SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

We aimed to identify the specific subset of tumor suppressor genes (TSGs) that are methylation-silenced during the earliest steps of hepatocarcinogenesis, and to further evaluate whether these genes can serve as predictive biomarkers of hepatocellular carcinoma (HCC) emergence. A total of 482 liver tissues including 177 pairs of HCCs and matched nontumor livers and 128 liver biopsies from chronic hepatitis C (CHC) patients were analyzed for quantitative methylation analysis in 24 TSG promoters and three MINT loci. The tumors were classified as early, less-progressed, and highly progressed HCCs using histology and radiological approaches. A subset of TSGs that harbored distinctly high levels of methylation in early HCCs were selected. Based on the methylation profiles of these genes, Kaplan-Meier analyses were performed to determine time-to-HCC occurrence in CHC patients. Subsequently, multivariate analysis was performed using age, gender, fibrosis stage, and number of methylated TSGs as covariates. Among TSGs analyzed, a subset of eight TSGs (HIC1, GSTP1, SOCS1, RASSF1, CDKN2A, APC, RUNX3, and PRDM2) demonstrated a distinct cluster by hierarchical clustering and receiver operating characteristic analyses. This subset of TSGs showed significantly higher methylation levels in the early HCCs (P < 0.0001). In the CHC patients, methylation frequencies in these TSGs were associated with shorter time-to-HCC occurrence (P < 0.0001), and number of methylated genes was an independent risk factor for HCC (hazard ratio = 5.21, 95% confidence interval = 2.25-11.76, P = 0.0002). Conclusion: Epigenetic inactivation of a subset of TSGs plays a critical role in the earliest steps of hepatocarcinogenesis. Furthermore, epigenetic inactivation of these genes in CHC provides a prognostic value for determining the risk for developing HCC later in life. (HEPATOLOGY 2012;56:994–1003)

Hepatocellular carcinoma (HCC) is one of the most common malignancies worldwide; however, the molecular mechanisms contributing to hepatocarcinogenesis remain unclear. It is widely accepted that HCC exhibits numerous genetic abnormalities, including chromosomal alterations, gene amplifications and mutations, as well as epigenetic alterations.1 Studies in the past have indicated that, although several chromosomal alterations are observed, frequent genetic alterations of individual cancer-related genes rarely occurs in HCC, and even when present, their role in the earliest steps of human hepatocarcinogenesis remains controversial.1 For instance, mutations in the tumor suppressor P53 have been observed in human HCC, but these alterations are exclusively detected in the advanced stages and are seldom present in the early stage neoplasms.2

In contrast to genetic defects, epigenetic alterations, such as hypermethylation of promoter CpG islands, occur far more frequently and are believed to constitute an essential mechanism of tumor suppressor gene (TSG) inactivation in HCC.3 Furthermore, aberrant methylation of genes is not only present in HCC but can also be found in patients with chronic hepatitis or cirrhosis, which suggests the notion that epigenetic signatures emerge at early stages in the development of this disease.4 In spite of its importance, the biological impact of individual methylation events in HCC development is sometimes difficult to appreciate because some of these may not be the real “drivers” of malignancy, but simply reflect a global methylation defect that is triggered by unrelated events elsewhere in the genome.5 Therefore, a better understanding of specific methylation alterations at different stages of HCC, particularly in the earliest steps of hepatocarcinogenesis, will provide important molecular insights into the stepwise accumulation of epigenetic alterations during HCC development.

To address this important gap in knowledge, we used a systematic and multipronged approach to identify the most important genes that are the targets of aberrant methylation in the earliest stages of hepatocarcinogenesis. First, we performed quantitative methylation analysis in a panel of putative HCC-related TSGs. Here we analyzed a large tissue cohort comprising different stages of HCC and the matched corresponding nontumor liver tissues. To ensure stringent classification of early stage HCC, using histological analysis and state-of-the-art imaging techniques, we carefully categorized all tumors into three subcategories as “early,” “less-progressed,” and “highly progressed” lesions. Second, after the initial screening of a large number of methylated genes, we narrowed down our focus to a subset of genes that harbored high levels of aberrant DNA methylation in the earliest stages of HCC and examined associations between aberrant methylation and the corresponding changes in the expression of these TSGs. The rationale for this approach was that if methylation-induced transcriptional inactivation of this subset of TSGs in early tumors would provide a definitive growth advantage, the established tumor tissues must carry considerable levels of methylation at these genes as a result of clonal expansion of the affected cells. Lastly, to further validate the specificity of the early tumor-related TSGs, we next examined their methylation status in tissues from patients with chronic hepatitis C (CHC). In this instance, we also determined the relationship between the methylation events at these genes in the CHC tissues and subsequent emergence of HCC in these patients. These investigations allowed us to confirm that aberrant methylation of these TSGs in CHC tissues served as an important risk factor for developing HCC, and further suggested that these genes may act as drivers during the initial steps of human hepatocarcinogenesis.

