Cancer-Associated Fibroblast-Targeted Nanomedicine in Solid Tumor Therapy: From Mechanisms of Therapeutic Resistance to Precision Stromal Modulation

Abstract

Solid tumors remain difficult to treat because their therapeutic resistance is shaped not only by malignant-cell heterogeneity but also by stromal barriers that prevent therapeutic agents and immune cells from effectively reaching and eliminating tumor cells. Among these stromal components, cancer-associated fibroblasts (CAFs) are among the most abundant and functionally influential cell populations in the tumor microenvironment. Through extracellular matrix remodeling, paracrine signaling, immune regulation, and metabolic crosstalk, CAFs profoundly influence therapeutic response and prognosis in solid tumors. As central stromal regulators, CAFs determine whether antitumor therapies can achieve effective tumor inhibition. This central regulatory role makes CAFs rational and increasingly important therapeutic targets in solid tumors. However, CAF-targeted therapy is complicated by the pronounced heterogeneity and plasticity of CAF populations. Distinct CAF subsets may exert divergent, or even opposing, effects on tumor progression and therapeutic response. Therefore, effective CAF-targeted therapy should move beyond nonspecific CAF depletion and instead focus on precise stromal modulation. Nanomedicine provides a powerful strategy to strengthen CAF-directed therapy because nanomaterials can be engineered to match the biological and spatial features of CAF-rich tumor stroma. By tuning size, charge, shape, porosity, surface ligands, biomimetic coatings, and stimulus-responsive release, nanocarriers can enhance stromal accumulation, bind CAF-associated targets, and convert CAF biology into actionable therapeutic selectivity. In this review, we summarize the biological origins, activation mechanisms, subtype heterogeneity, dual functions, and biomarker landscape of CAFs in solid tumors, and systematically discuss how CAFs drive therapeutic resistance through physical, biochemical, metabolic, and immune barriers. We then highlight current CAF-targeted nanomedicine strategies, including CAF-selective delivery, extracellular matrix remodeling, CAF reprogramming, immune microenvironment regulation, and combination with chemotherapy, radiotherapy, photothermal/photodynamic therapy, immune checkpoint blockade, and adoptive cell therapy. Finally, we discuss key translational challenges, including target specificity, deep stromal penetration, long-term safety, biomarker-guided patient stratification, scalable manufacturing, and AI-assisted nanocarrier optimization. Overall, CAF-targeted nanomedicine offers a promising route to transform the tumor stroma from a barrier to therapy into a modifiable therapeutic interface, thereby improving the precision and efficacy of solid tumor treatment.

Introduction

Therapeutic failure in solid tumors cannot be attributed solely to the genetic heterogeneity of malignant cells; stromal cells within the tumor microenvironment also profoundly shape drug delivery, immune responses, and treatment tolerance.^1,2 Cancer-associated fibroblasts (CAFs) represent one of the most abundant stromal cell populations in solid tumors and are extensively involved in extracellular matrix deposition, elevated interstitial pressure, vascular abnormalities, immune-cell exclusion, inflammatory cytokine secretion, and metabolic reprogramming.^3,4 In solid tumors such as pancreatic, breast, liver, lung, colorectal, and gastric cancers, CAF enrichment is frequently associated with tumor tissue stiffening, insufficient deep drug penetration, impaired T-cell function, and poor responses to chemotherapy, targeted therapy, and immunotherapy.^3,5–8 Thus, CAFs are not only key drivers of malignant progression in solid tumors but also critical intervention targets for improving nanomedicine delivery efficiency and enhancing the efficacy of combination therapies.

However, a major challenge in CAF-targeted therapy lies in the fact that CAFs do not constitute a functionally homogeneous cell population.^9–11 For a long time, CAFs were broadly regarded as tumor-promoting stromal cells, and pan-CAF markers such as FAP, α-SMA, and PDGFR were commonly used for their identification and targeting. In recent years, however, single-cell sequencing, spatial omics, and multi-omics studies have revealed pronounced heterogeneity among CAFs in terms of cellular origin, activation signals, spatial distribution, and functional consequences.^12,13 Certain CAF subsets can promote drug resistance and immune evasion through mechanisms including collagen deposition, the CXCL12/CXCR4 axis, IL-6/STAT3 signaling, the TGF-β pathway, and CAF–TAM crosstalk;^14–17 in contrast, other CAF subsets may help maintain tissue architecture, restrict tumor dissemination, or participate in antitumor immunity.^18,19 This gives rise to a critical knowledge gap: although current studies have substantially elucidated the diverse mechanisms by which CAFs contribute to tumor progression and therapeutic resistance, clear functional criteria remain lacking for determining which CAF subsets truly drive drug resistance and immune escape across different cancer types and under distinct therapeutic pressures, and which CAF states may preserve tissue restraint or support antitumor immunity. Because such functional stratification has not yet been effectively translated into target selection, delivery strategies, and rational combination-therapy design, existing CAF-targeted nanostrategies still face considerable limitations in selectivity, safety, and clinical translatability. Moreover, systematic reviews centered on these unresolved issues remain insufficient. Most existing CAF-targeted nanosystems have focused primarily on enhancing tumor accumulation or improving stromal penetration, while paying inadequate attention to CAF functional diversity, spatial constraints, and patient stratification, thereby limiting their progression from preclinical investigation to clinical application.²⁰

In light of these challenges, this review focuses on the dual significance of CAFs in therapeutic resistance and nanomedicine-based intervention in solid tumors, emphasizing that CAF-targeted strategies should shift from simple depletion toward precise modulation. We first summarize the origins, activation mechanisms, major subtypes, functional duality, and key biomarkers of CAFs, thereby providing a biological framework for understanding CAF-targeted nanotherapy. We then systematically discuss how CAFs compromise therapeutic efficacy in solid tumors through physical barriers, biochemical paracrine networks, immunosuppressive ecosystems, and metabolic interactions. Furthermore, we summarize current CAF-targeted nanomedicine strategies, including CAF-selective delivery, ECM remodeling, CAF reprogramming, reversal of immunosuppression, and their integration with chemotherapy, radiotherapy, photothermal/photodynamic therapy, and immune checkpoint blockade. Unlike reviews that examine CAF biology or nanodelivery technologies in isolation, this review seeks to integrate CAF functional diversity with the design logic of nanotherapeutic strategies, clarify unmet needs in the field, and propose that future CAF-targeted nanomedicine should place greater emphasis on subtype selectivity, locally controllable intervention, safety evaluation, and patient stratification. Such efforts may advance stromal-modulation strategies for solid tumors toward greater precision and clinical translatability.

Biological Basis of CAFs in Solid Tumors

CAFs are among the most important stromal cell populations within the microenvironment of solid tumors.²¹ In contrast to the traditional view that CAFs are merely “tumor-promoting stromal cells”, recent advances in single-cell sequencing, spatial transcriptomics, and multi-omics analyses have demonstrated that CAFs exhibit pronounced heterogeneity and plasticity in their cellular origins, phenotypes, spatial distribution, and functional states.²² CAFs can promote tumor progression through extracellular matrix remodeling, inflammatory factor secretion, immunosuppression, and metabolic crosstalk; however, in specific subtypes, tumor stages, or spatial niches, they may also restrict tumor dissemination, preserve tissue architecture, and support antitumor immunity.^23,24 Therefore, CAFs should be viewed as tumor microenvironment-regulating cells with diverse cellular states and functional outcomes, rather than as a uniform population intended solely for therapeutic depletion. Accordingly, elucidating the origins, activation mechanisms, subtype composition, dual functions, and biomarkers of CAFs is essential for understanding the design principles of CAF-targeted nanomedicine strategies. These CAF-derived cytokines, chemokines, growth factors, extracellular-matrix components, and matrix-remodeling enzymes constitute major stromal determinants of therapeutic resistance and provide multiple potential intervention points for CAF-targeted therapy (Table 1).

Table 1.

CAF-Derived Cytokines, Chemokines, Growth Factors, ECM Components, and Enzymes Associated with Therapeutic Resistance and Representative Intervention Strategies in Solid Tumors

Factor Category	Factor Name	CAF Source/Matrix Background	Core Function	Targeting/Intervention Strategy	Ref
Cytokine	IL-6	IL-6-secreting CAFs; iCAFs	Activates the IL-6R/JAK/STAT3 signaling pathway, promoting EMT, invasion, metastatic potential, and therapeutic resistance	IL-6/IL-6R blockade; JAK/STAT3 pathway inhibition	[25–27]
Cytokine	LIF	LIF-producing activated stromal fibroblasts/CAFs	Mediates CAF-tumor paracrine signaling via the LIF/LIFR axis, supporting tumor cell invasion, survival, and therapeutic adaptation	LIF/LIFR blockade; anti-LIF antibodies; LIFR-targeted inhibition	[28,29]
Cytokine	TGF-β	myCAFs; TGF-β-driven LRRC15⁺ CAFs; LRRC15⁺ myofibroblasts	Drives fibroblast activation, ECM remodeling, immune-exclusive tumor architecture, and reduces response to immune checkpoint blockade	TGF-β pathway blockade; TGF-β/PD-L1 dual targeting; LRRC15-directed CAF-targeted therapy	[30–32]
Chemokine	CXCL12/SDF-1	FAP⁺ CAFs; CXCL12⁺ CAFs; iCAFs	Establishes CXCL12-CXCR4 paracrine signaling, supporting tumor cell proliferation, migration, immunosuppression, and stromal remodeling	CXCL12/CXCR4 axis blockade; CXCR4 antagonists; CXCL12-targeted siRNA nanoparticles	[33–35]
Growth Factor	HGF	HGF-secreting CAFs; stromal fibroblasts	Activates MET/c-MET signaling, maintaining tumor cell survival, EMT, invasion, and resistance to targeted therapy or chemotherapy	HGF/MET pathway inhibition; MET inhibitors; combined RTK/MET blockade	[36–38]
Growth Factor	IGF-1/IGF Axis	IGF-1-expressing CAFs; CAFs regulating IGF/IGFBP	Promotes tumor cell invasion, survival, and modulation of therapeutic response via IGF axis-dependent stroma-tumor crosstalk	IGF ligand blockade; IGF1R pathway inhibition; chemotherapy combined with IGF axis-targeted therapy	[39–41]
ECM Component	Type I Collagen	Matrix-producing CAFs; ECM-remodeling CAFs; myCAFs	Promotes desmoplastic stromal architecture, tissue stiffness, vascular compression, and restricts intratumoral distribution of drugs and immune cells	Angiotensin pathway inhibition; collagen remodeling; stromal normalization strategies	[42–44]
ECM Component	Hyaluronic Acid (HA)	HA-rich desmoplastic stroma; matrix-producing CAFs; activated stromal fibroblasts	Forms a hydrated stromal barrier, impairing vascular function, restricting perfusion, and affecting intratumoral distribution of chemotherapeutics and nanodrugs	PEGPH20/PEGylated hyaluronidase; hyaluronidase-mediated stromal remodeling; HA-targeted delivery enhancement	[45–47]
ECM Component	Fibronectin/EDB-FN	ECM-remodeling CAFs; FN1⁺ matrix-producing CAFs; myCAFs	Organizes tumor ECM, promotes CAF-cancer cell interactions, supports directional tumor cell migration; EDB-FN provides a targetable stromal ECM site	EDB-FN-targeted nanoparticles; EDB-specific peptide/antibody-mediated delivery; fibronectin-targeted imaging or therapy	[48–50]
Enzyme	LOX/LOXL2	LOX⁺ CAFs; ECM crosslinking CAFs; activated fibroblasts/myofibroblasts	Catalyzes collagen crosslinking, promoting ECM stiffening and supporting invasive stromal remodeling in the tumor matrix	LOX/LOXL2 inhibition; pan-lysyl oxidase inhibitors; collagen crosslinking blockade	[43,51,52]

Abbreviations: CAFs, cancer-associated fibroblasts; c-MET/MET, MET proto-oncogene receptor tyrosine kinase; CXCL12, C-X-C motif chemokine ligand 12; CXCR4, C-X-C chemokine receptor 4; ECM, extracellular matrix; EDB-FN, extra-domain B fibronectin; EMT, epithelial-mesenchymal transition; FAP, fibroblast activation protein; HA, hyaluronic acid; HGF, hepatocyte growth factor; iCAFs, inflammatory CAFs; IGF, insulin-like growth factor; IGFBP, IGF-binding protein; IL, interleukin; IL-6R, interleukin-6 receptor; JAK, Janus kinase; LIF, leukemia inhibitory factor; LIFR, leukemia inhibitory factor receptor; LOX, lysyl oxidase; LOXL2, lysyl oxidase-like 2; LRRC15, leucine-rich repeat-containing 15; myCAFs, myofibroblastic CAFs; PD-L1, programmed death-ligand 1; PEGPH20, pegylated recombinant human hyaluronidase PH20; PMID, PubMed identifier; RTK, receptor tyrosine kinase; SDF-1, stromal cell-derived factor 1; siRNA, small interfering RNA; STAT3, signal transducer and activator of transcription 3; TGF-β, transforming growth factor beta.

Origins and Activation of CAFs

CAFs do not originate from a single cellular lineage; instead, they represent a heterogeneous cell population generated through the activation or transdifferentiation of multiple precursor cell types under persistent stimulation within the tumor microenvironment.⁵³ Tissue-resident fibroblasts are generally considered the predominant and most direct source of CAFs. In different organs, organ-specific stromal cells, such as pancreatic stellate cells and hepatic stellate cells, can also contribute to CAF formation in pancreatic cancer and liver cancer, respectively.²³ In addition, bone marrow-derived cells, including bone marrow mesenchymal stem cells and monocytes/macrophages, can migrate into the tumor microenvironment and differentiate into CAFs.⁵⁴ Epithelial cells may acquire fibroblast-like phenotypes through epithelial–mesenchymal transition (EMT), whereas endothelial cells may undergo endothelial–mesenchymal transition (EndMT) to obtain similar mesenchymal features.⁵⁵ Pericytes, adipocytes, mesothelial cells, and smooth muscle cells may also be converted into CAFs in specific tumor contexts. This diversity of cellular origins makes it difficult to define CAFs using a single marker and provides the biological basis for their subsequent subtype heterogeneity and functional divergence.²³

CAF activation is jointly driven by tumor cells, immune cells, the extracellular matrix, and the metabolic microenvironment.⁵⁶ Among these factors, the TGF-β/SMAD pathway serves as a central signaling axis that induces CAF activation and promotes the formation of the myCAF phenotype, accompanied by the upregulation of molecules such as α-SMA, collagen, and fibronectin.⁵⁷ PDGF/PDGFR signaling primarily promotes fibroblast proliferation, migration, and matrix production.⁵⁸ In contrast, inflammatory factors such as IL-1, IL-6, and TNF-α can induce inflammatory CAF phenotypes through pathways including NF-κB, JAK/STAT, and STAT3, thereby enhancing the cytokine-secreting capacity and immunomodulatory functions of CAFs.⁵⁹ Beyond soluble factors, hypoxia, oxidative stress, mechanical tension, and metabolic stress also contribute to CAF activation and sustain their activated state through signaling networks such as YAP/TAZ, HIF-1α, Hippo, Hedgehog, Notch, and Wnt.⁵⁷ Therapeutic interventions can further reshape CAF reprogramming patterns. For example, in primary liver cancer, transarterial chemoembolization (TACE) can promote the conversion of hepatic stellate cells into myCAFs by inducing hypoxia and oxidative stress, thereby increasing ECM synthesis and enhancing therapeutic tolerance.⁶⁰

Thus, CAF activation is not a one-time, static cellular transformation, but rather a dynamic reprogramming process shaped by tumor progression, therapeutic intervention, and microenvironmental stress. CAFs can undergo phenotypic switching in response to distinct signaling cues. This plasticity makes CAFs important therapeutic targets in solid tumors, while also requiring careful consideration in precision nanomedicine delivery and functional modulation.

CAF Subtypes

The marked heterogeneity of CAFs is central to understanding their biological functions and therapeutic relevance. With the rapid development of single-cell sequencing and spatial omics technologies, CAF classification has evolved from early, relatively coarse identification based on a limited set of markers, such as α-SMA and FAP, toward multidimensional stratification based on transcriptional states, spatial localization, functional properties, and microenvironmental interactions.^22,61 Current evidence generally suggests that CAFs do not conform to an absolutely unified classification system; instead, CAF subtype composition may vary across tumor types, disease stages, and even distinct spatial regions within the same tumor.²³ Accordingly, CAF subtype classification should be understood as a spectrum of recurrent functional states with distinct molecular markers, spatial distributions, and biological consequences (Figure 1).

Figure 1.

Major CAF functional states and subtype-associated molecular markers. The schematic illustrates recurrent core CAF states, including myofibroblastic CAFs, inflammatory CAFs, and antigen-presenting CAFs, together with additional specialized CAF subsets linked to vascular regulation, interferon response, metabolic remodeling, proliferation, reticular stromal programs, tumor-associated stromal organization, and epithelial–mesenchymal-transition-like features. Representative molecular markers and functional programs are summarized to highlight the heterogeneity and plasticity of CAF populations across solid tumors. Adapted from.⁶² Copyright © 2024 Kazakova, Lukina, Anufrieva, Bekbaeva, Ivanova, Shnaider, Slonov, Arapidi and Shender.

