Analysis of the Prognosis Prediction Ability of a Necroptosis-Related Gene Signature and its Relationship With the Hepatocellular Carcinoma Immune Microenvironment Using Bioinformatics Analysis and Experimental Validation

Abstract

Background: Hepatocellular carcinoma (HCC) is one of the most malignant cancers and has a poor prognosis. The immune microenvironment is closely related to the drug sensitivity of a tumor. Necroptosis was reported to be a key factor for HCC. The prognostic value of necroptosis-related genes and their association with the tumor immune microenvironment are still unknown. Methods: Necroptosis-related genes that could comprise a signature for predicting the prognosis of HCC cases were identified using univariate analysis and least absolute shrinkage and selection operator Cox regression analysis. The association between this prognosis prediction signature and HCC immune microenvironment was analyzed. The immunological activities and drug sensitivities were compared between different risk score groups identified using the prognosis prediction signature. The expression levels of the five genes comprising the signature were validated using RT-qPCR. Results: A prognosis prediction signature consisting of five necroptosis-related genes was constructed and validated. Its risk score was = (0.1634 × PGAM5 expression) + (0.0134 × CXCL1 expression) − (0.1007 × ALDH2 expression) + (0.2351 × EZH2 expression) − (0.0564 × NDRG2 expression). The signature was found to be significantly associated with the infiltration of B cells, CD4⁺ T cells, neutrophils, macrophages, and myeloid dendritic cells into the HCC immune microenvironment. The number of infiltrating immune cells and the expression levels of immune checkpoints in the immune microenvironment of high-risk score patients were higher. Sorafenib and immune checkpoint blockade were determined to be ideally suited for treating high-risk score patients and low-risk score patients, respectively. Finally, RT-qPCR results confirmed that the expression levels of EZH2, NDRG2, and ALDH2 were significantly down-regulated in HuH7 and HepG2 cells compared to those in LO2 cells. Conclusion: The necroptosis-related gene signature developed herein can classify patients with HCC according to prognosis risk well and is associated with infiltration of immune cells into the tumor immune microenvironment.

Keywords

hepatocellular carcinoma necroptosis signature prognosis immune infiltration

Introduction

Hepatocellular carcinoma (HCC) is one of the most common malignant tumors in the world, ranking second in terms of cancer-related mortality and fifth in terms of incidence among all solid tumors.¹ Due to the presence of hepatitis B virus, HCC is widespread in China, accounting for about 50% of newly diagnosed cases and deaths every year.² The clinical symptoms of early-stage HCC are not often obvious; most patients have already reached the advanced stage of HCC by the time they face obvious discomfort and visit a hospital for medical consultation, and the treatments at this stage are very limited.³ The advent of immunotherapy and targeted therapy has greatly improved the prognosis of patients with HCC.^4,5 The tumor heterogeneity in HCC cases causes great difficulty in predicting prognosis. With the development of next-generation high-throughput sequencing, a large number of genes have been identified. Some of these genes can be used as biomarkers for the accurate prognosis prediction, diagnosis, and follow-up of cancers and thus have great potential as targets for novel chemotherapeutic drugs. Moreover, the immune microenvironment of tumors is closely related to their drug sensitivity. As a result, there is an urgent need to construct a novel prognostic gene signature that can predict cancer prognosis and infiltration of immune cells into the tumor immune microenvironment.

Apoptosis is a ubiquitous programmed cell death mechanism in living cells, and chemotherapeutic drugs mainly inhibit tumor growth by inducing their apoptosis.⁶ Necroptosis was discovered as an alternative mode of programmed cell death that could be used to overcome apoptosis resistance in 2005.⁷ The morphological features of necroptosis bear some resemblance to those of necrosis, including swelling and altered permeability of the cytoplasmic membrane, the opening of selective ion channels in the membrane, and, ultimately, cell disintegration and death. Necroptosis is mainly activated by receptor-interacting protein kinase 1/3 (RIPK1/3), which subsequently triggers RIPK3-mixed lineage kinase domain-like pseudokinase (MLKL).⁸ Necroptosis plays a role in the occurrence as well as development of tumor cells.⁹ A previous study indicated that shikonin, the major chemical component of purple cromwell roots, can induce the expression of RIP1 and RIP3 in osteosarcoma, promote necroptosis, and inhibit tumor metastasis.¹⁰ In pancreatic ductal adenocarcinoma, RIPK1/RIPK3 induces immunosuppression by inducing C-X-C motif chemokine ligand 1 (CXCL1) and Minle signaling to promote pancreatic cancer progression.¹¹ On the contrary, necroptosis promotes adaptive immunity by providing antigenic and inflammatory stimuli to dendritic cells, which in turn activate CD8+ T cells and an antitumor immune reaction.¹² However, the identification of necroptosis-related genes involved in HCC and the elucidation of their ability to predict immune cell infiltration into the tumor immune microenvironment have not been systematically pursued.

Here, we aimed to develop a necroptosis-related gene signature for predicting the prognosis of patients with HCC and evaluating immune cell infiltration into the HCC tumor immune microenvironment by categorizing the patients into risk score groups using this signature. In this study, we first collected necroptosis-related genes from existing literature and downloaded its matched transcriptome and clinical information from the TCGA and ICGC databases. Then, we developed and validated a necroptosis-related gene signature using univariate Cox regression analysis and the least absolute shrinkage and selection operator (LASSO) Cox regression analysis. Next, we explored the association between the necroptosis-related gene signature and the tumor immune microenvironment in HCC. We also compared immunological activities and drug sensitivities between different risk score groups to enable the development of a strategy for targeted HCC therapy.

Materials and Methods

Data Acquisition

Harmonized RNA sequencing data (HTSeq-FPKM) and matching clinical information of HCC patients were downloaded from TCGA cohort (https://portal.gdc.cancer.gov/) and Japan Project of International Cancer Genome Consortium (the Liver Cancer-RIKEN JP, https://dcc.icgc.org/). These data were further reviewed. The following were the exclusion criteria: (i) patients who have not been diagnosed with HCC and (ii) patients who have incomplete clinical data. We focused on clinicopathological characteristics, including sex, age, TNM stage, and pathological tumor grade. A total of 56% (n = 211) HCC patients did not provide information regarding hepatitis, and 42% did not provide information on cirrhosis. To include as many patients as possible into our study, we did not pay attention to the comorbidities of hepatitis and cirrhosis in this study. From existing literature,¹³ we retrieved 17 necroptosis-related genes (Supplementary Table S1). The flow chart of this study is shown in Figure 1.

Figure 1.

Flow chart of the study.

