Liver cancer is the sixth most common cancer and the fourth leading cause of cancer-related death worldwide, posing a significant challenge to public health [1]. Hepatocellular carcinoma (HCC) accounts for approximately 90% of all primary liver cancers [2]. Genetic abnormalities, cellular context, and external environment play essential roles in the development of HCC. Interestingly, the main risk factors of HCC vary by region. For example, because of diverse traditional dietary habits and diseases susceptibility, the predominant causes of HCC are chronic hepatitis B virus (HBV) infection and aflatoxin exposure in China, whereas hepatitis C virus (HCV) infection and alcohol intake are likely the main risk factors in other countries [1, 3]. It has been reported that more than 120 million people carried hepatitis B surface antigen (HBsAg), and approximately 54% of HCCs are attributed to HBV infection in China [2, 4]. In addition, Non-alcoholic fatty liver disease (NAFLD) is increasingly a cause of cirrhosis and hepatocellular carcinoma in China with the popularity of the sedentary lifestyle and fast food culture brought by the expansion of urbanization in recent years [5]. Since HCC has its unique genetic and environmental background in China, it is vital to investigate the molecular mechanisms involved in the occurrence, progression, and metastasis of HCC from China to improve diagnostic and therapeutic strategies. Therefore, it is an unmet need to identify relevant genes and signaling pathways involved in the pathophysiology of HCC to achieve effective diagnosis and treatment in the early stage of HCC in China.

In recent years, bioinformatics analysis based on microarrays and high-throughput sequencing technologies has been widely used to identifying differentially expressed genes (DEGs) and functional pathways related to the occurrence and development of various diseases [6, 7]. However, given the high false-positive rates, it is difficult to obtain reliable results from independent microarrays or high-throughput sequencing analysis. Fortunately, to get reliable results, integrated bioinformatics analyses have been developed to perform a large-scale analysis of cross-platform microarrays or high-throughput data.

In this study, microarray datasets GSE84005 [8], GSE84402 [9], GSE101685 (unpublished), and GSE115018 [10] about HCC in China derived from Gene Expression Omnibus (GEO) database were downloaded and analyzed to obtain DEGs between HCC tissues and normal tissue. Subsequently, Gene Ontology (GO) functional enrichment analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis was performed. Furthermore, protein-protein interaction (PPI) networks were constructed to identify subnetworks and hub genes. Above all, this work is supposed to identify potential candidate genes and provide new therapeutic targets for the advancement of HCC from China.

In the present study, A total of 293 common DEGs, including 103 up-regulated genes and 190 down-regulated genes, were identified in HCC tissues compared with normal hepatic tissues. Interestingly, as shown in the PPI network, most of the genes with higher connectivity were up-regulated genes, which were mainly enriched in cell cycle, cell division, and mitotic nuclear division. It suggested that common DEGs participate in the proliferation and division of HCC cells. The common DEGs and subnetworks were associated with signaling pathways such as metabolic pathways and cell cycle. Under the regulation of various carcinogenic pathways, cancer cells undergo adaptive metabolic reprogramming to maintain a specific metabolic state that supports their uncontrolled proliferation [20]. The latest research first used integrated proteogenomic characterization of paired tumor and adjacent liver tissues to reveal liver-specific metabolic reprogramming in HBV-related HCC [21]. Also, given the evidence that the epidemics of obesity, diabetes, and metabolic syndrome were considered as contributory factors to the occurrence of HCC [1, 5], the changes in metabolic pathways are not only the result of the progression of HCC but also may engage in the development of HCC. The recovery of abnormal metabolism provides a new idea for prevention, diagnosis, and treatment of HCC in China. The 10 hub genes were identified in the PPI network, including CDK1, CCNB1, AURKA, CCNA2, KIF11, BUB1B, TOP2A, TPX2, HMMR and CDC45. Although the critical genes screened are not the same in many earlier reports, these studies with similar results show that cell cycle and metabolic pathways play an essential role in the occurrence and development of liver cancer [22–24].

Moreover, all of the hub genes’ over-expression was related to poor prognosis in the TCGA database. Interestingly, compared with White group, the Asian group’s overexpression predicted poorer survival outcomes in Kaplan-Meier Plotter. This shows the potential of these genes as prognostic markers for HCC in Asia (including China). This may contribute to discovering biomarkers and drug targets for HCC in China that guide clinical practice and benefit patients.

