IF: 0.644

Prediction of Biomarkers for Hepatocellular Carcinoma Through Microarray-Based DNA Methylation Analysis


Yan-Lan Zhang 1 , Zhong-Yue Han 2 , Xiu Pang 3 , Cheng Xu 4 , *


1 Department of Infectious Disease, People’s Hospital of Rizhao, Rizhao 276826, Shandong PR, China

2 Department of Gynecology, People’s Hospital of Rizhao, Rizhao 276826, Shandong PR, China

3 Digestive System Department, Ji Ning Hospitai of Tranditional Chinese Medicine, Jining 272000, Shandong PR, China

4 Department of Infectious Disease, People’s Hospital of Linyi, Linyi 276000, Shandong PR, China


Iranian Red Crescent Medical Journal: 20 (1); e14873
Published Online: January 7, 2018
Article Type: Research Article
Received: June 8, 2017
Revised: July 19, 2017
Accepted: October 21, 2017




Background: Aberrant DNA methylation of cytosine guanine dinucleotide sides (CpGs) is one of the earliest and most frequent alterations in cancer. However, there is no complete understanding of the methylome in hepatocellular carcinoma (HCC), and few studies comprehensively evaluated methylation signatures of HCC based on high-throughput platforms.

Objective: Based on the DNA methylation data of HCC, the current study aimed at identifying the specific DNA methylation biomarkers to diagnose HCC.

Methods: The current study used bioinformatics method based on the published microarray data of HCC was implemented in Linyi, Shandong Province, China in 2017. Using GSE57956 data downloaded from the Gene Expression Omnibus database, the differentially methylated genes between HCC and normal groups were identified. Next, hierarchical clustering was conducted to measure whether the differentially methylated genes could distinguish the HCC from the normal samples. Furthermore, functional enrichment analyses were respectively implemented for up- and down-methylated genes to further extract the potential biological processes based on DAVID tool.

Results: According to the cutoff threshold of ≥ 0.2 average beta-values difference, 1340 differentially methylated genes (1660 CpGs) were identified including 978 up-methylated and 682 down-methylated genes. Utilization of the up-methylated and down-methylated genes to enrich gene oncology (GO) terms and biological pathways led to the identification of several important function regions, and the most significant ones were innate immune response, transcription from RNA polymerase II promoter and neuroactive ligand-receptor interaction.

Conclusion: Functional terms (innate immune response, transcription from RNA polymerase II promoter, and neuroactive ligand-receptor interaction) might be as potential targets for HCC diagnosis and treatment.


Hepatocellular Carcinoma DNA Methylation Pathways, Biomarkers CpGs

Copyright © 2018, Iranian Red Crescent Medical Journal. This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/) which permits copy and redistribute the material just in noncommercial usages, provided the original work is properly cited
1. Background

Hepatocellular carcinoma (HCC) is the 5th most common cancer and the 3rd most common cause of cancer-related mortality worldwide (1). Unfortunately, in 2012, about 50% of the new cases occurring globally were in China (2). At present, the therapy methods for patient with the HCC are mainly focused on the operative treatments considered as the 1st choice and the most effective treatment method of HCC (3, 4). Although this method has certain curative effects on this disease, the radical treatment of HCC does not appear on the worldwide consensus (5).

Recently, several studies implicated how mRNA-based gene signatures were identified from HCC resection specimens and biopsies improve the prognostic performance of clinical and pathological variables (6, 7). In the lab, gene silencing or gene knockout was applied to detect the disease-related genes. Unfortunately, the process of selecting pathogenic genes is painful and time consuming; but, computational methods can solve this problem. Many previous studies adopted the comparison of genomics to identify differentially expressed genes (DEGs) to reveal the pathogenic processes of the disease through comparing the affected and control samples (8, 9). However, previous researches exhibited that many of the gene biomarkers detected from different studies on the same disease were frequently inconsistent (10, 11). Moreover, direct translation of these prognostic biomarkers into clinical decision making has not yet occurred. Thus, further understanding of HCC biology is needed to optimize prognostic accuracy and improve clinical management.

