Revisiting the immunosuppressive microenvironment in hepatocellular carcinoma: a functional taxonomy of immune cells and therapeutic targeting

Chronic inflammation fuels hepatocarcinogenesis and reshapes the hepatic immune microenvironment

Chronic inflammation is the dominant and unifying trajectory in hepatocarcinogenesis, and a common pathological intermediate across diverse etiological backgrounds. Regardless of the initial insult, most HCC patients experience a prolonged disease course characterized by chronic hepatitis, progressive fibrosis or cirrhosis, and eventual malignant transformation⁶. Persistent inflammatory signaling establishes a dynamic immunological battleground that progressively shapes a permissive microenvironment for tumor outgrowth, while recurrent cycles of hepatocyte injury and regeneration impose sustained cellular stress to drive DNA damage and transcriptional reprogramming, thereby remodeling the immune landscape⁷. These events promote cell-intrinsic malignant transformation as well as orchestrating extensive intercellular communication through cytokines, chemokines, metabolites, and extracellular vesicles to drive large-scale remodeling of the hepatic immune landscape.

Importantly, hepatocytes serve as both primary sensors of inflammatory cues and central architects of the evolving immune microenvironment. In viral hepatitis, continuous antigenic stimulation induces persistent activation of interferon signaling and inflammatory gene programs within hepatocytes, which then shapes antiviral T cell responses. However, during chronic disease progression, sustained hepatitis B virus (HBV) exposure promotes progressive T cell hyperactivation followed by functional exhaustion, whereas hepatitis C virus (HCV) infection preferentially impairs viral antigen presentation on hepatocytes, facilitating immune evasion through distinct molecular routes (Figure 1)⁴. These divergent immune escape strategies contribute to the etiological heterogeneity of the HCC immune microenvironment, while also converging on a shared immunosuppressive endpoint.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Key mechanisms driving immune microenvironment remodeling in hepatocellular carcinoma (HCC). Chronic inflammation, genetic mutations, epigenetic reprogramming, and metabolic dysregulation together shape the immunosuppressive tumor microenvironment in HCC. (A) Chronic inflammation: Chronic inflammatory stimuli, including HBV and HCV infection, and metabolic injury, promote hepatocyte-derived inflammatory cytokines, exosomes and metabolites, leading to dysfunction of NK cells and T cells as well as facilitating the establishment of an immunosuppressive TIME. (B) Genetic mutations: Driver mutations, including those in TERT, CTNNB1, TP53, and FGF19, promote the accumulation of TAMs and MDSCs, reduce the abundance of DCs, impair the activity of cytotoxic T-cells, and suppress adaptive immunity. (C) Epigenetic reprogramming: Epigenetic regulation operates in both tumor and immune cells through several distinct pathways. In tumor cells, methylation-associated suppression of miR-144 limits M1 macrophage polarization, whereas acetylation-associated regulation of IGFB1 enhances TGF-β-mediated inhibition of T-cell activity. In T cells, SWI/SNF-dependent activation of TOX promotes exhaustion. Conversely, gut microbiota-derived acetate induces histone acetylation at the ACC1 locus in TAMs and favors M1 polarization. (D) Metabolic remodeling: Tumor-derived fatty acids are taken up by TAMs through FATP5, enhancing TAM survival and immunosuppressive activity, and consequently suppressing the function of T-cells. Gut microbiota-derived iso-LCA inhibits NK cell activity. Altered glucose utilization impairs T-cell activity while supporting tumor cell survival. AC, acetylation; ACC1, acetyl-CoA carboxylase 1; CTNNB1, catenin beta 1; DC, dendritic cell; FATP5, fatty acid transport protein-5; FGF19, fibroblast growth factor 19; HBV, hepatitis B virus; HCV, hepatitis C virus; iso-LCA, isolithocholic acid; Me, methylation; MDSC, myeloid-derived suppressor cell; NK cell, natural killer cell; SWI/SNF, SWItch/sucrose non-fermentable chromatin-remodeling complex; TAM, tumor-associated macrophage; TERT, telomerase reverse transcriptase; TGFB1, transforming growth factor beta 1; TGF-β, transforming growth factor beta; TIME, tumor immune microenvironment; TOX, thymocyte selection-associated high mobility group box; TP53, tumor protein p53. Created in BioRender (www.biorender.com).

In addition to hepatocytes, liver-resident stromal and immune populations actively participate in inflammation-driven immune reprogramming. Liver sinusoidal endothelial cells possess intrinsic immunoregulatory properties and engage in direct crosstalk with T cells to promote peripheral immune tolerance⁸. Kupffer cells, the resident macrophage population in the liver, undergo profound phenotypic and functional reprogramming during chronic inflammation, and exhibit marked heterogeneity relative to monocyte-derived macrophages. Through their dynamic interactions with CD4⁺ T cells and other myeloid populations, Kupffer cells can exert both immune-activating and immunoregulatory functions, further shaping the inflammatory milieu throughout hepatocarcinogenesis^9,10. During fibrosis driven by chronic inflammation, hepatic stellate cells (HSCs) and their activated fibroblast-like derivatives act as central architects of the immunosuppressive TME. In addition to matrix remodeling, activated HSCs undergo immunometabolic reprogramming, enabling them to recruit and condition immune cells, thereby establishing a niche that promotes immune evasion and malignant progression¹¹. The chronically inflamed hepatic milieu itself also exerts a profound influence on local immune phenotypes other than through direct cell–cell crosstalk. The persistent presence of inflammatory cytokines, bioactive metabolites, and viral products provides a sustained source of immunological pressure to fuel immune reprogramming and reinforce remodeling of the TIME¹².

Genetic mutations reshape the immune landscape

Genetic alteration is a defining hallmark of cancer, and HCC has a distinct mutational landscape characterized by unique mutation spectra and frequencies. Large-scale whole-exome sequencing studies have identified recurrent driver alterations in genes such as telomerase reverse transcriptase (TERT), tumor protein p53 (TP53), and catenin beta 1 (CTNNB1), together with less frequent but biologically relevant events involving oncogenic loci, including fibroblast growth factor 19 (FGF19)¹³. Increasing evidence indicates that these genetic alterations drive tumor-intrinsic malignant phenotypes, and also actively shape the immune architecture of the TME. In particular, HCCs carrying distinct driver mutations appear to segregate into fundamentally different immune subtypes, suggesting that oncogenic signaling and immune remodeling are tightly coupled during hepatocarcinogenesis⁵.