The most widely recognized classical CAF subtypes include myofibroblastic CAFs (myCAFs), inflammatory CAFs (iCAFs), and antigen-presenting CAFs (apCAFs). myCAFs highly express molecules such as α-SMA, ACTA2, FAP, LRRC15, and TAGLN, are spatially located in close proximity to tumor cells, are induced by TGF-β signaling, and are characterized by ECM deposition, collagen synthesis, and stromal contraction.^63,64 myCAFs promote tumor invasion and therapeutic resistance by increasing stromal stiffness, creating invasive tracks, and activating signaling programs such as YAP/TAZ.⁶⁵ By contrast, iCAFs exhibit low α-SMA expression but high expression of inflammatory mediators, including IL-6, IL-8, CXCL12, CCL2, PDGFRα, and NLRP3, and are often located away from tumor nests, around blood vessels, or within hypoxic regions.^66,67 Through the recruitment of Tregs, MDSCs, and M2 macrophages, the suppression of CD8⁺ T cells, and the activation of pathways such as IL-6/STAT3 and CXCL12/CXCR4, iCAFs promote immunosuppression, tumor stemness, metastasis, and drug resistance.⁶⁷ apCAFs express MHC II, CD74, and HLA molecules but lack co-stimulatory molecules; they can induce Treg formation in pancreatic cancer, whereas in lung cancer, gastric cancer, and other tumor types, they may promote CD4⁺ T-cell activation and antitumor immunity.^68,69 Thus, the biological functions of apCAFs may be highly dependent on tumor type and spatial niche. It should be emphasized that myCAFs, iCAFs, and apCAFs are more appropriately understood as recurrent “core functional states” across cancer types rather than fixed cell types that are identical across all malignancies. Different solid tumors may develop cancer type-specific subtype combinations on the basis of these core states. For example, in primary liver cancer, in addition to myCAFs, iCAFs, and apCAFs, mCAFs, vCAFs, and lpCAFs can also be identified, which are respectively associated with stromal barrier formation, angiogenesis, and lipid metabolism;⁷⁰ in prostate cancer CAFs, metCAFs and nCAFs have been identified, suggesting that metabolic adaptation and neural-like adhesion/Hedgehog signaling may participate in the stromal ecology of this tumor type;⁷¹ in pancreatic cancer, pCAFs, sCAFs, and meCAFs are frequently highlighted as CAF states associated with proliferation, senescence, and metabolic reprogramming.⁷²

Beyond the three classical CAF subtypes, spatial transcriptomic studies have further demonstrated that CAF function is closely linked to spatial localization.⁶¹ Barrier-type or peritumoral CAFs are usually located at the tumor margin and may highly express molecules such as COL1A1, ACTA2, CXCL8, and TGFB1, thereby forming a dense extracellular matrix and an immune-exclusion barrier that restricts the entry of drugs and effector T cells into the tumor parenchyma.⁷³ Stromal or ECM-associated CAFs are mainly distributed deep within the tumor stroma and often express FAP, PDPN, MMP2, POSTN, collagen, and fibronectin; these CAFs primarily contribute to matrix remodeling, tissue stiffening, invasive-track formation, and impaired drug penetration.⁷⁴ Myeloid-enriched CAFs are frequently spatially colocalized with immunosuppressive cells such as M2 macrophages, neutrophils, and MDSCs, and reinforce the immunosuppressive microenvironment through CCL, CXCL, TGF-β, and related signaling pathways.⁷⁵ In contrast to these tumor-promoting spatial subtypes, TLS-associated CAFs are often located near tertiary lymphoid structures and may express molecules such as CD74, CCL19, and CCL21, thereby promoting T-cell and B-cell aggregation and correlating with improved responses to immunotherapy.⁷⁶ However, such spatial heterogeneity also exhibits cancer type-specific patterns. In primary liver cancer, myCAFs and mCAFs are mainly located at the tumor boundary, where they can establish physical and immune barriers, whereas vCAFs and lpCAFs are more likely to be enriched in the tumor core, where they participate in angiogenesis and metabolic crosstalk, respectively;⁷⁰ in breast cancer, different CAF-S1 subclusters may be distributed in peritumoral connective tissue, the tumor bed, intratumoral stromal nests, or adjacent normal lobules, forming distinct spatial neighborhoods with specific immune-cell populations;⁷⁷ in PDAC, dense stroma and immunosuppressive cells together constitute a highly exclusionary spatial ecosystem, making CAF-mediated ECM barriers and immune exclusion particularly prominent.⁷² Therefore, spatial targeting of CAFs should not be generalized solely on the basis of broad anatomical regions, but should instead be designed according to cancer type-specific tissue architecture and immune spatial maps. In addition to ligand density, CAF-targeted delivery is also affected by tumor heterogeneity.⁷⁸ Different tumors, and even different areas within the same tumor, may have different levels of stromal target expression, blood-vessel access, ECM density, and ECM pore size.^79–81 As a result, stromal targets such as FAP, PDGFRβ, fibronectin EDB, and tenascin-C may not be evenly distributed throughout the tumor, leading to strong nanoparticle binding in some regions but weak binding and poor penetration in others.⁸² Therefore, CAF-targeted nanocarriers should be designed with consideration of not only ligand–receptor binding, but also stromal structure, target distribution, and penetration barriers.

Meanwhile, an increasing number of CAF subtypes with specialized functions have been identified. For example, LRRC15⁺ CAFs often display TGF-β-dependent myCAF features and are associated with immune exclusion and resistance to immunotherapy;³² CD10⁺GPR77⁺ CAFs can sustain tumor stem-cell properties and mediate chemoresistance;⁸³ senCAFs express senescence-associated markers such as p16 and TSPAN8 and can promote drug resistance and immunosuppression through SASP factors;⁸⁴ glyCAFs or metabolic CAFs support tumor-cell growth and suppress immune responses through metabolites such as lactate, lipids, and amino acids.⁸⁵ Conversely, Meflin⁺ CAFs, rCAFs, certain CD105⁻ CAFs, ifCAFs, and TLS-associated CAFs may, in specific contexts, be associated with tissue homeostasis maintenance, enhanced antitumor immunity, or favorable prognosis.^61,86–89

Taken together, CAF subtypes do not merely represent a morphological classification, but instead constitute a functional stratification directly linked to tumor progression, immune status, therapeutic resistance, and barriers to drug delivery. This concept has important implications for CAF-targeted nanomedicine: nanocarriers should not take “CAF depletion” as their sole objective, but should, whenever possible, selectively recognize tumor-promoting CAF subtypes, locally block key pro-tumorigenic signals, or reprogram pathological CAFs into quiescent, tumor-restraining, or immune-supportive phenotypes. In other words, CAF subtype diversity and functional heterogeneity provide the biological rationale for selective CAF targeting rather than broad CAF depletion.

Dual Roles of CAFs

CAFs exert pronounced and context-dependent dual roles in solid tumors.⁹⁰ In most settings, CAFs promote tumor progression through stromal remodeling, immunosuppression, metabolic crosstalk, and paracrine signaling; however, under specific tumor stages, tissue contexts, or subtype states, CAFs may also restrain tumor dissemination, preserve tissue architecture, and facilitate antitumor immunity.⁶¹

The tumor-promoting functions of CAFs are reflected in multiple dimensions: by secreting factors such as IL-6, CXCL12, TGF-β, and LIF, CAFs activate signaling pathways including STAT3 and PI3K/AKT, thereby driving tumor proliferation, EMT, invasion, and metastasis;^91,92 by secreting collagen, LOX, MMPs, POSTN, and other matrix-associated molecules, CAFs remodel the ECM, resulting in stromal stiffening and the formation of physical barriers that hinder drug penetration and immune infiltration;⁹³ through TGF-β, IL-6, CXCL12, Galectin-9, PD-L1, FASL, and related mediators, CAFs induce T-cell exhaustion, apoptosis, and exclusion, while recruiting Tregs, MDSCs, and M2 macrophages to establish an immunosuppressive microenvironment;^94,95 additionally, CAFs can mediate resistance to chemotherapy, targeted therapy, endocrine therapy, and immunotherapy through exosomes, paracrine factors, and metabolite transfer.⁹⁶ By promoting tumor progression, angiogenesis, immune escape, and therapeutic resistance, CAFs compromise the efficacy of solid tumor therapies and therefore represent important targets for nanodelivery systems. However, an exclusive focus on the tumor-promoting roles of CAFs risks oversimplifying therapeutic design. In fact, certain CAF subtypes possess non-negligible tumor-restraining functions. First, some α-SMA⁺ myCAFs and the collagen matrix they secrete can form a mechanical barrier that limits the dissemination of tumor cells into surrounding tissues.⁹⁷ Second, some CAFs may help preserve tumor tissue architecture and maintain a relatively differentiated tumor state;⁸⁶ complete depletion of these CAFs may therefore increase tumor invasiveness or exacerbate immunosuppression. Third, CAF subsets such as apCAFs, TLS-associated CAFs, ifCAFs, Meflin⁺ CAFs, and rCAFs can, in specific contexts, promote antigen presentation, T/B-cell recruitment, or effector T-cell activation, and are associated with more favorable antitumor immune responses and therapeutic benefit.^61,86–89 In addition, PRELP⁺ CAFs, CD105⁻ CAFs, CD90⁺WISP-1-associated CAFs, and other subsets have also been proposed to suppress angiogenesis, tumor growth, or metastasis under certain conditions.^98,99

This duality is also dependent on cancer type. The same functional CAF state may produce distinct, or even opposing, consequences in different solid tumors. For example, in pancreatic cancer, apCAFs may promote Treg differentiation through MHC II-associated antigen-presentation-like activity and thereby contribute to immunosuppression; however, in certain immune-infiltrated or TLS-associated contexts, apCAFs/TLS-associated CAFs may instead facilitate T/B-cell recruitment and antitumor immunity.⁷² Similarly, myCAF-mediated collagen deposition may constitute a barrier to the entry of drugs and immune cells in stroma-rich tumors such as PDAC, but may also form a structural constraint that limits tumor-cell dissemination at certain stages.⁷² Therefore, whether CAFs exert tumor-promoting or tumor-suppressive effects cannot be determined independently of cancer type, disease stage, and spatial niche.

This duality has direct implications for CAF-targeted therapy. Previous CAF-targeting strategies have often relied on pan-CAF markers such as FAP and α-SMA to achieve broad depletion; however, such approaches may eliminate both tumor-promoting and tumor-restraining CAFs, thereby disrupting structural constraints within tumors, destabilizing the immune ecosystem, and even causing accelerated tumor progression or inconsistent therapeutic outcomes.¹⁰⁰ Conventional CAF markers, including α-SMA, FAP, S100A4, PDGFRα/β, and vimentin, have been widely used for CAF identification, but they are not absolutely CAF-specific, further increasing the risks associated with broad-spectrum targeting.¹⁰⁰ Therefore, rational CAF-targeted strategies should shift from “CAF elimination” toward “precise CAF modulation”.

CAF Biomarkers

CAF biomarkers provide the molecular foundation for CAF identification, patient stratification, and targeted nanocarrier design. However, a central challenge in CAF biomarker research is that no single marker is currently available that is absolutely CAF-specific, encompasses all CAF populations, and reliably distinguishes tumor-promoting from tumor-restraining functions. Therefore, CAF biomarkers should be interpreted as relatively specific molecular panels rather than isolated target labels. These biomarker-defined CAF populations are associated with diverse functional outputs, including tumor promotion, stromal remodeling, immune regulation, therapeutic resistance, and, in some contexts, tumor restraint (Figure 2).

Figure 2.

CAF biomarker-defined functions and therapeutic consequences in cancer. (A) Schematic representation of CAF heterogeneity based on biomarker-defined functional states. CAFs are broadly categorized as cancer-promoting, bidirectional, or cancer-restraining populations according to representative markers and their effects on tumor progression. These CAF subsets regulate tumor biology through cytokine secretion, exosome-mediated communication, ECM remodeling, metabolic reprogramming, pre-metastatic niche formation, and immunosuppressive signaling, thereby contributing to tumor proliferation, metastasis, invasion, cancer stem-cell phenotypes, epithelial–mesenchymal transition (EMT), drug resistance, and radiotherapy resistance. (B) Representative cancer-restraining CAF markers, including Meflin and CD146, are associated with suppressed tumor growth and may modulate therapeutic responses, such as immune-checkpoint blockade (ICB) and endocrine therapy with tamoxifen. (C) CAF-associated markers such as CAV-1, POSTN, and CD200 are implicated in treatment resistance, including resistance to chemotherapy or targeted therapy and radiotherapy. Adapted from. ¹⁰¹ © The Author(s) 2023.

Common pan-CAF markers include FAP, α-SMA/ACTA2, PDGFRα, PDGFRβ, FSP1/S100A4, vimentin, PDPN, CD90, CD34, CAV1, POSTN, fibronectin, and type I collagen.¹⁰¹ These markers can be used to identify activated fibroblasts or CAF-enriched regions and have also served as the targeting basis for many CAF-directed nanocarrier designs. For example, FAP is widely used in CAF-targeted delivery, diagnostic imaging, and immunotherapy-combination strategies because of its high expression in CAFs across multiple solid tumors;¹⁰² PDGFRβ, PDPN, CD90, and related markers are also commonly used for CAF identification and stromal targeting.¹⁰³ Nevertheless, these pan-CAF markers have substantial limitations: FAP can also be detected in certain contexts of normal tissue repair or fibrosis, α-SMA is expressed by smooth muscle cells and pericytes, vimentin is a broad mesenchymal-cell marker, and S100A4/FSP1 may also appear in immune cells or tumor cells.²³ Thus, a single pan-CAF marker is more suitable for preliminary CAF enrichment than for truly subtype-specific therapeutic targeting.

From the perspective of subtype specificity, myCAFs commonly express molecules such as α-SMA, ACTA2, COL1A1, MYL9, POSTN, MYH11, and SPP1. These markers are largely associated with extracellular matrix remodeling, collagen deposition, and stromal stiffening, and can be used to identify tumor regions characterized by pronounced fibrosis, restricted drug penetration, or evident immune exclusion.^104,105 iCAFs are mainly characterized by inflammatory and chemotactic mediators, including IL-6, IL-8, I, CCL2, CXCL1/2, PDGFRα, NLRP3, and SLC14A1.¹² These markers indicate that CAFs possess strong immunomodulatory and inflammation-amplifying functions, and they may provide an important basis for blocking immunosuppressive signals and improving T-cell infiltration in nano-immunotherapy. Typical markers of apCAFs include MHC II, CD74, HLA-DR, HLA-DP, and HLA-DQ. The function of this subtype should be interpreted in relation to cancer type and spatial niche, because apCAFs may either contribute to immunosuppression or participate in antitumor immune activation.¹⁰⁶

In addition, several function-associated markers are particularly important for precision targeting. LRRC15 often indicates a TGF-β-associated myCAF state and is linked to immune exclusion and poor responses to immunotherapy;¹⁷ CD10 and GPR77 mark CAF subsets associated with tumor stemness maintenance and chemoresistance;⁸³ Meflin/ISLR, Cav-1, CD146, and related markers can be used to identify tumor-restraining or homeostasis-maintaining CAFs;^86,107–109 CD105 can distinguish functionally distinct CAF populations in certain tumors; TSPAN8 and p16/CDKN2A are associated with senescence-related CAFs;¹¹⁰ and CCL19, CCL21, CXCL9, CD74, and HLA-DR may help identify TLS-associated CAFs or immune-supportive CAFs.¹¹¹ Because CAF populations defined by different markers may exert distinct, or even opposing, functions, CAF-targeted therapy should not rely on a single marker for identification. Instead, it should integrate multiple markers and evaluate the corresponding functional state.

In summary, the biological features of CAFs indicate that CAF-targeted nanomedicine should be designed on the basis of CAF heterogeneity, plasticity, and dual functionality. The multiple origins and diverse activation signals of CAFs explain their biological complexity; CAF subtype differentiation and spatial localization determine their functional divergence; the coexistence of tumor-promoting and tumor-restraining functions requires therapeutic strategies to avoid broad CAF depletion; and CAF biomarkers provide the molecular basis for targeted nanocarrier modification, stimulus-responsive release, combination therapy, and patient stratification. Therefore, future CAF-targeted nanotherapy should focus on selectively inhibiting tumor-promoting CAFs, preserving or inducing tumor-restraining CAFs, and achieving precise remodeling of the stromal ecosystem in solid tumors through locally controlled and multitargeted nanodelivery systems.

Mechanisms by Which CAFs Drive Therapeutic Resistance and Impaired Nanomedicine Delivery in Solid Tumors

On the basis of the biological features summarized above, CAFs contribute to therapeutic failure through several interconnected barriers. Dense extracellular matrix restricts drug and nanoparticle penetration; paracrine networks reshape tumor-cell survival and treatment tolerance; and immunosuppressive niches limit effective immune activation. This chapter organizes CAF-mediated resistance into physical, biochemical, and immune barriers, thereby linking CAF biology to the design rationale for CAF-targeted nanotherapeutic strategies.

Physical Barriers: ECM, Interstitial Pressure, and Impaired Nanomedicine Penetration

In solid tumors, CAFs can establish characteristic physical barriers through remodeling of the extracellular matrix (ECM). Activated CAFs are major sources of ECM components, including collagen, fibronectin, and hyaluronic acid.¹¹² Particularly in stroma-rich tumors such as pancreatic cancer and breast cancer, persistent matrix production and reorganization render tumor tissues dense and stiff. Both the abundance and cross-linking status of the ECM directly determine tissue stiffness, while increased collagen deposition and abnormal collagen alignment can narrow extracellular spaces and lengthen the diffusion path of drugs within the tumor stroma.^113,114 This barrier manifests not only as a structural impediment, but also as abnormal mechanical stress. In CAF-associated ECM, collagen networks provide a tensile scaffold, whereas hyaluronic acid increases tissue hydration pressure through its water-absorbing and swelling properties; together, these components promote the accumulation of solid stress and compress tumor blood vessels.¹¹⁵ Chauhan et al found that hyaluronic acid-mediated vascular compression depends on a collagen-rich stromal background, indicating that different ECM components do not act independently, but instead cooperate to remodel the mechanical landscape of tumors, thereby reducing vascular perfusion and limiting drug entry.^116,117 As tumor vessels become compressed and lymphatic drainage is impaired, interstitial fluid pressure increases, transvascular convective transport is weakened, and drugs are more likely to remain confined to perivascular regions rather than reaching tumor cell populations located far from blood vessels.¹¹⁸

This barrier is particularly important for nanomedicines. Although nanoparticles can enter tumor tissues through abnormal vascular permeability, their subsequent intratumoral diffusion is highly dependent on ECM pore size, fiber density, particle size, and surface properties.¹¹⁹ Dense collagen networks and hyaluronic acid-rich stroma can restrict particle movement, causing nanomedicines to exhibit pronounced perivascular accumulation rather than uniform penetration into the tumor parenchyma.¹²⁰ Importantly, passive nanoparticle accumulation driven by the enhanced permeability and retention (EPR) effect should not be equated with effective intratumoral delivery, particularly in highly desmoplastic tumors such as pancreatic ductal adenocarcinoma (PDAC). In PDAC, disorganized and compressed tumor vasculature, poor perfusion, elevated interstitial fluid pressure, and a collagen- and hyaluronic acid-rich extracellular matrix collectively weaken EPR-dependent extravasation and restrict subsequent interstitial diffusion. As a result, nanoparticles may accumulate predominantly in perivascular or stromal regions without reaching poorly perfused malignant cell nests, indicating that EPR-based tumor accumulation does not reliably predict therapeutic delivery efficiency in PDAC.^121–124 Studies using decellularized tissues and ECM models have also demonstrated that ECM-mediated restrictions on nanoparticle diffusion and cellular uptake are tissue-specific, suggesting that enhanced tumor accumulation alone does not necessarily translate into effective cellular delivery.¹²⁵ A representative example is the effect of losartan on the collagen barrier and nanomedicine distribution. Diop-Frimpong et al found that losartan inhibited collagen I production by breast cancer-derived CAFs and reduced stromal collagen levels in multiple stroma-rich tumor models; on this basis, the intratumoral distribution and therapeutic efficacy of pegylated liposomal doxorubicin, namely Doxil, were improved.¹²⁶ This finding highlights the CAF-mediated ECM barrier as a functional regulator of nanomedicine delivery efficiency, particularly in controlling nanoparticle transport beyond initial tumor access toward deep intratumoral penetration.

However, the CAF–ECM barrier should not be simply regarded as an entirely detrimental structure. Some studies have shown that excessive depletion of myofibroblast-like CAFs or fibrotic stroma can accelerate pancreatic cancer progression, enhance immunosuppression, and shorten survival.¹⁰⁰ Therefore, from a biological perspective, the physical barrier shaped by CAFs has a dual nature: on the one hand, it restricts the penetration of drugs and nanoparticles through ECM densification, vascular compression, and elevated interstitial pressure; on the other hand, certain stromal structures may also help restrain tumor dissemination or preserve tissue boundaries.^97,127 Accordingly, clarifying how distinct CAF subsets and their ECM products regulate local mechanics and drug transport is essential for understanding therapeutic resistance in solid tumors.

