Oncofetal reprogramming in hepatocellular carcinoma: linking developmental programs to cancer vaccines and immunotherapy

Abstract

Hepatocellular carcinoma (HCC) is the most common primary liver cancer and a leading cause of cancer mortality worldwide. Its pathogenesis reflects a combination of tumour-intrinsic heterogeneity and a profoundly immunosuppressive tumour microenvironment. Growing evidence shows that tumours recapitulate developmental programs to establish an oncofetal ecosystem, characterised by the re-expression of foetal antigens and foetal-like stromal and immune subsets. These features drive immune evasion and shape therapeutic response, contributing to immunotherapy outcomes in clinic. This review outlines mechanistic insights into oncofetal reprogramming across tumour, stromal, and immune compartments and evaluates therapeutic strategies that target these dependencies. We highlight emerging vaccine platforms, cellular therapies, and biologics targeting oncofetal antigens, with particular emphasis on mRNA–lipid nanoparticle vaccines and their potential to induce robust, durable antitumour immunity. We further discuss rational combinatorial strategies that integrate vaccines with immune checkpoint inhibitors. Finally, we discuss how overcoming liver tolerance and antigenic heterogeneity will be essential for effective oncofetal-directed therapies. Collectively, targeting the oncofetal ecosystem through coordinated vaccine, cellular, and immunotherapeutic strategies offers a path toward more durable responses and broader immunotherapy benefits in HCC.

INTRODUCTION

Hepatocellular carcinoma (HCC) remains a major global health challenge and represents the most common primary malignancy of the liver, characterised by high morbidity and mortality rates [1]. According to the World Cancer Research Fund, HCC is the sixth most frequently diagnosed cancer worldwide and the third leading cause of cancerrelated deaths [2]. HCC typically develops in the context of chronic liver inflammation, most often as a consequence of persistent hepatitis B or C virus infection. These ongoing inflammatory processes progressively drive hepatic fibrosis and cirrhosis, establishing a pathological environment conducive to malignant transformation [3]. The pathogenesis of HCC is highly complex, involving multiple molecular pathways and the gradual accumulation of genetic and epigenetic alterations, which together give rise to a heterogeneous mutational landscape.

Beyond these intrinsic tumour changes, the tumour microenvironment (TME) plays a crucial role in promoting tumour initiation, progression, and immune evasion. The immunosuppressive TME in HCC facilitates tumour growth and resistance to host immune surveillance [4]. Interestingly, this immunosuppressive state has been compared to that observed during foetal development, as both share molecular and cellular characteristics. This resemblance underpins the concept of an oncofetal ecosystem, in which tumour and stromal cells exhibit embryonic-like properties [5,6]. A hallmark of this oncofetal ecosystem is the expression of immunosuppressive mediators such as vascular endothelial growth factor A (VEGF-A) and programmed death-ligand 1 (PD-L1), alongside the presence of foetal-like reprogramming of tumour-associated macrophages (TAMs), endothelial cells (ECs), and cancer-associated fibroblasts (CAFs) [5,7]. These cellular and molecular components are characteristic of both tumour and foetal tissues yet are largely absent in healthy adult liver tissue [5,7]. The oncofetal features of the TME facilitate tumour development and are correlated with response to therapy in clinic [8]. Over recent years, several oncofetal antigens (OFAs) have been identified, including the well-characterised alpha-fetoprotein (AFP) and glypican-3 (GPC3) [9], as well as a growing list of novel and lesser-known candidates. These foetal-development proteins, absent in healthy adult tissues but re-expressed in tumours, are highly promising targets for immunotherapies and cancer vaccines.

Despite advances in systemic therapy, immune checkpoint inhibitors (ICIs), administered alone or in combination with anti-VEGF agents, currently form the first-line treatment for advanced HCC but have demonstrated only modest response rates [10-14]. This limited efficacy, coupled with a scarcity of curative options, underscores the urgent need for more effective and durable therapeutic strategies. In this review, we explore emerging treatment approaches for HCC including chimeric antigen receptor (CAR) technologies, with particular emphasis on messenger RNA (mRNA)-lipid nanoparticle (LNP) vaccines as a promising therapeutic platform (Fig. 1). We examine vaccine and immunotherapy modalities, combinatorial treatment strategies, and biomarker-driven patient stratification approaches aimed at optimising therapeutic outcomes. Importantly, we highlight the potential of targeting OFAs to enhance treatment efficacy and discuss strategies to overcome immune tolerance within the HCC microenvironment.

Figure 1.

Schematic diagram of tumour cell killing mechanisms mediated by OFA-targeted LNPs and CAR-engineered immune cells. CAR, chimeric antigen receptor; mRNA, messenger RNA; NK, natural killer; OFA, oncofetal antigen.

ONCOFETAL REPROGRAMMING IN HCC

Cancer development often recapitulates features of embryonic development, as tumour cells reactivate developmental signalling programs [6,15,16]. Similar to embryonic cells, tumour cells can self-renew and differentiate into multiple lineages [7,15]. These cells actively remodel the TME, generating a dynamic and heterogeneous niche [5,17]. Oncofetal reprogramming is defined by re-expression of foetal genes in cancer tissue [6]. In HCC, oncofetal reprogramming extends beyond malignant gene expression to include the surrounding stromal and immune compartments, collectively described as the oncofetal ecosystem [5,18,19]. This process involves reactivation of foetal genes that drive proliferation and immune evasion [20]. Notably, presence of oncofetal reprogramming is associated with drug response and clinical outcomes [16,21]. Importantly, both viral and nonviral aetiologies of liver cancer are associated with oncofetal reprogramming [5]. Further, foetal-like hepatocytes are more enriched in a subset of less differentiated tumours [22,23].

Within the TME, multiple cell types contribute to oncofetal reprogramming. TAMs, CAFs, and ECs collectively recreate a foetal-like immunoregulatory environment. Notably, plasmalemma vesicle-associated protein (PLVAP)⁺ ECs and folate receptor beta (FOLR2)⁺ TAMs reappear in HCC [5]. PLVAP is essential for macrophage development in the foetal liver, while FOLR2⁺ TAMs display transcriptional features similar to foetal liver macrophages, underscoring their developmental parallels [5,24]. Oncofetal cell states and their neighbourhoods are best documented by single cell and spatial mapping of patient tumours and we combined spatial profiling with in vitro functional assays to uncover VEGF/NOTCH signalling in this process [5]. Further, Wang et al. [25] provided perturbation evidence that NOTCH signalling controls macrophage polarisation. Accordingly, PLVAP⁺ ECs are proposed to promote FOLR2⁺ TAM activation through DLL4-NOTCH2 signalling, while hepatocytederived VEGF-A sustains PLVAP⁺ ECs, highlighting the central role of VEGF and NOTCH pathways in maintaining this ecosystem [25-27].

