Cancer-associated fibroblasts as a target in head and neck squamous cell carcinoma: bridging mechanisms to therapy with drugs, adjuvants, and nanotechnology

Abstract

The tumor microenvironment (TME) in head and neck squamous cell carcinoma (HNSCC) comprises a heterogeneous cellular landscape in which cancer cells, stromal cells, immune cells, and their secreted mediators collectively sustain a dynamic ecosystem. Cancer-associated fibroblasts (CAFs) and the extracellular matrix (ECM) cooperatively shape a desmoplastic stroma that drives resistance mechanisms and facilitates tumor metastasis. As central orchestrators of stromal remodeling, CAFs release cytokines, growth factors, and ECM-modifying enzymes that structurally and functionally alter the tumor niche. Concurrently, aberrantly crosslinked ECM components generate a mechanically rigid stroma that impedes drug penetration and restricts immune cell infiltration. This CAF-ECM axis not only sustains tumor progression but actively engages multiple resistance pathways that limit the efficacy of conventional therapies. A systematic literature search was conducted to identify studies reporting detailed mechanistic findings and translational relevance. Selective inhibitors targeting CAF-derived signaling pathways have shown capacity to suppress HNSCC progression, while ECM-directed interventions appear to improve drug delivery and attenuate tumor advancement. Engineered nanomaterials, evaluated in preclinical models, offer an additional layer of therapeutic precision by enhancing drug distribution within the stroma. Anti-fibrotic agents and immunomodulators have shown potential to disrupt CAF-ECM crosstalk and restore tumor sensitivity to standard anticancer regimens. Collectively, preclinical evidence supports the feasibility of stroma-targeted strategies in HNSCC and points toward combinatorial modalities and stromal reprogramming as clinically meaningful directions for future investigation.

Introduction

Head and neck squamous cell carcinoma (HNSCC) represents a substantial global health burden, accounting for approximately 4% of all cancers worldwide. Annual incidence exceeds 800,000 cases, with nearly 450,000 deaths attributed to these malignancies [1]. The disease disproportionately affects low- and middle-income regions, where tobacco use, alcohol consumption, and betel nut chewing remain the principal risk factors. High-income countries, by contrast, have witnessed a rise in human papillomavirus (HPV)-related oropharyngeal cancers, reflecting a notable shift in epidemiological patterns. Given its predilection for critical anatomical structures, HNSCC carries considerable morbidity, frequently producing functional impairments in speech, swallowing, and breathing [2, 3]. Late-stage diagnosis remains the norm rather than the exception, with more than 60% of patients presenting with advanced disease, a reality that continues to weigh heavily on survival outcomes [1].

Current management relies on surgery, radiotherapy, and chemotherapy, deployed individually or in combination to maximize therapeutic effect [4]. Advances in surgical precision and radiotherapy planning have reduced treatment-related complications, yet recurrence rates remain persistently high. Immune checkpoint inhibitors (ICIs) have emerged as a meaningful option for recurrent or metastatic disease; however, durable responses are confined to a subset of patients [5]. Molecular targeted therapy offers an additional avenue, though its clinical reach remains constrained [6]. Compounding these limitations, aggressive treatment regimens frequently produce debilitating sequelae, including chronic pain and tissue injury, that erode quality of life [7]. Collectively, these shortcomings underscore the pressing need for therapeutic strategies that improve efficacy while preserving healthy tissue integrity.

Increasing attention has turned to the tumor microenvironment (TME), and the stromal compartment in particular, as a central driver of HNSCC progression and treatment resistance. Across solid tumors, stromal cells and the extracellular matrix (ECM) establish a protective niche that sustains tumor growth, suppresses immune activity, and obstructs drug delivery [8]. Cancer-associated fibroblasts (CAFs), the dominant stromal cell population, actively remodel the ECM into a dense fibrotic network that physically shields tumor cells and promotes therapeutic evasion. This stromal barrier not only supports cancer cell survival but adapts dynamically to therapeutic pressure, thereby enabling disease recurrence through coordinated effects on immune and cancer cell signaling [9]. Targeting stromal components alongside cancer cells has consequently gained traction as a strategy to disrupt tumor-stroma interactions and improve treatment outcomes.

Several recent reviews have addressed CAF biology and the broader tumor stroma in solid tumors, with more limited coverage of HNSCC specifically. Although this review encompasses the full spectrum of HNSCC subsites, the majority of studies examined here investigated CAF interactions and therapeutic interventions in oral squamous cell carcinoma (OSCC). An important point of framing deserves emphasis: most CAF- and ECM-directed strategies in HNSCC remain at the preclinical stage, and clinical validation is still limited. This review therefore does not position stromal targeting as an established standard of care. Rather, these approaches are framed as mechanism-guided therapeutic opportunities whose value lies in identifying tractable stromal dependencies that may improve drug delivery, immune infiltration, or treatment sensitivity when integrated with existing modalities.

The aim of this review is to add translational value by linking mechanistic CAF programs to specific therapeutic rationales, clarifying why particular drug classes, adjuvants, and nanotechnology platforms may succeed or encounter limitations in HNSCC. We summarize the functional consequences of CAF activity and their secreted mediators, including high transforming growth factor-β (TGF-β), interleukin-6 (IL-6) and chemokine-rich secretomes, and ECM-remodeling outputs, within the HNSCC stroma. CAF markers and potential therapeutic interventions targeting these distinct functional effects are reviewed in turn. CAF biology in HNSCC is organized into recurring functional modules, each connected to measurable stromal barriers that directly impair drug transport and immune infiltration. These barriers are then mapped to rational therapeutic strategies, encompassing cytokine and chemokine blockade, CAF normalization approaches, ECM-modifying agents, and stimulus-responsive nanocarriers, including pH-, enzyme-, and hypoxia-responsive platforms, with particular attention to how each may be combined with radiotherapy and immune checkpoint inhibition in HNSCC.

Search strategy

A focused literature search was conducted across PubMed/MEDLINE, Embase, Scopus, Web of Science, and Google Scholar to identify studies relevant to CAF biology and stromal targeting in HNSCC. The search prioritized publications from the past 10 years, though seminal older studies were incorporated where necessary to establish foundational mechanistic context. Evidence from non-HNSCC squamous carcinomas was considered only when it supported the mechanistic plausibility of CAF and ECM biology.

Controlled vocabulary terms, including Medical Subject Headings (MeSH), were combined with free-text queries. Core search terms encompassed cancer-associated fibroblast, CAF, tumor-associated fibroblast, HNSCC, head and neck squamous cell carcinoma, and OSCC, among related descriptors. To capture mechanistic and therapeutic dimensions relevant to this review, additional terms were incorporated: TGF-β, IL-6, C-C motif chemokine ligand (CCL) 2, CCL5, CCL22, microRNA (miR)−21, signal transducer and activator of transcription 3 (STAT3), phosphoinositide 3-kinase (PI3K), Akt, SRY-box transcription factor (Sox), mitogen-activated protein kinase (MAPK), programmed death ligand 1 (PD-L1), epithelial-mesenchymal transition (EMT), ECM, hyaluronan, collagen, matrix metalloproteinase (MMP)−2, hypoxia, metabolism, immunosuppression, nanoparticle, small interfering RNA (siRNA), hyaluronidase, photothermal, and pH-sensitive. No language restrictions were applied during retrieval; however, eligibility assessment relied on full-text evaluation in all cases.

