Screening and identification of key biomarkers in hepatocellular carcinoma: Evidence from bioinformatic analysis.

Hepatocellular carcinoma (HCC) is one of the most common cancers worldwide. Intense efforts have been made to elucidate the pathogeny, but the molecular mechanisms of HCC are still not well understood. To identify the candidate genes in the carcinogenesis and progression of HCC, microarray datasets GSE19665, GSE33006 and GSE41804 were downloaded from Gene Expression Omnibus (GEO) database. The differentially expressed genes (DEGs) were identified, and function enrichment analyses were performed. The protein-protein interaction network (PPI) was constructed and the module analysis was performed using STRING and Cytoscape. A total of 273 DEGs were identified, consisting of 189 downregulated genes and 84 upregulated genes. The enriched functions and pathways of the DEGs include protein activation cascade, complement activation, carbohydrate binding, complement and coagulation cascades, mitotic cell cycle and oocyte meiosis. Sixteen hub genes were identified and biological process analysis revealed that these genes were mainly enriched in cell division, cell cycle and nuclear division. Survival analysis showed that BUB1, CDC20, KIF20A, RACGAP1 and CEP55 may be involved in the carcinogenesis, invasion or recurrence of HCC. In conclusion, DEGs and hub genes identified in the present study help us understand the molecular mechanisms underlying the carcinogenesis and progression of HCC, and provide candidate targets for diagnosis and treatment of HCC.

Introduction

Hepatocellular carcinoma (HCC), mainly induced by chronic hepatitis B virus (HBV) or hepatitis C virus (HCV) infection, hepatic cirrhosis or alcoholic liver diseases, is one of the most common malignancies with a rise in new cases worldwide each year. HCC has a higher rate in developing countries partly East Asia as compared to developed countries (1). Accumulating evidence has demonstrated that abnormal expression and mutation of genes are involved in the carcinogenesis and progression of HCC, including cyclin D1 (CCND1), epidermal growth factor receptor (EGFR), c-myc and Ras, as well as mutations of tumor-suppressor genes. It was found that the G870A polymorphism in exon 4 of the CCND1 gene may increase the risk of HBV-related HCC in the Chinese population (2). The chronic stimulation of EGFR plays a key role in the neoplastic conversion and development of HCC (3). A progressive increase in c-myc mRNA and protein was noted during the different steps of malignancy of HCC (4). Aberrant activation of different levels of the Ras pathway could play important roles in HCC. In addition, overexpression of H-ras, DNA copy number gains of B-Raf and hypermethylation of Ras binding proteins were found to be involved in the poor prognosis of HCC patients (5). However, due to the lack of effective diagnostic methods at the early stage of the disease, the mortality rate of HCC remains high. Therefore, it is crucial to understand the precise molecular mechanisms involved in the carcinogenesis, proliferation and recurrence of HCC and thus develop effective diagnostic and therapeutic strategies.

During the last decades, microarray technology and bioinformatic analysis have been widely used to screen genetic alterations at the genome level, which have helped us identify the differentially expressed genes (DEGs) and functional pathways involved in the carcinogenesis and progression of HCC. However, false-positive rates in independent microarray analysis make it difficult to obtain reliable results. Thus, in the present study, 3 mRNA microarray datasets from Gene Expression Omnibus (GEO) were downloaded and analyzed to obtain DEGs between liver cancer tissues and non-cancerous tissues. Subsequently, Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis and protein-protein interaction (PPI) network analyses were performed to help us understand the molecular mechanisms underlying carcinogenesis and progression. In conclusion, a total of 273 DEGs and 16 hub genes were identified, which may be candidate biomarkers for HCC.

Materials and methods

Microarray data

GEO (http://www.ncbi.nlm.nih.gov/geo) (6) is a public functional genomics data repository of high throughout gene expression data, chips and microarrays. Three gene expression datasets [GSE36668 (7), GSE18520 (8) and GSE14407 (9)] were downloaded from GEO (Affymetrix GPL570 platform, Affymetrix Human Genome U133 Plus 2.0 Array). The probes were converted into the corresponding gene symbol according to the annotation information in the platform. The GSE19665 dataset contained 10 HCC tissue samples and 10 non-cancerous samples. GSE33006 contained 3 HCC samples and 3 non-cancerous samples. GSE41804 contained 20 HCC samples and 20 non-cancerous samples.

