Nasopharyngeal carcinoma: current views on the tumor microenvironment's impact on drug resistance and clinical outcomes

Since EBV is present in virtually all poorly differentiated and undifferentiated non-keratinizing NPC tumor cells but rarely in normal cells, it represents a very specific diagnostic biomarker and an attractive therapeutic target. Numerous studies have demonstrated that EBV can induce genetic instability premalignant cells, promote tumor development through shaping an immunosuppressing TME, and facilitate tumor metastasis and invasion by inducing EMT. Given the oncogenic role of EBV in NPC, multiple clinical trials are initiated to explore the therapeutic value of targeting EBV [52]. In order to target EBV, peptide vaccine and viral vaccine, have been investigated in the context of NPC treatment [53]. Prophylactic EBV-vaccine aims to restrict primary infection and reduce the risk of EBV-associated diseases. To this end, vaccines targeting different EBV antigens have been developed and are under clinical investigation (NCT01094405, NCT01800071). The lytic antigen gp350 is an ideal candidate given its critical role in guiding primary B-cell infection and neonatal infection [54]. Therefore, vaccines against EBV gp350 are expected to prevent primary infection. Notably, the NIH recently announced an early-stage clinical trial to evaluate the safety and immune response of an EBV gp350-Ferritin nanoparticle vaccine. Another type of vaccine targets the latent life cycle antigens such as the EBV nuclear antigens (EBNAs) and latent membrane proteins (LMPs). This approach may be adapted to eradicate latently-infected reservoirs. Clinical trials have been conducted to evaluate the vaccines against two main EBV proteins, EBNA1 and LMP2 [55, 56]. For instance, a phase I clinical trial (NCT01256853) tested a recombinant vaccinia virus encoding an EBNA1/LMP2 fusion protein to boost T cell immunity [57]. The data show that the vaccine can induce a dose dependent EBNA1/LMP2 response with a safe profile, supporting the initiation of a phase II trial.

In addition to cancer vaccines, directing attention towards EBV through Adoptive Cell Therapy (ACT) has emerged as a new option with potential for superior effectiveness and a favorable safety profile. The potential of ACT was first noted by Smith et al., who suggested that it could potentially prevent tumor development and extend patient survival [58]. Targeting EBV-specific antigens LMP1 and EBNA1 with cytotoxic T lymphocytes (CTLs) represents one of the most promising immunotherapies for NPC. LMP1 and EBNA1 are highly immunogenic and successful targeting of EBV-specific antigens with CTLs may have long-lasting effects. Once activated, CTLs can continue to surveil the body for reappearance of EBV-infected cells, reducing the risk of recurrence. CTL-based therapies have fewer severe side effects compared to traditional chemotherapy or radiation therapy.

Other immune cells include dendritic cells (DC) which are potent antigen-presenting cells that play a central role in initiating and regulating anti-viral immune responses. Through genetic engineering DCs can present tumor antigens to naïve CD8⁺ T cells and turn them into tumor specific CTLs [59]. Currently, there are several clinical trials investigating the possibility of ACT against NPC and reported results demonstrated clinical benefits. A recent study by Li et al. reported that ACT is well tolerated following chemoradiotherapy in advanced NPC patients. More importantly, 19 of 20 patients exhibited an objective antitumor response, and 18 patients displayed disease-free survival more than 12 months after ACT, suggesting a sustained antitumor activity achieved by stimulating anti-EBV immune responses [60]. These observations align with results obtained from other clinical trials, including studies involving the combination of ACT with chemotherapy [35] and treatment in patients who did not respond to conventional therapies [36]. Furthermore, a systematic review conducted by Farooqi et al. reached a similar conclusion. In their analysis of 7 phase I/II clinical trials, they found minimal grade ≥ 3 adverse events, with the majority of such events being associated with chemotherapy [37]. Therefore, the combination of adoptive immunotherapy with traditional therapeutic approaches may achieve substantial treatment response while preserving a good safety profile.