Following duplicate removal, a two-stage screening process was employed. Titles and abstracts were first reviewed to exclude clearly irrelevant records, after which full texts were assessed against predefined inclusion and exclusion criteria. Discrepancies were resolved through discussion. Eligible study types included in vitro and in vivo experiments, patient-sample analyses, clinical trials, and substantive reviews. Inclusion required that articles directly address CAF biology, CAF-immune crosstalk, ECM remodeling, or CAF-driven oncogenic signaling in HNSCC. Editorials lacking primary data, case reports without mechanistic content, and non-squamous tumors without demonstrable CAF relevance were excluded. For each eligible study, data extraction covered study design, tumor subsite, CAF definitions, and key mediators such as cytokines and growth factors. Downstream signaling pathways, immune effects, ECM metrics, and tested interventions were also captured to support thematic synthesis and translational mapping. Immune variables of interest included PD-L1 expression, CD8 + T cell exhaustion, and immune cell recruitment patterns. Studies linking stromal biology to therapeutic testing were prioritized, though each was evaluated within its experimental context. Mechanistic studies, organoid models, and animal experiments were treated as hypothesis-generating evidence, and clinical inferences were drawn with appropriate caution. Findings with translational features, including patient-derived tumor models, organotypic systems, and combinatorial treatment designs, were given particular attention. It should be noted that evidence supporting CAF normalization, ECM modulation, and stromal drug delivery in HNSCC remains largely preclinical and limited in scope.

Current treatments for HNSCC: limitations and potential of novel approaches

HNSCC management encompasses surgery, radiotherapy, chemotherapy, immunotherapy, and targeted molecular therapy, with standard regimens remaining stage- and subsite-dependent. Despite meaningful advances across these modalities, tumor control remains suboptimal, undermined by tumor heterogeneity, a complex stromal architecture, and acquired resistance that emerges during treatment. The intricate anatomy and physiology of the head and neck region further constrain therapeutic response [10]. Continued progress in novel drug development, nanotechnology, and adjuvant strategies offers realistic prospects for addressing several of these barriers.

Radiotherapy serves as a primary, adjuvant, or palliative modality in HNSCC, and technological refinements now permit more precise dose delivery conforming to tumor contours, thereby reducing exposure to adjacent structures such as the salivary glands and spinal cord [11]. Stromal fibrosis and hypoxia, however, persistently compromise radiation efficacy, and treatment-related toxicities, including mucositis and xerostomia, remain a meaningful burden on patient quality of life [12, 13].

Systemic chemotherapy has long been central to HNSCC management, whether as induction therapy to reduce tumor burden, concurrently with radiotherapy to augment radiosensitivity, or as palliation in metastatic disease [14, 15]. Platinum-based regimens, cisplatin in particular, constitute the chemotherapeutic backbone, though dose-limiting nephrotoxicity and ototoxicity frequently necessitate substitution with carboplatin or taxanes such as paclitaxel in vulnerable patients [16]. Conventional cytotoxic agents also carry a substantial side-effect profile, encompassing mucositis and gastrointestinal toxicity [17]. Targeted molecular therapies, led by cetuximab, have expanded the treatment armamentarium and demonstrate greater utility when combined with radiotherapy in locally advanced disease [18], although their incremental benefit over chemoradiotherapy remains a matter of ongoing debate [19, 20]. Epidermal growth factor receptor (EGFR) inhibitors such as afatinib and angiogenesis inhibitors such as bevacizumab produce modest responses in HNSCC, with efficacy further limited by stromal-mediated drug resistance and the activation of compensatory signaling pathways [21].

ICIs represent an established therapeutic option in recurrent or metastatic HNSCC, functioning by reinvigorating exhausted cytotoxic CD8 + T lymphocytes (CTLs) through blockade of inhibitory programmed cell death protein 1 (PD-1)/PD-L1 interactions [22]. FDA-approved indications vary by agent and line of therapy. Nivolumab is approved for recurrent or metastatic HNSCC with disease progression on or after platinum-based therapy, without a PD-L1 expression requirement. Pembrolizumab is approved for first-line treatment of metastatic or unresectable recurrent HNSCC on the basis of the KEYNOTE-048 trial; monotherapy is indicated for PD-L1-expressing tumors selected by combined positive score (CPS), whereas pembrolizumab combined with platinum and 5-fluorouracil is approved as a first-line regimen regardless of PD-L1 status [23]. The approved indications have since extended beyond recurrent or metastatic disease: pembrolizumab is now approved as a perioperative regimen for adults with resectable locally advanced HNSCC whose tumors express PD-L1 (CPS ≥ 1) as determined by an FDA-approved test. This regimen comprises neoadjuvant pembrolizumab monotherapy, followed by adjuvant pembrolizumab combined with postoperative radiotherapy with or without cisplatin, and completed by pembrolizumab monotherapy [24]. Combinatorial approaches pairing ICIs with chemotherapy, radiotherapy, or stromal-depleting agents show promise in preclinical models through coordinated stromal remodeling and enhanced CTL infiltration. Nonetheless, tumor recurrence, limited response in a subset of patients, and the absence of reliable predictive biomarkers collectively underscore the complexity of immune-stromal crosstalk [25]. Deeper mechanistic understanding of the tumor stroma in HNSCC is therefore essential for developing more effective and rationally designed therapeutic strategies.

As noted above, all treatment modalities carry inherent limitations. Preclinical models have increasingly pointed to nanotechnology as a means of overcoming the biological and physical barriers that constrain HNSCC treatment. Nanoparticles are being engineered to address stromal obstacles characteristic of HNSCC, including elevated interstitial pressure, collagen and hyaluronan density, hypoxia, low pH, and immune exclusion. Through improvements in intratumoral deposition, controlled drug release, and cell-selective uptake, nanoparticles exploit the enhanced permeability and retention (EPR) effect to passively accumulate within tumors, facilitating antitumor drug delivery into the stroma [26]. Surface conjugation with targeting ligands such as folate or EGFR antibodies further enables selective binding to overexpressed receptors on cancer or stromal cells. Nanoparticles have also been explored as radiosensitizing agents [27]; gold nanoparticle-mediated radiosensitization has been demonstrated in vitro and in vivo in HNSCC models, including combinatorial strategies with cetuximab and radiotherapy [28]. These nano-radioenhancer approaches remain investigational and have not been established as approved radiosensitizers in HNSCC, though functionalized variants may offer additional efficacy gains [29]. Beyond radiosensitization, liposomes, polymeric nanoparticles, micelles, and silicon-based carriers have been proposed as delivery vehicles for chemotherapeutics, ICIs, tyrosine kinase inhibitors (TKIs), and agents used in photothermal or photodynamic therapy [30, 31], as well as for improving the bioavailability of naturally derived antitumor compounds [32].

