Unraveling resistance to immune checkpoint inhibitors in HNSCC: from mechanisms to combination therapies

Cancer immune surveillance and immune editing

The theory of cancer immune surveillance posits that the immune system maintains body homeostasis by recognizing and clearing tumor cells, but tumors can escape this process through immune editing^[36]. Immune editing consists of three stages: elimination, equilibrium, and escape. In HNSCC, immune editing primarily promotes resistance in the following mechanisms: (1) Antigen presentation defects: Tumor cells downregulate the expression of major histocompatibility complex I (MHC-I) or antigen-processing related transporter (TAP) to reduce the presentation of tumor-associated antigens (TAAs), thereby evading immune recognition^[37]. For example, the loss of MHC-I in HNSCC is significantly associated with resistance to PD-1 inhibitors^[38]; (2) Loss of immunogenic antigens: Under immune pressure, tumor cells selectively lose highly immunogenic antigens (such as HPV-related E6/E7 proteins) through epigenetic silencing or gene mutations, preventing recognition by T-cells^[39]; (3) Induction of immune tolerance: Tumor cells release immunosuppressive factors [such as interleukin (IL)-10 and transforming growth factor-beta (TGF-β)] that induce the expansion of regulatory T cells (Tregs), which suppress the function of effector T cells^[40].

Immune escape and immune suppression

The limited efficacy of immunotherapies in HNSCC can be attributed to a multifaceted network of immunosuppressive mechanisms operating within the TME. A well-characterized strategy employed by tumor cells to bypass the cytotoxic activity of T cells involves the high expression of PD-L1, which is often more pronounced in HPV-associated HNSCC. Ligation of PD-1 by PD-L1 effectively induces a state of functional exhaustion in antigen-specific T cells^[13]. The observation that therapeutic outcomes are not consistently predicted by PD-L1 expression levels alone suggests that resistance may be driven by the engagement of other inhibitory immune checkpoints, including cytotoxic T-lymphocyte-associated protein 4 (CTLA-4), lymphocyte activation gene-3 (LAG-3), and T cell immunoglobulin and mucin domain-containing protein 3 (TIM-3)^[41]. Beyond direct checkpoint interaction, the immunosuppressive niche is critically shaped by infiltrating myeloid cells. Tumor-associated macrophages (TAMs) suppress T-cell function through the release of soluble factors such as IL-10 and arginase-1. In parallel, myeloid-derived suppressor cells (MDSCs) induce T-cell apoptosis via the production of reactive oxygen species (ROS) and nitric oxide, creating a formidable barrier to an effective anti-tumor immune response^[42].

A fundamental distinction in the mechanisms of resistance to PD-1 vs. PD-L1 inhibitors stems from their differential binding specificities and consequent biological impacts. PD-1 inhibitors, including nivolumab and pembrolizumab, function by directly binding to the PD-1 receptor on T cells. This action provides broad-spectrum inhibition by preventing the receptor’s engagement with both of its known ligands, PD-L1 and PD-L2. The comprehensive blockade of PD-1 signaling can, however, trigger a compensatory feedback mechanism, leading to the pronounced upregulation of alternative co-inhibitory receptors such as LAG-3 and TIM-3 on exhausted T-cell populations^[41,43]. Additionally, resistance to PD-1 inhibitors frequently correlates with T-cell-intrinsic alterations, such as JAK1/2 mutations that disrupt IFN-γ signaling and reduce antigen presentation^[35].

Conversely, PD-L1 inhibitors (e.g., atezolizumab, durvalumab) bind specifically to PD-L1 on tumor or stromal cells, leaving the PD-1/PD-L2 axis intact. This allows persistent PD-L2-mediated immunosuppression through alternative receptors such as repulsive guidance molecule B (RGMb), promoting adaptive resistance^[44]. Tumor cells may also exploit PD-L1-independent pathways, including CD80 transendocytosis, which diminishes costimulatory signals for T-cell activation^[45]. TME further differentiates resistance: PD-L1 inhibitors are more susceptible to resistance driven by MDSCs expressing PD-L2 or V-domain Ig suppressor of T cell activation (VISTA), whereas PD-1 resistance is more associated with regulatory T-cell infiltration^[46].

Spatial heterogeneity amplifies these differences. PD-L1 expression is highly variable within tumors, and PD-L1 inhibitors require direct target engagement on cancer cells, leading to resistance from “immune deserts” lacking T-cell infiltration or PD-L1-negative clones that evade therapy^[47]. In contrast, PD-1 inhibitors act on circulating and tumor-infiltrating lymphocytes, making efficacy dependent on systemic T-cell functionality but vulnerable to peripheral tolerance mechanisms^[48]. Overcoming these resistance mechanisms may thus require tailored strategies - LAG-3/TIM-3 co-inhibition for PD-1 resistance, and targeting PD-L2 or myeloid compartments for PD-L1 resistance^[49].

