Distinct Diagnostic and Prognostic Values of Minichromosome Maintenance Gene Expression in Patients with Hepatocellular Carcinoma.

Background: The aim of the present study was to identify diagnostic and prognostic values of minichromosome maintenance (MCM) gene expression in patients with hepatocellular carcinoma (HCC).
Methods: The biological function of the MCM genes were investigated by bioinformatics analysis. The diagnostic and prognostic values of the MCM genes were investigated by using the data of HCC patients from the GSE14520 and The Cancer Genome Atlas (TCGA) databases.
Results: Bioinformatics analysis of the MCM genes substantiated that MCM2-7 genes were significantly enriched in DNA replication and cell cycle, and co-expressed with each other. These genes also co-expressed in HCC tumor tissue in both the GSE14520 and TCGA cohort. We also observed that the expression of the MCM2-7 genes was increased in tumor tissue, and diagnostic receiver operating characteristic analysis of MCM2-7 indicated that these genes could serve as sensitive diagnostic markers in HCC. Survival analysis in the GSE14520 cohort suggested that expression of MCM2, MCM4, MCM5, and MCM6 were significantly associated with hepatitis B virus-related HCC overall survival (OS). However, none of the MCM genes were associated with recurrence-free survival in the GSE14520 cohort. The validation cohort of TCGA suggested that the expression of MCM2, MCM6, and MCM7 were significantly correlated with HCC OS.
Conclusion: Our study indicated that MCM2-7 genes may be potential diagnostic biomarkers in patients with HCC. Among them, MCM2 and MCM6 may serve as potential prognostic biomarkers for HCC.

Introduction

Liver cancer is more common in males than females and has become the second leading cause of cancer-related death worldwide and in developing countries in 2012 1. Approximately half of the new cases and deaths involving liver cancer worldwide occurred in China in 2012. Moreover, liver cancer was the third leading cause of cancer-related death in China in 2015 1, 2. Therefore, the early detection and management of liver cancer would be valuable. Most liver cancers are diagnosed as hepatocellular carcinoma (HCC) 3. As with other cancers, hepatocarcinogenesis is also derived from genetic and environmental factors. Furthermore, genes that are dysregulated between tumors and normal tissues are the most promising source of diagnostic and prognostic biomarkers 4-6.

Minichromosome maintenance (MCM) genes play an essential role in DNA replication and include six highly related MCM genes (MCM2, MCM3, MCM4, MCM5, MCM6, and MCM7) 7, 8. Numerous studies have demonstrated that MCM genes play essential roles in various cancers, especially in cancer diagnosis and prognosis prediction 9-13. However, a comprehensive analysis of the diagnostic and prognostic values of MCM genes in HCC still needs further in-depth investigation. The aim of the present study was to identify the diagnostic and prognostic values of MCM gene expression in patients with HCC based on information from public databases and bioinformatics analysis.

Materials and Methods

Bioinformatics analysis of MCM genes

In order to investigate the biological functions and pathways involving the MCM genes, gene function enrichment analysis of MCM genes was performed using the Database for Annotation, Visualization and Integrated Discovery (DAVID, https://david.ncifcrf.gov/home.jsp, accessed December 15, 2017) version 6.814, 15. An enrichment P-value <0.05 was considered statistically significant. We also investigated the Gene Ontology (GO) terms of MCM genes by using the Biological Networks Gene Ontology tool (BiNGO) in Cytoscape_version 3.4.016. Investigation of gene-gene and protein- protein interactions of MCM genes were performed by GeneMANIA (http://www.genemania.org/, accessed December 15, 2017) 17, 18 and the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING, https://string-db.org/, accessed December 15, 2017) 19, 20, respectively.

Data source

The GSE14520 dataset of MCM gene expression and corresponding clinical data of hepatitis B virus (HBV)-related HCC were downloaded from the Gene Expression Omnibus database (https://www.ncbi. nlm.nih.gov/geo/query/acc.cgi?acc=GSE14520, accessed December 15, 2017) 21, 22. To validate the results obtained from GSE14520 and generalize these results to HCC, a gene expression dataset from HCC patients was obtained from The Cancer Genome Atlas (TCGA, https://portal.gdc.cancer.gov/, accessed December 15, 2017) and used as the verification cohort 23. The corresponding clinical information of TCGA HCC patients was downloaded from the University of California, Santa Cruz Xena browser (UCSC Xena: http://xena.ucsc.edu/, accessed December 15, 2017). The datasets included in the current study were downloaded from public databases, therefore there was no need for the study to be approved by an additional ethics committee.