Biochemical Barriers: CAF-Secreted Factors Driving Proliferation, Invasion, and Drug Resistance

In addition to ECM-mediated physical obstruction, CAFs can establish a sustained biochemical stimulatory milieu around tumor cells by secreting cytokines, chemokines, and growth factors.⁵⁶ Whereas physical barriers primarily determine whether therapeutic agents can “reach” tumor cells, biochemical barriers more directly shape how tumor cells respond to treatment. Soluble factors derived from CAFs can activate signaling pathways such as STAT3, PI3K–AKT, MAPK, and MET, thereby enhancing tumor-cell proliferation, migration, anti-apoptotic capacity, and stemness maintenance, ultimately reducing sensitivity to chemotherapy, radiotherapy, and targeted therapy.^25,128–130

IL-6 and LIF are representative CAF-secreted mediators that promote therapeutic resistance. CAF-derived IL-6 can activate the JAK/STAT3 pathway in tumor cells, thereby driving EMT, migration, and anti-apoptotic phenotypes.⁹¹ In models of invasive lobular breast cancer, IL-6 secreted by patient-derived CAFs induces STAT3 phosphorylation, suppresses estrogen signaling, and enhances tumor-cell motility and dissemination; conversely, IL-6-blocking antibodies attenuate STAT3 activation induced by CAF-conditioned medium.^91,131 LIF, which shares the JAK/STAT3 signaling axis with IL-6, is more prominently involved in maintaining cancer stem-cell properties and chemoresistance.¹³² In pancreatic cancer, CAF-derived LIF promotes stemness features and gemcitabine resistance by activating the LIFR/STAT3 pathway, whereas targeting LIF signaling significantly suppresses tumor progression and reverses drug resistance.¹³³

CAF-secreted chemokines can further promote invasion and microenvironmental remodeling.^92,134 The CXCL12/CXCR4 axis not only directly enhances tumor-cell migration and invasion, but also amplifies the CAF state itself. Ma et al found that CAF-derived CXCL12 induces normal fibroblasts to acquire CAF-like features through the STAT3 pathway, including upregulation of α-SMA and vimentin, enhanced proliferation and migration, and promotion of lung cancer xenograft growth and spontaneous lung metastasis in vivo.¹³⁵ These findings suggest that CAF-secreted factors do not act solely on tumor cells, but can also expand tumor-promoting stromal niches through positive feedback among stromal cells.

Distinct from the mechanisms described above, which mainly remodel malignant phenotypes within the microenvironment, CAF-secreted growth factors can directly provide alternative survival cues to tumor cells, thereby inducing bypass activation-mediated resistance during targeted therapy. In targeted therapy, tumor cells frequently evade pharmacological inhibition by activating alternative signaling pathways, a mechanism known as bypass activation.¹³⁶ CAF-secreted growth factors constitute an important source of such bypass signaling, with HGF being one of the most representative examples. Under normal conditions, BRAF-mutant melanoma depends on the BRAF–MEK–ERK signaling axis for proliferation and is sensitive to BRAF inhibitors.¹³⁷ However, Straussman et al found that CAF-derived HGF binds to the MET receptor on tumor cells and reactivates downstream MAPK and PI3K–AKT pathways. As a result, even when BRAF is pharmacologically inhibited, tumor cells can bypass the blocked target and continue to survive, thereby acquiring intrinsic tolerance.¹³⁸ Similarly, in EGFR-mutant non-small cell lung cancer, CAF-derived HGF and IGF-1 can induce EMT through the MET and IGF-1R pathways and bypass EGFR inhibitor blockade, leading to acquired resistance.¹³⁹ Thus, CAF-secreted growth factors can provide alternative survival signals without requiring additional genetic alterations in tumor cells themselves, thereby weakening the efficacy of targeted therapies. This mechanism also helps explain why, even in tumors driven by strong oncogenic mutations, inhibition of the dominant driver alone may fail because of microenvironment-mediated bypass activation.

However, the effects of CAF-secreted factors are not uniformly tumor-promoting. Their ultimate consequences are determined by CAF subtype composition, tumor genetic background, and therapeutic context. For example, Remsing Rix et al found in lung cancer models that CAFs or CAF-conditioned medium induced drug resistance in some cell lines, but enhanced drug sensitivity in others; this discrepancy was associated with the balance between CAF-secreted IGFs and IGF-binding proteins.⁴¹ Therefore, the so-called “biochemical barrier” should not be understood as the effect of any single fixed factor, but rather as a dynamic paracrine network shaped jointly by CAF heterogeneity, tumor-cell receptor status, and therapeutic pressure. It is precisely this networked and context-dependent nature that makes CAF-secreted factors a central yet mechanistically complex component of therapeutic resistance in solid tumors.

Immune Barriers: CAF-Mediated T-Cell Exclusion and Immune-Desert Formation

The third major mechanism by which CAFs contribute to therapeutic resistance is the establishment of immune barriers through remodeling of the tumor immune microenvironment. At its core, this process disrupts key steps of the cancer-immunity cycle, including T-cell infiltration, activation, and cytotoxic function, thereby preventing immune cells from efficiently entering tumor nests or leaving them functionally suppressed after infiltration.¹⁴⁰ Clinically, solid tumors can be broadly categorized into three immune phenotypes: immune-inflamed, immune-excluded, and immune-desert. In immune-excluded tumors, CD8⁺ T cells are largely retained at the tumor margin or within stromal compartments, whereas immune-desert tumors show little or no evident T-cell infiltration. Both phenotypes are generally associated with poor responses to immune checkpoint blockade (ICB) therapy.¹⁴¹

Quantitative Benchmarks of the Immune-Excluded Phenotype and Its Impact on Immunotherapy

From a quantitative perspective, the immune-excluded phenotype is relatively common across multiple solid tumors, although its reported prevalence varies substantially according to cancer type, specimen source, and classification criteria. In a cohort of advanced non-small cell lung cancer, AI-based spatial analysis of tumor-infiltrating lymphocytes (AI-TILs) on H&E whole-slide images classified tumors into inflamed, immune-excluded, and immune-desert phenotypes, accounting for 44.0%, 37.1%, and 18.9% of cases, respectively.¹⁴² In the TCGA ovarian cancer cohort, the distribution among 489 patients was 30.7% inflamed, 36.4% excluded, and 32.9% desert.¹⁴³ In primary triple-negative breast cancer, the excluded phenotype accounted for approximately 23.8% of 101 samples; however, in metastatic TNBC, the proportion of excluded tumors increased to 41%, while desert and inflamed phenotypes accounted for 37% and 21%, respectively.¹⁴⁴ Pancreatic ductal adenocarcinoma exhibits an even more prominent immune-excluded pattern: an AI-based spatial analysis of 304 resectable PDAC cases showed that immune-inflamed, immune-excluded, and immune-desert phenotypes accounted for only 9.9%, 85.2%, and 4.9% of cases, respectively.¹⁴⁵ Collectively, these findings indicate that the immune-excluded phenotype is observed across multiple solid tumors, with reported proportions generally ranging from approximately 20% to 40%, and reaching more than 85% in stroma-rich tumors such as pancreatic cancer.

Clinical studies have further quantified the impact of stromal abundance on immunotherapy efficacy. In stroma-rich tumors represented by NSCLC, the objective response rate (ORR) to chemoimmunotherapy was 51.2% in the high-stroma group (≥50%), significantly lower than the 72.1% observed in the low-stroma group (P = 0.031); median overall survival was also nearly halved in the high-stroma group (12.2 vs 24.0 months, P = 0.001).¹⁴⁶ In other solid tumors, stromal enrichment or CAF–TGF-β activation is likewise associated with poor or absent responses to immune checkpoint inhibitors. In a pembrolizumab-treated cohort of metastatic urothelial carcinoma, the patient subgroup enriched for stromal cells/products showed no objective responses, with an ORR of 0%, whereas the ORRs in the T-cell expression (TSE)-positive, TSE-neutral, and TSE-negative groups were 67%, 21%, and 0%, respectively.¹⁴⁷ Another study of atezolizumab-treated metastatic urothelial carcinoma similarly showed that non-response was associated with fibroblast TGF-β signaling signatures and CD8⁺ T-cell exclusion within collagen-rich peritumoral stroma.¹⁴⁸ In addition, integrated analyses of immune and stromal scores in lung adenocarcinoma, melanoma, and head and neck squamous cell carcinoma demonstrated that tumors with a low-immune/high-stroma profile had the lowest response rates across multiple immunotherapy cohorts.¹⁴⁹

CAF-Mediated T-Cell Exclusion and Dysfunction

The impact of CAFs on T-cell immunity is mainly reflected in two interconnected processes: the exclusion of T cells from tumor cell-enriched regions and the functional impairment of T cells that have already infiltrated tumor tissues. The former is characterized by the accumulation of T cells at the tumor margin or within stromal compartments, where they fail to establish effective contact with tumor cells. The latter is manifested by insufficient T-cell activation, restricted proliferation, reduced cytotoxic activity, and the progressive acquisition of an exhaustion-like phenotype. CAFs first reshape the spatial distribution of T cells by remodeling the extracellular matrix.¹⁵⁰ Specifically, CAFs promote the deposition of matrix components such as collagen and fibronectin, while increasing matrix density and tissue stiffness, thereby imposing physical barriers that hinder T-cell migration toward tumor nests.^151,152 In parallel, CAF-derived chemokines and matrix-associated signals can further alter T-cell positioning, retaining T cells within stromal regions or at the tumor periphery rather than allowing their penetration into tumor cell-enriched areas.¹⁵³

Even after T cells enter tumor tissues, CAFs can attenuate their effector functions through multiple mechanisms. CAF-derived immunosuppressive cytokines reduce T-cell activation and suppress the expression of cytotoxic molecules.¹⁵⁴ Immune checkpoint-related molecules expressed or induced by CAFs further dampen T-cell receptor signaling and cytotoxic function.⁹⁴ In addition, CAFs can remodel the local metabolic environment, exposing T cells to nutrient deprivation, adenosine accumulation, and dysregulated inflammatory signaling.¹⁵⁵ Among these mechanisms, TGF-β is particularly representative, as it not only promotes CAF activation and matrix remodeling but also directly contributes to T-cell exclusion and functional suppression.¹⁵⁶ Mariathasan et al demonstrated that fibroblast-associated TGF-β signaling in tumors is linked to T-cell exclusion and limited responsiveness to anti-PD-L1 therapy.¹⁴⁸ CAFs can also exacerbate T-cell dysfunction indirectly by modulating other immune cell populations. For example, they promote the accumulation of regulatory T cells, myeloid-derived suppressor cells, and immunosuppressive macrophages within the tumor microenvironment, thereby further suppressing the activation and cytotoxic capacity of CD8⁺ T cells.¹⁵⁷

Notably, distinct CAF subpopulations may play different roles in the formation of immune barriers, which may partly explain the heterogeneity of tumor immune phenotypes across cancer types. For instance, TGF-β-dependent LRRC15-positive CAFs have been reported to directly suppress CD8⁺ T-cell function and limit responses to immune checkpoint blockade.¹⁵⁶ In contrast, the role of antigen-presenting CAFs appears to be more context dependent. Some studies suggest that these CAFs may participate in T-cell regulation, whereas recent evidence indicates that, under specific conditions, they can enhance T-cell activation, proliferation, and cytotoxicity.^19,158 Therefore, CAF-mediated immune barriers should not be viewed simply as a consequence of increased overall CAF abundance. Rather, they should be interpreted in the context of CAF subpopulation composition, spatial organization, and tumor type. This perspective also suggests that future CAF-targeted strategies should move beyond indiscriminate CAF depletion and instead focus on selectively modulating specific CAF subpopulations and their barrier-forming functions.

Regulation of Macrophages, Dendritic Cells, and NK Cells by CAFs

In addition to directly restricting T-cell infiltration and effector function, CAFs can promote the establishment of a more comprehensive immune barrier within the tumor microenvironment by modulating macrophages, dendritic cells, and NK cells. These immune cell populations are respectively involved in inflammatory regulation, antigen presentation, and innate cytotoxicity, and therefore represent critical determinants of whether antitumor immune responses can be initiated and sustained. Thus, CAF-mediated immune barriers are not merely a consequence of spatial T-cell exclusion, but also reflect the coordinated suppression of immune priming, effector amplification, and tumor clearance.

At the level of the myeloid compartment, CAFs can promote monocyte recruitment and drive their differentiation into immunosuppressive tumor-associated macrophages (TAMs). CAF-derived factors, including M-CSF, CCL2, VEGF, and IL-6, have been implicated in macrophage recruitment and TAM formation.¹⁵⁹ Some studies further suggest that CAFs attract monocytes through IL-8 and enhance interactions among monocytes, tumor cells, and the stromal microenvironment through IL-6-dependent signaling.^160,161 These TAMs can subsequently secrete immunosuppressive mediators, promote stromal remodeling, and impair the effector functions of T cells and NK cells, thereby shifting the local tumor milieu from an antitumor inflammatory state toward a myeloid cell-dominated suppressive state.¹⁶¹ This shift is particularly relevant to immune checkpoint inhibitor therapy. Even after PD-1 or PD-L1 signaling is blocked, a parallel immunosuppressive program maintained by TAMs may persist within tumors. Consequently, T cells, although released from inhibitory checkpoint signaling, may still fail to fully restore cytotoxic activity in a microenvironment dominated by myeloid suppression.¹⁶²

CAFs can also impair the functions of dendritic cells (DCs) and NK cells, thereby disrupting both the initiation of antitumor immunity and the amplification of early effector responses. DCs are key antigen-presenting cells that bridge innate and adaptive immunity. CAFs can interfere with DC differentiation, maturation, and antigen-presenting capacity through cytokines, chemokines, and growth factors, and may even induce a tolerogenic DC phenotype.¹⁶³ CAFs can reduce the ability of DCs to present antigens and stimulate T cells, while simultaneously promoting an immunosuppressive microenvironment.¹⁶⁴ This directly compromises the biological foundation of immune checkpoint therapy, because anti-PD-1 treatment depends not only on the reinvigoration of pre-existing T cells, but also on sustained antigen presentation and pro-inflammatory signaling provided by DCs. When DC function is insufficient, therapy may enter a state in which inhibitory signals are relieved but effective immune priming remains inadequate.¹⁶⁵ Meanwhile, CAFs can suppress NK-cell activation and cytotoxicity through factors such as TGF-β, PGE2, and IDO, leading to reduced NK-cell proliferation, degranulation, and expression of activating receptors.¹⁶⁶ Impaired NK-cell function not only weakens early innate tumor killing, but also reduces tumor antigen release and support for DC function, thereby further limiting the transition of the tumor microenvironment toward an inflamed phenotype.

Overall, by modulating macrophages, dendritic cells, and NK cells, CAFs expand the immune barrier from T-cell exclusion into a multicellular suppressive network. Macrophage-driven myeloid suppression, DC dysfunction-mediated impairment of antigen presentation, and NK-cell inhibition-induced loss of innate cytotoxicity collectively diminish the immunological foundation required for immune checkpoint inhibitors to exert their effects. Therefore, the poor response of CAF-enriched tumors to immunotherapy cannot be attributed solely to impaired T-cell infiltration; rather, it also reflects broad restrictions on the initiation, amplification, and execution of the entire antitumor immune response.

Hypoxia Promotes the Immunosuppressive Phenotype of CAFs and Immune Desert Formation

The hypoxic microenvironment within solid tumors is one of the key drivers of tumor progression and immune evasion. As the most abundant stromal cell population in the tumor microenvironment, CAFs are highly sensitive to hypoxic signals. Hypoxia not only reshapes the metabolic state of CAFs but also induces them to acquire a strongly immunosuppressive phenotype, thereby promoting the formation of an “immune desert”, a state characterized by scarce effector T-cell infiltration and the absence of effective antitumor immune responses within tumor regions.¹⁶⁷ The hypoxic microenvironment directly regulates the functional reprogramming of CAFs through stabilization of hypoxia-inducible factor-1α (HIF-1α). Under hypoxic conditions, CAFs upregulate the secretion of multiple immunosuppressive cytokines and chemokines, including transforming growth factor-β (TGF-β), interleukin-10 (IL-10), CCL2, CCL22, and CXCL12.¹⁶⁸ Together, these factors recruit suppressive immune cell populations, such as regulatory T cells and tumor-associated macrophages, while directly inhibiting the activation and infiltration of CD8⁺ cytotoxic T cells.¹⁴⁰ Studies have shown that CXCL12 secreted by hypoxic CAFs can establish a peritumoral barrier that impedes T-cell migration into the tumor parenchyma, representing one of the key molecular mechanisms underlying immune desert formation.¹⁶⁹

In addition, hypoxic CAFs further reinforce immunosuppression through metabolic reprogramming. Hypoxia induces aerobic glycolysis in CAFs, leading to substantial lactate accumulation. Lactate not only acidifies the microenvironment and suppresses T-cell function, but also upregulates the expression of immunosuppressive molecules such as PD-L1 and arginase-1 in CAFs through histone lactylation.¹⁷⁰ Hypoxic CAFs can also secrete exosomes enriched with immunosuppression-associated non-coding RNAs, thereby remotely regulating the differentiation and function of immune cells.¹⁷¹ Meanwhile, hypoxic CAFs promote excessive deposition and crosslinking of extracellular matrix components, such as collagen and fibronectin, forming a dense stromal network that further physically restricts T-cell infiltration and motility, thus consolidating the structural basis of the immune desert.¹⁷² Targeting hypoxia-induced CAF activation pathways, such as HIF-1α inhibition or lactate metabolism intervention, or remodeling the stromal microenvironment may help reverse the immune desert state and restore antitumor immune responses.¹⁷³ A deeper understanding of hypoxia-driven CAF immunosuppressive mechanisms will provide a theoretical basis for developing novel combination therapeutic strategies aimed at overcoming immune exclusion. Thus, CAF-mediated immune resistance involves not only T-cell exclusion but also coordinated suppression of macrophages, dendritic cells, NK cells, neutrophils, and myeloid-derived suppressor cells, collectively reinforcing an immunosuppressive tumor microenvironment (Figure 3).

Figure 3.

CAF-mediated immune suppression and immune-cell crosstalk in the tumor microenvironment. The schematic illustrates how CAF-derived soluble factors and matrix-associated signals coordinate immune suppression by inhibiting natural killer-cell activity, impairing dendritic-cell maturation and antigen presentation, promoting M2 macrophage polarization and myeloid-derived suppressor-cell accumulation, inducing N2 neutrophil polarization and T-helper-cell skewing, and reducing CD8⁺ T-cell cytotoxicity and IFN-γ secretion. These interactions collectively contribute to immune exclusion, immune-desert formation, and resistance to immunotherapy. Adapted from.⁵⁶ © 2023 Sarkar, Nguyen, Gundre, Ogunlusi, El-Sobky, Giri and Sarkar.

Nanomaterials for Targeting CAFs in the Treatment of Solid Tumors

CAFs are important cellular components of the stroma in solid tumors and are critically involved in tumor progression, immunosuppression, and therapeutic resistance.²³ Nanomedicine materials designed to target CAFs must address several key issues simultaneously: whether the material can reach CAF-enriched regions, whether its selectivity can be enhanced through specific ligands or microenvironmental cues, whether it can be internalized by CAFs or release therapeutic payloads in their vicinity, and whether CAF modulation can improve drug distribution into deeper tumor regions. Accordingly, the advantages and limitations of different nanoplatforms largely arise from differences in these sequential processes. Therefore, the systemic distribution, CAF binding, stromal retention, and deep penetration of nanomedicines are jointly determined by nanoparticle type, size, shape, charge, porosity, surface modification, and interactions with antibodies or extracellular-matrix components (Figure 4).