ICIs represent a major class of immunotherapy across many cancers. Accumulating data has suggested the great potential of ICIs in the treatment of NPC. Most of the research are focusing on the development of programmed death-1/programmed death ligand-1 (PD-1/PD-L1) and cytotoxic T lymphocyte-associated protein 4 (CTLA-4) inhibitors. Tumor cells have the capacity to engage CTLA-4, thereby suppressing the initiation of T-cell responses and evading immune surveillance. Preclinical investigations have demonstrated that the inhibition of CTLA-4 can restore T-cell functions and facilitate the elimination of tumors. Ipilimumab and tremelimumab have been approved for adjuvant therapy of melanoma [61]. Ipilimumab is currently under phase 2 clinical trials for metastatic, recurrent, or last-staged NPC.CTLA-4 inhibitors are also under clinical investigation in other cancers, including lung cancer, prostate cancer, and NPC [62]. Cadonilimab is a bispecific antibody targeting both PD-1 and CTLA-4. The early phase clinical studies showed that cadonilimab achieved promising antitumor efficacy with a favorable safety profile, leading to its approval for recurrent or metastatic cervical cancer in China. It is now in multiple phase 2/3 clinical trials to investigate the antitumor efficacy in NPC [63]. PD-1 is an immune checkpoint found on T lymphocytes, whereas PD-L1 is present on both tumor cells and immune cells. Zhang and colleagues conducted an analysis of biopsies from 139 patients to assess the relationship between the expression levels of PD-1 and PD-L1 and treatment outcomes. Strikingly, PD-L1 which was identified in 95% of the patients was associated with an unfavorable prognosis. Co-expression of PD-1 and PD-L1 was associated with the poorest prognosis of disease-free survival. This finding agrees with studies in other cancers [64, 65]. Given that PD-1 and PD-L1 play critical roles in tumor survival and immune escape, treatment of NPC with ICIs has become a high interest research field. Many PD-1/PD-L1 inhibitors have advanced from preclinical studies to clinical trials for the treatment of recurrence/metastatic (R/M) NPC, including nivolumab, pembrolizumab, camrelizumab, toripalimab, tislelizumab [51]. The promising data have granted several approvals of PD-1 inhibitors for R/M NPC in combination with chemotherapy. A multinational study of nivolumab monotherapy in 44 patients demonstrated an overall response rate of 20.5%. In addition, the 1-year OS was 59% and 1-year PFS was 19.3%, suggesting that it has promising clinical activity in heavily pretreated R/M NPC. The study also showed that patients with PD-L1 positive tumors had higher response rate compared to those with PD-L1 negative tumors. In a phase 3 clinical study, toripalimab in combination with gemcitabine-cisplatin chemotherapy exhibited a significant improvement in median PFS (11.7 vs 8.0 months), and a 40% reduction in risk of death compared to placebo arm. The addition of toripalimab to chemotherapy had a similar incidence of grade ≥ 3 adverse events, with a more frequent immune-related adverse events (39.7%vs 18.9%) [66]. Notably, toripalimab is the first anti-PD-1 antibody that got approved for R/M NPC in China, representing a breakthrough of immunotherapy in NPC treatment [66]. Tislelizumab was approved by the China National Medical Products Administration for R/M NPC in 2022, based on the significant efficacy in phase 3 RATIONALE-309 trial.

TECs are characterized by their genetic instability, which contributes to the heterogeneity that promotes therapeutic resistance, tumor progression and metastasis. They are also known to regulate cytokine secretion and angiogenesis [78]. Tumor cells require oxygen and nutrients to survive and growth, therefore angiogenesis plays a critical role in tumor progression. TECs process increased capacities of proliferation, migration, and tube-formation in response to growth factors (e.g., VEGF and FGF2) and cytokines (e.g., CXCL1 and CXCL8), which may be attributed to the increased expression of growth factor and cytokine receptors such as TEK and VEGFR [79]. In EBV-positive NPC, LMP1 was shown to induce the expression of Cyclooxygenase-2 (COX-2) via NF-κB pathway. Murono et al. reported that LMP1 can increase the production of VEGF by acting together with COX-2, thereby contributing to angiogenesis in NPC [80]. EBV can also infect endothelial cells using monocytes as a shuttle [81]. The infected endothelial cells display upregulation of EBV lytic genes, promoting inflammation and vascular injury. Cheng et al. found that TECs co-cultured with cancer cells show constitutive activation of PI3K/AKT signaling pathway, resulted in cell survival and tube formation, creating a microenvironment that favors tumor growth [82]. Inhibition of PI3K and COX-2 signaling pathways can reverse the increased tube formation and induce apoptosis of endothelial cells. More importantly, TECs are irregular monolayers with impaired barrier functions compared to the normal endothelial cells, which could lead to vascular leakiness and provide a route for metastasis [83]. With angiogenesis, tumor cells can intravasate from primary site into blood circulation and extravasate into distant organs. Even though millions of tumor cells could be released into the circulation, only a few of them can successfully form distant metastatic nodes. The reason is that inadequate or inappropriate cell–matrix interaction when tumor cells detach from ECM in primary tumor cause them to enter cell-cycle arrest and rapid apoptosis, known as anoikis. Yadav et al. demonstrated that TEC could promote tumor metastasis by chaperoning tumor cells to distal sites [84]. The activated TECs can upregulate the expression of adhesion molecules and bind to tumor cells, thereby protecting the circulating tumor cells from anoikis and facilitating their movement to distant sites.