Among these HCCs, TP53- and CTNNB1-mutant tumors are the best-characterized molecular-immune subtypes to date. Compared with mutation-negative tumors, CTNNB1-mutant HCCs are consistently associated with a more strongly immunosuppressive microenvironment, characterized by increased infiltration of TAMs and CD8⁺ T cells with impaired cytotoxic function (Figure 1). This immunosuppressive phenotype has been linked to inferior clinical outcomes and poor responsiveness to immune checkpoint blockade^14,15. By contrast, although several studies have reported a relatively better response of TP53-mutant tumors to ICIs compared with CTNNB1-mutant cases, TP53 mutation is associated with extensive remodeling of the immune microenvironment. In addition to promoting tumor cell-intrinsic malignant proliferation, TP53 mutation has been implicated in the establishment of a myeloid-dominated immunosuppressive niche¹⁶. Moreover, multiple genetic alterations can cooperate to promote hepatocarcinogenesis and suppress dendritic cell (DC)-mediated immune activation¹⁷. TERT promoter mutations, the most frequent genetic alterations in HCC, have been increasingly associated with remodeling of the TIME. These mutations are highly enriched in cancer-prone dysplastic nodules, which exhibit a pronounced immune-desert phenotype characterized by markedly reduced infiltration of T cells, DCs, and macrophages¹⁸. This immunosuppressive landscape persists in very early HCCs with a high copy number alteration burden arising from these lesions, which also have a non-inflamed phenotype with reduced T cell and DC infiltration¹⁸. The immune microenvironments shaped by other less frequent driver mutations still need to be fully characterized, and this knowledge gap is probably attributable to the low incidence of these alterations in patient populations and the technical challenges associated with establishing appropriate experimental animal models. However, the widespread adoption of next-generation sequencing and rapid advances in molecular and genetic engineering technologies can greatly facilitate the systematic dissection of mutation-specific immune ecosystems in HCC.

Epigenetic reprogramming orchestrates immune dysfunction in the TIME

Epigenetic modifications are a crucial feature of tumor reprogramming. These alterations can occur early during precancerous stages as a preparation program, equipping tumor cells with profound genomic plasticity to facilitate malignant transformation and immune modulation¹⁹. Zhao et al. demonstrated that coordinated changes in microRNA and DNA methylation drive the conversion of immunologically “cold” tumors into “hot” ones, promoting M1 macrophage polarization and enhancing the cytotoxic function of tumor-infiltrating T cells (Figure 1)²⁰. Moreover, in HCC cells, inhibition of histone acetylation within the transforming growth factor beta (TGF-β) pathway restores the activity of CD8⁺ T cells and reduces Treg infiltration in the TME²¹. In addition, YTH N6-methyladenosine RNA binding protein 2 (YTHDF2), a classical reader of N6-methyladenosine, is upregulated in HCC via increased H3K4me3 and H3K27ac modifications, leading to elevated expression of PD-L1 and vascular endothelial growth factor A (VEGFA), which jointly promote immune evasion²².

Epigenetic reprogramming within immune cells also plays a pivotal role in reshaping their functional states. Overactivation of the SWItch/sucrose non-fermentable (SWI/SNF) complex in CD4⁺ T cells enhances its binding to thymocyte selection-associated high mobility group box (TOX), a key driver of T cell exhaustion²³. In macrophages, acetate produced by gut bacteria induces histone acetylation at the Acetyl-CoA Carboxylase 1 locus, reprogramming lipid metabolism to promote M1 polarization²⁴. These findings highlight the significance of epigenetic regulation for shaping the TIME and its potential as a therapeutic target.

Metabolic dysregulation establishes an immunosuppressive niche and reinforces immune evasion

Metabolic reprogramming has become a central theme in cancer research, particularly in HCC. The liver is the body’s primary metabolic organ and it coordinates a vast array of biochemical processes. Hepatocarcinogenesis disrupts this homeostasis by rewiring intrinsic hepatocyte metabolism as well as causing profound metabolic alterations in neighboring cells within the TME¹⁹. This reciprocal metabolic reprogramming is a key mechanism for inducing phenotypic switches, especially among immune cell populations, thereby actively shaping immunosuppressive landscapes.

Glucose metabolism is a pivotal driver of immune remodeling, and it influences the TME by modulating cellular transcriptional programs and directly altering the local abundance of metabolites (Figure 1). These changes collectively affect the infiltration and function of myeloid cells, and impair the cytotoxic efficacy of tumor-infiltrating T cells^25,26. Moreover, dysregulated lipid metabolism is a hallmark of metabolic dysfunction-associated steatotic liver disease (MASLD) and its progression to HCC, and it constitutes another critical axis of immune modulation. For instance, gut dysbiosis-induced perturbations in bile acid metabolism can suppress the antitumor activity of NK cells by disrupting cAMP response element-binding protein 1 (CREB1)-mediated transcriptional pathways²⁷. Furthermore, TAMs exhibit remarkable metabolic plasticity. In particular, the fatty acid transport protein-5-positive (FATP5⁺) TAM subpopulation leverages its enhanced phagocytic capacity to take up tumor-derived fatty acids, utilizing them as both an energy source and building blocks. This metabolic adaptation promotes their survival as well as reinforcing a pro-tumorigenic, immunosuppressive phenotype²⁸. In addition to glucose and lipids, it is increasingly recognized that other metabolic pathways, including ammonia handling, nucleotide biosynthesis, and glutamine metabolism, have roles in modulating the HCC immune milieu. Collectively, these findings highlight an extensive and intricate network of metabolic cross-talk between tumor and immune cells, suggesting novel avenues for therapeutic intervention.

Interconnected inflammatory, genetic, epigenetic, and metabolic drivers cooperatively reinforce immunosuppression

These seemingly distinct pathogenic mechanisms do not function independently but instead they converge within a highly interconnected regulatory network that collectively sustains the immunosuppressive HCC microenvironment. Among these interactions, crosstalk between metabolic reprogramming and epigenetic regulation is a critical regulatory axis. Metabolites such as acetyl-CoA and α-ketoglutarate serve as essential substrates or cofactors for epigenetic modifications, thereby linking cellular metabolic states to transcriptional regulation^24,29. Epigenetic alterations then reciprocally rewire metabolic programs in both tumor and immune cells, ultimately promoting immune dysfunction and disease progression³⁰. Similarly, reciprocal interactions between chronic inflammation and metabolic dysregulation constitute an important component of hepatocarcinogenesis. Persistent inflammatory stress reshapes hepatocellular metabolic programs, thereby promoting immune dysfunction and tumor progression³¹. Conversely, perturbations in key metabolic pathways induce endoplasmic reticulum stress and activate downstream inflammatory signaling pathways to establish a self-reinforcing inflammatory–metabolic circuit that promotes immunosuppressive microenvironment formation and disease progression³².

In addition to metabolic–epigenetic and inflammatory–metabolic crosstalk, chronic liver inflammation promotes widespread epigenetic remodeling to reshape transcriptional programs and facilitate inflammation-associated hepatocarcinogenesis¹⁹. Importantly, epigenetic alterations also cooperate with oncogenic mutations to drive malignant progression. By silencing tumor suppressor genes or activating oncogenic pathways, epigenetic regulators synergize with diverse genetic alterations in HCC, collectively reinforcing immune evasion and tumor progression³³.

