Adult stem cells (ASCs), also known as somatic or resident stem cells, are undifferentiated cells found in various tissues with limited self-renewal ability and differential potential (32). Examples of ASCs include HSCs, neural stem cells (NSCs), and mesenchymal stem cells (MSCs).

HSCs are predominantly found in bone marrow, peripheral blood (PB) and umbilical cord blood (UCB). They play a crucial role in regenerating blood cells and treating haematologic malignancies through HSC transplantation (HSCT), also known as bone marrow transplantation (33). This procedure, pioneered by E. Donnall Thomas in the late 1950s, involves infusing autologous or allogeneic stem cells to reconstruct a functional hematopoietic system (25). HSCT is now a standard treatment for various hematologic malignancies and have also been explored in solid tumours as immune cell progenitors, exhibiting synergistic effects with adoptive T cell immunotherapy and dendritic cell vaccines. Studies show that HSCs facilitate T cell trafficking, augment immunological rejection of invasive tumour cells, and can differentiate into immune-stimulating dendritic cells (DCs) within the TME (34). These findings highlight the multifaced role of HSCs not only as stem or progenitor cells for cell regeneration, but also as cells capable of synergistically enhancing the efficacy of immunotherapy against solid cancers. However, till today, there is limited evidence of HSCs in treating NPC patients. And thus, HSCs may have theoretical role against NPC, but no clinical study has confirmed its efficacy.

NSCs are self-renewable stem cells in the central nervous system, capable of differentiating into astrocytes, oligodendrocytes, and neural cells (35). Engineered NSCs have shown promise in treating brain cancers due to their ability to migrate to malignant brain masses (36). While the mechanism of NSC homing to tumour is not fully understood, hypoxia has been identified as a key factor. HIF-1α in the TME activates NSC chemoattractant production, such as cytokines, chemokines, and growth factors (37). This migratory behaviour positions NSCs as an effective tumour eradication therapy (38, 39). Similar to HSCs, currently no clinical study has confirmed NSC can eradicate NPC, and thus further study is warranted to investigate whether the NSC has its clinical value for NPC patients.

MSCs, derived from sources such as bone marrow, adipose tissues, PB, placenta, and umbilical cord, playing roles in tissue repair and regeneration due to its primary, non-specialised, non-haematopoietic, with rapid proliferation and differentiation potential. MSCs also exhibit immunomodulatory and anti-inflammatory properties, and are non-immunogenic, making them suitable for transplantation without immunosuppression. For example, bone marrow-derived MSCs inhibited CTL activation by downregulating natural killer group 2 member D protein (NKG2D) receptor expression and increasing the production of immunosuppressive factors (40). MSCs also promote Tregs differentiation to suppress allogeneic T cell response (41). MSCs also suppress tumour growth by modulating cell cycle signalling pathways and inducing apoptosis (42, 43). However, MSCs is a double edge sword in cancer therapy, in which MSCs can differentiate into cancer-associated fibroblasts (CAFs) in response to TGF-β released from tumour cells, leading to the promotion of tumour growth and progression (44, 45).

It is also worth noting that MSCs release exosomes that facilitate cell communication (46) and modulate disease progression (47, 48). In NPC, MSC-derived exosomes are associated with elevated expression of fibroblast growth factor (FGF)-19, enhancing tumour growth, migration, and metastasis through the activation of the FGF19-fibroblast growth factor receptor 4 (FGFR4)-dependent ERK signaling cascade and promotion of EMT (49). However, some reports suggest that these exosomes can inhibit tumour growth by reversing its therapy resistance. For instance, MSC-derived exosomes with overexpressed tumour suppressor microRNA-34c-5p (miR-34c) have shown potential therapeutic value in reversing radioresistance and inhibiting malignant behavior in NPC cells (50). Similarly, exosomes from engineered human MSCs with elevated expression of microRNA-125a-3p (miR-125a) demonstrated the ability to suppress migration and vasculogenic mimicry formation in NPC cells, suggesting potential applications in anti-angiogenic therapy for cancer treatment (51).

In summary, HSC and NSC are currently still under investigation whether its potential can be translated to bedside for NPC patients. The dual role of MSCs as tumours promoter or suppressor ( Table 1 ), may depend on tumour types, treatment doses, and duration. The role of MSCs as antitumour agents or therapeutic targets for NPC remains debated, and current evidence is insufficient to support the clinical application of MSC in NPC treatment. Further studies are needed to understand the complex interactions between MSCs and cancer, particularly the role of MSC-derived exosomes in carcinogenesis and therapy resistance.

Pluripotent stem cells (PSCs) can proliferate infinitely and differentiate into various specialised cell types. ESCs and induced pluripotent stem cells (iPSCs) are two types of human PSCs with significant implications for regenerative medicine and clinical research (55). The first human ESC line was derived by James Thomson and his team from in vitro-fertilised human embryos in 1998 (56). Due to ethical issues with ESCs (57), Takahashi and Yamanaka generated iPSCs from mouse embryonic and adult human somatic cells in 2006 and 2007, respectively (58, 59). These cells exhibit similar properties to ECSs, including pluripotent status, morphology, growth properties, surface antigen, gene expression, and telomerase activity, highlighting their potential in medical applications.