Among the various cell types present in the TME, Cancer-Associated Fibroblasts (CAFs) have been identified as significant mediators of cancer cell migration and invasion in numerous solid tumors. The primary origins of CAFs are the local normal fibroblasts. Kojima et al. found that tumor cells can produce TGF-β and SDF-1 to initiate and maintain the transformation of normal fibroblasts to CAFs. In addition, Kalluri et al. described that the endothelial cells may undergo endothelial-mesenchymal transition and convert to fibroblast-like cells, which accounts for a large proportion of CAFs [85]. Importantly, EBV-encoded LMP1 has been demonstrated to promote the transition of normal fibroblast to CAFs. Wu et al. reported that extracellular vesicles (EVs) packaged LMP1 can regulate CAF transition via activation of the NF-κB RelA/p65 signaling pathway [86]. Co-culture of EVs-activated CAFs enhances the proliferation, migration, and radiation resistance of NPC cell line HK1. A study by Davis et al. revealed another mechanism in which LMP1 transforms fibroblasts [87]. They showed that conditioned medium from MDCK-LMP1 cells increases cell motility and invasion in both epithelial and fibroblasts by stimulating TGF-β and ERK-MAPK signaling pathways. However, since these pathways have extensive crosstalk with other pathways and regulatory factors, the precise mechanism is unclear and requires further investigation. Functionally, CAFs are documented to enhance angiogenesis in tumors due to their large production of VEGFA within the TME [88]. Furthermore, it has been documented that CAFs maintain the capacity to stimulate angiogenesis independently of tumor cells, as they respond to PDGF-C. [89]. In addition, CAFs can promote tumor angiogenesis by recruiting endothelial progenitor cells via CXCL12/CXCR4 signaling pathway [90]. In NPC, CAFs are found to facilitate tumor migration and invasion, leading to worse prognosis. In support of this notion CAFs promote survival and confer RT resistance in NPC cells by secreting IL-8, which activates NF-kB signaling pathway and reduces DNA damage [91].

TILs are immune cells triggered by the host's immune response to fight against the tumor, including lymphocytes, DCs, macrophages, and mast cells. However, cancer cells possess a wide array of mechanisms to circumvent the immune response through an intricate and ever-changing interactions with TILs, promoting tumor progression, metastasis, and resistance to various drugs [92]. Therefore, TILs serve the host immune defense but can be hijacked by cancer cells to provide local support for the tumor.

TME also includes a non-cellular component known as the extracellular matrix (ECM). All the tumor cells, non-cancerous cells and immune cells are interacting within the ECM. ECM is a complex network composed of fibrous proteins, matrix proteins, and proteoglycans that provides structural support and facilitates tumor progression, intravasation, and metastasis [121]. It serves as a reservoir for various nearby cell-secreted cytokines, hormones, and pro-angiogenic growth factors, playing a significant role in angiogenesis and vascular stabilization [122]. Another key feature of ECM is the ability to help tumor cells escape immune surveillance. It is well-documented that ECM can modulate the activity of lymphocytes, such as migration, recognition/activation, and differentiation [123]. ECM deposition and increased stiffness can reduce lymphocyte replacement, limit T cell infiltration, and therefore providing a more favorable environment for tumor cell growth [124]. Because of the rapid tumor expansion and poor vascularization, the ECM is recognized as a hypoxic environment. In response to these conditions, tumor cells undergo a metabolic transition from oxidative phosphorylation to glycolysis. Recent findings suggest that the tumor-imposed glycolysis metabolism can restrict T cell functions, dampening the mTOR activity, glycolytic capacity, and IFN-γ production, causing T cell hypo-responsiveness during cancer [125].