Functional Enrichment Analysis of Necroptosis-Related Genes

Metascape (http://metascape.org) is a reliable open-access tool for gene annotation and functional enrichment analysis.¹⁴ The “Express Analysis” module of Metascape was utilized to identify the functions and pathways of necroptosis-related genes.

Consensus Clustering

The TCGA cohort was analyzed through the unsupervised clustering method using the “ConsensuClusterPlus” package with a maximum cluster number of six. The HCC samples in the TCGA cohort were divided into several groups. Kaplan–Meier survival curves were used to confirm the association between the groups and overall survival (OS).

Development and Validation of a Necroptosis-Related Gene Signature for HCC Prognosis Prediction

Necroptosis-related genes identified from existing literature were used to build an HCC prognosis prediction gene signature. We first conducted log-rank tests and univariate Cox proportional hazards regression to identify prognostic necroptosis-related genes using the “survival” and “survminer” R packages (P < .05). These prognostic necroptosis-related genes were subjected to LASSO Cox regression analysis using the “glmnet” R package. Then, we calculated the risk score for each patient in the TCGA cohort using the following formula: Risk score = Coef₁ × gene₁ expression + Coef₂ × gene₂ expression+ …… + Coef _n × gene _n expression. According to the median risk score, the TCGA cohort was divided into low-risk score and high-risk score groups. Survival analysis was then performed using the “survival” R package. Next, the prognostic prediction ability of this necroptosis-related gene signature was evaluated by constructing a time-dependent receiver operating characteristic (time-dependent ROC) curve using the “timeROC” R package. We also investigated associations between the risk score groups and clinicopathological characteristics of the patients. Next, univariate and multivariate Cox regression analyses were performed to confirm the role of our prognosis prediction signature as an independent prognostic factor for HCC. Finally, our necroptosis-related gene signature constructed using the TCGA cohort was validated using the ICGC cohort.

Gene Set Enrichment Analysis (GSEA)

GSEA is often used for the analysis and interpretation of genome-wide expression profiles.¹⁵ We used the GSEA software to identify signaling pathways in which genes of the newly established necroptosis-related gene signature were enriched. Gene sets with a P < .05, normalized enrichment score (NES)| > 1, and false discovery rate (FDR) < 0.25 were regarded to be notably enriched.

Association Between the Necroptosis-Related Gene Signature and the Tumor Immune Microenvironment

We first analyzed immune infiltration in the HCC tumor immune microenvironment of the TCGA cohort using immune cell subtype signatures downloaded from CIBERSORTx (https://cibersortx.stanford.edu/index.php).¹⁶ Next, we investigated associations between the necroptosis-related gene signature and typical infiltrating immune cells, including B cells, CD4⁺ T cells, CD8⁺ T cells, neutrophils, macrophages, and myeloid dendritic cells. Finally, we investigated correlations between the expression of genes constituting the necroptosis-related gene signature and genetic markers of infiltrating immune cells, which were collected from existing literature. The results are displayed as Spearman's correlation coefficients and the estimated statistical significances.

Comparison of Immunological Activities and Drug Sensitivities Between the High- and Low-Risk Score Groups

First, we compared the infiltration levels of various immune cell subtypes between the high- and low-risk score groups. Moreover, we compared the gene expression levels of immune checkpoint genes (LAG3, SIGLEC15, CTLA4, HAVCR2, PDCD1LG2, PDCD1, and TIGIT) between the high- and low-risk score groups.

Estimation of stromal and immune cells in malignant tumor tissues using expression data (ESTIMATE) is a tool for calculating the abundance of infiltrating stromal and immune cells into the tumor immune microenvironment; the results of these analyses are shown as stromal score and immune score, respectively.¹⁷ ESTIMATE scores were used to compare tumor purity between high-risk and low-risk groups.

Based on the two primary mechanisms of immune evasion, the tumor immune dysfunction and exclusion (TIDE, http://tide.dfci.harvard.edu/) score can predict cancer immunotherapy response; there is a negative correlation between the TIDE score and the response rate for immune checkpoint blockade.¹⁸ TIDE scores were compared between the high-risk and low-risk groups. Sorafenib is recommended as the first-line treatment for patients with advanced HCC. Thus, we predicted sorafenib sensitivity using the Genomics of Drug Sensitivity in Cancer (GDSC, https://www.cancerrxgene.org/) tool.¹⁹ Half-maximal inhibitory concentration (IC50) is the concentration of a drug required for 50% inhibition of its target cells. IC50 was used to evaluate chemotherapeutic response using the “pRRophetic” package. Sorafenib sensitivity was compared between the high- and low-risk score groups.

Construction of a Nomogram

To predict the 3-year and 5-year OS probabilities, we constructed a nomogram integrating the necroptosis-related gene signature with clinicopathological characteristics using the “rms” R package. The calibration plot was used to assess the consistency between the actual and predicted OS outcomes.

Cell Culture

The human HCC cell lines (HepG2, HuH7) and normal human liver cell line (LO2) were cultured in Dulbecco's Modified Eagle's Medium (DMEM, Gibco Invitrogen, catalog number C11960500CP, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (Gibco, catalog number 10099141, Australia), 10 000 units/mL penicillin, and 10 000 μg/mL streptomycin (Gibco Invitrogen, catalog number 15140122, Carlsbad, CA, USA). All the cell lines were purchased from the Cell Lines Service (Cellcook Biotech Co., Ltd, Guangzhou, China).

RNA Extraction, Reverse Transcription, and Quantitative Reverse Transcription PCR

Total RNA was isolated from cells using the RNA quick purification kit (ESscience, catalog number RN001, Shanghai, China) and then reverse transcribed to cDNA using the Fast All-in-One RT Kit (ESscience, catalog number RT001, Shanghai, China) for reverse transcription PCR (Bio-Rad T100); the cycling condition was 42 °C for 15 minutes. The 2× Super SYBR Green qPCR Master Mix (ESscience, catalog number QP002, Shanghai, China) was used for RT-qPCR using the Roche LightCycler 480 type II (Roche, Germany), according to the manufacturer's instructions; the cycling conditions were as follows: (a) preheating for 1 cycle at 95 °C for 10 minutes (b) amplification for 45 cycles at 95 °C for 30 seconds and 60 °C for 10 seconds; (c) 1 cycle at 72 °C for 30 seconds, 95 °C for 30 seconds, and 60 °C for 30 seconds; (c) cooling at 4 °C forever. The qPCR primer sequences are listed in Supplementary Table S2. Each sample was run in triplicate, and the relative expression levels of the five necroptosis-related genes constituting the prognosis prediction signature were compared with that of GAPDH. The 2^−△△ct method was used to calculate fold changes.