Besides, the immunosuppressive role of vascular endothelial growth factor (VEGF) extends beyond angiogenesis to include direct and indirect modulation of the immune landscape. VEGF signaling coordinates a complex immunosuppressive program by inhibiting dendritic cell (DC) maturation, reducing T-cell infiltration, and inducing inhibitory cells such as Tregs and MDSCs^[50]. At the vascular level, VEGF-A-induced abnormal angiogenesis generates chaotic, leaky vessels with poor perfusion, leading to persistent hypoxia. This stabilizes hypoxia-inducible factor alpha (HIF-1α), which upregulates PD-L1 expression on tumor-associated endothelial cells and malignant cells, fostering an environment that excludes effector immune cells and facilitates immune evasion^[51]. Concurrently, VEGF inhibits the activation of nuclear factor kappa-B (NF-κB) in hematopoietic progenitor cells, preventing DC differentiation. This leads to a reduced number of functional DCs, thus weakening the antigen - presenting ability and subsequent activation of T cells against tumor cells^[52]. These mechanisms collectively establish VEGF as a non-redundant orchestrator of ICI resistance beyond its angiogenic role.

Gene mutations and heterogeneity in tumor cells

The high mutational burden in HNSCC should theoretically enhance immunogenicity, but gene heterogeneity and clonal evolution lead to resistance in several ways: (1) Driver gene mutations: Mutations in tumor protein 53 (TP53), notch receptor 1 (NOTCH1), and related genes inhibit T-cell infiltration by activating the wingless-int (WNT)/β-catenin pathway^[53]. Phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha (PIK3CA) mutations promote PD-L1 expression, leading to an immunosuppressive phenotype^[54]; (2) Clonal heterogeneity: Multiple subclones exist within the tumor, and some subclones carrying immune escape-related mutations [e.g., Janus kinase 1/2 (JAK1/2) deletions] are selectively expanded under treatment pressure^[55]; (3) Genomic instability: Chromosomal instability (CIN) weakens the innate immune response by inducing defects in interferon signaling pathways [e.g., inactivation of the stimulator of interferon response CGAMP interactor 1 (STING) pathway]^[56].

Impact of TME

TME is a complex ecosystem composed of tumor cells, diverse stromal cells (including immune cells, fibroblasts, endothelial cells, etc.), and the extracellular matrix (ECM), which collectively play a critical role in the pathogenesis, progression, metastasis, and treatment response of HNSCC. The cellular components of the TME exhibit significant heterogeneity; for instance, CD8⁺ CTLs serve as the principal anti-tumor effector cells; however, their functional capacity is frequently impaired due to exhaustion. In contrast, Tregs primarily exert immunosuppressive activities. Moreover, TAMs and MDSCs facilitate tumor immune escape through the secretion of immunosuppressive mediators. The spatial imbalance in the distribution of these cellular components and their functional phenotypes - termed the spatial heterogeneity of the TME - represents a key determinant of therapeutic resistance, particularly to interventions such as ICIs, in HNSCC.

Cancer-associated fibroblasts (CAFs) are pivotal components of TME that actively contribute to immune evasion by secreting TGF-β, a process that directly suppresses T cell infiltration into tumor sites. TGF-β signaling in CAFs promotes the creation of physical and molecular barriers, which exclude CTLs from penetrating tumor nests and accessing cancer cells, thereby limiting effective anti-tumor immune responses^[57,58]. Specifically, CAF-derived TGF-β enhances the secretion of immunosuppressive factors and chemokines, such as C-X-C motif chemokine ligand 12 (CXCL12), that coat tumor cells and establish a milieu permissive for immune quiescence^[59]. This secretion further recruits Tregs, amplifying the immunosuppressive microenvironment through increased TGF-β levels and fostering Treg infiltration, which collectively exhausts and excludes effector T cells^[60]. Consequently, the TGF-β-dependent signaling in CAFs represents a key axis for modulating immune dynamics, highlighting its potential as a therapeutic target to overcome T cell exclusion and enhance immunotherapy efficacy in solid tumors.

Metabolic reprogramming, a hallmark of cancer and immune disorders, involves a shift from oxidative phosphorylation to glycolysis, leading to significant lactate accumulation in the TME^[61]. This excessive lactate buildup creates an acidic immunosuppressive milieu that directly impairs CTL function by interfering with metabolic pathways in immune cells, promoting CTL exhaustion and dysfunction^[62]. Specifically, lactate accumulation alters nutrient uptake and intracellular metabolic signaling in CTLs, disrupting their differentiation and activation^[63]. Mechanistically, lactate accumulation facilitates lactylation, a post-translational modification that modifies both histone and non-histone proteins, regulating gene expression through epigenetic changes such as histone H3 lysine 18 lactylation (H3K18la). This reshapes transcriptional programs in immune cells and suppresses CTL-mediated immunity^[64]. Lactate-derived lactylation acts as a key regulator in the immunosuppressive TME, reinforcing CTL exhaustion and limiting their persistence^[65]. Such metabolic-epigenetic crosstalk highlights how lactate not only functions as a metabolic byproduct but also orchestrates immunosuppressive signaling that inhibits CTL efficiency, presenting a therapeutic target for overcoming immunosuppression.

