMAD-dependent signaling cascade and interacting with other pathways, such as Notch and Wnt/β-catenin, thereby reinforcing the mesenchymal phenotype [48, 49]. Additionally, hypoxia, a hallmark of the TME, promotes EndMT by upregulating hypoxia-inducible factor-1α (HIF-1α), which enhances the expression of transcription factors that induce EndMT, including Snail, Slug, and Twist [50, 51].

Notably, EndMT-generated CAFs exhibit significant plasticity, enabling them to dynamically respond to environmental cues [52]. Although the extent of EndMT's contribution to the CAF population varies across different cancer types, it represents a crucial alternative mechanism for fibroblast activation within tumors. Further investigation is necessary to delineate the regulatory networks that govern EndMT and its role in shaping the tumor stroma.

2.1.3 Fibroblast Plasticity and Dedifferentiation in the TME

Beyond the recruitment and transdifferentiation of MSCs and endothelial cells, CAFs can also originate from the phenotypic plasticity and dedifferentiation of resident fibroblasts within the TME [53]. Traditionally considered terminally differentiated, fibroblasts have been shown to undergo dynamic reprogramming in response to tumor-derived signals, shifting among quiescent, activated, and CAF-like states [54]. This plasticity enables fibroblasts to adapt to the evolving TME and contributes to CAF heterogeneity.

Key regulators of fibroblast plasticity include TGF-β, PDGF, and FGF, which activate signaling pathways such as TGF-β/SMAD, JAK/STAT, and Wnt/β-catenin [55]. Under persistent stimulation, fibroblasts can lose their homeostatic identity and adopt a more primitive, mesenchymal-like state. This transition is characterized by the downregulation of quiescence markers, such as caveolin-1, and the upregulation of CAF markers, including α-SMA and FAP [56, 57]. Additionally, epigenetic modifications, including DNA methylation and histone remodeling, play a crucial role in stabilizing the CAF phenotype [58, 59].

Importantly, fibroblast dedifferentiation may serve as a reversible mechanism, allowing cells to transition between different activation states in response to tumor-derived cues [60]. This plasticity not only expands the potential sources of CAFs but also reinforces their heterogeneity, as fibroblasts from diverse tissue origins may respond differently to oncogenic signals [61]. Further investigation into the molecular drivers of fibroblast plasticity and dedifferentiation is essential for understanding CAF diversity and its impact on tumor progression.

2.2.1 CAF Subtypes: MyCAFs, Senescent CAFs, and Activated CAFs

CAFs exhibit profound functional and molecular heterogeneity, with distinct subtypes contributing to tumor progression through various mechanisms. MyCAFs, characterized by the expression of αSMA, play a central role in ECM remodeling and tissue contractility. In pancreatic ductal adenocarcinoma (PDAC), myCAFs promote immunosuppression by recruiting Treg cells and expressing programmed death-ligand 1 (PD-L1). In contrast, in breast cancer, they enhance tumor invasion by secreting CXCL12 and facilitating collagen contraction [62, 63]. Notably, epidermal growth factor receptor (EGFR)/ERBB2 signaling in PDAC myCAFs drives metastasis through an autocrine loop involving amphiregulin, highlighting their therapeutic vulnerability [12].

senCAFs, characterized by a senescence-associated secretory phenotype (SASP), secrete proinflammatory cytokines and ECM components that suppress antitumor immunity. In breast cancer, senCAFs inhibit natural killer (NK) cell activity through collagen deposition, which correlates with tumor recurrence [64]. In PDAC, p16⁺ senCAFs localized near ductal structures promote immunosuppression and chemoresistance, effects that can be alleviated by senolytic agents [65]. Additionally, hypoxia exacerbates their SASP in esophageal cancer, driving cancer stemness through the secretion of insulin-like growth factor 1 (IGF1) [66].

Activated CAFs, induced by TGF-β or inflammatory cytokines, exhibit protumorigenic phenotypes. In colon cancer, these cells reprogram the TME through the induction of EMT and glycolytic remodeling, thereby fostering immunosuppression [67]. In gastric cancer, activated CAFs secrete paracrine factors that enhance resistance to 5-fluorouracil [68]. Additionally, Snail1⁺ fibroblasts in colon cancer stimulate tumor cell migration via MCP-3-mediated signaling [69].

Interactions among CAF subtypes are dynamic and context dependent. For example, hypoxia can reprogram myCAFs into senCAFs, thereby amplifying immunosuppression. Targeting these subtypes offers therapeutic opportunities: senolytic agents can delay tumor growth by eliminating senCAFs, while inhibition of the EGFR preferentially suppresses metastatic myCAFs in PDAC. Additionally, spatial transcriptomics has shown that POSTN⁺ myCAFs in non-small-cell lung cancer (NSCLC) correlate with resistance to immunotherapy [70]. A deeper understanding of CAF heterogeneity and plasticity will be crucial for refining therapeutic strategies and improving clinical outcomes in cancer treatment (Figure 1).

2.2.2 Molecular Signatures of CAFs: Transcriptomic and Proteomic Profiling

The molecular heterogeneity of CAFs has been systematically dissected through transcriptomic and proteomic profiling, revealing distinct functional subtypes and regulatory mechanisms. Single-cell RNA sequencing (scRNA-seq) has emerged as a cornerstone technology for classifying CAFs into transcriptionally distinct subsets. For instance, pan-cancer analyses have identified four major subtypes: iCAFs, matrix CAFs (mCAFs), proliferative CAFs, and metabolic CAFs, each characterized by unique gene expression profiles. iCAFs are enriched in chemokines such as CCL11 and CXCL12, which mediate immune cell recruitment, whereas mCAFs exhibit high expression of ECM remodeling genes, including MMP11 and COL1A1, reflecting their role in tissue remodeling [71]. Notably, cancer type-specific CAF subtypes have been reported, such as vascular CAFs, marked by NOTCH3 and COL18A1 in breast cancer, and tumor-associated CAFs (tCAFs), enriched in MME and NDRG1 in pancreatic cancer [72].

The functional specialization of CAFs is further elucidated by lineage-tracing studies. For instance, macrophage–myofibroblast transition, driven by Smad3 signaling, generates CAFs with protumorigenic properties in NSCLC [73]. In gastric cancer, scRNA-seq has identified ECM CAFs that promote tumor invasion and correlate with poor prognosis. Similarly, glioblastoma CAFs have been shown to enhance stemness and angiogenesis through growth factor secretion and ECM remodeling [74].

Proteomic profiling complements transcriptomics by validating functional proteins and revealing spatial dynamics. For instance, periostin and CD36 have been identified as markers of CAFs in ovarian cancer, with their spatial distribution correlating with patient survival [75]. In pancreatic cancer, the LAMA5/ITGA4 axis drives acinar-to-ductal metaplasia, a precursor to carcinogenesis, through STAT3 activation [76]. Additionally, proteomics has uncovered NNMT as a master metabolic regulator in ovarian CAFs, mediating DNA methylation and contributing to drug resistance [59]. The integration of multiomics has provided deeper insights into CAF biology. A cross-species study identified three conserved CAF phenotypes: steady-state-like, mechanoresponsive, and immunomodulatory, highlighting their evolutionary conservation and therapeutic vulnerability [77]. In glioblastoma, fibronectin (FN1) has been validated as a key CAF-secreted factor that promotes mesenchymal transition and invasion [78].

Clinically, CAF-derived molecular signatures show promise as biomarkers. For instance, collagen XII has been shown to predict metastatic relapse in breast cancer [79], while APOE–LRP5 signaling correlates with poor outcomes in ovarian cancer [75]. Targeting CAF-specific pathways is emerging as a promising therapeutic strategy. The molecular heterogeneity of CAFs underscores their dynamic role in tumor progression. Although technical limitations and functional plasticity pose challenges, these findings highlight the necessity of integrating multiomics with spatial and functional analyses to fully elucidate CAF-mediated mechanisms. Such efforts will be critical for translating CAF heterogeneity into precision oncology.

2.2.3 Functional Roles in ECM Remodeling, Growth Factor Secretion, and Angiogenesis

CAFs exhibit remarkable molecular and functional diversity, leading to tumor-specific contributions in ECM remodeling, growth factor signaling, and angiogenesis [80]. These processes vary among CAF subpopulations and are influenced by cancer types, genetic mutations, and microenvironmental cues.