iPSCs can be genetically engineered into T cells or NK cells, which can effectively target and eradicate tumour cells. The generation of iPSC-derived T cells involves the induction of mesoderm specification and hematopoietic commitment, followed by T cell differentiation. Briefly, iPSCs collected from healthy donors are co-cultured on murine cell lines (C3H/10T1/2 or OP9) with morphogens, such as bone morphogenetic protein 4 (BMP4), VEGF, and fibroblast growth factor (FGF) to induce mesoderm specification. These cells are then co-cultured in a cocktail of cytokines to generate CD34⁺ HSCs (60). The CD34⁺ hematopoietic progenitors are transferred onto OP9 overexpressing the Notch ligand Delta-like 1 (DLL1) or DLL4 (OP9-DLL1 and OP9-DLL4) feeder cells to induce T cell development in the presence of cytokines (60). Similarly, iPSC-derived NK cells are generated from CD34⁺ hematopoietic progenitor cells using feeder-dependent culture method (61). However, murine-derived stroma feeder cells pose risks of cross-species contamination and complicate quality control due to the reliance on different serum and basal media. Therefore, feeder-free methods, such as using immobilized-DLL4 protein or DLL4-µbeads with lymphopoietic cytokines, have successfully induced T cell differentiation without feeder cells (62, 63). Researchers have also generated rejuvenated iPSC-derived antigen-specific T cells that retain the same antigen specificity and cytotoxicity as the original T cells, with higher proliferative capacity and reduced exhaustion markers (64, 65). iPSC-derived human papilloma virus (HPV)-specific CTLs have demonstrated more potent and sustained cytotoxic activity against cancer cells compared to original peripheral blood CTLs (66). Likewise, iPSC-derived EBV-specific CTLs persisted as central memory T cells in vivo for at least 6 months, continuously targeting EBV-associated lymphoma cells (67). Another study demonstrated that incorporating an inducible caspase-9 (iC9)-based suicide system into iPSC-derived EBV specific CTLs effectively suppressed tumour growth in vivo without compromising antigen-specific killing activity. This system also reduced cytokine release syndrome (CRS), enhancing the safety of T cell therapy (68).

NK cells derived from iPSCs exhibit greater cytotoxicity against a wide variety of cancers. By combining embryoid body (EB) formation with membrane-bound IL-21-expressing artificial APCs, functional NK cells can be efficiently produced from iPSCs. Briefly, iPSCs collected from healthy donors are grown under feeder-free conditions for a week, spun to aggregate into EBs, and cultured in feeder-free media with morphogens to induce the formation of HSC progenitors. The hematopoietic progenitor-containing EBs are then transferred to a feeder-free plate with NK differentiation media containing cytokines (namely IL-3, IL-15, IL-7, SCF, and Flt3L) to generate iPSC-derived NK cells (69). This method offers a safe and robust platform for producing clinically applicable immune cells. Preclinically, iPSC-derived NK cells have demonstrated effective cytotoxic responses against a variety of tumours in xenograft models. For example, iPSC-derived NK cells combined with anti-PD-1 therapy showed synergistic effects in ovarian cancer, delaying tumour progression and enhancing T cell recruitment and inflammatory cytokine production, transforming a “cold” TME into a “hot” TME (70). Although iPSC-NK cell therapy has shown promising results, it has yet to receive FDA approval. A recently completed phase I clinical trial (NCT03841110) involved evaluating the safety and efficacy of FT500, an off-the-shelf iPSC-NK cell product, in combination with one of three immune checkpoint inhibitors (nivolumab, pembrolizumab, or atezolizumab) in patients with solid tumours, including HNSCC (71). The administration of six doses of FT500 cells was well tolerated, with no serious adverse events, such as graft-versus-host disease (GvHD), cytokine release syndrome (CRS), neurotoxicity (NT), or host immune rejection. Nine out of 13 patients achieved stable disease, and one patient with Hodgkin lymphoma refractory to prior anti-PD-1 therapy experienced a 58% reduction in tumour size following the combination therapy (71). This trial provides clinical support for the high tolerability and potential antitumour efficacy of allogeneic iPSC-NK cell therapy.

Based on the evidence above, we advocate that iPSCs as a platform to provide an unlimited source of rejuvenated T cells or NK cells-based therapy. Such cell-based therapy holds promise as a safe and effective method for targeting NPC and other cancers. Further studies are warranted to investigate the possibility of introducing the CARs or transgenic T cell receptors (TCRs) into the iPSC-derived immune cells to further enhance their tumour-targeting capabilities while suicide gene-based safeguard system to ensure its safety.

Chimeric antigen receptor (CAR)-T cells are a type of immunotherapy involving the genetic modification of a patient’s T cells to express a synthetic receptor known as CAR, which includes an antigen-recognition domain (often an antibody single-chain variable fragment, scFv), and an intracellular signalling domain (typically CD3ζ) (89). These CAR-T cells can selectively target tumour associated antigen (TAA) via scFv recognition domain, leading to tumour cell elimination through the production of inflammatory cytokines and cytolytic effectors, achieving a long-term potent antitumour activity (89).