Identification of DEGs

The DEGs between HCC and non-cancerous samples were screened using GEO2R (http://www.ncbi.nlm.nih.gov/geo/geo2r). GEO2R is an interactive web tool that allows users to compare two or more datasets in a GEO series in order to identify DEGs across experimental conditions. The adjusted P-values (adj. P) and Benjamini and Hochberg false discovery rate were applied to provide a balance between discovery of statistically significant genes and limitations of false-positives. Probe sets without corresponding gene symbols or genes with more than one probe set were removed or averaged, respectively. logFC (fold change) >1 and adj. P-value <0.01 were considered statistically significant.

KEGG and GO enrichment analyses of DEGs

The Database for Annotation, Visualization and Integrated Discovery (DAVID; http://david.ncifcrf.gov) (version 6.7) (10) is an online biological information database that integrates biological data and analysis tools, and provides a comprehensive set of functional annotation information of genes and proteins for users to extract biological information. KEGG is a database resource for understanding high-level functions and biological systems from large-scale molecular datasets generated by high-throughput experimental technologies (11). GO is a major bioinformatics tool to annotate genes and analyze biological process of these genes (12). To analyze the function of DEGs, biological analyses were performed using DAVID online database. P<0.05 was considered statistically significant.

PPI network construction and module analysis

The PPI network was predicted using Search Tool for the Retrieval of Interacting Genes (STRING; http://string-db.org) (version 10.0) (13) online database. Analyzing the functional interactions between proteins may provide insights into the mechanisms of generation or development of diseases. In the present study, PPI network of DEGs was constructed using STRING database, and an interaction with a combined score >0.4 was considered statistically significant. Cytoscape (version 3.4.0) is an open source bioinformatics software platform for visualizing molecular interaction networks (14). The plug-in Molecular Complex Detection (MCODE) (version 1.4.2) of Cytoscape is an APP for clustering a given network based on topology to find densely connected regions (15). The PPI networks were drawn using Cytoscape and the most significant module in the PPI networks was identified using MCODE. The criteria for selection were as follows: MCODE scores >5, degree cut-off=2, node score cut-off=0.2, Max depth=100 and k-score=2. Subsequently, the KEGG and GO analyses for genes in this module were performed using DAVID.

Hub genes selection and analysis

The hub genes were selected with degrees ≥10. A network of the genes and their co-expression genes was analyzed using cBioPortal (http://www.cbioportal.org) (16,17) online platform. The biological process analysis of hub genes was performed and visualized using Biological Networks Gene Oncology tool (BiNGO) (version 3.0.3) plugin of Cytoscape (18). Hierarchical clustering of hub genes was constructed using UCSC Cancer Genomics Browser (http://genome-cancer.ucsc.edu) (19). The overall survival and disease-free survival analyses of hub genes were performed using Kaplan-Meier curve in cBioPortal. The expression profiles of TOP2A and CDK1 were analyzed and displayed using online database Serial Analysis of Gene Expression (SAGE; http://www.ncbi.nlm.nih.gov/SAGE). The relationship between expression patterns and tumor grades, hepatitis virus infection status, satellites and vascular invasion were analyzed using online database Oncomine (http://www.oncomine.com) (20–22).

Results

Identification of DEGs in HCC

After standardization of the microarray results, DEGs (1,558 in GSE19665, 1,330 in GSE33006 and 717 in GSE41804) were identified. The overlap among the 3 datasets contained 273 genes as shown in the Venn diagram (Fig. 1A), consisting of 189 downregulated genes and 84 upregulated genes between liver cancer tissues and non-cancerous tissues.

Figure 1.

Venn diagram, PPI network and the most significant module of DEGs. (A) DEGs were selected with a fold change >2 and P-value <0.01 among the mRNA expression profiling sets GSE19665, GSE33006 and GSE41804. The 3 datasets showed an overlap of 273 genes. (B) The PPI network of DEGs was constructed using Cytoscape. (C) The most significant module was obtained from PPI network with 15 nodes and 102 edges. Upregulated genes are marked in light red; downregulated genes are marked in light blue.

To analyze the biological classification of DEGs, functional and pathway enrichment analyses were performed using DAVID. GO analysis results showed that changes in biological processes (BP) of DEGs were significantly enriched in protein activation cascade, complement activation, defense response, mitotic cell cycle and cell cycle process (Table I). Changes in molecular function (MF) were mainly enriched in carbohydrate binding, oxidoreductase activity, mannose-binding, scavenger receptor activity and monosaccharide binding (Table I). Changes in cell component (CC) of DEGs were mainly enriched in the extracellular region, membrane attack complex and chromosome (Table I). KEGG pathway analysis revealed that the downregulated DEGs were mainly enriched in complement and coagulation cascades, glycolysis/gluconeogenesis and metabolic pathways, while the upregulated DEGs were mainly enriched in oocyte meiosis, cell cycle and progesterone-mediated oocyte maturation.

