Skip to main content
Download PDF

Integrated analysis reveals an immune evasion prognostic signature for predicting the overall survival in patients with hepatocellular carcinoma

Cancer Cell International volume 25, Article number: 101 (2025) Cite this article

Abstract

Background

The development of immunotherapy has enriched the treatment of hepatocellular carcinoma (HCC), but the efficacy is not as expected, which may be due to immune evasion. Immune evasion is related to the immune microenvironment of HCC, but there is little research on it.

Methods

We employed unsupervised clustering analysis to categorize patients from TCGA based on 182 immune evasion-related genes (IEGs). We utilized single-sample gene set enrichment analysis (ssGSEA) and CIBERSORT to calculate differences in immune cell infiltration between clusters. The differences in immune cells and immune-related pathways were assessed using GSEA. We constructed an immune escape prognosis signature (IEPS) using univariate Cox and LASSO Cox algorithms and evaluated the predictive performance of IEPS with receiver operating characteristic (ROC) curves and survival curves. Additionally, we established a nomogram for clinical application based on IEPS. IHC validated the expression of Carbamoyl phosphate synthetase 2, Aspartate transcarbamylase, and Dihydroorotase (CAD) and Phosphatidylinositol Glycan Anchor Biosynthesis Class U (PIGU) in HCC. We transfected liver cancer cell lines with siRNA and overexpression plasmids, and confirmed the relationship between CAD, PIGU, and the potential downstream TGF-β1 in HCC using qRT-PCR and Western blot. Finally, we validated the tumor response of CAD overexpression using an animal model.

Results

Unsupervised clustering analysis based on IEGs divided HCC patients from TCGA into two groups. There were significant differences in prognosis and immune characteristics between the two groups of patients. Scoring of TCGA patients using IEPS revealed that higher scores were associated with poorer overall survival (OS). Validation was performed using the ICGC database. TIME analysis indicated that patients in the high-IEPS group were in an immunosuppressive state, possibly due to a significant increase in Treg infiltration. Compared to normal liver cells, HCC cells expressed higher levels of CAD and PIGU. Cellular experimental results showed a positive correlation between CAD, PIGU and the potential downstream TGF-β1 expression. Animal experiments demonstrated that CAD significantly promoted tumor progression, with an increase in Treg infiltration.

Conclusion

IEPS has strong prognostic value for HCC patients, and CAD and PIGU provide perspectives on new biomarkers and therapeutic targets for HCC.

Introduction

Primary liver cancer is the third most common cause of cancer-related deaths worldwide and the sixth most frequently diagnosed cancer globally, with hepatocellular carcinoma (HCC) being the most common pathological type, accounting for 90% of cases [1, 2]. Current HCC treatments include surgery, liver transplant, transcatheter arterial chemoembolization (TACE), radiofrequency ablation, and systemic therapy. For early-stage HCC, surgical resection remains the most prevalent treatment modality. In the early stages of HCC, patients often have no any overt clinical symptoms, leading to a diagnosis at an advanced stage in 80% of HCC cases, which results in the loss of optimal surgical intervention opportunities. Date showed that patients with advanced HCC have a survival time of only 10 to 24 months following diagnosis [2, 3]. But in recent years, development of immunotherapy have brought new hope for advanced HCC. Nowadays a variety of immunotherapeutic drugs have been approved for first-line treatment of liver cancer, including nivolumab and tislelizumab, and multiple therapies that combining targeted drugs and immunotherapeutic drugs have been approved, such as atezolizumab in combination with bevacizumab, and camrelizumab in combination with apatinib [4,5,6]. In addition, many immunotherapeutic clinical trials are being carried out [7, 8]. Despite the increasing number of available drugs, many patients remain unresponsive to immunotherapy during the treatment process, and only a small fraction of patients can benefit from it.

The limited efficacy of immunotherapy of HCC is primarily due to the immune microenvironment of the liver [9]. Liver has an immunosuppressive property, which can weaken antigen responses mediated by T cells. For maintaining normal physiological homeostasis, liver’s immunosuppression is maintained by other resident immune cells, including Kupffer cells (KCs), dendritic cells (DCs), regulatory T cells (Tregs), and liver sinusoidal endothelial cells (LSECs) [10]. However, in advanced HCC, this setting is conducive to immune evasion in HCC, enabling cancer cells to dodge the host’s immune clearance [11]. Cancer-intrinsic immune evasion denotes the strategies that tumor cells use to elude host immune detection and destruction, facilitating their survival, proliferation, and resistance to immunotherapy [12, 13]. In advanced HCC patients, monocytes from HCC tissue highly express CD48, which acts on NK cells, causing them to become exhausted and ultimately die after a rapid and short activation [14]. Tumor-associated macrophages (TAMs) are key regulatory cells in the immune response of HCC, promoting immune evasion of HCC by secreting the inhibitory programmed death ligand 1 (PD-L1) to suppress T-cell activation and function [15]. Therefore, exploring the mechanisms of immune escape in the HCC immunological microenvironment holds great significance for the advancement of immunotherapy.

In our study, we leveraged the Cancer Genome Atlas (TCGA) database to precisely identify 18 immune evasion genes (IEGs) within HCC and conducted an in-depth analysis to elucidate their association with immunological characteristics of HCC. Furthermore, we used TCGA and the International Cancer Genome Consortium (ICGC) databases to construct and verify the Immune Evasion Prognostic Signature (IEPS) based on the IEGs, which is significantly associated with HCC prognosis and immune characteristics, and discovered that the genes included in IEPS [Carbamoyl phosphate synthetase 2, Aspartate transcarbamylase, and Dihydroorotase (CAD) and Phosphatidylinositol Glycan Anchor Biosynthesis Class U (PIGU)] promote the expression of TGF-β1, facilitating the recruitment of Treg in HCC.

Materials and methods

Data collection

From the TCGA database, 370 HCC patients were selected for clustering analysis and served as the training set for subsequent follow-up analysis. The ICGA cohort, which includes 231 HCC patients, was used as the validation set. The IEGs derived from previous study were listed in Supplementary Table 1 [16].

Unsupervised consensus cluster analysis

Use the “ConsensusClusterPlus” R package to perform consensus cluster analysis based on 18 IEGs. Compute the similarity distances employing the Euclidean distance metric, and aggregate the data through the k-means clustering algorithm, ensuring a constant inclusion rate of 80% for patient data across 100 iterations.

Construction and validation of IEPS

The IEGs were screened to a risk score signature using univariate Cox regression, Lasso regression analysis and multivariable Cox regression analysis, which were used to construct a IEPS for the HCC patients involving 2 IEGs (PIGU and CAD). The risk score of each sample was calculated using the following formula:\(\:IEPS\:score={\sum\:}_{i=1}^{n}{Coef}_{i}\text{*}{x}_{i}\)

In our analysis, patients from the TCGA and ICGC cohorts were stratified into two groups based on their IEPS scores and the optimal cut-off value. We employed PCA analysis and scatter plots using the R package “ggplot2” to visualize data from both databases. Survival analysis for the prognostic signature was performed using the R packages “survival” and “survminer,” which allowed us to generate the optimized cutoff and Kaplan Meier survival curves. The predictive accuracy was assessed through receiver operating characteristic (ROC) curves, risk plots, and the concordance index.

Analysis of tumor immune microenvironment (TIME)

We employed Single-sample Gene Set Enrichment Analysis (ssGSEA) via the R package “GSVA” to quantify the infiltration levels of immune cell types within the TIME of HCC patients. Additionally, we utilized Cell-type Identification by Estimating Relative Subsets of RNA Transcripts (CIBERSORT) to ascertain the relative abundance of these immune cells.

Pathway analysis

Gene Set Enrichment Analysis (GSEA) was conducted to identify differentially enriched pathways between the low-IEPS and high-IEPS groups. Pathways were considered significantly enriched if they met the following criteria: a normalized enrichment score greater than 1, a nominal p-value less than 0.05, and a false discovery rate q-value below 0.25.

Construction and valuation of nomogram

Develop a nomogram incorporating the independent risk factors identified through multivariate COX regression analysis to forecast overall survival (OS) in HCC patients. The nomogram’s accuracy was assessed using calibration curves and C-index within both the ICGC and TCGA cohorts. We utilized the “rms” package for the construction and validation of the nomogram. Additionally, the nomogram’s predictive accuracy was benchmarked against other prognostic factors using the ROC curve, employing the R package “timeROC”. Decision curve analysis (DCA) curve was used to assess the clinical utility.