Tumor heterogeneity in the TME is characterized by distinct immune phenotypes, including immune-inflamed tumors with abundant T cell infiltration and immune-desert tumors with a paucity of CD8⁺ T cells, leading to differential therapeutic outcomes. Immune-inflamed tumors are more responsive to immune checkpoint blockade (ICB) but can develop resistance through immunosuppressive niches formed by CAFs and macrophages, which spatially exclude CD8⁺ T cells and contribute to local immune evasion^[66]. In contrast, immune-desert tumors exhibit innate resistance to ICB due to an immune-deficient microenvironment incapable of initiating effective anti-tumor responses^[67-69]. Spatial heterogeneity, influenced by metabolic pressures and stromal interactions, promotes the emergence of drug-resistant clones and immune escape. For instance, the dynamic interplay between tumor cells and the ECM spatially restricts immune cell access, while desert regions correlate with non-expanded T cell receptor (TCR) clones and reduced MHC-I expression, exacerbating therapeutic resistance^[70-72]. Comprehensive spatial analysis underscores the need for personalized strategies, such as enhancing immune cell infiltration in desert microenvironments or disrupting immunosuppressive niches in inflamed tumors, to overcome resistance^[73,74]. Addressing this heterogeneity is critical for developing innovative immunomodulatory strategies.

Resistance to ICIs in HNSCC stems from multifactorial mechanisms, including tumor-intrinsic genomic heterogeneity, immunosuppressive TME dynamics, and compensatory immune checkpoint upregulation. Spatiotemporal mapping of resistance mechanisms via single-cell and spatial transcriptomics will be critical to identify context-specific therapeutic vulnerabilities for personalized interventions.

CTLA-4 inhibitors

Simultaneously blocking inhibitory checkpoints or activating stimulatory pathways can maximally enhance effector T-cell function [Table 4]. Although the combination of anti-PD-1/PD-L1 and anti-CTLA-4 agents has demonstrated superior efficacy over monotherapy in multiple tumor types, its clinical benefits in specific malignancies such as HNSCC require further precise delineation. The mechanistic basis for this synergy stems from the competitive binding of CTLA-4 to cluster of differentiation 80/cluster of differentiation 86 (CD80/CD86) ligands on APCs, which attenuates the costimulatory signals delivered through cluster of differentiation 28 (CD28) and consequently suppresses the initial activation of T cells. In contrast, anti-CTLA-4 antibodies not only block ligand engagement by binding CTLA-4 with high affinity but may also deplete or functionally inhibit Tregs within the TME, thereby alleviating immunosuppression and potentiating anti-tumor immunity. Nonetheless, clinical outcomes have been variable; for instance, the phase III EAGLE trial (NCT02369874) in R/M HNSCC revealed no statistically significant improvement in OS with durvalumab (anti-PD-L1) plus tremelimumab (anti-CTLA-4) compared to durvalumab monotherapy^[93]. Beyond the PD-1/PD-L1 and CTLA-4 pathways, other co-inhibitory receptors - including TIM-3, LAG-3, and T cell immune receptor with Ig and ITIM domains (TIGIT; Ig: immunoglobulin domain, ITIM: immunoreceptor tyrosine-based inhibitory motif), which are expressed on tumor-infiltrating lymphocytes - have emerged as promising next-generation targets for reversing T-cell exhaustion and reinvigorating anti-tumor immune responses.

LAG-3 inhibitors

LAG-3, an immunoinhibitory receptor structurally homologous to CD4, is upregulated on T cells following activation and maintains persistent expression under conditions of chronic antigen exposure, such as cancer and persistent viral infections^[109-111]. Its expression extends beyond T cells to encompass various immune cells, including NK cells, B cells, and DCs. On conventional T cells, LAG-3 functions as a critical negative regulator, suppressing T-cell activation and cytokine production, thereby contributing to a state of T-cell exhaustion^[112]. Additionally, LAG-3 is expressed on forkhead box P3 positive (FOXP3⁺) Treg cells, leading to an immunosuppressive state^[113]. Although LAG-3 blockade alone did not show benefits in preclinical studies, when combined with PD-1 inhibitors, it significantly enhanced tumor control in tumor models. The pivotal RELATIVITY-047 trial (NCT03470922), a global, randomized, double-blind phase II/III study, evaluated the combination therapy Opdualag (anti-LAG-3 plus anti-PD-1) vs. nivolumab monotherapy in patients with unresectable or metastatic melanoma. The results demonstrated a superior median PFS of 10.1 months for the Opdualag group compared to 4.6 months for the nivolumab group. The safety profile was manageable, with common immune-related AEs (irAEs) including thyroiditis and rash^[114]. Based on these findings, the U.S. FDA granted approval for this combination in 2022. Consequently, multiple phase I and II clinical trials are ongoing to further explore the efficacy of LAG-3 inhibitors, both as monotherapy and in combination regimens, in other malignancies, including HNSCC^[43].