This study is the first attempt that not only determines the role of individual methylation changes at different stages of hepatocarcinogenesis, but also specifically identifies the methylation events that orchestrate the earliest steps in HCC development. The findings reported in this study are of tremendous clinical significance because clarification of the impact of these epigenetic alterations could serve as a basis for the development of predictive biomarkers of tumor emergence and prevention in HCC.

Patients and Methods

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Patients.

A total of 482 liver tissues, including 177 HCCs and their matched corresponding nontumor livers were analyzed in this study. Among these, 300 liver tissues (150 pairs of HCCs and noncancerous livers) represented fresh-frozen samples that were obtained from surgical resection or biopsy, whereas 27 pairs of HCCs and noncancerous livers and 128 biopsy samples of CHC were formalin-fixed paraffin-embedded (FFPE) samples. According to our previous study, differences in sample processing between fresh- frozen and FFPE specimens did not affect the DNA quality for methylation analysis.6

We classified all HCCs into three stages as early HCC (eHCC), less-progressed HCC (less-pHCC), and highly progressed HCC (highly pHCC) as recommended by the International Consensus Group for Hepatocellular Neoplasia,7 particularly in the context of accurate categorization of early stage tumors. The reason for such careful subclassification was because larger tumors or tumors with hypervascular patterns at arterial phase are generally represented as progressive dedifferentiating lesions even when the pathological studies indicate these to be of well-differentiated histology.8 The details of the criteria of tumor stage and the clinical background of the patients of each stage of tumor are summarized in Table 1.

Table 1. Clinicopathological Features of the Patients in Each Stage of HCC
Clinical BackgroundEarly HCC (n = 26)Less-Progressed HCC (n = 37)Highly Progressed HCC (n = 114)
  • For the classification of tumor stages, we applied multiple criteria that combined histological tumor differentiation status, tumor size, and radiographic features obtained by the contrast-enhanced computed tomography (CE-CT). We classified tumors as early HCC (eHCC) if these lesions revealed well-differentiated HCC that were less than 2.0 cm in size and possessed a hypovascular pattern in arterial phase of CE-CT. Similarly, the HCCs were classified as less-progressed HCC (less-pHCC) if these were well differentiated, but measured greater than 2.0 cm or had hypervascular regions in the nodules in arterial phase of CE-CT. All HCCs that were moderately or poorly differentiated were categorized as highly progressed HCC (highly pHCC), regardless of their size or vascular pattern.

  • *

    Mean value (95% confidence interval) and median value (25%-75% percentile) are shown.

  • In all, 97 pairs of HCCs and their nontumor liver were newly corrected and 80 pairs of HCCs and nontumor livers were previously studied.

Age (y.o)*   
 Mean (95% CI)59.1 (55.1-63.1)63.3 (60.4-66.2)59.2 (57.3-61.2)
 Median (25%-75%)60.5 (55.8-64.5)63.0 (56.0-70.0)60.0 (54.0-66.0)
Gender   
 Male (119)172676
 Female (55)91135
 Missing (3)003
Hepatitis virus   
 HBV (39)6627
 HCV (112)192964
 HBV & HCV (3)003
 Negative (22)1219
 Missing (1)001
Background liver   
 Without cirrhosis (55)81532
 With cirrhosis (113)182174
 Missing (9)018

For the determination of aberrant methylation as a predictive marker for time-to-HCC occurrence, we analyzed an additional group of biopsy specimens from CHC cases without a prior history of HCC. These cases were randomly selected from an archival collection of 349 biopsy specimens using the inclusion and exclusion criteria listed in Supporting Table 1. Among these cases, we were able to extract usable DNA (both in terms of quantity and quality) from 128 biopsy specimens for quantitative MethyLight assays for all eight TSGs analyzed in this study. We specifically focused on CHC cases because hepatitis C virus (HCV)-related HCC had the propensity to carry high levels of methylation on CpG loci selected for this analysis.6 Liver fibrosis stage (F-stage) of each biopsy specimen was expressed using the METAVIR scoring system.9 All cases received antiviral interferon therapy during the clinical course after donating a liver biopsy. The endpoint was emergence of HCC and diagnosis was confirmed by histology. Patients were censored at the time of last clinical visit or death before developing HCC. The details of background of the patients, follow-up, and response to interferon therapy are shown in Supporting Table 1. Informed consent was obtained from each patient and the study was approved by the institutional review boards of all the involved institutions.

Methylation Analysis and Corresponding Gene Expression in Human HCC, Nontumor Liver, and HCC-Derived Cell Lines.