A critical translational limitation tempers this promise. Although nanoparticles are frequently described as leveraging the EPR effect, tumor delivery in humans is typically low and highly heterogeneous, and passive accumulation cannot be assumed to be consistent across patients or tumor types. A meta-analysis estimated median tumor delivery at approximately 0.7% of the injected dose, underscoring the magnitude of this gap [33]. In HNSCC, CAF-driven desmoplasia compounds this problem through elevated interstitial fluid pressure (IFP), collagen alignment and crosslinking, and hyaluronan-rich matrices that restrict extravasation and intratumoral penetration [34]. Clinically realistic nanomedicine strategies for HNSCC must therefore incorporate quantifiable delivery endpoints, including tumor uptake, penetration depth, spatial drug distribution, and tumor-to-blood ratios. Barrier-modulating combinations that reduce IFP or normalize ECM architecture, achieved through focused ultrasound, enzymatic digestion, or pharmacologic ECM remodeling, represent a parallel priority [35]. Patient stratification supported by companion imaging to identify tumors with permissive perfusion and permeability profiles adds another layer of translational rigor.

Mechanistically, the most compelling rationale for CAF- and ECM-focused nanomedicine centers on barrier modulation combined with targeted payload delivery, encompassing pH-responsive release within acidic stroma, MMP-responsive systems operating in protease-rich ECM, and enzyme-mediated matrix normalization to improve drug penetration. These strategies may be further integrated with immunotherapy or cytotoxic regimens to convert improved stromal access into measurable clinical benefit [36]. The path from mechanism to therapy is consequently indirect rather than linear: mechanistic data identify candidate vulnerabilities, preclinical interventions test whether those vulnerabilities are actionable, and clinical translation ultimately depends on safety profiles, patient selection criteria, delivery feasibility, and biomarker support. This staged logic is of particular importance for CAF-directed therapy, given that CAF states are heterogeneous and dynamic. A target that confers benefit in one stromal context may prove ineffective or detrimental in another [37].

Building on these advances, emerging strategies have proposed that durable tumor control in HNSCC requires simultaneous engagement of tumor-stroma biology. The HNSCC stroma is characterized by CAF-driven immune exclusion, TGF-β-dominant signaling, the C-X-C motif chemokine ligand (CXCL)12–CXC chemokine receptor (CXCR)4 chemokine axis, and ECM-mediated barriers to drug and immune cell access [38]. These properties provide the mechanistic basis for rational combinatorial strategies that target malignant cells and stromal programs in parallel.

Several investigational approaches attempt to overcome CAF- and ECM-driven immune exclusion and treatment resistance by pairing ICIs with agents directed at stromal signaling or immune trafficking. Bifunctional constructs intersecting the TGF-β axis represent one such direction; bintrafusp alfa, a fusion of PD-L1 blockade with a TGF-β trap, has been evaluated in HNSCC contexts encompassing radiation-based and immunotherapy-based regimens [39]. Chemokine axis intervention offers a complementary strategy: the CXCL12–CXCR4 pathway constitutes a mechanistically defined CAF-linked route for T-cell exclusion, and CXCR4 blockade is under broad investigation as an immunotherapy partner aimed at increasing intratumoral immune infiltration [40]. Inflammatory cytokine targeting, particularly IL-6 and IL-6 receptor (IL-6R) blockade, is being explored as a means to attenuate CAF-associated STAT3 activity and remodel the immune-stromal milieu, with the additional potential to augment checkpoint inhibition in HNSCC-relevant settings [41].

Therapeutic targeting of CAFs and the ECM should be regarded as a biologically rational but still-developing strategy. Current evidence supports mechanistic plausibility more convincingly than demonstrated clinical efficacy, and stromal interventions are therefore most likely to yield benefit as adjuncts to established therapies rather than as stand-alone modalities. The functional diversity of CAFs across different stromal contexts underlies this constraint; complete stromal ablation is neither feasible nor desirable. Functional reprogramming, selective pathway inhibition, and context-specific combination therapy represent more realistic and translatable goals [42].

HNSCC stroma

The tumor stroma in HNSCC constitutes a complex network of non-malignant cells and extracellular components that collectively govern tumor behavior. Comprising fibroblasts, immune cells, blood vessels, and ECM molecules, this stromal compartment functions as both a physical scaffold and a biochemical signaling hub [43]. Stromal elements are far from passive bystanders; they actively regulate tumor growth, invasion, and therapeutic resistance. In HNSCC, the stroma undergoes pathological remodeling characterized by excessive fibrosis, aberrant angiogenesis, and chronic inflammation. This altered microenvironment sustains cancer cell survival while dynamically interacting with therapeutic agents in ways that frequently blunt their efficacy. The dual role of the stroma, as both a promoter of malignancy and a barrier to treatment, underscores its significance as a therapeutic target [44].

CAFs and the remodeled ECM occupy a central position in stromal dysfunction. CAFs arise from activated resident fibroblasts or transdifferentiated cell types under the influence of tumor-derived signals, including TGF-β and inflammatory cytokines. Stroma-derived mediators, among them CAF-secreted prostaglandin E2 (PGE2), TGF-β, and ECM-bound chemokines, disrupt immune cell activation and recruitment, establishing an immunosuppressive milieu that enables tumors to evade immune surveillance. Beyond immune modulation, CAFs secrete growth factors, proteases, and ECM proteins that foster a stiff, desmoplastic matrix enriched in collagens, fibronectin, and hyaluronan [45]. The ECM in HNSCC becomes structurally disorganized, generating a hypoxic and acidic microenvironment in which elevated tumor stiffness plays a central mechanistic role. A rigid ECM limits the penetration of oxygen and glucose, alters the metabolic programs of both cancer cells and CAFs, and triggers glycolytic reprogramming with consequent lactate accumulation [46]. CAFs and the ECM are engaged in reciprocal signaling: CAFs drive ECM crosslinking and fiber alignment, while matrix stiffness in turn amplifies the pro-tumorigenic functions of CAFs. This feedback loop establishes a self-reinforcing stromal ecosystem that sustains tumor progression and consolidates therapeutic resistance [47].

The HNSCC stroma also exerts broad suppressive effects on antitumor immunity. Natural killer (NK) cells and cytotoxic CD8⁺ T lymphocytes (CTLs) are critical effectors of tumor elimination, yet both populations are frequently exhausted in CAF-rich tumors [48]. M1-polarized tumor-associated macrophages (TAMs) and N1 tumor-associated neutrophils (TANs), which exhibit tumoricidal activity, are typically outnumbered by their immunosuppressive counterparts, M2 TAMs and N2 TANs. While M1 macrophages and N1 TANs facilitate antigen presentation and CD8 + T lymphocyte activation, their immunosuppressive counterparts impair these processes and restrict the expansion of NK cells and CTLs through the release of IL-4, IL-10, IL-13, and TGF-β [49, 50]. T helper (Th) cell subsets are similarly shaped by the stromal environment: Th1 CD4 + T cells reinforce CTL activity through pro-inflammatory cytokine release, whereas Th2 cells counteract this response via IL-4 and IL-13 secretion [51]. Regulatory T cells (Tregs), Th22 cells, and Th17 cells further promote immune evasion through multiple cooperative interactions with CAFs and other immunosuppressive populations, mediated in part by TGF-β and related inhibitory cytokines [52]. Myeloid-derived suppressor cells (MDSCs) compound this suppressive landscape by dampening CTL activity and secreting inhibitory mediators such as arginase-1 [22]. (Fig. 1)

Fig. 1.