Figure 4.

Nanoparticle platforms and key physicochemical factors governing their biodistribution and biological interactions. The left panel summarizes representative NP platforms used for biomedical delivery, including virus-like particles, lipid nanoparticles, gold nanoparticles, liposomes, exosomes, polymeric micelles, membrane-derived nanoparticles, solid nanoparticles, and framework nucleic acids. The biodistribution, tissue penetration, cellular uptake, and therapeutic performance of these NPs are strongly influenced by their physicochemical properties and biological interactions. (A) NP size affects vascular transport, tissue extravasation, tumor penetration, cellular internalization, and systemic clearance. (B) NP shape influences circulation dynamics, margination behavior, cellular interactions, and transport within the extracellular matrix. (C) Surface charge regulates colloidal stability, protein adsorption, membrane association, immune recognition, and organ distribution. (D) Porosity determines surface area, cargo-loading capacity, release kinetics, and transport behavior in biological environments. (E) Surface modification strategies, including PEGylation, polymer coating, and protein- or peptide-based functionalization, can improve NP stability, prolong circulation, reduce nonspecific interactions, and enhance target specificity. (F) Interactions among NPs, antibodies, the extracellular matrix, and target cells further shape delivery efficiency, immune recognition, and therapeutic efficacy. Adapted from.¹⁷⁴ Copyright © 2025 Peng, Fang and Wang.

Key Mechanisms of CAF-Targeted Nanodelivery

FAP is one of the most widely used molecular targets for CAF-directed delivery. FAP is highly expressed in activated fibroblasts across multiple solid tumors and participates in matrix remodeling, tumor invasion, and immune regulation; therefore, it is frequently exploited to modify nanocarriers with antibodies, scFvs, small-molecule FAPI ligands, or FAP-cleavable peptides.¹⁷⁵ The development of FAPI PET further demonstrates that FAP-related signals can be used for tumor stromal imaging and may also support therapeutic delivery.¹⁷⁶ PDGFRβ, integrins, fibronectin EDB, tenascin-C, and other molecules can likewise serve as localization signals for CAFs or CAF-enriched matrices.^177–179 Ligand modification can increase the probability of nanoparticle binding to CAFs or the ECM; however, actual targeting efficiency is also shaped by ligand density, particle size, surface charge, PEG shielding, plasma protein corona formation, and CAF heterogeneity.¹⁸⁰ Insufficient ligand density may reduce binding efficiency, whereas excessive ligand density may increase nonspecific adsorption and clearance by the reticuloendothelial system.¹⁸¹ Moreover, targeting efficiency is constrained by inter- and intratumoral heterogeneity in stromal target expression, vascular accessibility, ECM pore size, and the distance between CAF-enriched regions and functional vessels. Heterogeneous expression of FAP, PDGFRβ, fibronectin EDB, tenascin-C, or other stromal targets may cause uneven nanoparticle binding among patients and within different regions of the same tumor. Therefore, CAF-targeted nanocarriers should be optimized not only for ligand–receptor affinity but also for patient-specific stromal architecture, spatial target distribution, and penetration depth.^49,182–185

After nanoparticles bind to CAFs, cellular internalization or local payload release is still required to produce therapeutic effects.¹⁸⁶ The entry of most nanomaterials into CAFs depends on endocytic pathways, including clathrin-mediated endocytosis, caveolin-associated endocytosis, macropinocytosis, and other noncanonical routes.^187–189 Carriers with smaller particle sizes, moderately positive surface charges, or exposed cell-penetrating peptides are generally more readily internalized by cells, but they may also increase uptake by normal tissues and trigger inflammatory responses.^180,190 Liposomes and lipid nanoparticles are well suited for the intracellular delivery of nucleic acids or small-molecule drugs,¹⁹¹ polymeric nanoparticles facilitate the design of endosomal escape and sustained-release structures,¹⁹² inorganic nanomaterials often act through photothermal therapy, magnetic hyperthermia, radiosensitization, or catalytic reactions,¹⁹³ and exosome- or cell membrane-coated systems exploit natural membrane structures for cell recognition and nucleic acid protection.¹⁹⁴ Thus, the differences among material platforms are not merely compositional, but also reflect distinct mechanisms of cellular uptake and payload release.

Intratumoral distribution remains a central limitation in CAF-targeted nanotherapy. CAF-enriched regions are often characterized by ECM deposition, tissue stiffening, and elevated interstitial pressure. After entering tumors, nanoparticles tend to be retained around blood vessels or in the outer stromal layers. This is particularly evident in highly fibrotic tumors such as pancreatic cancer, cholangiocarcinoma, and certain breast cancers, where simply prolonging circulation time does not ensure deep tumor penetration.^180,195,196 Larger carriers are advantageous for systemic circulation and tumor retention, whereas smaller carriers are more favorable for traversing dense stroma.¹⁹⁵ Therefore, a more rational design strategy is to first allow the nanosystem to accumulate within CAFs or CAF-enriched ECM, release antifibrotic or CAF-modulating agents to reduce collagen deposition and interstitial pressure, and subsequently promote the entry of smaller drug carriers or follow-up therapies into deeper tumor regions. NS-TAX@Lipo-VAC exemplifies this concept. This system encapsulates approximately 40-nm paclitaxel nanospheres within liposomes loaded with the TGF-β inhibitor vactosertib and uses the APTEDB peptide to target fibronectin enriched in pancreatic cancer stroma. After liposomal disintegration, VAC is released to suppress ECM hyperplasia, while the small paclitaxel nanospheres are released to promote deep penetration, establishing a cascade strategy in which the stromal barrier is first relieved and deep drug delivery is subsequently achieved.¹⁹⁷

Stimulus-responsive mechanisms represent another important approach for improving selectivity within CAF-enriched regions. pH-responsive materials are relatively mature and can exploit the acidic tumor interstitium and endolysosomal environment to promote payload release; however, acidity is not unique to CAFs, and therefore pH responsiveness alone offers limited specificity.¹⁹⁸ Enzyme-responsive systems are more closely aligned with the biological requirements of CAF targeting, especially when FAP, MMPs, or other matrix-associated enzymes are used to trigger ligand exposure, carrier disassembly, or drug release.¹⁹⁹ Redox-responsive systems are commonly used for intracellular release and can exploit differences in glutathione levels to promote drug release, but ligand-mediated targeting or local triggering is usually still required to improve CAF selectivity.²⁰⁰ Light-, magnetic field-, and X-ray-responsive systems can enhance spatiotemporal control and are suitable for inorganic nanomaterials and protein-based photosensitive platforms; however, tissue penetration depth, device requirements, and applicable tumor types may limit their broader use.¹⁹³ Therefore, the key principle in designing CAF-targeted materials should be to match the responsive signal with the actual biological features of CAF-enriched regions, rather than simply combining multiple responsive modules.

Organic Nanodelivery Systems

Liposomes and lipid nanoparticles are among the CAF-targeted platforms with relatively strong potential for pharmaceutical development. PEGylated liposomes can prolong systemic circulation, and surface conjugation with FAP- or ECM-targeting ligands may increase their retention in CAF-enriched regions.^201–203 The lipid membrane structure also facilitates the encapsulation of both hydrophilic and hydrophobic drugs and is suitable for multidrug co-delivery. For example, one study constructed GA&HA-modified liposomes for the co-delivery of berberine, a hydrophilic drug, and curcumin, a hydrophobic drug. This system successfully inhibited the activation of hepatic stellate cells into CAFs and blocked the “crosstalk” between CAFs and hepatocellular carcinoma cells, demonstrating the potential of liposomal multidrug co-delivery for microenvironmental modulation.²⁰⁴ Lipid nanoparticles are also suitable for delivering nucleic acid therapeutics such as siRNA, mRNA, and miRNA, but selective CAF targeting remains challenging.¹⁹¹ In vivo, LNPs often preferentially distribute to the liver, spleen, and immune cells; therefore, efficient delivery to CAFs requires further optimization of lipid composition, surface ligands, and routes of administration.^205,206 Overall, lipid platforms have relatively mature manufacturing processes, clinical experience, and pharmacokinetic evaluation pathways, making them suitable for constructing CAF-targeted delivery systems with considerable translational potential. Their main limitations include restricted deep penetration, substantial interference from the protein corona, and unstable selectivity across different CAF subtypes.^207,208

Polymeric nanoparticles offer extensive structural design flexibility. PLGA, PEGylated polymers, polyamino acids, dendrimers, and responsive copolymers can be engineered into pH-, enzyme-, or redox-sensitive systems, as well as multidrug co-loaded platforms. Owing to this structural tunability, polymeric nanoparticles have enabled diverse design strategies for CAF targeting. For example, Rodponthukwaji et al used a PLGA-PEG backbone to prepare nanoparticles loaded with the natural analogue 8-O-methylfusarubin and conjugated them with an anti-FAP antibody. In a three-dimensional tumor spheroid model, these nanoparticles exhibited FAP expression-dependent cellular uptake and cytotoxicity, illustrating the utility of polymeric materials for ligand conjugation and target-specific delivery.²⁰⁹ In another study, researchers designed a FAP-α-responsive block copolymer assembly, (α-GC/NAV)-CPC, capable of co-encapsulating navitoclax (NAV) and α-galactosylceramide. After reaching the tumor site, FAP-α-mediated cleavage of the responsive segment triggered disassembly, and the released drugs simultaneously depleted CAFs and activated NKT and T cells, leading to complete tumor regression without recurrence in 66.7% of mice in a triple-negative breast cancer model.²¹⁰ In addition, for CAF reprogramming rather than depletion, Wang et al synthesized a cathepsin B-responsive chondroitin sulfate–dasatinib prodrug polymer. Through responsive dasatinib release, this system converted CAFs into a quiescent state rather than killing them, thereby suppressing ECM production and promoting deep drug penetration; when combined with anti-PD-1 therapy, it further enhanced immunotherapeutic efficacy.²¹¹ These examples show that the flexibility of polymeric materials lies not only in compositional diversity, but more importantly in their capacity to support distinct therapeutic logics—including CAF depletion, CAF reprogramming, and multitarget combination therapy—through different configurations of responsive modules, targeting ligands, and drug combinations. However, increasing structural complexity also increases the difficulty of batch-to-batch consistency, quality control, and scalable manufacturing.²¹² Therefore, for clinical translation, polymeric nanoparticles require careful control over the number of functional modules and the complexity of synthetic procedures.

Protein- and albumin-based nanomedicines are characterized by favorable biocompatibility, and some protein carriers naturally interact with the tumor stroma or metabolically active cells.²¹³ Albumin can serve as a drug carrier and may also interact with stromal-associated proteins such as SPARC.²¹⁴ For CAF-targeted therapy, protein nanomedicines may be better suited for delivering antifibrotic agents, metabolic inhibitors, or small-molecule combinations, rather than merely serving as conventional chemotherapy carriers.²¹⁵ Albumin-bound paclitaxel is already a clinically used nanomedicine formulation, indicating that protein-based carriers have a certain pharmaceutical foundation. However, when albumin systems are used for specific CAF modulation, their uptake, release, and functional effects in CAFs must be directly demonstrated; CAF targeting should not be inferred solely from overall tumor accumulation.

Inorganic Nanomaterials

Inorganic nanomaterials include gold nanoparticles, iron oxides, mesoporous silica, carbon dots, black phosphorus, polydopamine, and metal–organic frameworks. Compared with liposomes and polymers, the advantages of inorganic materials arise primarily from their intrinsic optical, magnetic, thermal, imaging, and catalytic properties, and recent nanozyme studies further highlight that composition regulation, size control, morphology engineering, and surface modification can be used to tune catalytic activity, ROS management, immune modulation, and targeted drug delivery.^193,216 When applied to CAF targeting, these materials are often directed to stromal regions through FAP-, PDGFRβ-, or ECM-targeting ligands and then exert therapeutic effects through photothermal therapy, radiosensitization, ROS generation, or enzyme-responsive release.²¹⁷ For example, the FAP-targeted ferritin nanoparticle αFAP-Z@FRT adopts a protein nanocage structure and is combined with photodynamic therapy. This system uses a FAP-specific scFv to target CAFs, and after photosensitizer loading, eliminates CAFs under light irradiation while promoting antitumor immune responses.²¹⁸ Inorganic materials can also be used to address the spatial relationship between CAFs and the ECM. Regions with high FAP or MMP expression are usually regions of active ECM remodeling. If nanostructures can disassemble in these regions or release smaller therapeutic components, stromal localization and deep penetration can be achieved simultaneously. FAP-responsive carbon dot nanoassemblies embody this design concept. This system undergoes FAP-α-sensitive peptide-triggered disassembly, releases losartan to reduce collagen production, and allows carbon dots loaded with DOX and iron ions to continue diffusing and induce immunogenic cell death.²¹⁹ This example indicates that small-sized inorganic components can help overcome the stromal retention of larger particles. The limitations of inorganic materials are also evident, including liver and spleen accumulation, uncertainty regarding long-term clearance, potential metal ion toxicity, and the dependence of photothermal or photodynamic therapy on tissue penetration and equipment conditions.^220,221

Biomimetic Nanosystems

Exosomes and cell membrane-coated nanoparticles are representative biomimetic nanosystems, and recent biomimetic nanozyme designs have further shown that immune-cell membrane coating can support tumor accumulation, lymphoid-organ distribution, photodynamic ROS generation, dendritic-cell maturation, T-cell activation, and vaccine-like antitumor immunity.²²² Their advantages derive from natural membrane structures, retained membrane proteins, and intercellular recognition capabilities, making them suitable for delivering miRNAs, siRNAs, antifibrotic drugs, and metabolic modulators.²²³ Exosomes themselves participate in information exchange among CAFs, tumor cells, and immune cells; therefore, after engineering modification, they can be used to regulate CAF phenotypes.²²⁴ Engineered extracellular vesicles modified with an integrin α5-targeting peptide, IEVs-PFD/138, provide a representative example. This system carries pirfenidone and miR-138-5p, enters CAFs, and suppresses TGF-β-related signaling and collagen synthesis, thereby reducing ECM deposition and improving gemcitabine penetration in pancreatic cancer models.²²⁵ Cell membrane-coated nanoparticles typically use PLGA, lipid cores, or inorganic cores as drug-loading substrates, followed by coating with tumor cell membranes, fibroblast membranes, CAF membranes, or hybrid membranes.²²⁶ After membrane coating, nanoparticles may acquire immune-evasive, long-circulating, and homologous recognition capabilities. For example, Zang et al coated solid lipid nanoparticles with hybrid membranes derived from 4T1 cells and activated fibroblasts and co-loaded paclitaxel (PTX) with the glycolysis inhibitor PFK15. This system simultaneously targeted tumor cells and CAFs, blocked the metabolic support provided by CAFs, and enhanced chemosensitivity in fibrotic breast tumors.²²⁷ In another example, Chen et al constructed FAP-CAR cell membrane-coated PLGA nanoparticles (FAP-CAR-CM@PLGA-AB NPs), using a FAP single-chain antibody chimeric biomimetic shell to achieve dual targeting and elimination of CAFs and senescent CAFs, reduce TGF-β and collagen deposition, and improve therapeutic responses when combined with radiotherapy.²²⁸ The major limitation of biomimetic systems is standardization. Exosome source, cell culture conditions, isolation methods, drug-loading strategies, and the degree of membrane protein retention can all affect the final product; therefore, batch-to-batch consistency and quality control remain key challenges for clinical translation.^226,229

Overall, different CAF-targeted nanocarriers exhibit distinct strengths and limitations in targeting efficiency, pharmacokinetic behavior, and clinical translatability. In terms of targeting efficiency, polymeric nanoparticles and biomimetic nanosystems provide greater flexibility in structural engineering and cellular recognition, making them more suitable for developing ligand-mediated, enzyme-responsive, or cell membrane-mediated CAF-targeting strategies. By contrast, the CAF-targeting capacity of liposomes/LNPs and albumin-based carriers relies largely on surface functionalization and stromal accumulation within tumors, and their true CAF specificity often requires further validation. The major advantage of inorganic nanomaterials lies less in enhancing cellular uptake than in their ability to integrate imaging, photothermal therapy, photodynamic therapy, or radiosensitization for spatially controlled local treatment. From a pharmacokinetic perspective, liposomes/LNPs and albumin-based platforms have relatively mature foundations for evaluating circulation, biodistribution, and safety. In contrast, the in vivo behavior of polymeric, inorganic, and biomimetic systems is more readily influenced by structural complexity, degradation profiles, protein corona formation, cellular origin, and the extent of membrane protein retention; therefore, pharmacokinetic predictability and batch-to-batch consistency remain critical issues that require further optimization. Translational studies of nanomedicines also highlight that blood circulation, tissue distribution, target-cell accessibility, and clearance behavior are key determinants of in vivo therapeutic efficacy. From the perspective of clinical translation, liposomal, LNP, and albumin-based nanomedicines already have relatively well-established pharmaceutical development foundations; polymeric nanoparticles hold translational promise but require careful control of structural complexity; inorganic nanomaterials still need to overcome concerns regarding long-term safety and in vivo clearance; and exosome- or cell membrane-based biomimetic systems are mainly constrained by standardized preparation, quality control, and scalable manufacturing. Therefore, the selection of CAF-targeted nanocarriers should not be driven simply by increasing functional complexity, but should instead be guided by the therapeutic objective and balanced against targeting efficiency, pharmacokinetic predictability, and clinical translatability.

CAF-Targeted Nanotherapeutic Strategies for Precision Stromal Modulation

Following the discussion of nanomedicine materials, this chapter shifts the focus from carrier composition to therapeutic strategy. CAF-targeted nanotherapy can be organized according to the stromal function being modified: direct regulation of pathological CAF states, remodeling of dense extracellular matrix barriers, and reversal of CAF-associated immunosuppression. This strategy-based structure helps clarify how nanocarriers translate CAF biology into practical treatment designs and why precise modulation is preferable to indiscriminate stromal depletion. Current CAF-targeted nanotherapeutic strategies can therefore be broadly organized into blockade of CAF differentiation, selective CAF ablation, phenotypic reprogramming, and inhibition of CAF–cancer-cell crosstalk (Figure 5). Representative nanotherapeutic systems designed to implement these CAF-targeted strategies differ in targeting mechanisms, nanoplatform composition, therapeutic payloads, regulated CAF functions, tumor models, and reported antitumor efficacy (Table 2).

Figure 5.

CAF-targeted therapeutic strategies for precision stromal modulation. The schematic summarizes major strategies for CAF-directed intervention, including blockade of precursor-cell differentiation into tumor-promoting CAFs, selective ablation of pathogenic CAF subsets, induction of phenotypic switching toward tumor-restrictive CAF states, and inhibition of CAF–cancer-cell crosstalk. These approaches emphasize precise stromal modulation rather than broad CAF depletion, aiming to preserve beneficial CAF functions while suppressing CAF-mediated tumor progression and therapeutic resistance. Adapted from.²³ © The Author(s) 2023.

Table 2.

