Glucose metabolic reprogramming in HCC is characterized by the Warburg effect, in which glycolysis remains the predominant pathway supporting bioenergetic and biosynthetic demands even under aerobic conditions^[10]. This metabolic shift is mediated by coordinated mechanisms, including upregulation of glucose transporter 1 (GLUT1) to increase glucose uptake, ectopic expression and activation of hexokinase 2 (HK2) in place of the physiological isoform HK4 to enhance glucose phosphorylation efficiency, and elevated expression of pyruvate kinase M2 (PKM2) to augment glycolytic flux^[11,12]. These alterations are driven by aberrant activation of the Wnt/β-catenin and PI3K/AKT/mTOR signaling pathways and are further reinforced under hypoxia through hypoxia-inducible factor-1 alpha (HIF-1α)-mediated responses^[13,14]. Collectively, these changes establish a metabolic program that supports rapid tumor cell proliferation and metabolic flexibility. A major immunosuppressive consequence of the Warburg effect is glucose competition, which represents a central mechanism of immune evasion in HCC. Preferential glucose uptake by tumor cells markedly limits energy availability to CD8⁺ T cells and suppresses their effector functions, including immunological synapse formation, target cell killing, and clonal expansion^[15]. This suppression occurs because T cell activation depends on the glycolytic intermediate phosphoenolpyruvate to sustain the T cell receptor (TCR)-mediated Ca²⁺/nuclear factor of activated T cells (NFAT) axis^[16]. Glucose deprivation disrupts NFAT nuclear translocation and subsequent transcription of effector genes. Restoring glucose metabolic balance has been shown to enhance antitumor immune responses. For example, interferon-alpha (IFN-α) suppresses HIF-1α-driven glycolysis within the TME, thereby activating the mTOR/forkhead box M1 (FOXM1) axis in CD8⁺ T cells and upregulating CD27 expression, ultimately sensitizing HCC to ICIs^[17]. Similarly, inhibition of the deubiquitinase USP14, which stabilizes GLUT1 and promotes glucose uptake, represents a promising strategy to potentiate immunotherapy^[18]. Under metabolic stress or glucose scarcity, levels of the glycolytic intermediate fructose-1,6-bisphosphate (FBP) decline markedly, leading to activation of AMP-activated protein kinase (AMPK) signaling^[19,20]. AMPK activation induces dissociation of mechanistic target of rapamycin complex 1 (mTORC1) from the lysosome and subsequent inactivation, thereby impairing the anabolic metabolism required for cytotoxic CD8⁺ T cells and dendritic cells (DCs) to sustain immune effector functions and antigen presentation. Consequently, precise modulation of the AMPK/mTORC1 signaling pathway may represent an important strategy to restore immunometabolic balance and improve immunotherapeutic responsiveness in HCC. In addition, dysregulated glucose metabolism modulates immune checkpoint signaling and reinforces the immunosuppressive TME. Sorafenib-resistant HCC cells exhibit enhanced glycolysis through mTOR signaling, accompanied by upregulation of programmed cell death-ligand 1 (PD-L1) expression^[21]. Pharmacological inhibition of mTOR activation with rapamycin restores autophagic activity and downregulates PD-L1 expression, providing a rationale for dual targeting of metabolic pathways and immune checkpoints in drug-resistant HCC. Similarly, hyaluronan fragments derived from HCC cells induce tumor-associated macrophages (TAMs) to upregulate 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase 3 (PFKFB3), a rate-limiting glycolytic enzyme, thereby activating nuclear factor-kappa B (NF-κB) signaling and PD-L1 expression and generating a localized immune-privileged niche^[22]. PD-L1 signaling further promotes the immunosuppressive polarization of TAMs and amplifies immune evasion^[23]. Notably, immune checkpoint molecules can reciprocally regulate metabolic activity in immune cells. Upregulation of programmed cell death protein-1 (PD-1) and cytotoxic T lymphocyte antigen-4 (CTLA-4) cooperatively suppresses glycolysis in Teff cells, whereas PD-1 also enhances fatty acid oxidation (FAO) and lipolysis^[24,25]. This metabolic reprogramming limits Teff cell function and exacerbates the effects of metabolic competition. Comparable mechanisms have been observed in tumor-infiltrating natural killer (NK) cells in HCC, indicating that coordinated dysregulation of glucose metabolism and immune checkpoint signaling sustains an immunoevasive microenvironment across multiple immune cell types^[26].