Molecular subtyping: linking transcriptional programs to immune features

Molecular subtyping is one of the earliest classification frameworks developed for HCC, where clustering analysis of large-scale transcriptomic data is applied to categorize tumors based on overall transcriptional profiles. Early studies mainly showed that the proliferative capacity and activation of oncogenic pathways are important characteristics of HCC subtypes^37,38. Furthermore, a high degree of transcriptomic similarity was observed between specific subtypes of HCC and intrahepatic cholangiocarcinoma (ICC). In particular, the C1 and C2 subgroups defined within the respective classification systems for each tumor type exhibit convergent molecular and immune features, suggesting shared biological programs across these primary liver cancers. The C1 subgroup in both HCC and ICC is characterized by enrichment of mitotic checkpoint signaling pathways and chromosome instability, correlating with an aggressive phenotype and poor prognosis. By contrast, the C2 subgroup in both tumor types is characterized by elevated infiltration of CD4⁺ T cells and enhanced bile acid metabolism, which together contribute to more favorable clinical outcomes³⁶.

In 2023, Montironi et al. expanded this paradigm by integrating immune features and proposed three distinct transcriptomic classes: inflamed, intermediate, and excluded subtypes³⁹. They formally established correlations between global transcriptional patterns and immune infiltration levels in HCC. The inflamed class is characterized by a highly immunogenic microenvironment, with increases in tumor-infiltrating lymphocytes, a higher prevalence of tertiary lymphoid structures (TLS), and a reduced proportion of immunosuppressive cell subsets (Figure 2). These traits may be correlated with transcriptional signatures indicative of T-cell and B-cell activation, and enhanced antigen presentation. By contrast, the excluded class exhibits an immunologically cold microenvironment characterized by significantly lower infiltration of DCs, T cells, and NK cells. This subtype is enriched in patients with CTNNB1 mutations or aberrant Wnt–β-catenin pathway activation, aligning with the known association between this mutational background and immune exclusion. The intermediate class exhibits immune features between the two extremes, with a higher frequency of TP53 mutations.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Multi-dimensional classification of the hepatocellular carcinoma (HCC) immune microenvironment. HCC exhibits pronounced intertumoral heterogeneity in terms of the immune microenvironment, which can be categorized through complementary classification frameworks. (A) TIME of molecular subtypes: Molecular classification stratifies HCC into inflamed, intermediate, and excluded classes with distinct immune landscapes. The inflamed class is characterized by activated IFN signaling, enhanced lymphocyte chemotaxis, increased T-cell infiltration, M1-like TAMs and TLS formation. The intermediate class exhibits TP53 mutations, chromosomal instability, and intermediate immune infiltration. By contrast, the excluded class is enriched for CTNNB1 mutations and Wnt/β-catenin activation, accompanied by increased immunosuppressive cells and reduced cytotoxic lymphocyte infiltration. (B) TIME of immune subtypes: Immune-based classification identifies five representative TIME states. TIME-IA is enriched for effector lymphocytes and CXCL9⁺ TAMs. TIME-ISM is characterized by accumulation of immunosuppressive myeloid cells, including TANs, TAMs, and MDSCs. TIME-ISS is characterized by extensive stromal enrichment and infiltration of suppressive immune cells. TIME-IE represents an immune-desert phenotype with limited immune-cell infiltration. TIME-IR is characterized by enrichment of tissue-resident immune populations, including Kupffer cells and resident lymphocytes. (C) TIME for etiology subtypes: Distinct liver disease etiologies shape different immune landscapes. HBV-HCC is characterized by exhausted lymphocytes, dysfunctional NK cells and DCs, and increased Bregs and Tregs. HCV-HCC is characterized by monocyte and TAM enrichment with impaired DC activation. MASLD-HCC is associated with PMN-MDSCs, PD-1⁺ CD8⁺ T cells, Tregs, and NKT cells, whereas ALD-HCC is characterized by exhausted CD4⁺ T cells, M2 TAMs, dysfunctional Kupffer cells, and Th17-cell enrichment. ALD-HCC, alcohol-related liver disease HCC; Bregs, regulatory B cells; CTNNB1, catenin beta 1; CXCL9, C-X-C motif chemokine ligand 9; DC, dendritic cell; HBV, hepatitis B virus; HCV, hepatitis C virus; IFN, interferon; MASLD-HCC, metabolic dysfunction-associated steatotic liver disease-related HCC; NK, natural killer; NKT, natural killer T; PD-1, programmed cell death protein 1; PMN-MDSC, polymorphonuclear myeloid-derived suppressor cell; TAM, tumor-associated macrophage; TAN, tumor-associated neutrophil; TCR, T-cell receptor; Th17, T helper cell 17; TIME, tumor immune microenvironment; TIME-IA, tumor immune microenvironment immune activation; TIME-IE, tumor immune microenvironment immune exclusion; TIME-IR, tumor immune microenvironment immune residence; TIME-ISM, tumor immune microenvironment immune suppressive myeloid; TIME-ISS, tumor immune microenvironment immune suppressive stromal; TLS, tertiary lymphoid structures; TP53, tumor protein p53; Tregs, regulatory T cells. Created in BioRender (www.biorender.com).

Immune subtyping: defining TIME composition at single-cell resolution

The development of single-cell RNA sequencing (scRNA-seq) has enabled detailed profiling of the TIME in individual patients. This approach facilitates the direct integration of molecular features, clinical backgrounds, TIME composition, and patient outcomes. In 2022, Xue et al. established a large-scale single-cell atlas of HCC and proposed the TIMELASER classification system, which categorizes the HCC TIME into five distinct subtypes: immune activation (TIME-IA), immune suppressive myeloid (TIME-ISM), immune suppressive stromal (TIME-ISS), immune exclusion (TIME-IE), and immune residence (TIME-IR)⁵.

Importantly, TIME-ISM, TIME-ISS, and TIME-IE are all classified as immunologically “cold” phenotypes, but they appear to arise through distinct immunosuppressive architectures rather than representing biologically identical immune states. Collectively, these immunosuppressive TIME subtypes account for more than 50% of HCC cases, and TIME-IE alone represents approximately 37.8% of tumors⁵. Despite their shared non-inflamed phenotype, the immune microenvironmental compositions and dominant suppressive mechanisms differ substantially across these subtypes. TIME-ISM is characterized by the prominent accumulation of immunosuppressive myeloid populations, including tumor-associated neutrophils (TANs) and TAMs, indicating a predominantly myeloid-driven suppressive network. Stromal remodeling also appears to play a critical role in shaping this immunosuppressive TIME state by restricting effective immune cell trafficking, ultimately contributing to the formation of an immunosuppressive microenvironment or even an immune-desert phenotype⁴⁰. TIME-ISS is characterized by extensive stromal-cell enrichment accompanied by marked infiltration of immunosuppressive immune cells. By contrast, TIME-IE is defined by a profound immune-desert phenotype characterized by deficient infiltration of multiple immune subsets, including DCs, T cells, and NK cells, which indicates impaired immune cell recruitment and defective lymphocyte priming (Figure 2).