CAFs dynamically restructure the ECM through collagen deposition, degradation, and crosslinking, creating a mechanically and biochemically heterogeneous niche. For instance, in colorectal cancer (CRC) liver metastases, portal-derived CAFs adopt a contractile myofibroblastic phenotype and produce collagen IV to promote desmoplasia. This stromal stiffening paradoxically restricts cancer cell proliferation while enhancing metastatic colonization by inducing integrin-mediated signaling [81]. Conversely, in oral squamous cell carcinoma (OSCC), CAFs secrete small extracellular vesicles (sEVs) enriched in lysyl oxidase (LOX), which triggers collagen crosslinking and EMT via the p-focal adhesion kinase (FAK)/p-paxillin/YAP axis. Notably, LOX-positive sEVs are preferentially secreted by αSMA-negative CAFs, highlighting subpopulation-specific ECM-modifying strategies [82, 83]. Matrix metalloproteinases (MMPs) and tissue inhibitors of metalloproteinases (TIMPs) further emphasize CAF heterogeneity. In hepatocellular carcinoma (HCC), CAFs adjacent to tumors express MMP-9 and TIMP-1 in a spatially restricted manner, promoting ECM degradation at the invasion front while preserving structural integrity elsewhere [84, 85]. This dual role is mirrored in Ewing sarcoma, where CAF-like cells secrete protumorigenic ECM proteins, such as periostin, to support metastasis while simultaneously expressing TIMP-2 to limit excessive ECM breakdown [86]. Such functional specialization ensures precise ECM remodeling tailored to the stages of tumor progression.

CAFs secrete a diverse array of growth factors, cytokines, and chemokines that modulate tumor–immune crosstalk and metabolic reprogramming. However, their secretory profiles are not generic; rather, they align with specific CAF subtypes. For instance, TGF-β secreted by CAFs in HCC and CRC drives EMT and immunosuppression, yet its efficacy depends on CAF polarization: N2-polarized CAFs enhance TGF-β signaling to recruit immunosuppressive neutrophils, whereas M2-like CAFs prioritize the paracrine activation of PD-L1 expression on tumor cells [81, 84, 87, 88]. Similarly, FGF19 secretion in CRC liver metastases is restricted to inflammatory CAFs (CD14+), which activate the FGFR4–JAK2–STAT3 pathway to promote neutrophil extracellular trap formation. This process creates a nutrient-rich microenvironment by degrading extracellular DNA, in contrast to αSMA⁺ CAFs that primarily sustain fibrotic stroma [88, 89]. Notably, IGF2 secretion by CAFs in pancreatic cancer is associated with EMT-related CAFs, which upregulate PD-L1 to evade T-cell surveillance [6]. These examples illustrate how distinct CAF subpopulations encode specific growth factor signatures to orchestrate tumor progression.

CAF-driven angiogenesis is highly heterogeneous, with subpopulations employing distinct molecular and metabolic strategies to remodel the vasculature. In colon cancer, CAFs adjacent to oncogenic KRAS-mutant tumors secrete VEGF A (VEGFA) and HIF-1α, promoting endothelial sprouting through mTORC2-AKT activation. However, VEGFA expression is selectively upregulated in SULF1-negative CAFs, which lack heparan sulfate proteoglycan modification, thereby increasing VEGFA bioavailability [90, 91]. Conversely, in esophageal squamous cell carcinoma, CAF-derived milk fat globule–EGF factor 8 (MFGE8) binds to integrin receptors on endothelial cells, activating the PI3K/AKT/ERK pathways to enhance vascular permeability. This effect is amplified by lipid metabolism reprogramming in CAFs, which accumulate lipids to provide energy for endothelial proliferation [92]. Epigenetic regulation further diversifies CAF angiogenic functions. FTO-mediated m6A demethylation in conjunctival melanoma CAFs stabilizes VEGFA and EGR1 transcripts, whereas gut microbiota-derived butyrate suppresses angiogenesis by inhibiting histone deacetylase 3 and reducing SULF1 expression in CRC CAFs [91, 93]. The PDPN/CCL2/STAT3 feedback loop in CAFs sustains proangiogenic polarization in CRC; however, this is disrupted in LRRC15+ immunosuppressive CAFs, which instead secrete angiostatic factors such as TIMP-3 [94, 95]. Such contextual angiogenic programs underscore CAF heterogeneity as a critical determinant of vascular remodeling.

In summary, CAF heterogeneity enables specialized contributions to ECM remodeling, growth factor signaling, and angiogenesis, with each role tailored to specific tumor contexts. Unraveling the molecular determinants of these functional specializations will aid in developing precision therapies that disrupt the CAF–TME axis in a subtype-specific manner.

3.1.1 Expressing Immune Checkpoint Ligands

Cancer evades immune surveillance through various mechanisms, with the PD-1/PD-L1 signaling pathway playing a central role by inducing apoptosis in antigen-specific T cells while inhibiting apoptosis in Treg cells [116-119]. Consequently, numerous immunotherapeutic strategies targeting PD-1 and PD-L1 have been developed, leading to significant advancements in cancer treatment [120, 121]. Similar to PD-1/PD-L1, CTLA-4 is a crucial molecule in maintaining an immunosuppressive environment and is expressed in various tumors. Primarily found on T cells, CTLA-4 competes with CD28 for binding to CD80/CD86, thereby blocking the costimulatory signals necessary for T cell proliferation during the initiation phase of the immune response [122]. As the first approved immune checkpoint inhibitor (ICI), the humanized CTLA-4 antibody ipilimumab has transformed clinical cancer treatment, significantly improving the 10-year survival rate of patients with metastatic melanoma [123]. It is widely believed that cancer cells induce CTL apoptosis by expressing PD-L1. However, recent studies have revealed that CAFs can also express PD-L1, thereby contributing to CTL apoptosis.

In recent years, the discovery of PD-L1(+) CAFs in various solid tumors has deepened our understanding of the functions of CAFs. In addition to their previously known role in secreting cytokines to regulate PD-L1 expression in cancer cells, the direct inhibitory effect of CAFs on CTLs must also be acknowledged. In esophageal cancer, cancer cells and fibroblasts mutually enhance PD-L1 expression, contributing to tumor immune suppression. In vivo experiments have demonstrated that anti-PD-L1 antibodies promote the death of both CAFs and cancer cells, resulting in an increase in CD8+ T cells and a reduction in FoxP3+ Treg cells [124]. Due to cancer heterogeneity, PD-L1 expression in CAFs is not universal. Teramoto et al. [125] isolated PD-L1(+) CAFs in NSCLC, observing this subpopulation in 24.8% of patients. Patients with PD-L1(+) CAFs had a significantly better prognosis compared with those with PD-L1(−) CAFs. In vitro experiments further demonstrated that PD-L1 expression in CAFs is reversibly regulated by environmental stimuli, including IFN-γ produced by activated lymphocytes [125]. Zhang et al. [126] found that Jagged1 derived from glioma enhanced CAF proliferation and increased PD-L1 expression in vitro. In gliomas with Jagged1 expression, the levels of Notch1, c-Myc, and PD-L1 were significantly elevated. Furthermore, it was confirmed that Notch1 and PD-L1 expression were localized to CAFs in glioma tissues [126]. Although most other studies primarily attribute PD-L1 expression to cancer cells, we cannot overlook the direct expression of PD-L1 in CAFs, which may explain the differences in the efficacy of immunotherapy in some cancers. High-throughput omics technologies, such as single-cell sequencing, provide an intuitive perspective to observe the role of PD-L1(+) CAFs in the TME [127, 128].

3.1.2 Recruit Immunosuppressive Cells

Immunosuppressive cells within tumors include various cell types. Extensive research across different tumor types has identified Treg cells, MDSCs, tumor-associated macrophages (TAMs), and tumor-associated neutrophils (TANs) as the primary contributors to immune suppression in the TME [104, 106, 129-131]. These cells foster the immunosuppressive microenvironment by secreting inhibitory cytokines, promoting angiogenesis, and recruiting additional immunosuppressive cells [132-134]. CAFs can recruit various immunosuppressive cells through the secretion of multiple cytokines.

