From bench to bed: the tumor immune microenvironment and current immunotherapeutic strategies for hepatocellular carcinoma

MDSCs is a heterogeneous population of immature myeloid cells (IMCs), which are expanded in pathological conditions and up-regulate expression of immune suppressive factors, such as arginase and inducible nitric oxide synthase (iNOS or NOS2) [14]. Various tumor originated cytokines, such as G-CSF, GM-CSF, VEGF, MCP-1 and IL-1β, have been demonstrated to induce MDSCs infiltration [15]. Cell cycle related kinase (CCRK) represents a novel signaling target for cancer immunotherapy [16]. Emerging evidence also indicates the hepatoma-intrinsic CCRK upregulates interlukin-6 (IL-6) production through EZH2/NF-κB signaling, which consequently induce MDSCs accumulation in TME [17]. Hepatic carcinoma related tumor-associated fibroblasts (TAFs), a stromal part in HCC, can induce peripheral blood monocyte migration and differentiation into CD14 ⁺ HLA-DR ^−/low MDSCs by SDF-1α. TAFs mediate the generation of MDSCs via IL-6/STAT3 signaling [18]. In addition, local hypoxia has been identified as a key regulator that can promote MDSCs accumulation through the Chemokine C-C motif Ligand 26 (CCL26)/CX₃CR1 pathway [19]. Hypoxia-inducible factor 1α (HIF-1α) mediated ENTPD2 over-expression in HCC cells, has been proven to increase the extracellular level of 5′-AMP, which subsequently recruit MDSCs into TME [20].

As a powerful inhibitory immune modulator, infiltrated MDSCs in HCC damage effector T cells, expand immune checkpoink signaling, decrease NK cell cytotoxicity and cytokine production by diverse mechanisms [21, 22]. The MDSCs in fibrotic HCC tissue are notably correlated with reduced tumor infiltrating lymphocytes (TILs) and elevated tumorigenicity, aggressive phenotype, moreover, whose activation and infiltration contribute greatly to worse survival rate both in mouse model and HCC patients [23]. CD14⁺ HLA-DR ^−/low MDSCs can blunt HCC immunity through inducing activation of CD4⁺ CD25⁺ Foxp3⁺ regulatory T cells, inhibiting proliferation and IFN-γ secretion of CD3/CD28-stimulated autologous peripheral blood mononuclear cells (PBMCs) [24]. T cell function is impaired due to competition for energy resources (e.g. arginine and cysteine) with MDSCs [25], as well as involvement of several inhibitory receptor-ligand pathways in MDSCs-mediated immune evasion. Tumor-derived TGF-β triggers recruitment of MDSCs in a CXCL1/2/5- and CXCR2-dependent manner. The infiltrated MDSCs selectively suppress IFN-γ production deriving from NKT cells [26]. MDSCs also can express galectin-9 that binds to TIM-3 on T cells, inducing T-cell apoptosis [27]. In addition, it’s suggested that MDSCs in advanced HCC patients may interact with Kuppfer cells to induce PD-L1 expression, and mediate inhibited cytotoxicity and cytokine release of NK cells through the NKp30 receptor [28]. Taken together, MDSCs exert versatile immunosuppressive effects in HCC. Combined treatment with anti-PD-1/PD-L1 and concomitant targeting MDSCs (such as CCRK inhibition or p38 MAPK inhibitor) may synergistically enhance efficacy to eradicate HCC [17, 23]. In addition, recent evidence suggests radiation and IL-12 combination therapy (RT/IL-12) may elevate anti-tumor immunity in HCC by reducing MDSCs accumulation and the production of reactive oxygen species (ROS) [29]. Hence, MDSCs may serve as a potential target for resetting immunotorelant state in HCC tumors.