Association analysis and diagnostic value assessment

The comparison between HCC tumor tissues and adjacent normal liver tissues were evaluated by the Student's t-test. Pearson's correlation coefficient was used to evaluate correlations among genes in co-expression analysis and visualized by the corrplot package in the R platform. The additional analysis of MCM mRNA expression between normal liver tissue and primary liver cancer tissue was performed by Metabolic gEne RApid Visualizer (MERAV, http://merav.wi.mit.edu/, accessed December 15, 2017) 24. Diagnostic values of the MCM genes in distinguishing HCC tumors from adjacent normal liver tissue were performed using the receiver operating characteristic (ROC) curve calculated using SPSS software.

Survival analysis

All patients were divided into two groups according to the median value of gene expression levels in tumor tissues for survival analysis. Based on the survival analysis results of a single MCM gene, we also investigated the joint effects survival analysis of the MCM genes that were significantly correlated to HCC prognosis.

Prognostic signature construction

We investigated a prognostic model based on the expression of prognostic MCM genes. A prognosis risk score was established on the basis of a linear combination of gene expression levels multiplied by a regression coefficient(β) as the weight that was derived from a multivariate Cox proportional hazards regression model with the prognostic genes fitting the multivariate Cox regression model with OS as a dependent variable. The risk score formula was as follows: Risk score=expression of gene1 × β1gene1 + expression of gene2×β2gene2+…expression of Genen× βnGenen 25-28. Patients were divided into high and low risk groups according to the risk score median values. In order to evaluate the predictive accuracy of this gene expression-based prognostic signature in HCC outcome, a time-dependent ROC curve was constructed using the survivalROC package in the R platform 29.

Gene set enrichment analysis

To investigate the difference of biological functions and pathways between high and low expression groups of these prognostic MCM genes in HCC survival, gene set enrichment analysis (GSEA, http://software.broadinstitute.org/gsea/index.jsp, accessed December 15, 2017) 30, 31 was used to investigate potential mechanisms in the Molecular Signatures Database (MSigDB) of c2(c2.all.v6.1.symbols) and c5 (c5.all.v6.1.symbols) 32. The enrichment gene sets in GSEA that reached a nominal P-value <0.05 and false discovery rate (FDR) <0.25 were considered statistically significant.

Statistical analysis

FDRs in the GSEA were adjusted for multiple testing with the Benjamini-Hochberg procedure to control FDR 33-35. Univariate survival analysis of clinical features and MCM genes were compared using the log-rank test; those clinicopathological parameters significantly associated with OS (P < 0.05) were entered into the multivariate Cox proportional hazards regression model for adjustment, whereas, hazard ratios (HRs) and 95% confidence intervals (CIs) were used to assess the relative risk in different HCC patients that were stratified by the expression of the MCM genes. Co-expression relationships between MCM genes were assessed by the Pearson's correlation coefficient. All statistical analyses were conducted with SPSS software, version 20.0 (IBM Corporation, Armonk, NY, USA) and R 3.3.0. A P-value <0.05 was considered statistically significant.

Results

Bioinformatics analysis of the MCM genes

GO term enrichment analysis of the MCM genes, performed using DAVID, suggested that MCM genes were significantly enriched in DNA replication-related biological processes and the G1/S transition of the mitotic cell cycle (Figure ). However, the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis using DAVID indicated that all of the MCM genes were significantly associated with DNA replication and the cell cycle signaling pathway (Figure ). The directed acyclic graph of MCM genes that was constructed by BiNGO in Cytoscape also suggested that the most significant biological function of these genes was in DNA replication (Figure ). Gene-gene and protein-protein interaction networks substantiated that the MCM genes had a strong protein homology and co-expression with each other at both the gene and protein levels (Figure ).

In order to avoid the batch effect of microarray data in GSE14520, only the dataset of Affymetrix HT Human Genome U133A Array of GSE14520 was included in the current study. Because most of the patients in GSE14520 were HBV-related HCC, we excluded those patients without HBV infection reports and survival information. As a result, there were 212 HBV-related HCC tumor tissues and 204 adjacent normal liver tissues included in the current study, and all of the 212 HBV-related HCC patients had prognosis information. The raw data of the GSE14520 genome-wide expression profile were processed according to the manufacturer's guidelines (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSM362959). For multiple probe sets, the average value corresponding to the same gene was regarded as the gene's expression value and normalized by the limma package in R platform. In the validation cohort of HCC patients from TCGA, there were 371 primary tumor tissues and 50 adjacent normal liver tissues that were included in the current study. Of these, 370 HCC patients with prognosis information were used in the survival analysis. The RNA sequencing data of TCGA HCC genome-wide expression profile datasets were normalized by the DESeq package in the R platform.