FOLR2⁺ TAMs co-express CD163, MRC1, and TREM2, with the latter linked to poor survival [8,28,29]. These TAMs secrete IL-18, inducing Tregs to adopt IL-1β⁺ effector phenotypes that suppress CD8⁺ T cell responses [29]. In turn, Tregs reinforce TAM reprogramming via secretion of TGF-β, CSF-1, and IL-10, further promoting epithelialmesenchymal transition (EMT) and tumour invasion [29,30]. Additional macrophage subsets such as LYVE1⁺ tissueresident macrophages (TRM) and C1QC⁺ TRM-TAMs also share similarities with FOLR2⁺ TAMs, suggesting a broader presence of foetal-like reprogramming in myeloid cells [27,31]. CAFs contribute further – periostin⁺ fibroblasts recruit immunosuppressive Tregs through CXCL16-CXCR6 and CXCL12-CXCR4 signalling, reinforcing tolerogenic stromal niche [32]. Together, these cellular interactions recreate the key immunosuppressive features of the developing foetus. Overall, spatial/multi-omic studies define where these cells co-localise and associate with outcome [8], whereas functional perturbation is still needed to confirm which interactions are required in vivo.

Classical examples of oncofetal proteins include AFP and GPC3, both widely re-expressed in HCC and established as diagnostic markers [33,34]. Beyond them, a growing number of oncofetal genes have been implicated in HCC. Claudin-6 (CLDN6), normally restricted to embryonic stem cells, is reactivated in HCC, where it activates Hippo signalling and drives a lineage shift toward a sorafenib-resistant biliary phenotype [35,36]. Hepatic leukaemia factor (HLF), also foetal - rest r icted, transactivates c-Jun to promote tumour-initiating cell traits [37]. Notably, HLF suppression enhances sorafenib sensitivity [37]. Serine peptidase inhibitor Kazal type 1 (SPINK1), regulated by ELF3 and associated with CD133, promotes tumour initiation and chemoresistance through the EGFR-ERK-CDK4/6-E2F2 axis [38]. CD133⁺ tumourpropagating cells further display dedifferentiation and EMT features, underscoring the role of stem-like programs [38]. Delta-like homologue 1 (DLK1), highly expressed in foetal liver and HCC, correlates with tumour size and AFP levels [39,40]. Its role in both foetal growth and HCC progression highlights its function as a developmental driver aberrantly reactivated in malignancy.

Oncofetal transcription factors also shape HCC tumour biology. Epithelial splicing regulatory protein 2 (ESRP2), an oncofetal suppressor, promotes the foetal-to-adult RNA splicing switch [20]. Loss of ESRP2 drives HCC proliferation via oncofetal TAK1 splicing and activation of p38 MAPK [20]. Conversely, Sal-like protein 4 (SALL4), normally restricted to foetal hepatocytes, is aberrantly re-expressed in HCC, where it sustains pluripotency and self-renewal [20,41]. Similarly, several RNA-binding proteins including tripartite motif containing protein 71, lin-28 homolog B, IGF2 mRNAbinding proteins 1-3 (IMP1-3), and paternally expressed gene 10 (PEG10) are strongly enriched in foetal programs and contribute to hepatocarcinogenesis [42-44]. Among these, IMP3, absent from adjacent benign tissue, facilitates tumour aggressiveness and correlates with CD44 and AFP expression [43,44]. PEG10, primarily placental in origin, is upregulated in HCC and promotes aerobic glycolysis, linking oncofetal reprogramming to metabolic adaptation [45].

Therefore, the oncofetal signature can vary across disease stages in a gene-dependent manner. For example, AFP can be present before HCC development, such as during hepatocyte injury and regeneration [46], while GPC3 has high sensitivity for diagnosing HCC but is less sensitive in very well-differentiated HCC and the fibrolamellar variant [47]. Similarly, SALL4 is absent in normal differentiated hepatocytes and is preferentially detected in aggressive HCC [41].

Emerging omics technologies continue to identify novel oncofetal markers and pathways in HCC [48]. While many reactivated foetal programs have been characterised, others remain undiscovered and may provide fresh insights into tumour biology. A deeper understanding of the oncofetal ecosystem will be critical for developing strategies that not only detect HCC earlier but also disrupt the foetal-like programs that sustain tumour progression.

VACCINES, MRNA THERAPEUTICS, AND CELLULAR THERAPIES

Oncofetal antigen vaccines in HCC

A wide range of oncofetal markers have been identified across cancers, including HCC [35]. Targeting these antigens offers a practical and relatively safe immunotherapy strategy, as they are typically re-expressed only in tumour cells while absent or minimally expressed in normal adult tissues [49]. Classic examples in HCC include AFP and GPC3, both of which are highly expressed, immunogenic, and tumour-specific [50]. Vaccines against these targets aim to elicit cytotoxic CD8⁺ T cell responses, thereby mounting a tumour-directed immune attack. Although recent advances in HCC vaccination have shifted towards novel platforms, conventional vaccine approaches established the early foundation of this field. Among the earliest strategies were peptide vaccines, dendritic cell (DC) vaccines, DNA vaccines, and viral vector-based vaccines.

Peptide vaccines rely on small antigenic fragments and are relatively simple to manufacture with good safety profiles [51]. AFP-derived peptide vaccines were frequently combined with adjuvant peptides such as heat shock protein 70 (HSP70) or glycoprotein 96 (gp96) to improve antigen presentation and enhance CD8⁺ T cell priming [52,53], and early trials demonstrated that these vaccines were well tolerated and consistently induced antigen-specific T cells in HCC patients [54]. Similarly, GPC3-derived peptides were shown to be safe in clinical testing, with better efficacy observed when given as an adjuvant to surgery or ablation rather than as monotherapy in advanced disease [51,55]. To overcome the limited potency of single peptides, multipeptide cocktails incorporating GPC3 with a WD-repeat protein and potential tumour suppressor were developed, demonstrating strong tolerability and potential usefulness [56]. Fusion vaccines linking GPC3 to chemokines like XCL1 further improved DC recruitment and tumour control, and these effects were amplified when combined with anti-PD-1 therapy [57].

Parallel efforts explored DC-based vaccines, where ex vivo-loaded DCs present tumour antigens through MHC-I to activate CD8⁺ T cells [58]. In preclinical models, DCs pulsed with recombinant gp96 fused to GPC3 or PEG10, another OFA, induced potent T cell responses and suppressed tumour growth [59]. AFP-loaded DCs also showed early clinical promise, inducing immune activation with acceptable safety profile [60,61]. Broader DC vaccines incorporating AFP, GPC3, and melanoma-associated antigen-A1 (MAGE-1) elicited antigen-specific immune responses in patients, though with limited clinical benefit, suggesting that such approaches may be better suited to early-stage disease or require combination with immunomodulators [62].