Schematic of the tumor stroma in HNSCC. Collagen and fibronectin are major ECM components that promote tumor expansion. CAFs recruit and activate other immunosuppressive cells, which can exhaust NK cells and CD8 + T cells by releasing various inhibitory factors. (CAF: cancer-associated fibroblast; DC: dendritic cell; M1: M1 macrophage; M2: M2 macrophage; NK cell: natural killer cell; PD-L1: programmed death ligand 1; TGF-β: transforming growth factor-β; Th: T helper; Treg: regulatory T cell. Created with BioRender.)

CAF states in HNSCC

CAFs in HNSCC do not constitute a single functional entity. Spatial and single-cell transcriptomic analyses have identified multiple CAF populations with distinct immune and ECM consequences. Spatial context reveals that certain CAF programs co-localize with immune-desert tumor cores and correlate with reduced CD8 + T cell infiltration and diminished antitumor activity, providing a mechanistic basis for CAF-associated immune exclusion that extends beyond bulk stromal abundance [53].

Across HNSCC datasets, CAF programs exhibit distinct functional properties. ECM-producing and myofibroblastic CAFs represent a major class driving collagen-rich, mechanically stiffened matrices [54]. A separate inflammatory CAF population is enriched for IL-6 and chemokine signaling, coupled to STAT3-dependent tumor promotion and myeloid cell recruitment. Another CAF class associates with immune-structured niches characteristic of immune-hot microenvironments and may improve responsiveness to immunotherapy. Antigen-presenting-like CAFs, defined by major histocompatibility complex (MHC)-II gene expression, are capable of activating CD8 + T cells to detect and eliminate cancer cells.

These state definitions provide an HNSCC-specific framework for mapping CAF biology to therapeutic strategies. Tumors enriched in ECM and myofibroblastic programs may be susceptible to targeting of lysyl oxidase (LOX) and integrin–focal adhesion kinase (FAK) signaling, as well as Yes-associated protein (YAP)–transcriptional coactivator with PDZ-binding motif (TAZ) mechanotransduction. Inflammatory CAFs, by contrast, align with IL-6/Janus kinase (JAK)/STAT3 blockade and chemokine-axis inhibition, including the CXCL12–CXCR4 pathway, to facilitate immune infiltration and improve ICI efficacy [55]. Table 1 summarizes CAF subtypes in HNSCC, their core functions and intercellular interactions, released or expressed mediators, and potential therapeutic targets. (Table 1)

Table 1.

CAF states in HNSCC and potential targets for effective therapy [55]

CAF program/state (HNSCC)	Reproducible marker themes (examples)	Core functions	Immune/therapy links	Actionable nodes/intervention classes
Myofibroblastic/ECM-remodeling CAFs (myCAF/ECM-CAF-like)	Collagen-rich ECM genes; contractile features	ECM deposition, stiffness, barrier function	Physical immune exclusion; limits drug penetration; promotes hypoxia	LOX (crosslinking), integrin–FAK, ROCK, TGF-β
Inflammatory CAFs (iCAF-like)	IL6/chemokine programs; STAT3-linked	Cytokine/chemokine secretion; inflammatory milieu	Drives MDSC/TAM recruitment; supports PD-L1 upregulation via IL-6/JAK/STAT	IL-6R, JAK/STAT3, NF-κB modulation
Immune-niche/lymphoid-structured fibroblasts (FRC-like)	Chemokines/lymphoid organization programs (e.g., CCL19-like signatures)	Stromal support of immune aggregates/TLS-like niches	Associated with immune-hot neighborhoods and potentially better ICI responsiveness	Non-canonical NF-κB/lymphotoxin pathway concepts (context-dependent); combine with ICI
Antigen-presenting–like CAFs (apCAF-like)	MHC-II antigen presentation genes (HLA-DRA/DRB, CD74)	Antigen-presentation-like signaling; immune modulation	Can tune T-cell phenotypes (context-dependent across cancers)	Investigational; potential for combination with ICI and TGF-β modulation

DRA/DRB: chain paralogues in HLA; FRC: fibroblastic reticular cell; HLA: histocompatibility leukocyte antigen; NF-κB: nuclear factor of κB; ROCK: Rho-associated protein kinase; TLS: tertiary lymphoid structure

Interactions between CAFs and HNSCC cells: potential targets and modulators

CAFs and HNSCC cells communicate through a complex exchange of soluble factors and extracellular vesicles. CAFs release IL-6, TGF-β, and epidermal growth factor (EGF), which activate multiple signaling pathways in tumor cells. TGF-β, for instance, triggers the PI3K/Akt pathway and activates the Akt/SOX2 axis, leading to enhanced stemness [56]. CAF-derived TGF-β engages analogous mechanisms through SOX9 activation [57]. This signaling is inherently bidirectional: HNSCC cells activate normal fibroblasts via TGF-β release, while CAFs reciprocally stimulate EGFR/extracellular-signal-regulated kinase (ERK) and Akt pathways in HNSCC cells through TGF-β secretion, simultaneously driving CAF expansion and tumor cell proliferation [58]. Platelet-derived growth factor D (PDGF-D), another CAF-secreted mediator, regulates PI3K/Akt and ERK signaling in HNSCC cells, culminating in EMT [59]. IL-6 binds tumor cell receptors to activate JAK/STAT3 signaling [60] and can further engage NF-κB in HNSCC cells, conferring resistance to apoptosis [61]. TGF-β expression is positively correlated with IL-6 in HNSCC and associates with increased lymph node metastasis, tumor recurrence, and poorer patient survival [62]. Activation of these pathways has been directly implicated in drug resistance; TGF-β-expressing CAFs promote cetuximab resistance in HNSCC [63], and concurrent activation of EGFR and STAT3 has been proposed as a predictor of chemoresistance [64].

CAFs and HNSCC cells also interact through exosomes and noncoding RNAs. Tumor-derived exosomes can educate CAFs to release IL-6 and IL-17 [65]. CAF-derived exosomes carry microRNAs including miR-21, which downregulates phosphatase and tensin homolog (PTEN) in HNSCC cells, enhancing chemoresistance through apoptosis inhibition. PTEN suppression is coupled to overactivation of the PI3K/Akt pathway and its downstream effectors [66]. HNSCC-derived miR-21 can also drive mesenchymal stem cell polarization toward a CAF phenotype by upregulating alpha-smooth muscle actin (α-SMA) and fibroblast activation protein (FAP) [67]. Alterations in specific miRNA species further modulate these interactions: reduced miR-148b-3p in CAF-derived exosomes associates with upregulation of ATPase copper transporting alpha polypeptide (ATP7A), suppressing cuproptosis and enhancing cell migration in OSCC [68]. CAF-derived exosomes also carry miR-876-3p, which activates GATA-binding protein 1 (GATA1) and leads to overexpression of insulin-like growth factor binding protein 3 (IGFBP3), contributing to chemoresistance [69]. Additionally, CAF-released exosomes contain metastasis-associated lung adenocarcinoma transcript 1 (MALAT1), a long noncoding RNA that stabilizes β-catenin and NF-κB to promote metastasis through EZH2 binding and consequent STAT3 and Akt activation via inhibition of the von Hippel-Lindau tumor suppressor (VHL) [70].