TIM-3 inhibitors

TIM-3, a member of the T cell immunoglobulin mucin (TIM) family of immunoregulatory proteins, functions as a multifaceted inhibitory receptor within the TME. Initially characterized as a marker for IFN-γ-producing CD4⁺ (Th1) and CD8⁺ T cells, subsequent research has revealed that its expression extends to a diverse array of immune cells^[115]. These include Tregs, MDSCs, NK cells, and mast cells, indicating that TIM-3-targeted therapeutic strategies may exert immunomodulatory effects through multiple cellular pathways^[116-118]. Preclinical investigations have demonstrated that concurrent blockade of the TIM-3 and PD-1 pathways can induce significant tumor regression, a finding that has accelerated the clinical translation of anti-TIM-3 agents^[119,120]. Among the most advanced candidates are sabatolimab (MBG453) and cobolimab (TSR-022). Sabatolimab has received Fast Track designation from the FDA for use in combination with hypomethylating agents (HMAs) for the treatment of myelodysplastic syndromes (MDS)^[121]. Meanwhile, cobolimab is under evaluation in phase II/III clinical trials for various solid tumors, including lung cancer, hepatocellular carcinoma, and melanoma, exploring its efficacy in combination with PD-1 inhibitors and/or chemotherapeutic drugs. Encouragingly, a phase I trial investigating cobolimab in combination with a PD-1 inhibitor for lung cancer reported an ORR of 42.9%, alongside a favorable tolerability profile without serious treatment-emergent AEs^[122].

TIGIT inhibitors

TIGIT, a newly identified co-inhibitory receptor, was discovered in 2009^[123]. Its expression pattern varies across immune cell populations: it is transiently induced on T cells following TCR stimulation, while being constitutively expressed on NK cells and multiple T-cell subsets, such as Tregs, type 1 regulatory T (Tr1) cells, follicular helper T (Tfh) cells, and functionally impaired CD8⁺ T cells^[124]. Emerging evidence has further demonstrated TIGIT expression on regulatory B cells and innate immune cell populations^[125], suggesting that this receptor may exert regulatory effects in immune responses that extend beyond its well-documented roles. Upon engagement with its cognate ligands, TIGIT exerts suppressive effects on T-cell activation and impairs NK cell functionality^[126]. Moreover, within the TME, TIGIT mediates tumor immune escape through dual mechanisms: direct inhibition of effector CD8⁺ T cells and augmentation of the immunosuppressive properties of Tregs^[127]. A current phase II trial (CITYSCAPE-02, NCT03563716) is comparing the efficacy of PD-L1 inhibitors alone vs. a combination of TIGIT inhibitors with PD-L1 inhibitors in NSCLC patients. Results showed that the combination therapy group had a significantly higher ORR compared to the monotherapy group, with both PFS and OS improving^[128]. Besides NSCLC, other cancers are also being tested for the potential of TIGIT blockade as a therapeutic target. In the SKYCRAPER-09 trial (NCT03563716), the efficacy of tiragolumab combined with atezolizumab in R/M HNSCC was compared with a placebo^[129]. Additionally, multiple phase II/III clinical trials are ongoing in lung cancer, liver cancer, kidney cancer, gastrointestinal tumors, and HNSCC with anti-TIGIT/anti-PD-1/anti-PD-L1 combined with chemotherapy, and we look forward to the results of these trials in the future^[49].

B7 homolog 3 protein inhibitors

B7 homolog 3 protein (B7-H3) is a dual-function immune checkpoint molecule with both costimulatory and co-inhibitory properties, whose expression spans cancer cells and bone marrow-derived immune cells - including T lymphocytes, B lymphocytes, monocytes/macrophages, and DCs^[130]. Notably, elevated B7-H3 expression levels in tumor cells and immune-infiltrating cells within the TME have been linked to the development of tumor immune evasion^[131]. Emerging evidence suggests that B7-H3 promotes tumor proliferation, metastasis, and therapeutic resistance. In oral squamous cell carcinoma, B7-H3 overexpression is linked to increased tumor aggressiveness and higher mortality rates, with blockade suppressing tumor formation^[132]. These findings highlight B7-H3 as a promising therapeutic target. A phase I trial evaluated the B7-H3 inhibitor enoblituzumab combined with the anti-PD-1 inhibitor retifanlimab across multiple tumor types, showing an ORR of 33.3% in 18 HNSCC patients, including five partial responses (PRs) and one complete response (CR)^[133]. These promising phase I results have led to subsequent phase II/III trials to assess the safety and anti-tumor activity of combination therapy.