Statistical Analysis

The association between the necroptosis-related gene signature and clinicopathological characteristics of the patients was assessed using the χ² test or Fisher’s exact test. Correlation analysis and comparison of immune activity and drug sensitivity were performed using Spearman's correlation analysis and Wilcox test, respectively. The R software (version 4.1.0) and SPSS software (version 20) were used for all statistical analyses. Unless stated otherwise, a two-sided P < .05 was considered statistically significant.

Results

Baseline Characteristics

After excluding patients without complete clinical information, 338 patients from the TCGA cohort and 243 patients from the ICGC cohort were enrolled in the present study. The TCGA cohort and the ICGC cohort were regarded as the training cohort and validation cohort, respectively. The clinicopathological characteristics for the TCGA and ICGC cohorts are shown in Table 1.

Table 1.

Clinicopathological Characteristics of HCC Patients in the TCGA Cohort and ICGC Cohort.

Characteristic	TCGA cohort		ICGC cohort
Characteristic	Group	N	Group	N
Age (years)	≤60	167(49.4%)	≤68	120(49.4%)
	>60	171(50.6%)	>68	123(50.6%)
Gender	Male	231(68.3%)	Male	182(74.9%)
	Female	107(31.7%)	Female	61(25.1%)
Fustat	Alive	224(66.3%)	Alive	199(81.9%)
	Deceased	114(33.7%)	Deceased	44(18.1%)
TNM Stage	Stage I	168(49.7%)	Stage I	37(15.2%)
	Stage II	83(24.5%)	Stage II	109(44.9%)
	Stage III	83(24.5%)	Stage III	75(30.9%)
	Stage IV	4(1.3%)	Stage IV	22(9%)
Grade	G1	45(13.3%)
	G2	166(49.1%)
	G3	115(34.1%)
	G4	12(3.5%)
T stage	T1	170(50.3%)
	T2	84(24.9%)
	T3	74(21.9%)
	T4	10(2.9%)

Functional Enrichment Analysis and Consensus Clustering

We used the Metascape database for the functional enrichment analysis of necroptosis-related genes. The results indicated that the genes were significantly enriched in the necroptosis, TNF signaling pathway, alcoholic liver disease, lipid and atherosclerosis, Kaposi sarcoma-associated herpesvirus infection, and cytosolic DNA-sensing pathway functions (Figure 2A and B).

Figure 2.

Functional enrichment analysis of necroptosis-related genes using Metascape. (A) Bar graph showing GO and KEGG analysis results; (B) network graph showing GO and KEGG analysis results. Abbreviations: GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes.

To specifically investigate the potential functions of these necroptosis-related genes in HCC, we divided the TCGA cohort samples into several groups. As shown in Supplementary Figure S1, six appeared to be the optimal number of clusters (k) based on the expression similarities between necroptosis-related genes. The groups were thus named from C1 to C6. Our results showed a difference in overall survival values among the groups (P < .05).

Necroptosis-Related Prognostic Gene Signature Construction

Univariate Cox regression analysis was conducted to identify prognostic necroptosis-related genes in the training cohort. Seven candidates were identified (Table 2). These candidates were subjected to LASSO Cox regression analysis and used to construct a necroptosis-related gene signature for predicting OS in patients with HCC (Figure 3A and B). The results suggested that five of the genes were correlated with OS; phosphoglycerate mutase family member 5 (PGAM5), CXCL1, and enhancer of zeste 2 polycomb repressive complex 2 (EZH2) were found to have positive correlation coefficients, while aldehyde dehydrogenase 2 family (ALDH2) and NDRG family member 2 (NDRG2) were found to have negative correlation coefficients. The prognostic risk score for an HCC patient was thus calculated as follows: Risk score = (0.1634) × PGAM5 expression + (0.0134) × CXCL1 expression + (−0.1007) × ALDH2 expression + (0.2351) × EZH2 expression + (−0.0564) × NDRG2 expression. Furthermore, HCC patients from the training cohort and the validation cohort were divided into a high-risk score group and a low-risk score group, with the mean risk score being used as the cut-off value. As shown in Figure 3C to F, Kaplan–Meier curve analysis revealed that, compared with patients in the low-risk score group, those in the high-risk score group showed a worse prognosis in both cohorts (P < .05). Finally, the area under the curve (AUC) of the time-dependent ROC plot was used to evaluate the predictive performance of the gene signature. Notably, the AUC values of the necroptosis-related gene signature for 1-, 3-, and 5-year survivals were 0.67, 0.66, and 0.71, respectively, in the training cohort (Figure 3G). The AUC values for the 1-, 3-, and 4-year survivals were 0.76, 0.72, and 0.78, respectively, in the validation cohort (Figure 3H). Collectively, these results confirmed that our necroptosis-related gene signature had a good prognosis prediction ability.

Figure 3.

Construction of a prognosis prediction signature including five necroptosis-related genes using the training cohort (TCGA cohort) and its validation using the validation cohort (ICGC cohort). Cross-validation was performed to find the optimal lambda value in the LASSO Cox regression analysis (A). LASSO Cox regression analysis was performed to select radiomic features for constructing the prognosis prediction signature for HCC patients. Feature coefficients were plotted against the shrinkage parameter (lambda) (B). Risk score analysis of the five gene-based prognosis prediction signature. Risk score distribution (top), survival overview (middle), and heatmap (bottom) for patients in the high- and low-risk score groups in the TCGA cohort (C) and ICGC cohort (D) Kaplan–Meier estimates of patient OS obtained using the prognosis prediction signature in the TCGA cohort (E) and ICGC cohort (F) Time-dependent ROC curve for survival analysis of the high- and low-risk score groups in the TCGA cohort (G) and ICGC cohort (H). Different colors represent different years. Abbreviations: TCGA, The Cancer Genome Atlas; ICGC, International Cancer Genomics Consortium; HCC, hepatocellular carcinoma; LASSO, least absolute shrinkage and selection operator; OS, overall survival; ROC, receiver operating characteristic.

Table 2.

Univariate Cox Regression Analysis of Necroptosis-Related Genes in the TCGA Cohort.