Several studies have explored therapeutic disruption of these CAF-HNSCC interactions. Blockade of TGF-β signaling has been shown to reverse therapeutic resistance in OSCC [71], and suppress PI3K/Akt activity along with downstream metastasis mediators such as MMP-9 [72]. Resveratrol nanoparticles and IL-6 antibody treatment reduced vascular endothelial growth factor (VEGF) expression and restricted cancer stem cell (CSC) expansion in OSCC [73]. Curcumin inhibited stromal cell-derived factor 1 (SDF-1) and TGF-β expression and release, thereby suppressing EMT in OSCC [74]. (Fig. 2)

Fig. 2.

Interactions between CAFs and HNSCC cells that promote EMT, resistance to apoptosis, and resistance to anticancer drugs. HNSCC cells can induce the activation of CAFs. In turn, CAFs stimulate EMT, stemness, and migration of cancer cells by releasing TGF-β, IL-6, PDGF, and certain miRNAs such as miR-21 and MALAT1. (CAF: cancer-associated fibroblast; EGF: epidermal growth factor; EMT: epithelial-mesenchymal transition; ERK: extracellular-signal-regulated kinase; FAP: fibroblast activation protein; miR: microRNA; NF-κB: nuclear factor of κB; MALAT1: metastasis-associated lung adenocarcinoma transcript 1; PDGF: platelet-derived growth factor; PI3K: phosphoinositide 3-kinase; PTEN: phosphatase and tensin homolog; Sox: SRY-box transcription factor; STAT3: signal transducer and activator of transcription 3; TGF-β: transforming growth factor-β)

HNSCC cells constitute one of the primary sources of PD-L1 within these tumors, and PD-L1 expression in cancer cells can be further induced by CAFs and other immunosuppressive stromal populations. One study examined how a nanotechnology-enhanced resveratrol formulation suppresses PD-L1 expression in OSCC cells by disrupting paracrine crosstalk between TAMs and CAFs. Western blotting and immunofluorescence analyses demonstrated that cells treated with the nanoformulation in a co-culture system exhibited approximately 50% reduction in PD-L1 expression, an effect more pronounced than that observed with free, non-formulated resveratrol. IL-6 released by CAFs and M2-polarized TAMs was identified as a central driver of JAK2/STAT3/PD-L1 axis activation in cancer cells. IL-6 blockade disrupted this paracrine signaling loop, attenuating the amplification of PD-L1 expression and reducing pro-tumor stromal interactions [75].

CAFs in altered metabolism and hypoxia

Hypoxia is a hallmark of HNSCC, arising from aberrant vasculature, ECM stiffness, and rapid tumor cell proliferation. Within hypoxic niches, stabilized hypoxia-inducible factor-1α (HIF-1α) reprograms CAFs toward a pro-tumorigenic phenotype [76]. Hypoxia also reshapes tumor metabolism through multiple interconnected mechanisms. Hypoxic CAFs and HNSCC cells exchange exosomes enriched with noncoding RNAs that act bidirectionally to drive tumor progression. HIF-1α and NF-κB activation in HNSCC cells, for instance, elevates miR-192/215 levels, which in turn activates CAFs via TGF-β/SMAD signaling [77]. Hypoxic cancer cells additionally release miR-21, which targets YOD1, a tumor suppressor in CAFs; its downregulation enhances CAF activation and metastasis [78]. miR-5100, also released by hypoxic HNSCC cells, activates CAFs through STAT3/Akt signaling [79]. Conversely, hypoxic CAFs liberate serglycin, which stabilizes and activates Wnt/β-catenin signaling in cancer cells to support their expansion [80].

Stromal metabolic reprogramming in HNSCC begins at an early stage, preceding full fibroblast-to-CAF transition. Co-culture studies using normal oral fibroblasts and OSCC cells reveal that these fibroblasts undergo profound metabolic shifts early in the activation process, including a switch to glycolysis, increased reactive oxygen species (ROS) generation, and elevated lactate export driven by upregulation of monocarboxylate transporter 4 (MCT-4). In certain settings, fibroblasts transfer mitochondria to adjacent cancer cells. These early changes are accompanied by reduced AMP-activated protein kinase (AMPK) activity and decreased peroxisome proliferator-activated receptor gamma coactivator 1-alpha (PGC-1α) expression, alterations that prime fibroblasts for full CAF activation and disrupt metabolic homeostasis in favor of tumor anabolism [81]. CAFs also engage in bidirectional metabolic signaling with cancer cells: hepatocyte growth factor (HGF) released by CAFs drives a glycolytic shift in HNSCC cells, which in turn secrete basic fibroblast growth factor (bFGF) to further amplify HGF production, establishing a positive feedback loop that sustains glycolysis and promotes lactate accumulation [82]. Monocarboxylate transporters (MCTs) are central regulators of this metabolic coupling. HNSCC cells exhibit the Warburg effect, generating high levels of lactic acid, while CAFs expressing MCT-4 and related transporters facilitate lactate export and establish a metabolic symbiosis within the stroma. CAFs both clear lactate and utilize it as a fuel source, supporting mitochondrial energy production in the tumor. MCT inhibition has therefore been proposed as a therapeutic strategy to attenuate stromal metabolic remodeling and reduce tumor growth [83]. TGF-β contributes an additional layer of metabolic dysregulation by inducing senescence in CAFs and impairing autophagy, a state that supports the metabolic demands of HNSCC cells [84]. TGF-β receptor II (TGF-βRII) overexpression can further suppress glucose metabolism in CAFs through repression of pyruvate kinase muscle isozyme 2 (PKM2) [85].

Collectively, CAF-cancer crosstalk in HNSCC is deeply intertwined with hypoxia-linked metabolic programs that sustain tumor growth and adaptive survival. Early metabolic rewiring in adjacent fibroblasts reflects a proactive stromal contribution to cancer progression, well before the full activation of CAF programs. CAF-derived serglycin and HGF support glycolysis in tumor cells, while fibroblast remodeling primes an MCT-dependent lactate shuttle that sustains tumor bioenergetics and metabolic resilience. Hypoxia and mechanical rigidity amplify this remodeling, and cancer cell secretions feedback to further activate CAFs. Targeting the key nodes within these interactions, including HIF-1α/NF-κB, exosomal noncoding RNAs (ncRNAs), HGF/bFGF signaling loops, and MCTs, represents a rational approach to disrupting metabolic symbiosis in HNSCC progression.