Table I.

GO and KEGG pathway enrichment analysis of DEGs in HCC samples.

Term	Description	Count in gene set	P-value
Downregulated
GO:0072376	Protein activation cascade	16	5.27E-15
GO:0006956	Complement activation	14	5.47E-15
GO:0006952	Defense response	41	2.24E-09
GO:0030246	Carbohydrate binding	13	0.00398
GO:0005537	Mannose-binding	4	0.0216
GO:0016491	Oxidoreductase activity	18	0.0216
GO:0005615	Extracellular space	42	1.89E-11
GO:0005576	Extracellular region	80	2.41E-10
GO:0044421	Extracellular region part	69	6.95E-09
Hsa04610	Complement and coagulation cascades	10	1.37E-07
Hsa05020	Prion diseases	6	0.000106
Hsa00232	Caffeine metabolism	3	0.000665
Hsa01100	Metabolic pathways	25	0.00296
Hsa00010	Glycolysis/gluconeogenesis	5	0.00958
Upregulated
GO:0022402	Cell cycle process	45	4.36E-34
GO:0007049	Cell cycle	48	9.01E-34
GO:0000278	Mitotic cell cycle	41	6.08E-33
GO:0000793	Condensed chromosome	15	3.67E-13
GO:0005819	Spindle	17	3.67E-13
GO:0005694	Chromosome	23	1.05E-12
Hsa04110	Cell cycle	9	3.84E-07
Hsa04114	Oocyte meiosis	6	0.00058
Hsa04914	Progesterone-mediated oocyte maturation	5	0.00156

GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; DEGs, differentially expressed genes; HCC, hepatocellular carcinoma.

The PPI network of DEGs was constructed (Fig. 1B) and the most significant module was obtained using Cytoscape (Fig. 1C). The functional analyses of genes involved in this module were analyzed using DAVID. Results showed that genes in this module were mainly enriched in cell division, mitotic nuclear division and cell cycle (Table II).

Table II.

GO and KEGG pathway enrichment analysis of DEGs in the most significant module.

Pathway ID	Pathway description	Count in gene set	FDR
GO:0051301	Cell division	14	9.99E-19
GO:0000280	Nuclear division	13	4.42E-17
GO:0007067	Mitotic nuclear division	12	3.42E-16
GO:1903047	Mitotic cell cycle process	13	1.75E-14
GO:0000278	Mitotic cell cycle	13	6.11E-14
GO:0005819	Spindle	9	7.60E-11
GO:0000793	Condensed chromosome	7	1.36E-08
GO:0000777	Condensed chromosome kinetochore	6	1.98E-08
GO:0000779	Condensed chromosome, centromeric region	6	2.18E-08
GO:0015630	Microtubule cytoskeleton	10	3.12E-08
Hsa04110	Cell cycle	6	5.20E-08
Hsa04914	Progesterone-mediated oocyte maturation	4	3.91E-05
Hsa04114	Oocyte meiosis	4	8.19E-05

GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; DEGs, differentially expressed genes; FDR, false discovery rate.

Hub gene selection and analysis

A total of 16 genes were identified as hub genes with degrees ≥10. The names, abbreviations and functions for these hub genes are shown in Table III. A network of the hub genes and their co-expression genes was analyzed using cBioPortal online platform (Fig. 2A). The biological process analysis of the hub genes is shown in Fig. 2B. Hierarchical clustering showed that the hub genes could basically differentiate the liver cancer samples from the non-cancerous samples (Fig. 2C). Subsequently, the overall survival analysis of the hub genes was performed using Kaplan-Meier curve. HCC patients with BUB1, CCNB2, CDC20, CDK1, KIF20A, KIF2C, RACGAP1 and CEP55 alteration showed worse overall survival (Fig. 3A). Nonetheless, HCC patients with BUB1, CDC20, KIF20A, NUSAP1, RACGAP1, PRC1 and CEP55 alteration showed worse disease-free survival (Fig. 3B).

Table III.

Functional roles of 16 hub genes with degree ≥10.