A major advance in the understanding of immune regulation is the identification and characterization of a group of CD4(+) T cells, known as Treg cells, which play a critical role in preventing autoimmune responses, including organ-specific autoimmunity, systemic autoimmunity, and colitis. A large number of Treg cells are enriched in tumors, and these cells are highly activated and express elevated levels of coinhibitory molecules such as CTLA-4, PD-1, LAG3, and TIGIT [135-138]. Simultaneously, Treg cells can secrete IL-10, promoting CTL exhaustion [139]. The specific roles and mechanisms of Treg cells in the tumor immunosuppressive microenvironment have been extensively reviewed. Here, we focus on the impact of CAFs on this T cell population [140]. Research by Costa et al. [21] demonstrated that CAFs attract CD4+ CD25+ T lymphocytes by secreting CXCL12 and CAF-S1 and retain them through OX40L, PD-L2, and JAM2. Meanwhile, CAFs can enhance the inhibitory effect of Treg cells on the proliferation of effector T cells. Notably, this work identified four distinct CAF subpopulations in breast cancer, indicating the high heterogeneity of CAFs [21]. apCAFs are a specialized subtype of CAFs derived from mesothelial cells. In pancreatic cancer, mesothelial cells downregulate their mesothelial characteristics and acquire fibroblast-like features, thereby forming apCAFs. These apCAFs express MHC-II molecules and can directly interact with naive CD4+ T cells, inducing their differentiation into Treg cells in an antigen-specific manner [19]. Varveri et al. [141] reported that α-SMA+ CAFs can form immune synapses with Foxp3+ Treg cells within tumors. These α-SMA+ CAFs are capable of phagocytosing and processing tumor antigens, exhibiting a tolerogenic phenotype that promotes the antigen-specific arrest, activation, and proliferation of Treg cells. This study highlighted the immunosuppressive mechanism underlying the synapse formation between α-SMA+ CAFs and Treg cells, which occurs through an autophagy-dependent process [141]. Research by O'Connor et al. [142] indicates that CAFs provide TGF-β during T cell activation, leading to an expansion of activated Treg cells and a subsequent reduction in available IL-2. Inhibition of TGF-β signaling can prevent CAF-driven upregulation of CXCL13 and inhibit the expansion of the Treg cell population [142]. The same phenomenon was observed in SCLC of neuroendocrine origin [143].

However, some studies have shown that specific subgroups of CAFs do not promote Treg cells; instead, they can inhibit Treg cell activity. Zhao et al. [144] identified a novel CAF subgroup in OSCC that is CD68-positive and highly expresses CD68. The presence of high CD68+ CAFs was linked to tumor initiation. Interestingly, a higher proportion of tumor-supportive Treg cells was observed in patients with low CD68+ CAFs. Mechanistically, knockdown of CD68 in CAFs led to the upregulation of chemokines CCL17 and CCL22 in tumor cells, thereby enhancing Treg cell recruitment [144]. Similarly, Pei et al. [145] found that CAFs express CD1d and activate NKT cells under stress conditions. These NKT cells are considered crucial players in the body's antitumor immune response [145, 146]. This research highlights the critical role of CAFs in the proliferation and activation of Treg cells within the TME. However, due to the high heterogeneity of CAFs, some studies have also explored their potential in promoting antitumor immunity. Further classification of CAF subgroups and targeting specific markers for treatment may offer a promising approach to enhancing therapeutic efficacy.

MDSCs are an immature myeloid cell population that arises under pathological conditions and is characterized by their ability to suppress T-cell activity in cancer [147]. CAFs actively recruit MDSCs by secreting cytokines and chemokines, thereby contributing to the establishment of an immunosuppressive TME. For instance, Zhu et al. [148] demonstrated in HCC that CD36(+) CAFs secrete macrophage migration inhibitory factor to promote MDSC accumulation. Additionally, these CAFs enhance PD-L1 expression through STAT3 signaling, thereby suppressing CD8+ T cell activity and reinforcing the CAF–MDSC-driven immunosuppressive axis [148]. Similarly, Kumar et al. [149] found that CAFs produce GM-CSF and IL-6 to recruit polymorphonuclear MDSCs, which contribute to resistance against CSF1R blockade. This highlights CAF-driven MDSC regulation as a key mechanism underlying therapy resistance [149]. The crosstalk between CAFs and MDSCs involves multiple signaling pathways. Yang et al. [150] identified STAT3 activation in CAFs as a key driver of CCL2 upregulation, which recruits MDSCs through the CCL2–CCR2 axis and enhances their immunosuppressive functions. In lung squamous cell carcinoma, Xiang et al. [151] demonstrated that CAFs promote the differentiation of monocytic MDSCs through ROS/p38 MAPK pathway activation, establishing a link between oxidative stress and CAF-mediated immunosuppression. In pancreatic cancer, Bianchi et al. [152] confirmed that CAFs activate the CXCL1/CXCR2 axis to induce neutrophil-derived TNFα production, promoting an MDSC-enriched TME and highlighting the pivotal role of chemokine networks in tumor immunosuppression. CAFs also regulate MDSCs indirectly through epigenetic and exosomal mechanisms. Zhao et al. [153] reported in esophageal squamous cell carcinoma that CAF-derived IL-6, in synergy with exosomal miR-21, activates STAT3 signaling, driving MDSC expansion and conferring resistance to cisplatin. Additionally, CAF-secreted EVs deliver immunosuppressive cargo that directly enhances MDSC-mediated T-cell suppression, emphasizing the multifaceted complexity of CAF–MDSC interactions [154].

Targeting the CAF–MDSC axis may enhance immunotherapy efficacy. Khalaf et al. [155] suggested that neutralizing CAF-derived CCL2 or CXCL12 could help remodel the TME and alleviate immunosuppression. Recent work by Akiyama et al. [156] demonstrated that dual PDGFRα/β blockade reduces MDSC infiltration and synergizes with anti-PD-1 therapy, providing a preclinical rationale for combination strategies in cancer treatment. In summary, CAFs recruit and activate MDSCs through secreted factors, signaling pathways, and exosomal communication, forming a cascading immunosuppressive network. Therapeutic cotargeting of this axis has the potential to overcome current limitations in cancer immunotherapy.

TAMs are macrophages that play a crucial role in the formation of the TME and are commonly found in various tumors [157, 158]. TAMs are generally classified into two main subtypes within the TME: M1 and M2. M1 macrophages are involved in antitumor immunity, whereas M2 macrophages promote immunosuppression and support tumor progression [159]. The primary CAF subtypes (iCAF, myCAF, and apCAF) are essential components of the TME. Their interaction with TAMs is primarily mediated through cytokine secretion. iCAFs predominantly contribute to macrophage enrichment, whereas myCAFs play a major role in TAM polarization.

CAFs can recruit monocytes to the tumor site through various signaling pathways. Similar to the recruitment of MDSCs, CAFs may recruit monocytes to the tumor site via the CCL2–CCR2 signaling pathway. In a study on esophageal squamous cell carcinoma, Higashino et al. [160] found that CAFs induce M2 polarization of TAMs, with FAP expression being a key factor driving the tumor-promoting and immunosuppressive phenotype of CAFs. Similar results were observed in studies on intrahepatic cholangiocarcinoma [150]. The CXCL16 chemokine derived from CAFs may recruit monocytes, promote stromal activation, and facilitate tumor progression in TNBC [161]. Likewise, CXCL14 produced by CAFs can enhance macrophage recruitment to the tumor site, promoting prostate tumor growth [162]. Among the pathways involved in monocyte recruitment, the CXCL12–CXCR4 axis has been extensively studied. CAFs produce high levels of CXCL12 in the TME, and the CXCL12–CXCR4 axis plays a crucial role in recruiting monocytes to the tumor tissues [163]. The iCAF–TAM axis primarily activates the complement cascade pathway through the interaction of complement C5 and its receptor C5AR1. The C5 pathway serves as a crucial chemokine for recruiting immunosuppressive myeloid cells, ultimately leading to the inhibition of T-cell activity [164]. The interaction of the C3–C3aR iCAF–TAM axis has been further elucidated in melanoma, head and neck cancers, and breast cancer [165]. CD34+ CAFs produce C3 and convert it to its activated form (C3a) in the TME, facilitating the recruitment of C3aR+ circulating monocytes. CAFs primarily promote the M2 polarization of TAMs by secreting cytokines. In summary, a variety of cytokines, including IL-8, IL-33, IL-10, TGF-β, and CCL2, can be secreted by CAFs, and they have been shown to promote M2 polarization of TAMs in the TME [39, 166-169].

The role of TANs in cancer progression and tumor immunity remains controversial. In many solid tumors, a high neutrophil-to-lymphocyte ratio is associated with poor overall survival (OS). Extensive research indicates that neutrophils suppress both innate and adaptive immunity, thereby facilitating tumorigenesis [170-174].