DNA vaccines encoding OFAs were developed as another conventional approach. Plasmid DNA encoding AFP in combination with immune stimulators such as HSP70 or cytotoxic T-lymphocyte-associated protein 4 (CTLA4) produced strong AFP-specific cytotoxic T cell responses and antitumour effects in experimental models [63,64], while the addition of cytokines like IP-10 and IL-12 further enhanced tumour regression and prolonged survival [65]. GPC3-encoding plasmids also demonstrated tumour growth suppression and prolonged survival in vivo [66]. However, clinical translation has been limited: in a small two-patient study using an AFP DNA vaccine prime with adenoviral boost, treatment was well tolerated but patients experienced tumour recurrence within 9-18 months [67]. Comparative work suggested that AFP-based DC vaccines elicited stronger immune responses than AFP plasmid DNA vaccines, highlighting differences in potency between platforms [68].

Viral vector vaccines, leveraging the innate ability of viruses to infect cells and drive robust immune responses, were also tested [69]. AFP-expressing lentiviral vectors combined with ICIs such as anti-programmed death-1 (PD-1) or PD-L1 produced synergistic antitumour effects, significantly slowing disease progression, whereas either strategy alone was insufficient [70]. Bacteriophage-based vaccines targeting the oncofetal protein, aspartate β-hydroxylase, also inhibited tumour growth and improved survival in mice when combined with anti-PD-1 therapy [71]. Adenoviral vectors encoding both GPC3 and IL-12 yielded stronger tumour inhibition than either component alone [72], and Phase I/II studies of AFP-expressing adenoviruses confirmed tolerability without dose-limiting toxicity [67]. In addition, DCs engineered with lentiviruses encoding AFP produced stronger antitumour activity than AFP peptidepulsed DCs [73].

More recently, non-viral nanoparticle-based vaccines have emerged, offering alternative delivery strategies with improved flexibility. Liposomes carrying GPC3, either alone or together with the TLR7/8 agonist CL097, successfully induced GPC3-specific immunity and prevented progression of premalignant hepatic lesions in animal models [74,75]. Nanoparticles delivering siRNA against AFP reduced HCC cell viability in vitro and showed even greater efficacy when combined with angiogenesis inhibitors such as sorafenib or sunitinib, or with natural compounds like epigallocatechin gallate [76,77]. GPC3 antibody-conjugated copolymer nanoparticles loaded with luciferase-siRNA further demonstrated targeted delivery and efficient gene silencing in vivo [78].

Despite these advances, there has been minimal recent clinical or research activity on OFA-targeting vaccines in HCC. These strategies largely focused on single or limited antigen targets and, although they consistently demonstrated immunogenicity and antigen-specific responses, clinical efficacy was modest [79]. Challenges such as limited potency in advanced disease, antigen heterogeneity, high production costs, and scalability issues have likely dampened enthusiasm for their continued development. Additional barriers may include the lack of validated biomarkers to guide patient selection, an incomplete understanding of how best to integrate OFA vaccines with ICIs or other therapies, and limited evidence on whether single-antigen approaches can overcome the profoundly immunosuppressive HCC microenvironment. Consequently, the field has shifted its focus towards newer and more versatile technologies such as mRNA vaccines, personalised neoantigen-based strategies, and CAR-engineered cell therapies, that promise greater clinical impact and are better aligned with the modern immunotherapy landscape.

mRNA therapeutics and T cell-based strategies for oncofetal targets

The mRNA vaccine technology gained prominence in 2020, when Moderna’s Spikevax and BioNTech/Pfizer’s Comirnaty were both approved for use against COVID-19 [80]. The most common delivery platform is the LNP, a nanoscale carrier system composed of a mixture of lipid components, with the nucleic acid cargo (mRNA) encapsulated within the central core of the particle [80]. Notably, the first LNP system approved by the U.S. Food and Drug Administration (FDA) was Onpattro (Patisiran) in 2018, used for the treatment of hereditary transthyretin-mediated amyloidosis [81]. This approval represented a major breakthrough in nucleic acid delivery and the nanoparticle field, paving the way for subsequent LNP development.

In these systems, the mRNA encodes the antigen of interest, and vaccine efficacy depends largely on the immunogenicity of the selected antigen [82]. Consequently, OFAs represent attractive targets for these vaccines. Furthermore, this vaccine platform offers several advantages over the conventional systems discussed above, including rapid and scalable synthesis, reduced risk of genomic integration, and the ability to be engineered for organ-specific or cell type-specific targeting [83-85]. Importantly, several studies have also demonstrated that mRNA vaccines possess selfadjuvant properties, enabling activation of both innate and adaptive immune responses [49,86,87].

In recent years, several studies have employed the mRNALNP system to target OFAs in HCC. Pan et al. [88] reported an mRNA-LNP encoding a GPC3-IFN-α bispecific protein that suppressed tumour growth, prolonged survival, and showed synergistic effects with PD-1 blockade in mice. Similarly, Yang et al. [89] engineered mRNA-LNPs encoding GPC3-specific CAR constructs to generate liver-targeted macrophages (CAR-Ms). In combination with CD24-Siglec-G blockade, these CAR-Ms enhanced macrophage phagocytosis, reduced tumour burden, and prolonged survival in mouse HCC models.

The field could be further expanded to exploit this vaccine system for targeting other OFAs highly expressed in HCC. Numerous research groups continue to identify novel oncofetal genes and markers that could serve as ideal targets for mRNA vaccines. These oncofetal targets include those mentioned above, as well as ATP-binding cassette subfamily F member 1, Neighbour of Punc E11, granulinepithelin precursor, and multiubiquitin chain-binding protein 1 [90-93]. Several strategies can be employed to enhance the efficacy of mRNA vaccines and ensure effective CD8⁺ T cell activation and robust antitumour responses. These include incorporating multiple antigenic targets into a single vaccine, combining vaccines with immune modulators and/ or adjuvants, and modifying the vaccine to target specific immune cell populations [94-97].

Incorporation of multiple antigenic targets within a single vaccine can enhance antigenic breadth and may help address both intratumoural and intertumoural heterogeneity within individual patients or across patient populations [98]. Since not all OFAs are inherently immunogenic, antigen selection should be streamlined in silico, with candidates of high immunogenic potential screened using computational prediction tools to assess MHC class I binding affinity and predicted T cell receptor (TCR) interactions [99-101]. To further address interpatient heterogeneity, tumour genomic sequencing may then be used to identify patient-specific OFA signatures and determine the suitability of a multi-antigen vaccine or whether a more personalised approach is required [99]. Strong evidence suggests that effective cancer vaccines benefit from both targeting multiple antigens and tailoring antigen selection to the individual tumour landscape [102]. Nevertheless, despite the capacity of mRNA vaccines to encode multiple antigens, practical limits remain. Increasing the number of encoded targets may introduce antigenic competition during antigen processing and presentation by antigen-presenting cells, necessitating a careful balance between antigen selection and antigen number [103,104].