No.	Gene symbol	Full name	Function
1	BIRC5	Baculoviral IAP repeat containing 5	BIRC5 may prevent apoptotic cell death and is highly expressed in most tumors
2	BUB1	BUB1 mitotic checkpoint serine/threonine kinase	BUB1 promotes the progression of breast cancer
3	CCNB2	Cyclin B2	CCNB2 (cyclin B2) is associated with invasion, metastasis and poor prognosis of several cancers
4	CDC20	Cell division cycle 20	High expression of CDC20 is associated with development and progression of HCC
5	CDC25C	Cell division cycle 25C	CDC25C can regulate the G2/M transition in HCC cells
6	CDK1	Cyclin-dependent kinase 1	CDK1 can regulate the cell cycle progression, apoptosis and carcinogenesis of tumor cells
7	CEP55	Centrosomal protein 55	High expression of CEP55 can promote the proliferation of lung, breast and thyroid cancers
8	CXCL12	C-X-C motif chemokine ligand 12	High expression of CXCL12 in tumor cells may impede tumor spread
9	FOS	FBJ murine osteosarcoma viral oncogene homolog	FOS has been implicated as a regulator of cell proliferation, differentiation and transformation
10	KIF20A	Kinesin family member 20A	High expression of KIF20A is involved in the development and progression of various cancers
11	NUSAP1	Nucleolar and spindle associated protein 1	High expression of NUSAP1 is involved in the progression of prostate cancer
12	KIF2C	Kinesin family member 2C	KIF2C is overexpressed in various cancers and may be associated with the chemoresistance of ovarian cancer
13	RACGAP1	Rac GTPase activating protein 1	RACGAP1 plays a regulatory role in cytokinesis, cell growth and differentiation
14	PRC1	Protein regulator of cytokinesis 1	PRC1 may be a novel regulator of early HCC recurrence
15	SPC24	SPC24, NDC80 kinetochore complex component	High expression of SPC24 is associated with worse disease-free survival and overall survival in HCC
16	TOP2A	Topoisomerase (DNA) II α	TOP2A acts as a target for several anticancer agents and mutations of this gene have been associated with drug resistance

Figure 2.

Interaction network and biological process analysis of the hub genes. (A) Hub genes and their co-expression genes were analyzed using cBioPortal. Nodes with bold black outline represent hub genes. Nodes with thin black outline represent the co-expression genes. (B) The biological process analysis of hub genes was constructed using BiNGO. The color depth of nodes refers to the corrected P-value of ontologies. The size of nodes refers to the numbers of genes that are involved in the ontologies. P<0.01 was considered statistically significant. (C) Hierarchical clustering of hub genes was constructed using UCSC. The samples under the pink bar are non-cancerous samples and the samples under the blue bar are HCC samples. Upregulation of genes is marked in red; downregulation of genes is marked in blue.

Figure 3.

(A) Overall survival and (B) disease-free survival analyses of hub genes were performed using cBioPortal online platform. P<0.05 was considered statistically significant.

Among these genes, TOP2A and CDK1 showed the highest node degrees with 33, suggesting that they may play important roles in the carcinogenesis or progression of HCC. Using the data from cBioPortal, we noted that HCC patients who had an association of genomic alterations in TOP2A showed reductions in overall and disease-free survival. However, those observations were not statistically significant (P=0.506 for overall survival and P=0.124 for disease-free survival). In addition, the CDK1 alteration was significantly associated with worse overall survival but not disease-free survival (P=0.00111 for overall survival and P=0.280 for disease-free survival) (Fig. 3B). The expression profile of TOP2A and CDK1 in human tissue was displayed using SAGE. We found that TOP2A mRNA in brain, retina, breast, pancreas, liver, kidney, colon and peritoneum displayed higher levels as compared with the matched normal tissues (Fig. 4A). CDK1 mRNA in brain, retina, thyroid, breast, pancreas, liver and prostate displayed higher levels as compared with the matched normal tissues (Fig. 4B). Oncomine analysis of cancer vs. normal tissue showed that TOP2A and CDK1 were significantly overexpressed in HCC in the different datasets (Fig. 5A and B). In the Wurmbach Liver dataset, higher mRNA levels of TOP2A and CDK1 were associated with tumor grade, hepatitis virus infection status, satellites and vascular invasion (Fig. 6A-H).

Figure 4.

Expression profiles for (A) TOP2A and (B) CDK1 in human cancers analyzed using SAGE.

Figure 5.