We performed quantitative methylation analysis for the promoter CpG islands of 24 HCC-related tumor suppressor genes (HIC1, CASP8, GSTP1, SOCS1, RASSF1, CDKN2A, APC, RUNX3, PRDM2, SFRP2, CDH1, PTGS2, CACNA1G, RASSF2, RPRM, DCC, DAPK1, DPYD, CDKN2B, SFN, WRN, BLM, RECQL, RECQL5) and three methylated in tumor (MINT) loci (MINT1, MINT2, and MINT31) in both the tumor and the matched corresponding nontumor liver tissues. For the quantitative methylation analyses of various genes and MINT loci, we used combined bisulfite restriction assay (COBRA). The primer sequences, polymerase chain reaction (PCR) conditions, and restriction enzymes for 14 of 24 gene promoters (HIC1, RASSF1, CASP8, GSTP1, SOCS1, APC, RUNX3, PRDM2, CDH1, DPYD, WRN, BLM, RECQL and RECQL5) are summarized in Supporting Table 2, whereas the details for the remaining assays have been described previously.10, 11 We also quantified methylation levels of all 27 TSGs/CpG loci in a panel of 11 HCC cell lines (HLE, HLF, HepG2, PLC/PRL/5, SNU398, SNU423, SNU449, SNU475, HuH7, Hep3B, and Li-7). The TSGs that demonstrated methylation in more than four cell lines were subsequently analyzed for the corresponding messenger RNA (mRNA) transcripts using the StepOne real-time detection system (Applied Biosystems, Foster City, CA).

Methylation Analysis of a Subset of Genes in CHC Patients.

For the methylation status determination of a smaller subset of eight TSGs (HIC1, GSTP1, SOCS1, RASSF1, CDKN2A, APC, RUNX3 and PRDM2) that were identified as important targets of epigenetic inactivation in the early tumors, we performed quantitative MethyLight assays using the StepOne real-time detection system (Applied Biosystems). The PCR primers and probes used in this assay were described previously except for those of PRDM2.12 The primers and probe sequences for PRDM2 and conditions of MethyLight assays are described in Supporting Table 3.

Statistical Analysis.

To compare the differences in methylation levels in each tumor stage, one-way factorial analysis of variance (ANOVA) and post-hoc comparisons (Tukey-Kramer HSD multicomparison), or Dunnett's test were applied using Z-scores for normalization. For the discrimination of HCC from nontumor liver using methylation levels, we applied receiver operating curve (ROC) analysis and calculated the area under the curve (AUC) values. Hierarchical clustering analysis was also performed to identify a specific cluster of TSGs/CpG loci carrying the highest levels of methylation in tumor tissues. Correlation between methylation level and expression of the corresponding gene was evaluated using Spearman's rank correlation test. Time-to-HCC occurrence was estimated using Kaplan-Meier analysis, and univariate parameters were analyzed with a log-rank test. Variables with P < 0.05 on univariate analysis were further analyzed by Cox proportional-hazards regression to determine the independent determinants of outcome variables. All P-values were two-sided and P < 0.05 was considered statistically significant. All statistical analyses were performed using JMP v. 9.0 software (SAS Institute, Cary, NC).

Results

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Identification of Hypermethylated TSGs Responsible for the Earliest Events in HCC Pathogenesis.

At first, we examined the differences in the methylation levels between tumor and nontumor liver tissues in 27 different CpG loci (24 TSGs and three MINT loci). Sixteen of 24 genes and all three MINT loci showed significant differences in methylation levels between tumor and the corresponding nontumor liver tissues. The eight genes that did not reveal any significant differences in the methylation levels included DAPK1, DPYD, CDKN2B, SFN, WRN, BLM, RECQL, RECQL5, and were excluded from further analysis (data not shown).

We next studied the distribution of mean methylation levels of all CpG loci in different stages of HCC (early, less-progressed, and highly progressed) by comparing these with the matching nontumor liver tissues. The mean methylation levels and the 95% confidential intervals (CI) are listed in Supporting Table 4. Other than DCC, all TSGs showed significant differences in methylation levels among all stages of HCC (P < 0.0001 by ANOVA for 15 of 19 TSGs). Among these, a subset of eight gene promoters, HIC1, GSTP1, SOCS1, RASSF1, CDKN2A, APC, RUNX3, and PRDM2, revealed the most prominent differences in the methylation levels between nontumor liver and the eHCC (P < 0.0001 by Dunnett's test; Fig. 1). Interestingly, hierarchical clustering analysis using methylation levels of HCC on 19 TSGs/CpG loci further revealed that all eight eHCC-related TSGs were indeed classified as one distinct cluster and uniformly hypermethylated among eHCCs (Supporting Fig. 1). These data suggest that hypermethylation of this subset of genes, referred to as the “eHCC-related TSGs,” perhaps is critical for the initial steps of human hepatocarcinogenesis.

thumbnail image

Figure 1. Multiple comparisons of methylation levels at individual early HCC-related TSGs between noncancerous liver and different stages of HCC. This figure represents distribution of methylation levels in TSGs carrying prominent methylation in early HCCs (early HCC-related TSGs) in comparison to nontumor tissues. Methylation levels of each HCC stage are compared to that of noncancerous livers by Dunnett's multiple comparison tests. All comparisons depict significant differences with P < 0.0001 (highlighted in red). The green diamonds and the horizontal lines within the diamonds represent the mean values and the 95% CI. The box-and-whiskers plots denote 75% and 95% distributions, and the lines within the boxes denote median values. The horizontal dashed line indicates mean methylation levels of all tissues in each TSGs. NT, nontumor livers; E, early HCCs; L-Prog, less-progressed HCCs; H-Prog, highly progressed HCCs.