CAR-T cell development has evolved over generations with improved efficacy ( Figure 1 ). The first-generation CARs were developed in 1993 by Zelig Eshhar and consisted of a single signalling domain derived from the CD3ζ chain of the TCR (90). However, due to a low cytotoxicity, proliferation rate, and relatively short persistence in patients, the second-generation CARs were developed with an additional co-stimulatory domain (4-1BB/CD137 or CD28), resulting in enhanced T cell proliferation, cytotoxicity, and sustained response (91, 92). Soon, the third-generation CARs are further improved by including additional co-stimulatory domain, such as OX40/CD134 or CD137 (93, 94). CD28 and CD134 support the long-term T cell expansion and survival (95, 96), while CD137 to enhances T cell proliferation, IL-2 production, expression of anti-apoptotic genes to inhibit activation-induced cell death (AICD), and memory T cell development (97). Due to the undesirable clinical outcome, fourth-generation CARs, also known as T cell redirected for universal cytokine-mediated killing (TRUCKs) incorporated a nuclear factor of the activated T-cell (NFAT)-responsive promoter into the base of the second-generation CARs. This promoter can induce cytokine secretion, such as IL-7, IL-12, IL-15, and IL-18 upon CAR cell activation, augmenting T cell activation and killing of cancerous cells (98, 99). In the fifth-generation CARs, a truncated IL-2 receptor β (IL-2Rβ)-chain domain is introduced to provide a binding site for the transcription factor STAT3 and to activate JAK-STAT signalling pathway, further improving T cell activation, proliferation, and persistence (100).

In recent years, CAR T-cell therapy has demonstrated remarkable success in treating relapsed or refractory haematological malignancies. As of January 2024, the FDA has approved six CAR-T cell products: Abecma and Carvykti for multiple myeloma, Breyanzi and Yescarta for large B-cell lymphoma, Kymriah for non-Hodgkin lymphomas, and Tecartus for B-cell acute lymphoblastic leukaemia. Four target CD19, while Abecma and Carvykti target B cell maturation antigen (BCMA) (11). Despite the success in haematological malignancies, CAR-T cell therapies have not received clinical approval for the treatment of solid tumours, including NPC, due to challenges such as, the lack of specific antigens, on-target off-tumour effects, and the complexity of the TME (11). Encouragingly, clinical trials are ongoing. A Phase 1 clinical trial (NCT02915445) targeting epithelial cell adhesion molecule (EpCAM) in solid tumours, including NPC, showed promising antitumour efficacy and an acceptable safety profile, with two patients showing a partial response and three experiencing more than 23 months of progression-free survival (PFS) (80). Engineered T cells targeting LMP1 in EBV-positive NPC cells have also demonstrated increased IFN-γ and IL-2 production, effectively killing LMP1-overexpressing NPC cells in vitro and reducing tumour growth in vivo (73, 74). The safety and efficacy of these CAR-T cells have been evaluated in a Phase I/II clinical trial (NCT02980315) for the treatment of EBV-associated NPC. Additionally, an ongoing early phase I study (NCT04657965) is assessing these engineered T cells in patients with LMP1-positive infectious diseases and haematological malignancies.

It is also worth noting that numerous challenges limit the therapeutic efficacy of CAR-T cells in both hematological malignancies and solid tumours ( Figure 2 ). These challenges include antigen escape, poor tumour infiltration, low persistence, and the immunosuppressive nature of the TME (101). On-target off-tumour recognition is also a concern as many solid tumour antigens are expressed in normal tissues at varying levels. For example, HER2 CAR-T cells may induce severe side effects, including respiratory distress, cardiac arrest, and cytokine-release syndrome (CRS)-induced multi-organ failure. This is because HER2 is expressed not only by malignant cells, but also on normal epithelial cells in various tissues, making them off-targets by HER2 CAR-T cells (102). Therefore, selecting a suitable antigen for CAR-T cell therapy and enhancing the selective expression of CARs for targeting solid tumours are crucial considerations in current research. High rates of severe adverse events with fatal outcomes have prevented CAR-T cells from becoming first-line treatment. In a Phase I clinical trial (NCT01044069) for relapse B-cell ALL patients who received CD19-specific CAR T cells, potent antitumour efficacy of CAR-T cells was observed. However, 14 out of 53 patients experienced severe CRS, and tragically, one patient died from it (103). Neurotoxicity is also common after CAR-T cell therapy, with reports in 33 out of 53 patients, 22 of whom developed severe neurotoxicity within 28 days of CAR-T cell infusion (104). These challenges highlight the need for ongoing improvements and alternative approaches such as engineered TCR-T cells.

The TCR is a crucial component on the T cell surface for recognising antigens on infected or tumour cells. It consists of two antigen-binding peptide chains and three CD3 subunits (ζζ homodimers and δϵ and γϵ heterodimers), forming an octamer complex. Most T cells have αβ TCRs, while a smaller subset has γδ TCRs. The TCR chains comprises an extracellular region for antigen recognition, a transmembrane region, and a shorter cytoplasmic tail linked to the CD3 complex for signal transduction (105). However, due to their lack of co-stimulatory functions, TCRs often require additional co-stimulatory signals for effective T cell activation.

Similar to CAR-T cell therapy, TCR-T cell therapy involves genetically engineering T cells with TCR genes to specifically target tumour-specific antigens (TSAs) or TAAs. Unlike CAR-T cells, which recognise antigens directly in an MHC-independent manner, TCR-T cells use naturally occurring TCRs that recognise peptides presented by MHC molecules on the cell surface (106). This involves the proteasomal degradation of cellular proteins, transport into the endoplasmic reticulum (ER), and complex formation with MHC molecules. The peptide-MHC complexes are then presented on the cell surface for T cell surveillance (107). It has been reported that TCR-T cells exhibit nearly 100- to 1000-fold greater sensitivity in recognising peptide-HLA complexes compared to CAR-T cells, enhancing immune recognition and the tumour elimination (108, 109). This high specificity leads to more selective and regulated T cell activation, reducing the risk of excessive activation and cytokine production, thus resulting in milder treatment-related toxicity compared to CAR-T cell therapy (110, 111). Therefore, TCR-T cells offer an enhanced therapeutic strategy for targeting a broader range of malignancies compared to CAR-T cell therapy.