Interestingly, epigenetics is emerging as an attractive candidate for the biological studying of cancers. DNA methylation is the main epigenetic feature of DNA that participates in gene transcriptional regulation, genome stability, cell differentiation, and a wide variety of malignancies (12, 13). Since DNA methylation is stable and easily detected, it is regarded as the most promising diagnostic biomarker for cancer (14), compared with copy number variations (15), or gene/microRNA expression (16). Thus, more and more scholars concentrated on the DNA methylation in this disease. For example, many differentially methylated genes were discovered such as GABRA5 (17), APC, CDKN2A, and AKR1B1 (18), which may act as the potential biomarkers to detect HCC. Despite the fact that some diagnostic panels are identified, there is no complete understanding of the methylome in HCC, and few studies comprehensively evaluated methylation signatures based on high-throughput platforms (19).

Hence, to better understand the etiology of HCC, it was planned to identify the differentially methylated genes between HCC and normal groups. Next, hierarchical clustering was conducted to measure whether such differentially methylated genes could distinguish the HCC from the normal samples. Furthermore, functional enrichment analyses were respectively implemented on up-methylated and down-methylated genes to further extract the potential biological processes, based on DAVID tool. It can help to elucidate the pathogenesis and identity molecular biomarkers for HCC diagnosis and treatment.

2. Methods
2.1. Collection of DNA Methylation Data

The current pure bioinformatics study based on the published microarray data of HCC was implemented in Linyi, Shandong Province, China in 2017. The HCC samples were downloaded from Gene Expression Omnibus and not made by the authors. There are a total of 4348 datasets in the Gene Expression Omnibus. In the current study, under the filter conditions of “Homo sapiens”, “transcription profiling by array”, “HCC” and the assays, samples were not small, DNA methylation dataset of HCC (accession number: GSE57956) (20) deposited in the GPL8490 platform of Illumina HumanMethylation450 BeadChip, were downloaded from the Gene Expression Omnibus (GEO) in National Center of Biotechnology Information (NCBI) database (http://www.ncbi.nlm.nih.gov/gds/). In GSE57956, there were 120 samples including 59 HCC and 61 normal controls. Fifty-nine HCC and 61 adjacent non-tumorous liver tissues were requested from the National Cancer Centre of Singapore (NCCS)/SingHealth Tissue Repository when the patients’ written informed consent was obtained. The specific information about demographical variables and confounding factors are shown in the paper published by Mah et al. (17).

2.2. Data of Pretreating and Detecting Differentially Methylated Genes

The raw microarray data contained 27,578 CpG sites downloaded for further analysis. In order to make the data more accurate and understandable, it is imperative to remove the probes from the dataset when the probes met the following conditions: probes with the distance of single-nucleotide polymorphism (SNP) not greater than 2; probes with minor allele frequency (MAF) < 0.05; cross-hybridization probes; probes located in sex chromosome. After preprocessing, a total of 25,628 CpG sites were obtained for the analysis.

Lumi package (21) was used to pretreat the DNA methylation microarray data. Methylation at individual CpGs is reported as part of methylation changes (beta value), which is a quantitative measure of methylation for each CpGs site range 0 (no methylation) to 1 (complete methylation) (22). Thus, beta-mixture quantile normalization method was utilized to perform data normalization (23). In detail, firstly the beta values of HCC and normal samples were computed, respectively. Subsequently, the absolute value of difference of mean beta values between 2 groups were determined, called D. At the same time, t test was used to extract differentially methylated CpGs between the 2 groups. Differentially methylated CpGs were selected when the criteria was set at P < 0.05, and D > 0.5.

After the initial processing, further filtering analyses were conducted to promote a more stringent analysis, reduce the number of non-variable sites, and increase the statistical power of following analyses. Specifically, the sites with beta values no more than 0.2, and not less than 0.8 in all samples were deleted, and then, only the CpGs with the absolute β-values the difference between the HCC and normal groups ≥ 0.2 were retained. Finally, a cutoff threshold of ≥ 0.2 average beta-values difference was applied to identify CpGs with considerable methylation differences. The criterion applied in the current study was used by others previously (24-26).

2.3. Clustering Analysis of Differential Methylated Genes

Hierarchical clustering is a common method to determine clusters of similar data in multidimensional spaces (27). The heat map is an effective tool for data visualization, which represents a dataset with 2 dimensions and requires large display spaces when an input dataset contains a large amount of data or time steps (28, 29). In general, cancers with similar methylation profiles were clustered together. To verify the reliability of the differentially methylated genes and whether such CpGs could distinguish the HCC from the normal groups, hierarchical clustering was performed. The matrix of average beta-value levels of differentially methylated CpGs was generated between HCC and normal samples.