Download figure to PowerPoint

On the other hand, four TSGs (CASP8, MINT31, PTGS2, and CACNA1G) did not reveal any significant differences in the methylation between the nontumor liver and the eHCCs, but demonstrated significant hypermethylation in the less-pHCC (classified as less-pHCC-related TSGs; P < 0.0001 for CASP8 and CACNA1G, P = 0.0009 for MINT31 and P = 0.0183 for PTGS2). Furthermore, the methylation levels of these four genes continued to increase with the progression of the disease, indicating that these may play an important role in accelerating the progression of early stage HCCs to more advanced stages (Supporting Table 4 and Supporting Fig. 2A). At the remaining seven genes, the elevation in the methylation levels was specifically observed only in the highly pHCC in comparison to the nontumor liver tissues (classified as highly pHCC-related TSGs; RASSF2, MINT1, MINT2, RPRM, SFRP2, CDH1 and DCC; Supporting Table 4 and Supporting Fig. 2B).

Discrimination of HCC by ROC Analysis Using Methylation Levels of Various CpG Loci.

We previously reported that methylation of certain TSGs in HCC is not exclusively detected in the tumor tissue but could also be found in the nontumor livers, as well as in normal aging livers.6 These data question the specificity and the biological relevance of such methylation events as critical alterations underlying HCC pathogenesis. Therefore, in order to determine which methylation events specifically contribute to hepatocarcinogenesis most profoundly, we analyzed the quantitative methylation levels of various TSGs/CpG loci in their ability to successfully discriminate HCC from the nontumor livers using ROC analysis. It was intriguing to discover that six of eight eHCC-related TSGs represented AUC values higher than 0.80 (Table 2). In contrast, the AUC values of six of seven highly pHCC-related TSGs were below 0.60.

Table 2. ROC Curve Analysis for the Discrimination of HCC Tissues by Analyzing Methylation Levels of Each TSG/CpG Loci
TSG/CpG LocusROC Analysis
Best Threshold Limits (%)SensitivitySpecificityAUC Value
  1. Best threshold of methylation level (%) for the discrimination of HCC tissues from nontumor background liver, their sensitivity, specificity, and AUC values are shown.

  2. AUC values of more than 0.80 are indicated in bold with an underline. AUC values of more than 0.70 are listed in bold. Methylation levels of six of eight TSGs, which carried significant hypermethylation in early HCC, showed AUC values of more than 0.80, suggesting that prominent increase of methylation level took place during a dedifferentiation process from noncancerous hepatocyte to early stage of tumors.

APC22.00.84090.93100.90731
GSTP120.30.75000.91940.86620
RASSF126.00.68750.92490.84209
RUNX35.50.62500.93550.81019
PRDM21.00.72160.83260.80501
CDKN2A13.00.62570.95400.80729
HIC119.00.64770.93550.79658
SOCS135.40.57390.93550.74789
CACNA1G1.00.46400.98390.72616
MINT317.00.48000.88710.69984
PTGS29.70.28001.00.63813
RASSF21.20.26400.96770.61823
MINT18.00.22581.00.61290
CASP817.00.36000.87900.59168
MINT21.70.17891.00.58943
RPRM8.00.10670.98390.56406
SFRP28.30.16800.94350.55445
DCC8.00.14400.98390.55187
CDH13.70.28230.83870.54855

Methylation Levels in Early HCC-Related TSGs Inversely Correlate with Gene Expression in HCC Cell Lines.

The high levels of methylation in the eHCC-related TSGs suggest that their epigenetic alteration is not a passive phenomenon, but might serve as a driver event in HCC. To further confirm that methylation events at these TSGs lead to their transcriptional inactivation, we quantified the methylation levels and the corresponding gene expression in 11 different HCC-derived cell lines. It was of interest to note that all eight eHCC-related TSGs specifically harbored considerable levels of methylation in HCC cell lines in comparison to other TSGs/CpG loci (Fig. 2A). In addition, the expression of these TSGs inversely correlated with their methylation status in the HCC cell lines (Fig. 2B). This evidence confirmed the functional significance of promoter hypermethylation of these TSGs because it affected their gene expression and highlighted their specific critical role in the emergence of HCC.

thumbnail image

Figure 2. Methylation profiles in HCC cell lines and the correlation between methylation levels and expression of various TSGs. (A) Eleven different HCC-derived cell lines were examined for quantitative methylation levels for 24 TSGs and three MINT loci using COBRA. Following the extraction of genomic DNA, 2 μg of DNA was subjected to bisulfite modification treatment. Eight early HCC-related TSGs are highlighted in red letters. This set of TSGs represents higher levels of methylation compared to other TSGs/CpG loci. (B) TSGs carrying methylation in more than four HCC cell lines were subsequently examined for the changes in the gene expression of the corresponding TSG. Gene expression levels in the cell lines were quantified by comparing them with normal livers, for which the total RNA was obtained commercially. Gene-specific primers and probes were obtained from TaqMan Gene Expression Assays and expression of the GAPDH gene was used as an internal control. For the relative quantification, total RNA from normal liver was used as an internal calibrator, and the quantity of RNA was determined as a ratio of the target to that of the calibrator using a standard curve. As shown in this figure, the methylation levels were inversely correlated with the mRNA expression of the methylated gene.