Macrophages arise from bone marrow-derived circulating monocytes, and then reside in normal tissues. The two polarizing phenotypes M1 and M2, are highly plastic in response to complex stimuli. Substantial clinical data and experimental research confirmed that alternatively activated status macrophages, the M2 phenotype, stimulate tumor initiation, progression and malignant metastasis by various mechanisms [30, 31]. In HCC, some specific populations of the immunosuppressive tumor-associated macrophages (TAMs) have emerged as a research hotspot recently. The well-identified HCC-derived cytokines, such as IL-4, IL-13, CSF-1, CCL2, CXCL12, connective tissue growth factor (CTGF) [32‐34], induce TAMs differentiation from CCR2⁺ inflammatory monocytes, alternatively activated macrophages (AAMs) infiltration, then reduce innate or adaptive immunity [34, 35]. Osteopontin (OPN) expressed by HCC cells not only has a positive association with PD-L1 expression in HCC, moreover, it facilitates alternative activation and chemotactic migration of TAMs via CSF1-CSF1R pathway in TAMs [36]. HCC cells originated HIF-1α has been recently suggested to enhance IL-1β release by TAMs via TLR4/TRIF/NF-κB signaling pathway, which fosters EMT and immune evasion of HCC [37]. Crosstalk between MDSCs and TAMs results in decreased production of IL-6, IL-12, down-expression of MHCII, and elevated production of IL-10, a strong inhibitory mediator that impairs downstream CD8+ T cell and NK cell cytotoxicity [38]. TAMs-derived IL-10 also increases intratumoral Foxp3⁺ Tregs frequency, which then suppresses CD4⁺CD25⁻ T cells activation [38, 39]. TAMs in the peritumoral stroma of HCC have been shown to secrete multiple key proinflammatory cytokines (e.g. IL-1β, IL-6, IL-23, and TNF-α) and contribute to the expansion of interleukin-17-producing CD4⁺ T helper 17 cells (Th17), which suppress anti-tumor immunity by over-expressing several activation markers, such as PD-1, CTLA-4, and GITR [40]. In addition, TAMs are highly involved in other immune inhibitory regulations [41, 42]. TGF-β in the TME of HCC promotes the Tim-3 expression on TAMs, subsequently enabling the activated TAMs to facilitate tumor growth and immune tolerance via NF-κB signaling and downstream IL-6 production [43]. TAMs derived IL-6/STAT3 signaling also has been validated to sustain HCC carcinogenesis by promoting its carcinoma stem cells (CSCs)-liked characteristics [44].

Notably, recent evidence indicates that PD-1⁻ TAMs can capture anti-PD-1 monoclonal antibodies (aPD-1 mAbs) through Fcγ receptors (FcγRs) expressed on the surface binding to drug’s Fc domain glycan [45]. This novel investigation indicates that blockade of FcγRs before aPD-1 mAbs administration may substantially improve checkpoint blockade therapy.

Heterogeneity of TANs is a fundamental property that allows TANs to perform corresponding functions for adaptations to changing microenvironment. Similar to macrophages, neutrophils differently affect tumor biological behaviors depending on their polarization, either anti-tumoral (N1) and pro-tumoral (N2) phenotypes [46]. In some solid tumor models, such as lung cancer, metastatic renal cell cancer (mRCC) and melanoma, it was previously reported that TANs infiltration or neutrophil-lymphocyte ratio (NLR) closely correlate with tumor progression, which can serve as a significant predictor for monitoring patients with advanced tumor receiving anti-PD-1/PD-L1 immunotherapy [47, 48]. TANs activation is modulated by cytokines, such as Type I interferons (IFNs) and TGF-β [49]. TANs mainly suppress anti-tumor immunity via interacting with CD8⁺ T cells, inducing CD8⁺ T cells apoptosis through nitric oxide (NO) production mediated by tumor necrosis factor-α (TNF-α) [50].