Chemokine signaling may form a central regulatory axis underlying the establishment of these distinct TIME states. For example, CXCL9/10/11–CXCR3 signaling is significantly enriched in TIME-IA and is associated with enhanced effector lymphocyte recruitment, whereas activation of the CXCL1/3/8–CXCR2 axis may partially account for the extensive infiltration of immunosuppressive myeloid cells observed in TIME-ISM. Furthermore, specific genomic alterations appear to be correlated with particular TIME subtypes. Studies have shown that CTNNB1 mutations are highly enriched in TIME-IE, whereas KRAS proto-oncogene, GTPase (KRAS) and isocitrate dehydrogenase 1 (IDH1) mutations are more frequently observed in TIME-ISS³⁹.

Etiological subtyping: uncovering disease-specific immune microenvironments

Research into the etiology-specific heterogeneity of the HCC immune microenvironment has progressed slowly, partly due to limitations in animal models and the historical lack of clinical emphasis on etiological distinctions. However, advances in foundational models and the growing clinical recognition of etiological treatment-response variations have propelled this field into a new phase of accelerated discovery⁴¹.

Spatial proteomic analyses have shown that MASLD-related HCC (MASLD-HCC) is characterized by a more immunosuppressive microenvironment compared with viral hepatitis-related HCC⁴², with increased infiltration of polymorphonuclear MDSCs (PMN-MDSCs) and a significant reduction of CD8⁺ T cells (Figure 2). Furthermore, PD-1⁺ CD8⁺ T cells are more abundant in MASLD-HCC and even emerge during pre-neoplastic stages, suggesting that ICIs, particularly anti-PD-1 therapies, may risk triggering inflammatory activation and potentially promote malignant transformation⁴³. Alcohol-related liver disease HCC shares the pathological feature of steatosis with MASLD-HCC, but their immunomodulatory mechanisms are fundamentally distinct. Alcohol acts as an independent immunotoxic agent to directly impair both hepatocytes and immune cells, and it can induce an exhausted phenotype in CD4⁺ T cells and promote the expansion of tolerogenic, immunosuppressive cell populations⁴⁴.

Among viral etiologies, HBV- and HCV-related HCCs also shape distinct immune landscapes. HBV is a DNA virus that integrates into the host genome and sustains replication, leading to continuous production of viral proteins. This persistent antigenic stimulation drives immune cells into a state of hyperactivation followed by exhaustion, marked by effector cell dysfunction and increased infiltration of regulatory cells such as regulatory B cells and Tregs^45,46. By contrast, HCV primarily disrupts antigen presentation in hepatocytes and modulates the immune system via cytokines, exosomes, and other mechanisms, resulting in an immune-excluded microenvironment⁴⁷.

In summary, the diverse subtyping frameworks outlined above establish critical connections between patient heterogeneity and specific immune microenvironmental features in HCC. However, instead of representing mutually exclusive systems, these complementary classification strategies collectively provide a multidimensional understanding of the HCC TIME from molecular, cellular, and etiological perspectives. Thus, by integrating distinct layers of information, these frameworks can provide valuable guidance to facilitate patient stratification, treatment-response prediction, and the development of more precise immunotherapeutic strategies.

Constructors of the immunosuppressive network: myeloid cells and Tregs

HCC is a classic “cold” tumor characterized by an abundance of immunosuppressive myeloid cells, including TAMs, monocytes, neutrophils, MDSCs, and Tregs (Figure 3)⁴. In particular, TAMs constitute the most prevalent and functionally heterogeneous population, playing pivotal roles throughout HCC pathogenesis. The infiltration of these immune subsets into the tumor niche is tightly regulated by chemotactic cues. Spatial transcriptomic analyses of HCC have shown that C-C motif chemokine ligand 15 (CCL15) promotes macrophage recruitment and drives their polarization toward an M2-like, pro-tumorigenic phenotype⁴⁸. Conversely, the C-C motif chemokine ligand 19 (CCL19)-C-C motif chemokine receptor 7 (CCR7) axis facilitates the recruitment of M1-polarized macrophages, which correlates with the reversal of immunosuppression, delayed disease progression, and enhanced immunotherapy responsiveness⁴⁹. MDSC recruitment is governed by distinct chemokine pathways from TAMs. A specialized immunosuppressive subtype of tumor-associated endothelial cells expresses C-X-C motif chemokine ligand 12 (CXCL12), which recruits MDSCs via interaction with their surface receptor C-X-C motif chemokine receptor 4 (CXCR4)⁵⁰. Furthermore, CXCR2, another key receptor on MDSCs, is activated by multiple ligands. In HCC cells, upregulation of ring finger protein 2 (RNF2) enhances MDSC recruitment by modulating TNF receptor-associated factor 2 (TRAF2) ubiquitination and subsequent CXCL1 expression⁵¹. Furthermore, the accumulation of TGF-β in the hepatic microenvironment stimulates tumor cells to secrete CXCL5 and CCL22, which then promote the infiltration of pro-tumorigenic neutrophils and Tregs, respectively^52,53.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Construction of the immunosuppressive network in hepatocellular carcinoma (HCC). The immunosuppressive network in HCC is constructed through the coordinated recruitment and functional modulation of four major cell populations, i.e., TAMs, TANs, MDSCs, and Tregs, driven by tumor-derived and microenvironmental signals. (A) TAMs. M1 TAMs are recruited into the HCC microenvironment via the CCL15–CCR1 and CCL19–CCR7 axes. Upon recruitment, these cells undergo M2 polarization driven by tumor-derived lactate, N1-Ac-Spd, and exosomes. M2 TAMs upregulate PD-L1 and further respond to CCL2 via CCR2, reinforcing their immunosuppressive phenotype. (B) TANs. HCC cells recruit neutrophils through the CXCL1/2/5–CXCR2 axis. Activated TANs subsequently secrete CCL4, which enhances the malignant phenotype of HCC cells and establishes a positive feedback loop between tumor cells and neutrophils. (C) MDSCs. MDSCs are recruited through multiple signaling pathways, including endothelial cell-derived CXCL12–CXCR4 signaling, tumor-derived CXCL1/2–CXCR2 signaling, and S100A9-mediated signaling through the RAGE/TLR4 pathway. In addition, sorafenib enhances the immunosuppressive activity of MDSCs by promoting fatty acid oxidation and acetyl-CoA production. (D) Tregs. Tregs are recruited by multiple tumor- and macrophage-derived signals, including the CCL22–CCR4, CCL20–CCR6, and CCL1–CCR8 axes, together with TGF-β-mediated immunoregulatory signaling. These recruited Tregs exhibit enhanced expression of immune checkpoint molecules such as PD-1, further strengthening local immunosuppression. CCL, C-C motif chemokine ligand; CCR, C-C motif chemokine receptor; CXCL, C-X-C motif chemokine ligand; CXCR, C-X-C motif chemokine receptor; FAO, fatty acid oxidation; MDSC, myeloid-derived suppressor cell; N1-Ac-Spd, N1-Acetylspermidine; PD-1, programmed cell death protein 1; PD-L1, programmed death-ligand 1; RAGE, receptor for advanced glycation end products; S100A9, S100 calcium-binding protein A9; TAM, tumor-associated macrophage; TAN, tumor-associated neutrophil; TGF-β, transforming growth factor beta; TLR4, Toll-like receptor 4; Tregs, regulatory T cells. Created in BioRender (www.biorender.com).