In HCC, the Warburg effect drives excessive lactate production. Lactate is actively exported from tumor cells via monocarboxylate transporter 4 (MCT4), leading to markedly elevated extracellular lactate concentrations within the TME, which often reach 30-40 mM^[27,28]. Sustained lactate accumulation causes pronounced acidification of the TME and disrupts extracellular matrix architecture. These changes promote tumor cell migration and proliferation, induce angiogenesis, and enhance drug efflux and therapeutic tolerance. Importantly, lactate accumulation also interferes with antitumor immune responses through multiple mechanisms, thereby promoting HCC progression. Elevated lactate levels suppress the effector function of immune cells within the TME. High lactate concentrations markedly inhibit the proliferation, cytokine secretion, and cytotoxic activity of CD8⁺ T cells^[29,30]. Mechanistically, lactate uptake via MCT1 induces intracellular acidification and disrupts nicotinamide adenine dinucleotide (NAD⁺/NADH) redox homeostasis, thereby inhibiting key glycolytic enzymes, including GAPDH and PGDH. In addition, lactate-driven histone histone H3 lysine 18 (H3K18) lactylation promotes the expression of major vault protein (MVP), which competitively inhibits β-transducin repeat-containing protein (β-TRCP)-mediated ubiquitination and degradation of PD-L1, further weakening the antitumor activity of Teff cells^[31]. Accumulating evidence indicates that excessive lactate is a critical driver of T cell exhaustion in HCC^[32]. Notably, lower lactate concentrations may partially restore Teff cell function, suggesting potential avenues for metabolic modulation of antitumor immunity^[33]. Beyond T cells, lactate downregulates the activating receptor NKp46 on NK cells and reduces the production of IFN-γ, perforin, and granzyme B^[34]. DCs are also affected, as lactate impairs their maturation and activation, thereby limiting the expression of costimulatory molecules and proinflammatory cytokines^[35]. Accordingly, targeting lactate metabolism is emerging as an important strategy to alleviate effector immune cell suppression in HCC^[36]. Lactate accumulation also facilitates protumor immune functions within the TME. Regulatory T (Treg) cells exhibit pronounced metabolic plasticity and preferentially adopt a lactate-fueled metabolic program by downregulating glucose uptake, upregulating MCT1-mediated lactate transport, and enhancing lactate dehydrogenase A (LDHA) activity^[37]. These lactate-adapted Treg cells display enhanced immunosuppressive capacity. Moreover, lactate-induced lactylation of K72 on membrane-organizing extension spike protein (Moesin) activates downstream SMAD family member 3 (SMAD3) signaling, thereby promoting FOXP3 expression and stabilizing the immunosuppressive Treg cell phenotype^[38]. In TAMs, lactate promotes polarization toward the M2 phenotype through HIF-1α signaling, accompanied by upregulation of vascular endothelial growth factor (VEGF), arginase 1 (ARG1), and interleukin-10 (IL-10)^[39]. In addition, histone H3K18 lactylation enhances the expression of nuclear protein 1 (NUPR1), which sustains the M2 phenotype by suppressing ERK and JNK signaling pathways and ultimately exacerbates CD8⁺ T cell exhaustion^[40]. Interestingly, the gut-derived metabolite D-lactate has been implicated in driving TAM repolarization toward a proinflammatory M1 phenotype^[41]. However, whether D-lactate contributes to hepatic immune modulation via the portal circulation remains to be elucidated.