The facilitator role of TANs in pathological progression of HCC has become a topic of growing interest in recent years. Clinically, TANs play a key role in driving progression and poor prognosis of HCC, and NLR is an independent predictor of survival after hepatectomy in patients with HCC [51, 52]. The newest discovery shows that loss of hypoxia associated factor, HAF (encoded by SART1) results in inappropriate HIF-1 activation, and overproduction of downstream HIF-1 dependent chemokine, RANTES. HIF-1/RANTES upregulation accumulates TANs infiltration, which is associated with non-alcoholic steatohepatitis (NASH) driven HCC initiation and progression [53]. Moreover, recent studies suggested that TANs mediate the intratumoral infiltration of TAMs and regulatory T cells by overproducing some chemokines, such as CCL2 and CCL17, which then contributes to HCC progression, metastasis and resistance to sorafenib treatment [54]. A newly identified positive feedback loop implies that TANs induce HCC stem cell like characteristics via upregulating expression of miR-301b-3p in cancer cells, and maintain hyper-action in NF-kB signaling, lead to higher secretion level of C-X-C motif chemokine5 (CXCL5) and in turn recruit more TANs infiltration [55]. In general, TANs are strongly connected with immunosuppression in HCC, but direct interactions between TANs and other components in HCC tissue and the exact underlying mechanisms behind this regulation in HCC are not yet clear.

Regulatory T cells (Tregs) can be derived from peripheral blood T lymphocytes, resident T cells and other cellular sources. Its recruitment has been found to be induced by the CCR6 (CC chemokine receptor type 6)–CCL20 (CC motif chemokine 20) axis. Tregs activation is induced by T cell receptor (TCR) engagement concurrent with IL-10 and TGF-β signaling [59]. Apart from activation via pro-inflammatory signals, recent investigations elucidate that long noncoding RNAs (LncRNAs) may play pivotal roles in driving Tregs differentiation and implications during HCC progression [60]. Overexpressed Lnc-epidermal growth factor receptor (Lnc-EGFR) in Tregs binds to EGFR and prevents its ubiquitination by c-CBL, augmenting activation of its downstream AP-1/NF-AT1 axis in Tregs thus to promote immunosuppression in HCC [60]. Moreover, Amphiregulin (AREG), a multifunctional player, may enhance Tregs suppressive function via the EGFR motivation as well [61].

The frequencies of Tregs are associated with HCC invasiveness and have a crucial role in hampering the development of effective anti-tumor responses in HCC [57, 62]. Recent evidence indicates that CD4⁺CD25⁺ Tregs in HCC patients can trigger a compromised immune response through various mechanisms [63]. A typical subset, CD4⁺ CD25⁺ Foxp3⁺ Tregs, may impair CD8⁺ T cells killing capacity via inhibiting the release and production of granzyme A, B (GrA, B), and perforin [64], concurrently, they also selectively suppress certain molecules (such as TNF-α, IFN-γ) involved in CD8⁺ T cell activation [64, 65]. Additionally, high expression of IL-35 in HCC tissue has been implicated positively to correlate with another newly identified subtype, CD39⁺ Foxp3⁺ Tregs infiltration [66], which serves as a better independent predictive indicator for recurrence in HCC patients after curative resection.

The presence of CD8⁺ Cytotoxic T lymphocytes (CTLs) in HCC tissue is associated with improved survival. However, the efficacy of CTLs-mediated anti-tumor immune response is functionally limited through diverse mechanisms. Physical conditions (e.g. overload of lactic acid, low pH, hypoxia) [67], severe “metabolic competition” with tumor cells, a lack of CD4⁺ T cells help (moreover, interact with Tregs and other suppressor cells) [64, 68], and high expression of a large amount of immunoregulatory molecules in T cells or HCC cells (e.g. IL-10, Fas/FasL, CXCL17, VEGF, indoleamine-2,3-dioxygenase and so on) [67, 69‐71], may be responsible for restricted tumor-associated antigens (TAAs)-specific CD8⁺ T cell responses and poor IFN-γ production of CTLs [72, 73]. Apart from the classic immunosuppressive cells in TME, other components critically manipulate the functions of CTLs as well. Liver fibrosis, a prominent characteristic of HCC, impairs platelet-derived CD44 recognition by CD8⁺ T cells, reducing effector CD8⁺ T cells infiltration, and adhering to liver sinusoids to perform immunosurveillance [74]. Expression of Fas/FasL in CD8⁺ T cells positively correlates with HCC anti-tumor immunity [69]. Recent evidence indicates that tumor-derived vascular endothelial growth factor A (VEGF-A) and prostaglandin E2 (PGE2) cooperatively induce FasL expression in endothelial cells, which leads to excessive turnover of CD8 + T cells and reduce anti-tumor immune responses [71]. CD14⁺ dendritic cells (CD14⁺ DCs), a newly discovered immune regulator of HCC, has been suggested to suppress CTLs via IL-10 and indoleamine-2,3-dioxygenase (IDO) production, and the two cytokines play central roles in various physiological and pathological immune responses and inflammatory processes [75].

