The origin of liver cancer TICs remains debated. One prominent hypothesis suggests that TICs may stem from hepatic progenitor cells (HPCs). Specifically, CD34+ HPCs, essential for liver regeneration, have been proposed as a potential source of liver cancer stem cells (LCSCs). In a study by Park et al., CD34+ cells isolated from the PLC/PRF/5 hepatoma cell line were found to function as TICs, capable of generating HCC, cholangiocarcinoma (CC), and combined hepatocellular cholangiocarcinoma (CHC) in immunodeficient mice (17). These cells expressed both hepatic and hematopoietic markers, suggesting they might result from the fusion of hepatobiliary stem cells and myeloid precursors. This fusion likely provides tumor cells with cross-lineage characteristics, which contribute to tumor heterogeneity and malignant progression (18). Moreover, subpopulations of TICs that express specific surface antigens like CD133, CD44, and CD90 show distinct tumorigenic potentials, further contributing to the histological diversity observed in liver cancers (17).

An alternative theory involves the plasticity of mature hepatocytes. Evidence indicates that mature hepatocytes can serve as the origin of intrahepatic cholangiocarcinoma (ICC) cells. Mesencephalic astrocyte-derived neurotrophic factor (MANF), upregulated in ICC, promotes the transformation of mature hepatocytes into ICC cells via its interaction with CK19 and activation of the Notch signaling pathway (19). Furthermore, hepatocytes have the potential to transform directly into cancer cells following sequential genomic damage and dedifferentiate into precursor cells expressing progenitor cell markers (20).

The diagnosis of liver cancer TICs is fraught with challenges, primarily stemming from the lack of specific biomarkers. While markers such as CD133, EpCAM, and certain lncRNAs have been associated with liver TICs, their diagnostic accuracy remains constrained due to non-exclusive expression in other liver or tumor microenvironment cell types, which can lead to false positives or negatives (21). Compounding this issue is the complexity of the TME, where immunosuppressive factors—including immune cell infiltration, stromal components, and secreted molecules—can modify biomarker expression and compromise detection accuracy (22, 23). Additionally, the spatial and temporal heterogeneity of liver tumors, characterized by variations in genetic mutations, epigenetic changes, and cellular composition across different tumor regions, further hinders the development of universal diagnostic methods (24, 25). Clinically, the difficulty of early-stage detection in primary liver cancer, coupled with limited treatment options, exacerbates the challenge of identifying TICs at actionable stages. Classic biomarkers like AFP and des-gamma-carboxy prothrombin (DCP) exhibit low sensitivity and specificity for early detection (26), while newer biomarkers identified through proteomic, metabolomic, and genomic analyses hold promise but require validation for routine clinical use (27).

Therapeutic resistance poses a significant obstacle in managing liver TICs. A key example is the response of TICs to sorafenib, a first-line treatment for advanced hepatocellular carcinoma. While sorafenib initially suppresses the proliferation of certain HCC cell lines, it paradoxically causes a transient expansion of TIC populations. These TICs activate the YAP signaling pathway, a potent oncogenic transcriptional coactivator that enhances tumor cell survival, proliferation, and invasion (28). Resistant TICs display increased sphere-forming ability, higher proportions of EpCAM-positive cells, and upregulated YAP expression (29), all of which are associated with stemness and aggressive tumor behavior. Notably, combining sorafenib with a YAP inhibitor has shown synergistic anti-tumor effects, suggesting potential for dual pathway inhibition.

Epigenetic reprogramming also contributes to TIC-mediated resistance. In sorafenib-resistant HCC, EZH2 overexpression enhances NOTCH1 signaling through H3K27me3-mediated chromatin modifications, supporting TIC self-renewal and tumorigenicity (30). Inhibiting EZH2 can restore sorafenib sensitivity by suppressing NOTCH1 and weakening TIC properties, offering a promising strategy to overcome resistance.