Immunohistochemistry

Specimens of HCC and adjacent-tumor liver tissue were obtained from Sun Yat-sen Memorial Hospital of Sun Yat-sen University. Slides were subjected to dewaxing, rehydration, and antigen retrieval using microwave irradiation. Antibodies were applied and incubated overnight at 4 °C. Subsequently, the secondary antibodies were applied for 30 min at room temperature, followed by DAB staining and hematoxylin counterstaining. Finally, the slides were observed under the microscope. The antibodies used are summarized in Supplementary Table 2.

Cell source and culture

In our study, human normal hepatocyte line L-O2, HCC cell lines Huh7, SNU-449, Hep 3B2.1-7, MHCC97-H, Hep-G2, SK-Hep-1, and murine HCC cell lines Hepa 1–6 were purchased from the Chinese Academy of Sciences Cell Bank (Shanghai, China). Additionally, the use of these cell lines was approved by the ethical review board of Sun Yat-Sen University, ensuring compliance with ethical standards in research. All the cell lines were cultured in DMEM (Gibco, USA) supplemented with 1% penicillin/streptomycin and 10% fetal bovine serum (Moocow, China). Cells were maintained at 37 °C and 5% CO2.

Small interfering RNA transfection

To knock down the expression of PIGU and CAD, siRNA targeting them was constructed. The sequences are as follows: siPIGU#1(GCAAUCCAGGACUUCAAUATT), siPIGU#2(UAUUGAAGUCCUGGAUUGCTT), siCAD#1(TCCCATAACACCTCATTATGTAA) and siCAD#2(CCCATAACACCTCATTATGTAAC). HCC cell lines were transfected with siRNA using the transfection reagent (Neofect, China), and subsequent experiments were conducted 48 h after transfection.

Establishment of stable CAD-overexpressed Hepa1-6 cells

Transfect 293T cells with the CAD-overexpressed, pMD2G, and psPAX2 plasmids in a 6-well plate. After 24 h, collect the 293T cell culture medium, which was used to transfect Hepa1-6 cells to overexpress the CAD protein. After 48 h of infection, cells were selected using puromycin (Beyotime, China).

RNA extraction and qRT-PCR assays

RNA from cells was isolated using TRIZOL reagent (Magen, China) and subsequently reverse transcribed into cDNA with a reverse transcription kit (Vazyme, China). Quantitative real-time PCR (qRT-PCR) analyses were performed on QuantStudio Dx (Thermo Fisher Scientific, USA), with GAPDH serving as the reference gene. The primers utilized in this study are detailed in the online Supplementary Table 3.

Western blotting

The proteins of cell lines were extracted using the lysis solution (Beyotime, China), and the protein concentration were measured with the bicinchoninic acid kit (Beyotime, China). An equal volume of loading buffer (Beyotime, China) was used to denature the protein in the boiling-water for 3 min. Proteins were resolved by SDS-PAGE and subsequently transferred to a PVDF membrane (Thermo Fisher Scientific, USA). The membrane was blocked with a 5% BSA to prevent non-specific binding. The membrane was incubated overnight with primary antibodies at 4 °C, then 1-hour with secondary antibodies at room temperature. Bands were detected using enhanced chemiluminescence (ECL) and the ABC system. Band intensities were quantified utilizing ImageJ software (National Institutes of Health, USA), with GAPDH serving as an internal control. The primary antibodies are summarized in online Supplementary Table 2.

ELISA analysis

The supernatant of the HCC cells was collected by centrifugalization and the concentrations of TGF-β1 were determined using the ELISA Kits (Abcam, USA) in line with the manufacturers’ instructions.

Animal experiments

This study employed male C57BL/6 mice at 5 weeks of age purchased from Zhuhai BesTest Bio-Tech Company (Zhuhai, China). Hepa1-6 were prepared into a cell suspension at an appropriate concentration with PBS, and then mixed equally with Matrigel (Corning, USA). Mice were anesthetized with tribromoethanol (Sigma-Aldrich, USA) before the surgical procedure. An 8 mm longitudinal incision was made in the upper abdomen of the mice, and then a 25 µL cell suspension of Hepa1-6 cells stably transfected with CAD (2 × 106 per mouse) was injected into the left lobe of the liver in C57BL/6 mice using a microsyringe. Two weeks later, in vivo imaging was conducted on the mice, and the liver weights were assessed. The animal experiment was granted approval by the Animal Ethics Committee of Sun Yat-sen University.

Multiplex immunohistochemistry

Multiplex immunohistochemistry was used to evaluate the effect of CAD on the recruitment of Tregs within the tumor microenvironment (TME) in orthotopic HCC models. The reagent kits were purchased from Shanghai Record Biotech (Shanghai, China). Tissue slides were first deparaffinized using a warm chamber, and then hydrated through a series of graded ethanol solutions, followed by antigen retrieval using a citrate buffer (Servicebio, China). The tissues were incubated with a 5% BSA solution. Then the tissues were incubated with the primary antibody and left to incubate overnight at 4 °C, followed by the application of the secondary antibody for 30 min at room temperature. Then the slides were reacted with a TSA fluorescent dye solution and subsequently underwent another wash to remove unbound antibodies. Repeat the above steps to incubate with different antibodies. Finally, the slides were stained with DAPI solution for 10 min and observed under a fluorescence microscope. The antibodies used are summarized in online Supplementary Table 2.

Statistical analysis

We compared OS using the Kaplan Meier and log-rank test, and compared clinical characteristics between low-IEPS and high-IEPS groups with a chi-square test. Multivariate and univariate Cox regression assessed IEPS’s predictive independence among HCC patients’ clinical traits. Statistical analyses were conducted using IBM SPSS Statistics (Version 25.0) and R (Version 4.1.0).

Results

Consensus clustering identified two distinct subgroups of HCC based on IEGs

Figure 1 showed the detailed flow chart of our study. We gathered 182 previously reported IEGs. According to the data from HCC patients in the TCGA database, 18 IEGs were highly expressed in HCC compared to adjacent liver tissues (Fig. 2A). Using unsupervised clustering analysis based on 18 IEGs, we identified two clusters (IEG cluster 1 and IEG cluster 2) within the TCGA cohort (Fig. 2B-C; Supplementary Fig. 1A-H). Principal component analysis (PCA) revealed significant differences in IEGs expression between the two clusters. (Fig. 2E). We then analyzed the correlation between cluster classification and molecular pathological characteristics. Chi-square tests revealed that IEG cluster 2 was significantly associated with higher AFP levels (< 400 µg/L) and more advanced tumor pathological grades (Grade III-IV; Fig. 2D). Kaplan-Meier curves and the log-rank test further revealed that patients in Cluster 2 had significantly worse OS compared to those in IEG cluster 1 (Fig. 2G). KEGG analysis indicated significant enrichment of several tumor-associated pathways in Cluster 2, including the NOTCH and mTOR signaling pathways. Notably, immune-related pathways also exhibited significant differences between the two clusters (Fig. 2F). GSEA further revealed significant differences between the two clusters in immune-related pathways, such as the chemokine production, regulatory T cell differentiation, myeloid cell activation involved in immune response, and negative regulation of immune response (Fig. 2H). This suggested that patients in IEG cluster 2 may have an immunosuppressive state.

