Mutations in HCC
Mutations in each sample were presented in waterfall plot, and dissimilar colors indicate different mutation types. TP53, CTNNB1, and TTN were the commonly mutated genes (Supplementary Fig S1A). Missense single-nucleotide polymorphisms (SNPs), as well as C >T mutations, respectively, accounted for most of the categories (Supplementary Fig S1B–1D). Calculating mutation of every sample, median and maximum of mutations were 74.5 and 1250 separately (Supplementary Fig 1E). Moreover, counts of every variant type in dissimilar samples were presented informed of a box plot (Supplementary Fig 1F). Among the 364 HCC patients, TTN, CTNNB, TP53, MUC1, AL, PCLO, RYR2, MUC4, ABCA1 and APOB were the top 10 mutated genes (Supplementary Fig S1G).
Construction of the TTN mutation-associated prognostic signature in HCC patients
DEGs between the adjacent non-cancerous and HCC tissues were compared. Based on padj < 0.05 and |log2 FC| > 1 thresholds, 5977 DEGs were obtained. The volcano plot shows the DEGs (Figure 1A). As the second frequently mutated genes, TTN-mutation status was closely linked to HCC. Therefore, DEGs were explored between TTN mutated and wild-type groups. In summary, 19 upregulated and 330 down-regulated genes were found (Fig 1B). In total, 189 genes were differentially expressed in HCC samples, relative to normal samples (Fig 1C). To assess DEGs associated with OS outcomes in HCC patients, K-M and univariate Cox regression analyses were conducted among the 189 genes. At p < 0.05, 29 and 13 genes were identified, respectively (Fig 1D, 1E). As shown in Figure 1F, The VENN plot showed the common 5 prognostic associated genes (KCNA3, VAX1, MMP3, CXCL1 and TKTL1) in two analysis methods. According to the scores calculated by the risk assessment model, HCC patients were allocated into a high- and low-risk group (Fig 2A-2C). High risk patients had markedly worse DFS and OS outcomes relative low risk patients based on K-M analysis ((Fig 2D, 2E).
Kaplan-Meier analyses of survival outcomes based on the TTN mutation status
In addition to prognostic capacities of the five-gene signature, TTN-mutation groups were markedly negatively correlated with DFS and OS outcomes in HCC patients (Fig 3A, 3B). To investigate the TTN-mutation dependence of the five-gene signature, patients with HCC were allocated into high- and low-risk groups, respectively, in both TTN-mutation and wild type groups. Kaplan-Meier DFS curves and OS curves of the high- and low-risk groups based on the five-gene signature shared the same trends in the TTNWT and TTNMUT HCC cohorts, with high-risk groups had markedly poor outcomes (Fig 3C, 3D).
Enrichments of DEGs in high- and low-risk groups
The top 10 GO analyses results and KEGG analyses results are shown in a bar plot. Interestingly, DEGs were highly enriched in various immune-receptor associated BPs, including immune receptor activity and receptor ligand activity. KEGG pathway analyses showed high expression of genes relating to cytokine-cytokine receptor interaction pathway (Fig 4A). Furthermore, we conducted the GSEA enrichment analysis according to the riskscore level and some gene sets associated with cancer were markedly gathered in high-risk HCC patients, including “MYC targets V2”, “DNA repair”, “oxidative phosphorylation”, “MYC targets V1”, “glycolysis”, “G2M checkpoint”, “E2F targets”, “reactive oxygen species pathway”, “unfolded protein response” ( Fig 4B ).
Immune cell infiltrations
Tumor-infiltrating immune cells are the main components of the tumor microenvironment and play an important role in the occurrence, progression or metastasis of tumors. Therefore, assessing the potential correlation between different levels of immune cell infiltration in HCC enables observation of components and immune mechanisms that modulate the tumor microenvironment to influence antitumor immunity. To further characterize the complex immune status in solid tumors, ImmuCellAI was used for the prediction of the abundance of 24 immune cell types in high- and low-risk HCC patients based on RNA‐Seq data. The abundance information of various taxonomic groups in each sample were shown in Fig 5A. The abundances of several tumor-immune cell types, including Tfh、Th2、MAIT、Tr1、nTreg, between high-and low-risk HCC patients were markedly different (Fig 5B). What’s more, immune cell abundances were moderately to highly associated (Fig 5C). The abundance of CD4 naive T cells, exhausted T cells (exhausted), cytotoxic T cells(cytotoxic), Type 1 regulatory T cells (Tr1), natural regulatory T (nTreg), induced regulatory T (iTreg), type 2 helper T cells (Th2), mucosal‐associated invariant T cells (MAIT), central memory T cells, T cells follicular helper (Tfh), dendritic cells (DC), monocytes, natural killer cells (NK), CD4 T cells and CD8 T cells were generally high in low-risk group than in high-risk group. Contrastingly, the abundance of macrophages, Th17, and neutrophils were low in low-risk relative to the high-risk group (Fig 5D).
Correlations between the five-gene signature and clinic-pathological features
We next explored the independence of the risk score from the five-gene signature in regard to traditional prognostic indicators. Through univariate Cox regression analyses, M stage, T stage, TNM stage, and Risk score were markedly related with OS in HCC patients (Fig 6A). The above factors were subjected to multivariate Cox regression analyses, and the risk score was established to be an independent prognostic marker for HCC (Fig 6B). The risk score prediction model tended to exhibit better discriminative performance than other clinical factors (Fig 6C).
Verification of gene expressions in HCC cell lines
To validate the credibility of the gene signature, we evaluated CXCL5, VAX1, TKTL1, KCNA3 and MMP3 expression levels by RT-qPCR in human HCC and normal hepatocyte cell lines. Compared with normal cell line, MMP3 and CXCL5 mRNA expression were significantly higher in tumor cell lines ,p<0.05(Fig 7).