The combination of immunomodulators with vaccines has been shown to be more effective than vaccine monotherapy, as discussed in the previous section. These modulators can either be co-administered with the vaccine as separate entities or co-expressed within the same mRNA construct and delivered as a single formulation [94,105]. Because many OFAs are developmentally expressed selfantigens, a key translational challenge is pre-existing immune tolerance. However, this may be less limiting for AFP and GPC3, since AFP has been shown to be highly immunogenic [106], and GPC3 vaccine approaches have elicited antitumour immunity [59,107,108]. Notably, inclusion of an adjuvant (toll-like receptor 4 [TLR4] agonist) has been reported to disrupt immune tolerance in preclinical HCC models [107]. A more detailed discussion of strategies to overcome immune tolerance is provided in the next section. In humans, a number of mRNA-based cancer vaccines in trials have also been shown to be well-tolerated and immunogenic [109-111], with no issues in repeat intravenous dosing [111]. That said, the long-term safety and durability of responses with repeated intravenous mRNA-LNP dosing remain incompletely defined and warrant further study.

From a formulation perspective, immune tolerance can be addressed by directing LNP delivery toward immune cells such as antigen-presenting cells to induce a stronger and more durable CD8⁺ T cell response. This can be done through tuning lipid composition or decorating the LNP surface with targeting ligands/antibodies to recognise markers expressed on target immune cells [112,113]. As a result, targeted LNPs could also increase dose efficiency (lower doses for a given immune response) and reduce off-target transfection, potentially improving safety profiles. Furthermore, since LNPs delivered intravenously generally accumulate in the liver, liver tropism/accumulation and toxicity are a significant concern [114]. This can also be mitigated by targeted LNPs – for example, reduced hepatocyte transfection and enhanced immune-cell delivery have been reported with T cell-targeted LNPs [114]. Together, these features highlight the adaptability of mRNA vaccines for OFA-targeted HCC immunotherapy.

Beyond mRNA approaches, multiple cell-based therapies have been evaluated in preclinical HCC. Among these, CAR technologies – generated by fusing an antigen-binding domain with membrane-spanning and intracellular signalling domains – enable T cells, macrophages, or natural killer (NK) cells to recognise and eliminate antigen-expressing tumour cells [115]. GPC3- or AFP-targeted CAR-T cells, CAR-macrophages, and CAR-NK cells have demonstrated potent tumour killing effects, cytokine release, and survival benefits [116-125]. Notably, DLK1 – an imprinted OFA upregulated in HCC – was recently targeted by DLK1-directed CAR-T cells, which effectively suppressed DLK1-positive HCC cells both in vitro and in vivo [40]. Furthermore, dual- and multi-specific CAR-T cells targeting combinations such as GPC3 with NKG2D ligands, fibroblast activation protein, PD-1, or B7-H3 (CD276), have shown improved efficacy, persistence, and antitumour activity [119-123]. Likewise, adoptive transfer of TCR-engineered T cells expressing antigen-specific TCRs targeting AFP or GPC3 has yielded promising preclinical results and progressed into earlyphase clinical trials [126-129].

Importantly, these strategies extend beyond HCC. OFAs such as CLDN6, trophoblast glycoprotein (5T4), receptor tyrosine kinase-like orphan receptor 1, and mucin1 are now being targeted by mRNA vaccines or CAR-based therapies in ovarian, pancreatic, and other cancers [130-136]. Collectively, these findings highlight the evolution from early conventional vaccines to advanced mRNA and cell-based platforms (Fig. 2) [5,61,62,66-68,117,127,137-149]. Continued translational research is essential to address current gaps and fully harness the potential of OFA-based therapeutics. Table 1 summarises ongoing and completed clinical trials investigating cellular immunisation strategies targeting OFAs in HCC within the past five years (since January 2020). These data, retrieved from ClinicalTrials.gov as of 20 January 2026, were identified using a search strategy incorporating terms such as “oncofetal”, “CAR”, “T cell”, “TCR”, and the names of all markers examined in this review. To maintain focus on active research, trials listed as terminated, withdrawn, suspended, or with an unknown recruitment status were excluded.

Figure 2.

Timeline of key advancements in AFP/GPC3-directed vaccines, T cell engineering, and the parallel evolution of mRNA platforms for HCC. AFP, alpha-fetoprotein; CAR, chimeric antigen receptor; DC, dendritic cell; DNA, deoxyribonucleic acid; GPC3, glypican-3; HCC, hepatocellular carcinoma; HLA, human leukocyte antigen; LNP, lipid nanoparticle; MAGE-1, melanoma-associated antigen-1; MHC, major histocompatibility complex; mRNA, messenger ribonucleic acid; PFS, progression-free survival; TCE, T cell engager; TCR, T cell receptor.

Table 1.

Clinical trials of OFA-targeted cellular immunisation in HCC (since 2020)