Fig. 1
figure 1

Study flow chart

Full size image
Fig. 2
figure 2

The clustering of the IEG clusters of HCC. (A) Volcano plot of the highly expressed IEGs. (B) Consensus clustering was performed on HCC patients from the TCGA cohorts based on 18 IEGs. (C) The cumulative distribution function (CDF) plot indicated that the curve remained relatively flat at K = 2. The relative change in the area under the CDF curve between K and K-1 showed a more pronounced slope change after K values of 2 and 3. Consequently, K = 2 was selected as the optimal number of clusters. (D) The expression heatmap of IEGs was shown for the two clusters, along with clinical features. (E) PCA diagram illustrated the clustering between the two IEG clusters. (F) Heatmap showing differences in KEGG pathway between IEG cluster 1 and IEG cluster 2. (G) Survival curves for the two clusters were depicted. (H) Differences in immune-related pathways between IEG cluster 1 and IEG cluster 2. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001

Full size image

Immune characteristics of IEG clusters

Then we used ssGSEA to compare immune cell infiltration between the two clusters for immunological profiling. The box plot revealed that compared to IEG cluster 1, there was an increase in infiltration of activated CD4+ T cells, activated DCs, myeloid-derived suppressor cells (MDSCs), Tregs, and Type 2 helper T cells (Th2 cells) in IEG cluster 2 (Fig. 3A). Activated CD4+ T cells are implicated in the initiation of antitumor responses; however, their chronic activation can lead to immune exhaustion, thereby dampening immune reactions within the TME. Furthermore, they can differentiate into Th2 cells and Tregs, which exert immunosuppressive effects within the tumor [17]. DCs are considered to be a core component of the TME, capable of promoting antitumor responses. However, a specific subset of these cells, plasmacytoid dendritic cells (pDCs), can facilitate tumor growth through expressing immunosuppressive molecules such as PD-L1, ICOSL, and indoleamine 2,3-dioxygenase, or promoting the expansion of Tregs [18, 19]. MDSCs, Tregs, and Th2 cells are all implicated in promoting tumor progression. They play roles in the establishment of pre-metastatic niches, the stimulation of angiogenesis, and the enhancement of immune evasion mechanisms in tumors [20,21,22,23]. These results indicated the presence of a potential immunosuppressive microenvironment in HCC patients within IEG cluster 2. Analysis using CIBERSORT further substantiated this inference. In cluster 2, there was an increased infiltration of activated memory CD4+ T cells, follicular helper T cells (Tfh cells), Tregs, and macrophages, whereas B cells and NK cells were reduced (Fig. 3B). In B-cell lymphomas, Tfh cells modulate the TME through the secretion of cytokines, consequently facilitating tumor progression [24, 25]. Furthermore, we found that the infiltration of Tfh cells, Tregs, and M0 macrophages was positively correlated with the expression of the majority of IEGs (Fig. 3C). It is noteworthy that several algorithms have all demonstrated differences in the infiltration of Tregs. Numerous studies have shown that Tregs exerted immunosuppressive effects in tumors and promoted their progression [26,27,28]. Therefore, we speculated that Tregs contribute to the immunosuppressive microenvironment and immune evasion in HCC of patients of IEG cluster 2. In addition, we calculated the differences between the two clusters in immune cell infiltration and immune-related pathways using GSEA (Fig. 3D, E). In TIME of IEG cluster 2 patients, we observed increased infiltration of aDCs, iDCs, and Tregs, alongside reduced NK cell and macrophage infiltration. In terms of immune-related pathways, we were surprised to find that patients in IEG cluster 2 showed increased expression in the MHC class I pathway and HLA pathway, which are generally thought to enhance tumor immunity. However, at the same time, the expression of Type II IFN Response pathway was also enhanced. Current research suggested that tumor cells may increase the expression of PD-L1 through the IFN-γ signaling pathway, which can suppress the activity of T cells and thus help tumor cells evade the immune system’s attack [29, 30]. Collectively, IEGs exert a significant impact on the TME of HCC. Stratification of patients based on IEG profiles revealed distinct immunological landscapes, characterized by varying degrees of immune cell infiltration and enrichment of immune-related pathways within HCC, culminating in divergent clinical outcomes. These findings underscore the heterogeneity inherent within the TME of HCC and the pivotal role of IEGs therein, thereby warranting further investigation into the functional implications of IEGs in the context of HCC.

Fig. 3
figure 3

The difference of immune cell infiltration and pathway enrichment between two IEG clusters. (A) Analysis of differential immune cell infiltration levels using the ssGSEA algorithm. (B) Analysis of differential immune cell infiltration levels using the CIBERSORT algorithm. (C) Differential immune cell infiltration levels about the every single IEGs. (D) Differences in immune cell infiltration between IEG cluster groups predicted by the GSVA algorithm. (E) Differences in immune-related pathways between IEG cluster groups predicted by the GSVA algorithm. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001

Full size image

Establishment and validation of IEPS

To further identify key genes in the IEGs, we first applied univariate Cox regression analysis to 18 IEGs and found that 14 of them were significantly associated with OS in HCC patients in TCGA cohort (p < 0.05) (Table 1). Subsequently, through lasso algorithm analysis, we identified two genes (CAD and PIGU) for constructing the immune evasion prognostic signature (IEPS) (Fig. 4A, B). Multivariate COX regression analysis showed that CAD and PIGU were independently associated with the OS of HCC patients (p < 0.001). (Fig. 4C). Based on the formula, we calculated the IEPS score for HCC patients in the TCGA cohort and divide HCC patients into the high IEPS group (103 HCC patients) and the low IEPS group (267 HCC patients) based on the optimal cutoff value. Figure 4E shows the distribution of IEPS scores and survival status among the patients (Fig. 4E). Kaplan-Meier curves and log-rank tests demonstrated that the patients in the high IEPS group had a significantly worse OS compared to those in the low IEPS group (Fig. 4D). The ROC curves demonstrated that the IEPS score is a reliable indicator of OS, with the area under the curve (AUC) value for predicting 1-year OS being 0.768, 2-year OS being 0.705 and 3-year OS being 0.683 for patients in the TCGA cohort (Fig. 4F). Using the ICGC cohort as a validation set to assess the prognostic predictive power of IEPS, we achieved results that were similar to the TCGA cohort (Fig. 4G-I). Figure 4G and H demonstrated that the patients in the high IEPS group had significantly shorter OS compared to those in the low IEPS group. Furthermore, the ROC curve demonstrated that the IEPS score has excellent predictive power for prognosis in the ICGC cohort (The AUCs of 1, 2, and 3-year are 0.712, 0.685 and 0.731, p < 0.01) (Fig. 4I). The PCA plots clearly showed distinct clustering between the two subgroups (Fig. 4J, K). The above results indicated that the IEPS score can accurately predict the outcomes of HCC patients.

Table 1 The eighteen immune evasion-related genes
Full size table
Fig. 4
figure 4

Establishment and identification of the IEPS. (A-B) Lasso regression analysis showed that the number of variables corresponding to the best lambda value was 2. (C) The forest plot indicates the correlation between these two IEGs and patient OS. (D) The Kaplan–Meier curve showed that the OS of patients in the high-IEPS group was significantly worse in TCGA cohort. (E) Distributions of IEPS scores and survival status of HCC patients in the TCGA cohort. (F) ROC curves of IEPS for predicting the 1/2/3-year survival in the TCGA cohort. (G) The Kaplan–Meier curve showed that the prognosis of patients in the high-IEPS group was significantly worse in ICGC cohort. (H) Distributions of IEPS scores and survival status of HCC patients in the ICGC cohort. (I) ROC curves of IEPS for predicting the 1/2/3-year survival in the ICGC cohort. (J) PCA diagram illustrated the clustering of HCC patients of two groups in TCGA cohort. (K) PCA diagram illustrated the clustering of HCC patients of two groups in ICGC cohort

Full size image

Statistical analysis and subgroup analysis of IEPS

To investigate the relationship between the IEPS score and clinicopathological characteristics of HCC patients, we analyzed the changes in clinical characteristics of patients as the IEPS score increased. The heat map illustrated that the higher IEPS score correlated with advanced TNM stages, higher WHO grades, higher mortality rate and elevated α-fetoprotein (AFP) (Fig. 5A). Additionally, analysis revealed that female patients, as well as those with WHO grade III-IV, TNM stage III-IV, and high AFP levels (≥ 400 µg/L), had higher IEPS scores, with no significant differences observed in age subgroups (Fig. 5B-F). To evaluate the predictive power of IEPS across various HCC subgroups, we conducted survival analyses across different clinical subgroups. We found that high IEPS scores correlated with poorer prognosis in both TNM I-II and III-IV groups, as well as in WHOI-II and WHOIII-IV, AFP < 400 and ≥ 400 µg/L, female and male, and age < 65 and ≥ 65 years subgroups (Fig. 5G-P). This suggested that IEPS score reliably predicts HCC prognosis across diverse clinical subgroups.