Notably, immune checkpoint signaling, which involves enhancement of numerous inhibitory co-stimulatory molecules (e.g. PD-1, LAG-3, CTLA-4, Tim-3, 2B4), has been demonstrated to dramatically induce CTLs exhaustion [58, 76, 77]. More details will be discussed in the section “immune checkpoint pathways and related therapeutics”.

Natural killer (NK) cells constitute a large proportion of the innate immune system in the liver. As the first line of host defense against viral infections (e.g. HBV, HCV) and carcinogenesis, NK cells play a key role in maintaining the balance between immune defense and tolerance. Increasing evidence suggests that hypoxic stress in HCC tissue, switch of activating/inhibitory NK receptors (NKRs) and influences by immune regulatory components in TME, largely contribute to NK cells dysfunction, which significantly correlates with fatigue anti-tumor immunity and poor prognosis [78, 79].

α-Fetoprotein (AFP) overexpressed by HCC cells was demonstrated to indirectly impair interlukine-12 (IL-12) production from dendritic cells (DCs), which results in attenuated cytotoxic effector molecules release, decreased expression of natural killer group 2, member D (NKG2D), an activating receptor on NK cells, and subsequently inhibiting activation and ability of NK cells [80, 81]. A recent study also indicates AFP may exert dual effects on NK cells functions in a direct manner. Short-term exposure to AFP induces IL-2 hyperresponsive phenotype NK cells, accompanied with elevated secretion of IL-1β, IL-6 and TNF-α [82]. These pro-inflammatory cytokines were associated with a low recurrence rate and a prolonged overall survival (OS) of HBV-related HCC patients [83]. In contrast, extended effect of AFP negatively affects long-term NK cell viability [82].

Other modulators in TME of HCC also exert multiple effects on NK activities (Fig. 2.). As mentioned above, MDSCs and TAMs infiltration inhibit autologous NK cell cytotoxicity and cytokine production, and the suppression is mainly dependent on NKp30 on NK cells [28]. Tregs compete with NK cells for IL-2 availability and impair NK responses via cytokines release, such as IL-8, TGF-ß1 and IL-10, which then down-regulates expression of NKR ligands on hepatic stellate cells (HSCs) and inhibits their recognition by NKG2D [84]. Hepatocellular carcinoma-associated fibroblasts (CAFs or TAFs), has been shown to induce MDSCs generation through the IL-6/STAT3 axis and stromal cell-derived factor (SDF)-1α secretion [18]. In addition to its direct influence on immunosuppressive TME, CAFs-derived IDO and PGE2 attenuate NK cells-mediated TNF-α and IFN-γ production, which may be associated with persistent fibrosis in HCC and tumor cell immune evasion [85, 86].

