Among the potential targets for CAR-T therapy in HNSCCs, the ErbB family (also known as the epidermal growth factor receptor (EGFR) family), holds significant promise. This family is composed of EGFR, Human Epidermal Growth Factor Receptor 2 (HER2 or ErbB2), ErbB3 (HER3), and ErbB4 (HER4) (17, 18). All these receptors can be activated by binding to specific ligands. Upon this, the receptors form homo- or hetero-dimers, which activates their intrinsic tyrosine kinase activity and lead to multiple downstream signaling pathways that play a critical role in regulating cell growth, differentiation, survival, and migration.

The cytotoxic function of CAR-T cells targeted to EGFR against hypopharyngeal squamous cell carcinoma was verified as well in a preclinical study conducted by a team which had already successfully built and verified EGFR-CAR-T-cells (19). This research indicated a notable increase in cytokine secretion following the incubation of EGFR-CAR-T-cells with target tumor cells. This finding not only demonstrated the functional activity of these engineered T cells but also highlighted the enhanced cytotoxic effect specially against FaDu cells, a hypopharyngeal squamous cell carcinoma cell line. The EGFR-CAR-T cells exhibited a target cell lysis rate of 52.66%, further underscoring their potent therapeutic potential of treating HNSCCs.

In addition, T4 immunotherapy, a gene-modified immune cell treatment that specifically targets the ErbB homo- and heterodimers, has been suggested as a solution to the problems that CAR-T cell therapy encounters when treating solid tumors (20, 21). This immunotherapy is recommended because ErbB family members contributed significantly to the development of various types of cancer, including HNSCCs, and selectively targeting ErbB family members using monoclonal antibody therapy exhibits beneficial curative properties (22). It describes the engineering of patient T-cells by retroviral transduction to co-express a chimeric antigen receptor named T1E28z and a chimeric cytokine receptor named 4αβ (17). The T1E28z CAR has a promiscuous ligand of ErbB termed T1E that is connected to CD3ζ through a CD28 hinge/transmembrane and intracellular domain. Being positioned upstream of the CD28⁺ CD3ζ domain, T1E28z confers specificity on T cells against all EGFR homo- and heterodimers, the ErbB 2/3 heterodimer and all ErbB4 homo- and heterodimers. 4αβ chimeric cytokine receptor is created by fusing the ectodomain of IL-4 cytokine to the β chain used by IL-2 and IL-15. An ex vivo expansion system of CAR-T cells developed by Wilkie et al. suggested that a potent method for rapid proliferation of CAR-T cells is to cultivate them with IL-4 over a period of 7–10 days (23). One of the most challenging barriers of applying T4 immunotherapy is on-target off-tumor toxicity as ErbB family members are expressed in many healthy tissues. Pre-clinical evidence showed a possibility of reducing this toxicity by using the intra-tumoral route and a phase-I clinical study that recruited 13 patients who had been diagnosed with advanced HNSCCs has been developed and demonstrated the safety of intra-tumoral T4 (24). The whole trial is still under recruitment and expected to be completed by 2023 (NCT01818323) (25).

Although CD70 expression was high in 19% of HNSCC tumor biopsies, Park et al. utilized a retroviral human CD70 CAR construct to generate CD70 CAR-T and found that it presented the capacity of efficiently eliminating CD70-positive HNSCC cells, which might be related to the transduction efficiency, the CD4⁺/CD8⁺ T cell ratio, and a different response to IL-2 stimulation. Consistently, the infusion of CD70-targeted CAR-T cells into patients with clear-cell renal carcinoma showed an outcome of a disease control rate of 76.9%. Among the 14 patients enrolled, one patient had a long-lasting full remission that is still going on as of their 2-year follow-up (26). This outcome further encourages the feasibility of CD70-targeted CAR-T therapy for CD70-positive HNSCC patients.

The MUC1 gene was discovered in human breast cancer with high expression. It is expressed in barrier epithelia lining (27) and skin, immune cells and reproductive organs (28). Therefore, it plays multiple roles in protecting barrier epithelia as well as species propagation in mammals (27). It has been reported that CAR-T cells targeting the tumor MUC1 inhibit triple-negative breast cancer growth in murine models whereas there is minimal damage to normal breast epithelial cells (29). Furthermore, it has been demonstrated that MUC1 has a higher expression in HNSCCs than in adjacent normal tissues in a multivariate analysis (30). Since there was evidence that IL-22 contributed to inducing an elevation of MUC1 expression in colon cancer patients (31), Mei et al. constructed MUC1‐IL22 CAR-T cells and found they exerted more effective cytotoxic function than MUC1 CAR-T cells in vitro and in vivo (3).