Download figure to PowerPoint

DNA Methylation in CHC Patients and Their Risk for Developing HCC.

Our observations for the higher levels of methylation in the eHCC-related TSGs in the early stage tumors suggest that their epigenetic inactivation is not a passive phenomenon, but might serve as a driver event in HCC. Accordingly, we hypothesized that the presence of aberrant methylation of these genes in CHC patients may serve as a risk factor for subsequent emergence of HCC. Therefore, we next analyzed the methylation status of these eight TSGs in 128 liver biopsy tissues from CHC patients and performed Kaplan-Meier analysis to determine the duration of time-to-HCC occurrence. As sustained viral response (SVR) would reduce cancer risk dramatically,13 we not only specifically focused on non-SVR patients, but also analyzed the effect of response to interferon (nonresponse or relapse) on time-to-HCC occurrence in order to avoid an influence of antiviral interferon therapy. Standard risks of HCC in CHC such as age, gender, and F-stage were analyzed as well. We noted that CHC cases with hypermethylation of ≥5 genes showed a significantly reduced time-to-HCC occurrence in comparison to the patients with either ≤1 gene or 2-4 genes exhibiting methylation (P < 0.0001, log-rank test; Fig. 3).

thumbnail image

Figure 3. Kaplan-Meier curves for time-to-HCC occurrence in the biopsies from CHC patients. Kaplan-Meier survival curves were generated based on the methylation status of early HCC-related TSGs in CHC patients. Survival analyses for non-SVR cases are illustrated. Number of cases analyzed and the number of the cases with positive events (occurrence of HCC) were as follows: CHC with methylated TSG ≥5, total cases = 11, cases with HCC occurrence = 11; CHC with methylated TSG 4-2, total cases = 18, cases with HCC occurrence = 11; CHC with methylated TSG ≤ 1, total cases = 41, cases with HCC occurrence = 18, respectively. The P values were calculated using a log-rank test.

Download figure to PowerPoint

Older age (mean age ≥55 years), male gender, and progression of fibrosis (F3 or F4) were also significantly associated with shorter time-to-HCC occurrences. However, response to interferon therapy (nonresponse versus relapse) did not affect the time-to-HCC occurrence (P = 0.0089 for ≥55 years versus ≤54 years, P = 0.0324 for male versus female, P = 0.0121 for F4 or F3 versus F2-F0, and P = 0.3946 for NR versus relapse by log-rank test; Table 3). We subsequently conducted a multivariate analysis using the Cox's proportional-hazards regression model including age, gender, F-stage as variables, and compared these to the number of methylated TSGs. We noted that, whereas age and number of methylated TSGs were significant risk factors for shorter time-to-HCC occurrence, gender and F-stage were not associated with a significant risk for developing HCC (P = 0.0073 for age, P = 0.0742 for gender, P = 0.2849 for F-stage, and P = 0.0002 for the number of methylated TSGs of ≥ 5 versus ≤ 1, respectively; Fig. 4). In terms of the hazard ratios (HR) based on the number of methylated TSGs and the likelihood for the emergence of HCC in CHC patients, a dose-dependent effect was observed in which the HR of ≥5 methylated genes versus 1≤ methylated gene was the highest (HR = 5.21, 95% CI: 2.25-11.76), followed by that of ≥5 methylated genes versus 4-2 methylated genes (HR = 2.95, 95% CI: 1.12-7.80), and the lowest with that of 4-2 methylated genes versus 1≤ methylated gene (HR = 1.77, 95% CI: 0.78-3.88; Fig. 4).

thumbnail image

Figure 4. Hazard ratios (HRs) and P-values obtained from the multivariate analysis for each variable in CHC patients for the emergence of HCC. The P-values were calculated using Cox's proportional hazards regression model. A continuous variable of age was categorized into two groups with a mean age of ≥55 years or ≤54 years. F-stage was also categorized as F0-F2 and F3/F4. The total number of cases and the number of patients with HCC occurrence in each group and the associated P-values of univariate analyses calculated using a log-rank test are shown in Table 3.

Download figure to PowerPoint

Table 3. Univariate Analysis for the Contribution of Each Variable on Time-to-HCC Occurrence in Biopsy Specimens from Non-SVR Chronic Hepatitis C (CHC) Cases
 No. of Non-SVR CasesUnivariate
VariablesTotalWith Event*P valueHR (95% CI)
  • A continuous variable of age was categorized into two groups with a mean age of ≥55 years, or ≤54 years. F-stage was also categorized as F0-F2 and F3/F4.

  • *

    No. of cases with HCC occurrence after biopsy.

  • P-value by log-rank test. P values < 0.05 are shown in bold.

  • Nonresponse (NR), detectable for HCV-RNA during and after the treatment; relapse, no detectable HCV-RNA at the end of the treatment but detectable at 6 months after the end of the treatment.