Inducible T cell costimulatory agonists

The ICOS/inducible T cell costimulator (ICOS, also called CD278) pathway, an activation-induced T-cell costimulatory receptor, plays a critical role in modulating T-cell proliferation, survival, and differentiation^[134]. ICOS ligand (ICOSL) is expressed by APCs and various tumor cells within the microenvironment. ICOS ligation triggers downstream signaling that exerts pleiotropic effects across T-cell subsets, potentially enhancing CD8⁺ T-cell cytotoxicity while simultaneously amplifying the immunosuppressive function of Tregs^[135]. Although preclinical studies supported a synergistic effect between ICOS agonism and PD-1 blockade, clinical development faced challenges, as evidenced by the termination of the phase II/III INDUCE-3 (NCT04128696) and INDUCE-4 (NCT04428333) trials of the ICOS agonist feladilimab in R/M HNSCC due to insufficient efficacy^[108]. Subsequent research has shifted toward novel bispecific antibodies, such as izuralimab (XmAb23104), which targets both PD-1 and ICOS to concurrently block inhibitory signaling and provide agonist costimulation.

TLR agonists

Toll-like receptor 9 (TLR9) agonists have been widely studied in both monotherapy and combination therapy for cancer treatment. Combination therapy with TLR9 agonists and immunotherapy has shown a synergistic effect, enhancing anti-tumor immune responses. TLR9 TLR9 agonists are capable of activating B lymphocytes and plasmacytoid DCs (pDCs), which in turn enhances the secretion of chemokines and cytokines. This immunomodulatory effect optimizes the tumor immune microenvironment and facilitates T cell-driven anti-tumor immune responses^[136]. A phase Ib clinical trial (ClinicalTrials.gov: NCT02680184) investigated the therapeutic efficacy and safety profile of intratumoral administration of the Toll-like receptor (TLR) agonist Vidutolimod in combination with a PD-1 inhibitor. In this study involving 44 patients, 11 individuals achieved either a PR or CR, with a median PFS of 2.8 months. These results highlight the potential of TLR9 agonists to reverse PD-1 inhibitor resistance in patients with advanced melanoma^[137]. Another clinical study (NCT02521870) evaluated the safety and efficacy of SD-101 - a synthetic CpG oligodeoxynucleotide (CpG-ODN) acting as a TLR9 agonist - in combination with pembrolizumab for the treatment of R/M HNSCC. Among the 28 enrolled patients, the ORR was recorded at 22%^[138].

Epidermal growth factor receptor inhibitors

Epidermal growth factor receptor (EGFR) is an established therapeutic target for HNSCC. There is a theoretical rationale for combining EGFR inhibitors with ICIs. Preclinical studies have demonstrated that EGFR signaling can influence the TME by modulating cytokines such as IL-6, TGF-β, and progranulin, thereby suppressing immune-mediated anti-cancer effects. EGFR activation induces the expression of PD-L1 and other immunosuppressive factors^[139]. Cetuximab is a mAb used for HNSCC treatment that inhibits EGFR signaling. The NCT01218048 study revealed that after cetuximab treatment, PD-1 expression on CD8⁺ TILs increased significantly and was closely associated with treatment response, providing a theoretical basis for combination therapy. A phase II trial evaluated pembrolizumab plus cetuximab in platinum-refractory recurrent HNSCC (33 patients). The ORR was 45%, with a median duration of response (DOR) of 14.9 months and no severe treatment-related toxicity^[140]. Multiple phase I-II trials of PD-1/PD-L1 plus EGFR inhibitors in R/M HNSCC are complete [Table 5]; an ongoing phase III trial assesses cetuximab/radiotherapy with avelumab in locally advanced HNSCC^[146].

A meta-analysis of seven trials showed cetuximab + anti-PD-1 significantly improved ORR and 1-year OS in HPV-negative R/M HNSCC vs. monotherapy, but no benefit in HPV-positive disease. However, subgroup analysis indicated no significant effect in HPV-positive diseases^[147].

VEGFR inhibitors

Lenvatinib acts as a broad-spectrum receptor tyrosine kinase (RTK) inhibitor, which exerts robust inhibitory effects on major mediators of tumor angiogenesis and progression. These key regulatory targets include vascular endothelial growth factor receptors 1-3 (VEGFR1-3), fibroblast growth factor receptors 1-4 (FGFR1-4), platelet-derived growth factor receptor α (PDGFRα), as well as the KIT proto-oncogene, receptor tyrosine kinase (KIT) and ret proto-oncogene (RET) kinases. Clinically, this agent has secured approval from FDA for the treatment of several malignancies, specifically radioactive iodine-refractory differentiated thyroid carcinoma, advanced renal cell carcinoma, and unresectable hepatocellular carcinoma. The KEYNOTE-146 phase Ib/II trial (NCT02501096) assessed the efficacy and safety of pembrolizumab combined with lenvatinib in HNSCC, showing improved ORR and a median PFS of 7.6 months [95% confidence interval (CI) 4.2-12.6]. Treatment-related AEs primarily included fatigue, appetite loss, hypertension, diarrhea, and nausea, without severe toxic effects^[148]. Based on these findings, a phase III trial (LEAP-010, NCT04199104) was initiated to evaluate pembrolizumab with or without lenvatinib in R/M HNSCC^[149]. Results showed that in PD-L1 CPS ≥ 1 R/M HNSCC patients, the combination therapy significantly improved PFS and ORR, but OS was not prolonged. The safety profile was consistent with previous reports, although 156 (61.4%) patients in the combination therapy group experienced treatment-related AEs, compared to 45 (17.8%) in the placebo group.