The secretion of proinflammatory and immunosuppressive cytokines and chemokines can have a significant impact on immune cell activity and modulate the TME for tumor cell proliferation, immune escape, and apoptosis resistance. A multiplexed immuno-based profiling of cytokine markers in NPC [126] revealed elevated levels of IL-6, IL-8, TNF-α, VEGF, CXCL-10, and MIP-3α in NPC compared to healthy individuals. Moreover, the elevated cytokine levels were correlated with EBV DNA and demonstrated worse prognoses for OS. Several studies have shown that IL-6 and its receptor are broadly expressed among NPC cell lines [127]. IL-6 can promote NPC cell migration and invasion, which may be mediated by the regulation of MMP-2 and MMP-9. Moreover, the use of anti-human IL-6R antibody significantly inhibited NPC cell growth and invasion, suggesting blockade of IL6/IL6R is a potential therapeutic target to treat NPC metastasis. Several studies have found that IL-8 promoter T-251 T/A genetic variation is significantly associated to NPC risk [128‐130]. Compared to TT genotype, AA and AT genotypes were highly associated with susceptibility and aggressiveness of NPC, suggesting an important role of IL-8 in NPC progression. Along these lines, IL-8 can stimulate angiogenesis increasing blood supply to the tumor, promoting its growth and survival. The inflammatory cytokine TNF-α is recognized for its dual role in cancer, which depends on factors such as its concentration, duration of exposure, and the presence of other chemokines or cytokines within the TME. Short-term, localized administration of TNF-α has been demonstrated to exhibit anti-tumor effects, whereas prolonged expression of TNF-α can lead to a pro-tumorigenic state [131]. A multivariate analysis by Yu et al. revealed that elevated expression of TNF-α in primary NPC tissues was associated with increased risk of distant metastasis, particularly bone metastasis [132]. NPC cell lines CNE-2 and HK-1, as well as NPC biopsies, express IL-18 and CXCL-10 [133]. IL-18 can induce IFN-γ production from T cells and NK cells, which can lead to the activation of macrophages and other immune cells to secrete chemokines to initiate immune cell recruitment. Additional studies suggest that the elevated IL-18 level in TME can induce PD-1 expression on NK cells, a marker of functional exhaustion [134]. Therefore, despite the higher percentages for infiltrated NK cells, the functional exhaustion mitigates their cytotoxic effect and leads to poor prognosis.

Generally, cancer cells display competitive advantage over other cells for nutrients. It has been shown that EBV-encoded LMP1 can modulate NPC energy metabolism via the FGFR1 signaling pathway, characterized by increased glucose and glutamine consumption, LDHA activity, lactate production, and secretion of HIF-1α [69, 135]. LMP1 acts by upregulating the expression of FGFR1 in NPC cells through multiple downstream targets and signaling pathways. Consequently, the constitutive FGFR1 activation facilitates LMP1-mediated NPC cell transformation, migration, and invasion. These studies suggest that LMP1 promotes aerobic glycolysis through the regulation of metabolic enzymes and related genes. Since TILs also rely on glucose, it is believed that insufficient glucose uptake could contribute to the formation of immunosuppressive TME [136, 137]. Reinfeld and colleagues found that among myeloid cells, T cells, and tumor cells, myeloid cells exhibited the most significant capacity for capturing intratumoral glucose, followed by T cells, and tumor cells [113]. Notably, tumor cells displayed the highest glutamine uptake. Tumor cells take up glutamine and convert it to glutamate through mitochondrial oxidative phosphorylation to produce energy and support their growth and metastasis. Competition for glutamine uptake also exists between tumor cells and immune cells in the TME. Studies have demonstrated that competitive consumption of glutamine by tumor cells could lead to suppressed antitumor immune response in many cancers [138, 139]. Inhibiting glutamine uptake in tumor cells could potentially enhance glucose uptake, indicating that glutamine may serve as the limiting factor in the TME [140].