Age    
 ≥55 y.o45300.00892.54 (1.27-5.50)
 ≤54 y.o2510 1
Gender    
 Male40270.03242.03 (1.06 - 4.08)
 Female3013 1
F stage    
 F4 or F338250.01212.23 (1.19 - 4.36)
 F2-F03215 1
Response to interferon    
 NR42280.39461.34 (0.70 - 2.74)
 Relapse2812 1
No. of methylated TSGs    
 >51111< 0.00015.98 (2.61 - 13.3)
 4-21811 1.39 (0.64 - 2.91)
 <14118 1

Discussion

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

In this study we aimed to identify the contribution of DNA methylation changes as potential contributors in the earliest steps of human HCC. Inclusion of a reasonably large number of early stage tumors allowed us to carefully identify methylation patterns of genes that might be critical determinants during the initial phases of HCC development. We were able to successfully classify patterns of methylation progression of all the analyzed genes into three categories (as illustrated in the summarized model in Fig. 5), with eight genes being specifically hypermethylated in the eHCCs. Interestingly, the subset of these eHCC-related TSGs not only revealed a methylation-induced transcriptional inactivation in HCC-cell lines, but also best discriminated HCC tissues from the normal livers by ROC analysis. In addition, Kaplan-Meier analysis demonstrated that increased methylation of these genes in CHC patients was associated with reduced time-to-HCC occurrence and served as an independent risk factor to predict the emergence of HCC in the CHC patients. Taken together, these data highlight that epigenetic inactivation of this subset of growth regulatory genes is critical, and that these genes may act as “drivers” of neoplasia during the earliest steps of hepatocarcinogenesis.

thumbnail image

Figure 5. A schematic representation of consolidated model for the pattern of methylation progression in various stages of HCC development. This figure illustrates the progression of methylation events in (A) eight early HCC-related TSGs; (B) four less-progressed HCC-related TSGs; and (C) seven highly progressed HCC-related TSGs. NT, nontumor livers; Early, early HCCs; L-Prog, less-progressed HCCs; H-Prog, highly progressed HCC. The yellow circles represent HCC cells carrying methylation on the corresponding TSGs/CpG loci. The methylation of early HCC-related TSGs leads to the clonal expansion and establishment of early tumors with high levels of methylation of the corresponding TSGs. In contrast, only a small number of scattered cells carry methylation in the highly progressed HCC-related TSGs even in the most advanced stages of HCC, indicating the possibility that these could be passive methylation events.

Download figure to PowerPoint

DNA methylation has been considered an early event in hepatocarcinogenesis.1 However, the identification of individual methylation events that are critical for an accelerated tumor growth during the earliest steps of HCC development has not been investigated previously. A logical approach to address this issue must include analysis of the early stage tumors, for the identification of genes that are not only methylated more frequently, but also carry high levels of methylation. Recently, the International Consensus Group for Hepatocellular Neoplasia revisited the concept of classifying eHCC.7 In light of these recommendations, we were very stringent with our definition of early stage tumor (or eHCCs), and considered only the ones with well-differentiated histology with less than 2.0 cm in size, which was accompanied with a hypovascular imaging pattern. We reasoned that true critical methylation events must provide sufficient growth advantage in the affected cells to permit their subsequent clonal expansion as the tumor evolves. We feel that our stringent definition of “early HCCs,” together with the quantitative methylation analysis of a large panel of HCC-related TSGs, was a unique strength of our study that allowed us to uncover a subset of critical genes that control the tumor cell growth in the earliest steps of hepatocarcinogenesis.

Among the TSGs or MINT loci analyzed, eight gene promoters carried very high levels of promoter hypermethylation in all stages of HCC. However, methylation of these loci was particularly higher in the eHCCs and formed a distinct cluster by hierarchical clustering analysis, implying that the epigenetic inactivation of these genes was potentially important during the earliest stage of tumor development. Furthermore, we were also able to identify two additional subsets of genes, whose methylation strongly correlated with the less-pHCCs and highly pHCCs, suggesting that their inactivation might play a more essential role in the later stages of HCC. We compared the methylation levels of individual TSGs by ROC analysis in an attempt to select methylated genes that could discriminate HCC from nontumor livers with a high degree of sensitivity and specificity.14 Interestingly, six of eight eHCC-related TSGs possessed AUC values that were higher than 0.80, further lending credence to our other observations that these genes may trigger an aberrant methylation cascade during early hepatocarcinogenesis. In contrast, our observation for the lower AUC values obtained from two of four less-pHCC-related TSGs is suggestive of their role in the dedifferentiation of early tumors to an intermediate stage. Likewise, even lower AUC values for the seven highly pHCC-related TSGs suggests that conceivably the methylation alterations present in these genes is simply a reflection of a passive change that occurs elsewhere in the genome due to ensuing tumorigenesis.

