Clinicopathologic characteristics of HCC patients with cirrhosisBased on the degrees of liver fibrosis, the patients in the TCGA-LIHC cohort were divided into five groups as follows: no fibrosis, portal fibrosis, fibrous septa, nodular formation and incomplete cirrhosis, and established cirrhosis (Fig. 1A). We analyzed the relationship between the degree of liver fibrosis and various clinicopathologic features to demonstrate the correlation between liver fibrosis and HCC progression. (Table. S1). In particular, the established-cirrhosis group had the second highest proportion of HCC patients. Albumin levels were lower in the established-cirrhosis group than in the other four groups, indicating the fact that HCC patients with cirrhosis had low plasma levels (Fig. 1B). In addition, the prothrombin time was significantly shorter in the fibrous septa, nodular formation and incomplete cirrhosis, and established cirrhosis groups than in the no-fibrosis group (Fig. 1C). However, no significant differences were observed in terms of fetoprotein levels, body mass index (BMI), and vascular tumor type among the five groups (Fig. 1D–F).Fig. 1Clinical characteristics of patients with HCC with cirrhosis in the TCGA-LIHC cohort. (A) The proportion of HCC patients with different degrees of cirrhosis. (B–F) Comparison of albumin levels (B), prothrombin time (C), fetoprotein levels (D), BMI (E), and vascular tumor type (F) among HCC patients with different degrees of cirrhosis.Enrichment scores of hallmark gene sets in patients with HCC with cirrhosisTo identify cancer-related pathways enriched in different stages of cirrhosis, we calculated the enrichment scores of 50 hallmark gene sets in each patient in the TCGA-LIHC cohort using the ssGSEA algorithm (Fig. 2A). The results showed that patients in the nodular-formation-and-incomplete-cirrhosis group had the highest enrichment scores of numerous classical hallmarks of cancer, such as hypoxia, epithelial-mesenchymal transition (EMT), TGF-β signaling pathway, angiogenesis, and ROS production (Fig. 2A). Furthermore, we analyzed the relationships between the degrees of liver fibrosis and the above pathways based on the ssGSEA scores, and found that hypoxia and ROS pathways stood out (Fig. 2B–F). Liver fibrosis can cause in hypoperfusion of the hepatic lobules, closely related to decreased portal circulation, which ultimately results in intrahepatic hypoxia27. And hypoxia, which contributes to the poor prognosis of HCC patients, can increase the expression of hypoxia-inducible factors and activated downstream receptors, leading to the activation of HSCs, abnormal angiogenesis, EMT, and chronic inflammation, especially in advanced fibrosis27,28,29. Activation of HSCs plays an important role in the development of cirrhosis, and can be triggered by numerous signaling molecules, such as ROS and TGF-β30,31,32,33. In addition, TGF-β is involved in inducing EMT, which plays a crucial role in the progression of cirrhosis and metastasis in HCC patients34. Therefore, the above pathways were closely related to occurrence and progression of liver fibrosis, eventually leading to related complications including HCC.Fig. 2Enrichment levels of hallmark gene sets in patients with HCC with cirrhosis. (A) Heatmap demonstrating the enrichment levels of 50 hallmark gene sets for each cirrhosis degree. (B–F) Box plots demonstrating the enrichment levels of hypoxia (B), epithelial-to-mesenchymal transition (C), TGF-β signaling pathway (D), angiogenesis (E), and ROS (F) enrichment for each cirrhosis degree.