Liver cancer TICs rely on multiple molecular pathways to sustain their stemness and drive liver Cancer progression. A crucial pathway in this process is the Wnt/β-catenin pathway. For instance, LncTIC1 interacts with the N-terminal of β-catenin, inhibiting its phosphorylation and stabilizing the protein to activate Wnt/β-catenin signaling, thereby promoting TIC self-renewal and tumorigenesis (9). Similarly, TFAP4 amplifies this signaling cascade by directly binding to the promoters of DVL1 and LEF1, further reinforcing TIC characteristics (31). Additionally, activation of the Wnt/β-catenin pathway promotes M2 polarization of tumor-associated macrophages (TAMs), which facilitate immune evasion by secreting immunosuppressive cytokines such as IL-10 and TGF-β (32). Another critical regulator of liver cancer TICs is the interleukin-8 (IL-8) signaling axis. CD133+ liver TICs exhibit significant dysregulation of this pathway, evidenced by genomic and proteomic overexpression of IL-8 (8). This aberrant expression promotes endothelial cell proliferation, tube formation, and TIC stemness maintenance. Mechanistically, NTS binding to NTSR1/2 on CD133+ TICs triggers MAPK phosphorylation, driving IL-8 and CXCL1 production. IL-8 then acts via CXCR1/2 in an autocrine/paracrine manner to sustain MAPK activation, forming a positive feedback loop (8).

Epigenetic regulation by non-coding RNAs further modulates TICs stemness and metastatic potential. For instance, the circular RNA cia-MAF binds to the MAFF promoter and recruits the TIP60 histone acetyltransferase complex, driving MAFF expression which is a transcription factor critical for TIC self-renewal and metastasis. Likewise, the long non-coding RNA lncHOXA10 interacts with SNF2L and recruits the NURF chromatin remodeling complex to the HOXA10 promoter in liver cancer TICs, contributing to HOXA10 expression and liver cancer progression (33). These epigenetic mechanisms also synergize with canonical signaling pathways. For example, SHH signaling, activated by the PARD3-aPKC interaction, upregulates stemness genes such as SOX2 and correlates with advanced tumor stages (10) ( Figure 1 ).

The TME, which consists of a variety of non-cancerous cells, extracellular matrix components, and soluble factors, plays a critical role in influencing the behavior of liver cancer TICs. A case in point is CAFs, key stromal cells in the TME, interact with TICs to promote tumor progression. In a three-dimensional (3D) co-culture organoid model of primary murine liver tumors, the presence of CAFs significantly increased the number of LGR5+ TICs in the tumor compared to organoids lacking CAFs. This observation suggests that CAFs support LGR5+ liver cancer TICs, facilitating tumor formation, growth, and metastasis (34). Additionally, CAFs can enhance the stemness of TICs by secreting cytokines such as interleukin-6 (IL-6), which activate signaling pathways like IL-6-STAT3-NOTCH, thereby promoting tumor progression (35).

TICs interact with immune cells within the TME to establish an immunosuppressive milieu. Specifically, TREM2 macrophages are associated with glycolysis and PKM2 expression in HCC cells, likely mediated by the secretion of IL-1β, which promotes malignant phenotypes of HCC cells (36). Meanwhile, CD49f-high TICs recruit tumor-promoting neutrophils through the CXCL2-CXCR2 axis, establishing an immunosuppressive environment within the TME (12). This crosstalk between TICs and immune cells highlights the intricate tumor-immune microenvironment and holds important implications for the development of effective immunotherapy strategies (37).

Immunotherapy is a promising therapeutic approach for liver cancer TICs. Among various immunotherapeutic agents, immune checkpoint inhibitors (ICIs) are a prominent class. In the treatment of HCC, the combination of the anti-programmed cell death ligand 1 (PD-L1) antibody atezolizumab and the VEGF-neutralizing antibody bevacizumab has demonstrated significant efficacy. Consequently, this regimen is now recommended as the first-line systemic therapy for advanced HCC (15). Additionally, combining different ICIs, like nivolumab and ipilimumab, has yielded promising outcomes in second-line treatments, as evidenced by the CheckMate 040 study (38). Nevertheless, ICI monotherapy demonstrates limited efficacy in HCC, with objective response rates of 10–20% and no significant overall survival (OS) improvement observed in phase III clinical trials such as CheckMate 459 and KEYNOTE-240 (39).