Representative CAF-Targeted Nanotherapeutic Strategies for Precision Stromal Modulation in Solid Tumors, Organized by Intervention Logic, Nanoplatform Design, Payload, Regulated CAF Function, Tumor Model, and Key Efficacy Evidence

Sub-strategy	Nanoplatform	Targeting/Responsive Mechanism	Payload	Regulated CAF Funct ion/Pathway	Tumor Model	Key Efficacy/Evidence	Ref
Direct elimination of pathogenic CAF subsets	Anti-FAP scFv-coated PLGA nanoparticles	FAP targeting	Nintedanib + ABT-263	Eliminate activated and senescent CAFs	Breast cancer	Eliminates CAFs (86.7% inhibition), reverses post-RT immunosuppression	[228]
Induction of CAF quiescence/ reprogramming	FAP-targeted biomimetic nanosystem (FAP-C NPs)	FAP ligand-mediated targeting	Calcitriol	Activate VDR, suppress CAF activation	Breast cancer	Converts CAFs to quiescent (3× lipid droplets); +RT: 78.3% inhibition, intratumoral CD8⁺ T 2×, LN CTL 4.39×, Th 13.5×	[102]
	Platelet membrane liposome (B@R/PL) + NETs-modified liposome (P&P/NL)	Platelet membrane biomimetic + DNA-protein network	Berberine + Ginsenoside Rg3 + Paclitaxel + PCP	Downregulate α-SMA, inhibit CAF migration	Oral squamous cell carcinoma	Downregulates α-SMA, inhibits CAF migration (76.55%), reduces lung mets by 50%	[230]
In situ engineering of CAFs	FAP-sensitive nanoparticle (PN/MCG NPs)	FAP-responsive release	IL-15 plasmid	Convert CAFs to IL-15-secreting cells	Triple-negative breast cancer	Engineers FAP⁺ CAFs to secrete IL-15; +αPD-1: tumor volume ~176.5 mm3, CD8⁺ T ~7.5%, NK ~35%	[231]
Metabolic/signaling intervention	Artemisinin nanoplatform (AI-aLDL@A)	Anisamide-modified biomimetic LDL	Artesunate + ICG	Disrupt serine-MAPK axis in CAFs	Triple-negative breast cancer	Size 60 nm, loading 12%; +PTT reduces tumor volume vs PTT alone (p<0.0001); inhibits MAP2K7, reverses PTT resistance; safe	[232]
Direct ECM degradation	PH20-modified exosome (Exos-PH20@Cur)	Surface-displayed hyaluronidase PH20	Curcumin	Degrade HA + inhibit PRMT5-Smad3	Triple-negative breast cancer	Degrades HA (penetration 200 μm), normalizes CAFs (non-killing), 39.08% apoptosis	[233]
Reducing ECM production by CAFs	FAP-targeted retinoic acid nanoparticles	FAP-mediated targeting	Retinoic acid	Suppress Fap and Acta2 expression	Colorectal cancer	Suppresses CAF activation (Fap −70%, Acta2 −60%), increases oxaliplatin 2.5×	[234]
Biochemical + physical relaxation	ROS-responsive prodrug self-assembled nanoparticle (PTC NPs)	High ROS triggers release	Pirfenidone derivative + photosensitizer	Suppress CAF profibrotic signaling + mild hyperthermia	Triple-negative breast cancer	Eliminates CAFs (PDPN⁺ −68%), reduces stiffness (301→80.5 kPa), +Doxil: 87.6% inhibition	[235]
Blocking immunosuppressive signals	In situ nanovaccine (HCPT/S@CaP/HA)	Hyaluronic acid targeting	Hydroxycamptothecin + siCXCR4	Induce ICD + block CXCL12/CXCR4	Triple-negative breast cancer	Blocks CXCL12/CXCR4, tumor weight inhibition 72.45%, M1/M2=10.9, memory T +6.53%	[236]
	Lipid nanoparticle (LNCs@si-CXCL12)	Anti-FAPα antibody targeting	siCXCL12	Silence CXCL12 in CAFs	Esophageal squamous cell carcinoma	Silences CXCL12 in CAFs (~65%), restores CD8⁺ T, prolongs survival (p=0.037)	[34]
Active immune activation	FAP-α responsive nanoparticle	FAP-α cleavage triggers release	Navitoclax + α-galactosylceramide	Deplete CAFs + activate NKT cells	Triple-negative breast cancer	Depletes CAFs, activates NKT (5×), T cells (>4×), 66.7% complete regression	[210]

Abbreviations: ABT-263, navitoclax; α-galactosylceramide, an NKT-cell agonist; Acta2, actin alpha 2, smooth muscle; αPD-1, anti-programmed cell death protein 1 antibody; α-SMA, α-smooth muscle actin; CAFs, cancer-associated fibroblasts; CaP, calcium phosphate; CD8+ T, CD8-positive T cell; CTL, cytotoxic T lymphocyte; CXCL12, C-X-C motif chemokine ligand 12; CXCR4, C-X-C chemokine receptor 4; Doxil, pegylated liposomal doxorubicin; ECM, extracellular matrix; FAP, fibroblast activation protein; FAP-α, fibroblast activation protein alpha; HA, hyaluronic acid; HCPT, hydroxycamptothecin; ICD, immunogenic cell death; ICG, indocyanine green; IL-15, interleukin 15; LDL, low-density lipoprotein; LN, lymph node; MAP2K7, mitogen-activated protein kinase kinase 7; M1/M2, classically activated/pro-inflammatory macrophage to alternatively activated/immunosuppressive macrophage ratio; NETs, neutrophil extracellular traps; NK, natural killer; NKT, natural killer T cell; NPs, nanoparticles; PDPN, podoplanin; PH20, hyaluronidase PH20; PLGA, poly(lactic-co-glycolic acid); PRMT5, protein arginine methyltransferase 5; PTT, photothermal therapy; ROS, reactive oxygen species; RT, radiotherapy; scFv, single-chain variable fragment; siCXCR4, small interfering RNA targeting CXCR4; si-CXCL12/siCXCL12, small interfering RNA targeting CXCL12; Th, T-helper cell; VDR, vitamin D receptor.

Precision Modulation Strategies Targeting CAFs in Solid Tumors

Direct CAF depletion strategies mainly rely on relatively broad CAF markers such as FAP and are designed to eliminate FAP-high CAFs or therapy-induced activated and senescent CAFs. For example, one study developed a biomimetic nanoplatform in which a FAP single-chain antibody-modified cell membrane was used to encapsulate nintedanib and ABT-263, enabling simultaneous depletion of activated CAFs and radiotherapy-induced senescent CAFs. In a breast cancer model, this strategy achieved a tumor inhibition rate of 86.7% and reversed the immunosuppressive microenvironment after radiotherapy.²²⁸ However, because CAFs exhibit marked heterogeneity and certain subsets may exert tumor-suppressive functions, maintain tissue homeostasis, or restrain tumor progression, broad CAF depletion may also produce adverse consequences. Therefore, a more advanced direction is the precise elimination of clearly defined tumor-promoting CAF subtypes. In another study targeting HNRNPC-driven ITGA1-positive myofibroblastic CAFs in oral squamous cell carcinoma, researchers developed NIR-II imaging-guided AIE nanoparticles that selectively depleted this myCAF subtype under photodynamic irradiation and restored the function of exhausted CD8⁺ T cells.²³⁷ These studies indicate that CAF depletion strategies are gradually shifting from broad targeting represented by FAP toward precise intervention against CAF subsets with well-defined identities and tumor-promoting functions. Achieving this goal requires the integration of single-cell sequencing, spatial transcriptomics, and functional validation to accurately identify subtype-specific markers and define the functional boundaries of tumor-promoting CAFs, thereby distinguishing pathogenic CAFs that should be eliminated from protective or homeostasis-maintaining CAFs that should be preserved. Only when subtype definition and target selection are sufficiently precise can CAF depletion strategies relieve therapeutic resistance while minimizing the risk of damaging beneficial CAF subsets.

Compared with direct CAF depletion, inducing CAF quiescence or phenotypic reprogramming is more consistent with the biological complexity of CAFs. Fibroblasts themselves participate in extracellular matrix deposition, tissue repair, and structural support. Although CAFs acquire profibrotic, pro-invasive, and immunosuppressive functions in tumors, they should not be regarded simply as “harmful cells”. Therefore, extensive CAF ablation may disrupt stromal homeostasis and normal tissue support, whereas reprogramming activated CAFs toward a less activated and less immunosuppressive state, or partially restoring a normal fibroblast-like phenotype, represents a milder and more controllable intervention strategy. Following this rationale, calcitriol, retinoic acid, losartan, natural products, and metabolic inhibitors have all been used to modulate CAF phenotypes. Among them, the vitamin D receptor pathway represents a typical CAF-quiescence strategy: delivery of calcitriol through a FAP-targeted biomimetic nanosystem can activate vitamin D receptors on CAFs, promote CAF inactivation, reduce CAF-mediated tumor-cell stemness and immunosuppression, and increase cytotoxic T-cell infiltration by approximately twofold when combined with radiotherapy.¹⁰² Building on this concept, CAF reprogramming can be further integrated with immune activation. For example, a nanochaperone membrane delivery platform was used to co-deliver calcitriol and the chemokine CXCL9, which not only suppressed CAF activation but also actively recruited CD8⁺ T cells, thereby significantly enhancing chemoimmunotherapy efficacy in a pancreatic cancer model.²³⁸ In addition to classical nuclear receptor pathways, natural products and signal transduction inhibitors have also been used to attenuate the tumor-promoting functions of CAFs. Berberine and ginsenoside Rg3 encapsulated in platelet membrane liposomes blocked CAF-mediated tumor-cell migration by downregulating α-SMA expression and achieved microenvironmental reprogramming of oral squamous cell carcinoma when combined with photothermal chemotherapy.²³⁰ Digoxin, by inhibiting SMAD3 phosphorylation in CAFs, synergized with PD-L1-degrading nanofibers to shift CAFs toward a more immune-supportive state in hepatocellular carcinoma.²³⁹ Overall, the core advantage of CAF quiescence or reprogramming strategies is that they do not aim to completely eliminate CAFs. Instead, by suppressing profibrotic, promigratory, and immunosuppressive programs, they preserve the fundamental structural support provided by fibroblasts while improving the therapeutic microenvironment of tumors. Compared with depletion-based strategies, these approaches emphasize functional correction rather than cellular ablation and may therefore offer greater translational potential in terms of safety and long-term microenvironmental regulation.

A more cutting-edge strategy is to engineer CAFs in situ so that they actively exert antitumor immune functions. Conventional therapies often regard CAFs as targets to be suppressed, whereas in situ engineering seeks to exploit the stable distribution and persistent residence of CAFs within the tumor stroma and convert them into local regulators of antitumor immunity. For example, FAP-sensitive nanoparticles delivering an IL-15 plasmid can engineer FAP⁺α-SMA⁺ CAFs in situ, enabling them to continuously express IL-15 and alleviate immunosuppression. Gene set enrichment analysis showed enhanced immune-cell proliferation and activation.²³¹ Another study achieved CAF-targeted gene delivery through PD-L2 recognition and reprogrammed CAFs into cells expressing the CD86 co-stimulatory signal and producing anti-PD-L1 antibodies. This approach reactivated T-cell immunity after radiotherapy, generated central memory T cells, and prevented tumor recurrence.²⁴⁰

Metabolic and signaling-pathway interventions are also emerging as important directions for CAF modulation. CAF activation is not determined by a single marker, but is jointly driven by multiple signaling and metabolic pathways, including TGF-β/SMAD, Wnt/β-catenin, DDR2, MAPK, HIF-1α, and GLUT1-related glucose metabolism.⁶³ Therefore, in addition to depleting or quiescing CAFs through surface markers, directly blocking key molecular pathways that sustain the activated CAF state can weaken their profibrotic and tumor-promoting functions at the source. Based on this rationale, one study used a dual-targeting strategy involving CAF cell membranes and aniline to inhibit CAF activation and reduce collagen secretion by suppressing the Wnt/β-catenin pathway and IGF2 expression.²⁴¹ In response to the mechanism by which CAFs induce resistance to photothermal therapy in triple-negative breast cancer through the serine–MAPK axis, artesunate was used to disrupt serine homeostasis in CAFs, attenuate the cascade activity of GTPases in the MAPK pathway, and successfully reverse photothermal resistance.²³² Another mechanistic study revealed that hypoxic tumor-derived exosomes carrying HIF1A-AS2 activate HIF-1α by sponging miR-33 in colorectal cancer, thereby promoting the conversion of normal fibroblasts into CAFs. Silencing HIF1A-AS2 or overexpressing miR-33 effectively blocked this process, providing a new strategy for preventive CAF modulation.²⁴²

Therefore, precision modulation of CAFs can be summarized into three major routes: selectively eliminating CAF subsets with clearly defined tumor-promoting functions, attenuating the fibrotic and immunosuppressive programs of activated CAFs, and using engineering strategies to reshape the local immunomodulatory capacity of CAFs. These three approaches are not mutually exclusive and are often integrated within nanodelivery systems. Future studies should further clarify which CAF subsets in different solid tumors are most suitable for depletion, quiescence induction, or engineering, and should avoid reducing CAF targeting to uniform suppression of all CAFs.

Targeted Remodeling Strategies for Dense ECM Barriers in Solid Tumors

Continuous secretion of extracellular matrix (ECM) components, such as collagen and hyaluronic acid, by CAFs is a major driver of dense stromal barrier formation in solid tumors. This barrier not only physically restricts the penetration of nanomedicines, chemotherapeutic agents, and immune cells into deeper tumor regions, but also compresses blood vessels by increasing stromal stiffness, solid stress, and interstitial fluid pressure, thereby further reducing the delivery efficiency of therapeutic agents.²⁴³ Therefore, the central goal of ECM remodeling is not simply to “clear out” the stroma, but to reduce pathological matrix deposition and mechanical stress while preserving the basic tissue architecture, thereby restoring tumor accessibility. Based on the level of intervention, current strategies can be broadly divided into three categories: direct degradation of major ECM components to rapidly open penetration channels; regulation of CAF activation to reduce continuous ECM production at its source; and integration of biochemical inhibition with physical relaxation to simultaneously reduce stromal density and interstitial pressure.

Direct degradation of major ECM components represents the most straightforward and rapidly acting approach for barrier alleviation, particularly in solid tumors with extensive hyaluronic acid or collagen deposition, where drugs have difficulty reaching the tumor core. Because hyaluronic acid markedly increases interstitial viscoelasticity and tissue hydration pressure,²⁴⁴ its enzymatic degradation can rapidly improve tissue permeability. For example, one study modified exosome surfaces with hyaluronidase PH20 and loaded them with curcumin. PH20 degraded hyaluronic acid within the tumor stroma, endowing the exosomes with stronger deep-tissue penetration; meanwhile, curcumin promoted CAF normalization by inhibiting the PRMT5–Smad3 pathway, thereby reducing continuous ECM deposition at its source. In a breast cancer model, this strategy significantly increased CD8⁺ T-cell infiltration into the tumor core region.²³³ Similar to enzymatic degradation, physical approaches can improve tissue permeability by disrupting stromal architecture. Another study used a microwave sensitizer that generated reactive oxygen species and nitric oxide under microwave irradiation, which disrupted ECM structure while enhancing heat and mass transfer, thereby opening channels for immune cells to enter deep tumor regions.²⁴⁵ These studies suggest that direct enzymatic degradation or physical disruption of the ECM can rapidly improve tumor permeability, but the durability of this effect still depends on whether CAF-mediated matrix redeposition can be simultaneously suppressed.

Therefore, compared with simply disrupting pre-existing ECM, reducing ECM production at its source is considered a more durable remodeling strategy. CAFs are among the principal sources of abnormal ECM deposition, and stronger CAF activation is generally accompanied by more active synthesis of matrix components such as collagen, fibronectin, and hyaluronic acid. Based on this rationale, inhibiting CAF activation or inducing CAF quiescence can reduce stromal density while limiting barrier reformation. For example, losartan can reprogram CAFs toward a relatively quiescent state and reduce ECM deposition, thereby promoting the penetration of photothermal agents into deeper tumor regions and activating systemic immunity. This platform also possesses fluorescence and magnetic resonance dual-modal imaging capabilities, enabling visual monitoring of therapeutic delivery.²⁴⁶ Similarly, FAP-targeted retinoic acid nanoparticles significantly suppressed the expression of the CAF activation markers Fap and Acta2 in a colorectal cancer model, reduced collagen deposition, and increased intratumoral oxaliplatin levels by 2.5-fold.²³⁴ Compared with direct enzymatic degradation, the advantage of this type of strategy lies in targeting the cellular source of abnormal ECM accumulation, making sustained stromal remodeling more achievable.

However, the ECM barrier in solid tumors is not merely a problem of “excess matrix”; it also involves multiple mechanical obstacles, including collagen cross-linking, increased tissue stiffness, elevated solid stress, and abnormal interstitial fluid pressure.²⁴⁷ Therefore, a single enzymatic degradation or CAF-inhibition strategy is often insufficient to fully restore effective penetration of drugs and immune cells. In this context, integrating biochemical inhibition with physical relaxation into a single therapeutic platform has become a more systematic direction for ECM remodeling. For example, one ROS-responsive drug conjugate can self-assemble into nanoparticles and be specifically taken up by CAFs. In the high-ROS environment, it releases a pirfenidone derivative to suppress CAF profibrotic signaling while simultaneously generating a photosensitizer that physically softens the matrix under mild hyperthermia. By concurrently attenuating CAF activation and reducing ECM mechanical stiffness, this strategy disrupts the mutually reinforcing vicious cycle between CAFs and the ECM, lowers solid stress, and enhances the intratumoral delivery of doxorubicin liposomes.²³⁵ Another study used liquid metal nanomedicines that catalytically generated hydroxyl radicals under ultrasound activation. This approach not only directly damaged CAFs and inhibited collagen secretion, but also catalyzed the splitting of water molecules in interstitial fluid, thereby reducing tumor interstitial fluid pressure.²⁴⁸ By acting simultaneously on two key mechanical barriers—solid stress and interstitial fluid pressure—this strategy significantly improved immune-cell infiltration.

Overall, ECM remodeling strategies are evolving from simple degradation of pre-existing matrix toward integrated approaches that simultaneously control the source of ECM, improve tissue permeability, and relieve interstitial pressure. Direct enzymatic degradation or physical disruption can rapidly open channels for drugs and immune cells to enter tumors, whereas CAF inactivation helps reduce continuous ECM deposition at its source. The synergistic integration of these approaches is more likely to disrupt the positive feedback loop between CAFs and the ECM. In the future, enhancing tumor permeability while avoiding excessive disruption of normal stromal support will remain a key challenge for the clinical translation of ECM remodeling strategies.

Sensitization Strategies Targeting CAF-Mediated Immunosuppressive Microenvironments in Solid Tumors

CAFs not only restrict the infiltration of immune cells into deeper tumor regions through dense ECM barriers, but also directly suppress antitumor immune responses by secreting chemokines, cytokines, and metabolism-regulating molecules.²⁴⁹ Consequently, even when immunotherapeutic targets are present within tumor tissues, effector T cells may fail to infiltrate, may not be adequately activated, or may rapidly become functionally exhausted after entering the tumor. Therefore, sensitization strategies targeting CAF-mediated immunosuppression generally follow two major principles: one is to block immunosuppressive signals released by CAFs, thereby restoring T-cell migration and cytotoxic function; the other is to actively initiate immune responses, for example by activating innate immune pathways or inducing immunogenic cell death, thereby converting poorly immunogenic “cold tumors” into more treatment-responsive “hot tumors”.