Co-expression analysis of MCM genes in HCC tumor tissues was assessed by Pearson's correlation coefficient. All the MCM genes were co-expressed strongly with each other in both the GSE14520 and TCGA cohort (Figure ). When comparing the expression of MCM genes between tumor tissues and adjacent normal liver tissues, we observed that all MCM genes were significantly upregulated in HCC tumor tissue in both the GSE14520 and TCGA cohorts (Figure ). Additional comparison of the MCM genes expression between normal liver tissue and primary liver cancer tissue was performed by MERAV. We observed a marked increase of expression in all the MCM genes in liver tumor tissue (Figure ).

Because the MCM genes were significantly upregulated in HCC tumor tissue, the potential application of MCM genes in distinguishing HCC tumor tissues and adjacent normal liver tissues was also explored. The ROC analysis of MCM genes in the GSE14520 HBV-related HCC cohort indicated that all the MCM genes had high accuracy in distinguishing tumor tissues and adjacent normal liver tissues (the area under the curve [AUC] of the ROC curves of all MCM genes was >0.90, Figure ). The MCM genes of the TCGA HCC cohort showed a high accuracy in distinguishing tumor tissues and adjacent normal liver tissues (the AUC of the ROC curves of all MCM genes was >0.88; Figure ).

In the GSE14520 HBV-related HCC cohort, we observed that patients with advanced BCLC stage and cirrhosis were at significantly increased risk of HBV-related HCC death and recurrence (Table ). Male patients also have a high risk of recurrence in HBV-related HCC, whereas, patients with tumor sizes >5 cm and serum α-fetoprotein (AFP) >300 ng/ml also had a significantly increased risk of death (Table ). The other clinical features in the GSE14520 cohort do not show a significant association with HBV-related HCC recurrence-free survival (RFS) and overall survival (OS).

The survival analysis of MCM genes are shown in Figure and Table , suggesting that patients with a high expression of MCM genes in the GSE14520 cohort seem to have a longer RFS in HBV-related HCC (Table ) compared to patients with a low expression, however, the P values did not reach statistical significance. Patients with high expression of MCM2 (adjusted P =0.043; adjusted HR=1.587; 95% CI=1.016-2.480; Table ), MCM4 (adjusted P =0.043; adjusted HR=1.577; 95%CI=1.014-2.543; Table ), MCM5 (adjusted P =0.003; adjusted HR=1.991; 95%CI=1.272-3.117; Table ), and MCM6 (adjusted P=0.046; adjusted HR=1.572; 95% CI= 1.008-2.452; Table ) were significantly associated with OS in HBV-related HCC, after adjusting for tumor size, cirrhosis, and BCLC stage.

To verify and generalize the results obtained from the GSE14520 cohort, we also assessed the prognostic values of MCM genes expression in HCC OS prediction in the HCC patients from the TCGA cohort. The clinical characteristics of HCC patients in the TCGA cohort are summarized in Table . Patients with tumor stage III/IV (P <0.0001; HR=2.764; 95% CI=1.823-4.190; Table ) and without radical resection (P =0.007; HR=2.030; 95% CI=1.213-3.395; Table ) had a significantly increased risk of death from HCC, and this data was adjusted in the multivariate Cox proportional hazards model. Survival analysis of MCM genes in TCGA HCC patients are shown in Table and Figure . The mRNA expression of MCM2 (adjusted P =0.02; adjusted HR=1.574; 95% CI=1.073-2.309; Table ), MCM6 (adjusted P =0.015; adjusted HR=1.603; 95% CI=1.094-2.350; Table ), and MCM7 (adjusted P =0.003; adjusted HR=1.793; 95% CI=1.222-2.630; Table ) were significantly associated with HCC OS in the TCGA cohort.