Treatment	Target	Phase	Status	Planned enrolment (n)	Antigen requirement	Year started	Clinical trial identifier
CAR-T	GPC3	I	Recruiting	38	Tumour GPC3 positivity ≥25%	2021	NCT05003895
CAR-T (CATCH T cells)	GPC3/IL-15	I	Recruiting	27	GPC3-positive solid tumour	2021	NCT05103631
CAR-T (AGAR T cells)	GPC3/IL-15	I	Active, not recruiting	24	GPC3-positive solid tumour	2021	NCT04377932
CAR-T (C-CAR031)	GPC3	I	Recruiting	72	Possible tumour GPC3 expression	2021	NCT05155189
CAR-T	GPC3	N/A	Recruiting	20	GPC3-positive tumour	2022	NCT05620706
CAR-T (BOXR1030)	GPC3	I/II	Active, not recruiting	98	GPC3-positive solid tumours	2022	NCT05120271
CAR-T (Ori-C101)	GPC3	I/II	Recruiting	105	GPC3-positive tumour	2022	NCT05652920
TCR-T cell (ET140203)	AFP	I/II	Recruiting	15	Serum AFP >100 ng/mL	2022	NCT04634357
TCR-T cell (ECT204)	GPC3	I/II	Recruiting	20	GPC3-positive tumour	2022	NCT04864054
CAR-T (AZD5851)	GPC3	I/II	Active, not recruiting	94	GPC3-positive tumour	2023	NCT06084884
CAR-T (JWATM214)	GPC3	N/A	Recruiting	12	GPC3-positive tumour	2023	NCT05926726
CAR-T (EU307)	GPC3	I	Recruiting	12	GPC3-positive tumour	2023	NCT05783570
CAR-T (CT011)	GPC3	I	Recruiting	30	GPC3-positive tumour	2023	NCT06560827
CAR-T (C-CAR031)	GPC3	I/II	Recruiting	121	GPC3-positive tumour	2024	NCT06590246
CAR-mRNA-LNP (MT-303)	GPC3	I	Recruiting	70	GPC3-expressing cancers	2024	NCT06478693
CAR-NK cell (SN301A)	GPC3	Early I	Recruiting	12	GPC3-positive HCC	2024	NCT06652243
CAR-T (REVO-UWD-03)	GPC3	Early I	Recruiting	60	Tumour GPC3 expression ≥50%	2024	NCT06653023
CAR-T	GPC3	I/II	Not yet recruiting	30	GPC3-positive HCC	2024	NCT06641453
TCR-T cells (HRYZ-T102)	AFP	I	Recruiting	12	AFP-positive HCC/solid tumours	2024	NCT06515314
CAR-T (SC-CAR. GPC3xIL15.21 CAR T cells)	GPC3/IL-15/IL-21	I	Recruiting	21	GPC3-expressing solid tumour	2025	NCT07148050
CAR-T (21.15.GPC3-CAR T cells)	GPC3/IL-15/IL-21	I	Recruiting	21	GPC3-positive solid tumour	2025	NCT06198296
TCR-T Cell (super Hi-TCR-T cells)	Nectin4/NKG2DL/TROP2/B7H3/GPC3/FAP	I/II	Recruiting	30	Tumour Nectin4, NKG2DL, TROP2, B7H3 and GPC3 expression	2025	NCT06902389

Data represent a snapshot as of 20 January 2026 and were identified using search terms including “oncofetal”, “CAR”, “T cell”, “TCR”, and the names of all oncofetal markers discussed in this review. Trials with terminated, suspended, withdrawn, or unknown recruitment status were excluded.

AFP, alpha-fetoprotein; CAR, chimeric antigen receptor; GPC3, glypican-3; HCC, hepatocellular carcinoma; OFA, oncofetal antigen; TCR, T cell receptor.

Epitope-based vaccine strategies to overcome liver tolerance

The liver is inherently tolerogenic. Immune tolerance in the liver is largely regulated by specialised antigenpresenting immune cells, which paradoxically are the same populations responsible for recognising foreign antigens and initiating immune responses [150]. While these tolerogenic mechanisms protect the host from excessive responses to innocuous antigens, they also limit immunosurveillance (the immune system’s ability to eliminate tumour cells), thereby contributing to tumour progression [135]. This presents a particular challenge for cancer vaccines administered intravenously, as they must pass through the liver. Additionally, because OFAs are developmentally expressed self-antigens, immune tolerance can further reduce T cell responsiveness [151].

Several strategies have been investigated to overcome this tolerance barrier. The use of adjuvants such as 2′3′-cyclic guanosine monophosphate-adenosine monophosphate, resiquimod, and polyinosinic:polycytidylic acid with vaccines has shown promise, potentiating innate immune activation [150]. Another approach is to optimise antigen design by focusing on immunodominant peptides or epitopeoptimised peptides enriched for regions preferentially recognised by CD8⁺ T cells, thereby enhancing antigenspecific immune responses [152].

Over the years, multiple groups have identified T cell immunodominant epitope sites for human AFP [68,153-156]. These epitopes are often discovered using computational prediction algorithms capable of screening multiple candidate sites [157]. For example, Butterfield et al. [140,142,154] identified four immunodominant human leukocyte antigen (HLA)-A0201-restricted AFP-derived peptides that elicited both cytotoxic and cytokine-producing T cell responses in healthy individuals as well as AFP-positive HCC patient. In a pilot Phase-I clinical trial, HLA-A0201-positive patients with AFP-expressing HCC mounted T cell responses to most, if not all, of these epitopes [142]. A subsequent Phase-I/II clinical trial by the same group demonstrated that DC vaccines pulsed with the four HLA-A2-restricted AFP peptides induced strong AFP-specific immune responses [61]. These findings show that AFP-specific T cells are capable of recognising immunodominant AFP peptides presented via MHC-I despite high circulating levels of tumour-derived AFP [142]. Thimme et al. [158] further suggested that vaccines incorporating whole AFP containing multiple immunodominant epitopes may achieve superior clinical efficacy compared with targeting a limited set of epitopes.

In preclinical studies, computationally optimised AFP epitopes have further enhanced vaccine potency. Hong et al. [106] demonstrated that lentiviral vectors encoding an optimised AFP epitope, but not wild-type AFP, strongly activated CD8⁺ T cells and protected mice against tumour challenge and carcinogen-induced HCC. Remarkably, CD8⁺ T cells generated by this vaccine also cross-recognised native AFP peptides, effectively killing AFP-expressing tumour cells. Other groups have expanded these epitopebased strategies to DC platforms. DCs modified with human AFP peptides successfully induced peripheral blood mononuclear cells to differentiate into antigen-specific CD8⁺ T cells, with stronger responses than AFP alone [159-161]. Likewise, Dargel et al. [129] combined peptide prediction algorithms with mass spectrometry to identify immunodominant GPC3 epitopes. Transfection of DCs with HLA-A2 RNA and the predominant GPC3 peptide (GPC3-367) generated primary CD8⁺ T cells expressing transgenic TCRs that specifically recognised GPC3-367 on HLA-A2 [129]. These T cells killed GPC3-expressing hepatoma cells in vitro and slowed tumour growth in xenograft models, highlighting the potential of epitope-driven adoptive T cell therapy to bypass tolerance.

Beyond AFP and GPC3, similar epitope-based vaccine approaches have been explored in other cancers. In breast cancer, genome sequencing and predictive algorithms have been used to identify polyepitope neoantigens, which, when delivered as DNA vaccines, elicited robust neoantigen-specific immune responses [162]. When combined with ICIs, these vaccines achieved potent antitumour activity in preclinical models. Similarly, administration of two highly immunogenic epitopes from the oncofetal protein IMP-3 – frequently overexpressed in head-and-neck, lung, oesophageal, and several other malignancies – together with anti-PD-1 therapy, enhanced the expansion of IMP-3-specific CD8⁺ T cells [163]. In addition, overlapping peptide libraries spanning the oncofetal protein 5T4, expressed in prostate, renal, and colorectal cancers, have been used to map immunogenic regions capable of inducing IFN-γ-producing T cells in both mice and humans [164,165].