2.4. GO Analysis of Differentially Methylated Genes

As reported, GO analysis is broadly used as functional enrichment researches for large-scale genes. Database for Annotation, Visualization, and Integrated Discovery (DAVID) is a soft tool that can provide the researchers a comprehensive set of functional annotation to reveal the biological meaning of a large number of genes (30). In the current study, in order to find out enrichment function classification, GO enrichment analyses were performed for up- and down-methylated genes, respectively. Concretely, the Fisher exact test was firstly employed to classify the GO terms. Afterwards, Benjamini-Hochberg procedure was used to correct P values into false discovery rate (FDR) (31). Significant GO terms were identified when the threshold was set as FDR < 0.01.

2.5. Pathway Analysis of Differentially Methylated Genes

Pathway analysis is the 1st choice to gain insight into the underlying biology of differential expression genes and proteins, as it reduces complexity and increases explanatory power (32). In the current paper, the employed pathway database was the KEGG (http://www.geneme.jp/kegg/) that is a database of biological systems and provides a knowledge based reference to link genomes to life through the process of pathway mapping, which is to map a genomic of transcriptomic content of genes to KEGG reference pathways to infer systemic behaviors of the cell or the organism (33). In the current study, the significant pathways enriched by the differentially methylated genes were identified based on KEGG database and DAVID tool utilizing the Fisher test, and the cutoff criteria of significance was determined by FDR. Herein, FDR was set to 0.01. Similar to the GO analysis, the pathways were performed for up- and down-methylated genes.

3. Results
3.1. Differentially Methylated Genes

After quality control and normalization to remove 4 types of probes, overall 25,628 methylated sites were reserved from the chip dataset of 120 samples. A volcano plot showing the distribution of the 25,628 methylated CpGs was constructed, as described in Figure 1. After a preliminary screening, there were 7507 differentially methylated CpGs (comprising 5202 genes), including 4615 up-methylated and 2892 down-methylated sites. The volcano plot in Figure 1 helped to observe the distribution of the up-methylated and down-methylated genes.

Subsequently, those 7507 methylated CpGs initially identified as differentially methylated sites were subjected to further filtering. Based on the cut off threshold of ≥ 0.2 mean beta-values difference, a total of 1660 unique CpGs (involved in 1340 genes) were extracted to be differentially methylated between the 2 groups. Among these 1340 genes, 978 were differentially up-methylated and 362 were differentially down-methylated.

3.2. Hierarchical Clustering Analysis

In the current paper, hierarchical clustering analysis was applied to verify if the differentially methylated genes could distinguish the HCC samples from the controls. The heat map of the 1340 differentially methylated genes is exhibited in Figure 2; accordingly, the up-methylated and down-methylated distribution of the differentially methylated regions is shown in 120 samples. As expected, through the heat map, it was pretty clear that the HCC samples were separated from the normal controls.

3.3. GO Analysis

Overall, 25 GO functions were significantly enriched by differentially up-methylated genes, as shown in Table 1. These GO terms were ranked in ascending order on the basis of FDR values, and the top 3 significant functions were the innate immune response (FDR = 5.05E - 06), cell-cell signaling (FDR = 6.45E - 06), and neutrophil chemotaxis (FDR = 4.15E - 05). For down-methylated genes, GO functions mainly included transcription from RNA polymerase II promotor (FDR = 2.19E - 06), energy reserve metabolic process (FDR = 2.82E - 06), etc., as described in Table 2.