After performing survival analysis in both the GSE14520 and TCGA cohorts, we found that both the expression of MCM2 and MCM6 genes were significantly associated with HCC OS in these two cohorts. Therefore, we investigated the joint effects of MCM2 and MCM6 expression in the OS of HCC patients. In the GSE14520 cohort, patients with both low expression of MCM2 and MCM6 had a significantly decreased risk of death in HBV-related HCC (adjusted P =0.025; adjusted HR=0.562; 95% CI= 0.339-0.929; Table ), compared to patients with high expression of both MCM2 and MCM6. Similar results were found in the TCGA cohort (both the low MCM2 and MCM6 groups vs. both the high MCM2 and MCM6 groups, adjusted P =0.01; adjusted HR=0.584; 95% CI=0.388-0.881; Table ). In addition, we observed that patients with high MCM2 and low MCM6 had a significantly decreased risk of death in the TCGA HCC cohort (adjusted P =0.044; adjusted HR=0.444; 95% CI=0.202-0.977; Table ), compared to the patients with a high expression of both MCM2 and MCM6.

The MCM2 and MCM6 were associated with a significantly different survival in the joint effects survival analysis; however, the combination of MCM2 and MCM6 in prognostic prediction still needed further development. Our previous study divided the patients into high and low risk groups using a risk score model based on the expression of genes 25; therefore, MCM2 and MCM6 expression were used for further prognostic signature construction. In the GSE14520 cohort, the regression coefficient (β) that was derived from the multivariate Cox proportional hazards regression model and the risk score formula was: risk score = expression of MCM2 × 0.181 + expression of MCM6×1.552. Survival analysis of the prognostic signature in the GSE14520 cohort suggested that patients with a high risk score had a significantly increased risk of death in HBV-related HCC compared to the patients with a low risk score (adjusted P=0.026; adjusted HR=1.656; 95% CI=1.063-2.581; Table ). Time-dependent ROC analysis of the risk score indicated that the prognostic signature performed well in the HBV-related HCC OS prediction of the GSE14520 cohort, as the AUC of the ROC curve was 0.548, 0.598, 0.607, and 0.612 for 1-, 2-, 3-, and 5-year survival (Figure ), respectively.

The multivariate Cox proportional hazards regression model was used in the validation cohort of TCGA HCC patients with the following risk score formula: risk score=expression of MCM2 × 0.0878 + expression of MCM6×0.3056. Patients with a high risk score had a significantly increased risk of death in HCC (adjusted P=0.034; adjusted HR=1.512; 95% CI=1.033-2.213; Table ), compared to the patients with a low risk score. The AUC of the time-dependent ROC curve was 0.706, 0.673, 0.662, and 0.593 for 1-, 2-, 3-, and 5-year survival (Figure ), respectively.

GSEA

GSEA of MCM2 and MCM6 were also performed in both the GSE14520 and TCGA cohorts. The genome-wide expression profile dataset of the GSE14520 and TCGA cohorts were divided into two groups according to the median values of the MCM2 and MCM6 genes, respectively. GSEA results of the GSE14520 cohort are shown in Figure and Table , which suggested that both the high expression of MCM2 and MCM6 were significantly correlated with cell cycle process, P53 regulation pathway, liver cancer survival, liver cancer progression G1 and G2, and DNA repair. The MCM2 and MCM6 GSEA results in the GSE14520 cohort could also be validated in the TCGA HCC cohort, and high expressions of MCM2 and MCM6 were also significantly correlated with cell cycle process, P53 regulation pathway, liver cancer survival, liver cancer progression G1 and G2, and DNA repair (Figure -L and Table ).

Discussion

The MCM genes play a critical role in DNA replication 8, 36, 37. The MCM2-7 gene family is comprised of six structurally related proteins, which can form a hexameric complex, and this complex is an essential component in early G1 phase 8, 36, 37. Our gene function enrichment analysis also suggested that MCM2-7 genes were significantly enriched in DNA replication and cell cycle biological processes and pathways. Co-expression analysis demonstrated that MCM2-7 genes were strongly co-expressed with each other at both the gene and protein levels, as well as in HCC tumor tissues.

Extensive studies have reported that MCM2-7 genes are potential diagnostic markers in multiple cancers. Previous studies indicated that MCM2 is upregulated in colorectal cancer tumor tissue, and could be used as a diagnostic marker using immunocytochemical analysis from patients' tissues or colonocytes retrieved from the fecal surface 38, 39. Similar immunocytochemical detection of MCM2 in cells retrieved from urine also showed a diagnosis value in bladder cancer 40 and cervical cancer screening 41. The immunocytological evaluation of MCM3 can be used for early detection of oral squamous cell carcinoma 42. The potential diagnostic value of MCM5 has also been investigated in genito-urinary tract cancer 11, 43, oesophageal cancer 12, pancreaticobiliary malignancy 44, 45, and cervical cancer screening 41. In addition, MCM7 can be used for the early diagnosis of gastric cancer 46, and differential diagnosis between reactive mesothelial cells and malignant mesothelioma cells 47, 48. A study by Saydam et al. reported that MCM2-7 genes were upregulated in meningiomas tumor tissues and could serve as potential diagnostic markers 49. Consistent with the study by Saydam and his co-workers, our current study also observed that MCM2-7 genes were upregulated in HCC tumor tissues, and ROC analysis suggested that MCM2-7 genes may be potential diagnostic markers in HCC.