Gene	Hazard ratio	95% CI	P-value
RIPK1	1.4427	(1.0163,2.048)	.0403
RIPK3	1.0745	(0.7598,1.5195)	.6846
MLKL	1.2016	(0.8511,1.6964)	.2966
TLR2	1.1723	(0.8292,1.6575)	.3682
TLR3	0.8192	(0.5801,1.1568)	.2573
TLR4	1.0055	(0.712,1.4201)	.9752
TNFRSF1A	1.2499	(0.8839,1.7675)	.2070
PGAM5	1.7402	(1.2265,2.4692)	.0019
ZBP1	0.766	(0.5417,1.0832)	.1316
NR2C2	1.2465	(0.8815,1.7625)	.2125
HMGB1	1.0605	(0.7506,1.4984)	.7390
CXCL1	1.7454	(1.2231,2.4907)	.0021
USP22	1.1865	(0.8401,1.6756)	.3316
TRAF2	1.4485	(1.0242,2.0487)	.0362
ALDH2	0.6974	(0.4928,0.9869)	.0419
EZH2	2.0289	(1.4229,2.893)	.0001
NDRG2	0.6474	0.4573,0.9165)	.0142

Correlation of Risk Score with Clinicopathological Characteristics and Independent Prognostic Prediction Analysis

We first investigated the association between risk score groups and clinicopathological characteristics. Surprisingly, the risk score was significantly associated with tumor grade in the TCGA cohort and with gender and TNM stage in the ICGC cohort (Table 3). The clinicopathological characteristics of all patients are presented as a heat map (Figure 4A and B). Univariate and multivariate Cox regression analyses were applied to evaluate whether the prognosis prediction signature developed by us can predict prognosis independent of other clinicopathological characteristics. Results confirmed that the risk score derived from the necroptosis-related gene signature was an independent prognostic predictor in the TCGA cohort, and gender, TNM stage, and risk score were independent prognostic predictors in the ICGC cohort (P < .05) (Figure 5A–D).

Figure 4.

Heatmap of the prognosis prediction signature and clinicopathological characteristics of HCC patients in the TCGA cohort (A) and ICGC cohort (B). Abbreviations: TCGA, The Cancer Genome Atlas; ICGC, International Cancer Genomics Consortium; HCC, hepatocellular carcinoma.

Figure 5.

Risk score was found to be an independent prognostic factor for HCC patients. Univariate and multivariate Cox regression analyses were conducted. Forest plot for the univariate (A) and multivariate Cox regression (B) analyses of patient risk scores in the TCGA cohort. Forest plot for univariate (C) and multivariate Cox regression (D) analyses of patient risk scores in the ICGC cohort. Abbreviations: TCGA, The Cancer Genome Atlas; ICGC, International Cancer Genomics Consortium; HCC, hepatocellular carcinoma.

Table 3.

The Correlation Between Risk Score and Clinicopathological Characteristics of TCGA Cohort and ICGC Cohort.

	Parameters	Risk score		χ ²	P
	Parameters	Low	High	χ ²	P
TCGA cohort	Age			3.42	.06^a
	≤60	75(44.91%)	92(55.09%)
	>60	94(54.97%)	77(45.03%)
	Gender			2.8827	.079^a
	Male	123(52.25%)	108(46.75%)
	Female	46(42.99%)	61(57.01%)
	Grade			21.203	<.0001^a
	G1-G2	126(59.72%)	85(40.28%)
	G3-G4	43(33.86%)	84(66.14%)
	T stage			1.0139	.314^a
	T1-T2	131(51.57%)	123(48.43%)
	T3-T4	38(45.24%)	46(54.76%)
	TNM stage			1.8729	.171^a
	I-II	131(52.19%)	120(47.81%)
	III-IV	38(43.68%)	49(56.32%)
ICGC cohort	Age			0.6943	.405^a
	≤68	57(47.5%)	63(52.5%)
	>68	65(52.85%)	68(47.15%)
	Gender			5.0911	.024^a
	Male	99(54.4%)	83(45.6%)
	Female	23(37.7%)	38(62.3%)
	TNM stage			4.069	.0437^a
	I-II	81(55.48%)	65(44.52%)
	III-IV	41(42.27%)	56(57.73%)

Pearson Chi-squared test.

GSEA Analysis of Biological Function Comparing Risk Score Groups

Biological functions were compared between high- and low-risk score groups using GSEA. As shown in Supplementary Figure S2, cell cycle, E2F targets, G2M checkpoint, and mitotic spindle were found to be significantly enriched in low-risk score groups. Collectively, the results suggested that our necroptosis-related gene signature might be associated with the growth of HCC tumors.

Correlation Analysis of the Necroptosis-Related Gene Signature and Infiltration of Immune Cells into the Tumor Immune Microenvironment

The tumor immune microenvironment plays a critical role in tumor growth and development. Therefore, we further analyzed immune infiltration into the HCC tumor microenvironment in the TCGA cohort. The details of this immune infiltration are presented in Figure 6A. Next, we investigated the associations between the necroptosis-related gene signature and six typical infiltrating immune cells. In particular, the signature was significantly associated with the infiltration of B cells (r = 0.39, P = 6.53e-15), CD4⁺T cells (r = 0.41, P = 2.17e-16), neutrophils (r = 0.41, P = 8.92e-17), macrophages (r = 0.36, P = 5.61e-13), and myeloid dendritic cells (r = 0.43, P = 2.15e-18) (Figure 6B–G). Ultimately, we explored the relationship between the necroptosis-related gene signature and immune marker sets (Figure 7). Interestingly, PGAM5, CXCL1, and EZH2 showed significantly positive correlation with markers of monocytes, tumor-associated macrophages (TAMs), neutrophils, dendritic cells, Th1, Th2, and Th17, while ALDH2 showed significantly negative correlation with these markers and NDRG2 showed no correlation with most of these markers. Only EZH2 showed a significant correlation with markers of natural killer cells, such as killer cell immunoglobulin-like receptor, two domains; long cytoplasmic tail 3 (KIR2DL3) and killer cell immunoglobulin-like receptor, two domains; and long cytoplasmic tail 4 (KIR2DL4). We further analyzed the relationship between the necroptosis-related gene signature and markers of immunosuppressive cells. Interestingly, PGAM5, CXCL1, and EZH2 showed significant positive correlation with markers of M2 macrophages, regulatory T cells (Treg cells), and T cell exhaustion, such as CD163, forkhead box P3 (FOXP3), and hepatitis A virus cellular receptor 2 (HAVCR2). Collectively, these findings suggested that there is a close relationship between the necroptosis-related gene signature and immune infiltration into the HCC tumor immune microenvironment.

Figure 6.

Correlation analysis between prognosis prediction signature-based risk score and immune infiltration into the HCC tumor immune microenvironment. The infiltration levels of 22 immune cell types into the tumor immune microenvironment in the TCGA cohort (A). Each column represents a sample, and different colors represent different cells. The height of each colored column represents the abundance score of infiltrating immune cells in a sample. The box plot shows the differences in immune infiltration levels between the high- and low-risk score groups (B–G). Spearman correlation analysis between six typical infiltrating immune cells and risk score. The abscissa and ordinate in the figure represent the expression distribution of infiltrating immune cells and the risk score, respectively. B cells (B), CD4⁺ T cells (C), CD8⁺ T cells (D), neutrophils (E), macrophages (F), and myeloid dendritic cells (G). Abbreviation: TCGA, The Cancer Genome Atlas.