CD98hc, which is a strong related to T cell activation, has been proved to be highly expressed on the surface of radioresistant HNSCC cells (32). CD98hc-targeted UniCAR-T cell was developed and showed ideal ability of tumor cell elimination in a 3D setting of HNSCCs tumor spheroid (33). In addition, the infiltration capability of the UniCAR T cells was improved in the presence of the CD98hc targeted system and a synergistic effect was observed after treatment with fractionated irradiation followed by CD98hc-redirected UniCAR-T delivery.

Apart from these published results, several clinical trials of CAR-T therapy for the treatment of patients with head and neck cancer are underway, targeting novel antigens such as EpCAM, LAMP1, PSMA. We summarized in Table 1 .

Human tumor antigens can be classified into viral antigens, neoantigens that are derived from new, mutated genomic sequences, tumor-specific antigens (TSAs) that derive from unmutated proteins and that are either unique or specific for tumors, and tumor-associated antigens (TAAs) that are differentially expressed in tumor cells and less so in normal cells (34). The optimal target antigens for CAR-T therapy are those tumor-specific antigens (TSAs) exclusively expressed on malignant cells. Specific tumor-specific antigens (TSAs) are neoantigens presented on major histocompatibility complex (MHC) molecules that are suitable for personalized cancer vaccines instead of general CAR-T cell therapy (35). However, these antigens are highly heterogeneous among patients suffering from the same type of tumors, which causes challenges for CAR-T cell therapy. Schumacher, et al. suggest cancer exome-based identification of neoantigens for each patient and assess T cell recognition before generating CAR-T cells (36). Compared to hematological malignancies, targeting TAAs in solid tumors elicits stronger on-target off-tumor effects (37, 38).

When these two antigens can`t achieve satisfactory outcomes in vitro, other options for choosing target antigens for CAR-T might be considered. One alternative antigen is fibroblast activation protein (FAP) in HNSCCs. A novel strategy of CD40 agonist targeted to FAPα combined with radiotherapy has been proven to remodel the TME, activate the CD8⁺ T cells without the systemic toxicity in murine HPV-positive head and neck tumors. This model increased the overall survival and induced long-term antitumor immunity (39). A previous phase I clinical trial of malignant pleural mesothelioma locally treated with autologous anti-FAP-targeted CAR T-cells (NCT01722149) has been reported that no significant toxicity was observed and the activity of the patient’s redirected T cells was confirmed in vitro (40). Therefore, FAP-targeted CAR-T cells combined with other immune therapies could be a potential strategy for HNSCC patients.

The access of CARs to tumors and their infiltration into the tumor microenvironment (TME) are prerequisites for successful therapies. As with other solid tumors, treatment of HNSCCs must overcome two main barriers: the vascular and stromal barriers. Remodeling of the vascular system hinders immune cells’ ability to penetrate blood vessels to reach the tumor site (41). In the meantime, the stromal barrier of solid tumors consisting of abnormal extracellular matrix (ECM) or mismatched chemokine receptors of immune cells and tumor-derived chemokines also blocks T cell migration (35). Cancer-associated fibroblasts (CAF) within the TME facilitate the accumulation of abnormal ECM around the tumor, eventually creating a compact fibrotic environment. Activating TGF-β can trigger CAF to release extra ECM proteins, limiting T cell’s mobility. Therefore, researchers suppress TGF-β signaling and CAF differentiation, reducing EMC protein secretion and eventually encouraging immune cell infiltration into tumor tissues by inhibiting NOX4, a downstream molecule of TGF-β signaling (42). In the study of Trastuzumab-derived HER2-specific CARs for treating trastuzumab-resistant breast cancer, CAR-T cells can infiltrate the tumor matrix and eliminate tumors that are not accessible to antibodies (43).

In addition, Heparanase (HPSE) can degrade heparin sulfate proteoglycan in the ECM. CAR-T cells expressing HPSE were therefore designed to decompose the ECM and improve T cell infiltration and antitumor efficacy (44). In order to enhance CAR-T cell trafficking, direct injection of CAR-T cells into the tumor site could prevent cells’ circulation through the vascular transport system. For instance, in a study of IL13Rα2-targeted CAR-T cells against glioblastoma, local intracranial delivery of CAR T cells displayed superior anti-tumor efficacy as compared to intravenous administration (45). A similar scenario was reported in murine models with atypical teratoid/rhabdoid tumors (ATRTs) in the treatment with B7-H3-targeted CAR-T cells either intracerebroventricularly (ICV) or intratumorally (IT) compared to intravenously (IV) (46). Even though the lack of clinical lymphodepletion in murine models and the functional distinction between murine and human B7-H3, evidence of intratumaral delivery of CAR-T from animal experiments might provide fundamental data to support clinical trials.