In the present study, we observed that the expression of MCM2, MCM4, MCM5, and MCM6 were significantly associated with HBV-related HCC OS in the GSE14520 cohort, whereas expression of MCM2, MCM6, and MCM7 were correlated with HCC OS in the TCGA cohort. Joint effects survival analysis suggested that patients with low expression of both MCM2 and MCM6 had a significantly decreased risk of death in HBV-related HCC compared to the patients with high expression of both MCM2 and MCM6. In addition, the risk score model, which constructed based on the expression of MCM2 and MCM6 in the GSE14520 and TCGA cohorts, also could divided the patients into high- and low-risk groups, and patients with high risk scores were significant associated with a poor OS. However, the prognostic values of MCM2-7 genes in multiple cancers also have been reported in previous studies. Numerous studies have demonstrated that the high expression of MCM2 predicts a poor prognosis in patients with gastric cancer 50-52, lung cancer 10, 53, ovarian adenocarcinomas 54, and muscle-invasive urothelial bladder carcinomas 55. Additionally, the expression of MCM2 is also an independent predictor of recurrence in stage Ta/T1 bladder cancer 56. High expression of MCM3 and MCM4, identified by immunohistochemistry, were significantly associated with OS in patients with astrocytoma 57 and esophageal adenocarcinoma 58, respectively. Expression of MCM5 also increased markedly in lung cancer and cervical cancer, and patients with a high expression of MCM5 had a significantly increased risk of death 59, 60. Immunohistochemical staining of MCM6 showed a strong correlation between MCM6 expression and OS in patients with non-small cell lung carcinoma 61, low-grade chondrosarcoma 62, mantle cell lymphoma 63, and endometrioid endometrial adenocarcinoma 64, and these patients were significantly correlated with a poor OS. Furthermore, high MCM6 immunohistochemical staining significantly increased the risk of recurrence in patients with meningiomas, as well as correlated with the histological grade 65.

Similar results of MCM7 expression in cancer prognosis, identified by immunohistochemical staining, was found in non-small cell lung cancer 66, 67, colorectal cancer 68, 69, oral squamous cell carcinoma 70, HCC 71-73, and oesophageal squamous cell carcinoma 74. These studies suggested that the MCM7 gene may serve as a prognostic biomarker, and high MCM7 expression in these cancers were significantly associated with a poor OS. Consistent with the results of the MCM7 gene in cancer OS, high expression of MCM7 also significantly correlated with a poor RFS of colorectal cancer 68, gastric adenocarcinoma 75, pituitary adenoma 76 and meningiomas 77, and lymph node metastasis of oral squamous cell carcinoma 78. By reviewing these studies, a potential prognostic role for MCM genes in HCC was identified in the current study and was consistent with previous studies, which indicated that these MCM genes may serve as oncogenes in cancer. However, our findings still need further validation.

Due to the function of the MCM genes, they have been reported to play an important multi-aspect role in HCC, such as in diagnosis, progression, and prognosis. Previous studies substantiated that MCM2 was a novel marker to assess the progression from liver cirrhosis to HCC 79, and proliferation and metastasis of HCC cells could be inhibited by long noncoding RNA FTX through binding MCM2 and miR-374a 80. In addition, MCM2 could serve as a prognostic biomarker and therapeutic target for HCC 81, 82, and MCM7 also can act as a prognostic biomarker for HCC 71-73. Polymorphisms of MCM4 rs2305952 may be associated with susceptibility of HCC 83, and plasma MCM6 serves as a diagnostic biomarker for HCC patients, especially in patients with AFP-negative and small HCC 84.