Table 1. List of Significant GO Terms Enriched by the Up-Methylated Genes
GO TermsFDR-Value
Innate immune response5.05E-06
Cell-cell signaling6.45E-06
Neutrophil chemotaxis4.15E-05
Humoral immune response5.87E-05
Cell adhesion1.11E-04
Intermediate filament cytoskeleton organization1.86E-04
Arachidonic acid secretion2.43E-04
Negative regulation of endopeptidase activity2.44E-04
Lipid catabolic process4.24E-04
Regulation of immune response5.60E-04
Visual perception8.38E-04
Regulation of heart contraction1.06E-03
Cytoskeleton organization1.49E-03
Defense response to bacterium1.51E-03
Cardiac muscle contraction1.59E-03
Inflammatory response1.79E-03
Adaptive immune response1.83E-03
Phosphatidylethanolamine acyl-chain remodeling1.97E-03
Calcium ion transmembrane transport2.22E-03
Positive regulation of cytokine secretion2.39E-03
Potassium ion transmembrane transport2.55E-03
Cellular defense response2.58E-03
Intermediate filament organization2.79E-03
Phosphatidylinositol acyl-chain remodeling2.79E-03
Table 2. List of Significant GO Terms Involved in the Down-Methylated Genes
GO TermsFDR-Value
Transcription from RNA polymerase II promotor ppromptorprompopromoter2.19E-06
Energy reserve metabolic process2.82E-06
Cell fate commitment5.95E-05
Palate development8.76E-05
Adenylate cyclase-activating adrenergic receptor signaling pathway1.96E-04
Cellular response to BMP stimulus2.28E-04
Nephric duct morphogenesis2.57E-04
Embryonic forelimb morphogenesis3.31E-04
Aortic valve morphogenesis5.02E-04
Anterior/posterior pattern specification6.44E-04
Negative regulation of transcription from RNA polymerase II promoter1.05E-03
Negative regulation of cell proliferation1.99E-03
Negative regulation of neuron apoptotic process2.11E-03
Dorsal/ventral pattern formation2.46E-03
Positive regulation of transcription, DNA-templated3.69E-03
Cyclic nucleotide biosynthetic process4.83E-03
Synaptic transmission, dopaminergic6.14E-03
Post-embryonic development6.22E-03
Table 3. List of the Significant Pathways Enriched With the Up-methylated Genes, According to FDR < 0.01
PathwaysFDR Value
Pancreatic secretion2.45E-07
Fat digestion and absorption9.42E-06
Linoleic acid metabolism7.01E-05
Neuroactive ligand-receptor interaction significant pathway2.35E-04
Alpha-Linolenic acid metabolism2.43E-04
Arachidonic acid metabolism4.27E-04
Ether lipid metabolism1.23E-03
Cytokine-cytokine receptor interaction2.20E-03
Glutamatergic synapse3.29E-03
Adrenergic signaling in cardiomyocytes7.98E-03
Table 4. Significant Pathways Enriched by the Down-methylated Genes, Based on FDR < 0.01
PathwaysFDR Value
Neuroactive ligand-receptor interaction1.98E-05
Pancreatic secretion3.85E-04
Pathways in cancer5.17E-04
Dopaminergic synapse4.43E-03
Hippo signaling pathway4.92E-03
Gastric acid secretion5.52E-03
cAMP signaling pathway5.61E-03
Figure 1. Volcano Plot of the Methylation Data of 120 Samples Including 59 Hepatocellular Carcinoma and 61 Normal Controls
Volcano Plot of the Methylation Data of 120 Samples Including 59 Hepatocellular Carcinoma and 61 Normal Controls

This figure shows mean methylation differences between HCC and normal (x axis) versus log-transformed p-values (y axis). The blue points demonstrated the differentially methylated genes.

Figure 2. The Hierarchical Clustering Was Performed as the Heat Map, Which Showed the Differential CpGs Methylation Level Distribution of the High and Low Methylation in Samples
The Hierarchical Clustering Was Performed as the Heat Map, Which Showed the Differential CpGs Methylation Level Distribution of the High and Low Methylation in Samples

Column and row respectively represented the samples and the CpG sites, the color shows a quantitative measure of methylation for each CpGs site range 0 (no methylation) to 1 (complete methylation). In detail, green represents the low methylation and red represents the high methylation. Healthy controls and HCC samples are shown by blue and red bars, respectively.

3.4. Pathway Analysis

Based on FDR < 0.01, a total of 11 significant pathways were enriched with up-methylated genes including pancreatic secretion (FDR = 2.45E - 07), fat digestion, and adsorption (FDR = 9.42E - 06), linoleic acid metabolism (FDR = 7.01E - 05), neuroactive ligand-receptor interaction (FDR = 2.35E - 04), etc. Specific information is shown in Table 3.

It is noteworthy that down-methylated genes were significantly involved in 8 pathways presented in Table 4. Among these 8 pathways, the top 3 were neuroactive ligand-receptor interaction (FDR = 1.98E - 05), pancreatic secretion (FDR = 3.85E - 04), and pathways in cancer (FDR = 5.17E - 04).

