Expression landscape of epigenetic genes in human hepatocellular carcinoma

Abstract

Hepatocellular carcinoma (HCC) is the most common primary liver tumor, often arising in the context of chronic liver disease. Despite recent advances in systemic therapies, including the use of immune checkpoint inhibitors (ICIs), clinical outcomes remain suboptimal, with many patients exhibiting primary or acquired resistance. Accumulating evidence indicates that the dysregulation of epigenetic mechanisms contributes to HCC development, and may also play a crucial role in shaping the tumor immune microenvironment, influencing responses to treatments. In this study, we analyzed the expression profiles of a comprehensive set of epigenetic regulators across publicly available transcriptomic datasets of HCC and non-tumoral liver tissues. Our findings reveal a consistent dysregulation of key epigenetic modifiers, particularly those involved in DNA methylation and histone modification. Furthermore, our analysis underscores the need for a deeper understanding of the epigenetic landscape of HCC, as specific epigenetic patterns are directly associated with disease development, the major mutational, immune, and transcriptional subclasses of HCC, and patient clinical outcomes. Our study provides a foundation for integrating epigenetic biomarkers into patient stratification and therapeutic decision-making. A more comprehensive analysis of epigenetic alterations could pave the way for novel predictive markers and combination strategies that could enhance the efficacy of ICIs in HCC.

Keywords: Epigenetics; Gene expression; Hepatocellular carcinoma; Immune landscape; Oncofetal reprogramming.

Fig. 1

Heatmap of the expression of epigenetic genes grouped in families and according to liver disease classification (normal liver, chronic hepatitis, cirrhosis, dysplastic nodules and early and advanced HCC) in the GSE114564 dataset

Fig. 2

Epigenetic genes with statistically significant changes in expression between normal tissue and HCC from the indicated RNAseq and microarray datasets. Left panel, upregulated epigenetic genes; right panel, downregulated epigenetic genes

Fig. 3

Changes in the expression levels of epigenetic genes at different stages of hepatocarcinogenesis in the different datasets indicated

Fig. 4

A. Heatmap of the expression of the 44-gene signature of HNF4α transcriptional activity in human liver tissue samples from fetal, pediatric and adult donors (GSE111845). Lower panel shows violin plots representing the ssGSEA of the HNF4α signature scores in the three different groups. p < 0.05 (*), p < 0.001 (***). B. Heatmap of the expression of epigenetic genes differentially expressed in fetal, pediatric and adult liver tissues (GSE111845) statistically significant. C. Heatmap of the expression of epigenetic genes differentially expressed in fetal, pediatric and adult liver tissues (GSE61276) statistically significant

Fig. 5

Boxplots of the EpiG-Oncofetal and HNF4α signatures analyzed using ssGSEA in normal liver tissues and tissues at different stages of hepatocarcinogenesis in the indicated genesets. Pairwise comparisons were performed using the Wilcoxon rank-sum test with Benjamini–Hochberg (BH) adjustment for multiple testing. The preceding stage before hepatocellular carcinoma (HCC) development was used as the reference group in all comparisons. Statistical significance is indicated as follows: p < 0.05 (*), p < 0.01 (**), p < 0.001 (***). Comparisons with p > 0.05 are not shown

Fig. 6

A. Heatmap identifying five subgroups (EpiG1-T – EpiG5-T) of patients based on an unsupervised hierarchical clustering analysis of the TCGA-LIHC cohort according to the expression of the EpiG signature (93 + 25 epigenetic genes), left panel. Elbow plot for optimal selection of the number of clusters, and scatterplot of unsupervised DAPC representing the centroid and ellipses of 95% confidence interval, right panels. B. Kaplan–Meier curves showing the differences in overall survival between the different EpiGT subgroups identified in the TCGA-LIHC cohort. C. Distribution of tumor samples among different tumor stages, stages I—IV (left plot), or histological grades, G1-G4 (right plot), within the different EpiGT subgroups

Fig. 7

A. Distribution of tumor samples within each EpiGT subgroup according to the HCC subclassifications of Lee (A, B), Boyault (G1-G6), Chiang, Hoshida (S1, S2) and Roessler (A, B). B. Violin plots of the EpiG-HCC and HNF4α signatures analyzed using ssGSEA in the different EpiGT subgroups

Fig. 8

A. Violin plots showing the inflamed signature. B. Violin plots showing the IFNAP signature. C. Violin plots showing the Mutated b-catenin Gene Signature (MBGS), as analyzed using ssGSEA in the different EpiGT (TCGA-LIHC) and EpiGL (LIRI-JP) subgroups. p < 0.001 (***). For comparisons not reaching statistical significance (p > 0.05), exact p-values are reported

Fig. 9

Venn diagram showing the number of EpiGs specifically upregulated in the EpiG5 T and EpiG1-L subgroups, and the number and identity of EpiGs commonly upregulated in both subgroups

Fig. 10

A. Violin plots showing the expression of five immune-related genes (CCL4, CCL5, B2M, HLA-B and HLA-C) across the five EpiG subgroups identified in the TCGA-LIHC, upper panel and LIRI-JP, lower panel, cohorts. p < 0.05 (*), p < 0.01 (**), p < 0.001 (***). For comparisons not reaching statistical significance (p > 0.05), exact p-values are reported. B. Heatmap showing the expression of the indicated immune response-related genes in the different EpiG subgroups of the TCGA-LIHC and LIRI-JP cohorts. CTNNB1 mutational status and the expression of different immune-related signatures, as well as that of the fetal EpiGs (the 125 genes that are upregulated in the fetal vs adult liver shown in Fig. 4B) are also shown. C. Venn diagrams showing the number of EpiGs that show a statistically significant inverse correlation (R < −0.25) in their expression levels with that of the immune-related genes indicated in the TCGA-LIHC and LIRI-JP cohorts. The number and identity of the EpiGs that display this negative correlation simultaneously in both cohorts are also shown