GSEA in the current study indicated that MCM2 and MCM6 were significantly associated with liver cancer survival and progression, and the potential mechanism of MCM2 and MCM6 in HCC prognosis may involve signal pathway and biological processes of the cell cycle, DNA repair, and p53, which were correlated with their biological functions. As is well-known, MCM genes play a critical role in DNA replication and participate in the cell cycle process 8. Previous studies also demonstrated that the function of the MCM2 gene was to participate in the p53 pathway in non-small cell lung carcinomas 85 and in a mouse fibroblast 3T3 cell line86, followed by cellular apoptosis. Furthermore, immunocytochemistry of the MCM2 and p53 combination can be used for distinguishing benign cells from malignant cells in squamous cell carcinoma 87 and pancreaticobiliary adenocarcinoma 88. However, the functional correlation between p53 and MCM6 has not been reported in previous studies. Due to the co-expression and GSEA of MCM6 and MCM2, we concluded that MCM6 may participate in the p53 pathway by affecting MCM2 expression. However, this hypothesis still needs further experimental confirmation.

There are some limitations in the current study that need clarification. All data in the current study were obtained from public databases and the clinical parameters were incomplete; therefore, we could not perform a comprehensive survival analysis of MCM genes that considered all the potential prognostic factors of HCC in multivariate Cox proportional hazards regression model analysis. Second, due to the different sources of HCC patients and multiple factors that influence the HCC prognosis, we could not construct a unified risk score model that was based on MCM2 and MCM6 expression levels for prognosis prediction in patients with HCC. Third, by comparison with the previous study, the limitation of our current study was that it only investigated the association between the mRNA expression of the MCM genes and HCC prognosis; however, the relationship between the MCM protein level and HCC prognosis prediction still needs further exploration.

Despite these limitations, in the present study, we have identified and validated the diagnostic and prognostic values of the expression of the MCM genes in patients with HCC, and also investigated the potential mechanism of MCM2 and MCM6 in HCC prognosis through GSEA. Once these results are verified the diagnostic and prognostic values of MCM genes at the protein level, these genes may have a potential clinical application value in HCC diagnosis, cancer management and targeted therapy. However, prospective validation with a larger sample size is necessary before the MCM genes can be included in diagnosis and prognostic monitoring for patients with HCC.

Conclusions

In the present study, we found that all MCM genes were significantly upregulated in tumor tissue, and had a potential diagnostic value in patients with HCC. Survival analysis in the GSE14520 and TCGA cohorts suggested that MCM2 and MCM6 may serve as potential prognostic biomarkers in patients with HCC. Survival analysis of the risk score model and joint effects analysis indicated that the combination of MCM2 and MCM6 could also serve as an indicator for HCC prognosis prediction. However, our findings still need further validation, and the prognostic values of other MCM genes still need prospective validation in a larger number of patients.

Table 1

Clinical characteristics of HBV-related HCC patients in GSE14520 cohort

Variables	Patients (n=212)		RFS				OS
		No. of events	MRT (months)	HR (95% CI)	P		No. of events	MST (months)	HR (95% CI)	P
Age(years)
≤60	175	96	45	1			69	NA	1
>60	37	20	48	0.974(0.602-1.578)	0.916		13	NA	0.864(0.478-1.564)	0.63
Gender
Female	29	10	NA	1			8	NA	1
Male	183	106	40	2.143(1.120-4.100)	0.021		74	NA	1.704(0.821-3.534)	0.152
Multinodular
Single	167	90	49	1			59	NA	1
Multiple	45	26	28	1.216(0.785-1.883)	0.382		23	47	1.607(0.992-2.604)	0.054
Tumor Size&
≤5 cm	137	73	51	1			46	NA	1
>5 cm	74	43	28	1.409(0.966-2.056)	0.075		36	53	1.975(1.274-3.060)	0.002
Cirrhosis
NO	17	5	NA	1			2	NA	1
Yes	195	111	37	2.612(1.066-6.402)	0.036		80	NA	4.335(1.065-17.638)	0.041
BCLC stage
0	20	6	NA	1			2	NA	1
A	143	74	51	2.050(2.892-4.711)	0.091		48	NA	4.119(1.001-16.951)	0.05
B	22	15	26	4.019(1.550-10.421)	0.004		12	46	8.992(2.005-40.320)	0.004
C	27	21	8	6.163(2.477-15.333)	<0.001		20	13	18.993(4.419-81.632)	<0.001
Serum AFPφ
≤300 ng/ml	115	62	48	1			39	NA	1
>300 ng/ml	94	54	35	1.200(0.833-1.728)	0.328		43	NA	1.546(1.002-2.385)	0.049

Notes: &Information of tumor size was unavailable in 1 patients; φ Information of serum AFP was unavailable in 3 patients. HBV, hepatitis B virus; HCC, hepatocellular carcinoma; BCLC, Barcelona Clinic Liver Cancer; AFP, α-fetoprotein; MRT, median recurrence time; MST, median survival time; RFS, recurrence-free survival; OS, overall survival; HR, hazard ratio; CI, confidence interval; NA, not available.