To further enhance immunogenicity, multiepitope strategies have been explored. Martínez-Cortés et al. [166] developed the Variable Epitope Library (VEL) vaccine, which incorporates multiple mutated epitopes of the OFA/immature laminin receptor protein, expressed in breast, head-andneck, ovarian, brain, prostate, and haematologic malignancies. In mouse models, VEL vaccines elicited strong CD8⁺IFN-γ⁺ and CD4⁺IFN-γ⁺ T cell responses, increased tumour-infiltrating lymphocytes, and inhibited tumour growth and metastasis after a single dose [137]. Collectively, these studies highlight that multiepitope vaccines – whether delivered as peptides, DNA, or protein constructs – offer stronger and broader antitumour immunity than single-antigen vaccines. When combined with ICIs or other immunomodulators, such strategies may be essential to overcome tolerance and achieve durable tumour control.

IMMUNOMODULATORS AND BIOLOGICS

HCC is pathologically driven by extensive angiogenesis, primarily mediated by high VEGF expression, which contributes to the immunosuppressive TME. VEGF signalling recruits various immunosuppressive cells, including Tregs (such as CTLA4), FOLR2⁺ TAMs, PLVAP⁺ ECs, and myeloid-derived suppressor cells (MDSCs) via the VEGFNOTCH pathway, while concurrently inhibiting the activation of antigen-presenting cells and cytotoxic T lymphocytes [5,167]. As a result, VEGF represents an attractive target for immunotherapy, as its inhibition may indirectly reverse oncofetal reprogramming of FOLR2⁺ TAMs and PLVAP⁺ ECs, limit Treg expansion, and enhance overall antitumour immune responses. However, despite these advantages, monotherapy with VEGF inhibitors has demonstrated limited efficacy [7]. At high doses, anti-VEGF therapy causes excessive vascular pruning, which reduces perfusion and exacerbates hypoxia [7]. Interestingly, FOLR2 expression has been linked to hypoxia, where FOLR2⁺ TAMs and MDSCs serve as distinct markers of immunosuppressive states [168]. Activation of hypoxia-induced transcription factors within the hypoxic TME further promotes the release of proangiogenic growth factors such as VEGF-A, TGF-β, and PDGF [169]. These processes illustrate the complex interplay between VEGF signalling, hypoxia, and immune suppression, which collectively sustain tumour progression.

Given that monotherapy with anti-VEGF agents has achieved limited success in HCC, combining VEGF inhibition with ICIs, such as anti-PD-1/PD-L1 and/or anti-CTLA-4, has emerged as a promising therapeutic approach. HCC tumours are often infiltrated by immune cells that become functionally exhausted or suppressed through the upregulation of checkpoint receptors [170]. Notably, PD-1 is highly expressed not only on T cells but also on FOLR2⁺ macrophages, and PD-1/PD-L1 blockade has been shown to reduce the abundance of FOLR2⁺ TAMs, suggesting that patients with high FOLR2 expression could benefit from such treatment [171]. The combination of anti-PD-L1 and anti-VEGF has demonstrated synergistic effects by reversing VEGF-induced immunosuppression, enhancing antigen recognition, reprogramming the TME, and promoting T cell recruitment, while ICIs restore T cell-mediated tumour killing [7]. This mechanistic synergy underlies the success of the IMbrave150 trial, which led to the approval of the atezolizumab (anti-PD-L1) plus bevacizumab (anti-VEGF) combination as the first-line standard of care for advanced HCC by the U.S. FDA in 2020, achieving significant improvements in overall survival (OS) and progression-free survival (PFS) [12,172]. Preclinical studies further demonstrated that this combination reduces VEGFR2 expression and Treg populations while increasing CD8⁺ T cell infiltration [173].

Beyond VEGF and PD-L1 co-blockade, combinatorial inhibition of PD-1/PD-L1 and CTLA-4 has also been explored [174-176]. CTLA-4, expressed on activated T cells and Tregs, competes with CD28 for binding to CD80/CD86, while FOLR2⁺ TAMs are known to interact with Tregs through the CD86-CTLA-4 axis in HCC [5,177]. This relationship underscores the link between immune checkpoint activity, oncofetal reprogramming, and immunotherapy response. The CheckMate-040 trial investigated nivolumab (anti-PD-1) combined with ipilimumab (anti-CTLA-4) in patients with HCC who had progressed after sorafenib treatment, showing encouraging objective response rates that led to its accelerated FDA approval as a second-line treatment in 2020 [174,175]. This success laid the foundation for the HIMALAYA trial, in which tremelimumab (anti-CTLA-4) combined with durvalumab (anti-PD-L1) demonstrated significant survival benefit, resulting in its approval as an alternative first-line treatment for advanced HCC in 2022 [176]. This combination represents a viable option for patients with an increased risk of bleeding or resistance to anti-VEGF therapy [178].

Other immunotherapy combinations have produced mixed results. The LEAP-002 trial, evaluating pembrolizumab (anti-PD-1) with lenvatinib (a tyrosine kinase inhibitor), failed to achieve a significant OS improvement despite favourable response rates and PFS gains [179]. In contrast, a recently completed phase II trial (NCT04542837) combining anti-PD-1, anti-CTLA-4, and anti-VEGF therapy demonstrated promising efficacy in advanced HCC [180]. Additionally, ICIs are being evaluated in combination with ablation, vaccines, biologics, and as neoadjuvant treatments prior to surgery [181-183]. Beyond PD-1, PD-L1, and CTLA-4, other checkpoint molecules such as TIM-3 and LAG-3 are also upregulated in HCC, contributing to immune escape [184,185]. High expression of these molecules is associated with poor prognosis and reduced survival. Nevertheless, tumours with high checkpoint expression, while highly immunosuppressive, may also show better responsiveness to ICIs and a greater potential for TME reprogramming, emphasising the importance of biomarker-guided therapeutic strategies and patient stratification [7,8,186].

Beyond checkpoint inhibitors, monoclonal antibody therapeutics targeting OFAs are emerging (Fig. 3). These can be categorised into three major formats based on structure and function: monospecific, bispecific, and antibody-drug conjugate (ADC) [187]. Monospecific antibodies bind a single antigen. Indeed, all approved ICIs are monospecific antibodies. Among the OFAs targeted in HCC, GPC3 is the most extensively studied. Codrituzumab (GC33/RO5137382/RG7686) is a humanised anti-GPC3 IgG1 monoclonal antibody that interacts with CD16 on NK cells to induce antibody-dependent cellular cytotoxicity and complement-dependent cytotoxicity, thereby inhibiting tumour growth [188,189]. Phase-I and -II clinical trials demonstrated that codrituzumab was well tolerated and exhibited antitumour activity [190-192]. However, it did not significantly improve OS or PFS. Several additional GPC3-targeting antibodies – HN3, YP7, and HS20 – developed by the Ho Laboratory at the National Cancer Institute have shown potent antitumour effects in preclinical studies by disrupting Wnt/β-catenin and YAP signalling via interference with GPC3’s co-receptor functions, resulting in tumour growth inhibition [193,194]. Beyond GPC3, monoclonal antibodies targeting PLVAP and DLK1 have also been developed [195,196]. Anti-PLVAP antibodies have demonstrated enhanced tumour necrosis and suppression with minimal systemic toxicity [195], while anti-DLK1 (CBA-1205) exhibited durable antitumour efficacy in combination with lenvatinib and favourable toxicity profiles [196]. Additionally, combining anti-PLVAP antibody therapy with gemcitabine and cisplatin further enhanced tumour growth inhibition [197].