Fig. 5
figure 5

Clinicopathological features and stratification analysis of the IEPS. (A) Heatmap of the associations between the expression levels of the two IEGs and clinicopathological features in the TCGA cohort and ICGC cohort. (B-F) Patients with different clinicopathological features (including gender, AFP levels, WHO grade, TNM stage, but not age) had different levels of IEPS scores. (G-P) The IEPS retained its prognostic value in multiple subgroups of HCC patients (including patients with male or female, WHO grade III-IV or I-II, TNM stage III-IV or I-II, age < 65 or ≥ 65 years and AFP < 400 or ≥ 400 ug/L)

Full size image

Construction and validation of the nomogram based on IEPS

We conducted univariate and multivariate Cox regression analyses on the TCGA and ICGC cohorts to evaluate the potential of IEPS scores as independent prognostic factors in patients with HCC (Fig. 6A, B). The univariate Cox analysis including age, gender, AFP levels, grade, stage, and IEPS score showed that IEPS is a strong risk factor for OS [Hazard Ratio (HR): 3.485, 95% Confidence Interval (CI): 2.336–5.198, p < 0.001]. In the multivariate COX analysis, we confirmed that IEPS is an independent risk factor for OS (HR: 3.100, 95% CI:2.024–4.748, p < 0.001). This result was validated on the ICGC cohort (HR: 5.190, 95% CI: 2.417–11.144, p < 0.001 in the univariate Cox analysis; HR: 3.938, 95% CI: 1.810–8569, p < 0.001 in the multivariate COX analysis).

Fig. 6
figure 6

Development and validation of prognostic nomogram base on IEPS. (A) Univariate and multivariate analyses revealed that IEPS score was an independent prognostic predictor in the TCGA cohort. (B) Univariate and multivariate analyses revealed that IEPS score was an independent prognostic predictor in the ICGC cohort. (C) Nomogram based on IEPS and TNM stage. (D) Calibration plots of the nomogram for predicting the probability of OS at 1, 2, and 3 years in the TCGA cohort. (E) Calibration plots of the nomogram for predicting the probability of OS at 1, 2, and 3 years in the ICGC cohort. (F-H) Time-dependent ROC curves for the nomogram based on IEPS score and TNM stage in the TCGA cohort (for predicting 1, 2, and 3-year OS)

Full size image

Subsequently, to better apply IEPS in clinical practice, we developed a nomogram based on the IEPS score and tumor stages (Fig. 6C). The calibration curves showed excellent agreement between predicted and actual survival rates at the 1-, 2-, and 3-year marks (Fig. 6D, E). The ROC curves demonstrated that the predictive accuracy of the nomogram for patient prognosis has been further improved compared to IEPS (AUCs at 1, 2, and 3-year reached 0.777, 0.719 and 0.744 in the TCGA cohort, AUCs at 1, 2, and 3-year reached 0.834, 0.749 and 0.766 in the ICGC cohort) (Fig. 6F-H, Supplementary Fig. 1I-K). Moreover, DCA curves showed that the prognostic nomogram has a high net benefit for the clinical application (Supplementary Fig. 1L-M). The above results indicate that the nomogram we developed, based on IEPS and stage, has an exciting predictive capability, which is convenient for clinical application.

Immunological characteristics of two IEPS subgroups

Next, we want to clarify the immunological characteristics of IEPS groups. The box plot and Sankey diagram showed that compared to IEG cluster 1, the majority of HCC patients within IEG cluster 2 demonstrate elevated IEPS scores, and exhibit a trend towards decreased OS (Fig. 7A, B). To further comprehend the immunological variance between high-IEPS and low-IEPS groups, GSEA was utilized to determine infiltration of immune cells and enrichment of immune-related pathways within the two subgroups. In the high IEPS group, the enrichment of the regulatory T cell differentiation was observed, which implied the presence of an immunosuppressive condition (Fig. 7C). In the low IEPS group, immune-related pathways such as complement activation, the classical pathway, and the lectin pathway are predominantly activated. These pathways play a pivotal role in immune defense, suggesting an enhanced immune response within the low IEPS group. (Fig. 7D). Furthermore, we examined the differences in immune cell infiltration and immune-related pathways between the high IEPS group and the low IEPS group using the GSVA method. The results showed that compared to the low score group, the high score group had more infiltration of Tregs and macrophages, but fewer NK cells and neutrophils (Fig. 7E). In terms of immune-related pathways, the high IEPS group showed elevated expression in MHC class I and Cytolytic activity pathway, but lower expression in Type I IFN Response and Type II IFN Response pathway (Fig. 7F). These results further suggested that the high IEPS group may be in a state of immunosuppression.

Fig. 7
figure 7

Immune characteristics of two IEPS subgroups. (A) The difference in IEPS score among IEG clusters. (B) Sankey plot of IEG subtype distribution in groups with different IEPS scores and survival status. (C-D) GSEA was used to reveal the IEPS-related pathways enriched in the IEPS subgroups. (E) Difference of immune cell infiltration between two groups via GSEA analysis. (F) Difference of immune-related pathways between two groups via GSEA analysis. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001

Full size image

CAD and PIGU promote the infiltration of Tregs in HCC

The development of tumor is closely related to the immune microenvironment. Next, to depict the TIME landscape of the high-IEPS group, we used ssGSEA to examine the immune cell infiltration associated with PIGU and CAD in HCC (Fig. 8A). Pearson correlation analysis indicated that the infiltration levels of Tregs and M0 macrophages were positively correlated with the expression of PIGU and CAD. Further validation using the ICGC dataset yielded similar conclusions (Fig. 8B). Figure 8C-F illustrated the relationship of the expression of Tregs, macrophages and IEPS scores. This suggests that Tregs may be key immune cells facilitating immune evasion in the TME of high IEPS score group and may also be a reason for the poor response to immunotherapy in HCC, which was consistent with the result we obtained earlier. Furthermore, we compared the expression of several important immune checkpoints (ICs) between the low-IEPS and high-IEPS groups. Including PDCD1, CTLA4, TIM-3, CTLA-4, LAG3, TIGIT, VISR, OX40 and OX-40 L. The results showed that the expression of all the above checkpoints is higher in the high IEPS score group compared to the low IEPS score group, indicating an immunosuppressive state in the high IEPS score group, which may have a potential better response to immune checkpoint inhibitor therapy (Fig. 8G-N). In summary, we conducted a comprehensive analysis of IEGs and found that they play a significant role in the formation of the immunosuppressive microenvironment in HCC, with CAD and PIGU being the core genes. The IEPS model and nomogram constructed based on IEGs have strong clinical value, as they can accurately predict the OS of HCC patients and assess the TIME in HCC. It will help us develop personalized treatment plans for HCC patients and improve their prognosis.

Fig. 8
figure 8

Characteristics of immune cells and immune checkpoints between two groups. (A) Correlation between the CAD, PIGU and immune cells in the TCGA cohort based on CIBERSORT. (B) Correlation between the CAD, PIGU and immune cells in the ICGC cohort based on CIBERSORT. (C-F) The association between IEPS score and the infiltration of Tregs and M0 macrophages. (G-N) Comparison between the expression of several prominent immune checkpoints in the high- and low-IEPS subgroups of the TCGA cohort. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001

Full size image

CAD and PIGU promote the expression of TGF-β1 in HCC

Subsequently, we intend to further investigate the functions these two genes in HCC. Immunohistochemical assays reveal that, in comparison to the adjacent-tumor liver tissue, the expression levels of both genes are markedly elevated in the tumor samples (Fig. 9A). CAD, a multi-enzyme protein, initiates and regulates pyrimidine de novo biosynthesis, significantly impacting nucleotide balance, cell growth, and proliferation [31]. Studies have shown that CAD can bind with GOT1 and UMPs to form a pyrimidine complex, which is regulated by AMPK and plays an important role in tumor progression [32]. However, there is still a lack of research on the role of CAD in the immune microenvironment of tumors. Our preceding analysis has demonstrated a correlation between CAD expression and an increased infiltration of Tregs within the TME. This association may be pivotal to the role in establishing an immunosuppressive microenvironment in HCC. Prior research has showed that in HCC, tumor-derived TGF-β1 fosters the differentiation of Tregs, consequently inducing immune tolerance within the TME [33]. Therefore, we further investigated whether there is a correlation between CAD and TGF-β1. The data in TCGA database showed the expression of CAD in HCC was significantly positively correlated with that of TGF-β1 (Fig. 9H). Then we conducted cellular experiments to investigate the relationship between CAD and TGF-β1 in HCC cell lines. We detected the expression of CAD in normal liver cell lines and HCC cell lines using qRT-PCR (Fig. 9B). The results showed that the expression of CAD in Huh7 and SNU-449 were the lowest among all cell lines, so we transfected them with CAD overexpression plasmids to analyze the potential role of CAD in HCC. The results indicated that the mRNA expression of TGF-β1 increased along with the upregulation of CAD expression (Fig. 9C, D). The Western blot results indicated that this was not only at the RNA level but also at the protein level (Fig. 9E-G). Additionally, we knocked down CAD using siRNA in Hep3B2.1-7 and MHCC97-H, which were with the highest basic expression level of CAD. The results showed that TGF-β1 decreased along with the downregulation of CAD (Fig. 9I-M). In addition, we also examined whether the secretion of TGF-β1 by HCC cells was affected by CAD. The ELISA results indicated that increased expression of CAD enhanced the secretion of TGF-β1, while decreased expression reduced the secretion of TGF-β1(Fig. 9N-O). Moreover, we found that PIGU promotes the expression of TGF-β1 at both the mRNA and protein levels and increased the secretion of TGF-β1 (Supplementary Fig. 2).