HCC shows etiology-associated heterogeneity in glucose metabolic reprogramming. In virus-related HCC, viral proteins directly enhance key glycolytic nodes, thereby promoting a shift toward a Warburg phenotype. Both hepatitis B virus (HBV)- and HCV-related HCC exhibit markedly increased HK2 activity. In HBV-related HCC, this increase is driven by hepatitis B virus X protein (HBx)-mediated activation of the NF-κB p65/HK2 axis. In HCV-related HCC, it arises from a direct interaction between HCV nonstructural protein 5A (NS5A) and HK2^[42,43]. In addition, HBV pre-S2 deletion mutants activate the mTOR signaling cascade, thereby upregulating GLUT1 expression^[44]. These mechanisms may help explain the convergence between HBV- and HCV-related HCC in the transcriptomic landscape and immune profiles^[45]. Notably, the higher prevalence of β-catenin mutations in HCV-related HCC predisposes tumor cells to a hepatocyte-like subtype characterized by reduced glycolysis and enhanced FAO^[46,47]. This metabolic shift is typically accompanied by downregulation of chemokine expression, impaired DC recruitment, and reduced CD8⁺ T cell infiltration, resulting in an immune-excluded TME^[48]. By contrast, HBV-related HCC more commonly shows preserved immune cell infiltration with functional suppression, whereas HCV-related HCC exhibits more pronounced metabolic and immunological stratification^[49]. In alcohol-related HCC, the association between β-catenin mutation and an immune-excluded phenotype appears more prominent, while the immunosuppressive effects mediated by dysregulated glucose metabolism are comparatively attenuated^[50]. In metabolic dysfunction-associated steatotic liver disease (MASLD)/metabolic dysfunction-associated steatohepatitis (MASH)-related HCC, metabolic reprogramming preferentially involves aberrant activation of lactate metabolism. Elevated expression of GLUT1 and MCT4 correlates with increased tumor invasiveness and adverse clinical outcomes, although the upstream regulatory mechanisms remain incompletely defined^[51]. Compared with virus-related HCC, MASLD/MASH-related HCC may be enriched for specific nonviral driver genetic events. For example, activin A receptor type 2A (ACVR2A) loss occurs more frequently in MASLD/MASH-related HCC and enhances lactate production and efflux through suppression of SMAD signaling, followed by upregulation of LDHA and MCT4^[52]. Furthermore, MASH-related HCC more readily establishes a specialized spatial interaction network among TAMs, MDSCs, and CD8⁺ T cells, suggesting that myeloid cell-mediated TME remodeling constitutes a principal mechanism of immune suppression^[53]. Given the limited benefit of ICIs in MASH-related HCC, therapeutic strategies aimed at dismantling myeloid cell-driven spatial immune sequestration may be critical for enhancing immunotherapeutic responsiveness^[54].

HCC cells undergo progressive metabolic reprogramming to establish a dynamically coordinated network of fatty acid synthesis (FAS), uptake, and oxidation^[56]. This network is tightly regulated to meet the demands imposed by a hypoxic and nutrient-deprived TME. Enhanced de novo lipogenesis, driven by upregulation of ATP citrate lyase (ACLY), acetyl-CoA carboxylase (ACC), and fatty acid synthase (FASN), is regulated by the PI3K/AKT/mTORC1 signaling pathway through activation of sterol regulatory element-binding protein 1 (SREBP1)^[57,58]. This regulatory circuitry enables tumor cells to preserve membrane integrity and sustain prosurvival signaling under metabolic stress. Concurrently, elevated expression of CD36 and fatty acid transport protein 4 (FATP4) enhances exogenous lipid uptake, thereby alleviating biosynthetic pressure while modulating membrane fluidity to promote tumor cell motility and invasiveness^[59]. Carnitine palmitoyltransferase 1A (CPT1A) and CPT2 regulate energy production and lipid load balance by controlling FAO, a process critical for HCC proliferation, redox homeostasis, and metabolic adaptation^[60,61]. As a consequence of increased lipid synthesis and catabolism in HCC cells and associated stromal compartments, together with chronic inflammation and hypoxia-driven lipid mobilization, the HCC TME becomes enriched in fatty acids^[62]. Within this lipid-rich TME, CD8⁺ T cells lack sufficient enzymatic capacity to efficiently catabolize excess fatty acids, leading to lipotoxic accumulation and subsequent functional exhaustion^[63]. Notably, epigenetic silencing of the E3 ubiquitin ligase Riplet in HCC aberrantly enhances FAS in tumor cells, resulting in excessive palmitate production that drives terminal CD8⁺ T cell exhaustion through signal transducer and activator of transcription 3 (STAT3) palmitoylation-dependent signaling^[64]. Prolonged energy deprivation further induces metabolic reprogramming in CD8⁺ T cells, characterized by increased CD36 expression and enhanced uptake of oxidized low-density lipoprotein (oxLDL). This process activates p38/CCAAT enhancer-binding protein β (CEBPβ) signaling, leading to upregulation of transferrin receptor 1 (TFR1), increased