CAF-derived chemokines and cytokines are key drivers of the immunosuppressive microenvironment, among which the CXCL12/CXCR4 axis is particularly representative. CXCL12 secreted by CAFs can establish exclusionary signals within the tumor stroma, thereby restricting the migration of CD8⁺ T cells into the tumor parenchyma.²⁵⁰ Thus, blocking this signaling axis can not only remove barriers to T-cell entry, but also enhance antitumor effects induced by immunotherapy and chemotherapy. Based on this mechanism, one study constructed an in situ cancer nanovaccine co-loaded with hydroxycamptothecin and CXCR4-targeting siRNA. This system induced immunogenic tumor-cell death through hydroxycamptothecin, while simultaneously reducing CXCL12 secretion by CAFs and silencing CXCR4 expression in tumor cells, thereby achieving bidirectional blockade of the CXCL12/CXCR4 axis. In a breast cancer model, this strategy enhanced the infiltration and activity of cytotoxic T lymphocytes, increased the proportion of memory T cells in the spleen, and inhibited tumor recurrence and spontaneous lung metastasis.²³⁶ Furthermore, another study used lipid nanoparticles to deliver CXCL12-targeting siRNA, specifically silencing CXCL12 expression in CAFs of esophageal squamous cell carcinoma, restoring CD8⁺ T-cell migratory capacity, and improving mouse survival.³⁴ In addition to chemokines, inflammatory factors secreted by CAFs also influence immunotherapy responses. For example, one study found that inhibition of CAF autophagy reduced IL-6 production, which in turn decreased USP14 expression and ultimately upregulated PD-L1 levels in tumor cells; targeting CAF autophagy with chloroquine liposomes enhanced the efficacy of immunochemotherapy.²⁵¹ These studies indicate that intervening in CAF-derived immunosuppressive signals can improve immunotherapy efficacy at two levels: relieving immune exclusion and restoring T-cell function.

However, relieving CAF-mediated immunosuppression alone may not be sufficient to generate robust antitumor immunity. For cold tumors characterized by insufficient immune-cell infiltration, weak antigen presentation, or inadequate dendritic-cell activation, additional active immune stimulation is often required. Accordingly, increasing numbers of strategies now integrate CAF modulation with innate immune activation, immunogenic cell death, or immune-cell recruitment, thereby shifting the therapeutic logic from “releasing suppression” to “actively initiating immunity”. Taking the cGAS–STING pathway as an example, a pH-gated hydrogel can co-deliver pirfenidone and manganese–curcumin nanoparticles. In this system, pirfenidone modulates the CAF phenotype and attenuates the fibrotic barrier, while manganese ions activate the cGAS–STING pathway and promote dendritic-cell maturation; meanwhile, near-infrared laser irradiation induces photothermal effects and reactive oxygen species production, further amplifying antitumor immune responses. After treatment, CD8⁺ T-cell infiltration in tumors was enhanced, and proinflammatory cytokines such as TNF-α, IFN-γ, and IL-6 were significantly upregulated.²⁵²

Building on this rationale, some platforms further integrate CAF-barrier disruption with immunostimulant delivery within a single system, enabling immune activation to occur more efficiently in deeper tumor regions. For example, a carrier-free nanopolymer covalently linked a mitochondria-targeted photosensitizer with a STING agonist. In a pancreatic cancer model, the photosensitizer first disrupted the CAF-associated stromal barrier and triggered immunogenic cell death, thereby creating conditions for deep penetration of the STING agonist; subsequent STING activation promoted dendritic-cell maturation, reprogrammed macrophages toward the M1 phenotype, and enhanced cytotoxic T-cell infiltration, ultimately driving the conversion of cold tumors into hot tumors.²⁵³ Similarly, FAP-α-responsive nanoparticles co-delivering navitoclax and α-galactosylceramide used navitoclax to deplete CAFs and reduce tumor interstitial fluid pressure, thereby creating conditions for immune-cell entry; α-galactosylceramide then activated natural killer T cells, increasing NKT-cell abundance by fivefold and T-cell abundance by more than fourfold, and achieved tumor eradication in 66.7% of mice.²¹⁰ Another study used magnetic nanoparticles to induce CAF ferroptosis through iron and copper ions, while reshaping the chemokine secretion profile of CAFs, promoting dendritic-cell maturation and CD8⁺ T-cell activation; therapeutic efficacy was further validated in colorectal cancer patient-derived xenograft and organoid models.²⁵⁴

Overall, sensitization strategies targeting CAF-mediated immunosuppression are evolving from the simple blockade of individual immunosuppressive factors toward integrated modulation of T-cell entry, antigen presentation, innate immune activation, and effector-cell cytotoxicity. Blocking CAF-derived signals such as CXCL12/CXCR4 and IL-6 can relieve immune exclusion, whereas combining these approaches with STING activation, immunogenic cell death, NKT-cell activation, or CAF ferroptosis can further amplify local antitumor immunity. Compared with simply increasing the dose of immunotherapeutic agents, these strategies place greater emphasis on remodeling the CAF-dominated immunosuppressive microenvironment, thereby offering a more targeted and mechanistically informed approach for sensitizing solid tumors to immunotherapy.

CAF-Targeted Nanomedicine in Combination Therapy

Because CAFs simultaneously influence drug penetration, immune activation, and cellular trafficking, their modulation is most valuable when integrated with established therapeutic modalities, particularly radiotherapy- or phototherapy-based nanoplatforms that induce sustained ROS generation, immunogenic cell death, macrophage repolarization, and durable antitumor immunity.²⁵⁵ This chapter discusses how CAF-targeted nanosystems can be combined with immune checkpoint blockade and adoptive cell therapy. The central goal of such combinations is to synchronize stromal barrier relief with local immune reactivation, rather than simply adding another therapeutic component to an already complex regimen.

CAF-Targeted Nanomedicine Strategies for Enhancing the Efficacy of Immune Checkpoint Inhibitors (ICIs)

Immune checkpoint inhibitors (ICIs), such as anti-PD-1/PD-L1 antibodies, reactivate T-cell-mediated antitumor immunity by blocking inhibitory immune signals and have become important therapeutic modalities for a variety of solid tumors.²⁵⁶ However, in CAF-enriched fibrotic solid tumors, the clinical efficacy of ICIs remains suboptimal, largely because of poor intratumoral drug penetration, insufficient T-cell infiltration, and a persistent immunosuppressive microenvironment.¹⁷ CAF-targeted nanomedicine strategies offer a unique opportunity to overcome these barriers: they not only facilitate more efficient delivery of ICIs to tumor sites, but also remodel CAF-dominated physical and immune barriers, thereby fundamentally broadening the therapeutic window and durability of ICI-based therapy.

First, CAF-targeted nanocarriers can enhance the local delivery and therapeutic efficacy of ICIs within dense tumor stroma by attenuating stromal barriers. Following systemic administration, conventional ICIs are often restricted by the dense ECM deposited by CAFs, resulting in insufficient drug exposure within the tumor core and limited therapeutic benefit.²⁵⁷ CAF-targeted nanosystems can selectively accumulate in CAF-enriched regions through recognition of CAF-associated markers such as FAP and PDGFR, or through biomimetic membrane camouflage. By modulating CAFs and loosening the stromal architecture, these platforms can open interstitial penetration routes and enable ICIs to reach their sites of action more efficiently. For example, cancer cell-derived extracellular vesicles loaded with the GLUT1 inhibitor BAY-876 reduced ECM stiffness by 46.7% and promoted the infiltration of CD3⁺CD8⁺ T cells into the tumor core, increasing their infiltration by approximately 3.2-fold. When combined with anti-PD-L1 therapy, this strategy reduced tumor volume by 78.4% in an LLC lung cancer model and significantly prolonged survival.²⁵⁸ These findings suggest that CAF metabolic intervention and ECM softening can expand the effective range of ICIs from the tumor periphery into the tumor core.

Second, CAF-targeted nanoplatforms can amplify the immune-activating effects of ICIs by synergistically remodeling the immunosuppressive microenvironment. CAFs not only establish physical stromal barriers, but also secrete immunosuppressive mediators such as TGF-β, CXCL12, and IL-6, and recruit suppressive immune cell populations. By integrating ICIs with CAF-modulating agents, chemokines, or immunostimulatory modules, nanocarriers can simultaneously relieve CAF-mediated immunosuppression and actively reinforce antitumor immunity. The nanochaperone system Cal@nChap-CXCL9 co-delivers calcitriol and CXCL9, in which calcitriol induces quiescence of activated CAFs and relaxes the ECM, while CXCL9 establishes a chemokine gradient that promotes CD8⁺ T-cell recruitment. In the Panc02 pancreatic cancer model, Cal@nChap-CXCL9 treatment increased the proportion of CD8⁺ T cells to 18.7% ± 1.5% among CD45⁺ cells. When combined with anti-PD-1 therapy, the tumor growth inhibition rate reached 69%, with a combination index Q value of 1.21, indicating synergistic efficacy. In a large-tumor model, Cal@nChap-CXCL9 combined with gemcitabine and anti-PD-1 almost completely suppressed tumor proliferation and significantly prolonged median survival.²³⁸ The carrier-free nanoremodeler QN NPs integrates quercetin-mediated CAF normalization, Mn²⁺-induced ICD/STING activation, and NLG919-mediated IDO1 inhibition. In the 4T1 breast cancer model, QN NPs increased the proportion of CD8⁺ T cells, reduced the proportion of Foxp3⁺CD25⁺ Tregs, and significantly inhibited primary tumor growth. When combined with anti-PD-L1 therapy, both the number and size of lung metastases were markedly reduced; at the experimental endpoint, all mice in the combination group survived, whereas 3/4 mice in the control group had died.²⁵⁹

Finally, some CAF-targeted or TME-responsive nanoplatforms can further broaden the therapeutic window of ICI-based combination therapy by improving drug loading, tumor accumulation, and programmed release. PmMN@Om&A, a platelet membrane-coated magnetic metal–organic framework, co-delivers oxymatrine and astragaloside IV with a total drug loading of 33.77 wt%. In an HCC model, PmMN@Om&A combined with α-PD-1 achieved a tumor inhibition rate of 84.15% and prolonged mouse survival. The study also noted that the clinical objective response rate of PD-1 inhibitor therapy in HCC patients is typically only 15%–20%, underscoring the translational rationale for improving ICI sensitivity through dual modulation of CAFs and TILs.²⁶⁰ The stimulus-responsive polymeric prodrug DPPA-TRPP/Tab integrates CAF inhibition, ROS modulation, ICD induction, and PD-L1 blockade within a single platform. In vivo, this system suppressed CAF formation, downregulated Tregs, and promoted T-lymphocyte infiltration, ultimately achieving a 60% complete tumor regression rate and long-term immune memory in mice.²⁶¹

CAF-Targeted Nanomedicine for Enhancing Adoptive Cell Therapy in Solid Tumors

Adoptive cell therapy, particularly chimeric antigen receptor T-cell (CAR-T) therapy, has achieved remarkable progress in hematological malignancies; however, its efficacy in solid tumors remains substantially constrained by the physical and immunological barriers established by CAFs.^262,263 Based on this rationale, recent nanomedicine strategies have mainly evolved along two directions: first, using nanoplatforms to deplete or modulate CAFs, thereby enabling ex vivo-manufactured CAR-T cells to function more effectively; and second, using nanocarriers to directly deliver CAR components in vivo, generating CAF-targeted CAR-T cells within the body and simplifying the conventional CAR-T manufacturing workflow.

In the context of ex vivo–manufactured CAR-T cells, prior depletion or attenuation of CAFs using nanoparticles represents a relatively straightforward strategy to sensitize tumors to CAR-T therapy. The rationale is that weakening the CAF-derived physical barrier and immunosuppressive milieu may facilitate CAR-T cell access to tumor cells and enhance effector molecule release. For example, in triple-negative breast cancer (TNBC), CAFs highly express fibroblast activation protein (FAP), whereas tumor cells express folate receptor α (FRα). Based on this differential expression pattern, researchers developed a dual-targeting combination strategy: FAP antibody-coated nanoparticles, termed anti-FAP@OMF-NPs, were loaded with 8-O-methylfusarubin to selectively eliminate CAFs, while FRα-specific CAR-T cells were generated to directly kill tumor cells. In this study, the surface expression rate of anti-FRα-CAR on T cells was 37.61% ± 7.31%, and the resulting CAR-T product was predominantly composed of CD8⁺ cytotoxic T cells, accounting for 65.75% ± 0.495% of the population. In a 3D TNBC–CAF co-culture heterospheroid model, combined treatment with anti-FAP@OMF-NPs and FRα CAR-T cells enhanced the disruption of MDA-MB-231/130CAF and MDA-MB-231/132CAF heterospheroids, accompanied by increased secretion of IFN-γ, granzyme A, and granzyme B. These findings indicate that CAF-targeted nanoparticles can potentiate the cytotoxic activity of FRα CAR-T cells in CAF-enriched tumor environments.²⁶⁴

Nevertheless, conventional CAR-T therapy still depends on the isolation of patient-derived T cells, followed by ex vivo transduction, expansion, and reinfusion. This process is technically complex, costly, and often requires lymphodepleting preconditioning. To further simplify the therapeutic workflow, several studies have begun to explore nanocarrier-based strategies for generating CAR-T cells directly in vivo. Unlike the approach of first preparing CAR-T cells ex vivo and then remodeling the tumor microenvironment, this strategy delivers CAR-encoding information directly to endogenous T cells, enabling them to acquire CAF-targeting capacity within the patient. In pancreatic cancer, researchers developed anti-CD5-targeted lipid nanoparticles, termed tLNPs, encapsulating FAP-CAR mRNA. In vitro, 60.6% of T cells expressed FAP-CAR after treatment with CD5-tLNP-FAP-CAR, exceeding the 53.9% expression achieved by retrovirally transduced FAP-CAR-T cells. On day 3 after a single intravenous injection, FAP-CAR⁺ T cells were detected in more than 45% of splenic CD3⁺ T cells, more than 69% of peripheral blood T cells, and more than 35% of tumor-infiltrating T cells. The abundance of in vivo–generated CAR-T cells exceeded that observed after infusion of 1 × 10⁷ ex vivo retrovirally transduced FAP-CAR-T cells. On day 10 after treatment, the mean tumor volume was reduced by 74% in the in vivo CAR-T group and by 48% in the ex vivo CAR-T group; the endpoint tumor weights were 0.182 g and 0.304 g, respectively, both lower than the 0.462 g observed in the PBS group. Mechanistically, the in vivo CAR-T strategy reduced FAP⁺ stromal cells by more than 60% and increased endogenous intratumoral CD8⁺ and CD4⁺ T cells by 1.5-fold and 2.5-fold, respectively.²⁶⁵ Another study employed an mRNA-LNP strategy encoding FAP-CAR to reprogram host immune cells in vivo to target FAP⁺ CAFs in solid tumors. In vitro transfection experiments showed that human FAPCAR mRNA and mouse FAPCAR mRNA endowed 29.7% and 37.3% of splenocytes, respectively, with FAP-binding capacity. At 24 h after in vivo injection of hFAPCAR mRNA-LNPs, a distinct population of FAP-binding-positive cells was detectable among splenocytes. This strategy induced tumor regression across multiple solid tumor models, and its antitumor efficacy was further enhanced when combined with 5-FU and immune checkpoint inhibitors. In a colon cancer model, mice cured by the LNP + 5-FU + ICI combination completely rejected rechallenge with homologous MC38 tumors, whereas heterologous E0771 tumors continued to grow, suggesting that this strategy can induce antigen-specific immune memory.²⁶⁶

In addition to the two strategies described above, some nanoplatforms do not directly introduce exogenous CAR-T cells or CAR-encoding information, but instead induce T-cell responses with adoptive transfer potential through CAF-targeted therapy. Ferritin nanoparticle-mediated FAP-targeted photodynamic therapy (PDT) exemplifies this concept. In this approach, researchers conjugated a FAP-specific single-chain antibody and the photosensitizer ZnF16Pc to ferritin nanoparticles to construct αFAP-Z@FRT, enabling selective accumulation in FAP-positive CAFs and photodynamic cytotoxicity upon light irradiation. In a single-tumor 4T1 model, αFAP-Z@FRT-mediated PDT achieved a tumor inhibition rate of 65.3% on day 21 after treatment. In a bilateral tumor model, the primary tumor inhibition rate reached 90.2% on day 23 after treatment, while the volume of non-irradiated distant tumors decreased to 88.8 ± 25.1 mm³, markedly lower than the 859.7 ± 72.8 mm³ observed in untreated animals. Immunological analyses showed that, after PDT, the CD8⁺/Treg ratio increased by approximately 1.6-fold, 3.2-fold, and 5.4-fold in primary tumors, distant tumors, and draining lymph nodes, respectively. When combined with anti-PD-1 therapy, the CD8⁺/Treg ratio further increased by 4.8-fold and 5.7-fold in primary and distant tumors, respectively. ELISpot analysis showed that 4T1-specific IFN-γ⁺ T cells increased from 38.0 per million splenocytes to 108.5 per million splenocytes, and further to 185.0 per million splenocytes after combination with anti-PD-1. CAF-specific IFN-γ⁺ T cells increased from 9.0 per million splenocytes to 67.5 and 76.5 per million splenocytes, respectively. In adoptive transfer experiments, T cells isolated from 4T1 tumor-bearing mice treated with αFAP-Z@FRT PDT were transferred into nude mice bearing A549 tumors; on day 43, the tumor volume was 216.5 mm³, lower than the 542.1 mm³ observed in mice receiving T cells from PBS-treated donors. T cells derived from donors treated with αFAP-Z@FRT PDT plus anti-PD-1 showed a comparable tumor-suppressive effect, with an endpoint tumor volume of 218.6 mm³. These results indicate that FAP-targeted PDT not only eliminates CAFs and suppresses local tumor growth, but also induces antitumor T-cell responses with adoptive transfer activity.²¹⁸

Overall, CAF-targeted nanomedicine can enhance adoptive cell therapy through three interconnected mechanisms. First, nanoparticles can deplete or modulate CAFs, thereby removing stromal and immunosuppressive barriers that limit the infiltration and effector function of ex vivo-manufactured CAR-T cells. Second, lipid nanoparticles can deliver CAR mRNA directly to endogenous T cells, enabling in vivo acquisition of CAF-targeting capacity and reducing the complexity of conventional CAR-T manufacturing. Third, FAP-targeted photodynamic therapy and related approaches can induce endogenous T-cell responses with demonstrable adoptive-transfer potential. Together, these strategies represent a progression from microenvironmental support for conventional CAR-T therapy, to direct in vivo CAR-T generation, and further to the induction of transferable antitumor immunity through CAF-targeted treatment. In the future, selecting appropriate combinations of nanoplatforms and adoptive cell products according to CAF abundance, spatial distribution, and immunosuppressive intensity in different solid tumors will be an important direction for improving the efficacy of adoptive cell therapy in solid tumors.