Immunosuppression by CAFs: potential targets

CAFs orchestrate immunosuppression in HNSCC through diverse mechanisms that collectively cripple antitumor immunity. This suppression is mediated by cytokines, growth factors, noncoding RNAs (ncRNAs), and immune checkpoint molecules released by CAFs, with phosphorylated STAT3 in CAFs serving as a central regulator of this immunosuppressive program [86]. CAF-derived TGF-β suppresses CTL function by downregulating granzyme B and perforin expression and additionally impairs the maturation of both CTLs and NK cells [87]. IL-10 and PGE2 are further soluble mediators produced by CAFs that, together with TGF-β, establish an immunosuppressive microenvironment that shields tumors from immune surveillance [88]. FAP-positive CAFs contribute to CTL exhaustion through CXCL12 expression; CXCL12 engages CXCR4 on CTLs and attenuates cytokine and granzyme release by these cells.

The PD-1/PD-L1 axis occupies a central position in CAF-mediated immune evasion in HNSCC. Upregulation of α-SMA in CAFs correlates with increased PD-L1 expression in OSCC [89]. Molecular analyses indicate that PDGF exposure induces CCL25 expression and Akt phosphorylation in CAFs, leading to PD-L1 upregulation [90], while FAP represents an additional key regulator of PD-L1 expression in this population [91]. Beyond direct immune suppression, CAFs also diminish the efficacy of ICIs. An OSCC tumor model study employing high-dimensional cellular profiling and RNA sequencing (RNA-seq) examined cellular and molecular changes following anti-PD-1 antibody treatment. PD-1 blockade was associated with a notable increase in CAF numbers, which subsequently enhanced collagen synthesis. A negative correlation was observed between CAF accumulation and infiltrating CTLs, and these changes were accompanied by expansion of Tregs and M2 TAMs. Taken together, these findings suggest that while PD-1 blockade may initially delay tumor progression through CTL activation, CAF-mediated stromal remodeling can ultimately undermine therapeutic efficacy [92].

These observations confirm that CAFs impose multilayered immunosuppression that directly impairs CTL and NK cell function. The responses documented in OSCC models indicate that ICI monotherapy produces limited antitumor activity in a stromal-rich context. Combining ICIs with CAF-targeting strategies, cytokine and chemokine blockade, and ECM remodeling approaches may therefore be necessary to achieve durable and effective antitumor immunity.

The recruitment and activation of immunosuppressive immune cells by CAFs: potential targets

CAFs establish a protective niche for immunosuppressive cell populations through coordinated chemokine secretion. CCL2 and CCL5 released by CAFs engage CCR2 and CCR5 on MDSCs, promoting their mobilization from bone marrow into the tumor [88]. Within the TME, MDSCs express arginase-1 and nitric oxide synthase 2 (NOS2), enzymes that deplete arginine and generate ROS, collectively impairing CTL function [93]. Activated MDSCs further contribute to resistance against anti-PD-1/PD-L1 therapy through upregulation of additional checkpoint molecules such as T cell immunoreceptor with immunoglobulin and ITIM domain (TIGIT) [94]. CAFs also secrete CCL2, CCL5, and CCL18, which act through CCR2, CCR5, and CCR8, respectively, to recruit monocytes and skew their polarization toward an M2 TAM phenotype. Upregulation of FERM domain-containing kindlin 2 (FERMT2) in CAFs plays a central role in this M2 TAM recruitment program [95].

CAF-derived signals also drive Treg expansion and modulate the immunosuppressive functions of neutrophils and Th17 cells. CCL22 secreted by CAFs engages CCR4 on Tregs, increasing their stromal infiltration, while CAF-derived CCL22 and IL-1β cooperate to upregulate Foxp3 and expand the Treg compartment [96]. TGF-β promotes the expansion of both Tregs and Th17 cells through overlapping mechanisms. One study demonstrated that CAF-derived IL-6 and TGF-β synergistically expand and activate both populations: IL-6 activates JAK2/STAT3 in Th17 cells to induce their differentiation and IL-17 A production, and IL-17 A in turn activates CAFs, establishing a self-amplifying feedback loop. In vivo, these interactions depleted CTLs and promoted cancer cell proliferation [97].

Collectively, these findings underscore the central role of CAFs in orchestrating immunosuppression in HNSCC. The process is driven by CCL2, CCL5, and miR-21, which sustain STAT3-dependent MDSC recruitment and expansion, while CAF-derived signals simultaneously skew monocytes toward M2-like TAMs and promote Treg and Th17 differentiation. The resulting network of self-reinforcing immunosuppressive loops represents a tractable set of targets; disrupting these interactions may prevent CTL exhaustion and restore functional antitumor immunity. (Fig. 3)

Fig. 3.

Interaction between CAFs with Tregs, MDSCs, and TAMs in HNSCC. Release of chemokines such as CCL2 and CCL5 recruits monocytes and induces polarization toward M2 TAMs. CAFs can also activate and induce the infiltration of Tregs and MDSCs. These cells potentiate CAF activation and exhaust CTLs via PD-L1 and TGF-β. These interactions increase resistance to apoptosis in HNSCC cells. (CAF: cancer-associated fibroblast; CCL: C-C motif chemokine ligand; CTL: cytotoxic CD8 + T lymphocyte; IFN-γ: interferon-γ; M2 TAM: M2 tumor-associated macrophage; MDSC: myeloid-derived suppressor cell; NF-κB: nuclear factor of κB; PD-L1: programmed death ligand 1; STAT3: signal transducer and activator of transcription 3; TGF-β: transforming growth factor-β; Th: T helper; TNF-α: tumor necrosis factor-α; Treg: regulatory T cell)

ECM in HNSCC

The ECM in HNSCC is a dynamic scaffold composed of collagens, including types I, III, and IV, along with fibronectin, hyaluronan, and proteoglycans. CAFs are the principal drivers of ECM development and remodeling, secreting matrix components and crosslinking enzymes that produce excessive crosslinking and progressive stiffening. This aberrantly remodeled ECM generates a hypoxic, mechanically rigid microenvironment that promotes tumor migration, invasion, and survival [98]. Increased collagen alignment and hyaluronan accumulation elevate IFP, impairing drug perfusion and restricting immune cell infiltration [99].

Beyond collagen abundance, ECM crosslinking is a primary determinant of matrix stiffness and downstream mechanical signaling. LOX family enzymes catalyze covalent collagen crosslinking, stiffen the matrix, align collagen fibers, and raise IFP. CAFs are the dominant stromal source of LOX in HNSCC, and CAF-derived LOX promotes malignant progression through FAK phosphorylation-dependent signaling. This LOX-stiffness-FAK axis provides a direct mechanistic link between CAF activity and stiffness-induced invasion and therapy resistance, motivating pharmacologic targeting of LOX activity and integrin/FAK signaling, either alone or in combination with cytotoxic and immunotherapeutic regimens [100]. Mechanical cues from a stiff ECM are transduced into pro-tumorigenic transcriptional programs through integrins, RhoA-ROCK, and nuclear mechanosensors, with YAP/TAZ acting as central integrators. In the fibrotic tumor stroma, YAP/TAZ activation reinforces CAF activity and ECM production, establishing a feed-forward loop in which matrix stiffness drives further collagen deposition, fiber alignment, and stromal immune exclusion. This mechanistic framework positions FAK- and ROCK-directed strategies alongside YAP/TAZ disruption within the same stromal reprogramming toolbox as cytokine blockade targeting TGF-β and IL-6/STAT3, and chemokine-axis inhibition targeting CXCL12/CXCR4 [101]. The ECM also serves as a reservoir for growth factors including TGF-β and VEGF, and these structural and biochemical alterations collectively generate a pro-tumorigenic niche that sustains HNSCC progression [102]. Hypoxia and ECM rigidity further induce metabolic remodeling in both CAFs and cancer cells, promoting lactate production and establishing a low-pH stromal environment [103].