Recently, aberrant methylation of TSGs was reported in early HCC that had an HBV-positive genotype.15 Interestingly, four of five methylated TSGs reported in that report also belong to the subset of our eight eHCC-related TSGs. In HCC cell lines a considerable degree of DNA methylation was observed in these TSGs regardless of HBV status, and their methylation levels were inversely correlated with expression of the corresponding genes (Fig. 2). Therefore, although the analysis of human HCC tissues indicate that methylation levels were higher in HCV-related than HBV-related tumors (Supporting Fig. 3), epigenetic inactivation of eHCC-related TSGs might play a role for both types of virus-related neoplasms. It was reported that cancer-specific promoter methylation mostly targeted genes, which normally had low baseline steady-state levels of expression.16 Therefore, it is possible that unique profiles of methylated genes in HCC, which could be involved in multiple oncogenic pathways, attribute to the baseline expression in normal hepatocytes. However, our results also indicate that infection with different hepatitis-related viruses might cause a selective pressure for DNA methylation-induced transcriptional inactivation of certain TSGs. Genome-wide analyses of epigenetic alterations may help clarify the differences in the methylation profiles of HCC with various etiologies.

Another interesting feature of our study is that some of the eHCC-related TSGs are known to carry low levels of methylation even in the background liver of HCC patients, especially HCV-positive cases. Therefore, it is reasonable to speculate that some of these clones with methylation-silenced TSGs already exist in patients with chronic hepatitis, and may act as “seeds” for the development of early stage tumors.5 As the number of methylated TSGs in hepatitis tissues is related to the risk of HCC, stepwise accumulation of methylation-inactivated genes must be required during the initial steps of hepatocarcinogenesis. The eight genes that showed high levels of methylation in eHCC in our study are involved in a variety of functions, suggesting that multiple pathways might be involved in emergence of eHCC.

A previous report indicated that DNA methylation alterations in HCC had an important prognostic and therapeutic implication for this malignancy.17 In addition, a recent report suggested that aberrant DNA methylation of certain genes could predict recurrence-free survival of HCC patients who underwent hepatectomy.18 However, the determination of recurrence-free survival of HCC is challenging because tumor recurrence typically has two components: true metastasis and metachronous tumors that arise de novo.19 Therefore, analysis of hepatitis tissues without prior history of HCC provide a more logical and prudent substrate to evaluate the usefulness of methylation events at critical TSGs/CpG loci for assessing the risk of HCC development. To the best of our knowledge, none of the previous studies have investigated associations between epigenetic alterations in hepatitis tissues and determined the future risk of developing de novo HCC. This is another unique strength of our study in which we addressed this issue and discovered that an increased number of methylated TSGs was associated with shorter time-to-HCC occurrence. In addition, the increased number of methylated TSGs was an independent risk factor even when the analysis eliminated the effect of interferon therapy (Fig. 4). In contrast, multivariate analysis in our study revealed that F-stage, which was previously believed to a better predictor of HCC in CHC patients, failed to emerge as an independent risk factor when simultaneously analyzed with the number of methylated TSGs. This observation is in line with the report suggesting that CpG island methylator phenotype in HCC might also be associated with background liver in patients with cirrhosis.10 Recently, it was reported that oxidative stress caused by HCV induced liver fibrosis,20 and it also led to epigenetic inactivation of TSGs by way of DNA methylation.21 From this viewpoint, it is conceivable to speculate that epigenetic inactivation of TSGs might be a more critical event that may take precedence over the effect of F-stage on the risk of HCC emergence. Interestingly, we also discovered a dose-dependent effect of the number of methylated TSGs on the HRs for the times-to-HCC occurrence, suggesting that sequential inactivation of these tumor suppressors may lead to the formation of de novo HCC in preneoplastic liver. However, our study cannot rule out the inadvertent bias in patient selection given the retrospective nature of our study, in which CHC cases with higher alanine aminotransferase or α-fetoprotein levels might have been prone to more frequent to liver biopsies. Therefore, we determined the rate of fibrosis progression using F-stages at the time of initial biopsies and HCC occurrence (Supporting Table 5). The mean progression of F-stage was 0.07 ± 0.10 unit/year for all cases and 0.14 ± 0.10 unit/year for noncirrhotic cases, which was somewhat higher than that reported by Shiratori et al.22 However, the mean age and proportion of advanced F-stage in our cohort is also higher than those reported by Shiratori et al., which might help explain the higher incidence of HCC observed in our study. Nonetheless, to address these important issues we are currently planning an independent validation of our results in which we will interrogate the significance of these methylation markers for the prediction of HCC emergence in a prospective multicenter patient cohort.

In summary, in this study we systematically characterized methylation patterns in a large panel of TSGs and identified a subset of genes that play a critical role in the earliest steps in hepatocarcinogenesis. Furthermore, we provide additional evidence that epigenetic inactivation of these genes in patients with CHC is an important and independent risk factor for predicting emergence of HCC late in life. Given the robustness of data presented in this article, we believe that these results are of significance, from both basic and clinical perspectives. From a basic research standpoint, these data provide a molecular understanding on the contribution of individual TSGs that are targets of transcriptional inactivation in the earliest steps of hepatocarcinogenesis. From a clinical viewpoint, in addition to the antiviral and antiinflammatory therapies, our data suggest that potential inclusion of epigenetic therapies in future might serve as effective adjuncts for CHC patients with refractory disease in reducing their risk for developing HCC.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Author contributions: Study concept and design (N.N. and A.G.); acquisition of data (N.N.); analysis and interpretation of data (N.N., I.I., and T.N.); statistical analysis (N.N.); provision of samples (N.N., I.I., and T.N.); drafting of the article (N.N., M.K., and A.G.).