4. Discussion

HCC, as one of the leading causes of cancer-related death, is the main event leading to death in patients with cirrhosis (34). Due to the high mortality and the high recurrence rate, the prognosis of patients with HCC is still frustrating (35). In addition, the molecular pathogenesis of HCC studied by different scholars was equivocal and inconsistent. Significantly, analyzing DNA methylation data were widely used to explore the abnormally methylated genes associated with HCC and enabled the identification of targets for therapeutic strategies. Therefore, to elucidate the pathogenesis and identify molecular biomarkers for HCC, bioinformatics analysis was conducted based on methylated data. Specifically, differential methylation genes were identified, and cluster analysis was, then, implemented. Finally, GO and pathway analyses for differential methylation genes were performed. As expected, through the heat map, it was quite clear that the disease samples were separated from the controls. This fully demonstrated that these differentially methylated genes could serve as a distinguishing label, which could be applied to diagnosis biomarkers.

After conducting GO analysis for up-methylated genes, several GO terms considered as the highly related to HCC were screened out. Among these, innate immune response was the most significant GO term. In detail, immune response refers to the defense responses when human are faced with the variation of themselves or xenogenous components. In the midst of this, the innate immune response is the 1st line of defense against infectious diseases, and the principal challenge for the host is to detect the pathogen and mount a rapid defensive response (36). Up to now, many scholars studying HCC proved that when human is infected with the hepatitis B virus (HBV), the immune response is triggered starting at the antiviral T-cells quickly (37). In addition, other evidence accumulated that precise evaluation of local immune responses could be useful to predict prognosis for HCC (38). Moreover, a former study exhibited that the prognosis of HCC was predicted based on the unique immune response signature (39). Accordingly, the researches on the immune response in HCC collectively proved that it may act as a good revelation of the development of this disease.

For down-methylated genes, the most important GO term was transcription from RNA polymerase II promoter. In detail, the RNA polymerase II promoter is a key component in the regulation of gene expression (40). A previous study demonstrated that the strongest promoter precedes the surface antigen gene composed of the mature HBV genome coding sequence (41). Notably, the HBV is the major etiology of HCC (42). In addition, a large noncoding RNA is implemented as a marker for murine HCC, which is usually produced by RNA polymerase II (43). Therefore, it was inferred that the GO term of RNA polymerase II promoter plays an important role in the development of HCC.

According to pathway analysis, neuroactive ligand-receptor interaction was the most abnormal pathway for the down-methylated genes in the current study. As reported, neuroactive steroid could influence the modulation of GABA receptor, and GABA receptors are capable of controlling cell proliferation (44). Cell proliferation is the hallmark of cancer (45), and suppressing cell proliferation has anti-cancer effect on HCC (46). Therefore, the current study findings indicated that the pathways of neuroactive ligand-receptor interaction might be closely related to the initiation and progression of HCC.

5. Conclusion

Currently, many studies identified the CpG sites differentially methylated between the HCC and adjacent non-tumorous tissues (47). Nevertheless, the role of epigenetic changes such as aberrant DNA methylation in HCC remains largely unclear. Based on DNA methylation data, it was tried to extract the candidate gene and pathway biomarkers for HCC. It was the main strong point in the current study. However, several limitations should be considered in in the current study analysis. Firstly, the data utilized in the current study were downloaded from the GEO database, not made by the authors. Secondly, the sample size was small. Moreover, the results were bioinformatics-based, but these findings were not proved using animal experiments, which was the main weak point. Thus, further experimental studies are still urgently needed. Although, there were some drawbacks, it was believed that the differential methylation genes and functions might provide a clue to understand the potential etiology of HCC. Additionally, this DND-methylation method might provide analysis on other corresponding studies. In conclusion, functional terms (innate immune response, RNA polymerase II promoter, and neuroactive ligand-receptor interaction) might act as potential therapeutic targets for HCC.