Over the past decade, therapies involving the EBV-specific T cell infusion have shown remarkable clinical responses in treating post-transplant lymphoproliferative disease (PTLD), a life-threatening complication following organ transplantation (112). Subsequent studies have explored the use of TCRs specific to EBV viral antigens, such as EBNAs and LMPs, in treating EBV-associated malignancies, including NPC. For example, retroviral transduction of engineered EBV antigen-specific TCRs into primary human T cells have demonstrated increased IFN-γ production and efficient lysis of tumour cells (113). Cho and co-workers developed a TCR specific to LMP1 from HLA-A*0201 transgenic mice immunised with the minimal epitope LMP1₁₆₆ (TLLVDLLWL). Infusion of these engineered T cells into immunocompromised mice revealed specific activation by low peptide concentrations and efficient recognition of LMP1-expressing tumour cells, demonstrating high avidity for antigen recognition in EBV-associated malignancies (23). Promising outcomes were also observed with LMP2-specific TCR, where transduced CTLs increased IFN-γ production and specific lysis of target cells. Infusion of these CTLs significantly attenuated the growth of LMP2-expressing tumours in vivo, improving survival rates in tumour-bearing mice (75). Additionally, transgenic T cells expressing LMP2-specific TCRs showed high avidity antigen-specific functions, including proliferation, cytotoxicity, and cytokine release (IFN-γ, TNF-α, and IL-2) against EBV-positive NPC cells. These engineered T cells effectively suppressed LMP2-positive tumour growth in an immunocompromised mouse model and specifically recognised and targeted LMP2-expressing NPC cells in advanced NPC patients (76). Taken together, these findings highlight the potential of TCR-T cell therapies in redirecting T cells to recognise EBV antigens, demonstrating efficacy in combating EBV-associated malignancies such as NPC.

At present, TCR-T cell therapies have not received FDA approval for clinical application in solid tumours other than melanoma. However, extensive efforts are ongoing to evaluate their safety and efficacy in other cancers, including NPC. TCR-T cells have also been developed to target cancer-associated viral antigens, such as EBV, HPV, and hepatitis B virus (HBV), and are being tested in multiple clinical trials. In a phase I/II clinical trial (NCT02280811) with HPV16-E6-specific TCR-T cells, two of 12 chemotherapy-refractory, metastatic HPV16-positive cancer patients experienced objective tumour responses without off-target toxicities or dose-limiting toxicities (114). In another phase I clinical trial (NCT03899415), HBV-targeted TCR T cells for HBV-related advanced HCC post-hepatectomy or radiofrequency ablation showed promising results. Among the eight patients, one achieved a partial response lasting 27.7 months, with only mild adverse events reported. Additionally, circulating HBV antigens and HBV DNA loads were reduced in all patients after TCR-T cell infusions, indicating effective targeting (115). For NPC, multiple clinical trials are focusing on TCR-T cells targeting EBV viral antigen such as LMP1 and LMP2. An ongoing phase I/II study (NCT05587543) is investigating LMP2-specific IL-12-secreting TCR-T cells in EBV-positive metastatic/refractory NPC patients. Another phase I trial (NCT03925896) is evaluating the safety and efficacy of LMP2-specific TCR T cells for HLA-A2, HLA-A11, and HLA-A24 recurrent and metastatic NPC patients. Additionally, a phase II study (NCT03648697) is assessing the safety and tolerability of LMP1, LMP2 and EBNA1-specific TCRs (YT-E001) in HLA-A02:01-/24:02-/11:01-positive recurrent or metastatic NPC patients. These high affinity TCRs targeting specific EBV antigens were screened from healthy donors and transduced into autologous T cells via lentiviral vectors.

Despite their promise, TCR-T cell therapies face challenges, including antigen selection, tumour heterogeneity, T cell exhaustion, and safety concerns. Ideal target antigen should be selectively and homogeneously expressed in tumours and presented via MHC class I to minimise off-target effects and treatment-related toxicity. However, solid tumours often exhibit antigenic variability, increasing the risk of antigen escape and the emergence of TCR-T cell-resistant tumours (116, 117). The immunosuppressive TME further impedes TCR-T cell infiltration and their efficacy in eliminating tumour cells (118). Additionally, prolonged TCR stimulation can induce T cell exhaustion, characterised by upregulated immune checkpoint proteins, such as PD-1, CTLA-4, T cell immunoglobulin and mucin domain-containing protein (TIM)-3, and LAG-3 (119). Tumours can also evade TCR-T cell recognition through mechanisms, such as downregulation of loss of MHC class I molecules through genetic alterations or epigenetic silencing (120). For example, non-responsive patients in a phase I/II clinical trial (NCT02858310) targeting HPV-16 E7 antigen showed tumour resistance due to defects in antigen presentation and interferon response pathways (121). Therefore, developing screening assays to identify genomic alterations associated with treatment resistance is important in order to improve TCR-T cell therapy efficacy.

Tumour-infiltrating lymphocytes (TILs) represent a subset of intratumour lymphocytes, and their adoption for treating advanced solid tumours has shown promising clinical outcomes. TIL therapy involves isolating these cells from the tumour, cultivating them in vitro, and reinfusing them into the patient with a high dose of IL-2 after lymphodepletion to enhance T cell survival and target tumour cells (122, 123). The efficacy of TIL therapy was first established by Steven Rosenberg in 1988 (124), and has now shown encouraging outcomes in NPC.