Table 2

Prognostic values of MCM genes expression in HBV-related HCC of GSE14520 cohort

Gene expression	Patients (n=212)	RFS							OS
		NO. of event	MRT (months)	Crude HR (95% CI)	Crude P	Adjusted HR (95% CI)	Adjusted P §		NO. of event	MST (months)	Crude HR (95% CI)	Crude P	Adjusted HR (95% CI)	Adjusted P §
MCM2
Low	106	57	51	1		1			34	NA	1		1
High	106	59	30	1.200(0.834-1.728)	0.326	1.125(0.776-1.629)	0.534		48	NA	1.693(1.090-2.629)	0.019	1.587(1.016-2.480)	0.043
MCM3
Low	106	56	48	1		1			36	NA	1		1
High	106	60	36	1.268(0.880-1.826)	0.202	1.306(0.905-1.885)	0.154		46	NA	1.502(0.971-2.324)	0.068	1.516(0.976-2.354)	0.064
MCM4
Low	106	56	51	1		1			35	NA	1		1
High	106	60	30	1.255(0.872-1.807)	0.222	1.285(0.891-1.854)	0.179		47	57	1.596(1.030-2.474)	0.037	1.577(1.014-2.543)	0.043
MCM5
Low	106	53	57	1		1			32	NA	1		1
High	106	63	32	1.392(0.966-2.007)	0.076	1.427(0.985-2.066)	0.06		50	54	1.857(1.191-2.895)	0.006	1.991(1.272-3.117)	0.003
MCM6
Low	106	58	48	1		1			35	NA	1		1
High	106	58	36	1.152(0.800-1.657)	0.448	1.111(0.765-1.613)	0.58		47	57	1.584(1.022-2.455)	0.04	1.572(1.008-2.452)	0.046
MCM7
Low	106	60	48	1		1			35	NA	1		1
High	106	56	30	1.069(0.743-1.539)	0.719	0.987(0.677-1.437)	0.945		47	NA	1.549(1.000-2.401)	0.05	1.387(0.885-2.174)	0.154

Notes: §Adjusted for tumor size, cirrhosis, BCLC stage; MCM, minichromosome maintenance; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; MRT, median recurrence time; MST, median survival time; RFS, recurrence-free survival; OS, overall survival; HR, hazard ratio; CI, confidence interval; NA, not available.

Table 3

Clinical characteristics of HCC patients in TCGA cohort

Variables	Patients (n=370)	No. of events	MST (days)	Crude HR (95% CI)	P
Age(years)
≤60	177	55	2532	1
>60	193	75	1622	1.246(0.879-1.766)	0.217
Sex
female	121	51	1490	1
male	249	79	2486	0.817(0.573-1.164)	0.262
Alcohol consumption a
NO	234	84	1694	1
YES	117	40	1624	1.026(0.703-1.496)	0.896
Ishak fibrosis score b
0 - No Fibrosis	74	30	2131	1
1,2 - Portal Fibrosis	31	9	1372	0.917(0.429-1.962)	0.823
3,4 - Fibrous Speta	28	6	NA	0.682(0.281-1.654)	0.397
5 - Nodular Formation and Incomplete Cirrhosis	9	2	1386	0.750(0.177-3.167)	0.695
6 - Established Cirrhosis	69	17	NA	0.766(0.418-1.403)	0.388
Tumor Stage c
I	171	42	2532	1
II	85	26	1852	1.427(0.874-2.330)	0.155
III/IV	90	48	770	2.764(1.823-4.190)	<0.0001
Histologic Grade d
G1	55	18	2116	1
G2	177	60	1685	1.181(0.697-2.000)	0.537
G3	121	43	1622	1.233(0.711-2.140)	0.456
G4	12	5	NA	1.693(0.626-4.584)	0.3
Serum AFP e
≤400 ng/ml	213	62	2456	1
>400 ng/ml	64	22	2486	1.055(0.645-1.724)	0.832
Radical resection f
R0	323	110	1852	1
R1/R2/RX	40	17	837	2.030(1.213-3.395)	0.007
Micro Vascular Invasion g
NO	206	60	2131	1
YES	108	36	2486	1.351(0.892-2.047)	0.155
Child-Pugh score h
A	216	59	2542	1
B/C	22	9	1005	1.614(0.796-3.270)	0.184

Notes: a Information of alcohol consumption was unavailable in 19 patients; b Information of ishak fibrosis score was unavailable in 159 patients; c Information of tumor stage was unavailable in 24 patients; d Information of histologic grade was unavailable in 5 patients; e Information of serum AFP was unavailable in 93 patients; f Information of radical resection was unavailable in 7 patients; g Information of micro vascular invasion was unavailable in 56 patients; h Information of Child-Pugh score was unavailable in 132 patients; HCC, hepatocellular carcinoma; TCGA, The Cancer Genome Atlas; OS, overall survival; MST, median survival time; HR, hazard ratio; CI, confidence interval; AFP, α-fetoprotein; NA, not available.