Lactate is one of the tumor cell metabolites that can serve as a prognostic parameter for metastasis and overall survival of patients. Clinical study showed that high lactate level in the TME inhibits lactate export in T cells, which hinders their metabolism and function [140]. In addition, lactate could interfere with the function of DCs and TAMs, as well as inhibit monocyte migration and cytokine (TNF, IL-6) release [141]. Colegio et al. showed that tumor-derived lactic acid can act through HIF-1α to induce the expression of VEGF and the M2-like polarization of TAMs [142]. In addition, lactate promotes tumor cell proliferation by upregulating the expression of arginase 1 in TAMs. Arginase can promote tumor cell proliferation by depleting arginine from the microenvironment, inducing arginine autotrophy in certain tumor cells, suppressing the immune response, and contributing to an immunosuppressive tumor microenvironment.

Another important metabolite of NPC cells is indoleamine 2,3-dioxygenase (IDO), a cancer-associated enzyme that catabolizes intracellular tryptophan to produce kynurenine [143]. The process of tryptophan degradation and kynurenine production together contribute to immunosuppressive effect, leading to immune escape of tumors. It is well-established that T cells are highly sensitive to tryptophan depletion, which causes a mid-G1 arrest [144]. Kynurenine in the TME can act on the ligand-activated transcription factor aryl hydrocarbon receptor to drive the differentiation of Tregs, tolerogenic myeloid cells and PD-1 upregulation in CD8⁺ T cells [145]. Moreover, selective inhibition of the receptor could impede the progression of IDO-overexpression tumors, making it a new target for immunotherapy.

Exosomes, a subset of EVs that originate from either the endosome or the cell's plasma membrane, play a pivotal role in facilitating cell-to-cell communication. They are considered one of the foremost mechanisms for establishing the diverse characteristics of the TME [146]. To date, accumulated evidence has suggested that exosomes play a significant role in NPC progression, metastasis, and resistance to therapies. In fact, NPC-derived exosomes and CNE2-derived exosomes are enriched in PFKFB3, which facilitates endothelial cell proliferation, migration, and angiogenesis through the activation of the ERK and the AKT pathways [147]. In addition, highly metastatic NPC cells could transfer EGFR-rich EV to poorly metastatic NPC cells to enhance their tumor metastasis potential [148]. Mechanistically, the EGFR-rich EV could induce EGFR upregulation and ROS downregulation via the PI3K/AKT signaling pathway. NPC cells infected with EBV can stimulate tumor progression by using the exosome system for the transfer of signaling molecules, viral proteins, and microRNAs. EBV LMP1 oncoprotein, was reported to be secreted from EBV-positive NPC cells via exosomes to modulate the TME through intracellular trafficking [149]. Studies showed that LMP1 can promote EV formation by upregulating the levels of syndecan-2 and synaptotagmin-like-4 through activation of the NF-κB signaling pathway [150]. This is further supported by investigations revealing that syndecan-2 can interact with syntenin to promote the formation of EVs, while synaptotagmin-like-4 can regulate the release of EVs. NF-κB activation is required to initiate the production of LMP1-packaged exosome, which contribute to the epithelial-mesenchymal transition (EMT) potential of EBV-negative recipient NPC cells [151]. Furthermore, inhibition of NF-κB significantly repressed exosome LMP1 secretion and limited NPC lung metastasis in nude mice. Other EBV-related exosomes include LMP2A [152], BART1 [153], HMGA2 [154]. These exosomes are found to promote tumor metastasis and modulate immunosuppressive TME.

Recently, additional evidence confirmed that TME-derived exosomes are involved in NPC development. Shi et al. found that mesenchymal stem cells (MSCs) secrete FGF19-rich exosomes [155]. Co-incubation study suggested that NPC cells can pick up the exosomes, leading to enhanced cell proliferation, migration, and tumorigenesis. Mechanistically, FGF19-rich exosomes can activate the FGF19-FGFR4-dependent ERK signaling cascade and modulate EMT process. Another study reported that MSCs could be used as a therapeutic tool in NPC. MSCs transfected with miR-34c, a tumor suppressor miRNA, could produce exosomes that attenuate NPC proliferation, migration, invasion and EMT process [156]. Notably, miR-34c-overexpressing exosomes significantly sensitized NPC cells to RT.