intracellular iron accumulation, and heightened oxidative stress. These changes culminate in lipid peroxidation and accelerated T cell dysfunction^[65]. In parallel, the PD-1/STAT3 signaling pathway promotes FAO by upregulating CPT1A while suppressing glycolysis, thereby further weakening the antitumor activity of CD8⁺ T cells^[66]. Collectively, STAT3 represents a critical signaling node linking fatty acid metabolism to immunosuppression and constitutes a potential therapeutic target in HCC^[67]. Compared with Teff cells, Treg cells exhibit greater metabolic plasticity and adaptability. In lipid-rich environments, tumor-infiltrating Treg cells markedly upregulate CD36, thereby enhancing fatty acid uptake and activating peroxisome proliferator-activated receptor β/δ (PPARβ/δ) signaling, which drives FAO and mitochondrial biogenesis^[68]. This metabolic reprogramming enables Treg cells to maintain energy homeostasis and immunosuppressive capacity within the lactate-rich and glucose-deprived microenvironment. Moreover, FOXP3 directly activates CD36 transcription, reinforcing this lipid-driven immunoregulatory phenotype and further impairing antitumor immunity^[69]. Accumulating evidence supports CD36 as a promising therapeutic target for enhancing immunotherapeutic responsiveness in HCC^[70]. Fatty acid metabolic remodeling in myeloid cells further contributes to immune evasion. In DCs, a lipid-rich environment promotes lipotoxic accumulation and markedly impairs antigen processing and cross-presentation^[71]. Upregulation of scavenger receptor-A (SR-A) facilitates excessive lipid uptake and further dampens DC-mediated activation of adaptive immunity^[72]. In TAMs, fatty acid-binding protein 1 (FABP1) is markedly upregulated during HCC progression and cooperates with PPARγ to activate CD36-dependent FAO, thereby metabolically supporting M2 polarization^[73]. In late-stage tumors characterized by hypoxia and necrosis, HCC-derived exosomes deliver FABP5 to TAMs, thereby activating PPARγ and suppressing PPARα-mediated FAO. This metabolic shift favors lipid storage and lipid droplet accumulation, stabilizing the immunosuppressive M2 phenotype^[74]. These findings indicate that FABP-mediated lipid metabolic plasticity dynamically regulates TAM polarization and immune evasion through spatiotemporally controlled mechanisms. In MDSCs, FAO flux is likewise markedly elevated, whereas genetic deletion of CD36 or pharmacological inhibition of STAT3 significantly attenuates their immunosuppressive activity^[75]. Of particular clinical relevance, sorafenib, a frontline therapeutic for HCC, exhibits a nonclassical resistance mechanism by activating PPARα signaling and enhancing FAO flux in MDSCs^[76]. This metabolic reprogramming not only drives the expression of immunosuppressive mediators but also impairs MDSC differentiation into macrophages, ultimately dampening antitumor immunity.

Aberrant cholesterol metabolism, particularly overactivation of de novo cholesterol biosynthesis, is a hallmark of HCC and contributes to enhanced lipid raft-associated signaling in tumor cell membranes^[77,78]. Elevated expression of 3-hydroxy-3-methylglutaryl-CoA reductase (HMGCR) promotes tumor proliferation by sustaining c-Myc and FOXM1 signaling, whereas persistent activation of SREBP2 facilitates epithelial-mesenchymal transition (EMT) and promotes immune evasion. Sustained activation of the cholesterol biosynthetic pathway concurrently suppresses cholesterol efflux, as reflected by impaired transcriptional activity of liver X receptors (LXRs) and marked downregulation of the downstream transporters ABCA1 and ABCG1^[79,80]. This suppression disrupts intracellular cholesterol distribution and metabolic feedback regulation, establishing a self-reinforcing positive feedback loop that further amplifies cholesterol biosynthesis. In addition, cholesterol esterification mediated by sterol O-acyltransferase 1 (SOAT1) and intracellular redistribution driven by GRAM domain-containing protein 1A (GRAMD1A) exacerbate cholesterol retention and recirculation, ultimately promoting the formation of a cholesterol-rich TME^[81,82]. Excessive cholesterol accumulation induces CD8⁺ T cell exhaustion. Mechanistically, cholesterol overload disrupts lipid metabolic homeostasis in CD8⁺ T cells, triggers endoplasmic reticulum (ER) stress, and activates the ER stress sensor XBP1^[83]. XBP1 subsequently drives the expression of immunosuppressive receptors, including PD-1, TIM-3, and LAG-3, thereby impairing antitumor immunity. Conversely, pharmacological inhibition of SOAT1 to block cholesterol esterification increases free cholesterol in the plasma membrane, alleviates ER stress, enhances TCR signaling, and promotes immunological synapse formation, thereby markedly augmenting CD8⁺ T cell cytotoxicity^[84]. Unlike fatty acids, cholesterol often acts as a proinflammatory mediator in TAMs. Despite elevated systemic cholesterol levels in the HCC TME, intracellular cholesterol content in TAMs is markedly reduced^[85,86]. This paradox largely reflects tumor-derived immunosuppressive