KCs have previously been demonstrated to constitute an important part in maintaining liver immune homeostasis. Some studies reported that IL-10-mediated suppression of KC-derived inflammatory TNF-α and NO production contribute to attenuation of hepatitis [87]. Although KCs were once regarded as a powerful line of defense against tumors in the liver, recently, KCs have commonly been explored as pro-carcinogenic stakeholders in the context of HCC, more underlying mechanisms about their immune regulator roles, and KCs-related innate or adaptive immune response have been gradually uncovered. Current investigations indicate that altered functions of KCs are mainly influenced by pro-inflammatory signals and other suppressive cells (e.g. MDSCs) [88]. Triggering receptors expressed on myeloid cells-1 (TREM-1) expressed by KCs, is a crucial factor in HCC initiation. New studies suggest that the potential ligand for TREM-1, high mobility group Box 1 (HMGB1) released by necrotic hepatocytes, is likely involved in activating KCs pro-inflammatory signaling and promoting HCC progression [89]. Autophagy-defective KCs, a novel non-parenchymal liver cellular degradation deficiency, has been shown to promote liver fibrosis, inflammation and hepatocarcinogenesis during the pre-neoplastic stage via enhancing the mitochondrial ROS/NF-κB/IL-1 pathway [90]. In addition, KCs-derived galectin-9, the natural ligand for the T cell immunoglobulin domain and mucin domain protein 3 (Tim-3), leads to expansion of CD4⁺ CD25⁺ FoxP3⁺ Tregs, contraction of CD4⁺ effector T cells, and apoptosis of CTLs in HCC [91]. Galectin-9 not only mediates T-cell senescence in HBV-associated HCC, significantly contributes to the inflammatory reactions and HCC immune escape [92], but notably also represents a potential biomarker of liver fibrosis and may emerge as a novel immunotherapeutic target for treating HCC and liver viral infections [92, 93].

Similar to NK cells, as another major player of innate immunity, DCs serve as professional antigen-presenting cells that are able to prime T-cells against tumor associated antigens (TAAs) involved in HCC progression. Recently, DCs have been an area of high interest as novel vaccines based on DCs have been developed and widely used in treating solid tumors including prostate cancer, melanoma, renal cancer and HCC [94]. DCs engineered with tumor-associated antigens (TAAs), which have been clarified by numerous in vitro and in vivo studies, are regarded as promising vaccines in HCC immunotherapy. In addition, autologous DCs pulsed ex vivo with the lysate of autologous tumor cells, HepG2 cells and telomerase peptides, have been evaluated in human clinical trials.

Recently, the Dendritic cell (DC)-derived exosomes (DEXs) and tumor cell–derived exosomes (TEXs), which elicit tumor regression in autochthonous HCC mouse models, form a new class of cell-free vaccines and extend options for HCC immunotherapeutic vaccines [95, 96] (Table 1.).

PD-1, a regulator immunoglobulin expressed on activated CD4⁺, CD8⁺ T cells, B cells and NK cells, plays an important role in maintaining immune tolerance and repressing cytotoxicity of T lymphocytes [123]. Co-inhibitory signals in lymphocytes are mediated by binding of PD-1 to its ligands PD-L1 (B7-H1) and PD-L2 (B7-DC) [124]. In HCC, it’s clear that an increase in the number of both circulating and intratumoral PD-1⁺ CD8⁺ T cells predict high postoperative recurrences and poorer prognosis. It is also known that up-regulation of PD-L1 on HCC cells, which is induced by various cytokines, particularly IFN-γ, in turn contributes to impairing anti-tumor immunity and promotes CD8⁺ T cells apoptosis [110]. New in vitro and in vivo discoveries indicate that PD-1 overexpressed myeloid cells, such as DCs, suppress T cell responses in HCC. CD8⁺ T cells can be more potently activated to secrete IL-2 and IFN-γ via adoptive transfer of PD-1-deficient DCs [111].

Clinically, a representative phase 1/2 dose escalation and expansion trial on PD-1 immune checkpoint inhibitors Nivolumab (CheckMate 040 study) showed a promising role for immunotherapy in the treatment of advanced HCC, and relevant results were presented at the 2017 ASCO annual meeting [125]. In dose-escalation phase (enrolled number = 48), the objective response rate (ORR) was 15%, the disease control rate (DCR) was 58% and the median time to progression was 3.4 months. In dose-expansion phase (total number = 214; in 4 cohorts), generally, the ORR was reported as 20%, the DCR was 64%, the median time to progression was 4.1 months, and the 6-month and 9-month progression-free survival rates were 83 and 74% respectively. A subsequent CheckMate-040 based analysis compared the ORR and survival between intent-to-treat (ITT) overall population and Asian cohort. It suggested that Nivolumab showed similar mOS and manageable safety profile both in ITT population and Asian patients [126].