  • 1. El-Serag HB, Rudolph KL. Hepatocellular carcinoma: epidemiology and molecular carcinogenesis. Gastroenterology. 2007;132(7):2557-76. doi: 10.1053/j.gastro.2007.04.061. [PubMed: 17570226].
  • 2. Wang YC, Wei LJ, Liu JT, Li SX, Wang QS. Comparison of Cancer Incidence between China and the USA. Cancer Biol Med. 2012;9(2):128-32. doi: 10.3969/j.issn.2095-3941.2012.02.009. [PubMed: 23691468].
  • 3. Kudo M. Local ablation therapy for hepatocellular carcinoma: current status and future perspectives. J Gastroenterol. 2004;39(3):205-14. doi: 10.1007/s00535-003-1280-y. [PubMed: 15064996].
  • 4. Brudvik KW, Vauthey JN. Peeling-off technique or bile duct resection of tumor thrombus in hepatocellular carcinoma: does future liver-directed treatment justify less extensive surgery?. Ann Surg. 2015;261(5). e126. doi: 10.1097/SLA.0000000000001210. [PubMed: 25775077].
  • 5. McCormack L, Petrowsky H, Clavien PA. Surgical therapy of hepatocellular carcinoma. Europ J Gastroenterol Hepatol. 2005;17(5):497-503.
  • 6. Villanueva A, Hoshida Y, Battiston C, Tovar V, Sia D, Alsinet C, et al. Combining clinical, pathology, and gene expression data to predict recurrence of hepatocellular carcinoma. Gastroenterology. 2011;140(5):1501-12 e2. doi: 10.1053/j.gastro.2011.02.006. [PubMed: 21320499].
  • 7. Hoshida Y, Villanueva A, Sangiovanni A, Sole M, Hur C, Andersson KL, et al. Prognostic gene expression signature for patients with hepatitis C-related early-stage cirrhosis. Gastroenterology. 2013;144(5):1024-30. doi: 10.1053/j.gastro.2013.01.021. [PubMed: 23333348].
  • 8. Han H, Bearss DJ, Browne LW, Calaluce R, Nagle RB, Von Hoff DD. Identification of differentially expressed genes in pancreatic cancer cells using cDNA microarray. Cancer Res. 2002;62(10):2890-6. [PubMed: 12019169].
  • 9. Liu F, Zhou D, Zhang Q, Wang B. Comments on the article of Bu, et al entitled "P16INK4a overexpression and survival in osteosarcoma patients: a meta analysis". Int J Clin Exp Pathol. 2015;8(3):3384-5. [PubMed: 26045875].
  • 10. Ganguly A, Shields CL. Differential gene expression profile of retinoblastoma compared to normal retina. Mol Vis. 2010;16:1292-303. [PubMed: 20664703].
  • 11. Reis AH, Vargas FR, Lemos B. More epigenetic hits than meets the eye: microRNAs and genes associated with the tumorigenesis of retinoblastoma. Front Genet. 2012;3:284. doi: 10.3389/fgene.2012.00284. [PubMed: 23233862].
  • 12. Portales-Casamar E, Lussier AA, Jones MJ, MacIsaac JL, Edgar RD, Mah SM, et al. DNA methylation signature of human fetal alcohol spectrum disorder. Epigenetics Chromatin. 2016;9:25. doi: 10.1186/s13072-016-0074-4. [PubMed: 27358653].
  • 13. Portela A, Esteller M. Epigenetic modifications and human disease. Nat Biotechnol. 2010;28(10):1057-68. doi: 10.1038/nbt.1685. [PubMed: 20944598].
  • 14. Zhao Y, Sun J, Zhang H, Guo S, Gu J, Wang W, et al. High-frequency aberrantly methylated targets in pancreatic adenocarcinoma identified via global DNA methylation analysis using methylCap-seq. Clin Epigenetics. 2014;6(1):18. doi: 10.1186/1868-7083-6-18. [PubMed: 25276247].
  • 15. Jiang F, Todd NW, Li R, Zhang H, Fang H, Stass SA. A panel of sputum-based genomic marker for early detection of lung cancer. Cancer Prev Res (Phila). 2010;3(12):1571-8. doi: 10.1158/1940-6207.CAPR-10-0128. [PubMed: 20864512].
  • 16. Zhu J, Yao X. Use of DNA methylation for cancer detection: promises and challenges. Int J Biochem Cell Biol. 2009;41(1):147-54. doi: 10.1016/j.biocel.2008.09.003. [PubMed: 18834953].