In addition, multiple studies have investigated the efficacy of combining VEGFR inhibitors with ICIs in the treatment of metastatic or recurrent HNSCC. For instance, a phase 1/2 trial evaluating ramucirumab (a mAb targeting VEGFR2) in combination with pembrolizumab for recurrent or metastatic HNSCC demonstrated encouraging results. In this trial, ORR reached 55%, highlighting the superiority of the combination therapy over single-agent immunotherapy in improving response rates^[150]. Additionally, a phase II trial of pembrolizumab combined with cabozantinib (a multi-targeted RTK inhibitor including VEGFR) achieved promising outcomes, meeting the primary endpoint of improved ORR^[151]. Cabozantinib modulates immune cell function within the TME by targeting VEGFR and other pathways, thereby synergizing with pembrolizumab to enhance tumor control.

Indoleamine-2,3-dioxygenase 1 inhibitors

Indoleamine-2,3-dioxygenase 1 (IDO1), a key immunomodulatory enzyme, catalyzes the initial and rate-limiting step of tryptophan catabolism along the kynurenine pathway, thereby suppressing CTL activity and contributing to an immunosuppressive TME^[152]. In HNSCC, elevated expression of IDO1 is frequently observed within malignant cells, as well as in TAMs and DCs, and this overexpression is correlated with unfavorable patient prognosis^[153]. IDO1 plays a role in T cell exhaustion and inhibition while promoting Treg cell activation and MDSC infiltration^[154,155]. The selective IDO1 inhibitor epacadostat was developed to counteract this mechanism. Studies have also suggested that IDO1 activity is closely related to ICI resistance. A phase I/II clinical trial evaluating epacadostat combined with pembrolizumab in HNSCC showed an ORR of 34%^[156]. However, the CheckMate 9NA/ECHO-310 phase III study (NCT03342352), which assessed the same treatment combination in melanoma, yielded poor results, leading to early study termination^[157].

Antibody–drug conjugates

Antibody–drug conjugates (ADCs) are hybrid therapeutic agents composed of mAbs and small-molecule cytotoxic payloads, where the two components are covalently linked via a chemical linker. The mAb moiety of ADCs is specifically designed to bind to tumor-specific antigens (TSAs) or TAAs, enabling targeted delivery of the cytotoxic agent to cancerous cells. This unique targeted mechanism endows ADCs with two key advantages: enhanced safety profiles and an expanded therapeutic window compared to conventional chemotherapeutics. Furthermore, the antibody moiety of ADCs interacts with immune effector cells in the TME, triggering antibody-dependent cellular cytotoxicity (ADCC) to enhance anti-tumor effects^[158]. A NCT04223856 trial evaluated Enfortumab Vedotin (ADC) plus Pembrolizumab as first-line therapy for advanced urothelial carcinoma. Results showed that the combination group had significantly longer PFS (12.5 vs. 6.3 months) and OS (31.5 vs. 16.1 months) than the standard chemotherapy group^[159]. The combination of ADCs and ICIs may offer a new potential treatment approach for R/M HNSCC. Based on data from other cancer types, further research into this combination therapy is warranted.

Oncolytic viruses

Oncolytic viruses represent a distinctive category of viruses characterized by their ability to selectively proliferate within cancer cells, trigger apoptotic cell death in these malignant targets, while leaving normal, non-tumorigenic tissues intact^[160]. RP1, a herpes simplex virus type 1 (HSV-1)-based oncolytic immunotherapeutic agent, is engineered to express two key bioactive molecules: human granulocyte-macrophage colony-stimulating factor (GM-CSF) and the chimeric fusion protein Gibbon Ape Leukemia Virus envelope glycoprotein (GALV-GP-R). This dual-expression design strengthens the eradication of tumor cells by potentiating the activation of systemic anti-tumor immune responses. The phase I/II IGNYTE study (NCT03767348) assessed the therapeutic efficacy of RP1 in combination with nivolumab among melanoma patients who had previously experienced treatment failure with PD-1 inhibitors. Study outcomes indicated that at the 12-month follow-up, the ORR reached 33.6%. Further stratification showed an objective response rate of 31.4% and a CR rate of 12.2%. Notably, the mean DOR surpassed 35 months, and the combined regimen exhibited favorable tolerability in the study cohort^[161]. Beyond RP1, other oncolytic viruses - including vaccinia virus, vesicular stomatitis virus, and measles virus - are currently under clinical investigation for the treatment of head and neck cancer^[162,163].