signals that activate the LXRα/ABCA1 axis, thereby enhancing cholesterol efflux and disrupting intracellular cholesterol homeostasis. This metabolic reprogramming maintains TAMs in a low-cholesterol, semimature state characterized by impaired antigen-presenting capacity and a predominantly immunosuppressive phenotype. Notably, exogenous cholesterol can synergize with tumor-derived signals to further reinforce this immunosuppressive reprogramming. In MDSCs, although mechanistic studies remain limited, available evidence indicates that cholesterol homeostasis is a key determinant of immunosuppressive activity. Genetic ablation of cholesterol efflux-related genes attenuates immunosuppression^[87]. Moreover, activation of LXRβ by LXR agonists upregulates apolipoprotein E (ApoE) and cholesterol efflux pathways, leading to dysregulated lipid efflux and induction of ApoE/low-density lipoprotein receptor-related protein 8 (LRP8)-dependent apoptosis in MDSCs, thereby restoring antitumor immunity^[88,89]. Future studies should delineate the dynamic regulation of cholesterol homeostasis to enable precise modulation of the immunometabolic network in HCC. High-cholesterol conditions also promote pathological accumulation of oxysterols, which further reinforces the immunosuppressive milieu in HCC. Mechanistic studies indicate that accumulation of 25-hydroxycholesterol (25-HC) activates NF-κB and STAT3 signaling, increases FABP4 expression, and promotes matrix metalloproteinase (MMP) release, thereby enhancing tumor cell migration and invasiveness^[90]. In addition, 27-HC activates LXR signaling to modulate inflammatory cytokine profiles in TAMs and lymphocytes, while concurrently triggering the NF-κB/Twist1 axis to induce EMT and potentiate immune evasion^[91].

Distinct etiological contexts engage different upstream regulatory networks, thereby shaping discrete lipid metabolic subtypes. MASLD/MASH is frequently accompanied by pathological lipid accumulation, which impairs mitochondrial function and activates cell death signaling pathways^[92]. To adapt to lipotoxic stress, HCC cells downregulate CPT2 to suppress FAO, thereby attenuating Src/JNK-mediated apoptotic signaling and gaining a survival advantage^[93]. Concurrently, oxLDL uptake induces lipophagy, reprograms intracellular lipid flux, and activates a yes-associated protein (YAP)-dependent proproliferative transcriptional program^[94]. Upregulation of SOAT1 enhances cholesterol esterification and lipid droplet storage and, through SREBP2-mediated feedback regulation, remodels membrane cholesterol distribution, ultimately promoting EMT and metastasis^[95]. Collectively, in MASLD/MASH-related HCC, lipid metabolic reprogramming may establish a positive feedback loop that shifts from metabolic adaptation to metabolic dependency, thereby facilitating lipid accumulation-associated immune evasion. Lipid-rich microenvironments induce chronic inflammation and drive aberrant immune activation, which exacerbates hepatic fibrosis and tissue injury and ultimately promotes tumorigenesis^[96]. MASLD/MASH-related HCC exhibits heterogeneous immune regulatory patterns, giving rise to subpopulations with differential immunotherapeutic responses. For example, an endothelial cell-dominant subset characterized by high FABP4 expression promotes vascular normalization and enhances CD8⁺ T cell infiltration^[97]. In contrast, an intratumoral steatotic phenotype displays an immune-enriched yet functionally suppressed TME marked by T cell exhaustion and enhanced immune checkpoint signaling^[98]. These observations provide a basis for precision stratification. In contrast to MASLD/MASH-related HCC, lipid metabolic reprogramming in HBV-related HCC is directly driven by viral factors. HBx modulates SREBP-mediated lipogenic pathways while enhancing autophagic flux, thereby coordinately supporting viral persistence and tumor progression^[99,100]. Moreover, a high-risk subtype of early-stage HBV-related HCC characterized by elevated SOAT1 expression and dysregulated cholesterol metabolism has been associated with poor prognosis^[82]. In HCV-related HCC, lipid metabolic remodeling preferentially involves diacylglycerol O-acyltransferase 1 (DGAT1)-mediated lipid droplet biogenesis, enhanced viral protein localization, and suppressed triglyceride turnover, thereby establishing a lipid droplet-dependent metabolic niche that sustains viral replication and tumor progression^[101]. Although immune cells may initially exhibit heightened reactivity in the context of viral infection, lipid metabolic dysregulation frequently drives their transition toward an exhausted state^[102]. In alcohol-related HCC, ethanol oxidation leads to abnormal NADH accumulation, thereby promoting lipogenesis and suppressing FAO^[103]. Lipid peroxidation products further impair mitochondrial function and establish a self-amplifying cycle of oxidative stress and metabolic disruption. The immune landscape of alcohol-related HCC is predominantly characterized by chronic inflammation and a myeloid-skewed protumor immune milieu^[104].