Figure 3.

Types of antibody therapeutics targeting GPC3 in HCC. GPC3, glypican-3; NK, natural killer.

Bispecific antibodies (BsAbs) combine two antibody specificities within a single molecule, enabling dual targeting of distinct antigens [194]. Typically, BsAbs bind a tumourspecific antigen – often an OFA such as GPC3 – and an activating receptor on immune cells, thereby recruiting and stimulating immune cells in close proximity to tumour cells [198]. Most GPC3-targeting BsAbs developed for HCC engage immune effector receptors such as CD3, NKp46, 4-1BB, or CD16A, facilitating T cell or NK-cell-mediated cytotoxicity [199-202]. GPC3/CD3 BsAbs have demonstrated strong antitumour activity, inducing efficient T cell recruitment, tumour lysis, and local TME inflammation [203]. A GPC3/CD3 bispecific probe has also been engineered for theragnostic use, allowing both tumour detection and treatment with favourable preclinical results [204]. Beyond GPC3, a BsAb targeting CLDN6 and GPC3 (6PHU3) has been developed, showing potent tumour killing and improved survival in preclinical models [205].

ADCs have also shown promise in HCC. Anti-GPC3 antibodies linked to duocarmycin (hYP7-DC) or pyrrolobenzodiazepine (hYP7-PC) exhibited strong potency in GPC3-positive models, with hYP7-DC demonstrating synergistic activity with gemcitabine in vitro and in vivo [206]. Similarly, an HN3-mPE24 immunotoxin, consisting of an anti-GPC3 antibody fused to a modified Pseudomonas exotoxin, induced marked tumour regression and improved survival in preclinical studies [207]. Outside of GPC3, a CLDN6-DM1 ADC incorporating the cytotoxic agent mertansine also showed robust antitumour efficacy, both as a single agent and in combination with sorafenib [36]. Table 2 summarises ongoing and completed clinical trials of oncofetal-targeted antibody therapies in HCC registered on ClinicalTrials.gov since January 2020. Records were compiled from searches cur-rent to 20 January 2026, using keywords including “oncofetal”, “antibody”, “monoclonal”, “bispecific”, and the names of all oncofetal markers discussed in this review. Trials marked as terminated, withdrawn, suspended, or with unknown recruitment status were excluded to focus on active research.

Table 2.

Clinical trials of OFA-targeted biologics in HCC (since 2020)

Treatment	Target	Phase	Status	Planned enrolment (n)	Antigen requirement	Year started	Clinical trial identifier
Humanised monoclonal antibody (CBA-1205)	DLK1	I	Recruiting	66	Not specified	2020	NCT06636435
Recombinant antibody (CSR02)	PLVAP/tissue factor	Early I	Recruiting	43	Not specified	2021	NCT04601428
Humanised bispecific T cell-redirecting antibody (ERY974)	GPC3/CD3	I	Active, not recruiting	179	Not specified	2021	NCT05022927
Humanised monoclonal antibody-drug conjugate (TORL-4-500)	DLK1	I	Recruiting	70	Not specified	2023	NCT06005740
Humanised monoclonal antibody-drug conjugate (ZW251)	GPC3	I	Recruiting	100	Not specified	2025	NCT07164313
Trispecific T cell-engaging antibody (AZD9793)	GPC3/CD3/CD8	I/II	Recruiting	304	GPC3-positive tumour	2025	NCT06795022

Data represent a snapshot as of 20 January 2026 and were identified using search terms including “oncofetal”, “antibody”, “monoclonal”, “bispecific”, and the names of all oncofetal markers discussed in this review. Trials with terminated, suspended, withdrawn, or unknown recruitment status were excluded.

DLK1, delta-like homologue 1; GPC3, glypican-3; HCC, hepatocellular carcinoma; OFA, oncofetal antigen; PLVAP, plasmalemma vesicle-associated protein.

Overall, immunomodulatory strategies encompassing both checkpoint blockade and antibody-based therapeutics hold great promise for reshaping the HCC treatment landscape. VEGF inhibition can remodel the immunosuppressive vasculature, ICIs can reinvigorate cytotoxic T cell responses, and OFA-targeting biologics expand precision immunotherapy options. Future research should focus on developing novel checkpoint inhibitors guided by current immune signatures, exploring additional OFA targets for biologics, and testing synergistic combinations that integrate vaccines, biologics, and ICIs. Patient stratification based on oncofetal and immune checkpoint marker expression will be essential for guiding personalised therapy and optimising clinical outcomes. These integrated approaches are likely to define the next generation of immunotherapeutic strategies in HCC, leveraging angiogenesis blockade, oncofetal reprogramming, and immune activation to improve survival and long-term disease control.

CONCLUSIONS AND FUTURE PERSPECTIVES

Currently, vaccine and biologic options targeting OFAs in HCC remain limited. Over the past five years, most clinical efforts have focused on T cell-based therapies rather than vaccines and biologics. The rapid success of mRNA-LNP vaccines against COVID-19 has highlighted the transformative potential of this platform in oncology. Compared to conventional vaccine systems, mRNA-LNP vaccines offer several advantages, including precise antigen delivery, enhanced tumour-specific killing, reduced off-target toxicity, scalable manufacturing, and the capacity to encode multiple antigens or immunomodulatory molecules within a single formulation [208]. Leveraging this technology for HCC could accelerate the development of vaccines directed against OFAs, opening a new avenue for effective immunotherapy.

A major barrier to vaccine efficacy in HCC remains the profoundly immunosuppressive TME. Even when the target antigen is immunogenic, the TME can limit T cell infiltration, promote T cell exhaustion, and ultimately drive resistance to immunotherapy [209]. To overcome these challenges, a rational approach would be to combine vaccines with immunomodulators, such ICIs and/or biologic agents, to achieve synergistic effects on TME remodelling and tumour elimination. Repeated vaccine dosing may also be necessary to sustain durable antitumour immunity rather than relying on a single administration. Supporting this concept, preclinical and early clinical studies combining mRNA vaccines with ICIs, biologics, or adoptive T cell therapies have shown encouraging results [210,211].