Clinical Translation Status

The clinical translation of CAF-targeted therapy remains uneven but instructive. FAP-directed platforms and stroma-modulating agents have shown that stromal targets can be exploited for drug localization, immune activation, or delivery improvement, yet clinical outcomes also indicate that marker expression alone is insufficient to guarantee efficacy. This chapter separates FAP-centered clinical strategies from broader nanomedicine and stromal-modulation approaches, highlighting how clinical evidence can guide future patient selection, target validation, and combination design.

Clinical Trials Targeting FAP: From Direct Inhibition to Delivery Platforms

FAP is highly expressed in cancer-associated fibroblasts, whereas its expression is relatively limited in most normal tissues. It has therefore long been regarded as a therapeutic target with tumor-stroma selectivity. The clinical translation of FAP-targeted therapy has broadly progressed through three stages. Early studies mainly attempted to directly interfere with FAP using antibodies or small-molecule inhibitors. Subsequently, the therapeutic rationale gradually shifted toward using FAP as a tumor-stromal localization marker to deliver effector molecules, including cytokines and immune agonists, into the tumor microenvironment. More recently, the success of FAPI imaging has further accelerated the development of FAP-targeted radioligand therapy. Overall, clinical outcomes with FAP-targeted therapy have been heterogeneous, and the value of FAP appears to lie more in its role as a delivery and localization platform than as a purely functional target for direct inhibition. These clinical studies collectively indicate that FAP-targeted and stromal-modulating approaches have heterogeneous therapeutic outcomes, with stronger translational potential when FAP is used as a localization or delivery platform rather than as a simple depletion target (Table 3).

Table 3.

Representative Clinical Evidence for FAP-Targeted Therapies, Nanomedicines, and Stromal-Modulating Agents Relevant to CAF-Targeted or CAF-Associated Tumor-Stroma Modulation

Drug	Targeting Mechanism	Cancer Type (line)	Phase	Key Results	Status	Ref
Sibrotuzumab	Humanized anti-FAP mAb	Metastatic colorectal cancer (mCRC)	Early Phase II	ORR 0%, only 2 SD; did not meet minimum efficacy criteria	Failed	[267]
Sibrotuzumab	Humanized anti-FAP mAb	Various FAP-positive advanced cancers	Phase I	No objective responses; anti-drug antibodies (HAHA) in 12.5% of patients	Failed	[268]
Talabostat (Val-boroPro)	Small molecule FAP/DPPIV inhibitor	Metastatic colorectal cancer (pretreated)	Phase II	ORR 0%, 21% SD (median 25 weeks); partial FAP enzyme inhibition	Failed (minimal clinical activity)	[269]
Simlukafusp alfa + atezolizumab	FAP-IL2v immunocytokine	Pancreatic ductal adenocarcinoma (PDAC), 2L	Phase I/IIb	ORR 0–7.1%; limited efficacy	Failed/limited activity	[270]
Simlukafusp alfa + cetuximab	FAP-IL2v	Recurrent/metastatic head and neck SCC (HNSCC)	Phase Ib	ORR 7% (4/58); pharmacodynamic T cell expansion but low antitumor activity	Failed (insufficient for further development)	[271]
Simlukafusp alfa + pembrolizumab	FAP-IL2v	Advanced/metastatic melanoma (mostly CPI-pretreated)	Phase Ib	CPI-pretreated: ORR 6.7% (5/75, all PR); CPI-naïve: ORR 25%; median PFS 3.1 mo	Failed (discontinued due to lack of activity)	[272]
Simlukafusp alfa + atezolizumab	FAP-IL2v	Recurrent/metastatic cervical SCC	Phase II	ORR 27% (12/45, including 3 CR); median PFS 3.7 mo	Clinically active	[273]
Simlukafusp alfa + atezolizumab	FAP-IL2v	Esophageal SCC (pretreated)	Phase II	ORR 20.6% (7/34); median PFS 1.9 mo	Limited activity	[274]
RO7300490	FAP-targeted CD40 agonist bispecific antibody	Advanced/metastatic solid tumors	Phase I (first-in-human)	ORR 0%, DCR 42.5%; induced intratumoral DC-LAMP⁺ DC increase and tertiary lymphoid structures	Limited activity (no objective response)	[275]
177Lu-FAPI-XT	FAPI radioligand therapy (RLT)	Advanced sarcoma and other solid tumors (failed standard therapy)	Phase I	0% CR/PR, 35.7% SD; median PFS 4.63 mo; short tumor retention (15.0 h)	Failed/needs optimization	[276]
177Lu-EB-FAPI (177Lu-LNC1004)	FAPI-RLT with albumin-binding moiety	Metastatic radioiodine-refractory thyroid cancer (mRAIR-TC)	Phase I	ORR 25% (3/12 PR), DCR 83%; tumor half-life extended to 92.5 h	Therapeutic potential (strategy optimization)	[277]

Abbreviations: 2L, second-line therapy; 177Lu, lutetium-177; CAFs, cancer-associated fibroblasts; CPI, checkpoint inhibitor; CR, complete response; DCR, disease control rate; DC, dendritic cell; DC-LAMP, dendritic cell lysosome-associated membrane glycoprotein; DPPIV, dipeptidyl peptidase IV; FAP, fibroblast activation protein; FAPI, fibroblast activation protein inhibitor; FAP-IL2v, FAP-targeted interleukin-2 variant; HAHA, human anti-human antibody; HNSCC, head and neck squamous cell carcinoma; mAb, monoclonal antibody; mCRC, metastatic colorectal cancer; mo, months; mRAIR-TC, metastatic radioiodine-refractory thyroid cancer; ORR, objective response rate; PDAC, pancreatic ductal adenocarcinoma; PFS, progression-free survival; PR, partial response; Ref, reference; RLT, radioligand therapy; SCC, squamous cell carcinoma; SD, stable disease.

Early Clinical Results of Anti-FAP Antibodies and Small-Molecule Inhibitors

FAP-targeted therapy was initially based on a relatively straightforward hypothesis: FAP is not only a marker of tumor stroma, but may also functionally contribute to tumor growth and invasion. Accordingly, early clinical development focused on anti-FAP antibodies and inhibitors of FAP-related enzymatic activity, with the aim of producing antitumor effects by recognizing or blocking FAP. However, clinical results did not demonstrate sufficient therapeutic efficacy for this strategy.

Sibrotuzumab is a humanized anti-FAP monoclonal antibody derived from the murine F19 antibody.²⁷⁸ In an early Phase II study involving patients with metastatic colorectal cancer, 25 patients received weekly intravenous infusions of 100 mg for 12 weeks. Among 17 evaluable patients, no complete or partial responses were observed, only two patients achieved stable disease, and most patients still experienced tumor progression. The study failed to meet the prespecified minimum efficacy criterion and therefore did not support further development.²⁶⁷ Another phase I dose-escalation study enrolled 26 patients with FAP-positive advanced cancers and likewise observed no objective tumor responses; some patients also developed anti-drug antibodies, suggesting that repeated administration may be affected by immunogenicity.²⁶⁸

The small-molecule inhibitor talabostat also failed to overcome the limited efficacy of early FAP-targeted therapy. Talabostat was the first FAP/DPPIV inhibitor to enter clinical trials and was evaluated in a phase II study of patients with metastatic colorectal cancer who had failed prior chemotherapy. After 28 patients received oral talabostat at 200 μg twice daily, no objective responses were observed, and only six patients achieved stable disease, with a median duration of 25 weeks. Although peripheral blood FAP enzymatic activity was significantly inhibited, enzymatic inhibition did not translate into clear clinical benefit.²⁶⁹

These results indicate that FAP expression and FAP enzymatic activity cannot be directly equated with an effective therapeutic target. Simply recognizing FAP-positive stromal cells or inhibiting FAP-related enzymatic activity may be insufficient to disrupt the support provided by tumor stroma for tumor progression and immune escape. Consequently, subsequent studies gradually moved away from treating FAP as a standalone inhibitory target and instead began to use it as a localization gateway within the tumor stroma to enhance the local enrichment of other therapeutic molecules in the tumor microenvironment.

Cancer Type-Dependent Activity of the FAP-IL2v Immunocytokine

Following the insufficient efficacy of early anti-FAP antibodies and small-molecule inhibitors, the rationale for FAP-targeted therapy gradually shifted from direct FAP inhibition toward FAP-mediated local delivery. Simlukafusp alfa (RO6874281) is an immunocytokine developed in this context. It fuses an IL-2 variant with an anti-FAP antibody and is designed to concentrate cytokine activity within the tumor microenvironment through FAP expression in the tumor stroma, thereby enhancing local NK-cell and CD8⁺ T-cell responses while reducing the systemic toxicity associated with conventional IL-2 therapy.²⁷⁹

Available clinical results show that simlukafusp alfa has not demonstrated consistent pan-cancer efficacy, and its clinical activity appears to depend strongly on tumor type and immune microenvironmental context. In pancreatic ductal adenocarcinoma, recurrent or metastatic head and neck squamous cell carcinoma, and melanoma previously treated with immune checkpoint inhibitors, combination therapy produced certain pharmacodynamic changes, such as expansion of immune effector cells or local immune activation. However, overall objective response rates were low and did not generate sufficient clinical benefit to support continued development.^270–272 These findings suggest that, in tumors with profound immunosuppression, impaired effector-cell function, or prior immunotherapy failure, simply enhancing local IL-2 signaling is insufficient to reverse complex treatment resistance.

By contrast, clearer efficacy signals have emerged in some squamous cell carcinomas. Recurrent or metastatic cervical squamous cell carcinoma is a particularly notable example: simlukafusp alfa combined with atezolizumab achieved a confirmed objective response rate of 27%, including complete responses. Response activity was also observed in esophageal squamous cell carcinoma, although the durability of disease control remained limited.^273,274 These results indicate that FAP-mediated IL-2 delivery is not ineffective per se, but requires a suitable tumor immune context to translate into clinically meaningful responses.

FAP-Targeted CD40 Agonist Bispecific Antibody RO7300490

The clinical results of FAP-IL2v suggest that FAP has value as a localization marker, but therapeutic efficacy depends on whether the delivered molecule can trigger a sufficiently effective immune response within the tumor microenvironment. Along this line of development, FAP-targeted strategies have been further extended to immune co-stimulatory pathways. RO7300490 is a FAP-targeted bispecific CD40 agonist antibody designed to activate dendritic cells through FAP-mediated local crosslinking, thereby promoting antigen presentation and T-cell-mediated antitumor immunity.²⁸⁰

In the first-in-human phase I clinical trial, 80 patients with advanced or metastatic solid tumors received intravenous RO7300490 every two weeks at doses ranging from 16 to 1100 mg. Treatment-related adverse events occurred in 66.3% of patients, most of which were grade 1–2; grade 3–4 treatment-related adverse events accounted for only 3.8%, and no grade 5 treatment-related adverse events occurred. Radiolabeled RO7300490 showed rapid and sustained uptake in tumor tissues, and tumor biopsies indicated increased densities of DC-LAMP-positive dendritic cells and B cells after treatment, along with the formation of local tertiary lymphoid structures colocalized with dendritic cells.²⁷⁵ These pharmacodynamic findings demonstrate that RO7300490 can induce immune-modulatory effects within the tumor microenvironment. However, no objective responses were observed in this study, and the disease control rate was 42.5%, suggesting that local immune activation induced by monotherapy remains insufficient to produce clear tumor regression.²⁷⁵

FAPI Radioligand Therapy

In addition to immune-delivery strategies, FAPI radioligand therapy represents another important route for the clinical translation of FAP targeting. FAPI PET/CT imaging has shown high tumor uptake and low background signal across multiple solid tumors, providing a foundation for therapeutic translation. However, a high tumor-to-background ratio in diagnostic imaging does not necessarily guarantee therapeutic efficacy. For radioligand therapy, the key question is not only whether the agent can enter the tumor, but also whether it can remain within tumor tissue long enough to deliver an adequate radiation dose.

The first-in-human dose-escalation study of ¹⁷⁷Lu-FAPI-XT enrolled 14 patients with advanced solid tumors refractory to standard therapy, predominantly patients with sarcoma. After receiving 3.7–5.55 GBq per cycle, patients showed no dose-limiting toxicities or grade 3 or higher treatment-related adverse events, indicating favorable safety. In terms of efficacy, RECIST assessment showed no complete or partial responses; only five patients achieved stable disease, corresponding to a disease stabilization rate of 35.7%. In the fibrosarcoma subgroup, four patients achieved stable disease, with a disease stabilization rate of 50%. The mean effective half-life of tumor lesions in this study was 15.0 hours, and the median progression-free survival was 4.63 months.²⁷⁶ These results demonstrate the feasibility and safety of ¹⁷⁷Lu-FAPI-XT radioligand therapy, while also suggesting that limited tumor retention may restrict therapeutic efficacy.

To improve the retention of FAPI in tumor tissues, subsequent studies have attempted to optimize pharmacokinetics through albumin-binding structures. ¹⁷⁷Lu-EB-FAPI, also known as ¹⁷⁷Lu-LNC1004, prolongs blood circulation and tumor retention through conjugation with an albumin-binding group. In a first-in-human dose-escalation study involving patients with metastatic radioiodine-refractory thyroid cancer, 12 patients received 2.22–4.99 GBq per cycle. The treatment was well tolerated, the mean effective half-life of tumor lesions was extended to 92.5 hours, and the mean absorbed tumor dose reached 8.50 Gy/GBq. In terms of efficacy, the partial response rate was 25%, the stable disease rate was 58%, and the disease control rate reached 83%.²⁷⁷

Compared with ¹⁷⁷Lu-FAPI-XT, the results of ¹⁷⁷Lu-EB-FAPI suggest that prolonging tumor retention time may be an important strategy for improving the efficacy of FAPI radioligand therapy. Current evidence indicates that FAPI radioligand therapy has certain advantages in safety, but its efficacy is jointly influenced by tumor retention time, absorbed radiation dose, radionuclide selection, and indication selection. Future studies should further clarify which tumor types are most suitable for FAPI radioligand therapy and how molecular-structure optimization and rational combination therapy can improve therapeutic dose delivery.

Nanomedicines and Stroma-Modulating Agents Targeting CAFs and Tumor Stroma

Unlike direct FAP-targeted therapies, broader strategies for CAF- and tumor-stroma-directed treatment also include approaches aimed at optimizing drug delivery and modulating the stroma. These strategies do not necessarily eliminate CAFs; rather, they attempt to improve drug entry, distribution, and action within the dense tumor stroma, or to enhance the efficacy of existing treatments by interfering with CAF-related inflammatory and drug-resistance signals. Pancreatic cancer is the most representative disease model in this area because of its abundant tumor stroma, high interstitial pressure, poor vascular perfusion, and pronounced drug delivery barriers. Consequently, nanomedicines, hyaluronan-degrading agents, and inhibitors of CAF-related signaling pathways have all undergone extensive clinical evaluation in pancreatic cancer.

Nab-Paclitaxel

Nab-paclitaxel is a nanoparticle formulation of paclitaxel that uses human serum albumin as a carrier. Compared with conventional solvent-based paclitaxel, nab-paclitaxel was designed to improve drug solubility, tissue distribution, and intratumoral delivery.²⁸¹ Because stromal proteins such as SPARC may mediate drug accumulation via albumin binding, nab-paclitaxel has also been considered to act by modulating the tumor stroma and drug delivery processes in stroma-rich pancreatic cancer.²⁸²

The MPACT trial is the pivotal Phase III randomized controlled trial that established the clinical value of nab-paclitaxel in metastatic pancreatic cancer. The study enrolled 861 patients with metastatic pancreatic cancer and compared nab-paclitaxel plus gemcitabine versus gemcitabine alone as first-line therapy. Results showed that the combination significantly prolonged median overall survival (8.5 vs. 6.7 months; hazard ratio 0.72), with 1-year survival rates of 35% vs. 22% and 2-year survival rates of 9% vs. 4%. Median progression-free survival was also significantly improved (5.5 vs. 3.7 months), and objective response rates were 23% vs. 7%. Although the combination increased the incidence of grade ≥3 neuropathy and neutropenia, neuropathy was mostly reversible, with a median recovery time of 29 days.²⁸³

The significance of this trial lies in demonstrating that formulation improvement and combination chemotherapy can enhance treatment outcomes in pancreatic cancer. The clinical benefit likely arises not from a single stromal factor but from a combination of albumin carrier properties, synergistic effects of paclitaxel and gemcitabine, improved intratumoral drug exposure, and changes in the stromal state. Thus, nab-paclitaxel serves as a representative example of successful drug delivery optimization in stroma-rich tumors and provides a rationale for the development of other nanomedicines in pancreatic cancer.

Liposomal Irinotecan

Following nab-paclitaxel, liposomal irinotecan further illustrates the value of nanomedicines in pancreatic cancer therapy. Liposomal irinotecan encapsulates the topoisomerase I inhibitor irinotecan in liposomes, with the primary goals of prolonging drug circulation time, modifying drug release behavior, increasing exposure of the active metabolite in tumor tissue, and reducing non-target tissue toxicity.²⁸⁴ Unlike direct CAF-targeting agents, liposomal irinotecan focuses on pharmacokinetic optimization to enhance the efficacy of cytotoxic drugs within the complex tumor microenvironment. A systematic review and meta-analysis encompassing 21 studies with a total of 3044 patients with locally advanced unresectable or metastatic pancreatic cancer evaluated liposomal irinotecan-based second-line regimens. The results showed that liposomal irinotecan combined with 5-fluorouracil and leucovorin significantly improved progression-free survival, overall survival, and objective response rate. The regimen was associated with increased grade ≥3 neutropenia, anemia, hypokalemia, diarrhea, and vomiting.²⁸⁵ This study achieved therapeutic gains in advanced pancreatic cancer by modifying drug delivery and tumor exposure, but did not demonstrate inhibition of a specific stromal target. Similar to nab-paclitaxel, these results support a clear translational direction: in stroma-rich tumors with limited drug delivery, optimizing formulation and pharmacokinetic strategies may yield more reliable clinical benefits than targeting a single stromal component.

Clinical Results of PEGPH20 Targeting Hyaluronan

Unlike nanomedicines that improve drug delivery, PEGPH20 represents a strategy that directly modulates tumor stromal components. Hyaluronan is abundantly deposited in the pancreatic cancer stroma, where it increases interstitial pressure, compresses blood vessels, and limits the entry of chemotherapeutic drugs into the tumor tissue.¹¹⁵ Therefore, degrading hyaluronan with PEGPH20 can theoretically lower tumor interstitial pressure, improve vascular permeability, and enhance the delivery of drugs such as nab-paclitaxel and gemcitabine. The phase II HALO 202 study provided early positive signals for this strategy. It evaluated PEGPH20 combined with nab-paclitaxel and gemcitabine in previously untreated patients with metastatic pancreatic ductal adenocarcinoma. In patients with high hyaluronan expression, the PEGPH20 combination significantly improved progression-free survival (hazard ratio 0.51), with objective response rates of 45% vs. 31% and median overall survival of 11.5 months vs. 8.5 months, suggesting a favorable trend.²⁸⁶ These results indicate that stroma-directed therapies may require biomarker-based patient selection, and that patients with high hyaluronan levels are more likely to experience clinical benefit from improved drug delivery. However, the subsequent phase III HALO 109–301 study failed to replicate the phase II results. That study enrolled 494 patients with hyaluronan-high metastatic pancreatic ductal adenocarcinoma and compared PEGPH20 plus nab-paclitaxel/gemcitabine versus placebo plus the same chemotherapy regimen. Results showed median overall survival of 11.2 months vs. 11.5 months (hazard ratio 1.00) and median progression-free survival of 7.1 months for both groups (hazard ratio 0.97). Although the objective response rate was higher in the PEGPH20 arm (47% vs. 36%), this did not translate into a survival benefit, and the incidence of grade ≥3 fatigue, muscle spasms, and hyponatremia was increased.²⁸⁷

The transition from positive phase II signals to negative phase III results for PEGPH20 illustrates that modulating a single stromal component does not necessarily lead to long-term survival improvement. The tumor stroma is highly heterogeneous and capable of dynamic adaptation; even after hyaluronan reduction, improved drug delivery may still be limited by other stromal components, vascular architecture, immunosuppression, and tumor cell resistance mechanisms. Therefore, the failure of PEGPH20 does not completely negate stroma-directed therapies, but rather highlights the need for more refined patient selection, better combination regimens, and a more comprehensive understanding of the multifunctional nature of the tumor stroma.