Identification of gene modules related to cirrhosis in the TCGA-LIHC cohort
To identify genes related to the initiation and development of cirrhosis, co-expressed gene networks were constructed using the WGCNA method, and the relationship between each gene module and cirrhosis severity was examined via Pearson correlation analysis in the TCGA-LIHC cohort (Fig. 3A–C). A total of 10 gene modules were identified, namely, MEblue, MEturquoise, MEred, MEbrown, MEyellow, MEgreen, MEpink, MEblack, MEmagenta, and MEgray (The genes of each gene module were shown in Table. S2). We found that regardless of the no fibrosis group, MEbrown and MEgreen exhibited the strongest and most significant correlation with portal fibrosis and fibrous septa, respectively (P < 0.05, Fig. 3D). Further analyses showed that genes in MEbrown were primarily involved in G-protein coupled receptor signaling pathway and several immune-related functions, including T-cell activation, B-cell receptor signaling, neutrophil and monocyte chemotaxis, macrophage differentiation, and cytokine-mediated signaling (Fig. 3E). In addition, genes in MEgreen were primarily associated with ribosome assembly and translation (Fig. 3F). The above pathways or biological process may be involved in the progression of liver fibrosis, which the genes in MEbrown and MEgreen may play a potential role in.Fig. 3Identification of cirrhosis-related gene modules in the TCGA-LIHC cohort. (A, B) Determination of the soft threshold and the relationship between soft threshold and connectivity. (C) Gene dendrogram and modules: each leaf represents a gene, whereas each branch represents a co-expression module. (D) Correlation between different cirrhosis degrees and gene modules. (E, F) Enrichment of MEbrown (E) and MEgreen (F) in biological processes.Enrichment degree of cirrhosis-related gene modules in different cell typesThe scRNA-seq data of 10 patients with HCC were extracted from the GSE149614 dataset to identify cell types involved in the occurrence and development of cirrhosis. After filtration, normalization, dimensionality reduction, and clustering, a total of 34,015 cells were selected and categorized into 9 subpopulations (Fig. 4A, S1). Based on the expression of marker genes (Table. S3), the 9 cell subpopulations were identified as follows: hepatocytes (APOA2+APOC3+AHSG+TTR+), macrophages (C1QA+FCER1G+AIF1+), NK/T cells (NKG7+CD3D+), plasma B cells (IGHG1+MZB1+), proliferative hepatocytes (TOP2A+MKI67+), endothelial cells (PLVAP+CLDN5+), myofibroblasts (ACTA2+TAGLN+COL1A1+LUM+), epithelial cells (EPCAM+KRT19+), and B cells (MS4A1+LY9+) (Fig. 4B).Fig. 4Enrichment levels of cirrhosis-related gene modules in different cell types based on scRNA-seq. (A) Single-cell profiles of patients with primary HCC in the GSE149614 dataset. (B) Bubble plot demonstrating the expression of marker genes in each cell type. (C, D) AUCell enrichment scores of MEbrown (C) and MEgreen (D) in each cell type.The results of the previous analyses suggested that the genes in MEbrown and MEgreen may be involved in liver fibrosis. We further used the AUCell algorithm to evaluate the enrichment degree of genes in MEbrown and MEgreen at the above single-cell levels to explore key cell types. Notably, the scores of MEbrown showed higher enrichment in macrophages, NK/T cells, B cells, and plasma cells (Fig. 4C), whereas the scores of MEgreen showed higher enrichment in hepatocytes, epithelial cells, and proliferative hepatocytes (Fig. 4D). These results suggested that the genes in MEbrown and MEgreen genes may play an important role in the development of cirrhosis, which may be closely related to the above immune cells or abnormal proliferative hepatocytes (malignant hepatocytes).Molecular mechanism of immune cells involved in liver fibrosisTo further investigate the mechanisms through which the immune cells contribute to the development of cirrhosis in HCC, we used the CellChat package to identify ligand-receptor pairs in myofibroblasts and main immune cells, including macrophages, NK/T cells, plasma B cells, and B cells, based on the above scRNA-seq data. Several signaling pathways, including TGF-β, fibroblast growth factor (FGF), NOTCH, WNT, EGF, bone morphogenetic protein (BMP) and hepatocyte growth factor (HGF), have be confirmed to be involved in liver fibrosis. For example, TGF-β is widely considered as a crucial mediator in tissue fibrosis35, and WNT, FGF and NOTCH pathways can play an important role in cell proliferation, differentiation and tissue remodeling36,37,38. Initially, we evaluated the probability of several key signaling pathways involving main ligands-receptor pairs among different cell types (Fig. 5A). The results revealed that there were significant differences in the key pathways of enrichment in different cell types, including several signaling pathways associated with the development of fibrosis, such as TGF-β, FGF, NOTCH, EGF, BMP, and HGF signaling pathways (Fig. 5B–G).Fig. 5Analysis of the molecular mechanisms