A novel approach for regional delivery of CAR-T cells to solid tumors has also been reported when using implantable biopolymer scaffolds to eliminate heterogeneous tumors. The concurrent transfer of CAR-T cells and the stimulator of INF genes (STING) agonist has the potential to enhance the immune response to tumor cells that do not express the CAR-specific target molecule while locally providing growth factors to CAR-T cells can protect them from the hostile TME during the initial phase of tumor priming (47).

Various immunosuppressive substances and cells that decrease the elicited antitumor immunity are produced and recruited in TME by cancer cells, cancer-associated fibroblasts, and stromal cells (48). In a previous study, the locomotion of CAR-T cells in the solid tumor microenvironment was carefully addressed (49).

Local delivery of cytokines and chemokines by CAR-T cells within the TME, such as IL-22, IL-18, IL-12, and the combination of CCL19 and IL-7, or checkpoint blocking agents, can help in overcoming barriers to T cell infiltration and functioning (50). Besides the exogenous addition of IL-22, CAR-MUC1-IL22 T cells, the fourth-generation CAR-T cells that could release the IL-22 cytokine, may also boost recognition and cytotoxic potential against HNSCC (3).

T-cell exhaustion is significantly influenced by the activation of the transcription factor family NR4A, which is connected to the production of immunosuppressive molecules in solid tumors such as PD-1 and TIM3. CAR-T cells can exhibit better tumor-killing activity and enhanced persistence with the knockout of NR4A (51).

Toxicities associated with CAR-T cell therapy can be classified into two major categories. (1) “On-target off-tumor” toxicity results from specific interactions between the CAR and its target antigen expressed by non-malignant cells. (2) general toxicities, which are related to T cell activation and subsequent systemic release of high levels of cytokines, including cytokine release syndrome (CRS), immune effector cell-associated neurotoxicity syndrome (ICANS), hemophagocytic lymph histiocytosis (HLH), and/or macrophage activation syndrome (MAS). ICANS and CRS are the two most severe and unanticipated reactions to CAR-T cell therapy within the category of general toxicities, as reported in the literature (52). CRS, a systemic inflammatory response, is the most common adverse effect caused by CAR-T cell therapy. The most prevalent symptom of CRS following CAR-T cell infusion is fever, which can be accompanied by nausea, fatigue, hypotension, and cardiac dysfunction (53). ICANS is linked with the disturbance of the blood-brain barrier and the rise of cytokine levels in the cerebrospinal fluid. Usually, ICANS arise alongside or after CRS (54). The clinical signs of ICANS linked to CAR-T cell therapy cover encephalopathy, memory impairment, seizures, speech impairment, tremors, headaches, language disturbance, and motor weakness (55, 56). In contrast to CRS, the exact underlying pathophysiology of ICANS is currently vaguely comprehended (57).

Several strategies have been developed to address the challenge of toxicity, such as multitarget CARs, affinity-optimized CARs, inhibitory CARs (iCARs), incorporation of suicide genes into CAR-T cells, and utilization of transient RNA-expressing CARs (58). Kosti et al. developed a CAR-T cell capable of detecting hypoxia. This approach relies on the expression of pan-ErbB-targeted CARs in the hypoxic site. The results demonstrate that CAR molecule expression can only be detected in the hypoxic tumor site of HNSCC, while it is undetectable in normal tissues (59).

The manufacturing process for CAR-T takes about three to four weeks, which can be a fatal waiting time for patients whose disease might continue to be aggressive. Adding bridging chemotherapy prior to CAR-T is currently regarded to be an essential to eradicate tumor cells and boost the efficacy of CAR-T cell therapy in hematologic malignancies (35, 72) Furthermore, the combination of CAR-T therapy with chemotherapy reduces immune suppressor cells in the TME and enhances T cell trafficking to tumors with upregulation of chemokines including CCL5, CXCL9 and CXCL10 (73). However, the combination of chemotherapy with CAR-T cell therapy poses challenges. For instance, the susceptibility of naïve T-cells towards chemotherapy may decrease CAR-T cell efficacy. Besides, chemo-agent cisplatin via HSF1-HSP90 axis enhances the expression of functional PD-L1 in oral squamous cell carcinoma (74). Therefore, the combination effects may be dependent on the dose and the treatment schedule (73).