In addition to their abundances, the functional phenotypes and specific mechanisms associated with these immunosuppressive network builders are critical for understanding and overcoming immunotherapy resistance in HCC. The CXCL2-CXCR2 axis recruits TANs but also fundamentally reprograms their function. Tumor-derived CXCL2 induces TANs to secrete CCL4, which then stimulates tumor cells to upregulate CD155, thereby enhancing tumor immune evasion and stemness⁵⁴. Similarly, tumor-derived S100 calcium-binding protein A9 (S100A9) recruits and activates PMN-MDSCs. In PMN-MDSCs, S100A9 drives a pro-tumorigenic transcriptional program, establishing a self-reinforcing immunosuppressive loop by secreting S100A9. Metabolic reprogramming further promotes myeloid suppression⁵⁵. Sorafenib treatment unexpectedly enhances fatty acid oxidation in myeloid cells, promoting MDSC differentiation and augmenting their immunosuppressive capacity, and this mechanism potentially underlies tyrosine kinase inhibitor resistance⁵⁶. Among lymphoid suppressors, tumor-enriched CCR4+ Tregs exhibit a more potent immunosuppressive phenotype characterized by elevated IL-10 and PD-1 expression and marked stemness. Their enhanced suppressive function is linked to profound chromatin remodeling, which sustains their intra-tumoral survival and ability to inhibit CD8+ T cells⁵⁷. TAMs are also extensively reprogrammed by tumor-derived signals. In particular, tumor cells can transfer functional PD-L1 via exosomes to TAMs, conferring an enhanced capacity to suppress CD8+ T cells. Moreover, tumor-derived lactate triggers oxidative stress in TAMs, rewiring their acetate metabolism, leading to acetate release and fueling tumor cell metastasis⁵⁸.

Furthermore, these immunosuppressive populations do not function in isolation but rather engage in intricate crosstalk within the TIME, forming a cohesive and potent immunosuppressive network. CCL2 is a critical chemokine for monocyte–macrophage trafficking and a key orchestrator of this network. In HCC, tumor cells with ETS variant transcription factor 4 (ETV4) upregulation exhibit elevated PD-L1 expression and secrete high levels of CCL2. This dual action recruits both TAMs and MDSCs, which together establish an immunosuppressive niche that facilitates tumor metastasis⁵⁹. In addition, tumor cells leverage cytokines to reshape the function of macrophages and promote immunosuppression. Tumor-derived TNF-α reprograms lipid metabolism in macrophages and induces CCL20 secretion, which subsequently activates the CCL20/CCR6 axis to recruit Tregs, thereby amplifying the overall inhibitory nature of the microenvironment⁶⁰. Moreover, tumor-derived metabolites can co-opt this collaborative axis. The secretion of N1-acetylspermidine by tumor cells promotes macrophage polarization toward a pro-tumor phenotype. These polarized macrophages upregulate CCL1, which specifically recruits CCR8+ Tregs into the tumor⁶¹. This coordinated recruitment of both innate (macrophages) and adaptive (Tregs) suppressors leads to synergistic suppression of the function of CD8⁺ T cells. Moreover, IL-17A derived from neutrophil extracellular traps acts on macrophages via the IL-17 receptor, driving their polarization toward an M2-like phenotype and collectively accelerating HCC progression⁶².

In HCC, tumor-induced microenvironmental remodeling recruits and reprograms immune cells into an immunosuppressive state. These cells directly suppress antitumor immunity while forming a self-amplifying network through intercellular communication, thereby promoting tumor survival and immune escape. This integrated process encompassing recruitment, phenotypic reprogramming, and network reinforcement collectively maintains the immunosuppressive landscape of HCC.

Executors of antitumor immunity: innate and adaptive cytotoxic lymphocytes

The antitumor immune response critically depends on the effector functions of innate and cytotoxic lymphocytes, including classical subsets such as NK cells, natural killer T (NKT) cells, and CD8⁺ T cells, as well as newly identified populations like cytotoxic CD4⁺ T cells⁶³. However, these effectors represent primary targets for suppression within the TIME. Consequently, in HCC, their activity is frequently compromised by restricted tumor infiltration or immune exhaustion. This paradigm is robustly supported by large-scale single-cell transcriptomic analyses of HCC, which have consistently identified an expanded compartment of exhausted CD8⁺ and CD4⁺ T cells alongside Tregs, coupled with reduction in functionally competent T cell infiltration⁶⁴.

CD8⁺ T cells are the central cytotoxic component of adaptive immunity and the primary effector population that allows ICIs to exert their antitumor activity. However, in HCC, the functional potential of these cells is systematically undermined through a complex network of immunosuppressive mechanisms. Impaired chemokine signaling constitutes a major barrier to effective CD8⁺ T cell recruitment. In particular, downregulation of the Fas-associated death domain in non-inflamed “cold” HCC tumors leads to diminished CCL5 secretion, thereby curtailing the initial influx of CD8⁺ T cells⁶⁵. This deficit is exacerbated by an antagonistic interplay between interferon regulatory factors 1 and 2 within tumor cells, which attenuates the CXCL10-CXCR3 signaling axis⁶⁶. This pathway is critical because it mediates chemotactic recruitment and also sustains the intracellular metabolic programs and effector molecule synthesis required for robust cytotoxic function⁶⁷. In addition to impaired recruitment, CD8⁺ T cell infiltration are confronted by physical sequestration within the tumor stroma. Spatial transcriptomic profiling has demonstrated the formation of discrete tumor immune barriers, where immunosuppressive cell types collaborate to isolate tumor nests⁶⁸. A key example involves secreted phosphoprotein 1-positive (SPP1+) TAMs co-localizing with cancer-associated fibroblasts (CAFs) to create a restrictive boundary. Parallel mechanisms are observed in MASLD-HCC, where expanded populations of Tregs and CAFs establish a functionally similar barrier via tumor necrosis factor superfamily member 14 (TNFSF14)-tumor necrosis factor receptor superfamily member 14 (TNFRSF14) interactions, directly contributing to clinical resistance against immunotherapy⁶⁹. At a molecular level, CD8⁺ T cell dysfunction is further determined by tumor-derived metabolic perturbations. Aberrant activation of methionine metabolism in HCC cells driven by methionine adenosyltransferase 2A overexpression results in the accumulation of methyl-donor metabolites, including S-adenosylmethionine and methylthioadenosine⁷⁰. These compounds act as epigenetic modulators, altering chromatin accessibility in CD8⁺ T cells and silencing genes essential for their effector identity. Concurrently, fatty acids and eicosanoids within the lipid-rich TME hijack the transcriptional machinery of T cells^30,71. This metabolic interference promotes the upregulation of canonical exhaustion markers such as PD-1 and T-cell immunoglobulin and mucin domain-containing protein 3 (TIM-3) while also repressing the expression of cytotoxic mediators, effectively locking CD8⁺ T cells into a state of profound functional exhaustion.