Fig. 9
figure 9

Verifying the association between two IEGs and TGF-β1 using IHC and cell experiments. (A) Expression level of CAD and PIGU in 8 pairs of HCC and liver tissues using IHC (White scale bar is 50 μm). (B) Basic expression of CAD in different HCC cell lines. (C) qRT-PCR detected the over-expression efficiency of CAD in Huh7 and SNU-449 cell lines. (D) qRT-PCR detected the changes in the expression levels of TGF-β1 before and after CAD over-expressed in Huh7 and SNU-449 cell lines. (E-G) Western blot analysis of the changes in the expression levels of TGF-β1 before and after CAD over-expressed in Huh7 and SNU-449 cell lines. (H) Correlation between CAD and TGF-β1 in TCGA database. (I) qRT-PCR detected the over-expression efficiency of CAD in Hep 3B2.1-7 and MHCC97-H cell lines. (J) qRT-PCR detected the changes in the expression levels of TGF-β1 before and after CAD knockdown in Hep 3B2.1-7 and MHCC97-H cell lines. (K-M) Western blot analysis of the changes in the expression levels of TGF-β1 before and after CAD knockdown in Hep 3B2.1-7 and MHCC97-H cell lines. (N). ELISA detected the change of the of TGF-β1 in the supernatant of Huh7 and SNU-449 cell lines before and after CAD over-expressed. (O). ELISA detected the change of the of TGF-β1 in the supernatant of Hep 3B2.1-7 and MHCC97-H cell lines before and after CAD knockdown. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001

Full size image

CAD promotes the increase of Tregs in orthotopic HCC model

To further validate the function of CAD, CAD-overexpressing Hepa1-6 cells and vector control cells were orthotopically transplanted into the livers of C57BL/6 mice for intrahepatic tumor implantation experiments. After the tumors had grown for three weeks, the mice were subjected to in vivo imaging, followed by dissection to remove the liver for observation and comparison of tumor sizes (Fig. 10A). The results showed that the tumors in the overexpression group were significantly larger than those in the control group (Fig. 10B-E). Liver tissues were collected for mIHC, and the results indicated that CAD promoted the expression of TGF-β1 and increased the infiltration of Tregs in the TME (Fig. 10F). Synthesizing the results, our experiments confirm that CAD and PIGU contribute to the infiltration of Treg cells by promoting the expression of TGF-β1, leading to an immunosuppressive microenvironment in HCC. This could be a key factor in the poor prognosis of patients. CAD may potentially serve as a novel therapeutic target for HCC, activating the patients’ antitumor immune response and enhancing the efficacy of immunotherapy.

Fig. 10
figure 10

Effects of CAD on the growth of orthotopic HCC model and infiltration of Tregs. (A) Scheme of orthotopic implantation mice HCC model. (B-C) In vivo imaging of mice. (D-E) Photograph of mice livers from orthotopic implantation model after harvest. (F) Multiplex immunohistochemistry staining of liver tissues (White scale bar is 50 μm). *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001

Full size image

Discussion

The TME and immune tolerance of the liver may lead to immune evasion of tumor cells. Research into the mechanisms of immune evasion contributes to the advancement of immunotherapy for HCC. In our study, by intersecting the 182 IEGs with the genes highly expressed in HCC according to the TCGA database, we obtained a gene set containing 18 IEGs. Based on this gene set, we identified two distinct HCC clusters through consensus clustering analysis, with patients in cluster 2 exhibiting poorer prognosis and more malignant tumor biological behavior. Further analysis of immune cell infiltration and immune-related pathways revealed a significant immunosuppressive microenvironment in cluster 2. This highlighted the substantial heterogeneity of the immune microenvironment in HCC, which is notably associated with IEGs. Exploring these mechanisms is beneficial for the development of HCC treatment.

To further identify key genes, based on the TCGA-LIHC dataset, we applied Lasso-Cox regression analysis to screen two genes (PIGU and CAD) from these 18 genes and established an new signature (IEPS) based on them. The results showed that in both the TCGA training and ICGC validation cohorts, high IEPS group patients exhibited advanced TNM stages, shorter OS and higher mortality. This enables us to identify patients with poor prognosis and advanced disease at the time of diagnosis, and to implement personalized treatment strategies based on IEPS, such as increasing the frequency of follow-ups for patients with high IEPS scores to monitor progression of HCC. Furthermore, in order to better utilize the IEPS in the clinical practice, we also conducted univariate and multivariate Cox regression analyses with IEPS score as an independent factor for patients, and the results indicated that IEPS is an independent prognostic factor for HCC patients. Based on this, we developed a clinical nomogram model to predict the 1-year, 2-year, and 3-year OS for HCC patients and validated its accuracy and the clinical utility using calibration curves and DCA curves.

CAD is a multifunctional protein that participates in the first three rate-limiting steps of pyrimidine nucleotide synthesis. Studies have shown that the absence of Argininosuccinate synthase 1 (ASS1) activates CAD, leading to increased pyrimidine synthesis and contributing to tumor progression [34]. In glioblastoma, the CHIP protein promotes the ubiquitination and degradation of CAD, inhibiting the progression of the tumor, and targeting CAD can inhibit intracellular pyrimidine synthesis, thereby suppressing tumor progression [35, 36]. In human cell lines lacking MLH1 or MSH6, the rate of CAD gene amplification is increased by 50 to 100 times, increasing susceptibility to cancer [37]. In this study, the high expression of CAD in HCC indicated its role in promoting the progression of HCC, which is consistent with previous research. This provided a strong supplement for the clinical use of CAD as a therapeutic target for HCC.

PIGU, a crucial component of the GPI-T complex, which is involved in the biosynthesis of a specific type of membrane anchor, is the first cancer-linked part of this complex to be described in scientific research [38, 39]. In bladder cancer, the protein PIGU contributes to cancer growth by increasing the activity of the urokinase receptor, which is essential for the advancement of the disease [40]. According to the comprehensive field Synopsis and systematic meta-analyses of genetic association studies, the PIGU gene increases susceptibility to cutaneous melanoma [41]. PIGU and the asparagine-linked glycosylation protein 5 homolog (ALG5) may undergo a gene fusion that could lead to the development of prostate cancer [42]. In breast cancer, PIGU promotes the exhaustion of T cells in the TME, contributing to tumor progression In breast cancer, PIGU promotes the exhaustion of T cells in the TME, contributing to tumor progression [43]. In this study, PIGU was upregulated in the high score group and was significantly associated with poor patient prognosis, which is consistent with the recent research findings.