The efficacy of another anti-PD-1 monoclonal antibody, Pembrolizumab, was assessed in a phase II, open-label trial (KEYNOTE-224). In this study, Pembrolizumab was proven to be effective and well-tolerated in Sorafenib-experienced patients with advanced HCC, and PD-L1 expression level may act as a useful predictive bio-marker in selecting interested HCC patients. A total of 104 enrolled patients in this study represented 8-month median duration of response (mDOR), with median time to response of 2 months [127].

In addition, another phase 3 randomised clinical trial of nivolumab mono-therapy compared with sorafenib in the first-line setting is ongoing (NCT02576509). Moreover, combination therapies of anti-PD-L1 antibody (Duvalumab) with anti-CTL4–4 antibody (Tremelimumab) for unresectable HCC are under study as well (NCT02519348).

Other reported combination immunotherapy studies are encouraging and really open new avenues for HCC treatment [128‐130], however, additional strategies are needed to uncover more sensitive predictive biomarkers besides PD-1/PD-L1 axis, optimize treatment selection, and improve HCC patients immune response. (More data of terminated or ongoing clinical trials are available in Table 3.).

Cytotoxic T lymphocyte protein 4 (CTLA-4), an inhibitory co-receptor that is expressed by activated T cells and is constitutively present on Tregs, has great affinity for competing with CD28 by binding to its ligands, CD80 and CD86, on antigen presenting cells (APCs). CTLA-4 plays a critical part in controlling CD4⁺ T cells function. In HCC and many other types of cancer, CTLA-4 suppresses the proliferation of T cells that have undergone TAA recognition and differentiation [131]. Additionally, inside HCC tissues, CTLA-4 further mediates immunosuppression by inducing Tregs activity and IDO and IL-10 productions in DCs [75].

Many clinical trials of antibodies targeting CTLA-4 are ongoing. A pilot clinical investigation testing anti-tumor and anti-viral effects of Tremelimumab in patients with HCC and HCV infection showed strong signs of anti-tumour efficacy (NCT01008358). The treatment presents a reliable safety profile, as no immune-related adverse events occurred. The median time to progression (TTP) was 6.48 months, and median overall survival (OS) was 8.2 months. The partial response rate was observed as 17.6%, and had a remarkable disease control rate of 76.4%. Moreover, 36% of the patients with AFP levels > 100 ng/ml showed more than 50% drop after Tremelimumab therapies [132]. In another phase 1 clinical study that tests the safety and effectiveness of Tremelimumab with radiofrequency (RFA) (NCT01853618), the median TTP and median OS were respectively 7.4 months (95% CI 4.7 to 19.4 months) and 12.3 months (95% CI 9.3 to 15.4 months). The 6-week HCC biopsies showed a clear increase in CD8⁺ T cells infiltration demonstrating that the combination of Tremelimumab with RFA in advanced HCC is feasible and results in stronger anti-tumor immunity [133].

Mucin domain-containing molecule-3 (Tim-3) is a transmembrane protein that is expressed on IFN-γ-secreting Th1 cells, NK cells and CTLs [113]. Tim-3 interacts with its soluble ligand galectin-9, and then negatively regulates T cell responses [91]. The expression of Tim-3 is increased in T cells infiltrating in chronic HBV infection [134], and the Tim-3/galectin-9 pathway consistently predicts poor prognosis in patients with HBV-associated HCC [76].

Lymphocyte activation gene 3 protein (LAG-3), a member of the immunoglobulin super-family proteins, which often binds MHC class II molecules with high affinity, represses the co-stimulatory functions of T cells [135]. Clinically, dual blockade of LAG-3 with anti-PD-1 therapy is being tested in a Phase I trial (NCT01968109) (Table 3.).

The immunosuppressive roles of both Tim-3 and LAG-3 in chronic viral hepatitis and HCC have been uncovered recently. However, their clinical values need to be further elucidated.