Figure 7.

Heatmap showing the correlations between necroptosis-related genes constituting the prognosis prediction signature and genetic markers of infiltrating immune cells in the TCGA cohort. Blue and red colors represent negative correlation and positive correlation, respectively. *P < .05, **P < .01, and ***P < .001. Abbreviation: TCGA, The Cancer Genome Atlas.

Comparison of Immunological Activities and Drug Sensitivities Between Risk Score Groups

Based on the results mentioned above, we further performed a comparison of immunological activities and drug sensitivities between the high-risk score group and low-risk score group. First, we compared the infiltration levels of 22 immune cell types between different risk score groups. Consistent with the correlation analysis results between the necroptosis-related gene signature and immune infiltration, significant differences in the infiltration levels of B cells (P < .05), T cells, CD4 cells memory activated (P < .001), Tregs (P < .001), T cells gamma delta (P < .01), NK cells resting (P < .01), monocytes (P < .001), M0 macrophages (P < .001), M2 macrophages (P < .001), dendritic cells resting (P < .001), mast cells resting (P < .001), and neutrophils (P < .05) were found between the high- and low-risk score groups (Figure 8).

Figure 8.

Comparison of 22 types of immune cells infiltrating the tumor microenvironment between risk score groups in the TCGA cohort. Box plot showing the different immune infiltration levels between the high- and low-risk score groups. Blue and red colors represent the low-risk score and high-risk score groups, respectively. *P < .05, **P < .01, and ***P < .001. Abbreviation: TCGA, The Cancer Genome Atlas.

Considering the positive correlation between the necroptosis-related gene signature and T cell exhaustion, we compared the gene expression levels of immune checkpoints between different risk score groups at the same time. As shown in Figure 9A to E, it was shocking to observe that high-risk score patients had higher expression levels of LAG3 (P < .01), CTLA4 (P < .001), HAVCR2 (P < .001), PDCD1 (P < .001), and TIGIT (P < .001).

Figure 9.

Comparison of immune checkpoint expression levels between the high- and low-risk score groups. LAG3 (A), CTLA4 (B), HAVCR2 (C), PDCD1 (D), and TIGIT (E). *P < .05, **P < .01, and ***P < .001. Abbreviations: LAG3, lymphocyte-activation gene 3; CTLA4, cytotoxic T-lymphocyte-associated protein 4; HAVCR2, hepatitis A virus cellular receptor 2; PDCD1, programmed cell death 1; TIGIT, T cell immunoreceptor with Ig and ITIM domains.

We also utilized the ESTIMATE tool to determine the abundances of stromal and immune cells infiltrating into the HCC tumor immune microenvironment and comparison them between high- and low-risk score groups. We found that high-risk score patients had higher immune scores (P < .01), which suggested that they had higher immune cell infiltration into the tumor immune microenvironment (Figure 10A). Contrarily, there was no significant difference in stromal immune scores between high- and low-risk score groups (Figure 10B).

Figure 10.

Comparison of immune infiltration levels and drug sensitivities between the high- and low-risk score groups in the TCGA cohort. Comparisons for immune scores (A), stromal scores (B), TIDE scores (C), and sorafenib IC50 values (D) are presented. *P < .05, **P < .01, and ***P < .001. Abbreviations: TCGA, The Cancer Genome Atlas; TIDE, tumor immune dysfunction and exclusion; IC50, half-maximal inhibitory concentration.

The TIDE tool can predict cancer immunotherapy response. The comparison of TIDE scores between the high- and low-risk score groups indicated that there is a high infiltration of dysfunctional and immune-excluded T cells into the HCC tumor immune microenvironment of high-risk score patients (Figure 10C). This result suggested that, compared with low-risk score patients, high-risk score patients are less sensitive to immune checkpoint blockade. We also predicted the sorafenib sensitivity of patients with HCC using GDSC and found that high-risk score patients had a lower IC50 than low-risk score patients. This result suggested that sorafenib is appropriate for treating high-risk score patients with HCC (Figure 10D). Collectively, the above results indicate that the necroptosis-related gene signature developed in the present study might be a biomarker of response to immune checkpoint blockade and sorafenib.

Constructing of a Nomogram Based on the Risk Score

Because the risk score obtained using our necroptosis-related gene signature was found to be an independent prognostic predictor, a prognostic nomogram integrating the signature with clinicopathological characteristics was further developed (Supplementary Figure S3A). Next, calibration curves for 3-year and 5-year OS were plotted and revealed that the nomogram-predicted survival outcomes closely corresponded with actual survival outcomes (Supplementary Figure S3B).

RT-qPCR Validation of the Five Genes Comprising the Necroptosis-Related Gene Signature in HCC Cell Lines

We used RT-qPCR to validate the expression of the five genes comprising the novel necroptosis-related gene signature developed in the present study in HCC cell lines. It was found that the expression levels of EZH2, NDRG2, and ALDH2 were significantly down-regulated in the HuH7 and HepG2 cell lines compared to those in the LO2 cell line (Figure 11).

Figure 11.

The mRNA expression levels of the necroptosis-related genes constituting the prognosis prediction signature in HuH7, HepG2, and LO2 cells.

Discussion

HCC is one of the most malignant digestive tumors and the second leading cause of cancer mortality in the world.¹ Many systems have been suggested for evaluating the prognosis of HCC, including BCLC and TNM staging.²⁰ However, an increasing number of studies have demonstrated that even HCC patients at the same TNM stage have different survival times.²¹ Immune infiltration into the tumor immune microenvironment is closely related to the drug response of the cancer.²² Thus, a precise prognosis prediction system is critical for deciding the treatment course of patients with HCC. Necroptosis plays an important role in the development of tumors, but whether the expression levels of necroptosis-related genes can predict patient prognosis and evaluate immune infiltration into the tumor immune microenvironment is still unclear. In the present study, we established a novel necroptosis-related gene signature to predict the prognosis of patients with HCC. Univariate and multivariate Cox regression analyses confirmed that the signature can act as an independent prognostic factor for HCC. Moreover, we found a close relationship between the signature and immune infiltration into the tumor immune microenvironment. Ultimately, we found significant differences in immune infiltration levels and drug sensitivities between risk score groups based on the prognosis prediction signature, which could provide targeted therapy and immunotherapy strategies for HCC.