EBV antigens in NPC have led to the exploration of EBV-specific T cells as an alternative treatment. Studies showed that TILs isolated from NPC patients consists of a high frequency of CD4⁺ T cells that produce IFN-γ in response to EBNA1, aiding tumour regression (125). A number of clinical studies have delved into assessing the effectiveness of EBV-specific T cells in patients with NPC. These investigations have revealed promising trends, notably in the alteration of EBV DNA copy numbers and the augmentation of CTL levels, indicating a potential therapeutic benefit of EBV-specific T cell therapy in NPC patients. For instance, a phase I clinical study targeting EBV-positive NPC resistant to RT and chemotherapy with EBV-specific CTLs showed acceptable safety and modest objective responses. Among the 10 participants, two displayed partial responses and four maintained stable disease states (81). Building on this success, a subsequent phase II study in 23 recurrent or refractory NPC patients showed well-tolerated autologous EBV-CTL infusion, with 1-year and 2-year PFS rates of 65% and 52%, and overall survival rates of 87% and 70%, respectively (82). Furthermore, in a phase III clinical trial (NCT02578641), when EBV-CTLs were used to treat NPC patients after completing a first course of chemotherapy, the response rate increased to 71.4%, including three complete responses and 22 partial responses (83). A phase I study (NCT01462903) evaluating the safety and antitumour activity of TILs following CCRT in locoregionally advanced NPC (LANPC) patients showed promising results, including sustained antitumour activity, extended disease-free survival (> 12 months in 18 out of 20 patients), and enhanced EBV antigen-specific T cell responses (84). However, a completed phase II study (NCT00834093) evaluating EBV-specific CTLs in patients with recurrent, metastatic NPC demonstrated a poor overall response rate (ORR) and a median progression-free survival (PFS) of only 2.2 months. Of the 28 patients enrolled, 21 were treated, only one patient achieved a complete response, two experienced stable disease, and the remaining patients had disease progression. Interestingly, two patients who had previously failed the same chemotherapy regimen showed a renewed and robust response to chemotherapy after receiving EBV-CTL immunotherapy (85). These findings suggest that the efficacy of EBV-CTL immunotherapy may be enhanced when used in combination with other conventional therapies. Combination therapies, such as EBV-specific CTL with PD-1 blockade, have shown potential, as evidenced by a complete response with no evidence of disease relapse for 22 months in a metastatic EBV-positive NPC patient, suggesting potential for synergistic combination therapies (86). An ongoing phase I clinical trial (NCT02065362) takes a step further by examining the antitumour activity of TGF-β resistant, EBV-specific CTLs in patients with EBV-positive NPC. These genetically modified T cells incorporate a dominant negative receptor (DNR), conferring resistance to TGF-β and augmenting their efficacy in eliminating tumour cells. The ongoing or completed clinical approaches utilising EBV-specific TILs for treating NPC are summarised in Table 4 .

Despite promising outcomes, TIL-based therapy encounters significant challenges. Initial steps involve invasive surgical resection for TIL isolation can be distressing and poses risks to patients (126). The heterogeneity of TILs in terms of antigen specificity and differentiation stages may affects their effectiveness against tumours, impacting the overall success of the therapy (127). Challenges in TIL expansion and preparation include the need for specialised facilities and technical expertise for TIL culture and expansion (128), and patient factors such as age and overall health may influence the success of ex vivo expansion (129). Furthermore, the lengthy process of isolation, expansion, and reintroduction into the patients’ body is around 3 weeks to 3 months, considered a prolonged duration that inevitably generates delays for patient intervention (130). Ongoing research aims to optimise expansion protocols and shorten production times, while exploring combination therapies involving TILs and conventional treatments that hold potential for synergistic efficacy in cancer therapy.

Natural killer (NK) cells offer a promising alternative for cellular immunotherapy beyond T cells. As a key component of the innate immune system, NK cells play a vital role in cancer immune surveillance, eradicating tumour cells in an antigen-independent manner without requiring HLA matching (131). Their cytotoxic effects involve releasing perforin and granzymes, and activating killer activating receptors (KARs), leading to apoptosis via the expression of death ligands, such as TNF-α, FasL, and TRAIL. NK cells also modulate immunity by producing cytokines and chemokines, including IFN-γ, IL-10, CCL3, CCL4, and CCL5 (132). Several studies have confirmed the close relationship between NK cells and cancer development, with higher NK cell activity correlating with reducing susceptibility to oncogenic virus infection and improved survival (133, 134).

Adoptive transfer of autologous NK cells for cancer treatment is feasible due to easy sourcing and low risk of GVHD, but its limited tumour regression restricts clinical application (135). Researchers now focus on allogeneic NK cells, which offer advantages such as MHC-unrestricted immune recognition and lower GVHD incidence compared to CAR-T cell therapy. These cells, derived from PB, UCB, hESCs, iPSCs, or NK-92 cell lines, provide versatility for large-scale manufacturing and cryopreservation, allowing off-the-shelf availability (131). Among these, PB is the most accessible source but has limitations such as low cell counts (136), reduced proliferation and short lifespan (137), decreased cytotoxicity after cryopreservation (138), donor variability (139), and heterogenous NK cell populations, may potentially lead to variable NK cell function (140).