Multiple studies have demonstrated that CAFs recruit immunosuppressive neutrophils through specific chemokine signaling pathways, thereby shaping an immunosuppressive TME. In HCC, CAFs were shown to secrete CXCL12, which directly recruits CXCR4+ neutrophils that impair CD8+ T cell function via arginase-1 (ARG1) production and reactive oxygen species (ROS)-mediated suppression [6]. Spatial transcriptomic analysis in CRC further revealed neutrophils densely clustered within CAF-enriched immune hubs, where their spatial colocalization with CAFs correlates with upregulated PD-L1 expression and adaptive immune resistance [175]. The functional interplay between CAFs and neutrophils extends beyond chemotaxis; in breast cancer models, fibroblast-derived IL-33 activates neutrophil ST2 receptors, driving IL-10/TGF-β production that polarizes Th2-type immunity [167]. scRNA-seq studies in breast cancer additionally identified bidirectional CXCL8–CXCR1/2 axis communication between CAFs and neutrophils, amplifying N2 polarization characterized by ARG1+ and CCL2+ phenotypes that inhibit dendritic cell (DC) maturation [176]. Recent spatial mapping in lung cancer uncovered CAF-organized fibrotic barriers that spatially confine immunosuppressive neutrophils at tumor-invasive fronts while excluding cytotoxic T cells, a spatial configuration associated with poor clinical outcomes [177]. Similarly, pancreatic cancer liver metastasis models demonstrated stable ICAM-1/β2-integrin-mediated adhesions between CAFs and neutrophils that cooperatively block T cell transmigration [178]. Clinically, CAF-mediated neutrophil recruitment mechanisms are particularly significant in premetastatic niche formation, with CRC studies identifying CXCL1/2–CXCR2 and CSF3R as potential therapeutic targets to disrupt this axis [179]. Emerging spatial intervention strategies suggest that disrupting CAF-neutrophil cluster colocalization could enhance PD-1 blockade efficacy, evidenced by a 3.2-fold increase in treatment response in preclinical models through microenvironmental reprogramming [175]. Collectively, these findings position CAF-neutrophil crosstalk as a critical orchestrator of immunosuppression across multiple malignancies, offering novel therapeutic opportunities through spatial and functional modulation of tumor-stromal interactions.

In summary, CAFs contribute to the formation of an immunosuppressive TME by recruiting immunosuppressive cells through cytokine secretion and other mechanisms, thereby inhibiting antitumor immunity. However, the diverse functional roles of different CAF subpopulations highlight the complexity of their interactions. Further investigation using omics technologies to elucidate the crosstalk between specific CAF subtypes and immunosuppressive cells is crucial. These insights may pave the way for novel therapeutic strategies targeting CAF-mediated immunosuppression in cancer immunotherapy.

3.1.3 Secretion Inhibitory Factor

In addition to recruiting immunosuppressive cells through cytokine secretion, CAFs can directly suppress the activity of antitumor immune cells, particularly CD8+ T cells, by releasing immunosuppressive cytokines. TGF-β, a key cytokine secreted by CAFs, plays a crucial role in shaping the immunosuppressive TME [115, 180]. The activation of the TGF-β signaling pathway in the TME has been extensively documented in previous reviews [181, 182]. The study by Li et al. [183] demonstrates that Ln-γ2 is transcribed and activated via the JNK/AP1 signaling pathway in response to TGF-β1 secreted by CAFs. This process alters T cell receptor expression, thereby preventing T cell infiltration into tumor nests [183]. More importantly, TGF-β directly suppresses the function of effector T cells, further contributing to immune evasion in the TME. In melanoma, research by Ahmadzadeh et al. [184] demonstrated that TGF-β1 not only inhibits the acquisition of effector functions in human memory CD8+ T cells and tumor-infiltrating lymphocytes but also suppresses their expression.

Compared with effector T cells, the antitumor effects of NK cells have been studied to a lesser extent. NK cells can rapidly kill multiple adjacent cells that display surface markers associated with cancer. This unique ability, combined with their capacity to enhance antibody and T-cell responses, positions NK cells as key anticancer agents [185-188]. Despite the various mechanisms tumors may develop to evade NK cell attacks, recent advancements in CAR-NK cell therapy have shown promising potential as a novel treatment approach [189]. CAFs suppress NK cell activation by secreting prostaglandin E2 (PGE2) and indoleamine 2,3-dioxygenase (IDO), thereby inhibiting key activating receptors such as DNAM-1, NKp44, and NKp30. This immunosuppressive mechanism has been observed in liver, breast, and CRCs [190-192]. Studies have demonstrated that CAF inhibitors can enhance the efficacy of CAR-NK cell therapy by reducing IL-6 secretion from CAFs, thereby mitigating immunosuppressive effects in the TME [193]. A study has also shown that in gastric cancer, CAFs induce ferroptosis in NK cells by regulating iron metabolism, which contributes to immune evasion [194]. In breast cancer, a distinct subpopulation of senescent CAFs secretes a specialized ECM that specifically impairs NK cell cytotoxicity, thereby facilitating tumor progression [64]. Similarly, CAFs express Dickkopf-1, a Wnt/β-catenin inhibitor, which suppresses NK cell activation and cytotoxicity by downregulating AKT/ERK/S6 phosphorylation [195]. Remarkably, CAFs can induce their own lysis and downregulate the expression of activating receptors on NK cells through ligand–receptor interactions. This process promotes cancer cell evasion from NK cell surveillance, further contributing to tumor immune escape [196].

In PDAC, NetG1-associated CAFs demonstrate intrinsic immunosuppressive properties that inhibit NK cell-mediated tumor cell killing. This suppression is regulated by the NetG1 downstream signaling pathway, which involves AKT/4E-BP1, p38/FRA1, vesicular glutamate transporter 1, and glutamine synthetase [197]. Similar inhibitory effects have also been observed in endometrial cancer [198]. Notably, radiotherapy does not eliminate this suppression, as CAFs maintain their ability to inhibit NK cell activity even after treatment [199].

Recent studies have demonstrated that CAFs inhibit NK cells through various mechanisms. This complexity is evident not only in the heterogeneity across different tumor types but also in the diverse regulatory pathways within the same tumor, as seen in breast cancer. Such variability poses a challenge to the precision of tumor therapy. A potential strategy to enhance treatment efficacy is to consider multiple pathways when selecting therapeutic targets, thereby overcoming CAF-mediated immune suppression.

Regulating the immune microenvironment of tumors is a critical function of CAFs. By secreting cytokines to recruit immunosuppressive cells, directly inhibiting antitumor immune cells, and expressing immune checkpoints, CAFs play a pivotal role in maintaining the immunosuppressive TME. However, the mechanisms involved are highly complex, with significant heterogeneity observed not only between different tumor types but also among subtypes within the same tumor. To counter the immunosuppressive effects of CAFs, multiple immunotherapeutic approaches should be employed to target as many pathways as possible while also developing precise therapies aimed at specific CAF subtypes. Simultaneously, the potential dual roles of CAFs in both promoting and inhibiting tumors must be carefully considered, and further research is needed to better understand the underlying mechanisms (Figure 2).

3.2.1 Matrix Production and Remodeling

Activation of the TGF-β pathway induces the transformation of fibroblasts into CAFs within the TME [115, 213]. CAFs involved in ECM deposition exhibit myofibroblast-like characteristics and express ECM-related genes, including those encoding collagen, proteoglycans, and glycoproteins [214, 215]. Additionally, CAFs secrete enzymes such as LOXs and hydroxylases, which catalyze the cross-linking of collagen and elastin, as well as the posttranslational modifications of ECM proteins, thereby contributing to the structural and mechanical properties of the tumor stroma [83, 216].

The deposited ECM contributes to tumor drug resistance by acting as a physical barrier, limiting drug penetration and reducing therapeutic efficacy. In breast cancer, high-density stroma significantly suppresses T-cell proliferation compared with low-density stroma. Prolonged culture in a high-density stromal environment leads to an increased CD4+/CD8+ T-cell ratio. Notably, the surrounding collagen density does not influence cancer cell proliferation [217]. Under the mechanical pressure exerted by a rigid ECM, infiltrating T lymphocyte populations undergo significant alterations in quantity, surface marker expression, subset composition, and gene expression profiles [218]. The motility of cancer cells is inherently limited, preventing them from traversing the ECM solely through their own mechanisms [219]. Studies have shown that CAFs expressing α-SMA can acquire the ability to contract the ECM within the TME. This contractile capability facilitates tumor cell penetration through the dense ECM, thereby promoting tumor invasion [220, 221]. This clarifies why the dense ECM exerts a greater impact on immune cells than on tumor cells and how CAFs facilitate tumor cell migration within the ECM.