Innate and innate-like lymphocytes, including NK cells, NKT cells, γδ T cells, and tissue-resident populations such as mucosal-associated invariant T cells, play vital roles in the frontline antitumor defense of HCC. However, single-cell transcriptomic studies of HCC have demonstrated that impaired hepatic infiltration and functional attenuation are often observed in these innate lymphocyte subsets within the TIME⁷². NK cells comprise the most abundant innate cytotoxic population and are subject to multifaceted suppression. Their recruitment is often limited due to disrupted interferon signaling and gut microbial homeostasis, thereby reducing the production of key chemokines, including CXCL10 and CXCL11^66,73. In addition to impaired trafficking, HCC cells directly suppress the functions of NK cells through specific receptor–ligand interactions. For example, the upregulation of transmembrane 4 L six family member 5 (TM4SF5) on tumor cells promotes lysosomal degradation of the activating receptor signaling lymphocytic activation molecule family member 7 (SLAMF7) on NK cells in a process that depends on N-glycosylation⁷⁴. In addition, enhanced engagement between tumor Siglec-9L and NK cell Siglec-9 activates intracellular inhibitory signaling motifs⁷⁵. Furthermore, HCC-driven dysregulation of bile acid metabolism alters the gut microbiota and elevates iso-lithocholic acid (iso-LCA), which disrupts transcriptional programs in NK cells and downregulates effector molecules²⁷. These mechanisms collectively impair NK cell cytotoxicity and cytokine secretion. NKT cells face analogous barriers. Liver endothelial stellate cells can recruit NKT cells via the CXCL16–CXCR6 axis after bile acid–mediated activation of leucine-rich repeat-containing G-protein-coupled receptor 5 (LGR5) or farnesoid X receptor (FXR), but the pathological accumulation of cholesterol in HCC promotes lipid peroxidation within NKT cells, thereby compromising their proliferation and cytotoxic function^76,77. Collectively, these findings indicate that the balance between bile acids and cholesterol plays a critical role in HCC progression. γδ T cells are selectively depleted in HCC, probably due to tumor-induced metabolic reprogramming, particularly in fatty acid metabolism⁷⁸. Intriguingly, hepatic stellate cells can activate γδ T cells in vitro, but their spatial distribution appears to be mutually exclusive with that of γδ T cells within tumors, suggesting contextual regulation of these two cell populations in vivo⁷⁹. Similarly, other tissue-resident lymphocytes, including mucosal-associated invariant T cells and conventional innate lymphoid cells, lose their prototypic antitumor phenotypes in HCC patients. This functional decline is closely linked to metabolic remodeling of the liver microenvironment and widespread dysregulation of transcriptional programs within these immune compartments^80,81.

In addition to the classical cytotoxic lymphocytes, emerging evidence indicates the significant potential of CD4⁺ cytotoxic T lymphocytes (CD4⁺ CTLs) as a non-canonical effector subset in HCC. Their clinical relevance was highlighted by early demonstrations that the frequency of CD4⁺ CTLs in both the circulation and liver tissue is an independent predictor of improved disease-free and overall survival in HCC patients⁸². Preliminary translational observations also suggest that factors may constrain their activity. For instance, a plasma cell profiling study found a significant inverse correlation between the abundance of immunosuppressive IL-10⁺ B cells and the levels of CD4⁺ CTLs⁸³. Moreover, within the HCC microenvironment, elevated B-cell lymphoma 6 (BCL6) expression has been implicated in upregulating co-inhibitory molecules on CD4⁺ CTLs, thereby directly suppressing their cytotoxic potential⁸⁴. At present, the evidence remains limited, but these findings collectively highlight CD4⁺ CTLs as a potentially important and underexplored component of the antitumor immune response in HCC, warranting further investigation to elucidate their precise roles and associated regulatory mechanisms.

In summary, cytotoxic effector cells are the principal executors of antitumor immunity, but their functions are profoundly constrained within the HCC microenvironment. Therefore, understanding the cell-specific mechanisms that underlie this suppression is essential for reversing immunotherapy resistance and ultimately improving patient outcomes.

Orchestrators of immune responses: helper cells and tertiary lymphoid structures

In addition to immunosuppressive networks and frontline cytotoxic effectors, the immune microenvironment of HCC also depends on key components that coordinate antitumor immunity. In this section, we highlight the immunoregulatory components with predominantly immunostimulatory roles, including DCs, helper CD4⁺ T cells, B cells, plasma cells, and TLS, and discuss how their functions in organizing and sustaining antitumor immune responses are reshaped or impaired in HCC.

DCs are recognized as the most potent specialized antigen-presenting cells, but they exhibit considerable heterogeneity and functional plasticity within the TME of HCC⁸⁵. Single-cell profiling has identified a distinct state in HCC characterized by the expansion of lysosome-associated membrane glycoprotein 3-positive (LAMP3⁺) DCs in patients⁷². These cells have an enhanced capacity to migrate to lymph nodes, theoretically implicating them as key initiators of adaptive anti-tumor immunity. However, emerging evidence suggests that this potential is frequently subverted in HCC, and the DC function is often driven toward immunosuppression. In scirrhous HCC, hypoxia-induced tumor cells impair the activity of DCs via the SPP1–CD44 axis, thereby hindering T cell activation⁸⁶. Hypoxia also promotes direct interactions between Tregs and conventional type 2 DCs (cDC2s), leading to the downregulation of human leukocyte antigen-DR (HLA-DR) on cDC2s and compromised antigen presentation⁸⁷. Furthermore, soluble fibrinogen-like protein 2, which is typically secreted by Tregs, can suppress the expression of immunostimulatory molecules on DCs and disrupt their transcriptional programs, ultimately inhibiting CD8⁺ T cell activation⁸⁸. Plasmacytoid DCs (pDCs) appear to be another aspect of DC-mediated immunoregulation. In the post-resection liver, pDCs release abundant IFN-α, which acts on HCC cells to induce C-X3-C motif chemokine ligand 1 (CX3CL1) secretion and subsequently recruit MDSCs, thereby forming an immunosuppressive milieu⁸⁹. In another immunosuppressive circuit, hypoxia-driven adenosine secretion from tumor cells recruits pDCs into the TME. These recruited pDCs can subsequently promote the expansion of Tregs, which further suppresses the function of CD8⁺ T cells⁹⁰. Collectively, these findings highlight the profound functional heterogeneity of DC subsets in HCC. Their canonical immunostimulatory roles are frequently hijacked by the TME, driving their differentiation toward states that favor immune tolerance.