In contrast, UCB-derived NK cells offer several advantages over PB, including easier collection and long-term cryopreservation (141), better proliferation (142), enhanced bone marrow homing ability (143), and lower GVHD incidence (144). However, they have higher CD161 expression, limiting maturity and response to foreign antigens (143). The lower expression of adhesion molecules such as intercellular adhesion molecule (ICAM)-1 also limit their capacity to form conjugates with target cells (142, 145). They also exhibit higher expression of inhibitory molecules such as NKG2A/CD94, lower expression of CD16 (receptor that facilitates ADCC), CD57 (NK cell maturation marker), and KIR2DL4 (activating receptor), and reduced production of perforin, granzyme B, IFN-γ, and cell surface FasL and TRAIL, reducing cytotoxicity against tumour cells (146). However, cytokines such as IL-2, IL-12, IL-15, and IL-18 can restore UCB-NK cell cytotoxicity (147), and monoclonal antibodies such as monalizumab can enhance NK cell activity by inhibit the inhibitory function of NKG2A (148).

As discussed earlier, hESCs/iPSCs can differentiate into NK cells, offering an alternative source for allogeneic NK cells by providing homogenous cell populations that can be grown indefinitely. Their potential for genetic engineering with CARs makes them a promising strategy to generate standardised, off-the-shelf NK cells with enhanced expansion, in vivo persistence, and improved antitumour activity (149, 150). Immortalised NK cell lines such as NK-92, which can be expanded ex vivo, are also promising (151, 152). NK-92 shows high antitumour activity and are the only cell line approved for CAR-NK-92 clinical trials (153). It was established in 1992 from a non-Hodgkin lymphoma patient, showing characteristics of early NK cells with the expression of CD56, CD2, and CD7, but lack CD3 (154). The continuous growth of NK-92 is IL-2 dependent, and despite lacking CD16, they exhibit significant cytotoxicity due to the expression of activating receptors, including NKp30, NKp46, NKG2D, and CD28, with almost complete loss of inhibitory killer cell immunoglobulin-like receptors (KIRs) on the surface, except for KIR2DL4. NK-92 cells also express high levels of cytotoxic effector molecules, such as perforin, granzyme B, FasL, TRAIL, and TNF-α, consistently inducing high cytotoxic activity against tumour cells (155). In order to minimise the risk of developing lymphoma in recipients due to the origin of NK-92 cells, they require irradiation before infusion to prevent continuous growth while maintaining their cytotoxicity (156). Multiple phase I trials (NCT00990717, NCT00900809) have evaluated the safety and efficacy of irradiated NK-92 cells, showing favourable clinical outcomes, further reinforcing the potential of their therapeutic application (157, 158). However, irradiation may limit their expansion capacity, potentially diminishing overall antitumour efficacy (159).

There is growing research on NK cell-based therapy for NPC. LMP2 antigens are potential targets for NK cells, with targeted clearance of LMP2-containing cells showing robust antitumour effects and minimal toxicity to normal cells. During EBV infection, LMP2A induces F3 expression via the PI3K/Akt signalling pathway, negatively affecting NK cell activation (77). F3 promotes platelet aggregation, potentially aiding cancer metastasis and evading NK cell surveillance by downregulating NK2GD ligands and suppressing IFN-γ release (160). Additionally, platelet-derived TGF-β further enhances tumour metastasis by inhibiting the expression of CD226 and CD96 on NK cell surface, protecting tumour cells from being recognised by NK cells (161). Inhibition of F3 with the administration of NK-92MI (independent of exogenous IL-2) or UCB-derived NK cells has been shown to restore NK cell antitumour function in an NPC xenograft mouse model (77). UCB-derived NK cells also suppressed brain metastasis in a recurrent NPC patient post chemoradiotherapy, showing significant reduction over two years, with sustained efficacy (87). Combining RT with PD-1 inhibition synergistically increases NK cell-mediated killing of NPC cells in vitro and in vivo (78). Additionally, LMP1 induced B7-H3 expression via PI3K/AKT/mTOR pathway activation, leading to a reduction of NK cell cytotoxicity. Deleting B7-H3 in tumour cells and using anti-PD-L1 treatment restored NK cell function and enhanced cytotoxicity against NPC cells in xenograft models, suggesting the potential when combined with NK cell-based immunotherapy and immune checkpoint blockade against EBV-associated NPC (79). In a recent Phase I/II study (NCT02507154), Lim and co-workers assessed the safety and efficacy of combining cetuximab (anti-EGFR) with autologous expanded NK cells in EGFR-positive, recurrent and/or metastatic NPC patients who had failed at least two prior lines of chemotherapy. The treatment was well tolerated, with three out of seven patients achieved stable disease after the first NK cell infusion and experiencing a relatively long time to disease progression, lasting up to 19 months with two NK cell infusions (88).

NK-92 cells, known for their long-term cryopreservation capability and uniform population, are increasingly integrated with cancer-targeting CARs (CAR-NK-92). The first-generation CAR-NK cells feature a single signalling domain (CD3ζ), which is insufficient for potent killing. In contrast, the second- and third-generation CAR-NK cells incorporate additional co-stimulatory molecular motifs, such as CD28 or CD137 (4-1BB) for improved efficacy (162). For instance, anti-CD-19/CD22 bispecific CAR-NK-92 cells incorporating a CD3ζ/CD137 signalling domain exhibited increased cytotoxicity against B cell lymphoma cells compared to anti-CD19 CAR-NK-92 cells alone (163). Third-generation CAR-NK-92 cells targeting HER1 in HNSCC demonstrated enhanced antitumour immune responses characterised by increased INF-γ secretion and CD107a expression, a degranulation marker. Despite their enhanced killing activity, challenges such as PD-L1 upregulation and expansion of CD44v6-positive (putative CSC marker) on surviving HNSCC cells have been reported (164). This suggests that relying solely on CARs targeting a single TAA may not be effective as monotherapy, potentially leading to tumour relapse and treatment resistance. Therefore, combination therapy approaches are necessary to enhance CAR-NK cell efficacy. For instance, second-generation CAR-NK-92 cells co-expressing anti-HER2 and soluble PD-1 significantly increased NK cell and T cell infiltration and effector molecule release, enhancing immunological antitumour efficacy in PD-L1⁺Her2⁺ breast cancer cells (165). Similarly, a third-generation CAR targeting epidermal growth factor receptor (EGFR) in NK-92 cells, combined with the kinase inhibitor cabozantinib, effectively lysed EGFR-positive renal cell carcinoma (RCC) cells and improved tumour homing (166).