It is widely believed that the cell cycle is closely linked with the advancement of cancers, while the disruption of the cell cycle is a characteristic of tumor cells. In this study, thirteen common DEGs containing half of ten hub genes, including CCNB1, CDK1, CDC45, BUB1B and CCNA2, enriched in the cell cycle. This result suggests that cell cycle plays a vital role in the development of HCC and provides new targets for identifying serological markers and therapies of HCC in China. Cyclins and cyclin-dependent protein kinases (CDKs) are important regulators for cell cycle progression [25]. CCNB1, usually called cyclin B1, is a key regulator of G2/M in the cell cycle [26]. Some studies have found that CCNB1 expression is increased in different types of cancer, such as breast cancer [27] and gastric cancer [28]. CDK1 is a member of serine/threonine protein kinases, which forms a complex with CCNB1 to regulate the mitotic process and maintain the mitotic state [29]. It has been reported that CDK1 is not only overexpressed in diffuse large B-cell lymphoma and melanoma but also highly expressed in colorectal cancer and prostate cancer [30]. Previous studies have shown that CDK1/CCNB1 inhibits the p53 signaling pathway and regulate the development of HCC [8]. An in vitro study demonstrated that HBV X protein (HBx) induces G2/M phase arrest and apoptosis through continuous activation of CDK1/CCNB1 kinase [31]. Other studies reported that both CCNB1 and CDK1 are overexpressed in HBV-related HCC tissues and are associated with poor survival [26].

CCNA2, usually called cyclin A2, binds and activates cyclin-dependent kinase 2 and promotes transition through G1/S and G2/M in the cell cycle. Some studies have implied that CCNA2 is overexpressed in many types of cancers [32]. Other studies have revealed that CCNA2 is overexpressed in HCC and may be relevant to poor prognosis [33], supporting our results. HBV integration is common in HBV-related HCC and may play an important role in the occurrence and development of HCC. In 1990, the study of Wang et al. first reported the integration of CCNA2 gene and hepatitis B virus in HCC [34]. Recent studies have demonstrated that adeno-associated virus type 2 (AAV2) infection induces insertion mutations in tumors. CCNA2, as one of the insertion target genes of AAV2, is overexpressed in HCC, promoting cell cycle progression and showing its potential carcinogenic function [35]. Recently, Bayard et al. firstly described the recurrent fusion of the CCNA2 gene in the non-cirrhotic liver cancer genome, which leads to oncogene activation by truncating a regulatory N-terminal domain [36].

BUB1B encodes a kinase involved in spindle checkpoint function and plays a role in delaying the onset of anaphase and ensuring proper chromosome segregation. Previous studies discovered that over-expression of BUB1B in tumor tissues predicts a poor prognosis of pancreatic ductal adenocarcinoma and adrenal carcinoma, while the low expression of BUB1B is associated with poor survival in patients with colon adenocarcinoma and lung cancer [26]. The up-regulation of BUB1B in tumor tissues of patients with HCC predicts poor OS and relapse-free survival (RFS) [37], which is consistent with our results. Nevertheless, its specific role in the development of HCC is still not completely clear, and further experiments are necessary.

CDC45 (cell division cycle 45) plays a vital role in the initiation and extension of DNA replication in eukaryotic chromosomes [38]. The expression of CDC45 increased in tongue squamous cell carcinomas, and its level was positively correlated with grades of precancerous lesions in epithelial dysplasia [39]. Recently, the overexpression of CDC45 was found to predict poor prognosis in Asian HCC and HBV-related HCC [40, 41], similar to our research results.

In addition to CCNB1, CDK1, CDC45, BUB1B, and CCNA2, we also identified five hub genes in Chinese liver cancer, namely AURKA, KIF11, TOP2A, TPX2, and HMMR, which play a crucial role in regulating the cell cycle. AURKA is a cell cycle-regulated kinase that appears to be involved in microtubule formation and/or stabilization at the spindle pole during chromosome segregation. AURKA plays multiple roles in regulating cancer development, while its oncogenic roles might vary in different types of cancer. In the majority of solid tumors, AURKA works mainly through overriding cell cycle checkpoints and promoting cell cycle progression [42]. Chen et al. found that AURKA is up-regulated in HCC tissues, which is associated with pathological stage and distant metastasis [43]. Interestingly, AURKA is involved in tumor metastasis after radiotherapy for HCC. This is might because AURKA enhances the radiation resistance of HCC by activating the NF-κB signaling pathway [44].