The rigidity of the ECM also influences various immune cells, modulating their infiltration, activation, and functional properties within the TME. Studies have shown that the dense ECM in tumors not only directly suppresses the activity of CD8+ T cells but also affects the function of immunosuppressive cells, such as TAMs, thereby enhancing their inhibitory effects on T cells [222, 223]. The primary component of the ECM is collagen. In an in vivo model, knocking out the collagen gene resulted in a decreased OS rate [224, 225]. However, when a LOX inhibitor was combined with the chemotherapy drug gemcitabine for the treatment of early-stage PDAC, it improved OS in mice by reducing metastasis [226]. These findings suggest that ECM remodeling may have a dual role in promoting tumor invasion while also restricting tumor progression.

The ECM facilitates paracrine signaling with tumor cells, activating a series of key biochemical pathways associated with tumor invasion and aggressiveness. These activated pathways include FAK, WNT, and MAPK, leading to EMT, disrupting cell polarity, and ultimately resulting in increased tumor proliferation and invasive capabilities. In the paracrine mechanism, EVs serve as crucial mediators. EVs secreted by CAFs promote tumor invasion and metastasis through multiple pathways, including exosomes and secreted proteins, by activating the FAK pathway [83, 227, 228]. Similarly, the activation of the FAK pathway, the WNT pathway plays a significant role in tumor invasion and metastasis. Through the mediation of the ECM, the activated WNT pathway can enhance tumor invasion by strengthening the stemness of cancer cells and promoting EMT [229-231]. The MAPK pathway is not only a key signaling pathway activated by the paracrine mechanisms of CAFs, but studies have also shown that its activity can distinguish different CAF subtypes in PDAC, highlighting its crucial role in paracrine signaling. Furthermore, MAPK pathway activation is associated with poor prognosis in ovarian cancer and promotes invasion and metastasis in lung cancer [232-235].

CAFs promote tumor invasion and metastasis by secreting ECM components and through ECM-mediated paracrine signaling. However, under certain conditions, a stiff ECM can paradoxically restrict tumor cell invasion, an intriguing phenomenon observed in numerous studies. This highlights the challenges associated with simply eliminating ECM-producing CAFs or employing ECM-targeted therapies. The contractile effect of CAFs within the TME warrants particular attention, as it represents a mechanism by which tumor cells regulate ECM stiffness. Disrupting this contractile effect could potentially leverage the ECM to suppress tumor cell invasiveness, thereby transforming CAFs into therapeutic allies. Additionally, studies indicate that reducing the primary ECM component, collagen, does not effectively treat tumors. Instead, targeting enzymes responsible for collagen cross-linking may offer a more promising approach. Similarly, inhibiting the complex paracrine signaling pathways within the ECM could theoretically mitigate fibroblast-induced tumor progression, but identifying precise therapeutic targets is essential. In this regard, spatial omics provides a powerful tool for gaining deeper insights into ECM dynamics and fibroblast–tumor interactions.

3.2.2 Angiogenesis

CAFs play a crucial role in tumor angiogenesis. In tumors, CAFs and tumor-associated blood vessels (TABVs) have been observed to colocalize [236, 237]. CAFs serve as the primary source of tumor-derived VEGFA; however, they can also support tumor angiogenesis through VEGFA-independent mechanisms [238-241]. For instance, CAF-derived PDGFC sustains angiogenesis by stimulating the secretion of proangiogenic growth factors such as FGF2 and osteopontin [242-244]. Furthermore, the CAF secretome enhances tumor angiogenesis by attracting vascular endothelial cells and recruiting monocytes from the bone marrow. Studies have also demonstrated that CAFs promote angiogenesis by activating the PI3K/AKT and TGF-β signaling pathways, further underscoring their critical role in tumor vascularization [92, 245]. The integrated stress response (ISR) is a homeostatic mechanism that links cell growth and survival to bioenergetic demands. Verginadis et al. [95] demonstrated that the activation of the ISR stimulates CAFs surrounding blood vessels, thereby driving tumor angiogenesis. Tumor-associated stromal cells, derived from mesenchymal tissues, play a crucial role in the formation of TABVs, with CAFs serving as key mediators. CAFs contribute to the TME by providing the ECM, secreting cytokines, and recruiting bone marrow-derived cells. Currently, antiangiogenic therapies primarily target VEGF and VEGFR, and in certain cancer types, their combination with chemotherapy or immunotherapy has significantly improved survival outcomes [246]. Enhancing treatment efficacy by integrating more precise therapeutic strategies targeting CAFs holds great promise and warrants further investigation.

4.1.1 Positron Emission Tomography

PET combined with computed tomography (CT) is a widely utilized nuclear medicine technique that uses positron-emitting radioisotope-labeled compounds to visualize and evaluate biochemical processes in vivo [280]. This integration offers critical insights into both metabolic activity and the precise anatomical localization of diseases. The most commonly used tracer, fluorine-18 fluorodeoxyglucose (¹⁸F-FDG), facilitates the visualization of glucose uptake in tumors and metastases. Since cancer cells exhibit increased metabolic activity, they absorb more tracer, appearing as brighter regions on imaging [281-283]. However, ¹⁸F-FDG PET has limitations in distinguishing malignant tumors from benign inflammatory or infectious processes due to the nonspecific uptake of glucose [280]. In contrast, FAP inhibitor (FAPI) tracers function independently of glucose metabolism, minimizing background signals in regions such as the stomach, intestines, and lungs [284]. FAPI exhibits stable uptake, high image quality, and superior clinical performance across various tumor types, positioning it as a promising alternative to ¹⁸F-FDG for oncologic imaging. Various enzyme inhibitors targeting FAP, including the widely used FAPI-04 and FAPI-46, as well as the more selective UAMC-1110, have been thoroughly discussed in previous reviews [284]. In this section, we highlight recent advances in the field, focusing on the novel small-molecule tracer OncoFAP, which represents a promising development in FAP-targeted imaging and therapy.

The use of peptide and pseudo-peptide-based targeting methods for FAP represents a promising alternative to enzyme inhibitors and has the potential for clinical application. OncoFAP is a highly specific ligand for FAP, offering broad tumor-targeting capabilities. In animal models, Millul et al. [285] reported that OncoFAP demonstrated effective targeting of tumors. Backhaus et al. [286] conducted an experiment involving 12 patients with various types of tumors to evaluate the use of ⁶⁸Ga–OncoFAP–DOTAGA–PET/CT and PET/MRI, which showed favorable biodistribution and kinetics. The tracer demonstrated high and reliable uptake rates in primary cancers, binding to human FAP at sub-nanomolar concentrations; due to its low molecular weight, the tracer accumulated rapidly and displayed low immunogenicity [286]. It is noteworthy that relatively successful preclinical studies have been reported for applications in both PET/CT and targeted radionuclide therapy (TRT). In a preclinical study, the modified ¹⁷⁷Lu–OncoFAP (OncoFAP-23) showed improved targeting and higher tumor uptake [287]. OncoFAP combined with IL-2 targeting resulted in increased tumor uptake [288]. While it may be premature to suggest that OncoFAP will replace the established FAPI-04, its development warrants close attention. Clinical case reports on OncoFAP have demonstrated promising efficacy and safety [289, 290]. However, several key challenges must be addressed before it can be broadly adopted in clinical practice. The first challenge is scalability. While dimeric and trimeric forms of OncoFAP exhibit improved tumor retention, their synthesis involves complex chemical coupling steps that require stringent control over process stability and purity, potentially limiting feasibility for large-scale production. The second challenge involves regulatory hurdles. Clinical translation of OncoFAP necessitates progression through multiple trial phases to establish safety and efficacy. Phase III trials, in particular, require large patient cohorts and extended follow-up periods, resulting in considerable time and financial investment. Furthermore, variations in regulatory frameworks and approval standards across countries may prolong the review process. The third challenge is high production costs. The synthesis of OncoFAP relies on high-purity intermediates, with significant expenses associated with raw materials and purification procedures. Addressing these three challenges will be critical for advancing the clinical development and widespread application of OncoFAP.