Five necroptosis-related genes were included in our prognosis prediction signature, including PGAM5, CXCL1, ALDH2, EZH2, and NDRG2. Emerging data have revealed that necroptosis is a double-edged sword, as it is involved in cancer initiation as well as progression.²³ Previous studies have demonstrated that therapeutic reagents, including 5-fluorouracil, can promote necroptosis to eliminate cancer cells.²⁴ The mechanism through which 5-fluorouracil triggers necroptosis and thereby exerts anti-cancer effects has also been observed in the treatment of HCC. For example, metronomic capecitabine, a pro-drug of 5-fluorouracil was able to improve the survival of HCC patients after sorafenib failure; this may be a potential alternative plan for advanced HCC patients.^25,26 Contrarily, necroptosis may promote tumor progression by activating inflammation, inducing ROS production, and suppressing tumor immunity.^11,27,28

PGAM5 is a mitochondrial serine /threonine protein phosphatase that is involved in mitochondria homeostasis.²⁹ Yuan et al demonstrated that up-regulation of PGAM5 could induce 5-fluorouracil resistance by inhibiting HCC cell apoptosis.³⁰ Consistent with a previous study, we found that PGAM5 was significantly associated with macrophages in the present study.³¹ Although there was no significant association between PGAM5 expression and natural killer cells, Yuan et al showed that PGAM5 could activate natural killer T cells, which are involved in liver inflammation and immune responses to tumors.³²

CXCL1 is a member of the CXC chemokine family, which can recruit a variety of immune cells, especially chemotactic neutrophils and other non-hematopoietic cells, and play an important role in regulating immune and inflammatory responses. In the present study, we obtained similar results. CXCL1 was found to be associated with the infiltration of many immune cells, including TAMs, neutrophils, and dendritic cells, into the tumor immune microenvironment as well as T cell exhaustion. CXCL1 is widely recognized for promoting immune suppression in many cancers through the recruitment of TAMs,^33,34 cancer-associated fibroblasts,³⁵ and tumor-associated neutrophils.³⁶ Moreover, a study revealed that CXCL1 could influence the efficacy of chemotherapies against breast cancer.³⁷

ALDH2 is a subtype of mitochondrial ALDH and plays a vital role in the detoxification of ethanol.³⁸ Abnormal ethanol metabolism is an important factor affecting HCC occurrence. It is believed that, after prolonged exposure to alcohol, ALDH2-deficient hepatocytes produce large amounts of harmful, oxidized mitochondrial DNA via extracellular vesicles that can be transported to neighboring HCC cells and, subsequently, activate multiple oncogenic pathways.³⁹ ALDH2 indirectly regulates the immune system through its role in aldehyde metabolism and acetaldehyde adduct formation. In the present study, we found a significant negative correlation between ALDH2 expression and the infiltration of immune cells, including T cells and B cells, into the tumor immune microenvironment. Zhang et al found that the down-regulation of ALDH2 increased CD3⁺ and CD8⁺ T lymphocyte infiltration into the tumor immune microenvironment to inhibit tumor growth and progression and that there was a significant synergistic therapeutic effect of ALDH2 inhibition and PD-1 blockade in a mouse colorectal cancer model.⁴⁰ A recent study demonstrated that ALDH2 can promote immune escape by stabilizing the PD-L1 protein via direct interaction with the intracellular segment of PD-L1 and inhibition of its degradation in alcohol-induced cancer.⁴¹

EZH2 is an important methyltransferase that acts as a core catalytic subunit of polycomb repressive complex 2 (PRC2) to promote tumor proliferation and metastasis by regulating transcriptional activity and drug resistance.⁴² EZH2 was also found to be associated with sorafenib resistance in our present study. Interestingly, we found that our necroptosis-related gene signature-based risk score could predict the sensitivity of HCC tumors to sorafenib. Studies have shown that EZH2 overexpression can lead to sorafenib resistance in HCC, and the specific mechanism underlying this effect may involve the regulation of the expression of insulin-like growth factor-1 receptor, which lies downstream to EZH2.⁴³ Wu et al showed that EZH2 could lead to tamoxifen resistance by silencing growth regulation by estrogen in breast cancer 1 (GREB1).⁴⁴ In the present study, there was a strong correlation between EZH2 expression and the infiltration of most immune cells into the tumor immune microenvironment. TAMs can be divided into two classes: type 1 macrophages (M1 macrophages) and type 2 macrophages (M2 macrophages). M2 macrophages are activated by Th2 cytokines, which play a critical role in pro-tumor activities, such as anti-inflammatory, tissue remodeling, tumor proliferation, invasion, and metastasis activities.⁴⁵ EZH2 expressed in glioblastoma multiforme cells was found to suppress cancer immunity, and the inhibition of EZH2 expression was found to shift the TAM population from M1 to M2 phenotype.⁴⁶ EZH2 expression in immune cells plays a dual role in anti-cancer immunity. EZH2 is necessary for dendritic cell activation and mediates epigenetic modification in allergen immunotherapy.⁴⁷ Moreover, EZH2 expression in Tregs was found to suppress anti-cancer immunity.⁴⁸ One study demonstrated that the inactivation of EZH2 could upregulate the differentiation of Th1 and Th2 cells.⁴⁹ Another recent study found that inhibition of EZH2 in prostate cancer cells inhibited endogenous dsRNA formation in cells, resulting in enhanced STING-interferon-stimulated gene responses, immune cell activation, and sensitivity to PD-1 blockade.⁵⁰

NDRG2 is a member of the N-myc downstream-regulated gene family and is related to tumor development.⁵¹ Kang et al found that NDRG2 inhibition promoted angiogenesis in HCC via vascular endothelial growth factor A.⁵² In the present study, we found a negative correlation between NDRG2 expression and PDCD1 expression. Lee et al found that overexpression of NDRG2 in breast cancer cells inhibited PD-L1 expression and restored the T cell proliferation ability inhibited by PD-L1.⁵³

Another important aspect of the present study is the comparison of immunological activities between different prognosis prediction signature-based risk score groups. We found that the high-risk score group had increased infiltration of immune cells, particularly LAG3, CTLA4, HAVCR2, PDCD1, and TIGIT, into the tumor immune microenvironment as well as higher T cell exhaustion. Thus, a high expression of immune checkpoint genes means that the immunological activity in HCC patients with a high-risk score was suppressed. There have been several advancements in cancer immunotherapy since the development of the first generation immunosuppressant, Ipilimumab, an anti-cytotoxic T lymphocyte-associated protein 4 (CTLA-4) antibody that was approved in 2011.⁵⁴ Recently, atezolizumab plus bevacizumab combination therapy was developed as a first-line therapy for unresectable HCC tumors.⁵⁵ The availability of biomarkers for predicting cancer immunotherapy response is limited. In our present study, we found that high-risk score patients were less sensitive to immune checkpoint blockade than low-risk score patients. Overall, our findings suggest that the risk score based on our prognosis prediction signature could provide doctors with a clinical strategy for targeted therapy and immunotherapy against HCC.