of immune cells involved in fibrosis in HCC. (A) The possibility of different immune cell types acting on the singling pathway where the ligand-receptor pairs of myofibroblasts reside. (B–G) Ligand-receptor pairs in the TGF-β (B), FGF (C), NOTCH (D), BMP (E), EGF (F), and HGF (G) signaling pathway.Because ligand-receptor pairs are the key links of signaling pathway transmission, we further analyzed the communication between different cells in HCC based on several key ligand-receptor pairs of TGF-β, FGF, NOTCH, EGF, BMP, and HGF signaling pathways. In particular, we found that the interaction between macrophages and myofibroblasts was the strongest through the above signaling pathways, such as the TGFB1-(TGFBR1 + TGFBR2) pair in the TCG-β signaling pathway, the FGF7-FGFR1 and FGF7-FGFR2 pairs in the FGF signaling pathway, the JAG1-NOTCH3 pair in the NOTCH signaling pathway, the BMP2-(BMPR1B + BMPR2) and BMP2-( BMPR1B + ACVR2B) pairs in the BMP signaling pathway, the AREG-EGFR pair in the EGF signaling pathway, and the HGF-MET pair in the HGF signaling pathway (Fig. 5B–G).Molecular mechanisms of hepatocytes involved in liver fibrosisHCC is characterized by high heterogeneity, and different types of cells, including hepatocytes, proliferative hepatocytes and epithelioid cells, may have malignant characteristics in HCC tissues39. CNV is one of the most prominent features of tumor cells, so the study chose to analyze these cell types from the perspective of genomic variation to reveal their malignant characteristics40. The inferCNV package was used to analyze the CNVs profiles of hepatocytes, proliferative hepatocytes, and epithelial cells, with B cells serving as a reference. The results demonstrated that numerous genomic regions in these three cell types were notably amplified or deleted, based on the above scRNA-seq data (Fig. S2). The above results revealed all hepatocytes, proliferative hepatocytes and epithelial cells in primary HCC tissues were malignant in the samples included in this study.To further explore the mechanisms through which these malignant cells contributed to cirrhosis, we used the CellChat package to identify ligand-receptor interactions between myofibroblasts and hepatocytes, proliferative hepatocytes, or epithelial cells. Figure 6A shows the probability of key signaling pathways involving main ligands-receptor pairs in different malignant cell types. Like immune cells, there were significant differences in the key pathways of enrichment in different malignant cell types. including several important signaling pathways associated with the development of fibrosis (Fig. 6A), such as TGF-β, FGF, vascular endothelial growth factor (VEGF), WNT, NOTCH, and BMP signaling pathways (Fig. 6B–G).Fig. 6Molecular mechanisms of malignant cells involved in fibrosis in HCC. (A) The possibility of different malignant cells acting on the singling pathway where the ligand-receptor pairs of myofibroblasts reside. (B–G) Ligand–receptor pairs in the TGF-β (B), FGF (C), VEGF (D), WNT (E), NOTCH (F), and BMP (G) signaling pathway.Specifically, we found that the interactions between hepatocytes and myofibroblasts, and proliferative hepatocytes and myofibroblasts were the strongest through the above signaling pathways, such as the TGFB1-(TGFBR1 + TGFBR2) pair in the TCG-β signaling pathway, the FGF5-FGFR2 and FGF5-FGFR1 pairs in the FGF signaling pathway, the VEGFB-VEGR1 and VEGFA-VEGFR1 pairs in the VEGF signaling pathway, the JAG1-NOTCH3 pair in the NOTCH signaling pathway and the BMP2-(BMPR1B + BMPR2) pair in the BMP signaling pathway between hepatocytes and myofibroblasts, and the WNT3A-(FZD8 + LRP6) pair in the WNT signaling pathway between proliferative hepatocytes and myofibroblasts (Fig. 6B–G).Spatial distribution of different cell types in HCC tissuesThe spatial transcriptomic data of patients with HCC were extracted from the GSM7021870 dataset to explore the spatial distribution of different cell types in HCC tissues, providing a reference for further exploring the mechanism of liver fibrosis (Fig. 7A–F, S3). The results indicated that hepatocytes constituted the majority of cells within HCC tissues, with hepatocytes and myofibroblasts being closest to each other (Fig. 7A, B). Additionally, plasma B cells, macrophages and proliferative hepatocytes were in proximity to myofibroblasts (Fig. 7A, C–E). These results suggested that hepatocytes, proliferative hepatocytes, macrophages, and plasma B cells may be spatially close to myofibroblasts and potentially participate in the process of stimulating myofibroblast fibrotic activity, which may be related to the development of liver fibrosis.Fig. 7Spatial distribution of different cell types in HCC tissues. (A) Distribution of myofibroblasts in HCC tissues. (B–F) Distribution of hepatocytes (B), plasma B cells (C), macrophages (D), proliferative hepatocytes (E), and endothelial cells (F) in HCC tissues.
Multi-omic analysis identifies the molecular mechanism of hepatocellular carcinoma with cirrhosis