  • 17. Lambert MP. Aberrant DNA methylation of imprinted loci in hepatocellular carcinoma and after in vitro exposure to common risk factors. Clinic Epigenet. 2015;7(1):15.
  • 18. Yamada N, Yasui K, Dohi O, Gen Y, Tomie A, Kitaichi T, et al. Genome-wide DNA methylation analysis in hepatocellular carcinoma. Oncol Rep. 2016;35(4):2228-36. doi: 10.3892/or.2016.4619. [PubMed: 26883180].
  • 19. Villanueva A, Hoshida Y, Toffanin S, Lachenmayer A, Alsinet C, Savic R, et al. New strategies in hepatocellular carcinoma: genomic prognostic markers. Clin Cancer Res. 2010;16(19):4688-94. doi: 10.1158/1078-0432.CCR-09-1811. [PubMed: 20713493].
  • 20. Mah WC, Thurnherr T, Chow PK, Chung AY, Ooi LL, Toh HC, et al. Methylation profiles reveal distinct subgroup of hepatocellular carcinoma patients with poor prognosis. PLoS One. 2014;9(8). e104158. doi: 10.1371/journal.pone.0104158. [PubMed: 25093504].
  • 21. Du P, Zhang X, Huang CC, Jafari N, Kibbe WA, Hou L, et al. Comparison of Beta-value and M-value methods for quantifying methylation levels by microarray analysis. BMC Bioinformatics. 2010;11:587. doi: 10.1186/1471-2105-11-587. [PubMed: 21118553].
  • 22. Wu P, Farrell WE, Haworth KE, Emes RD, Kitchen MO, Glossop JR, et al. Maternal genome-wide DNA methylation profiling in gestational diabetes shows distinctive disease-associated changes relative to matched healthy pregnancies. Epigenet. 2016;(just-accepted):00.
  • 23. Teschendorff AE, Marabita F, Lechner M, Bartlett T, Tegner J, Gomez-Cabrero D, et al. A beta-mixture quantile normalization method for correcting probe design bias in Illumina Infinium 450 k DNA methylation data. Bioinformatics. 2013;29(2):189-96. doi: 10.1093/bioinformatics/bts680. [PubMed: 23175756].
  • 24. Fryer AA, Emes RD, Ismail KM, Haworth KE, Mein C, Carroll WD, et al. Quantitative, high-resolution epigenetic profiling of CpG loci identifies associations with cord blood plasma homocysteine and birth weight in humans. Epigenetics. 2011;6(1):86-94. doi: 10.4161/epi.6.1.13392. [PubMed: 20864804].
  • 25. Glossop JR, Nixon NB, Emes RD, Haworth KE, Packham JC, Dawes PT, et al. Epigenome-wide profiling identifies significant differences in DNA methylation between matched-pairs of T- and B-lymphocytes from healthy individuals. Epigenetics. 2013;8(11):1188-97. doi: 10.4161/epi.26265. [PubMed: 24005183].
  • 26. Byun HM, Siegmund KD, Pan F, Weisenberger DJ, Kanel G, Laird PW, et al. Epigenetic profiling of somatic tissues from human autopsy specimens identifies tissue- and individual-specific DNA methylation patterns. Hum Mol Genet. 2009;18(24):4808-17. doi: 10.1093/hmg/ddp445. [PubMed: 19776032].
  • 27. Suzuki R, Shimodaira H. Hierarchical Clustering with p-Values via Multiscale Bootstrap Resampling. R Packag Ver. 2009:1.2-1.
  • 28. Deu-Pons J, Schroeder MP, Lopez-Bigas N. jHeatmap: an interactive heatmap viewer for the web. Bioinformatics. 2014;30(12):1757-8. doi: 10.1093/bioinformatics/btu094. [PubMed: 24567544].
  • 29. Kumatani S, Itoh T, Motohashi Y, Umezu K, Takatsuka M, editors. Time-Varying Data Visualization Using Clustered Heatmap and Dual Scatterplots. Information Visualisation (IV), 2016 20th International Conference. 2016. IEEE; 2016. p. 63-8.
  • 30. Huang DW, Sherman BT, Lempicki RA. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Natur Potocol. 2008;4(1):44.
  • 31. Benjamini Y, Drai D, Elmer G, Kafkafi N, Golani I. Controlling the false discovery rate in behavior genetics research. Behav Brain Res. 2001;125(1-2):279-84.
  • 32. Khatri P, Sirota M, Butte AJ. Ten years of pathway analysis: current approaches and outstanding challenges. PLoS Comput Biol. 2012;8(2). e1002375. doi: 10.1371/journal.pcbi.1002375. [PubMed: 22383865].