Radiation therapy (RT) also plays a vital role in the management of HNSCCs and is the primary treatment for nasopharyngeal carcinoma. Radiation modulates tumor immune microenvironments via the release of inflammatory mediators to attract and mediate immune cells. On the other hand, TAAs produced by irradiated tumor cells can be captured by antigen-presenting cells (APCs) (75). Furthermore, activated T cells circle and sequentially have an effect on unirradiated metastases, which is called the abscopal effect (76). With the increasing use of CAR-T, synergistic effects highlight a potential strategy. Further preclinical evidence has been described that NKG2D-based CAR-T cells in mouse glioma models demonstrated efficacy, long-term persistence, and synergistic activity in combination with radiotherapy (77).

Although curative radiotherapy is challenging as a subset of patients are radioresistant (78), combination of conventional radiotherapy with CAR-T therapy turns out encouraging outcomes. Mechanically, RT delivered after CAR T cells resulted in expansion of TCR repertoires, which is an abscopal-like response in a case of a patient with multiply relapsed, refractory myeloma (79). Low-dose RT delivered before CAR T cells supports CD19-targeting CAR T cells in a murine leukemia model (80). In particular, Köseer et al. constructed a 3D culture system of HNSCCs tumor spheroid and demonstrated that the antitumor effect of a CD98hc-retargeted UniCAR-T system was not impaired by irradiation (33). From fractionated irradiation followed by UniCAR-based immunotherapy in Cal33 radioresistant spheroids, they observed that increased UniCAR T expansion with increased EGFP signal and enhanced UniCAR T function with increasing granzyme B and IFN-γ after this combination therapy, indicating an intriguing strategy for the therapy of patients with high-risk HNSCCs (32).

The feasibility of therapeutic mRNA vaccination against cancer was first demonstrated in 1995 (81). Since then, plenty of studies on mRNA-pulsed DC vaccines have emerged. There are two main mRNA-based immunotherapy approaches. The first strategy is to directly deliver mRNA, involving naked mRNA or mRNA complexed with lipid nanoparticle (LNP), by different administration routes (local injection, such as intramuscular, subcutaneous, and intradermal injection, are major routes for mRNA vaccines in clinical studies). The latter strategy is based on the administration of mRNA-modified DCs or chimeric antigen receptor (CAR) T cells. mRNA vaccines have been utilized in the treatment of aggressive and less accessible solid tumors, however, DC mRNA vaccines are determined by multiple factors. mRNA-encoded proteins are processed to antigenic peptides that associate with MHC class I molecules targeting CD8⁺ T cells only, not CD4⁺ T helper cells, to boost antitumor responses. Additionally, posttranscriptional modifications of peptide epitopes (82, 83) and the duration of the antigen peptide-loaded DC presented, DC-cell density, injection site might be considered.

A recent study documented that Reinhard et al. introduced the developmentally regulated tight junction protein claudin 6 (CLDN6) as a CAR target in solid tumors and demonstrated that CAR-T cell-amplifying RNA vaccine was a suitable approach in treating with CLDN6⁺ lung tumors, without cytokine release syndrome (CRS) (84). Furthermore, this RNA vaccine is efficient in in vivo proliferation and in improving CAR-T cell persistence. Based on data from the Cancer Genome Atlas (TCGA), CLDN6 expression was significantly upregulated in head and neck squamous cell carcinoma (HNSC) (85). Recently, the ongoing clinical trial (EudraCT No. 2019-004323-20) holds a promising future in using CAR-T cell immunotherapy to treat HNSC patients although barriers including high tumor heterogeneity and the maintenance of T cell immune responses in the TME have yet to be addressed.