Oncomine analysis of cancer vs. normal tissue of (A) TOP2A and (B) CDK1. Heat maps of TOP2A and CDK1 gene expression in clinical hepatocellular carcinoma samples vs. normal tissues. 1. Hepatocellular carcinoma vs. normal liver, Chen, Mol Biol Cell, 2002 (20). 2. Hepatocellular carcinoma vs. normal liver, Roessler, Cancer Res, 2010 (21). 3. Hepatocellular carcinoma vs. normal liver, Roessler, Cancer Res, 2010 (21). 4. Hepatocellular carcinoma vs. normal liver, Wurmbach, Hepatology, 2007 (22).

Figure 6.

Association between the expression of TOP2A and CDK1 and tumor grade, hepatitis virus infection status, satellites and vascular invasion in the Wurmbach Liver dataset. (A-D) TOP2A mRNA expression in HCC compared to normal liver tissues. (E-H) CDK1 mRNA expression in hepatocellular carcinoma samples.

Discussion

Hepatocellular carcinoma (HCC) is the fifth most common malignant tumor worldwide and its mortality rate has increased in recent years (1). The main etiology factors of HCC include chronic infection with hepatitis viruses, gene mutations, cell damage, alcoholic liver diseases and aflatoxin poisoning (23). The most common cause is chronic infection with hepatitis B virus (HBV) or hepatitis C virus (HCV), accounting for over 80% of HCC cases. However, the molecular mechanisms of HCC remain poorly understood. Cell cycle regulators play important roles in HCC. Mutation of cyclin D1 (CCND1), c-myc or Ras, hypermethylation of cyclin D2 (CCND2) promoter and aberrant expression of p53 or p21 have been reported to be involved in HCC (24,25). In addition, splicing alterations of NT5E, Sulf1 or SLC39A14 have been reported to be associated with HCC (26–28). Most cases of HCC without an early finding are not candidates for curative therapies, which may be one of the reasons for the poor patient prognosis. Thus, potential markers for diagnosis and treatment with high efficiency are urgently demanded. Microarray technology enables us to explore the genetic alterations in HCC, and has been proved to be a useful approach to identify new biomarkers in other diseases.

In the present study, 3 mRNA microarray datasets were analyzed to obtain DEGs between liver cancer tissues and non-cancerous tissues. A total of 273 DEGs were identified among the 3 datasets, including 189 downregulated genes and 84 upregulated genes. GO and KEGG enrichment analyses were performed to explore interactions among the DEGs. The upregulated genes were mainly enriched in oocyte meiosis, mitotic cell cycle and cell cycle, while the downregulated genes were mainly enriched in protein activation cascade, complement activation and complement and coagulation cascades. Previous studies have reported that dysregulation of the cell cycle process and mitotic cell cycle play important roles in the carcinogenesis or progression of tumors (24,25,29). In addition, recent studies have brought forward a tumor-promoting role for complement activation (30). Moreover, oxidoreductase activity often plays a major role in antioxidant defense and can encode tumor suppressors that are frequently altered in tumors (31,32). In a word, all these theories are consistent with our results. GO enrichment analysis revealed that changes in the most significant modules were mainly enriched in cell division, nuclear division and mitotic cell cycle process, while changes in KEGG were mainly enriched in cell cycle, progesterone-mediated oocyte maturation and oocyte meiosis.

We selected 16 DEGs as hub genes with degrees ≥10. Among these hub genes, TOP2A and CDK1 showed the highest node degrees with 33. TOP2A, which forms breaks in double-stranded DNA and alters DNA structure during transcription, has been shown to be correlated with early age onset, microvascular invasion, shorter patient survival, chemoresistance and recurrence in HCC (33,34). Thus, it is regarded as a target for anticancer agents, such as epirubicin, doxorubicin, etoposide and temozolomide (35–37). In HER2-amplified breast cancer, HER2 and TOP2A genes are frequently co-amplified (38). However, TOP2A overexpression in HCC is independent from HER2 amplification or overexpression (39). In addition, TOP2A overexpression has also been found in lung, colon and ovarian cancers, and may be regarded as a valuable biomarker for diagnosis, treatment and prognosis of tumors (40–42). In the present study, PPI network showed that TOP2A directly interacts with CDK1, RACGAP1, BIRC5 and PRC1, indicating a key role of TOP2A in HCC. Cyclin-dependent kinase 1 (CDK1), a serine/threonine kinase, regulates cell cycle progression by binding to cyclin B to form a complex called cyclin B-CDK1. miR-582-5p was found to regulate the progression of HCC by inhibiting the expression of CDK1 (43). CDK1 overexpression has also been found in lung, pancreas and other cancers. In addition to its role in cell cycle progression, cyclin B-CDK1 could interact with apoptin and regulate the subcellular localization of apoptin, leading to apoptosis and carcinogenesis (44). Survivin inactivates and blocks CDK1 by increasing the level of Wee1 and thus, inactivates and blocks the pro-apoptotic activity of CDK1 (45). Cyclin B-CDK1 is also involved in connecting mitotic arrest and apoptosis by mediating the phosphorylation of anti-apoptotic Bcl-2 (46). We assessed the expression of TOP2A and CDK1 in relation to overall and disease-free survival. Gene alteration in TOP2A showed reductions in overall and disease-free survival. However, in the present study, those observations were not statistically significant. In addition, the CDK1 alteration was significantly associated with worse overall survival, but not disease-free survival. Nevertheless, several clinical studies have shown that overexpression of TOP2A is significantly associated with shorter survival times (37,39). We speculate that the reason may be that survival analyses in cBioPortal were performed on the basis of the relationship between gene mutation and prognosis, while gene overexpression usually arises from mutation or amplification. Thus, overexpression of TOP2A in HCC may arise from gene amplification rather than mutation, and further research is needed to confirm our hypothesis. Oncomine analysis showed that higher mRNA levels of TOP2A and CDK1 were associated with tumor grade, hepatitis virus infection status, satellites and vascular invasion, indicating vital roles of TOP2A and CDK1 in the carcinogenesis or progression of HCC.