Currently, only a few registered clinical trials on the US Clinical Trials Registry (ClinicalTrials.gov) investigate CAR-NK-92 for hematological or solid tumours. For instance, a phase I clinical trial (NCT02944162) assessed the safety profile and clinical efficacy of third-generation CAR-NK-92 cells targeting CD33 in relapsed/refractory AML (167). Although no significant clinical efficacy was observed due to the decreased cytotoxic potential following irradiation, the therapy demonstrated safety with a high cell dose (five billion cells per patient) (167). Another ongoing phase I clinical trial (NCT03383978) is evaluating the safety and tolerability of a second-generation HER-2 specific CAR-NK-92 for recurrent glioblastoma. The trial has shown a favourable safety profile with local intracerebral injections of up to 1 x 10⁸ HER2-CAR-NK-92 cells (168). New advancements include high-affinity Fc-receptor-expressing NK cells (haNKs), derived from NK-92 cells, which can be genetically engineered to express CARs, augmenting their targeting capabilities and cytotoxic potential (169). CAR-haNK cells targeting PD-L1 have demonstrated successful recognition and targeting of heterogeneous tumour cell populations, including both T cell-sensitive and T cell-resistant tumour cells, in an HNSCC mouse model, effectively overcoming immune resistance (170). A phase II clinical trial (NCT04847466) is evaluating the effectiveness of irradiated PD-L1 CAR-haNK cells in combination with the PD-1 inhibitor pembrolizumab plus N-803 (ALT-803, an IL-15 agonist) in patients with recurrent or metastatic gastric cancer and HNSCC. Preliminary studies (NCT04050709) have shown PD-L1 CAR-haNK cells at a dose of 2 x 10⁹ cells twice weekly are well-tolerated, with no dose-limiting toxicities or CRS, supporting their advancement to phase II clinical trial (171). While studies have not yet evaluated CAR-NK products in NPC patients, the promising therapeutic efficacy observed in other solid tumours suggests that exploring genetic engineering of the ex vivo expanded NK cells could offer clinical benefits in NPC treatments.

Despite their promise, CAR-NK-92 cells require irradiation before infusion, which affects their viability and cytotoxicity more rapidly than non-irradiated cells (172). Therefore, the dosage and impact of irradiation on CAR-NK-92 cells needs to be carefully considered in future clinical trials. Given the short lifespan of irradiated CAR-NK-92 cells, shortening the interval between infusions may improve therapeutic efficacy. Similar to CAR-T cells, CAR-NK-92 cells exhibit target-dependent cytotoxicity and may induce on-target, off-tumour toxicity if target antigens are expressed in normal tissues (173). To address this, strategies like incorporating suicide genes into CAR constructs can serve as a safety switch, enabling control over engineered cell activity and facilitating elimination if necessary. This approach could improve safety and effectiveness of CAR therapies (174, 175).

Chemokines are small cytokines that regulate immune cell migration and trafficking. The chemokine expression profile of solid tumours is complex and influenced immune cell recruitment to the TME and tumour growth (176). Strategies targeting chemokine and their receptors are increasingly used in immunotherapy to enhance the effectiveness of CAR-engineered immune cells. For instance, CAR-T cells engineered with IL-7 and CCR2b (7x2b CAR-T) showed improved survival and migration to the tumour site by boosting IFN-γ, IL-2, and granzyme production (177). Similarly, co-expressing CXCR1 in CAR-NK cells targeting NKG2D ligands exhibit better tumour trafficking and growth inhibition (178). Additionally, genetically modified oncolytic viruses (OVs) can alter the tumour chemokine profile, aiding in antigen presentation and boosting chemokine production to recruit immune cells into the TME (179). Studies show that intratumoral administration of a CXCL11-armed tumour selective vaccinia OV increases tumour-specific CTLs and granzyme B production, while reducing immunosuppressive cytokines in the TME of a syngeneic mouse mesothelioma model, leading to enhanced cytotoxic activities of CTLs (180). In another study, Moon and co-workers evaluated the synergistic effects of CXCL11 and mesothelin-redirected CAR-T cells. While CAR/CXCL11 showed limited T cell trafficking, VV.CXCL11, an oncolytic vaccinia virus producing CXCL11, effectively increased T cell infiltration and improved antitumour efficacy after adoptive T cell therapy (181). Thus, combining oncolytic virotherapy with adoptive T cell transfer holds promise for enhancing NPC therapy efficacy.