NPC is a cancer whose pathogenesis is highly correlated with EBV, therefore EBV protein and miRNA are thought to be one of the mechanisms of chemoresistance in NPC patients [158]. LMP1, a transmembrane protein of the EBV virus, is found to be expressed in 70% of NPC. It can induce chemoresistance by sequestering Pim-1 in the cytoplasm or activating the PI3K/AKT signaling pathway [159, 160]. Yang et al. reported that LMP1 can upregulate the expression of miRNA-21 via the PI3K/AKT/FOXO3a signaling pathway. In turn, high level of miR-21 could lead to cisplatin resistance through the inhibition of pro-apoptotic factors PDCD4 and FasL [161]. Meanwhile, studies also show that LMP1 can inhibit the DNA damage response through DNA-PK/AMPK signaling, and LMP1 positively regulates the expression of the cancer stem cell (CSC) marker CD44 and several stemness-associated genes (Nanog, Oct4, Bmi-1, and SOX2), which induces the development of CSCs and contributes to RT resistance in NPC [162, 163].

Among the EBV virally-encoded miRNAs, some are associated with chemotherapeutic drug resistance, such as miR-BART22, miR-BART7-3p, and miRBART5; while others are associated with RT resistance, miR-BART4, miR-BART7, and miR-BART8-3p [164‐166]. Studies have revealed that miR-BART22 promotes MYH9 expression through the PI3K/AKT/c-Jun pathway, leading to EMT and cisplatin resistance in vitro and in vivo [167]. Others have reported that miR-BART7-3p inhibits PTEN, activates the PI3K/AKT/GSK-3β signaling pathway, and promotes the expression and nuclear aggregation of β-catenin, which in turns facilitates EMT [168]. Similarly, another study demonstrated that miRBART7-3p targets SMAD7 and triggers the TGF-β pathway, inducing cancer stem-like cells and chemoresistance to fluorouracil, cisplatin, and paclitaxel [169]. By downregulating p53 and simultaneously up-regulating apoptosis regulator, miR-BART5 expression reduces the sensitivity of NPC cells to pro-apoptotic drugs such as doxorubicin and etoposide [170].

Cisplatin is one of the standard treatments for patients with advanced NPC. Cisplatin has the capability to engage with the DNA within tumor cells, creating both intra-strand and inter-strand links. This interference disrupts DNA replication and transcription processes, ultimately leading to the destruction of tumor cells [171]. Consequently, abnormal DNA damage repair function is one of the main mechanisms of cisplatin resistance. FOX family is a group of evolutionarily conserved proteins that can bind to DNA to regulate transcription [172]. The transcription factor FoxM1 is a critical proliferation-associated transcription factor that is widely spatiotemporally expressed during the cell cycle [173], and FOXM1, FOXC2 and FOXQ1 have been reported to be overexpressed and involved in the occurrence of EMT and chemoresistance in NPC [174, 175]. Studies have shown that downregulation of FoxM1 inhibited MRN-ATM-mediated DNA repair and subsequently increased the sensitivity of NPC cells to cisplatin. Another study revealed that MRN-ATM-mediated DNA repair was inhibited by the downregulation of FOXM1, which subsequently increased the sensitivity of NPC cells to cisplatin [176]. In addition, a specific non-coding RNA, the circular RNA CircCRIM1, was found to be overexpressed in NPC cells to suppress the inhibitory effect of miR-422a on FOXQ1, thereby inducing chemoresistance towards docetaxel [177]. It has been reported that circRNAs are abundant and stably expressed in exosomes and play critical roles in mediating chemotherapy resistance in various cancers [178, 179]. Therefore, further study is warranty to study the role of other circRNAs in NPC drug resistance.