The imaging efficacy of FAP-targeted imaging is closely linked to the abundance of FAP in specific cancers. FAPI demonstrates strong imaging performance in cancers characterized by high stromal content, exhibiting uptake comparable to the commonly used ¹⁸F-FDG PET/CT, but with a higher tumor-to-background ratio (TBR), particularly in the assessment of brain metastases. However, in cancers with high heterogeneity, such as lung cancer, FAPI has not yet demonstrated a clear advantage in imaging primary lesions. Nonetheless, it offers significant benefits in imaging metastases and distinguishing malignant tumors from benign lesions.

The diagnostic efficacy of FAPI PET/CT in breast cancer has been extensively investigated. Several clinical studies have demonstrated the advantages of FAPI PET/CT over ¹⁸F-FDG PET/CT, particularly regarding uptake and TBR. Sahin et al. [291] examined the advantages of ⁶⁸Ga–FAPI PET/CT in imaging invasive lobular breast cancer (ILC). Their retrospective analysis of 23 female ILC patients with hormone-positive, HER2-negative tumors showed that ⁶⁸Ga–FAPI PET/CT had higher TBR and SUVmax in primary tumors, as well as improved sensitivity and uptake for detecting axillary lymph nodes and distant metastases [291]. In a separate study involving 20 female breast cancer patients, FAPI imaging demonstrated significantly higher SUVmax and TBR values for primary tumors, lymph nodes, lung, and bone metastases compared with FDG [292]. Elboga et al. [293] compared ⁶⁸Ga–FAPI-04 and ¹⁸F-FDG PET/CT in 48 breast cancer patients and found ⁶⁸Ga–FAPI PET/CT to be superior in detecting primary breast cancer SUVmax. Zheng et al. [283] also confirmed FAPI PET/CT's superiority over FDG in detecting primary breast tumors based on SUVmax and TBR. Although Ballal et al.’s study [294] of 47 breast cancer patients using ⁶⁸Ga–DOTA.SA. FAPI did not show significant differences in some indicators, it still outperformed ¹⁸F-FDG overall. The high density of stromal components and the abundant presence of CAFs in breast cancer, along with the stable expression of FAP, are likely key factors contributing to the excellent imaging performance of FAPI PET in this tumor type [97, 295].

In contrast to breast cancer, the situation regarding lung, gastric, and head and neck cancers differs. Due to tumor heterogeneity, FAPI did not demonstrate a significant advantage in the uptake of primary lesions. However, it showed a clear benefit over ¹⁸F-FDG PET/CT in terms of the TBR, with this advantage being particularly pronounced in metastatic lesions. FAPI PET/CT and FDG PET/CT exhibited no significant differences in imaging primary lung cancer. In a study of 34 advanced LC patients, Wang et al. [296] found no significant differences in delineating the primary tumor and metabolic volume between the two methods. Similarly, Giesel et al. [297], in a study of 41 primary tumors (including nine LC patients), observed no significant difference in uptake between FAPI and FDG at the primary tumor level (p = 0.429), although the comparison did not include uptake rates among LC patients. A prospective analysis by Wu et al. [298] in 28 newly diagnosed NSCLC patients also found no difference between FAPI PET/CT and FDG PET/CT in detecting LC. The high heterogeneity of lung cancer may lead to variable FAP expression among different patients, which could explain the inconsistent imaging results observed with FAPI.

Due to the limited number of human trials available for FAPI PET/CT, both in terms of case volume and large-scale comparative studies, assessing its effectiveness for specific types of cancer remains challenging. In gastric, head and neck, and CRCs, FAPI has demonstrated superior imaging performance compared with ¹⁸F-FDG PET/CT in certain clinical trials. However, due to the limitations in study scale, it is premature to conclude that FAPI can replace ¹⁸F-FDG. The imaging performance of FAPI PET/CT in gastric cancer appears to vary across different studies. In a 2022 study by Lin et al., ⁶⁸Ga–DOTA–FAPI-04 PET/CT showed superior detection efficacy compared with ¹⁸FDG PET/CT in 56 patients with long-term proven gastric carcinomas, as evidenced by a higher SUVmax (10.3 vs. 8.1, p = 0.004) and TBR (11.6 vs. 5.8, p < 0.001) in the primary tumor. Similar results were found in two other studies [299, 300]. However, Kuten et al. [301] conducted a clinical study involving 10 patients with gastric adenocarcinoma undergoing initial staging. They found that, although ⁶⁸Ga–FAPI-04 PET/CT demonstrated a significant advantage in TBR values compared with ¹⁸FDG PET/CT (p = 0.007), there was no significant difference in SUVmax (p = 0.139) [301]. Another study compared the performance of ⁶⁸Ga–FAPI-04 PET/CT with ¹⁸F-FDG PET/CT in 21 patients with pathologically proven newly diagnosed or recurrent gastric adenocarcinoma. For primary gastric cancer, no significant differences were observed in SUVmax (p = 0.247) and TBR (p = 0.117) between the two groups. Interestingly, FAPI demonstrated significantly better imaging performance for other metastases, including lymph nodes, peritoneum, bone metastasis, and liver metastasis [302]. Contradictory results have also been observed in head and neck cancers. While several studies have demonstrated the effectiveness of FAPI in imaging these cancers [303, 304], other studies suggest that FAPI does not provide a significant advantage over ¹⁸F-FDG [305, 306]. In the diagnosis of CRC, the diagnostic value of FAPI PET/CT is evaluated for both the primary lesion and disease staging, focusing on parameters such as SUVmax and TBR [307, 308]. However, some comparative studies have highlighted the nonsignificance of certain key parameters, indicating the need for more clinical research to validate the effectiveness of FAPI in CRC [309-311]. Studies utilizing FAPI PET/CT have also been reported in the diagnosis of other tumors, such as cholangiocarcinoma [312], cervical cancer [313], HCC [314], pancreatic neuroendocrine tumors [315], and glioma [316, 317]. However, due to the limited number of cases, no definitive conclusion can be drawn regarding the effectiveness of FAPI in these cancers.

Metabolic variability in the liver is likely a key factor influencing the stability of FAPI PET tracers. To address this challenge, optimizing the molecular structure and introducing functional modifications represent promising strategies. Specifically, the incorporation of rigid linkers [318] and metabolic blocking groups may help reduce susceptibility to hepatic enzyme degradation and minimize nonspecific liver uptake. In parallel, the development of multimeric probes and refinement of albumin-binding strategies can further decrease liver accumulation while enhancing tumor-specific uptake, thereby improving the overall imaging performance of FAPI tracers [319].

FAPI PET has demonstrated superior performance over FDG PET in detecting metastatic lesions, particularly in the brain, digestive tract, and liver, primarily due to its higher TBR. This advantage is largely attributed to the limitations of FDG PET in regions with high physiological glucose metabolism, which can obscure lesion detection. Additionally, FDG PET is less effective in identifying low-metabolic tumors, as these lesions exhibit minimal glucose uptake. In contrast, the relatively stable expression of FAP within such lesions enables FAPI PET to offer improved sensitivity and diagnostic accuracy in detecting both high- and low-metabolic metastatic sites [296, 320, 321].

In conclusion, FAPI PET/CT is an emerging and promising diagnostic tool for cancer, demonstrating its diagnostic value in multiple clinical studies across various cancer types. However, current research is insufficient to support its adoption as a guideline-altering imaging method to replace ¹⁸F-FDG PET/CT. The variability in expression across different cancer types may limit its utility as a broad-spectrum diagnostic approach. Nevertheless, its superior performance in certain cancers suggests the potential to replicate the success of PSMA PET/CT in prostate cancer.

4.1.2 Fluorescence Imaging

In addition to the aforementioned PET/CT imaging, fluorescence visualization technology plays a crucial role in medical research and clinical applications, particularly in tumor imaging and treatment navigation. Fluorescent probes have emerged as indispensable research tools [322] in contemporary chemical biology and biomedical fields due to their advantages, including high spatiotemporal resolution, noninvasiveness, and in situ visualization. These probes have been successfully applied to the dynamic visualization of important biomarkers in living cells and organisms, as well as to the precise diagnosis and treatment of major human diseases, such as in fluorescence-guided surgery [323].