RACGAP1, a Rac- and Cdc42-specific GAP, can suppress Rac and activate RhoA, leading to the promotion of pseudopod extension and invasion (47). Moreover, RACGAP1 shows a relationship with AURKA, a negative prognostic indicator in gastric cancer (48). BIRC5, also called survivin, is overexpressed and plays important roles in cell division and proliferation in a majority of cancers including HCC. PRC1 is involved in the microtubule organization in eukaryotes and is upregulated in breast tumors (49). Recent research has found that it can be a novel regulator of early HCC recurrence via potentiating Wnt signaling (50). High expression of CDC20 is associated with sex, differentiation, tumor-node-metastasis (TNM) stage, and p53 expression of HCC (51). The protein kinase CDC25C acts as an activator of cyclin B-CDK1 that regulates the G2/M transition in HCC cells (52). HCC patients with relatively low CXCL12 mRNA levels exhibit worse overall survival (53). A recent study found that the transcriptional activity of c-FOS could be induced by membrane melanoma cell adhesion molecule (MCAM), and it is important for MCAM-induced liver tumorigenesis (54). High expression of SPC24 is associated with worse disease-free and overall survival in HCC patients (55). High expression of KIF20A is involved in the development and progression of various cancers such as pancreatic, bladder and breast cancer. A recent study found that the glioma-associated oncogene 2/KIF20A axis is crucial for the proliferation of human HCC cells (56).

Literature retrieval results showed that the interaction among HCC and hub genes NUSAP1, CEP55, BUB1, CCNB2, KIF2C and RACGAP1 has not been widely reported. NUSAP1 regulates mitosis, and high expression of NUSAP1 is involved in the progression of prostate cancer (57). CEP55 is a centrosome-associated protein, which plays an important role in regulating the cell cycle. Overexpression of CEP55 was found to promote the proliferation of several cancers, such as pulmonary adenocarcinoma, breast cancer and anaplastic thyroid carcinoma (58–60). BUB1, a component of the spindle assembly checkpoint, is overexpressed and plays important roles in the progression of breast cancer (61). However, BUB1 has been reported to have a controversial role in spindle assembly checkpoint activation (62,63), which needs further investigation. CCNB2 is overexpressed in bladder, lung and colorectal cancers, and is associated with invasion, metastasis and poor prognosis of cancers (64,65). KIF2C plays an important role in the segregation of chromosomes in mitosis, and is overexpressed in various cancers and may be associated with the chemoresistance of ovarian cancer (66,67). In addition, we also performed hierarchical clustering for hub genes. Results showed that these hub genes differentiated HCC samples from non-cancerous samples, and may be candidates for diagnostic biomarkers. Moreover, alteration of BUB1, CDC20, KIF20A, RACGAP1 and CEP55 is involved in worse overall and disease-free survival, indicating that these genes may play important roles in the carcinogenesis, progression, invasion or recurrence of HCC.

In conclusion, the present study was designed to identify DEGs that may be involved in the carcinogenesis or progression of HCC. A total of 273 DEGs and 16 hub genes were identified and may be regarded as diagnostic biomarkers for HCC. However, further studies are needed to elucidate the biological function of these genes in HCC.