Aberrant activation of glutamine metabolism is a hallmark of HCC^[106,107]. Key components of this pathway, including the glutamine transporter SLC1A5, kidney-type glutaminase GLS1, and glutamine synthetase (GS), are frequently upregulated in HCC tissues. By contrast, GLS2, the liver-type glutaminase that maintains metabolic homeostasis in normal hepatic tissue, is often silenced or markedly downregulated. Notably, in β-catenin-mutant HCC, GS overexpression paradoxically restrains tumor progression by limiting aberrant accumulation of free glutamate and suppressing hyperactivation of mTORC1 signaling^[108]. This observation underscores the importance of nitrogen flux homeostasis, governed by the balance between glutamine synthesis and catabolism, in shaping the metabolic adaptability and invasive potential of HCC cells. Beyond supporting anabolic demands, glutamine metabolism sustains HCC cell survival under persistent oxidative stress by fueling glutathione synthesis and replenishing the tricarboxylic acid (TCA) cycle^[109]. These processes maintain nicotinamide adenine dinucleotide phosphate (NADPH)-dependent redox homeostasis required for lipid and protein stability, thereby conferring resistance to ferroptosis. More importantly, glutamine metabolic reprogramming contributes directly to immune evasion. Nutrient competition suppresses effector immune cell function because CD8⁺ T cells depend heavily on glutamine metabolism for activation. Glutamine deprivation disrupts mitochondrial homeostasis, as reflected by reduced membrane potential, accumulation of reactive oxygen species (ROS), and ultrastructural damage^[110]. These alterations activate apoptotic programs, leading to downregulation of effector molecules, upregulation of inhibitory receptors, and eventual T cell exhaustion. Interestingly, systemic blockade of glutamine metabolism simultaneously impairs glycolysis and oxidative phosphorylation (OXPHOS) in tumor cells^[111]. In contrast, CD8⁺ T cells maintain energy and redox homeostasis by upregulating acetate utilization and oxidative metabolism and are reprogrammed into a memory-like state with long-term survival potential. This divergent metabolic response suggests a strategy for selectively alleviating tumor-associated immunosuppression. Similarly, NK cells are functionally compromised under glutamine-limiting conditions because of suppressed expression of the metabolic regulator c-Myc^[112]. Metabolic stress can also enhance the function of immunosuppressive cell populations. In TAMs, glutamine metabolism promotes an immunosuppressive M2 phenotype through generation of α-ketoglutarate (α-KG), which drives epigenetic remodeling. In addition, glutamate provision fuels the TCA cycle and sustains OXPHOS^[113,114]. Unlike effector immune cells, TAMs retain the capacity for de novo glutamine synthesis, conferring a metabolic advantage within the nutrient-deprived TME. Persistent GS overexpression in TAMs is therefore considered essential for maintaining the M2 phenotype. Moreover, glutamine synthesized by TAMs can be utilized by HCC cells as a carbon source to support proliferation and metabolic demands. Conversely, intermediates derived from tumor glutamine metabolism can reprogram macrophages to further reinforce immunosuppressive functions. Notably, tumor-derived glutamate induces chemokine CCL2 expression, thereby promoting recruitment of C-C motif chemokine receptor 2 (CCR2⁺) TAMs and exacerbating HCC progression^[115]. In MDSCs, severe glutamine scarcity acts as a metabolic stress signal that drives tumor cells to upregulate granulocyte colony-stimulating factor (G-CSF) and granulocyte macrophage colony-stimulating factor (GM-CSF), thereby systemically reconfiguring the TME to favor MDSC expansion and differentiation^[116]. However, MDSC dependence on glutamine is highly dose- and time-dependent^[117]. Systematic delineation of the underlying regulatory network in the multidimensional context of the TME remains necessary.