The ATP-binding cassette (ABC) transporter is a superfamily of transmembrane proteins, which are widely expressed in cell membranes [180]. Its involvement in anti-cancer drug efflux is a common cause of chemoresistance [181]. Currently, ABCB1/P-gp and ABCC1/MRP1, ABCC5, and ABCG2/BCRP have been found to be overexpressed in NPC drug-resistant cells [174, 182]. ABC transporters have the capacity to expel a broad spectrum of anticancer drugs from within cells, giving rise to a multidrug resistance phenotype in cancer cells [183]. For instance, ABCB1 and ABCG2 are key resistance factor to chemotherapeutic drugs paclitaxel and docetaxel, as well as EGFR inhibitors gefitinib and erlotinib. Notably, there is a strong association between cancer stem cells and the emergence of chemoresistance [184, 185], and in fact both ABCG2 and ABCB1 are expressed in cancer stem cells. The isolated cancer-stem like side population cells from NPC demonstrated increased expression of ABCG2 and the anti-apoptotic factor Bmi-1. These factors play a role in enhancing multidrug resistance and elevating the survival rate of tumor cells [186]. Therefore, the evidence provides a reference for the use of chemotherapy in NPC.

Another common chemo-resistance mechanism is the inhibition of apoptosis in NPC cells [187, 188]. Eukaryotic translation initiation factor 4E (eIF4E) has long been recognized as a quantitative restriction initiator of mRNA translation. Genome-wide analysis by Truitt et al. showed that elF4E dosage plays a critical role in the translational program induced by oncogenic transformation [189]. Compared to parental cells, high expression of eIF4E was found in cisplatin-resistant NPC cells, which promoted the translation of c-Myc, cell cycle protein D1, and VEGF to circumvent drug-induced apoptosis. In addition, in vitro and in vivo studies have confirmed that TBL1XR1, an essential transcriptional cofactor, can promote anti-apoptosis activity by activating the NF-κB pathway. Hence, its high expression can trigger NPC cells to become resistant to cisplatin-induced apoptosis [190].

CSCs are cells within a tumor that can self-renew and form heterogeneous cell populations that lead to heterogeneous branching of cancer cells in the tumor. It can be characterized using different cancer stem cell markers [191]. BEX3, a receptor-associated protein for the CSC marker CD271, is overexpressed in cisplatin-resistant NPC cells and is associated with drug resistance through the activation of the MAPK pathway [192]. In the cisplatin-resistant NPC xenograft, treatment with nontoxic level of cisplatin led to significantly increase in BEX3 level. In addition, BEX3 can induce the expression of OCT-4, another cancer stem cell marker, which contributes to chemoresistance through activation of the AKT pathway and the STAT3 pathway [193]. In addition, CSCs can also induce drug resistance by scavenging ROS from oxidative stress, activating anti-apoptotic pathways, and protecting the microenvironmental niche [194]. The phenotype of CSCs may be closely related to EMT, which renders epithelial cells mesenchymal properties characterized by the loss of epithelial markers (e.g., E-cadherin, α-cadherin) and the gain of mesenchymal markers (e.g., Vimentin, fibronectin, N-cadherin). EMT process not only enhances tumor cell migration and invasion, but also induces drug resistance to chemotherapy.

Recently, the TGF-β pathway has been implicated in chemoresistance through EMT or maintenance of tumor-induced cell heterogeneity [195]. In vitro study shown that shRNA-mediated downregulation of TGF-β induces phosphorylation of PTEN and AKT, thereby increasing cisplatin resistance. In contrast, overexpression of TGF-β sensitized NPC cells to cisplatin. In fact, out of the four identified miRNAs associated with an increased risk of advanced nasopharyngeal cancer, MiR-449b has the capacity to modify the TGF-β pathway, leading to the development of cisplatin resistance in NPC [196].

Exosomes are EVs that carry information including proteins, mRNAs, miRNAs, and function as intercellular communication [197]. Co-incubation of doxorubicin-resistant human microvascular endothelial cells-derived exosomes with NPC cells can induce the proliferation, migration, EMT, and the development of chemotherapy resistance in NPC cells [198]. Moreover, the exosomes induced the expression level of ABC transporters ABCB1 and ABCC1 that are capable of causing multidrug resistance. In addition, other studies have demonstrated that LMP1-positive exosomes extracted from EBV-infected NPC cells can promote chemoradiation resistance [86]. The specific regulatory pathways related to drug resistance remain to be elucidated and warrant future studies.