It is well known that immune checkpoint molecules interactions with their ligands prevent recognition between T cells and antigen-presenting cells (86–88). PD-1 and CTLA-4 display distinct functions since they utilize different structural motifs to bind and recruit phosphatases for signal blockade. Blocking of PD-1-PD-L1 but not CTLA-4-B7 interactions enhances tolerized T cell interactions with antigen-bearing DCs and facilitates the phosphorylation of TCR signaling molecules (89). In an orthotopic mouse model of pleural mesothelioma, high doses of both CD28- and 4-1BB-based second-generation CAR T cells eradicated tumors effectively, whereas PD-1 upregulation in the tumor microenvironment inhibited T cell function. Using PD-1 antibody, PD-1 shRNA blockade, or a PD-1 dominant negative receptor restored the function of CD28 CAR-T cells (90). Clinically, a recent phase I clinical study conducted by Adusumilli et al. provided satisfactory median overall survival results for using pembrolizumab and CAR-T together when treating patients with malignant pleural disease (91). An alternative method of targeted delivery of a PD-1-blocking scFv by CAR-T cells not only enhanced the anti-tumor efficacy of CAR-T but also improved bystander tumor-specific T cells in xenogeneic mouse models of PD-L1⁺ hematologic and solid tumors (92). Clinically, a recruiting phase I trial of T4 immunotherapy of head and neck cancer (NCT01818323) is designed to use panErbB-specific CAR T cells intratumorally and involves the administration of nivolumab. This clinical trial might provide fundamental evidence for the toxicity and persistence of CAR-T combined PD-1 inhibitor. Interestingly, Lee et al. evaluated CD19-targeting CAR-T cells in the context of four different checkpoint combinations-PD-1/TIM-3, PD-1/LAG-3, PD-1/CTLA-4, and PD-1/TIGIT-and found that CAR T cells with PD-1/TIGIT downregulation exerted synergistic antitumor effects while dual downregulation of PD-1/LAG-3 severely damaged antitumor activity. Mechanically, downregulation of PD-1 enhances short-term effector function, whereas downregulation of TIGIT is primarily responsible for remaining a less exhausted state. The efficacy and safety of PD-1/TIGIT-downregulated CD19-targeting CAR-T cells are being assessed in patients with large B cell lymphoma (NCT04836507) (93). The utilization of dual immune checkpoint blockades might provide novel perspectives for the treatment of patients with solid tumors in CAR-T therapy. However, it is vital to carefully monitor the potential risk such as CRS and ICANS.

Conversely, CD80/CD86-targeted CAR-T (CTLA4-CAR), which comprises the extracellular and transmembrane portions of human CTLA4, the cytoplasmic region of human CD28, and the intracellular domains of human CD3z, was designed to destroy CD80/CD86-associated tumor cells in vitro and vivo. CTLA4-CAR T cells promoted IL-2 and IFN-γ secretion and suppressed tumor growth in xenograft models of malignant B cells. Furthermore, CTLA4-CAR T cells were found to infiltrate in residual tumors and are cytotoxic to myeloid-derived suppressor cells (MDSCs) without signs of severe CRS in murine models (94). Similarly, CTLA-4 tail fusion has superior anti-tumor efficacy in a relapsed leukemia model (95).

Apart from immune checkpoints, co-stimulatory members of the TNFRSF molecule (4-1BB, OX40, CD27, CD40, HVEM, and GITR) display unique functional roles in CAR-T therapy for solid tumors. CD40 ligand-modified CAR T cells displayed superior antitumor efficacy, promoted upregulation of co-stimulatory markers CD80 and CD86 on CD40⁺ lymphoma cells, educated antigen-presenting cells, enhanced recruitment of immune effectors, and mobilized specific tumor-recognizing T cells without outward signs of toxicity (96). Mata et al. introduced an inducible costimulatory (iCO) molecule consisting of a chemical inducer of dimerization (CID)-binding domain and the MyD88 and CD40 signaling domains into CAR-T and found that the CAR-T cells improve the efficacy of CAR T-cell therapy for solid tumors (97). Golubovskaya et al. designed CAR with a glucocorticoid-induced TNFR-related protein (GITR) co-stimulatory domain and recorded that EGFR-GITR-CD3 CAR-T cells were cytotoxic against EGFR⁺ pancreatic and ovarian cancer cells with higher levels of IFN-γ (98). These results suggest co-stimulatory molecules could be a potential option to enhance the activity of CAR-T in HNSCCs.

Oncolytic adenoviruses are one of the most employed oncolytic viruses as they can elicit immune-stimulating signals that can substantially eliminate local immunosuppression, cause tumor cell lysis, and improve antitumor immune responses (99). However, limitations such as short persistence, host antiviral immune response, are impeding the utilization of the clinic (100). To overcome them, helper-dependent adenoviral vectors (HDAds) are deleted of all viral genes and can be armed with approximately 34 kb of therapeutic transgenes, which have demonstrated better efficacy for CAR-T immune therapy (101). Further evidence suggests that the feasibility of encoding the PD-L1 blocking antibody and IL-12p70 (CAd12_PDL1) combined with HER2-CAR-T cells infusion in HNSCC xenograft and orthotopic models. Results showed controlled primary and metastasized tumor growth and prolonged survival (102). In addition to this, a phase I trial involving patients with HER2-positive cancer including HNSCCs, utilizing CAdVEC (a genetically modified oncolytic viral strain of human adenovirus (Ad) with potential immunostimulating and antineoplastic activities) coupled with HER2-specific CAR-T cells, is in recruitment (NCT03740256). We have summarized different strategies to enhance the efficacy of CAR-T cell therapy (Figure 3).