Collagen and fibronectin are major ECM constituents that drive tumor cell proliferation and invasion. Clinical evidence indicates that collagen expression correlates with poor prognosis in OSCC [104], and collagen deposition can initiate tumorigenesis and expand OSCC cell populations [105]. Collagen also upregulates discoidin domain receptor 1 (DDR1), a collagen-activated receptor tyrosine kinase in HNSCC, whose overexpression enhances cancer cell migration, inhibits apoptosis, and increases chemotherapy resistance [106]. Fibronectin exerts similar pro-tumorigenic effects: its overexpression associates with increased invasion in HNSCC, and it induces VEGF to facilitate metastasis [107]. Fibronectin activates integrin α4-FAK signaling in tumor cells, driving proliferation and migration [108], promotes M2 TAM recruitment to amplify intratumoral immunosuppression [109], and engages the Akt/NF-κB pathway in HNSCC cells to confer resistance to apoptosis and antitumor drugs [110]. Hyaluronan, another key stromal component, contributes independently to drug resistance and invasion. The hyaluronan-rich matrix binds CD44 on cancer cells, activating PI3K/Akt and EGFR pathways along with downstream ERK1/2 signaling to enhance survival and chemoresistance [111]. CAF-derived hyaluronan synthase 2 (HAS2) upregulates MMPs and induces EMT, further promoting metastasis and cell migration [112].

Several studies have explored ECM components and conditions as therapeutic targets or drug delivery platforms. The acidic pH, hypoxia, and elevated expression of MMPs, collagen, hyaluronan, and fibronectin characteristic of the HNSCC stroma have each been proposed as actionable vulnerabilities. Two studies engineered pH-responsive nano-graphene platforms to deliver doxorubicin and a photothermal agent in OSCC models, demonstrating substantially enhanced drug release under the low-pH conditions of the tumor stroma and improved photothermal therapy efficacy [113, 114]. MMP-2 and MMP-9 have also been explored as triggers for controlled drug release: one study employed an MMP-2-responsive hydrogel enabling co-delivery of doxorubicin and imatinib in OSCC, producing significant tumor suppression and improved survival in tumor-bearing mice [115], while another demonstrated that MMP-2 quantum dot biosensors effectively deliver a photothermal agent to suppress tumors in response to laser irradiation [116]. ECM-degrading nanodrug platforms, including collagenase-based nanoparticles capable of improving drug penetration into the tumor stroma, warrant further investigation. (Fig. 4)

Fig. 4.

Targeting low pH and elevated concentrations of MMP-2 and MMP-9 in HNSCC as a strategy to deliver photothermal agents and chemotherapies. Unlike normal tissues, tumors have abnormal vasculature that facilitates nanoparticle accumulation. pH-sensitive and MMP-sensitive nanoparticles enable selective drug delivery to the tumor stroma. (MMP: matrix metalloproteinase; ROS: reactive oxygen species; Created with BioRender)

Taken together, these findings position the ECM in HNSCC as an active, instructive compartment whose CAF-driven remodeling, encompassing collagen crosslinking, fibronectin deposition, and hyaluronan accumulation, generates a biophysical and biochemical niche that directly drives invasion, EMT, and resistance to anticancer drugs. The underlying mechanisms are mechanistically distinct yet functionally convergent: DDR1 activation by collagen, integrin α4/FAK and Akt/NF-κB signaling induced by fibronectin, and CD44-mediated PI3K/Akt and ERK activation by hyaluronan each contribute to immunosuppression and therapeutic resistance in HNSCC. Through these pathways, ECM components enhance tumor cell survival and recruit immunosuppressive cells, reinforcing a feed-forward, immune-evasive microenvironment. The demonstrated responsiveness of low-pH and MMP-rich matrices to pH-sensitive and MMP-triggered nanocarriers provides proof of principle that ECM attributes are exploitable for targeted drug delivery (Table 2).

Table 2.

Summary of studies targeting checkpoint molecules, autophagy, growth factors, and metabolism in CAFs in HNSCC

Experimental model	Target/mechanism	Key findings	Refs.
General HNSCC model	MCTs/Lactate Shuttle	Disrupting the lactate shuttle through targeted MCT inhibition undermines the metabolic symbiosis between CAFs and cancer cells, impeding tumor growth.	[83]
OSCC tumor models	PD-1/PD-L1 Immune Checkpoint	PD-1 blockade increased CAF abundance, which upregulated collagen synthesis and correlated inversely with CTL infiltration, suggesting CAFs can undermine ICI efficacy.	[92]
In vitro and in vivo tongue squamous cell carcinoma models (fibroblast-tumor co-culture; mouse xenografts)	Artemisinin derivatives targeting PDGF	Treatment preserved fibroblast normalcy, reduced CAF activation markers, and impaired pro-tumorigenic stromal functions, resulting in slowed tumor growth and reduced invasive behavior.	[117]
Human HNSCC patient-derived CAFs and orthotopic xenografts; molecular cell assays	Autophagy in CAFs	Inhibiting CAF secretory autophagy curtailed release of pro-tumorigenic proteins, decreased tumor cell proliferation, and improved cisplatin response.	[118]
Assembled oral cancer organoids containing CAFs, endothelial cells, and tumor cells	Nicotinamide N-methyltransferase (NNMT) in CAFs	NNMT silencing or inhibition in CAFs reduced VEGFA production and endothelial network formation, lowering tumor angiogenesis in organoids and impairing pro-tumorigenic growth cues.	[119]
Murine tumor models and stromal profiling following CSF-1R blockade	CSF-1R	CSF-1R blockade reduced pro-tumorigenic TAMs but provoked CAF-mediated PMN-MDSC infiltration that neutralized antitumor effects; combined CXCR2 blockade improved efficacy.	[120]
Preclinical tumor models treated with synthetic retinoid plus ICI (mouse and ex vivo assays)	Meflin-positive CAF targeting by Am80 (synthetic retinoid)	Stromal preconditioning reduced immunosuppressive CAF and macrophage features, increased CTL infiltration, and enhanced responsiveness to PD-1 blockade.	[121]
OSCC cell lines, CAF co-cultures, and metastasis assays (in vitro and mouse metastasis models)	Galectin-1	Galectin-1 targeting lowered CAF-derived CCL2, decreased TAM recruitment, diminished metastatic dissemination, and impaired tumor cell motility.	[122]
In vitro invasion assays and patient sample correlation studies	CCL7 signaling in CAFs	CCL7 blockade reduced tumor cell invasive behavior in co-culture models, implicating CCL7 as a key paracrine driver of stroma-dependent OSCC invasion.	[123]
OSCC cell lines, CAF co-cultures, and murine models treated with HDAC6 inhibitors	HDAC6	Pharmacologic HDAC6 blockade decreased IL-1β levels, reprogrammed the TME toward reduced MDSC abundance, and limited tumor progression.	[124]