Table 4

Prognostic values of MCM genes expression in HCC OS of TCGA cohort

Gene expression	Patients(n=370)	NO. of event	MST (days)	Crude HR (95% CI)	Crude P	Adjusted HR (95% CI)	Adjusted P §
MCM2
Low	185	54	2116	1		1
High	185	76	1397	1.782(1.256-2.529)	0.001	1.574(1.073-2.309)	0.02
MCM3
Low	185	60	2116	1		1
High	185	70	1372	1.580(1.114-2.242)	0.010	1.456(0.992-2.136)	0.055
MCM4
Low	185	62	1791	1		1
High	185	68	1397	1.408(0.997-1.990)	0.052	1.302(0.895-1.894)	0.167
MCM5
Low	185	59	1.791	1		1
High	185	71	1622	1.386(0.980-1.960)	0.065	1.299(0.893-1.891)	0.172
MCM6
Low	185	54	2131	1		1
High	185	76	1372	1.842(1.297-2.615)	0.001	1.603(1.094-2.350)	0.015
MCM7
Low	185	53	2131	1		1
High	185	77	1149	1.852(1.304-2.632)	0.001	1.793(1.222-2.630)	0.003

Notes: §Adjusted for tumor stage and radical resection. MCM, minichromosome maintenance; HCC, hepatocellular carcinoma; OS, overall survival; MST, median survival time; HR, hazard ratio; CI, confidence interval; TCGA, The Cancer Genome Atlas.

Table 5

Joint effects analysis of MCM2 and MCM6 expression in HCC patients OS

Group	MCM2	MCM6	Patients	NO. of event	MST	Crude HR (95% CI)	Crude P	Adjusted HR (95% CI)	Adjusted P §
GSE14520 cohort			n=212		months
A	High	High	86	39	53	1		1
B	High	Low	20	9	NA	0.874(0.423-1.805)	0.716	0.803(0.384-1.679)	0.56
C	low	High	20	8	NA	0.747(0.349-1.598)	0.452	0.791(0.365-1.714)	0.552
D	Low	Low	86	26	NA	0.537(0.327-0.883)	0.014	0.562(0.339-0.929)	0.025
TCGA cohort			n=370		days
a	High	High	153	69	1005	1		1
b	High	Low	32	7	2542	0.393(0.181-0.857)	0.019	0.444(0.202-0.977)	0.044
c	low	High	32	7	NA	0.431(0.198-0.938)	0.034	0.442(0.190-1.028)	0.058
d	Low	Low	153	47	2116	0.502(0.346-0.728)	<0.001	0.584(0.388-0.881)	0.01

Notes: § Adjusted for tumor size, cirrhosis, BCLC stage in GSE14520 cohort; and adjusted for tumor stage and radical resection in TCGA cohort. MCM, minichromosome maintenance; HCC, hepatocellular carcinoma; OS, overall survival; MST, median survival time; HR, hazard ratio; CI, confidence interval; TCGA, The Cancer Genome Atlas; NA, not available.

Table 6

Survival analysis of MCM gene expression prognostic signature in HCC patients

Variables	Patients	NO. of event	MST	Crude HR (95% CI)	Crude P	Adjusted HR (95% CI)	Adjusted P §
GSE14520	n=212		months
Low risk	106	35	NA	1		1
High risk	106	47	57	1.643(1.060-2.547)	0.026	1.656(1.063-2.581)	0.026
TCGA	n=370		days
Low risk	185	55	2131	1		1
High risk	185	75	1397	1.751(1.234-2.485)	0.002	1.512(1.033-2.213)	0.034

Notes: § Adjusted for tumor size, cirrhosis, BCLC stage in GSE14520 cohort; and adjusted for tumor stage and radical resection in TCGA cohort. MCM, minichromosome maintenance; MST, median survival time; HR, hazard ratio; CI, confidence interval; TCGA, The Cancer Genome Atlas.