Nanomedicine approaches for cancer treatment

The insufficient accumulation of ICIs within HNSCC tumors constitutes a critical barrier to their therapeutic effectiveness. This limitation is mainly attributed to aberrant tumor vascularization, elevated interstitial fluid pressure, and a dense ECM^[164]. Nano-drug delivery systems (NDDSs) have emerged as a promising strategy for the targeted delivery of diverse ICIs, with the majority of current research focusing on PD-1/PD-L1 inhibitors. These nanosystems enable the site-specific delivery of various ICI types - including antibodies, nucleic acids, and small-molecule inhibitors - to tumor tissues, thereby boosting anti-tumor potency. Furthermore, NDDSs can be engineered for dual ICI co-delivery, such as the simultaneous transport of PD-1/PD-L1 and CTLA-4 inhibitors^[165]. Collectively, NDDSs have not only enhanced the therapeutic efficacy of ICIs but also mitigated the incidence of irAEs. Nanomedicine approaches address the pharmacokinetic limitations that have hampered traditional combination therapies. For example, glutathione (GSH)-sensitive polyethylene glycol-poly-L-lysine (PEG-PLL) micelles are surface-modified with angiopep-2 peptide [to penetrate the blood-brain barrier (BBB)] and co-load anti-PD-L1 and paclitaxel (PTX) to achieve chemo-immunotherapy for glioma^[166]; Hollow mesoporous organosilicon nanoparticle-coated Fe-MOF loaded with doxorubicin (DOX), which can locally release DOX after surgery to induce ICD, combined with anti-PD-1 to inhibit postoperative recurrence and brain metastasis of 4T1 breast cancer^[167]. Man@pSiNPs-erastin (mannose-modified porous silicon nanoparticles loaded with erastin), targeting TAM to induce ferroptosis, combined with anti-PD-1 to inhibit liver cancer^[168]; Cu₂WS₄-PEG nanozyme induces ferroptosis through the Kelch-like ECH associated protein 1 (KEAP1)/nuclear factor erythroid 2–related factor 2 (NRF2)/heme oxygenase 1 (HMOX1)/glutathione peroxidase 4 (GPX4) pathway, and enhances immunoradiotherapy in combination with radiotherapy and anti-PD-L1^[169].

In conclusion, NDDSs have unique advantages and potential in the treatment of R/M HNSCC, and various development strategies can help them maximize therapeutic effects and reduce side effects in treatment.

Neoadjuvant ICB

Neoadjuvant ICIs leverage intact tumor antigenicity to prime systemic anti-tumor immunity prior to surgery. This approach induces in situ formation of tertiary lymphoid structures (TLS) and clonal T-cell expansion, enhancing micrometastasis eradication. TLS are lymph node-like structures formed in chronic inflammatory or TMEs. While not endogenously formed at birth, these structures exhibit functional similarities to lymph nodes: they recruit immune cell populations - including T lymphocytes, B lymphocytes, and DCs - and facilitate localized immune responses. Within the TLS microenvironment, DCs capture TAAs and present them to naive T cells. Subsequently, the activated T cells undergo clonal proliferation under the stimulation of cytokines (e.g., IL-2), yielding a large pool of effector T cells that share identical antigenic specificity. This process enhances the recognition and cytotoxic capacity against tumor cells. In tumors, the presence of TLS is often correlated with improved immune infiltration and better prognosis, as they provide an organized immunological niche where immune cells can be activated and proliferate. For resectable HNSCC patients undergoing neoadjuvant nivolumab therapy, the surgical procedure was successfully accomplished in roughly 88% of participants across all study arms. The median duration of follow-up stood at 38.3 months. With respect to 36-month EFS outcomes, distinct differences were observed between the pembrolizumab and control groups: in the population with a CPS of 10 or higher, the EFS rates were 59.8% and 45.9%, respectively; whereas in the CPS ≥ 1 subgroup, the rates were 58.2% vs. 44.9%; and in the total population, 57.6% vs. 46.4%^[170]. Mechanistically, ICIs expanded tumor-reactive T-cell clones post-surgery, reduced MDSCs, and induced systemic CD8⁺ memory differentiation driving abscopal effects^[171]. Phase III trials (e.g., KEYNOTE-689, NCT03765918) now validated pathologic response as a surrogate endpoint for OS^[172,173].

Microbiome modulation: the gut-immune axis as a therapeutic target

Based on provided literature, the gut microbiome significantly modulates ICI efficacy in cancers, including HNSCC, through intricate mechanisms involving immune crosstalk: (1) Molecular Mimicry: Bacterial peptides can structurally resemble tumor antigens. For instance, peptides derived from Bacteroides fragilis exhibit cross-reactivity with TP53 antigens^[174]. This mimicry potentially triggers or modifies T-cell responses targeting tumors; (2) Metabolite Production: Microbial metabolites profoundly influence immune cell differentiation and function. Short-chain fatty acids (SCFAs), produced by commensal bacteria during fiber fermentation, promote the differentiation and function of Tregs^[175]. Conversely, the microbial metabolite inosine enhances anti-tumor Th1 responses, potentiating ICI activity^[176]; (3) DC Priming: Specific commensal bacteria directly prime DCs, essential for initiating adaptive immunity. Akkermansia muciniphila, a species associated with positive ICI responses, enhances DC function leading to the upregulation of IL-12. This, in turn, promotes the recruitment and activation of cytotoxic CD8⁺ T cells within the TME^[177].