Anti-IL-6R and Other CAF-Related Signaling Modulation Strategies

The results of PEGPH20 suggest that directly reducing a single stromal component may be insufficient to improve long-term outcomes. Thus, another approach to CAF-directed therapy is to modulate the inflammatory, drug-resistance, and drug-barrier signals in which CAFs participate. The IL-6/STAT3 pathway is a central axis for CAF-mediated inflammatory responses, tumor cell survival, and treatment resistance.²⁸⁸ Tocilizumab, an anti-IL-6 receptor antibody, has been explored to determine whether inhibiting IL-6 signaling can improve chemotherapy efficacy in pancreatic cancer. A phase I study enrolled 10 patients with gemcitabine/nab-paclitaxel-refractory metastatic pancreatic cancer to evaluate the safety and preliminary activity of tocilizumab combined with gemcitabine/nab-paclitaxel rechallenge. No dose-limiting toxicities were observed, and the recommended dose of tocilizumab was 8 mg/kg. Among the 10 patients, 4 showed tumor shrinkage during the first treatment cycle and were classified as responders.²⁸⁹ Paired biopsies further supported the biological rationale of this strategy. In responders, increased cleaved PARP expression in tumor cell nuclei, reduced proliferative CAFs, and increased intratumoral drug infiltration were observed. Decreased phosphorylated STAT3 expression correlated with increased drug infiltration, suggesting that IL-6/STAT3 pathway inhibition may improve the entry and action of chemotherapeutic drugs by attenuating CAF-related barrier functions. However, the small sample size of this study places it in the early exploratory stage, and it cannot be concluded that the anti-IL-6R strategy has definitive efficacy. Its value lies mainly in indicating that CAF-directed therapy can shift from simply altering stromal components to modulating key inflammatory, pro-resistance, and drug-delivery-limiting signals. Future validation will require larger studies that integrate CAF subtype analysis, IL-6/STAT3 activity measurements, drug infiltration assessments, and clinical outcomes.

Challenges and Future Perspectives

Despite rapid progress in CAF biology and nanomaterial engineering, the clinical translation of CAF-targeted nanomedicine remains constrained by delivery uncertainty, safety concerns, and incomplete functional stratification of CAF subsets. Future development should integrate material design with spatial biology, biomarker-driven patient selection, and computational optimization. This final chapter summarizes major translational challenges and outlines how artificial intelligence may help convert CAF heterogeneity from an obstacle into a design variable.

Delivery, Safety, and Biological Heterogeneity Challenges in CAF-Targeted Nanomedicine

After systemic administration, nanoparticles must pass through a series of biological barriers, including blood circulation, protein corona formation, clearance by the mononuclear phagocyte system, vascular extravasation, and interstitial diffusion. Non-target accumulation in organs such as the liver and spleen markedly reduces the effective dose that ultimately reaches tumors.¹⁷⁴ For CAF-targeted nanosystems, this challenge is even more pronounced. CAFs are often located deep within the stroma, around blood vessels, or at the tumor margin, and some CAF subsets are associated with collagen deposition, stromal stiffening, and elevated interstitial pressure. Consequently, even when nanoparticles successfully enter tumor tissue, they may still fail to reach the intended CAF subpopulations.²⁹⁰ Non-target accumulation not only weakens therapeutic efficacy but also increases safety risks. Poorly degradable materials may induce chronic inflammation, strongly charged nanoparticles may compromise hemocompatibility, and low-level accumulation after repeated dosing may amplify long-term toxicity.^291–293 In stroma-rich tumors such as pancreatic cancer, the limitations of the EPR effect further exacerbate delivery barriers. Compressed blood vessels, insufficient perfusion, and high interstitial pressure generated by the gel-like fluid phase of hyaluronic acid together create multiple transport obstacles, causing nanoparticles to accumulate predominantly in perivascular regions. Although PEGPH20 can degrade hyaluronic acid, its failure in a phase III clinical trial indicates that modulation of a single stromal component is unlikely to produce consistently improved clinical outcomes.²⁸⁷ Therefore, CAF-targeted nanomedicines require systematic evaluation of whole-body organ distribution, intratumoral spatial distribution, uptake by specific CAF subtypes, and payload release sites. At the same time, patient stratification, transvascular transport strategies such as iRGD-mediated transcytosis, and approaches that enhance diffusion within the stroma should also be incorporated into translational design.

The immunogenicity and biocompatibility of nanomaterials are also intertwined with CAF heterogeneity and plasticity, creating additional translational barriers. Modifications such as PEGylation and cationic lipids can improve circulation, but they may also induce anti-PEG antibodies, accelerated blood clearance, or complement activation.²⁹⁴ Anti-PEG antibodies have been reported to accelerate clearance of PEGylated nanomedicines through complement-related mechanisms, making this issue a recognized obstacle to clinical translation. CAF-targeting markers, such as FAP and α-SMA, are not absolutely specific and may also be expressed by normal reparative cells or pericytes; direct cytotoxic targeting may therefore damage beneficial cell populations.^295,296 A further challenge lies in the functional diversity of CAF subtypes within tumors. myCAFs, iCAFs, apCAFs, and populations such as LRRC15⁺ CAFs, CD10⁺GPR77⁺ CAFs, and senCAFs may play distinct roles in tumor promotion, immunosuppression, tumor restraint, or tissue homeostasis, whereas most current strategies still rely on pan-CAF markers and therefore cannot precisely distinguish among these functional states.^24,53 Although CAF reprogramming is a promising strategy, its long-term stability remains uncertain. Under TGF-β signaling, hypoxia, or therapeutic pressure, CAFs can undergo phenotypic switching; short-term reductions in α-SMA or IL-6 do not necessarily indicate stable reversal of the stromal ecosystem.¹² Phenotypic rebound after treatment withdrawal, enrichment of iCAFs after myCAF suppression, and compensatory tumor-cell growth through CAF-independent pathways may all occur.²⁹⁷ Therefore, safety and efficacy assessments should not rely solely on short-term tumor shrinkage or decreases in a limited set of markers. Instead, they should comprehensively evaluate CAF subtype composition, spatial organization, ECM status, immune-cell distribution, and global changes in the tumor transcriptome, while also establishing rigorous quality-control systems covering batch-to-batch consistency, protein corona formation, ligand stability, and related parameters. Only through such integrated evaluation can CAF-targeted nanomedicine progress from a mechanistic concept toward reliable clinical translation.

Future development of CAF-targeted nanomedicines should move beyond strategies that rely solely on single, broadly expressed pan-CAF markers, such as FAP, α-SMA, and PDGFR, because these markers are insufficient to distinguish CAF subpopulations with divergent therapeutic functions. Instead, multi-marker profiling should be integrated with functional phenotypic characterization to precisely identify tumor-promoting CAFs, tumor-suppressive CAFs, and immune-supportive CAFs, thereby avoiding indiscriminate CAF depletion and improving targeting specificity.¹² In this context, multi-omics integration is essential for translating CAF biology into rational nanocarrier design.²⁹⁸ Single-cell sequencing enables the delineation of CAF subpopulations and the identification of subtype-specific transcriptional programs, including those associated with matrix remodeling, immunosuppression, cytokine secretion, and therapeutic resistance, thereby facilitating the selection of CAF subsets that are truly actionable for intervention.²⁹⁹ Spatial transcriptomics further defines the localization of these CAF subsets relative to tumor nests, blood vessels, immune cells, and dense stromal regions, providing spatial guidance for nanocarrier engineering, such as perivascular retention, deep stromal penetration, ECM remodeling, or localized immune activation.^300,301 Proteomic analysis can validate whether transcriptomic targets are translated into actual protein expression and help prioritize membrane proteins, secreted factors, and ECM-associated molecules that are more suitable for ligand modification, antibody conjugation, or stimuli-responsive payload release.^302,303 Metabolomic profiling can reveal CAF-associated metabolic features, such as lactate accumulation, lipid remodeling, amino acid metabolic reprogramming, and hypoxia adaptation, which may inform the design of metabolism-responsive nanocarriers or combination strategies aimed at reversing CAF-mediated immunosuppression and therapeutic resistance.³⁰⁴ In addition, imaging technologies, including FAP-based molecular imaging, spatial imaging, and multimodal nanoparticle tracking, can be used to evaluate stromal target abundance, intratumoral distribution, drug penetration, and therapy-induced stromal remodeling, thereby supporting patient stratification and dynamic monitoring of therapeutic response.^305,306 At the level of material design, CAF-targeted nanosystems should prioritize biodegradable and highly biocompatible components; optimize particle size and surface modification to balance stromal penetration with CAF selectivity; and minimize nonspecific protein adsorption and off-target accumulation.^{208,307–309} Collectively, these strategies are expected to enhance the precision, safety, and clinical translational potential of CAF-targeted nanomedicines for solid tumor therapy.

Artificial Intelligence-Guided Design and Precision Optimization of CAF-Targeted Nanomedicine

In recent years, the deep integration of artificial intelligence (AI) with nanomedicine has profoundly reshaped the research and development paradigm of nanotherapeutics and has provided new technical tools for CAF-targeted nanotherapy. Conventional nanocarrier design relies heavily on iterative experimental optimization. However, when confronted with CAF heterogeneity, the structural complexity of ECM barriers, and the multifactorial coupling of nano–bio interactions, traditional approaches are often inefficient and insufficient for truly precision-guided design. By integrating multimodal data, constructing predictive models, and accelerating design iteration, AI has the potential to systematically address key bottlenecks across target identification, nanocarrier optimization, and patient stratification. By integrating target identification, molecular docking, QSAR modeling, PBPK simulation, pharmacokinetic prediction, and iterative nanocarrier optimization, AI-assisted workflows may accelerate the rational design of CAF-targeted nanomedicine (Figure 6).

Figure 6.

Artificial intelligence-assisted PBPK and QSAR modeling for nanomedicine optimization. The workflow illustrates how nanotumor databases, nanoparticle physicochemical properties, tumor-treatment variables, random forest models, deep neural networks, QSAR modeling, and PBPK simulation can be integrated to predict tissue biodistribution and tumor-delivery efficiency of nanomedicines. This computational framework supports data-driven optimization of nanocarrier design, dosing strategy, and translational evaluation. Adapted from,³¹⁰ © 2025 The Author(s).

Artificial intelligence can accelerate the development of CAF-targeted therapies by converting nanocarrier design from empirical optimization into a predictive and biologically informed process. At the nano–bio interface, machine learning enables the prediction of protein corona formation, an early biological event that shapes circulation time, immune recognition, targeting efficiency, and ultimately the delivery performance of nanomedicines. For example, Canchola et al established the Protein Corona Database (PC-DB) and used interpretable models such as LightGBM and XGBoost to identify particle size, zeta potential, and incubation time as key determinants of protein corona composition.³¹¹ Liu et al further emphasized that AI-assisted protein corona analysis can help decode complex nano–bio interactions and improve nanomedicine safety and efficacy.³¹² In this way, AI provides a rational basis for selecting nanoparticle physicochemical properties that are more likely to support stable systemic circulation and effective delivery to CAF-rich tumor microenvironments.

Beyond early nano–bio interactions, AI further supports CAF-targeted therapy by predicting in vivo nanoparticle fate and integrating CAF spatial heterogeneity into therapeutic design. Based on a nano–tumor database, Mi et al trained a deep neural network model to predict nanoparticle biodistribution and tumor delivery efficiency, supporting its use as a high-throughput prescreening tool for nanomedicine formulation design.³¹³ In a separate study, an ensemble learning model optimized by an adaptive tree-structured Parzen estimator, namely AdaBoost KNN, provided another computational strategy for predicting nanomedicine biodistribution across multiple organ targets.³¹⁴ More directly, Pan et al developed NanoNet, which incorporates FAP immunostaining to predict nanoparticle distribution at pixel-level resolution, indicating that the spatial organization of CAFs can serve as an important determinant of nanocarrier behavior.³¹⁵ Meanwhile, AI-driven multi-omics and spatial analyses are helping to define functionally distinct CAF subtypes and their clinical relevance. Zheng et al identified ERS-CAFs as an immunoregulatory CAF population associated with immune exclusion and therapeutic resistance.³¹⁶ while another study in head and neck squamous cell carcinoma used AI-based multi-omics integration to construct a CAF-related prognostic signature and screen candidate therapeutic agents.³¹⁷ Together, these advances suggest that AI can promote CAF-targeted therapy by linking nanocarrier optimization, biodistribution prediction, CAF spatial mapping, and CAF subtype stratification, thereby supporting the transition from broadly CAF-targeted approaches toward subtype-specific and spatially precise therapeutic modulation.

Conclusions

CAFs link tumor biology, stromal mechanics, immune suppression, and nanomedicine delivery into a single regulatory network. The central message of this review is that CAF targeting should shift from broad depletion toward subtype-aware and spatially controlled modulation. By integrating CAF functional heterogeneity with nanocarrier design, future strategies may selectively weaken tumor-promoting stromal programs, preserve beneficial stromal restraint, enhance deep tumor penetration, and improve the efficacy of chemotherapy, immunotherapy, radiotherapy, and cell-based therapies. For clinical translation, however, several key challenges remain, including the lack of robust biomarkers for patient stratification, the difficulty of distinguishing tumor-promoting from tumor-restraining CAF states in individual patients, and the need to balance stromal remodeling with safety. In addition, scalable manufacturing, reproducible quality control, long-term toxicity evaluation, and rational clinical trial design will be essential for determining which CAF-targeted nanomedicines can provide meaningful therapeutic benefit. Such an approach is essential for advancing CAF-targeted nanomedicine from promising preclinical concepts toward safer and more clinically translatable treatment strategies for solid tumors.

Funding Statement

This work was supported by the Special Scientific Research Fund of the First Affiliated Hospital of Chengdu Medical College [grant numbers CYFY2021YB07], the Key project of Chengdu Medical College-The First Affiliated Hospital of Traditional Chinese Medicine of chengdu Medical college Joint Scientific Research Fund [grant numbers 2022LHZYYB-16] and the Scientific Research Project of Chengdu Medical College [grant numbers CYZYB25-04].

Abbreviations

ACTA2, actin alpha 2, smooth muscle; AI, artificial intelligence; α-SMA, α-smooth muscle actin; apCAFs, antigen-presenting cancer-associated fibroblasts; CAFs, cancer-associated fibroblasts; CAR-T, chimeric antigen receptor T cell; CAV1, caveolin 1; CCL, C-C motif chemokine ligand; CD, cluster of differentiation; CD8⁺ T cells, CD8-positive T cells; c-MET/MET, MET proto-oncogene receptor tyrosine kinase; COL1A1, collagen type I alpha 1 chain; CSCs, cancer stem cells; CXCL, C-X-C motif chemokine ligand; CXCL12/SDF-1, C-X-C motif chemokine ligand 12/stromal cell-derived factor 1; CXCR4, C-X-C chemokine receptor 4; DCs, dendritic cells; Doxil, pegylated liposomal doxorubicin; EDB-FN, extra-domain B fibronectin; ECM, extracellular matrix; EGFR, epidermal growth factor receptor; EMT, epithelial–mesenchymal transition; EndMT, endothelial–mesenchymal transition; EPR, enhanced permeability and retention; EVs, extracellular vesicles; FAP, fibroblast activation protein; FAPI, fibroblast activation protein inhibitor; FSP1/S100A4, fibroblast-specific protein 1/S100 calcium-binding protein A4; HA, hyaluronic acid/hyaluronan; HCC, hepatocellular carcinoma; HGF, hepatocyte growth factor; HIF-1α, hypoxia-inducible factor 1 alpha; HLA, human leukocyte antigen; iCAFs, inflammatory cancer-associated fibroblasts; ICB, immune checkpoint blockade; ICIs, immune checkpoint inhibitors; IDO, indoleamine 2,3-dioxygenase; IFN-γ, interferon gamma; IGF, insulin-like growth factor; IGF-1R, insulin-like growth factor 1 receptor; IL, interleukin; JAK, Janus kinase; LIF, leukemia inhibitory factor; LIFR, leukemia inhibitory factor receptor; LNPs, lipid nanoparticles; LOX, lysyl oxidase; LRRC15, leucine-rich repeat-containing 15; MAPK, mitogen-activated protein kinase; MDSCs, myeloid-derived suppressor cells; MHC, major histocompatibility complex; MMPs, matrix metalloproteinases; mRNA, messenger RNA; myCAFs, myofibroblastic cancer-associated fibroblasts; NF-κB, nuclear factor kappa B; NK cells, natural killer cells; NPs, nanoparticles; NSCLC, non-small cell lung cancer; ORR, objective response rate; PBPK, physiologically based pharmacokinetic; PD-1, programmed cell death protein 1; PD-L1, programmed death-ligand 1; PDAC, pancreatic ductal adenocarcinoma; PDGF, platelet-derived growth factor; PDGFR, platelet-derived growth factor receptor; PDPN, podoplanin; PDT, photodynamic therapy; PEG, polyethylene glycol; PET, positron emission tomography; PFS, progression-free survival; PGE2, prostaglandin E2; PI3K, phosphoinositide 3-kinase; PLGA, poly(lactic-co-glycolic acid); PMID, PubMed identifier; PTT, photothermal therapy; QSAR, quantitative structure–activity relationship; RECIST, Response Evaluation Criteria in Solid Tumors; RNA, ribonucleic acid; ROS, reactive oxygen species; RT, radiotherapy; RTK, receptor tyrosine kinase; SASP, senescence-associated secretory phenotype; scFv, single-chain variable fragment; siRNA, small interfering RNA; SMAD, suppressor of mothers against decapentaplegic; STAT3, signal transducer and activator of transcription 3; STING, stimulator of interferon genes; TACE, transarterial chemoembolization; TAMs, tumor-associated macrophages; TCGA, The Cancer Genome Atlas; TGF-β, transforming growth factor beta; TILs, tumor-infiltrating lymphocytes; TLS, tertiary lymphoid structure; TME, tumor microenvironment; TNBC, triple-negative breast cancer; TNF-α, tumor necrosis factor alpha; Tregs, regulatory T cells; VEGF, vascular endothelial growth factor; YAP/TAZ, Yes-associated protein/transcriptional coactivator with PDZ-binding motif.

Data Sharing Statement

No new data has been generated, all references are cited in the manuscript.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Disclosure

The authors declare that there are no competing interests associated with the manuscript.