The TME plays a regulatory role in tumor phenotype. The infiltration of immune cells is crucial for the immune evasion of tumor cells and the induction of inflammation, and it is a key characteristic of the TME [44]. We explored the relationship between immune-related pathways and immune cells infiltration between IEPS subgroups using a variety of tools. The results showed that the regulatory T cell differentiation pathway was enriched in high IEPS group. Additionally, there was a decrease in neutrophils and NK cells in the high IEPS group, but an increase in macrophages and Tregs in it. This indicates a significant immunosuppressive state in the high IEPS group. Neutrophils can suppress tumor cells through direct cytotoxic effects. For example, they can secrete reactive oxygen species (ROS), myeloperoxidase (MPO), interferon-gamma (IFN-γ) to directly inhibit tumor progression [45]. Furthermore, neutrophils can also act as antigen-presenting cells (APCs), stimulating anti-cancer T cell responses [46,47,48]. NK cells are pivotal in anti-tumor immunity, directly killing tumor cells and swiftly releasing cytokines and chemokines to rally other immune cells and bolster T and B cell responses [49]. The role of macrophages in tumors is multifaceted. In the TME, macrophages typically differentiate into two phenotypes: M1 and M2. M1 macrophages have pro-inflammatory effects and can directly kill tumor cells or eliminate them through antibody-dependent cellular cytotoxicity (ADCC). M2 macrophages, linked to immunosuppression and tumorigenesis, enhance tumor growth, invasion, and metastasis, dampen T cell anti-tumor responses, and stimulate angiogenesis, potentially advancing tumor progression [50]. Tregs are a distinct lymphocyte subset critical for immune homeostasis, autoimmunity prevention, and inflammation reduction [51]. In tumors, Tregs bolster immune evasion by secreting inhibitory cytokines like IL-10, TGF-β1, and IL-35, upregulating checkpoint receptors such as CTLA-4, LAG-3, and PD-1, and using granzyme to eliminate effector cells [52,53,54,55]. Our study revealed the immune cells and pathways that play a significant role in immune suppression and immune evasion within the TME of HCC, which may potentially contribute to the development of new treatments for HCC, particularly immunotherapies.

Immune checkpoints are vital in cancer immunotherapy, but only a fraction of HCC patients respond to immune checkpoint inhibitors (ICI) [56]. This urges us to conduct an individual assessment of patients to determine whether they should add other adjuvant therapies to enhance the efficacy of ICIs. This suggests that we can perform an IEPS score on HCC patients before immunotherapy to determine whether they should undergo immunotherapy, and also can assess the efficacy of ICIs using IEPS.

There is limited understanding of how Tregs are recruited to HCC. Previous studies have indicated that tumor-induced cellular lipid reorganization and TNFα-mediated tumor fatty acid uptake recruit CCR6+ Tregs to HCC tissues, promoting the progression of HCC [57]. In addition, there was the research indicating that MicroRNA-15a/16 − 1 suppresses the recruitment of liver Tregs and enhances liver CTLs, thereby reducing the immune suppression in HCC [58]. Our study suggests that CAD and PIGU are associated with an increase in Treg infiltration in HCC. Based on existing research, we hypothesize that PIGU and CAD may promote the recruitment of Tregs by enhancing the secretion of TGF-β1. We have demonstrated through cellular experiments that CAD and PIGU promote the expression and secretion of TGF-β1. Furthermore, animal experiments indicate that CAD promotes tumor progression in HCC and the recruitment of Tregs within the TME.

This study proposes that IEPS has a new prognostic value in HCC and provides a new set of indicators for personalized precision treatment for patients. PIGU and CAD provide perspectives on new biomarkers and therapeutic targets for HCC. However, our research has certain limitations. Though our study comprehensively assessed and validated the IEPS, limitations such as a constrained sample size, partial data, and limited clinical validation restrict the broad applicability of our results. These limitations will guide the direction of our future research.

Conclusions

Our study revealed that the IME of HCC exhibits significant heterogeneity, with IEGs playing a crucial role. CAD and PIGU act as key players in this context. The IEPS and nomogram models we developed based on PIGU and CAD can accurately predict the OS and IME status of HCC patients, providing new tools for the personalized treatment evaluation and follow-up of liver cancer patients. Mechanistically, we found that CAD and PIGU promote the infiltration of Tregs in the TME by increasing the transcription and translation of TGF-β1. CAD and PIGU has the potential to become a new therapeutic target for HCC.

Data availability

The HCC data used to support the analysis in this study were downloaded from TCGA (http://portal.gdc.cancer.gov) and ICGC (https://dcc.icgc.org/). The additional data can be obtained from the corresponding author on a reasonable request.

References

  1. Bray F, Laversanne M, Sung H, Ferlay J, Siegel RL, Soerjomataram I et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. Cancer J Clin. 2024;74(3):229–63.

  2. Villanueva A. Hepatocellular carcinoma. N Engl J Med. 2019;380(15):1450–62.

    Article  PubMed  CAS  Google Scholar 

  3. Yeung YP, Lo CM, Liu CL, Wong BC, Fan ST, Wong J. Natural history of untreated nonsurgical hepatocellular carcinoma. Am J Gastroenterol. 2005;100(9):1995–2004.

    Article  PubMed  Google Scholar 

  4. Gordan JD, Kennedy EB, Abou-Alfa GK, Beal E, Finn RS, Gade TP et al. Systemic therapy for advanced hepatocellular carcinoma: ASCO guideline update. J Clin Oncology: Official J Am Soc Clin Oncol. 2024;42(15):1830–50.

  5. Yi M, Zheng X, Niu M, Zhu S, Ge H, Wu K. Combination strategies with PD-1/PD-L1 Blockade: current advances and future directions. Mol Cancer. 2022;21(1):28.

    Article  PubMed  PubMed Central  Google Scholar 

  6. Lee A, Keam SJ, Tislelizumab. First Approval Drugs. 2020;80(6):617–24.

    PubMed  CAS  Google Scholar 

  7. Chen Z, Lan X, Du C, Xiao H. Adjuvant therapy in hepatocellular carcinoma: the IMbrave050 trial. Lancet (London England). 2024;404(10453):656.

    Article  PubMed  CAS  Google Scholar 

  8. Tabrizian P, Abdelrahim M, Schwartz M. Immunotherapy and transplantation for hepatocellular carcinoma. J Hepatol. 2024;80(5):822–25.

  9. Llovet JM, Pinyol R, Yarchoan M, Singal AG, Marron TU, Schwartz M et al. Adjuvant and neoadjuvant immunotherapies in hepatocellular carcinoma. Nat Reviews Clin Oncol. 2024;21(4):294–311.

  10. Thomson AW, Knolle PA. Antigen-presenting cell function in the tolerogenic liver environment. Nat Rev Immunol. 2010;10(11):753–66.

    Article  PubMed  CAS  Google Scholar 

  11. Du G, Dou C, Sun P, Wang S, Liu J, Ma L. Regulatory T cells and immune escape in HCC: Understanding the tumor microenvironment and advancing CAR-T cell therapy. Front Immunol. 2024;15:1431211.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  12. Evdokimova V, Gassmann H, Radvanyi L, Burdach SEG. Current state of immunotherapy and mechanisms of immune evasion in ewing sarcoma and osteosarcoma. Cancers. 2023;15(1):272.

  13. Vinay DS, Ryan EP, Pawelec G, Talib WH, Stagg J, Elkord E, et al. Immune evasion in cancer: mechanistic basis and therapeutic strategies. Sem Cancer Biol. 2015;35(Suppl):S185–98.

    Article  Google Scholar 

  14. Wu Y, Kuang DM, Pan WD, Wan YL, Lao XM, Wang D, et al. Monocyte/macrophage-elicited natural killer cell dysfunction in hepatocellular carcinoma is mediated by CD48/2B4 interactions. Hepatology (Baltimore MD). 2013;57(3):1107–16.

    Article  PubMed  CAS  Google Scholar 

  15. Zheng H, Peng X, Yang S, Li X, Huang M, Wei S, et al. Targeting tumor-associated macrophages in hepatocellular carcinoma: biology, strategy, and immunotherapy. Cell Death Discovery. 2023;9(1):65.

    Article  PubMed  PubMed Central  Google Scholar 

  16. Lawson KA, Sousa CM, Zhang X, Kim E, Akthar R, Caumanns JJ, et al. Functional genomic landscape of cancer-intrinsic evasion of killing by T cells. Nature. 2020;586(7827):120–6.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  17. Miggelbrink AM, Jackson JD, Lorrey SJ, Srinivasan ES, Waibl-Polania J, Wilkinson DS et al. CD4 T-Cell exhaustion: does it exist and what are its roles in cancer?> Cancer Clin Res. 2021;27(21):5742–52.

  18. Swiecki M, Colonna M. The multifaceted biology of plasmacytoid dendritic cells. Nat Rev Immunol. 2015;15(8):471–85.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  19. Del Prete A, Salvi V, Soriani A, Laffranchi M, Sozio F, Bosisio D, et al. Dendritic cell subsets in cancer immunity and tumor antigen sensing. Cell Mol Immunol. 2023;20(5):432–47.