CD4⁺ T cells primarily function as orchestrators of immune responses, modulating diverse immune cells through direct contact and cytokine secretion. Their functional states exhibit significant heterogeneity and they are closely linked to clinical outcomes. Magen et al. identified a cellular triad formed by CXCL13⁺ CD4⁺ T helper cells, PD-1⁺ CD8⁺ T cells, and DCs, which correlated with patient responsiveness to PD-1 blockade therapy⁹¹. Similarly, Zhao et al. reported that long-term HCC survivors have elevated frequencies of CD4⁺CXCR5⁺CD57⁺ T cells, which are associated with high serum levels of IL-21 and IFN-γ, and their presence is negatively regulated by factors such as high Treg infiltration and HBV DNA load⁹². The functionality of CD4⁺ T cells is susceptible to metabolic and environmental perturbation. In a MASLD-HCC mouse model, accumulated fatty acids, particularly linoleic acid, induced oxidative stress within CD4⁺ T cells, severely impairing their antitumor capacity⁹³. Moreover, the loss of CXCR6 diminishes the secretion of inflammatory cytokines by CD4⁺ T cells, thereby attenuating their ability to work in collaboration with NKT cells to prevent the malignant transformation of senescent hepatocytes⁹⁴. At the molecular level, key regulators govern the activation and fate of CD4⁺ T cells. The RNA-binding protein matrin 3 (MATR3) acts as an epigenetic checkpoint, where it is essential for initiating CD4⁺ T cell activation, but can also promote exhaustion by remodeling chromatin to upregulate TOX expression. These observations suggest that precise control of MATR3 expression is required to maintain the functional integrity of CD4⁺ T cells²³. Similarly, choline acetyltransferase (ChAT) plays a pivotal role in CD4⁺ T cell differentiation. ChAT⁺ T cells produce acetylcholine in an autocrine/paracrine manner to ensure appropriate activation, whereas ChAT deficiency drives CD4⁺ T cells toward exhaustion or Treg phenotype⁹⁵.

B cells and plasma cells also exhibit substantial functional heterogeneity within the HCC microenvironment, where they can exert both immunostimulatory and immunosuppressive functions depending on their differentiation states and spatial organization. In addition to regulatory B cells, specific plasma cell subsets have been implicated in distinct immunological outcomes. For instance, intratumoral IgG⁺ plasma cells have been associated with enhanced TAM infiltration and the establishment of a protumorigenic microenvironment, whereas circulating IgG1⁺ plasma cells have been correlated with favorable responses to immune checkpoint blockade therapy^96,97. B cells have antibody-producing and regulatory functions, but can also serve as specialized antigen-presenting cells that are capable of priming and sustaining T-cell activation, and they have close spatial interactions with intratumoral T cells, thereby contributing to effective antitumor immunity^98,99. Collectively, these findings highlight the profound functional heterogeneity of B-cell populations in HCC, and the importance of their coordinated interactions with other immune cells within the TME. These organized immune cell interactions are particularly exemplified by TLSs, which are ectopic lymphoid aggregates that have emerged as a promising research focus in antitumor immunity and immunotherapy in HCC. Studies have shown that the abundance and maturity of TLSs are negatively correlated with the recurrence risk, and their presence predicts improved prognosis in patients receiving PD-1 blockade¹⁰⁰. This predictive value may be due partly to the enriched populations of PD-1⁺ CD8⁺ T cells and NK cells within TLSs, which exhibit heightened responsiveness to ICIs¹⁰¹. Consistently, integrated spatial transcriptomic and radiomic analyses have also associated high TLS density with favorable responses to immunotherapy, accompanied by a remodeled transcriptional landscape in HCC^102,103. Recent mechanistic insights have helped to elucidate how TLSs are induced and function within the liver. Hepatic arterial infusion chemotherapy can promote TLS formation, and the CXCL12–CXCR4 axis has been identified as a key regulatory pathway¹⁰⁴. In particular, CXCR4-expressing stromal and immune cells, including matrix metallopeptidase 2-positive (MMP2⁺) fibroblasts, folate receptor beta-positive (FOLR2⁺) macrophages, and T central memory-like CD4⁺ T cells, are recruited by CXCL12 secreted from fibroblasts and endothelial cells, before coalescing into TLS-like aggregates and undergoing further activation. B cells are essential TLS components that also determine the structure’s functional impact. Spatial transcriptomics indicates that tumor-derived tryptophan metabolites can accumulate within TLSs to disrupt normal B cell differentiation, and thus impair the function of TLSs¹⁰⁵. Furthermore, tumor-secreted Dickkopf WNT signaling pathway inhibitor 1 (DKK1) can inhibit CCL19⁺ fibroblasts and plasma cells, ultimately suppressing TLS formation and the associated antitumor immune response¹⁰⁶.

In summary, the functional capacity of immune orchestrators, such as DCs, helper T cells, B cells, plasma cells and TLSs, reflects a dynamic struggle within the HCC microenvironment. These populations inherently retain the potential to coordinate protective antitumor immunity, but their efforts are systematically countered by tumor-driven mechanisms comprising functional suppression, phenotypic reprogramming, and physical exclusion. The resulting impairment of immune coordination further reinforces the immunosuppressive network, creating a feed-forward loop that dampens effector cell recruitment, activation, and cytotoxicity.

Therefore, overcoming the state of profound immune dysfunction will require a multi-layered therapeutic strategy. An effective approach must simultaneously disrupt key immunosuppressive pathways, restore the regulatory and stimulatory functions of immune orchestrators, and re-activate the cytotoxic potential of effector lymphocytes. Successfully rebalancing this triad by dismantling suppression, reinstating coordination, and unleashing execution is a central challenge and opportunity for next-generation immunotherapies in HCC.

Current immunotherapeutic strategies targeting the HCC TIME have transformed the treatment landscape

For patients with unresectable, advanced HCC, systemic therapy constitutes the primary treatment modality when locoregional approaches are no longer feasible or provide diminishing clinical benefit. Current systemic regimens are predominantly based on two drug classes: multi-kinase inhibitors (MKIs) and ICIs, which are both capable of reshaping the immunosuppressive TME (Table 1)².

The therapeutic landscape was first transformed by the 2008 SHARP trial, which established the MKI sorafenib as an effective systemic option¹⁰⁷. This success facilitated the development of subsequent MKIs, including lenvatinib, regorafenib, and cabozantinib, as well as anti-VEGF pathway antibodies, employed either as monotherapies or in combination strategies². A second paradigm shift occurred in 2017, when the CheckMate-040 study led to FDA approval of the anti-PD-1 antibody nivolumab for second-line treatment, marking the formal introduction of ICIs into HCC management¹¹⁶. This advance was further solidified by the IMbrave150 trial, where the combination of atezolizumab (anti-PD-L1) and bevacizumab (anti-VEGF) obtained superior clinical outcomes compared with sorafenib, establishing immune-based combination therapy as a first-line standard^41,118. More recently, positive outcomes in two pivotal phase III trials, HIMALAYA and CheckMate-9DW, have led to the formal establishment of dual immune checkpoint blockade as a new first-line treatment standard for advanced HCC^112,119. This strategy combines anti-PD-1/PD-L1 antibodies with an anti-CTLA-4 antibody.