Solid tumours are enriched with extracellular matrix (ECM), stromal cells, and immunosuppressive cells, creating barriers that hinder immune cell penetration and infiltration (182). To tackle the problem, researchers have engineered CAR-T cells to target ECM components and cancer-associated fibroblasts (CAFs). For instance, CAR-T cells expressing heparanase (HPSE) can break down ECM components and reduce fibrosis, improving immune cell infiltration and antitumour effects (183). Fibroblast activation protein (FAP), a marker distinguishing CAFs from their normal counterparts, has been found in over 90% of epithelial cancers, including NPC, and is often correlated with poor prognosis (184). Preclinical studies shows that CAR-T cells targeting FAP can eliminate CAFs, suppress myeloid-derived suppressor cells (MDSCs) recruitment, and enhance CTL and CAR-T cell survival (185). At present, only two clinical trials have been conducted using anti-FAP CAR-T cells. A phase I clinical trial in malignant pleural mesothelioma (NCT01722149) reported that localized injection of these CAR-T cells was well tolerated with ongoing antitumour immune responses (186, 187). Another phase I clinical trial (NCT03932565) is evaluating the safety of fourth-generation CAR-T cells targeting Nectin4 and FAP, combined with IL-7, CCL19, or IL-12 for advanced Nectin4-positive solid tumours. Pre-clinical studies show that Nectin4-targeted CAR-T cells with IL-7 and CCL19 help prevent CAR-T cell exhaustion by reducing immune checkpoint expression, while FAP-targeted CAR-T cells with IL-12 enhance immune cells recruitment (188). These combination therapies show potential for improving treatment outcomes in NPC.

Over the past few decades, bispecific antibodies (BsAbs), especially bispecific T cell engagers (BiTEs), have proven effective in treating hematologic malignancies by binding two different antigens to direct immune cells to tumour (189). BiTEs facilitate direct interaction between T cells and tumour cells directly by binding CD3 on T cells and TAA on tumour cells, bypassing the need for antigen-presenting cells (APCs) (190). This approach helps overcome antigen loss and variability, as seen with EGFR variant III (EGFRvIII) CAR-T cells secreting BiTEs to target glioblastoma cells while activating bystander T cells, enhancing the antitumour response against heterogeneous tumours (191). A study by Yin and co-workers showed that BiTE-secreting T cells EGFR and interleukin-13 receptor alpha 2 (IL13Rα2) exhibited superior antitumour activity with higher sensitivity and specificity compared to their CAR-T counterparts in glioblastoma model (192). Overall, these findings suggest that BiTE-secreting CAR-T cells could be a promising approach to address challenges associated with antigen heterogeneity in solid tumours. While research on using BiTEs to enhance CAR-T efficacy in targeting NPC is limited, targeting LMPs expressed on EBV-infected cell surfaces, but not on normal cells, could be explored as a potential approach for BsAbs. This concept is supported by second-generation CAR-T cells targeting LMP1 in LMP1-positive NPC cells, demonstrating specific killing of NPC cells and inhibition of tumour growth in xenograft model (73).

An increasing body of evidence suggests that CAR toxicity may be linked to the synthetic nature of the receptor design (194). To address this, researchers have constructed a natural multi-chain immunoreceptor CAR using DAP12 instead of CD3ζ. DAP12, a 12-kDa transmembrane adaptor protein with a single immunoreceptor tyrosine-based activation motif (ITAM), was originally found to activate NK cells and is involved in transmitting activating signals from various receptors (195). Studies show that DAP12-based CARs offer superior antitumour activity and safety than CD3ζ-based CARs, with improved antigen-specific cytotoxicity, TIL proliferation, reduced toxicity, and lower production of pro-inflammatory cytokines (196, 197), highlighting the potential of DAP12 in mitigating the risk of CRS. For instance, a phase I clinical trial (ChiCTR1800016584) of CD19-KIRS2/DAP12-BB CAR-T cells reported complete responses in all patients with low toxicity (198). Similarly, CAR-NK cells incorporating DAP12 have shown promising results in treating solid tumours. In a recent study, Xiao and co-workers constructed a CAR-NK cell by combining the NKG2D receptor with DAP12, which showed significant therapeutic effects and lower toxicity in mice with solid tumours. This approach also led to positive outcomes in three patients with metastatic colorectal cancer (199). Building on these results, a pilot clinical trial (NCT03415100) has been launched to evaluate the safety and feasibility of these CAR-NK cells for treating metastatic solid tumours.

Another strategy to enhance the safety of CAR-based cell therapy is the engineering of suicide genes such as inducible caspase 9 (iCasp9) into the CAR construct. These suicide genes serve as a safety switch that can induce cell death upon activation by an external agent, such as drug or antibody (174, 175). For instance, the iCasp9 gene is often used in CAR-based cell therapy in conjunction with a small, bio-inert molecule AP1903 (Rimiducid), which acts as a chemical inducer of dimerization (CID) (200). When administered, AP1903 binds to the CID domain fused to iCasp9, leading to the formation of homodimers and subsequent activation of caspase 9. This activation triggers apoptotic cell death specifically in the CAR-engineered immune cells that express high levels of the transgene, allowing for the selective removal of inappropriately activated cells and thus providing a safety mechanism to manage potential toxicities or adverse events associated with CAR-T cell therapy (200). The efficacy of iCasp9 in eliminating CAR-engineered immune cells to counteract serious adverse event in CAR-based cell immunotherapy has been demonstrated in several preclinical studies (175, 201). Furthermore, several early phase clinical trials (NCT05239143, NCT03016377, NCT03696784 and NCT03721068) are ongoing to evaluate the safety and efficacy of this approach in patients with hematological malignancies or solid tumours, including NPC.