We must acknowledge potential limitations of our analyses. First, although our prognosis prediction signature was validated using an independent ICGC cohort, its predictive probability still needs to be confirmed. Second, the mechanism underlying the association between our prognosis prediction signature and immune infiltration into the HCC tumor immune microenvironment needs to be further explored. Finally, the ability of our prognostic prediction signature to predict the sensitivity of HCC tumors to sorafenib and immune checkpoint blockade should be validated using large clinical trials.

Conclusion

Through the systematic analysis of necroptosis-related genes, we established and validated a novel five-gene prognosis prediction signature for classifying patients with HCC into high- and low-risk score groups. Moreover, we found a strong correlation between the prognosis prediction signature-based risk score and infiltration of immune cells into the HCC tumor immune microenvironment. Finally, our findings revealed that our prognosis prediction signature could predict the response of patients with HCC to sorafenib and immune checkpoint blockade, providing a foundation for the development of targeted therapies and immunotherapies.

Supplemental Material

sj-docx-1-tct-10.1177_15330338231182208 - Supplemental material for Analysis of the Prognosis Prediction Ability of a Necroptosis-Related Gene Signature and its Relationship With the Hepatocellular Carcinoma Immune Microenvironment Using Bioinformatics Analysis and Experimental Validation

Supplemental material, sj-docx-1-tct-10.1177_15330338231182208 for Analysis of the Prognosis Prediction Ability of a Necroptosis-Related Gene Signature and its Relationship With the Hepatocellular Carcinoma Immune Microenvironment Using Bioinformatics Analysis and Experimental Validation by Renguo Guan, Jie Mei and Rongping Guo in Technology in Cancer Research & Treatment

Footnotes

Abbreviations:

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Ethical Approval

Ethical Approval is not applicable for this article. Because all the liver sample information was downloaded from public databases, including The Cancer Genome Atlas (TCGA) and International Cancer Genomics Consortium (ICGC).

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the National Natural Science Foundation of China (No. 82172579); Clinical Trials Project (5010 Project) of Sun Yat-sen University (No. 5010-2017009).

Informed Consent

Not applicable, because there are no human subjects in this article and informed consent is not applicable.

Human and Animal Rights

Not applicable, because this article does not contain any studies with human or animal subjects.

ORCID iD

Renguo Guan

Supplemental Material

Supplemental material for this article is available online.

References

Siegel

Miller

Fuchs

Jemal

. Cancer statistics, 2021. CA Cancer J Clin. 2021;71(1):7-33.

Feng

Zong

Cao

. Current cancer situation in China: good or bad news from the 2018 Global Cancer Statistics? Cancer Commun. 2019;39(1):22.

Llovet

Montal

Sia

Finn

. Molecular therapies and precision medicine for hepatocellular carcinoma. Nat Rev Clin Oncol. 2018;15(10):599-616.

Huang

Yang

Chung

Dennison

Zhou

. Targeted therapy for hepatocellular carcinoma. Signal Transduct Target Ther. 2020;5(1):146.

Sangro

Sarobe

Hervas-Stubbs

Melero

. Advances in immunotherapy for hepatocellular carcinoma. Nat Rev Gastroenterol Hepatol. 2021;18(8):525-543.

Carneiro

El-Deiry

. Targeting apoptosis in cancer therapy. Nat Rev Clin Oncol. 2020;17(7):395-417.

Degterev

Huang

Boyce

, et al. Chemical inhibitor of nonapoptotic cell death with therapeutic potential for ischemic brain injury. Nat Chem Biol. 2005;1(2):112-119.

Declercq

Vanden Berghe

Vandenabeele

. RIP kinases at the crossroads of cell death and survival. Cell. 2009;138(2):229-232.

Gong

Fan

Luo

, et al. The role of necroptosis in cancer biology and therapy. Mol Cancer. 2019;18(1):100.

10.

Deng

Liao

, et al. The anti-tumor effect of shikonin on osteosarcoma by inducing RIP1 and RIP3 dependent necroptosis. BMC Cancer. 2013;13(1):580.

11.

Seifert

Werba

Tiwari

, et al. The necrosome promotes pancreatic oncogenesis via CXCL1 and Mincle-induced immune suppression. Nature. 2016;532(7598):245-249.

12.

Yatim

Jusforgues-Saklani

Orozco

, et al. RIPK1 and NF-kappaB signaling in dying cells determines cross-priming of CD8(+) T cells. Science. 2015;350(6258):328-334.

13.

Wang

Liu

. Identification and validation a necroptosis-related prognostic signature and associated regulatory axis in stomach adenocarcinoma. Onco Targets Ther. 2021;14:5373-5383.

14.

Zhou

Pache

, et al. Metascape provides a biologist-oriented resource for the analysis of systems-level datasets. Nat Commun. 2019;10(1):1523.

15.

Subramanian

Tamayo

Mootha

, et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A. 2005;102(43):15545-15550.

16.

Steen

Liu

Alizadeh

Newman

. Profiling cell type abundance and expression in bulk tissues with CIBERSORTx. Methods Mol Biol. 2020;2117:135-157.

17.

Yoshihara

Shahmoradgoli

Martinez

, et al. Inferring tumour purity and stromal and immune cell admixture from expression data. Nat Commun. 2013;4(1):2612.

18.

Jiang

Pan

, et al. Signatures of T cell dysfunction and exclusion predict cancer immunotherapy response. Nat Med. 2018;24(10):1550-1558.

19.

Yang

Soares

Greninger

, et al. Genomics of Drug Sensitivity In Cancer (GDSC): a resource for therapeutic biomarker discovery in cancer cells. Nucleic Acids Res. 2013;41(D1):D955-D961.

20.

Chen

Chang

. Survival predictability between the American Joint Committee on cancer 8th edition staging system and the Barcelona clinic liver cancer classification in patients with hepatocellular carcinoma. Oncologist. 2021;26(3):e445-e453.

21.

Wang

Zhong

, et al. A modified staging of early and intermediate hepatocellular carcinoma based on single tumour >7 cm and multiple tumours beyond up-to-seven criteria. Aliment Pharmacol Ther. 2019;49(2):202-210.

22.

Petitprez

Meylan

de Reynies

Sautes-Fridman

Fridman

. The tumor microenvironment in the response to immune checkpoint blockade therapies. Front Immunol. 2020;11:784.

23.

Yan

Wan

Choksi

Liu

. Necroptosis and tumor progression. Trends Cancer. 2022;8(1):21-27.

24.

Wang

Jin

Yang

, et al.