Clinically, modulation strategies such as Fecal Microbiota Transplantation (FMT) exploit these mechanisms to overcome resistance. Transplanting microbiota from ICI-responsive donors to refractory patients enriches beneficial taxa (e.g., Ruminococcaceae) and has demonstrated the capacity to double objective response rates compared to control groups (32% vs. 15%) in refractory cancers^[178]. Importantly, the microbiome’s impact extends beyond the gut. Within the HNSCC context, a high abundance of Fusobacterium in the salivary microbiome has been negatively correlated with resistance to PD-1 inhibitors, suggesting that modulating both gut and local oral/respiratory microbiota represents a viable therapeutic avenue to enhance ICI efficacy and circumvent resistance^[179]. Collectively, targeting the microbiome-immune axis is emerging as a critical strategy to optimize immunotherapy outcomes.

Multimodal therapeutic innovations

CAR-NK cell therapy: precision off-the-shelf immunity
Chimeric antigen receptor-modified NK (CAR-NK) cells are a pivotal breakthrough for overcoming solid tumor resistance. Via natural cytotoxicity receptors, they exert MHC-unrestricted cytotoxicity, and their PD-1^- and TIM-3^- phenotype avoids cytokine release syndrome (CRS) and neurotoxicity. The University of Texas MD Anderson Cancer Center spearheaded a phase I/IIa clinical study (NCT03056339) to assess the safety and efficacy of cord blood-derived cluster of differentiation 19 (CD19)-targeted CAR-NK cells (designated CAR19/IL-15 NK cells) in patients with relapsed or refractory CD19-positive B-cell hematological malignancies, such as non-Hodgkin lymphoma (NHL) and chronic lymphocytic leukemia (CLL). Results from this trial demonstrated a cumulative ORR of 48.6% across the cohort, with substantial variability by disease subtype: 100% ORR in low-grade NHL, 67% in CLL, and 41% in diffuse large B-cell lymphoma (DLBCL). The 1-year OS was 68%, with no graft-versus-host disease (GVHD) or neurotoxicity observed^[180]. Critically, CAR-NKs persisted > 90 days post-infusion, confirming durable engraftment^[181]. Targeting diverse antigens [e.g., EGFR deletion variant III (EGFRvIII), MHC class I polypeptide-related sequence A (MICA) and MHC class I polypeptide-related sequence B (MICB), natural killer group 2 member D (NKG2D) ligands] enhances tumor specificity while leveraging “off-the-shelf” availability^[182].

STING agonists: activating innate immune surveillance
Stimulator of interferon genes (STING) agonists reprogram immunosuppressive TMEs by binding cyclic guanosine monophosphate-adenosine monophosphate (cGAMP), triggering TANK binding kinase 1 (TBK1)/interferon regulatory factor 3 (IRF3) phosphorylation and type I interferon production. This signaling cascade facilitates the maturation of DCs and enhances the infiltration of CD8⁺ T cells into the TME. In the context of HNSCC, a phase II clinical trial (NCT03937141) investigated the combination of ADU-S100 and pembrolizumab in adult patients with PD-L1-positive R/M HNSCC. The findings indicated that the ADU-S100 plus pembrolizumab regimen exhibited favorable tolerability and successfully induced abscopal effects^[183]. Novel delivery systems - including pH-sensitive nanoparticles targeting TAMs and sustained-release intratumoral implants (e.g., MK-2118) - optimize bioavailability and reduce systemic toxicity^[1,184].

Epigenetic modulators: resetting immune evasion mechanisms
Histone deacetylase inhibitors (HDACi; e.g., entinostat) and enhancer of zeste homolog 2 (EZH2) inhibitors (EZH2i; e.g., tazemetostat) reverse epigenetic silencing of immunogenic pathways. HDACi downregulate Treg-suppressive genes [FOXP3, cytotoxic T-lymphocyte-associated protein 4 (CTLA4)], while EZH2i reactivate endogenous retroviruses, inducing “viral mimicry” via dsRNA sensing pathways. The ongoing phase II ENCORE601 trial (NCT02437136) evaluates entinostat plus pembrolizumab in multiple tumors. In 76 treated NSCLC patients, the ORR was 9.2% (below the pre-specified threshold), median PFS was 2.8 months (95%CI 1.5-4.1), and median OS was 11.7 months (95%CI 7.6-13.4), with no new toxicities (including irAEs) observed. However, the combination missed its primary efficacy endpoint^[185]. Demethylation of the interferon gamma (IFNG) promoter in circulating T cells served as a predictive biomarker for therapeutic response^[186]. These agents also enhance chimeric antigen receptor T-cell (CAR-T)/NK cell function by reducing exhaustion and promoting memory phenotypes^[187].

In conclusion, a growing body of clinical and preclinical evidence indicates that synergistic treatment modalities can induce potent and durable anti-tumor responses in R/M HNSCC. This approach provides a potential pathway to achieving long-term disease control in historically intractable HNSCC subsets. Future research must focus on optimizing these combinations to maximize efficacy while diligently managing the associated toxicities.