Early surgical resection of tumors is often the optimal choice for treating diseases. However, there remains a local recurrence rate of 15–20% within the first 5 years after surgery for lung cancer [324, 325]. During surgery, accurately identifying and delineating the tumor margin is crucial for ensuring complete resection and minimizing the risk of recurrence [326]. However, traditional visual inspection and imaging methods have limitations in accurately depicting the tumor boundary, especially for tumors that are challenging to detect or have complex shapes. Intraoperative molecular imaging (IMI) offers a solution by providing real-time, high-resolution images of the tumor and its surroundings during surgery [327, 328]. By utilizing tumor-targeting fluorescent dyes and ultra-sensitive camera systems, surgeons can visualize the tumor margin with unprecedented clarity. This capability enables more precise surgical resection, reducing the risk of leaving behind microscopic tumor cells and improving patient outcomes.

Early-generation FAP-targeted probes, such as FAPI-04, exhibited limitations for fluorescence imaging due to rapid blood clearance and short tumor retention times. To overcome these challenges, probes modified with albumin-binding moieties, such as 4-(p-iodophenyl)-butyric acid or Evans blue (EB), were developed to prolong systemic circulation and enhance tumor accumulation [329-331]. Additionally, incorporating hydrophilic peptide linkers has been employed to reduce off-target organ retention and improve overall pharmacokinetic profiles. Recent research has shifted toward expanding the utility of these probes beyond imaging alone, exploring their potential in therapeutic applications. For example, fluorescent probes conjugated with photodynamic therapy agents can generate singlet oxygen under light activation to kill tumor cells [332] while simultaneously offering real-time intraoperative imaging for precise delineation of tumor margins. Furthermore, multimodal platforms, such as the integration of PET/CT with fluorescence imaging, are being developed to facilitate both diagnosis and image-guided treatment in patients with resectable malignant tumors.

To date, there have been no reported cases of targeted FAP fluorescence technology being applied to human IMI. However, several examples of its application in cancer animal models exist. Liu et al. [333] reported a highly sensitive fluorescent probe, cv-fap, for detecting FAPα activity in melanoma, with a detection limit of 5.3 ng/mL. This probe exhibits a high binding affinity and overall catalytic efficiency, enabling it to effectively differentiate between normal cells in mice and malignant melanoma cells in tumor-bearing mice [333]. In the case of fibrosarcoma, Tansi et al. [334] developed activatable FAP-targeting immunoliposomes (FAP-IL) for image-guided detection of spontaneous metastases in mouse models. For HCC, a dual-modal probe validated in both cellular and animal models of liver cancer has been mentioned [335]. Additionally, a fluorescent probe validated at both the cellular and animal levels using the MCF-7 breast cancer cell line was also referenced in the context of breast cancer [148]. Dual-modal probes combining PET/CT imaging and fluorescence imaging have gained increasing attention in tumor diagnosis and treatment research in recent years, with numerous preclinical studies emerging [336-339]. To address the challenge of off-target effects resulting from tumor heterogeneity, dual-modal probes offer a promising solution. For partially resectable tumors, preoperative PET/CT screening to identify high-expression populations, coupled with the selective use of fluorescence for intraoperative navigation, offers a feasible approach. This strategy may help overcome the clinical challenge of unstable imaging.

In summary, the imaging targets for CAFs in cancer diagnosis and treatment mainly focus on FAP, including PET/CT, fluorescence imaging, and other modalities. We have observed the promising potential of FAP-targeted PET/CT in cancer diagnosis, which may serve as a viable alternative to FDG PET/CT for certain cancers. However, the lack of existing clinical trials prevents us from drawing definitive conclusions. Multicenter clinical trials with a larger number of cases are necessary. Fluorescence imaging is primarily aimed at resectable tumors, and the application of dual-modal probes may facilitate the integration of tumor diagnosis and treatment, enhancing precision in both areas.

5.1.1 Transforming Growth Factor-β

Blocking TGF-β is currently the most extensively studied therapeutic approach targeting CAFs. Mechanistic studies have highlighted the immunosuppressive role of TGF-β within the TME, making its combination with PD-1/PD-L1 blockade a promising therapeutic strategy. Additionally, TGF-β signaling is a crucial regulator of CAF maturation. Inhibiting this pathway holds promise for normalizing CAFs or inducing a dormant state, potentially mitigating their tumor-promoting effects [373].

LY2157299 (galunisertib) is a small molecule inhibitor of TGF-β that blocks CAF activation and suppresses its immunosuppressive effects. This drug has been evaluated for efficacy and safety in various cancers, including liver cancer and malignant glioma, both as a monotherapy and in combination with other treatments. However, the results have been mixed, with varying therapeutic effects [374-388]. Ongoing clinical trials are currently assessing its potential in lung, liver, and pancreatic cancers. In response to the drug resistance associated with TGF-β application in pancreatic cancer, some studies have suggested that autocrine chemokines secreted by CAFs can induce their transformation into an inflammatory phenotype. These inflammatory phenotype CAFs promote treatment resistance through paracrine signaling, further complicating therapeutic strategies [389]. LY3200882 is a TGF-β receptor inhibitor that has been tested in clinical trials for various advanced cancers [390, 391]. Studies have demonstrated that its combination with immunotherapy can enhance its therapeutic effect in TNBC [392]. Additionally, LY3200882 promotes the release of interferon-β by macrophages, thereby improving the efficacy of radiotherapy in tumor-bearing mice with head and neck cancer and lung cancer [393]. Vactosertib is an ALK5 antagonist targeting the TGF-β receptor. Several studies are currently underway to explore its therapeutic potential. Notably, it has shown promise in the treatment of desmoid tumors when combined with imatinib [394-396].

5.1.2 CXCL12/CXCR4 and CSF1

CAFs are considered the primary source of CXCL12, which exerts an immunosuppressive effect by binding to its receptor, CXCR4 [397]. Targeting the CXCL12/CXCR4 pathway derived from CAFs has been shown to be effective in mouse models of pancreatic and CRCs, leading to increased T-cell infiltration [398]. Similar results were observed in targeted therapies that antagonize CSF1/CSF1R signaling, indicating the potential for macrophage reprogramming [399, 400]. Studies in melanoma have also demonstrated that targeting CXCR4 results in a reduction of myeloid cells and the activation of NK cells [401]. BL-8040 is a CXCR4 antagonist that has shown therapeutic potential in pancreatic cancer when combined with pembrolizumab and chemotherapy. In healthy individuals, it can rapidly recruit CD34+ cells [402, 403]. AMD3100 is another CXCR4 antagonist with a mechanism of action similar to that of BL-8040. It is used in the treatment of various diseases, including but not limited to cancer. In multiple myeloma and lymphoma, AMD3100 has demonstrated notable therapeutic efficacy [404-410]. The primary function of this class of drugs is to enhance stem cell mobilization and promote T-cell infiltration within the TME.

5.1.3 Other Pathways

IL-6 is a key immunosuppressive factor secreted by iCAFs. IL-6 antagonists have been widely used in the treatment of autoimmune diseases. In cancer therapy, their combination with PD-L1 blockade has shown therapeutic efficacy in mouse models of melanoma and pancreatic cancer by disrupting the immunomodulatory interactions mediated by IL-6 [411, 412]. IL-6 secreted by CAFs can increase the levels of TIMP-1 in the extracellular environment. TIMP-1, in turn, activates the STAT3 signaling pathway, which is considered a key molecular mechanism promoting the proliferation and invasion of breast cancer cells [413, 414].

The Hedgehog (Hh) signaling pathway plays a critical role in the proliferation and invasion of various tumors. While largely inactive in normal adult tissues, it becomes reactivated within the TME [415-417]. CAFs can stimulate Hh signaling through paracrine mechanisms [182, 418]. Smoothened (SMO), a key transmembrane protein in the Hh pathway, mediates this activation, leading to aberrant expression of downstream Gli transcription factors [419, 420]. Several SMO inhibitors have been approved and have demonstrated clinical efficacy in tumors such as basal cell carcinoma [421-424]. A representative drug is vismodegib (GDC-0449), a first-in-class small molecule inhibitor targeting the Hh signaling pathway. In a phase I clinical trial, vismodegib demonstrated a response rate of 58% in patients with advanced basal cell carcinoma, highlighting its therapeutic potential [402]. Gli is the central effector molecule of the Hh signaling pathway. Direct inhibition of Gli activity can effectively disrupt the maintenance of tumor stemness and suppress metastatic potential [425, 426]. GANT61 has shown therapeutic efficacy in osteosarcoma, ovarian cancer, and lung cancer by inhibiting the transcriptional activity of Gli, thereby suppressing tumor growth and progression [427-429]. Hh acyltransferase (Hhat), the enzyme responsible for palmitoylating Sonic hedgehog (Shh), has emerged as a novel therapeutic target for inhibiting Shh signaling in pancreatic cancer cells. By blocking Hhat activity, the posttranslational modification and subsequent signaling of Shh can be disrupted, potentially attenuating tumor growth and progression [430, 431]. RU-SKI 43, a small-molecule inhibitor of Hhat, has demonstrated potential therapeutic efficacy in preclinical models of pancreatic cancer, lung cancer, and breast cancer. By blocking the palmitoylation of Shh, RU-SKI 43 effectively disrupts downstream Hh signaling, thereby impairing tumor growth and progression in these cancer types [432-434].