    Article  PubMed  PubMed Central  Google Scholar 

  20. Lin S, Zhang X, Huang G, Cheng L, Lv J, Zheng D, et al. Myeloid-derived suppressor cells promote lung cancer metastasis by CCL11 to activate ERK and AKT signaling and induce epithelial-mesenchymal transition in tumor cells. Oncogene. 2021;40(8):1476–89.

    Article  PubMed  CAS  Google Scholar 

  21. Qi Y, Zhang L, Liu Y, Li Y, Liu Y, Zhang Z. Targeted modulation of myeloid-derived suppressor cells in the tumor microenvironment: implications for cancer therapy. Biomed Pharmacother. 2024;180:117590.

    Article  PubMed  CAS  Google Scholar 

  22. Shang Q, Yu X, Sun Q, Li H, Sun C, Liu L. Polysaccharides regulate Th1/Th2 balance: A new strategy for tumor immunotherapy. Biomed Pharmacother. 2024;170:115976.

    Article  PubMed  CAS  Google Scholar 

  23. Watson MJ, Vignali PDA, Mullett SJ, Overacre-Delgoffe AE, Peralta RM, Grebinoski S, et al. Metabolic support of tumour-infiltrating regulatory T cells by lactic acid. Nature. 2021;591(7851):645–51.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  24. Guilloton F, Caron G, Ménard C, Pangault C, Amé-Thomas P, Dulong J, et al. Mesenchymal stromal cells orchestrate follicular lymphoma cell niche through the CCL2-dependent recruitment and polarization of monocytes. Blood. 2012;119(11):2556–67.

    Article  PubMed  CAS  Google Scholar 

  25. Laurent C, Dietrich S, Tarte K. Cell cross talk within the lymphoma tumor microenvironment: follicular lymphoma as a paradigm. Blood. 2024;143(12):1080–90.

    Article  PubMed  CAS  Google Scholar 

  26. Kumagai S, Itahashi K, Nishikawa H. Regulatory T cell-mediated immunosuppression orchestrated by cancer: towards an immuno-genomic paradigm for precision medicine. Nat Reviews Clin Oncol. 2024;21(5):337–53.

    Article  Google Scholar 

  27. Kumagai S, Itahashi K, Nishikawa H. Regulatory T cell-mediated immunosuppression orchestrated by cancer: towards an immuno-genomic paradigm for precision medicine. Nat Reviews Clin Oncol. 2024;21(5):337–53.

  28. Nie P, Cao Z, Yu R, Dong C, Zhang W, Meng Y et al. Targeting p97-Npl4 interaction inhibits tumor Treg cell development to enhance tumor immunity. Nat Immunol. 2024; 25 (9):1623–36.

  29. Knopf P, Stowbur D, Hoffmann SHL, Hermann N, Maurer A, Bucher V, et al. Acidosis-mediated increase in IFN-γ-induced PD-L1 expression on cancer cells as an immune escape mechanism in solid tumors. Mol Cancer. 2023;22(1):207.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  30. Zhao T, Li Y, Zhang J, Zhang B. PD-L1 expression increased by IFN-γ via JAK2-STAT1 signaling and predicts a poor survival in colorectal cancer. Oncol Lett. 2020;20(2):1127–34.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  31. Del Caño-Ochoa F, Ramón-Maiques S. The multienzymatic protein CAD leading the de Novo biosynthesis of pyrimidines localizes exclusively in the cytoplasm and does not translocate to the nucleus. Nucleosides Nucleotides Nucleic Acids. 2020;39(10–12):1320–34.

    Article  PubMed  Google Scholar 

  32. Yang C, Zhao Y, Wang L, Guo Z, Ma L, Yang R, et al. De Novo pyrimidine biosynthetic complexes support cancer cell proliferation and ferroptosis defence. Nat Cell Biol. 2023;25(6):836–47.

    Article  PubMed  CAS  Google Scholar 

  33. Shen Y, Wei Y, Wang Z, Jing Y, He H, Yuan J, et al. TGF-β regulates hepatocellular carcinoma progression by inducing Treg cell polarization. Cell Physiol Biochemistry: Int J Experimental Cell Physiol Biochem Pharmacol. 2015;35(4):1623–32.

    Article  CAS  Google Scholar 

  34. Rabinovich S, Adler L, Yizhak K, Sarver A, Silberman A, Agron S, et al. Diversion of aspartate in ASS1-deficient tumours fosters de Novo pyrimidine synthesis. Nature. 2015;527(7578):379–83.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  35. Wang X, Yang K, Wu Q, Kim LJY, Morton AR, Gimple RC et al. Targeting pyrimidine synthesis accentuates molecular therapy response in glioblastoma stem cells. Sci Transl Med. 2019;11:eaau4972.

  36. Li G, Xiao K, Li Y, Gao J, He S, Li T. CHIP promotes CAD ubiquitination and degradation to suppress the proliferation and colony formation of glioblastoma cells. Cell Oncol (Dordrecht). 2024;47(3):851–65.

    Article  CAS  Google Scholar 

  37. Chen S, Bigner SH, Modrich P. High rate of CAD gene amplification in human cells deficient in MLH1 or MSH6. Proc Natl Acad Sci USA. 2001;98(24):13802–7.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  38. Fujita M, Kinoshita T. GPI-anchor remodeling: potential functions of GPI-anchors in intracellular trafficking and membrane dynamics. Biochim Biophys Acta. 2012;1821(8):1050–8.

    Article  PubMed  CAS  Google Scholar 

  39. Hong Y, Ohishi K, Kang JY, Tanaka S, Inoue N, Nishimura J, et al. Human PIG-U and yeast Cdc91p are the fifth subunit of GPI transamidase that attaches GPI-anchors to proteins. Mol Biol Cell. 2003;14(5):1780–9.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  40. Guo Z, Linn JF, Wu G, Anzick SL, Eisenberger CF, Halachmi S, et al. CDC91L1 (PIG-U) is a newly discovered oncogene in human bladder cancer. Nat Med. 2004;10(4):374–81.

    Article  PubMed  CAS  Google Scholar 

  41. Chatzinasiou F, Lill CM, Kypreou K, Stefanaki I, Nicolaou V, Spyrou G, et al. Comprehensive field synopsis and systematic meta-analyses of genetic association studies in cutaneous melanoma. J Natl Cancer Inst. 2011;103(16):1227–35.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  42. Pflueger D, Terry S, Sboner A, Habegger L, Esgueva R, Lin PC, et al. Discovery of non-ETS gene fusions in human prostate cancer using next-generation RNA sequencing. Genome Res. 2011;21(1):56–67.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  43. Wu H, Wu Z, Li H, Wang Z, Chen Y, Bao J, et al. Glycosylphosphatidylinositol anchor biosynthesis pathway-based biomarker identification with machine learning for prognosis and T cell exhaustion status prediction in breast cancer. Front Immunol. 2024;15:1392940.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  44. de Visser KE, Joyce JA. The evolving tumor microenvironment: from cancer initiation to metastatic outgrowth. Cancer Cell. 2023;41(3):374–403.

    Article  PubMed  Google Scholar 

  45. Zhang M, Qin H, Wu Y, Gao Q. Complex role of neutrophils in the tumor microenvironment: an avenue for novel immunotherapies. Cancer biology & medicine. 2024;21(10):849–863.

  46. Gosselin EJ, Wardwell K, Rigby WF, Guyre PM. Induction of MHC class II on human polymorphonuclear neutrophils by granulocyte/macrophage colony-stimulating factor, IFN-gamma, and IL-3. J Immunol (Baltimore Md: 1950). 1993;151(3):1482–90.

    Article  CAS  Google Scholar 

  47. Smith WB, Guida L, Sun Q, Korpelainen EI, van den Heuvel C, Gillis D, et al. Neutrophils activated by granulocyte-macrophage colony-stimulating factor express receptors for interleukin-3 which mediate class II expression. Blood. 1995;86(10):3938–44.