Together, these clinical developments illustrate how TME-targeting strategies have evolved from theoretical concepts to established therapeutic standards in HCC. The current challenge entails refining these approaches to more effectively overcome primary and adaptive resistance mechanisms, thereby increasing the proportion of patients who obtain durable benefit from immunotherapy.

Emerging and potential therapeutic avenues for TIME reprogramming

In addition to the established combination regimens and analogous agents currently in clinical use, many novel therapeutic strategies for HCC are under investigation at various preclinical and clinical stages. In this section, we focus on those that directly target the TME. These emerging approaches, which include both novel combination regimens and innovative immunotherapeutic modalities, aim to overcome the limitations of current therapies by reprogramming the immunosuppressive milieu or engaging complementary antitumor mechanisms.

Targeting immune cell recruitment to remodel the immunosuppressive TIME

The main pathway that governs immune cell recruitment is the chemokine–chemokine receptor axis, which makes it a rational target for therapeutic intervention. In particular, the CXCL12–CXCR4 axis is pivotal for establishing an immunosuppressive microenvironment by recruiting various inhibitory cell types, thereby contributing to an immunologically “cold” tumor phenotype (Figure 4). Preclinical models demonstrate that CXCR4 blockade increases the infiltration of activated T cells as well as reduces the presence and suppressive output of TAMs and MDSCs, ultimately sensitizing tumors to ICIs^50,120. These findings are supported by a phase II trial in metastatic pancreatic cancer, highlighting its potential translational relevance for HCC¹²¹. Similarly, CXCR2 is highly expressed on neutrophils and PMN-MDSCs, and it represents another promising target. Multiple preclinical studies in mouse models confirm that CXCR2 inhibition decreases the intratumoral accumulation of suppressive myeloid populations and enhances the efficacy of anti-PD-1 therapy^54,122,123. By contrast, targeting the widely expressed receptors CCR2 and CCR5 is established in HIV therapy, but remains contentious in oncology. Several studies have reported that CCR2/CCR5 dual inhibitors combined with anti-PD-1/PD-L1 can suppress tumor progression and reduce the infiltration of TAMs, MDSCs, and Tregs, but clinical translation has yielded mixed results^59,124,125. A phase II trial in pancreatic cancer showed promise but failures in colorectal cancer and MASLD models highlight the context dependence of this approach and the need for careful patient stratification¹²⁶. Other chemokine receptors, including CCR1, CCR4, and CCR8, are also under preclinical investigation. Their inhibition has been shown to modulate the recruitment of monocytes or Tregs, thereby reshaping the immunosuppressive TME and potentiating ICIs in preclinical settings^53,61,127. In particular, early-phase clinical trials have begun to explore the potential of applying CCR4 inhibition in combination with PD-1 blockade as a novel strategy for solid tumors¹²⁸.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Emerging therapeutic strategies targeting the hepatocellular carcinoma (HCC) immune microenvironment. Multiple investigational approaches have demonstrated the potential to overcome immunotherapy resistance and achieve durable antitumor responses in preclinical models or early-phase clinical trials for HCC. These strategies include conventional and innovative approaches targeting immune cell recruitment, reprogramming immune cell function, and novel immunotherapeutic modalities. (A) Conventional strategies targeting immune cell recruitment: Conventional approaches inhibit chemokine signaling pathways to reduce the recruitment of immunosuppressive cell populations, including TAMs, TANs, MDSCs, and Tregs, through representative targets such as CXCR4, CXCR2, CCR2/5, and CCR4. (B) Innovative strategies targeting immune cell recruitment: Innovative chemokine-based delivery systems, including CXCL9 fusion proteins, CXCL9-loaded nanoparticles, and chemokine-containing microparticles, are designed to enhance the recruitment of cytotoxic T cells, remodel CAFs, and promote TLS formation. (C) Reprogramming immune cell function: Representative metabolic and signaling interventions, including spironolactone, EVT0185, galunisertib, MTX3937, and R162, reprogram the functional states of immune cells by targeting distinct immunosuppressive pathways within the HCC TIME. (D) Novel immunotherapeutic modalities: Emerging immunotherapeutic approaches include chimeric antigen receptor therapy, T-cell receptor-engineered T cells, oncolytic virotherapy, and tumor vaccines, which target multiple immune-cell populations within the HCC TIME. ACLY, ATP-citrate lyase; CAF, cancer-associated fibroblast; CCR, C-C motif chemokine receptor; CXCR2, C-X-C motif chemokine receptor 2; CXCR4, C-X-C motif chemokine receptor 4; CXCL9, C-X-C motif chemokine ligand 9; DC, dendritic cell; GDH1/GLUD1, glutamate dehydrogenase 1; iso-LCA, iso-lithocholic acid; MDSC, myeloid-derived suppressor cell; NK, natural killer; TAM, tumor-associated macrophage; TAN, tumor-associated neutrophil; TCR, T-cell receptor; TGFβR1, transforming growth factor beta receptor 1; TIME, tumor immune microenvironment; TLS, tertiary lymphoid structure; Tregs, regulatory T cells; Siglec-9, sialic acid-binding immunoglobulin-like lectin 9. Created in BioRender (www.biorender.com).

Enhancing the recruitment of immunostimulatory and effector cells through engineered chemokine delivery is another promising therapeutic avenue, although technically challenging. The development of chemokine analogs or agonists has been hindered by issues such as poor in vivo stability, promiscuous receptor interactions, and complex signaling dynamics, leading to relatively slow progress in this field. Nevertheless, several innovative preclinical studies provide proof-of-concept for this approach. Lugassy et al. engineered a stabilized CXCL9 fusion protein by adding an N-terminal glutamine tag, which protected it from proteolytic degradation by dipeptidyl peptidase-4 in vivo¹²⁹. This modification preserved its chemotactic function, and the construct exhibited potent T-cell recruitment and antitumor efficacy across multiple murine tumor models, highlighting its broad therapeutic potential. In a more sophisticated delivery strategy, Chen et al. developed an acid-responsive nanoparticle platform designated as Cal@nChap-CXCL9¹³⁰. This system co-encapsulated calcitriol (Cal) and CXCL9, and released its payload specifically within the acidic TME. The sequential release of Cal (to reprogram CAFs) and CXCL9 (to recruit T cells) synergized with PD-1 blockade, significantly improving therapeutic outcomes in preclinical models. In addition to single chemokine approaches, platforms designed to induce TLS formation are also emerging. One example utilizes hydrogel-based microparticles co-loaded with multiple lymphoid chemokines (CCL19, CCL21, and CXCL13) and lymphotoxin-α1β2¹³¹. This combination successfully induces TLS-like structure formation in tumors, demonstrating the potential of multi-factor delivery systems as a future direction for chemokine-based immunotherapy.