Under physiological conditions, hepatic tryptophan catabolism proceeds primarily through the kynurenine pathway, which is catalyzed by tryptophan-2,3-dioxygenase (TDO2) to maintain systemic immunometabolic homeostasis^[118]. In HCC, however, this pathway is aberrantly upregulated. Both TDO2 and indoleamine-2,3-dioxygenase 1 (IDO1) are markedly overexpressed in tumor tissues compared with normal liver. Their upregulation is driven, respectively, by oncogenic signaling pathways such as Wnt/β-catenin and NF-κB and by mediators associated with chronic inflammation. Increased expression of these enzymes correlates closely with higher tumor burden and poorer histological differentiation^[119,120]. Notably, some studies have interpreted IDO1 overexpression as a metabolically induced negative feedback mechanism secondary to immune activation, rather than solely as a protumorigenic factor^[121]. For example, in inflamed microenvironments enriched in CD8⁺ T cells, IFN-γ induces IDO1 expression in tumor cells through the JAK2/STAT1 signaling pathway, thereby restraining excessive cytotoxic responses. With tumor progression, the TME shifts toward an immune-excluded phenotype, during which IDO1 expression transitions to a sustained, tumor-intrinsic immunoevasive program that further reinforces the immunosuppressive network^[122]. Persistent tryptophan depletion leads to the accumulation of uncharged transfer RNAs (tRNAs), which activates general control nonderepressible 2 (GCN2) and initiates an amino acid starvation response^[123]. This response induces cell-cycle arrest in CD8⁺ T cells and NK cells, promotes differentiation of naïve CD4⁺ T cells into Treg cells, and facilitates tumor immune evasion. In parallel, kynurenine, an endogenous ligand of the aryl hydrocarbon receptor (AHR), sustains AHR activation and mediates broad immunosuppressive effects. In HCC cells, excessive AHR signaling promotes tumor cell proliferation and acquired drug resistance through STAT3 and NF-κB signaling^[124,125]. This process further increases IDO1 expression and establishes a self-reinforcing positive feedback loop. Within adaptive immunity, AHR activation upregulates PD-1 and reprograms transcriptional networks associated with T cell exhaustion, thereby impairing CD8⁺ T cell cytotoxicity^[126,127]. At the same time, AHR suppresses the transcription factors RORγt and T-bet, inhibiting T helper 1 (Th1) and Th17 differentiation while promoting FOXP3 and IL-10 expression, which collectively biases CD4⁺ T cells toward a Treg cell phenotype. Within innate immunity, AHR signaling downregulates key NK cell-activating receptors, including NKG2D and NKp46, and activates ROS-dependent mitochondrial apoptotic pathways, thereby promoting NK cell exhaustion^[128,129]. In DCs, AHR activation reduces the expression of costimulatory molecules and IL-12 while increasing IL-10, IDO1, and inhibitory receptors such as ILT3 and ILT5, thereby promoting a tolerogenic phenotype^[130]. Importantly, the immunological consequences of AHR activation are highly ligand dependent, with distinct endogenous and exogenous ligands eliciting divergent or even opposing effects^[131,132]. Accordingly, comprehensive elucidation of ligand-specific AHR engagement and potential allosteric regulatory mechanisms is essential for the rational design of selective AHR modulators aimed at precise therapeutic targeting of this pathway.

In HCC, arginine metabolism is extensively reprogrammed at the systemic level. Suppression of the urea cycle, marked by downregulation of carbamoyl phosphate synthetase 1 (CPS1) and argininosuccinate synthase 1 (ASS1), compromises endogenous arginine regeneration^[133,134]. As a compensatory response, upregulation of the arginine transporter SLC7A1 enhances extracellular arginine uptake, whereas increased expression of asparagine synthetase (ASNS) facilitates arginine influx through amino acid exchange mechanisms^[135]. Concurrently, reduced ARG1 expression limits arginine flux into the ornithine pathway, leading to substantial intracellular arginine accumulation^[136]. Excess arginine can directly bind to and modulate the transcriptional and splicing activity of RNA-binding motif protein 39 (RBM39), thereby reshaping metabolic circuits and establishing a proproliferative positive feedback loop that sustains tumor growth^[137]. Arginine depletion driven by metabolic dysregulation also has profound immunological consequences. It induces compensatory upregulation of SLC7A1 in T cells^[138]. However, because T cells depend strongly on extracellular arginine and have limited capacity for de novo arginine synthesis, they exhibit impaired survival, proliferation, and effector differentiation under arginine-limiting conditions. Moreover, activated hepatic stellate cells induce ARG1 expression in C-X3-C motif chemokine receptor 1 (CX3CR1⁺) TAMs through retinoic acid signaling^[139,140]. In parallel, CD14⁺ human leukocyte antigen-DR isotype (HLA-DR^-/low) MDSCs, which are enriched in the HCC microenvironment, display markedly elevated arginase activity. Together, these processes further exacerbate local arginine scarcity and drive T cell hyporesponsiveness or functional exhaustion. Notably, ARG1 also competes with inducible nitric oxide synthase (iNOS) for arginine as a substrate, thereby limiting NO production and impairing the cytotoxicity of M1-like macrophages^[141]. Metabolic reprogramming of arginine utilization in TAMs and MDSCs also increases polyamine synthesis, which contributes to the establishment of an immunosuppressive milieu. Polyamines enrich intracellular and extracellular compartments with organic cations that stabilize nucleic acids and ribosomes, thereby supporting the high metabolic throughput required for DNA replication, RNA transcription, and protein synthesis in proliferating HCC cells^[142]. More importantly, polyamines promote hypusination of eukaryotic translation initiation factor 5A (EIF5A) and selectively enhance translation of proteins involved in OXPHOS and mitophagy^[143]. This translational bias shifts immune cell metabolism toward an OXPHOS-dominant state, driving Teff cell exhaustion while promoting differentiation of Treg cells and M2-like macrophages. In addition, polyamines activate IDO1 signaling, reprogramming DCs toward a tolerogenic phenotype and reinforcing the immunosuppressive programs of MDSCs^[144,145]. Accordingly, targeting dysregulated arginine metabolism in HCC is emerging as a promising therapeutic strategy. Three principal approaches have been proposed, including arginine deprivation to exploit tumor auxotrophy, augmentation of arginine bioavailability to restore immune competence, and inhibition of polyamine biosynthesis or export to counteract immunosuppression^[146].

HBV- and HCV-related HCC exhibit marked divergence in glutamine metabolic reprogramming. HBV-related HCC predominantly depends on exogenous glutamine uptake. SLC1A5 has been identified as an independent risk factor, and its elevated expression correlates closely with enrichment of protumorigenic immune subsets and upregulation of immune checkpoint molecules^[147]. By contrast, HCV-related HCC preferentially relies on endogenous glutamine synthesis, reflected by high GS expression that is more frequently observed in well-differentiated tumors^[148]. GS overexpression is strongly associated with β-catenin mutations and, in the context of urea cycle impairment, constrains tumor growth signaling by maintaining nitrogen homeostasis^[149]. These findings suggest that distinct glutamine metabolic phenotypes in HCV-related HCC are determined by specific molecular backgrounds. Activation of IFN-I signaling represents a key molecular feature of virus-related HCC and may promote immune evasion through amino acid metabolic reprogramming. IFN-I selectively induces hepatocellular TDO2 expression, thereby integrating tryptophan catabolism into the antiviral response program^[150]. In parallel, IFN-I suppresses hepatic urea cycle flux, impairs arginine regeneration, and causes systemic hypoargininemia^[151]. Notably, HCV-related HCC shows significant enrichment of IFN-stimulated genes and is more likely than HBV-related HCC to exist in a state of chronic antiviral signaling activation^[152]. In MASLD/MASH-related HCC, glutamine metabolism is dynamically reprogrammed during disease progression. During hepatic fibrogenesis, stromal cell-mediated glutaminolysis via GLS1 is markedly enhanced, providing energy and biosynthetic precursors to support cellular proliferation and collagen synthesis^[153]. Meanwhile, emerging tumor nodules activate GS-dependent glutamine synthesis to maintain metabolic homeostasis and sustain growth potential^[154]. Cell type-specific glutamine metabolic reprogramming may facilitate metabolic coupling among stromal cells, myeloid cells, and tumor cells, thereby intensifying nutrient competition within the TME. In this context,α-KG-dependent epigenetic remodeling and OXPHOS-supported metabolic activity jointly promote M2 polarization of TAMs, conferring tolerogenic antigen-presenting functions within the TME^[155]. Furthermore, both MASLD/MASH-related and alcohol-related HCC display immunosuppressive phenotypes associated with arginine metabolic dysregulation^[156,157]. In MASLD/MASH-related HCC, urea cycle dysfunction secondary to metabolic disequilibrium constitutes the principal driver, whereas in alcohol-related HCC, marked expansion of ARG1⁺ granulocytic myeloid-derived suppressor cells (G-MDSCs) predominates. Although arginine metabolic reprogramming converges on similar terminal immunosuppressive phenotypes across etiologies, the upstream regulatory mechanisms remain heterogeneous. This mechanistic divergence provides a rationale for differential immunotherapeutic responses across etiologies and underscores the necessity of etiology-informed stratified targeting of arginine metabolic pathways.