Despite the encouraging efficacy of anti-PD-1/PD-L1 therapy, many patients failed to response or developed acquired resistance towards ICIs [199, 200]. Potential drug resistance mechanisms toward ICIs can result from impaired T-cell proliferation and functions. In fact, genetic alteration in granzyme, gasdermin, and IFN were enriched in refractory NPC tumors [201]. Specifically, the granzymes GZMD and GZMB are known to induce apoptosis of cytotoxic T cells and NK cells. Cancer cells resistance to ICIs treatment may occur through the dysfunctional pyroptosis pathway and weakened cytotoxic lymphocyte function. Using an NPC-PDX mouse model to test the combination of anti-PD-1 antibody nivolumab and anti-CTLA-4 antibody ipilimumab [202]. Interestingly, human proinflammatory cytokines including IFN-γ and IL-6 were significantly upregulated in plasma, accompanied by a decrease in CD4/CD8 ratio. Moreover, the isolated TILs were responsive to stimulation but remain insufficient to induce antitumor effect in vivo. Drug resistance was observed in this NPC-PDX model, which could be mediated by other inhibitory checkpoints such as LAG3 and TIM3 as well as immunosuppressive molecules in the TME [202].

Others have reported activation of alternative immune checkpoints as drug resistance mechanisms to PD-1/PD-L1 blockade. The alternative pathways include LAG3, TIM3, TIGIT/CD155, and VISTA [203]. In addition, adenosine receptor signaling can suppress various immune cells to create an immunosuppressive niche. Chen et al. reported that tumor cells could develop drug resistance against PD-1/PD-L1 blocking antibodies by upregulating CD38, which is induced by all-trans retinoic acid and IFN-β in the TME [204]. CD38 suppresses CD8⁺ T-cell function via adenosine receptor signaling and therefore represents a major mechanism of acquired immunotherapy resistance. Long non-coding RNAs (lncRNAs) with immune-related functions in the TME have been observed to influence both immune cell infiltration and the response of cancer cells to anti-PD-1 immunotherapy [183]. In a comprehensive analysis of the whole genome expression, Tang and colleagues discovered a notable correlation between the lncRNA AFAP1-AS1 and the expression of PD-1 in patients with NPC [205]. Co-expression of AFAP1-AS1 and PD-1 in TILs predicted the poorest prognosis and distant metastasis at relapse.

The resistance to ICIs could also be attributed to the establishment of immunosuppressive tumor microenvironment resulting from immune modulatory effects of CAFs, TAMs, and myeloid-derived suppressor cells (MDSCs) [206]. Angiogenesis factor can directly promote immune suppression by suppressing the function of APCs and immune effector cells, or by enhancing the effect of Tregs, MDSCs, and TAMs [207]. Those immunosuppressive cells can reciprocatively drive angiogenesis to create a viscous cycle of impaired immune activation. Interestingly, expression of annexin A2 on NPC cells can lead to immunosuppressive responses by interacting with DCs [208]. Annexin A2 acts as a ligand for DC-SIGN DCs and therefore activates DCs to release extremely high levels of IL-10. The release of IL-10 into TME causes immunosuppressive responses including CD8⁺ T cell dysfunction, Treg expansion, and inhibition of proinflammatory IL-12 [59].

The understanding of ACT resistance mechanism remains limited, and more research is needed given the important impact TME has on antitumor immunity. Along these lines, increased frequency of MDSCs can affect the outcome of EBV-specific T cell therapy in NPC [209]. Indeed, MDSCs can expand in response to cytokine stimulation under pathological conditions and cause antigen specific or non-specific suppression of T-cell response [210]. In addition, the limited ACT effectiveness can be caused by a failure of tumor specific CTLs to expand or to survive in vivo. Tumor cells expressing Fas can utilize the Fas/FasL pathway to induce T cell apoptosis and escape immune defenses [211]. Dotti et al. reported that EBV-CTLs are highly sensitive to Fas/FasL-mediated apoptosis and that knockdown of Fas significantly improve the efficacy of EBV-CTLs in NPC and Hodgkin lymphoma [212].