Necroptosis in cancer: an angel or a demon?

Tumour Biol. 2017;39(6):1010428317711539.

25.

Marinelli

Granito

Piscaglia

Renzulli

Stagni

Bolondi

. Metronomic capecitabine in patients with hepatocellular carcinoma unresponsive to or ineligible for sorafenib treatment: report of two cases. Hepat Mon. 2013;13(9):e11721.

26.

Granito

Marinelli

Terzi

, et al. Metronomic capecitabine as second-line treatment in hepatocellular carcinoma after sorafenib failure. Dig Liver Dis. 2015;47(6):518-522.

27.

Deng

, et al. Neoalbaconol induces cell death through necroptosis by regulating RIPK-dependent autocrine TNFalpha and ROS production. Oncotarget. 2015;6(4):1995-2008.

28.

Karin

. Nuclear factor-kappaB in cancer development and progression. Nature. 2006;441(7092):431-436.

29.

Kanamaru

Sekine

Ichijo

Takeda

. The phosphorylation-dependent regulation of mitochondrial proteins in stress responses. J Signal Transduct. 2012;2012:931215.

30.

Cheng

Qian

Ding

, et al. High PGAM5 expression induces chemoresistance by enhancing Bcl-xL-mediated anti-apoptotic signaling and predicts poor prognosis in hepatocellular carcinoma patients. Cell Death Dis. 2018;9(10):991.

31.

Ng Kee Kwong

Nicholson

Pavlidis

Adcock

Chung

. PGAM5 Expression and macrophage signatures in non-small cell lung cancer associated with chronic obstructive pulmonary disease (COPD). BMC Cancer. 2018;18(1):1238.

32.

Kang

Bang

Han

, et al. Regulation of NKT cell-mediated immune responses to tumours and liver inflammation by mitochondrial PGAM5-Drp1 signalling. Nat Commun. 2015;6(1):8371.

33.

Kemp

Carpenter

Steele

, et al. Apolipoprotein E promotes immune suppression in pancreatic cancer through NF-kappaB-mediated production of CXCL1. Cancer Res. 2021;81(16):4305-4318.

34.

Wang

Sun

Wei

Cen

DuBois

. CXCL1 is critical for premetastatic niche formation and metastasis in colorectal cancer. Cancer Res. 2017;77(13):3655-3665.

35.

Miyake

Hori

Morizawa

, et al. CXCL1-mediated Interaction of cancer cells with tumor-associated macrophages and cancer-associated fibroblasts promotes tumor progression in human bladder cancer. Neoplasia. 2016;18(10):636-646.

36.

Yuan

Zhu

Zheng

Cao

Liu

. Tumor-derived CXCL1 promotes lung cancer growth via recruitment of tumor-associated neutrophils. J Immunol Res. 2016;2016:6530410.

37.

Acharyya

Oskarsson

Vanharanta

, et al. A CXCL1 paracrine network links cancer chemoresistance and metastasis. Cell. 2012;150(1):165-178.

38.

Yang

Zhang

Ren

. ALDH2 Polymorphism and ethanol consumption: a genetic-environmental interaction in carcinogenesis. Adv Exp Med Biol. 2019;1193:229-236.

39.

Seo

Gao

, et al. ALDH2 deficiency promotes alcohol-associated liver cancer by activating oncogenic pathways via oxidized DNA-enriched extracellular vesicles. J Hepatol. 2019;71(5):1000-1011.

40.

Zhang

. The role of ALDH2 in tumorigenesis and tumor progression: targeting ALDH2 as a potential cancer treatment. Acta Pharm Sin B. 2021;11(6):1400-1411.

41.

Zhang

Xia

Wang

, et al. Aldehyde dehydrogenase 2 mediates alcohol-induced colorectal cancer immune escape through stabilizing PD-L1 expression. Adv Sci (Weinh). 2021;8(10):2003404.

42.

Eich

Athar

Ferguson

III Varambally

. EZH2-targeted therapies in cancer: hype or a reality. Cancer Res. 2020;80(24):5449-5458.

43.

Zhang

Sun

, et al. Enhancer of zeste 2 polycomb repressive complex 2 subunit promotes sorafenib resistance of hepatocellular carcinoma though insulin-like growth factor 1 receptor. Anti-Cancer Drugs. 2019;30(7):e0746.

44.

Zhang

Cenciarini

, et al. Tamoxifen resistance in breast cancer is regulated by the EZH2-ERalpha-GREB1 transcriptional axis. Cancer Res. 2018;78(3):671-684.

45.

Genard

Lucas

Michiels

. Reprogramming of tumor-associated macrophages with anticancer therapies: radiotherapy versus chemo- and immunotherapies. Front Immunol. 2017;8:828.

46.

Yin

Qiu

Huang

Peng

. EZH2 suppression in glioblastoma shifts microglia toward M1 phenotype in tumor microenvironment. J Neuroinflammation. 2017;14(1):220.

47.

Wen

, et al. Epigenetic modification of enhancer of zeste homolog 2 modulates the activation of dendritic cells in allergen immunotherapy. Int Arch Allergy Immunol. 2019;180(2):120-127.

48.

Duan

Guo

. EZH2: a novel target for cancer treatment. J Hematol Oncol. 2020;13(1):104.

49.

Tumes

Onodera

Suzuki

, et al. The polycomb protein Ezh2 regulates differentiation and plasticity of CD4(+) T helper type 1 and type 2 cells. Immunity. 2013;39(5):819-832.

50.

Morel

Sheahan

Burkhart

, et al. EZH2 inhibition activates a dsRNA-STING-interferon stress axis that potentiates response to PD-1 checkpoint blockade in prostate cancer. Nat Cancer. 2021;2(4):444-456.

51.

Kim

Lim

Kim

. N-myc Downstream-Regulated Gene 2 (NDRG2) function as a positive regulator of apoptosis: a new insight into NDRG2 as a tumor suppressor. Cells. 2021;10(10):2649.

52.

Wang

Liu

Wang

Kang

. NDRG2 inhibition facilitates angiogenesis of hepatocellular carcinoma. Open Med. 2021;16(1):742-748.

53.

Lee

Lim

Yang

. NDRG2 expression in breast cancer cells downregulates PD-L1 expression and restores T cell proliferation in tumor-coculture. Cancers (Basel). 2021;13(23):6112.

54.

Topalian

Weiner

Pardoll

. Cancer immunotherapy comes of age. J Clin Oncol. 2011;29(36):4828-4836.

55.

Finn

Qin

Ikeda

, et al. Atezolizumab plus bevacizumab in unresectable hepatocellular carcinoma. N Engl J Med. 2020;382(20):1894-1905.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.92 MB