KIF11 encodes a motor protein belonging to the kinesin-like protein family, which is involved in various kinds of spindle dynamics. Previous studies have discovered over-expression of KIF11 in a variety of cancers and suggested poor survival, while another study found that chromosome instability caused by KIF11 silencing or inhibition may contribute to the development of cancers [45]. A study showed that KIF11 overexpression was significantly correlated with HCC progression and prognosis in the TCGA database [46], which is consistent with our results. However, it is necessary to further investigate the function of KIF11 and its exact mechanism in HCC.

TOP2A encodes a DNA topoisomerase, an enzyme that controls and alters the topologic states of DNA during transcription. Among all forms of topoisomerase, TOP2A is mainly involved in cell proliferation and overexpressed in a variety of cancers, and its overexpression causes the poor prognosis of these malignant tumors. In this study, TOP2A was found to be overexpressed in HCC, and its expression levels are positively correlated with poor prognosis, which was consistent with previous research [47]. Previous studies also revealed that TOP2A expression level was closely related to histological grade, microvascular invasion, and early onset. Furthermore, TOP2A was overexpressed in HBV-related HCC, which has close association with serum AFP [48].

TPX2 has been considered as a critical factor in mitosis and spindle assembly due to the Ran-regulated microtubule-associated protein properties and its control of the Aurora-A kinase [49]. It has been reported that TPX2 is overexpressed in many types of cancer, which is correlated with poor prognosis [50]. Our study found that overexpression of TPX2 predicted a poor prognosis of patients with HCC in China. Previous clinical studies have shown that the expression of TPX2 in liver cancer tissue is significantly related to tumor-node-metastasis stage, tumor number, differentiation, and stage [51]. Above all, TPX2 could be a novel prognostic biomarker and a potential therapeutic target for HCC.

HMMR, also known as RHAMM, is one of the few known hyaluronan receptors. However, a recent review indicated that HMMR encodes evolutionarily conserved homeostasis, mitosis, and meiosis regulator instead of a hyaluronan receptor [52]. Additionally, HMMR is decreased in most healthy tissues but increased in hyperplastic tissues. It has been reported that HMMR is overexpressed in HCC tissues compared with normal tissues, and its level is associated with poor prognosis [53], which confirms our results. However, further laboratory experiments are needed to investigate the importance of HMMR in the development of HCC.

Prior to our study, there has been some research on bioinformatics analysis of key genes in HCC [26, 48, 53]. Nevertheless, our study still has several obvious advantages: Firstly, the datasets we selected are from different cities in China. Therefore, the identified genes have unique guiding significance for the early diagnosis and precise treatment of patients with HCC in China. Secondly, the microarray datasets were searched from January 1st, 2016 to October 30th, 2019, to eliminate the errors caused by the imbalance of high-throughput sequencing technology due to the results of some outdated datasets are relatively inaccurate. Thirdly, before identifying DEGs, each dataset’s sample was normalized, which enables accurate comparisons of expression levels between and within samples in this study. Finally, we used the TCGA, HPA, and other databases to verify the results obtained from the GEO database, and the combined use of those databases made the results more convincing than a single database. Unfortunately, our research still has the following deficiencies. Firstly, the four datasets we selected contain only 144 samples and cover four big densely populated cities in China. Secondly, our results were only based on the analysis of GEO, TCGA, and other public databases and have not been verified by molecular biology experiments. Thirdly, our study is an integrated analysis of HCC from China, without further subgroup analysis according to different pathogenesis and tumor types. Therefore, in further study, we will enlarge samples and supplement more multi-center clinical data and add some molecular biology experiments if possible.

In summary, this study firstly screened out common DEGs and signaling pathways involved in the occurrence and development of HCC in China, on the basis of integrated bioinformatics analysis. In addition, the present study opened up new horizons for the specific etiology and molecular mechanisms of HCC and provided candidate biomarkers and new therapeutic targets for HCC in China.