Targeting key molecules within CAF-associated signaling pathways has produced promising therapeutic outcomes, particularly when combined with ICIs, as this strategy can effectively mitigate immune suppression within the TME. However, due to tumor heterogeneity, a universal or broad-spectrum therapeutic approach has yet to be established. Identifying patient populations and tumor types most likely to benefit from such treatments remains a crucial area for further investigation. In conclusion, evidence from numerous clinical studies supports the viability of targeting CAF-related pathways as a strategy in cancer therapy.

5.2.1 Fibroblast Activation Protein

Eliminating FAP-positive CAFs (FAP⁺ CAFs) has become a key focus in the development of CAF-targeted therapeutic strategies. Since FAP is a membrane protein specifically expressed by CAFs and is involved in nearly all tumor-promoting processes, targeting FAP represents a promising approach to disrupt the protumorigenic functions of the tumor stroma [437-440]. Various therapeutic strategies targeting FAP are currently under investigation, including CAR-T cell therapy, monoclonal antibody therapy, vaccines, and radionuclide-based treatments.

The use of antibodies to eliminate FAP+ CAFs was one of the earliest attempts, primarily aimed at enhancing antitumor immunity. It is important to note that targeting FAP to eliminate the tumor stroma does not inherently lead to antitumor effects. In mouse models of pancreatic cancer, the ablation of FAP⁺α-SMA⁺ CAFs paradoxically resulted in accelerated tumor progression and increased infiltration of Treg cells [436]. The therapeutic efficacy of FAP-targeted antibody monotherapy has not met expectations [441, 442]. This class of agents is primarily represented by the monoclonal antibody F19 and its humanized derivative, sibrotuzumab. Existing studies on sibrotuzumab have primarily focused on evaluating its safety profile, while its therapeutic efficacy has been unsatisfactory [443]. These findings suggest that, although FAP antibodies demonstrate tumor accumulation, their cytotoxic efficacy against tumor cells is limited. Recent studies indicate that conjugating FAP antibodies with immunotoxins may provide a promising strategy to enhance their therapeutic potential. In mouse models of breast cancer and melanoma, FAP antibody-conjugated immunotoxins have exhibited robust antitumor efficacy. However, to date, no clinical trials in humans have been reported, and further research is needed to evaluate their safety and therapeutic potential in clinical settings [444, 445]. Compared with the use of antibodies targeting FAP alone, antibody-conjugated radionuclide therapy and enzyme inhibitor-conjugated radionuclide therapy have shown more promising potential in both preclinical and clinical trials. These approaches offer enhanced therapeutic efficacy and could represent more effective strategies for targeting CAFs.

TRT for FAP, a subspecialty within nuclear medicine, is rapidly advancing. As a pan-cancer target, FAP holds significant potential and should not be overlooked in the development of precision cancer therapies. The earliest achievement in FAP-TRT was the development of the ¹³¹I-labeled monoclonal antibody F19. Several FAP-TRTs are currently being applied to various cancer types, including pancreatic cancer, lung cancer, and gastrointestinal tumors. Covalent targeted radioligand (CTR) represents a novel drug modality that selectively binds radioactive ligands to tumors. This technology enhances tumor uptake and retention of radioligands while minimizing their absorption in circulation and healthy tissues. Cui et al. [446] developed an innovative CTR targeting FAP by incorporating a latent warhead into the ligand. This modification enables covalent bonding to the target protein without compromising affinity, resulting in the irreversible modification of the chelating agent carrying the radionuclide to the target. Upon reaching the tumor, the CTR first binds noncovalently to the target, then accelerates covalent (irreversible) bonding via the proximity effect, potentially enhancing affinity and reducing the tumor's drug clearance rate. Unbound CTR is rapidly excreted from the body. In mice with subcutaneous tumors, the drug demonstrated promising tumor control efficacy [446].

In pancreatic cancer, multiple FAP-targeted probes, such as TEFAPI-06 and 07, as well as ¹⁷⁷Lu–FAPI-46, have demonstrated improved tumor retention and significant tumor growth inhibition in mouse models. Clinical trials using ⁹⁰Y–FAPI-46 in pancreatic cancer have shown disease stabilization, while ¹⁷⁷Lu–FAPI-46 has exhibited favorable targeting and pharmacokinetics [447-453]. In thyroid cancer, ¹⁷⁷Lu–EB–FAPI has resulted in partial remission and disease stabilization in patients with metastatic disease [331, 454, 455]. Studies on sarcoma [451, 453, 456, 457] and glioblastoma [330, 458, 459] also report promising results with FAP-targeted therapies, though adverse events such as hematological toxicity have been observed. Other cancers, including lung and breast cancers, have also shown positive responses to FAP-based TRT, with some patients experiencing significant disease control and reduced metastasis [460-467]. The combined use of TRT and ICIs presents promising therapeutic potential. In preclinical models, treatment with ¹⁷⁷Lu–LNC1004 has been shown to induce a transient upregulation of PD-L1 expression on tumor cells, thereby sensitizing tumors to ICI therapy and enhancing overall antitumor efficacy [468]. However, the small sample sizes and limited clinical data highlight the need for further studies to establish the safety and efficacy of these treatments across larger cohorts.

CAR-T cell therapy has been United States Food and Drug Administration-approved for certain forms of lymphoma and leukemia, demonstrating significant success in hematological malignancies. However, its application in solid tumors remains challenging [469-472]. One of the main obstacles is the identification and selection of specific tumor antigens that are consistently expressed across the TME. Unlike tumor cells, CAFs are genetically stable components of the TME, making them an attractive target for CAR-T cell therapy. Several preclinical studies have demonstrated the feasibility of CAR-T cell therapy targeting FAP, but the results of clinical trials are still pending. Addressing the considerable heterogeneity in solid tumors poses a significant challenge [473, 474]. A phase I clinical trial of CAR-T cell therapy targeting FAP was conducted in patients with malignant pleural mesothelioma. The results indicated promising safety, with two-thirds of the patients surviving the treatment [475]. Two additional studies are underway, and their results are worth monitoring (NCT03932565, NCT03198546). However, the path to effectively implementing CAR-T cell therapy targeting FAP in broader clinical applications remains challenging and requires further development and research.

Tumor vaccines represent an emerging therapeutic approach in oncology, designed to activate antigen-presenting cells and T lymphocytes, which play a crucial role in initiating the cancer immune cycle. By stimulating the immune system, tumor vaccines are considered one of the most promising therapies for enhancing antitumor immunity [476-479]. Currently, both DNA vaccines and DC vaccines targeting FAP are in the preclinical research stage. In animal models, these vaccines have demonstrated strong CD8⁺ T cell activation, high targeting specificity, and promising tumor-suppressive effects [480-482]. Due to the lack of clinical trials, the efficacy and safety of FAP-targeted vaccines cannot yet be confirmed. Further validation through human studies is essential to evaluate their clinical potential.

5.2.2 α-Smooth Muscle Actin

α-SMA, another marker of CAFs, is considered a potential biomarker in HER2-positive breast cancer [483]. However, research targeting α-SMA remains in the preclinical stage, with contradictory results observed in mouse models. On one hand, in a mouse model of breast cancer, docetaxel-conjugated nanoparticles targeting α-SMA-expressing stromal cells effectively inhibited metastasis, demonstrating potential therapeutic efficacy [484]. On the other hand, the targeted elimination of FAP+α-SMA+ CAFs leads to enhanced tumor stemness and increased infiltration of Treg cells, accelerating tumor progression [435]. It is important to note that α-SMA, as a marker of smooth muscle cells, is widely expressed in normal tissues such as blood vessels and the digestive tract [485-487]. This broad expression raises significant toxicity concerns in in vivo experiments. Although current organ-specific delivery strategies may offer potential improvements [488-491], the efficacy and safety of α-SMA-targeted therapies in humans remain unsatisfactory. Overall, the therapeutic application of α-SMA targeting across various cancer types is viewed with skepticism.