Another key consideration is immune tolerance to OFAs, which can compromise long-term vaccine efficacy. Strategies such as targeting immunodominant or optimised epitopes may enhance response durability. Host genetics also play a crucial role: polymorphisms in HLA alleles across different ethnic and regional populations influence the binding affinity of tumour antigens and the quality of T cell responses, underscoring the need to design vaccines that account for population-specific immunogenetics [212]. Incorporating multiple tumour-associated antigens, including OFAs, within a single formulation may further limit immune escape and improve tumour control.

Timing and disease stage are also important considerations for antigen selection. For example, an OFA-targeted vaccine could be considered prophylactically in high-risk individuals (e.g., those with chronic hepatitis B or C) prior to malignant transformation, since evidence suggests that elements of the oncofetal microenvironment, such as foetal-like stromal and immune programs, can already be detected in chronically inflamed or cirrhotic liver [213]. Furthermore, AFP, which can be present in cirrhotic patients before malignant development, has been associated with subsequent HCC risk [214]. Accordingly, a risk model was developed to predict HCC development using AFP measurements in patients with hepatitis C virus-related cirrhosis [215]. Integrating such risk stratification with knowledge of premalignant oncofetal marker expression could help identify patients most likely to benefit from preventive vaccination approaches in advanced liver disease (Fig. 4).

Figure 4.

Disease stage-specific use of OFA vaccines in the prevention or treatment of HCC. AFP, alpha-fetoprotein; GPC3, glypican-3; HCC, hepatocellular carcinoma; OFA, oncofetal antigen; TME, tumour microenvironment.

Conversely, therapeutic vaccines targeting OFAs present in both the tumour and TME may be deployed in established HCC. Depending on disease stage and the targets confirmed by tumour profiling (e.g., IHC/RNA), vaccination could be used in the adjuvant setting (after curative resection) to reduce relapse risk – either as monotherapy or alongside immunotherapy – or in unresectable/advanced disease as part of combination regimens, including in patients with limited benefit from prior immunotherapy (Fig. 4).

While T cell therapies continue to show promise, their widespread application in HCC is constrained by high manufacturing costs, complex logistics, and limited scalability. In this context, mRNA LNPs encoding CAR or TCR constructs could offer a more scalable and flexible alternative. This strategy may be particularly relevant given the modest efficacy of CAR-T cell therapy in solid tumours [216], suggesting that integration with mRNA vaccine platforms could enhance potency and help overcome immune tolerance.

Ultimately, there remains a critical need for novel vaccines and combinatorial immunotherapy strategies that improve response rates and survival outcomes in HCC. mRNA vaccines are particularly attractive due to their low toxicity profile, rapid and adaptable production, and potential for personalised design. However, key refinements are needed before their successful translation into late-phase clinical testing. These include optimisation of prime-boost schedules, rational antigen selection, and strategic combinations with other immunotherapies. Incorporating patient stratification based on biomarker expression and molecular profiling – an approach currently being explored in several clinical trials – will further guide treatment selection and combination strategies, ultimately improving patient outcomes (Fig. 5). Collectively, integrating mRNA vaccine platforms with OFA targeting and immune modulation represents a promising next frontier in HCC immunotherapy.

Figure 5.

Biomarker-guided stratification strategy for selecting OFA-targeting therapeutic modalities. ADC, antibody-drug conjugate; AFP, alpha-fetoprotein; CAR, chimeric antigen receptor; CLDN6, claudin-6; CTLA4, cytotoxic T-lymphocyte-associated protein 4; DLK1, deltalike homologue 1; GPC3, glypican-3; IHC, immunohistochemistry; mRNA, messenger ribonucleic acid; OFA, oncofetal antigen; PD-1/PDL1, programmed death-1/programmed death-ligand 1; PLVAP, plasmalemma vesicle-associated protein; POSTN, periostin; RNA, ribonucleic acid; SALL4, Sal-like protein 4; TME, tumour microenvironment.

Abbreviations

5T4

trophoblast glycoprotein

ABCF1

ATP-binding cassette subfamily F member 1

ADC

antibody-drug conjugate

AFP

alpha-fetoprotein

ASPH

aspartate β-hydroxylase

BsAb

bispecific antibody

CAF

cancer-associated fibroblast

CAR

chimeric antigen receptor

cGAMP

2′3′-cyclic guanosine monophosphate-adenosine monophosphate

CLDN6

claudin-6

CTLA4

cytotoxic T-lymphocyte-associated protein 4

DC

dendritic cell

DLK1

delta-like homologue 1

DNA

deoxyribonucleic acid

EC

endothelial cell

EGCG

epigallocatechin gallate

EMT

epithelial-mesenchymal transition

ESRP2

epithelial splicing regulatory protein 2

FAP

fibroblast activation protein

FDA

Food and Drug Administration

FOLR2

folate receptor beta

GEP

granulin-epithelin precursor

gp96

glycoprotein 96

GPC3

glypican-3

HCC

hepatocellular carcinoma

HLA

human leukocyte antigen

HLF

hepatic leukaemia factor

HSP70

heat shock protein 70

ICI

immune checkpoint inhibitor

IHC

immunohistochemistry

iLRP

immature laminin receptor protein

IMP1-3

IGF2 mRNA-binding proteins 1-3

LAG-3

lymphocyte-activation gene 3

LIN28B

lin-28 homolog B

LNP

lipid nanoparticle

MAGE-1

melanoma-associated antigen-1

MCB1

multiubiquitin chain-binding protein 1

MDSC

myeloid-derived suppressor cell

MHC

major histocompatibility complex

mRNA

messenger ribonucleic acid

MUC1

mucin 1

NOPE

Neighbor of Punc E11

NK

natural killer

OFA

oncofetal antigen

OS

overall survival

PD-1/PD-L1

programmed death-1/programmed death-ligand 1

PEG10

paternally expressed gene 10

PFS

progression-free survival

PLVAP

plasmalemma vesicle-associated protein

Poly(I:C)

polyinosinic:polycytidylic acid

POSTN

periostin

R848

resiquimod

RNA

ribonucleic acid

ROR1

receptor tyrosine kinase-like orphan receptor 1

SALL4

Sal-like protein 4

SPINK1

serine peptidase inhibitor Kazal type 1

TAM

tumour-associated macrophage

TCE

T cell engager

TCR

T cell receptor

TLR4

toll-like receptor 4

TME

tumour microenvironment

Treg

regulatory T cell

TRIM71

tripartite motif containing protein 71

TRM

tissue-resident macrophage

VEGF

vascular endothelial growth factor

VEL

Variable Epitope Library