Adoptive cell therapy (ACT) and peptide- or dendritic cell (DC)-based vaccines are also under active investigation. ACT, including chimeric antigen receptor T-cell (CAR-T) therapy targeting tumor-associated antigens (TAA) like glypican-3 (GPC3) and alpha-fetoprotein (AFP), has demonstrated encouraging results in early-phase clinical trials by specifically recognizing and eliminating tumor cells without MHC restriction (40).

Peptide-based vaccines, exemplified by GPC3-targeted formulations, induce cytotoxic T lymphocyte (CTL) infiltration into tumors and can improve overall survival in HCC patients (41). Similarly, DC vaccines targeting tumor antigens or neoantigens help remodel the immunosuppressive TME, by activating tumor-reactive T cells, showing good safety and immunogenicity profiles in early trials (42, 43). However, the liver’s unique immune microenvironment limits immune cell infiltration and function, reducing the effectiveness of these therapies. Therefore, identifying predictive biomarkers for immunotherapy response remains crucial.

Multiple therapeutic strategies are being developed to address the challenges posed by liver cancer TICs. The Wnt/β-catenin pathway is crucial in liver cancer, and 6-C-(E-phenylethenyl) naringenin (6-CEPN) inhibits TIC self-renewal, migration, and invasion by activating GSK3β, leading to the degradation and suppression of the Wnt/β-catenin pathway (52). In addition, selective tyrosine kinase inhibitors, such as Indo5, target the c-Met and Trk receptor tyrosine kinases, which are frequently co-activated in HCC and associated with poor prognosis (53). By blocking these pathways, Indo5 disrupts the related signaling and suppresses tumor growth. Equally important, Bio-SS-TS, a compound combining biotin and taraxasterol (TS), exploits the overexpression of biotin receptors on tumor cells for targeted delivery (54). It enhances mitochondria-dependent apoptosis and inhibits tumor development.

In parallel, genome editing also holds great promise in targeting liver cancer TICs. Marayati et al. used CRISPR/Cas9 to knockout PIM3 in hepatoblastoma cells, leading to reduced cell proliferation, viability, and stemness (55). Regarding non-coding RNAs, they are emerging as crucial therapeutic targets. Li et al. engineered an artificially designed lncRNA that functions as a molecular sponge, sequestering multiple oncogenic miRNAs and inhibiting tumor growth in xenograft HCC mouse models (56). Yao et al. demonstrated that miR-186 suppresses the expansion of liver CSCs by targeting PTPN11 (57). These advances suggest promising avenues for personalized TIC-targeted therapies.

Overcoming the liver’s immunosuppressive microenvironment is a key area for future research. Recent studies have identified several mechanisms by which TICs facilitate immune suppression and resistance (58). The overexpression of CD155 on TICs interacts with immune checkpoint receptors, leading to T cell exhaustion. Combining CD155 blockade with anti-PD-1/PD-L1 therapy has shown enhanced anti-tumor responses in preclinical models (12).

Additionally, a unique subset of TAMs expressing CD19 has been discovered in HCC. CD19+ TAMs exhibit immunosuppressive functions by upregulating PD-L1 and CD73, thereby inhibiting T cell activity, and direct targeting with anti-CD19 CAR-T cells has been shown to suppress tumor growth (59).

The immunosuppressive TME in HCC is further characterized by diverse immune cell populations and signaling pathways that impede effective immune responses (7). Strategies to overcome this include the use of inhibitors targeting the TGF-β/IL-10 pathway and metabolic modulators to enhance immune activation (60, 61). Moreover, accurately assessing and predicting tumor responses to immunotherapy remain critical challenges. Artificial intelligence (AI) integrating imaging, pathology, genomics, and clinical data show promise in predicting treatment responses and survival outcomes in HCC patients undergoing immunotherapy (62). Furthermore, MRI-based response criteria, particularly 3D quantitative enhancement tools (qEASL), have significant potential for predicting overall survival and identifying non-responders (63).