    Article  PubMed  CAS  Google Scholar 

  48. Reinisch W, Lichtenberger C, Steger G, Tillinger W, Scheiner O, Gangl A, et al. Donor dependent, interferon-gamma induced HLA-DR expression on human neutrophils in vivo. Clin Exp Immunol. 2003;133(3):476–84.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  49. Maskalenko NA, Zhigarev D, Campbell KS. Harnessing natural killer cells for cancer immunotherapy: dispatching the first responders. Nat Rev Drug Discovery. 2022;21(8):559–77.

    Article  PubMed  CAS  Google Scholar 

  50. Pan Y, Yu Y, Wang X, Zhang T. Tumor-Associated macrophages in tumor immunity. Front Immunol. 2020;11:583084.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  51. Lee W, Lee GR. Transcriptional regulation and development of regulatory T cells. Exp Mol Med. 2018;50(3):e456.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  52. Toomer KH, Malek TR. Cytokine signaling in the development and homeostasis of regulatory T cells. Cold Spring Harb Perspect Biol. 2018;10:a028597.

  53. Maj T, Wang W, Crespo J, Zhang H, Wang W, Wei S, et al. Oxidative stress controls regulatory T cell apoptosis and suppressor activity and PD-L1-blockade resistance in tumor. Nat Immunol. 2017;18(12):1332–41.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  54. Budhu S, Schaer DA, Li Y, Toledo-Crow R, Panageas K, Yang X et al. Blockade of surface-bound TGF-β on regulatory T cells abrogates suppression of effector T cell function in the tumor microenvironment. Sci Signal. 2017;10:eaak9702

  55. Kalia V, Penny LA, Yuzefpolskiy Y, Baumann FM, Sarkar S. Quiescence of memory CD8(+) T cells is mediated by regulatory T cells through inhibitory receptor CTLA-4. Immunity. 2015;42(6):1116–29.

    Article  PubMed  CAS  Google Scholar 

  56. Chen K, Shuen TWH, Chow PKH. The association between tumour heterogeneity and immune evasion mechanisms in hepatocellular carcinoma and its clinical implications. Br J Cancer. 2024;131(3):420–9.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  57. Wang Y, Chen W, Qiao S, Zou H, Yu XJ, Yang Y, et al. Lipid droplet accumulation mediates macrophage survival and Treg recruitment via the CCL20/CCR6 axis in human hepatocellular carcinoma. Cellular & molecular immunology. 2024;21(10):1120–30.

  58. Liu N, Chang CW, Steer CJ, Wang XW, Song G. MicroRNA-15a/16– 1 prevents hepatocellular carcinoma by disrupting the communication between Kupffer cells and regulatory T cells. Gastroenterology. 2022;162(2):575–89.

    Article  PubMed  CAS  Google Scholar 

Download references

Acknowledgements

We express our gratitude to all colleagues and students involved in this study, as well as the institutions that provided funding for this research.

Funding

This research was funded by National Natural Science Foundation of China (No. 82073045, 82103090, 82403734), Science and Technology Program of Guangzhou (No.202201020311), Guangdong Basic and Applied Basic Research Foundation (No. 2024A1515010579, 2023A1515010745, 2023A1515220131, 2022A1515012391), China Postdoctoral Science Foundation (2023M744042, 2024T171073), Beijing Xisike Clinical Oncology Research Foundation (Y-MSDPU2022-0826).

Author information

Author notes

  1. Jiahua Wen, Kai Wen and Meng Tao contributed equally to this work.

Authors and Affiliations

  1. Department of Hepatobiliary Surgery, Sun Yat-Sen Memorial Hospital, Sun Yat-Sen University, Guangzhou, 510120, China

    Jiahua Wen, Kai Wen, Meng Tao, Zhenyu Zhou, Weidong Wang, Zian Huang, Qiaohong Lin, Huoming Li, Haohan Liu, Yongcong Yan & Zhiyu Xiao

  2. Guangdong Provincial Key Laboratory of Malignant Tumor Epigenetics and Gene Regulation, Sun Yat-Sen Memorial Hospital, Sun Yat-Sen University, Guangzhou, China

    Jiahua Wen, Kai Wen, Meng Tao, Zhenyu Zhou, Xing He, Weidong Wang, Zian Huang, Qiaohong Lin, Huoming Li, Haohan Liu, Yongcong Yan & Zhiyu Xiao

  3. Department of Biliary and Pancreatic Surgery, Sun Yat-Sen Memorial Hospital, Sun Yat-Sen University, Guangzhou, 510120, China

    Xing He

Authors
  1. Jiahua Wen
    View author publications

    You can also search for this author inPubMed Google Scholar

  2. Kai Wen
    View author publications

    You can also search for this author inPubMed Google Scholar

  3. Meng Tao
    View author publications

    You can also search for this author inPubMed Google Scholar

  4. Zhenyu Zhou
    View author publications

    You can also search for this author inPubMed Google Scholar

  5. Xing He
    View author publications

    You can also search for this author inPubMed Google Scholar

  6. Weidong Wang
    View author publications

    You can also search for this author inPubMed Google Scholar

  7. Zian Huang
    View author publications

    You can also search for this author inPubMed Google Scholar

  8. Qiaohong Lin
    View author publications

    You can also search for this author inPubMed Google Scholar

  9. Huoming Li
    View author publications

    You can also search for this author inPubMed Google Scholar

  10. Haohan Liu
    View author publications

    You can also search for this author inPubMed Google Scholar

  11. Yongcong Yan
    View author publications

    You can also search for this author inPubMed Google Scholar

  12. Zhiyu Xiao
    View author publications

    You can also search for this author inPubMed Google Scholar

Contributions

ZX, YY and HL conceived and designed this study. JW, KW and MT collected and analyzed the relative data. ZZ wrote the paper. XH, WW and ZH conducted the experiments. QL and HL revised the manuscript. All authors read and approved the final manuscript.

Corresponding authors

Correspondence to Haohan Liu, Yongcong Yan or Zhiyu Xiao.

Ethics declarations

Ethics approval and consent to participate

The research involving human subjects underwent review and approval by Sun Yat-Sen Memorial Hospital (Ethical number: SYSKY-2023-646-01). The written informed consent was obtained from the patients prior to their involvement in the study. The animal experiment was granted approval by the Animal Ethics Committee of Sun Yat-sen University (Ethical number: SYSU-IACUC-2024-002173).

Consent for publication

All authors consented for publication.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Supplementary Material 1

Supplementary Material 2

Supplementary Material 3

12935_2025_3743_MOESM4_ESM.docx

Supplementary Material 4: Supplementary Fig. 1 (A-H) A The consensus score matrix of all samples of HCC patients in TCGA database. (I-K) Time-dependent ROC curves for the nomogram based on IEPS score and TNM stage in the ICGC cohort (for predicting 1, 2, and 3-year OS). (L-M) DCA plots were conducted to determine the clinical utility of the nomogram for assessing the clinical benefits for patients in the TCGA and ICGC cohorts. Supplementary Fig. 2. (A) qRT-PCR detected the over-expression efficiency of PIGU in Huh7 and SNU-449 cell lines. (B) qRT-PCR detected the changes in the expression levels of TGF-β1 before and after PIGU over-expressed in Huh7 and SNU-449 cell lines. (C-E) Western blot analysis of the changes in the expression levels of TGF-β1 before and after PIGU over-expressed in Huh7 and SNU-449 cell lines. (F) qRT-PCR detected the over-expression efficiency of PIGU in Hep 3B2.1-7 and MHCC97-H cell lines. (G) qRT-PCR detected the changes in the expression levels of TGF-β1 before and after PIGU knockdown in Hep 3B2.1-7 and MHCC97-H cell lines. (H-J) Western blot analysis of the changes in the expression levels of TGF-β1 before and after PIGU knockdown in Hep 3B2.1-7 and MHCC97-H cell lines. (K) ELISA detected the change of the of TGF-β1 in the supernatant of Huh7 and SNU-449 cell lines before and after PIGU over-expressed. (O) ELISA detected the change of the of TGF-β1 in the supernatant of Hep 3B2.1-7 and MHCC97-H cell lines before and after PIGU knockdown. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Wen, J., Wen, K., Tao, M. et al. Integrated analysis reveals an immune evasion prognostic signature for predicting the overall survival in patients with hepatocellular carcinoma. Cancer Cell Int 25, 101 (2025). https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s12935-025-03743-9

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s12935-025-03743-9

Keywords

Cancer Cell International

ISSN: 1475-2867

Contact us