CCL7: C-C motif chemokine ligand 7; CSF-1R: colony-stimulating factor 1 receptor; CXCR2: C-X-C motif chemokine receptor 2; HDAC6: histone deacetylase 6; MCT: monocarboxylate transporter; NNMT: nicotinamide N-methyltransferase; PMN-MDSC: polymorphonuclear myeloid-derived suppressor cell; VEGFA: vascular endothelial growth factor A

Nanotechnology represents a promising translational platform for HNSCC, though its clinical potential warrants careful framing. pH-responsive, MMP-responsive, and ligand-guided nanoparticle systems have each demonstrated encouraging results in preclinical settings, yet these findings do not guarantee clinical success. Tumor heterogeneity, off-target distribution, manufacturing complexity, and variable stromal architecture may limit reproducibility across patient populations. Nanocarriers should therefore be regarded as potential tools for stromal modulation rather than definitive therapeutic solutions at this stage. Their most immediate translational value likely lies in improving local drug concentration, reducing systemic toxicity, and supporting combination regimens targeting CAF signaling, ECM remodeling, or immune resistance.

Conclusion

CAFs promote HNSCC tumor growth and therapeutic resistance through a range of interconnected mechanisms. Through the release of growth factors and cytokines, including TGF-β, IL-17, and IL-6, CAFs activate PI3K/Akt, STAT3, Sox family, and MAPK pathways in HNSCC cells, driving EMT, metastasis, and drug resistance. CAF-mediated ECM remodeling, characterized by collagen crosslinking and hyaluronan deposition, generates a mechanically stiff and hypoxic niche that reduces drug penetration and shields tumor cells from immune detection. This stiffened matrix amplifies hypoxia, EMT, metabolic reprogramming, and metastatic potential, positioning the ECM as an active, pro-tumorigenic compartment sculpted by CAFs. ECM constituents such as collagen, fibronectin, and hyaluronan serve as reservoirs for growth factors that sustain EMT, invasion, and apoptosis resistance, while simultaneously impairing drug and immune cell penetration. Hypoxia-inducible factor 1-alpha (HIF-1α), induced by hypoxia and metabolic crosstalk, activates NF-κB and exosomal ncRNA release, which in turn engage TGF-β/STAT3/Akt signaling in CAFs and drive glycolytic reprogramming of adjacent fibroblasts, establishing a monocarboxylate transporter (MCT)-dependent lactate shuttle that sustains tumor bioenergetics.

These CAF-derived factors also orchestrate broad immunosuppression. TGF-β directly exhausts CTLs and impairs their maturation, while CCL2, CCL5, CCL22, IL-6, and miR-21 recruit immunosuppressive cell populations into the TME. CAFs, M2 TAMs, MDSCs, Th17 cells, and Tregs form mutually reinforcing immunosuppressive networks that upregulate PD-L1 and drive CTL exhaustion via the PD-1/PD-L1 axis. These interactions sustain continuous production of TGF-β, IL-6, and related mediators, progressively consolidating an immunosuppressive stroma that facilitates EMT and cancer cell proliferation by shielding tumor cells from NK- and CTL-mediated cytotoxicity. CAF-derived TGF-β, IL-10, and PGE2 suppress granzyme B and perforin expression and impair CTL and NK maturation, while CXCL12-CXCR4, PDGF/Akt, and FAP activity in FAP-positive CAFs upregulate stromal PD-L1 and physically exclude CTLs from the tumor core.

Therapeutic targeting of CAFs in HNSCC may be achieved through modulation of expressed surface molecules, secreted cytokines, growth factors, and ncRNAs. Small-molecule inhibitors and epigenetic modulators have shown capacity to reduce pro-tumorigenic and pro-angiogenic factor production in preclinical models, limiting invasion and tumor spread. Direct CAF targeting may also curtail MDSC and monocyte recruitment and their polarization toward M2 TAMs by disrupting relevant chemoattractant signals. CAFs may blunt single-agent myeloid therapies such as colony-stimulating factor 1 receptor (CSF-1R) blockade by recruiting polymorphonuclear MDSCs (PMN-MDSCs), supporting the rationale for combination regimens. Collectively, these data favor CAF normalization or functional silencing over indiscriminate stromal depletion, which may be detrimental in certain stromal contexts.

Future investigations should prioritize CAF subtype resolution through single-cell RNA sequencing and spatial transcriptomics to delineate the distinct functions of individual subsets and enable patient-selective targeting. CAFs carrying high TGF-β, IL-6, or CCL2/CCL5 signatures, for instance, may represent tractable targets in tumors with pronounced stromal rigidity, addressable through siRNA-loaded nanoparticles or CRISPR-based platforms. Inhibiting downstream effectors, including STAT3, PI3K/Akt, and Sox programs, in cancer cells may further attenuate CAF-driven EMT, and dual targeting strategies such as combined TGF-β blockade and IL-6R antagonism may disrupt bidirectional stromal-tumor crosstalk while reducing stromal-induced chemoresistance. ECM normalization strategies warrant equal attention: hyaluronidase- or collagenase-loaded nanoparticles may improve drug penetration and immune infiltration, while targeting MMP-2 and MMP-9 may reduce metastasis and partially restore tissue architecture. pH-sensitive nanoparticles capable of delivering small molecules, siRNAs, and adjuvants locally to CAFs and the ECM offer a mechanism for translating these stromal targets into actionable drug delivery strategies. Dismantling the HNSCC stroma will ultimately require a multipronged approach that integrates precise stromal targeting, immune modulation, and metabolic disruption, supported by technological innovation to ensure delivery precision and selectivity. Addressing the CAF-immune-ECM interaction network holds the potential to remodel the TME in ways that meaningfully improve the efficacy of existing and emerging antitumor therapies.

CAFs and the ECM represent credible therapeutic targets in HNSCC, though current evidence supports cautious optimism rather than immediate clinical adoption. The preponderance of available data is mechanistic or preclinical, demonstrating that stromal signaling governs EMT, immune exhaustion, metabolic remodeling, and resistance to anticancer agents, and suggesting that selective stromal modulation may improve therapeutic response. CAF heterogeneity, context-dependent stromal functions, and limited clinical validation nevertheless remain substantial barriers to translation. The most defensible translational strategy is therefore to develop CAF- and ECM-directed interventions as biomarker-guided adjuncts within rational combinations involving radiotherapy, chemotherapy, immunotherapy, or targeted delivery systems. Meaningful progress will depend on subtype-resolved stromal profiling, improved preclinical model fidelity, and early-phase clinical studies designed to evaluate safety, pharmacodynamic endpoints, and patient selection criteria.

Author contributions

All authors were involved in first draft preparation. M A and M N reviewed and edited final manuscript. All authors confirmed final draft.

Funding

None.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

No human or animal study was performed in this study.

Consent for publication

Not applicable.

AI declarations

The manuscript has been checked and edited using AI-assisted Grammarly.

Competing interests

The authors declare no competing interests.