  • 33. Kanehisa M, Araki M, Goto S, Hattori M, Hirakawa M, Itoh M, et al. KEGG for linking genomes to life and the environment. Nucleic Acid Res. 2007;36(suppl_1):D480-4.
  • 34. Bruix J, Gores GJ, Mazzaferro V. Hepatocellular carcinoma: clinical frontiers and perspectives. Gut. 2014;63(5):844-55. doi: 10.1136/gutjnl-2013-306627. [PubMed: 24531850].
  • 35. Zhong JH, Xiang BD, Gong WF, Ke Y, Mo QG, Ma L, et al. Comparison of long-term survival of patients with BCLC stage B hepatocellular carcinoma after liver resection or transarterial chemoembolization. PLoS One. 2013;8(7). e68193. doi: 10.1371/journal.pone.0068193. [PubMed: 23874536].
  • 36. Aderem A, Ulevitch RJ. Toll-like receptors in the induction of the innate immune response. Nature. 2000;406(6797):782-7. doi: 10.1038/35021228. [PubMed: 10963608].
  • 37. Brinkhoff B, Ostroumov D, Heemcke J, Woller N, Gurlevik E, Manns MP, et al. Microsphere priming facilitates induction of potent therapeutic T-cell immune responses against autochthonous liver cancers. Eur J Immunol. 2014;44(4):1213-24. doi: 10.1002/eji.201343794. [PubMed: 24338782].
  • 38. Gao Q, Qiu SJ, Fan J, Zhou J, Wang XY, Xiao YS, et al. Intratumoral balance of regulatory and cytotoxic T cells is associated with prognosis of hepatocellular carcinoma after resection. J Clin Oncol. 2007;25(18):2586-93. doi: 10.1200/JCO.2006.09.4565. [PubMed: 17577038].
  • 39. Budhu A, Forgues M, Ye QH, Jia HL, He P, Zanetti KA, et al. Prediction of venous metastases, recurrence, and prognosis in hepatocellular carcinoma based on a unique immune response signature of the liver microenvironment. Cancer Cell. 2006;10(2):99-111. doi: 10.1016/j.ccr.2006.06.016. [PubMed: 16904609].
  • 40. Butler JE, Kadonaga JT. The RNA polymerase II core promoter: a key component in the regulation of gene expression. Genes Dev. 2002;16(20):2583-92. doi: 10.1101/gad.1026202. [PubMed: 12381658].
  • 41. Tang H, Delgermaa L, Huang F, Oishi N, Liu L, He F, et al. The transcriptional transactivation function of HBx protein is important for its augmentation role in hepatitis B virus replication. J Virol. 2005;79(9):5548-56.
  • 42. Baffy G, Brunt EM, Caldwell SH. Hepatocellular carcinoma in non-alcoholic fatty liver disease: an emerging menace. J Hepatol. 2012;56(6):1384-91. doi: 10.1016/j.jhep.2011.10.027. [PubMed: 22326465].
  • 43. Lin R, Maeda S, Liu C, Karin M, Edgington TS. A large noncoding RNA is a marker for murine hepatocellular carcinomas and a spectrum of human carcinomas. Oncogene. 2007;26(6):851-8. doi: 10.1038/sj.onc.1209846. [PubMed: 16878148].
  • 44. Watanabe M, Maemura K, Oki K, Shiraishi N, Shibayama Y, Katsu K. Gamma-aminobutyric acid (GABA) and cell proliferation: focus on cancer cells. Histol Histopathol. 2006;21(10):1135-41. doi: 10.14670/HH-21.1135. [PubMed: 16835836].
  • 45. Urrego D, Tomczak AP, Zahed F, Stuhmer W, Pardo LA. Potassium channels in cell cycle and cell proliferation. Philos Trans R Soc Lond B Biol Sci. 2014;369(1638):20130094. doi: 10.1098/rstb.2013.0094. [PubMed: 24493742].
  • 46. Xu J, Zhu X, Wu L, Yang R, Yang Z, Wang Q, et al. MicroRNA-122 suppresses cell proliferation and induces cell apoptosis in hepatocellular carcinoma by directly targeting Wnt/beta-catenin pathway. Liver Int. 2012;32(5):752-60. doi: 10.1111/j.1478-3231.2011.02750.x. [PubMed: 22276989].
  • 47. Mah WC, Lee CG. DNA methylation: potential biomarker in Hepatocellular Carcinoma. Biomark Res. 2014;2(1):5. doi: 10.1186/2050-7771-2-5. [PubMed: 24635883].