References

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information
  • 1
    Nishida N, Goel A. Genetic and epigenetic signatures in human hepatocellular carcinoma: a systematic review. Curr Genomics 2011; 12: 130-137.
  • 2
    Nishida N, Fukuda Y, Kokuryu H, Toguchida J, Yandell DW, Ikenega M, et al. Role and mutational heterogeneity of the p53 gene in hepatocellular carcinoma. Cancer Res 1993; 53: 368-372.
  • 3
    Nishida N, Nishimura T, Nagasaka T, Ikai I, Goel A, Boland CR. Extensive methylation is associated with beta-catenin mutations in hepatocellular carcinoma: evidence for two distinct pathways of human hepatocarcinogenesis. Cancer Res 2007; 67: 4586-4594.
  • 4
    Kondo Y, Kanai Y, Sakamoto M, Mizokami M, Ueda R, Hirohashi S. Genetic instability and aberrant DNA methylation in chronic hepatitis and cirrhosis—a comprehensive study of loss of heterozygosity and microsatellite instability at 39 loci and DNA hypermethylation on 8 CpG islands in microdissected specimens from patients with hepatocellular carcinoma. HEPATOLOGY 2000; 32: 970-979.
  • 5
    Ushijima T, Okochi-Takada E. Aberrant methylations in cancer cells: where do they come from? Cancer Sci 2005; 96: 206-211.
  • 6
    Nishida N, Nagasaka T, Nishimura T, Ikai I, Boland CR, Goel A. Aberrant methylation of multiple tumor suppressor genes in aging liver, chronic hepatitis, and hepatocellular carcinoma. HEPATOLOGY 2008; 47: 908-918.
  • 7
    International Consensus Group for Hepatocellular Neoplasia. Pathologic diagnosis of early hepatocellular carcinoma: a report of the international consensus group for hepatocellular neoplasia. HEPATOLOGY 2009; 49: 658-664.
  • 8
    Takayasu K, Muramatsu Y, Mizuguchi Y, Moriyama N, Ojima H. Imaging of early hepatocellular carcinoma and adenomatous hyperplasia (dysplastic nodules) with dynamic CT and a combination of CT and angiography: experience with resected liver specimens. Intervirology 2004; 47: 199-208.
  • 9
    [No authors listed.] Intraobserver and interobserver variations in liver biopsy interpretation in patients with chronic hepatitis C. The French METAVIR Cooperative Study Group. HEPATOLOGY 1994; 20: 15-20.
  • 10
    Shen L, Ahuja N, Shen Y, Habib NA, Toyota M, Rashid A, et al. DNA methylation and environmental exposures in human hepatocellular carcinoma. J Natl Cancer Inst 2002; 94: 755-761.
  • 11
    Nagasaka T, Koi M, Kloor M, Gebert J, Vilkin A, Nishida N, et al. Mutations in both KRAS and BRAF may contribute to the methylator phenotype in colon cancer. Gastroenterology 2008; 134: 1950-1960.
  • 12
    Weisenberger DJ, Siegmund KD, Campan M, Young J, Long TI, Faasse MA, et al. CpG island methylator phenotype underlies sporadic microsatellite instability and is tightly associated with BRAF mutation in colorectal cancer. Nat Genet 2006; 38: 787-793.
  • 13
    Shindo M, Di Bisceglie AM, Hoofnagle JH. Long-term follow-up of patients with chronic hepatitis C treated with alpha-interferon. HEPATOLOGY 1992; 15: 1013-1016.
  • 14
    Hua D, Hu Y, Wu YY, Cheng ZH, Yu J, Du X, et al. Quantitative methylation analysis of multiple genes using methylation-sensitive restriction enzyme-based quantitative PCR for the detection of hepatocellular carcinoma. Exp Mol Pathol 2011; 91: 455-460.
  • 15
    Um TH, Kim H, Oh BK, Kim MS, Kim KS, Jung G, et al. Aberrant CpG island hypermethylation in dysplastic nodules and early HCC of hepatitis B virus-related human multistep hepatocarcinogenesis. J Hepatol 2011; 54: 939-947.
  • 16
    Hahn MA, Hahn T, Lee DH, Esworthy RS, Kim BW, Riggs AD, et al. Methylation of polycomb target genes in intestinal cancer is mediated by inflammation. Cancer Res 2008; 68: 10280-10289.
  • 17
    Calvisi DF, Ladu S, Gorden A, Farina M, Lee JS, Conner EA, et al. Mechanistic and prognostic significance of aberrant methylation in the molecular pathogenesis of human hepatocellular carcinoma. J Clin Invest 2007; 117: 2713-2722.
  • 18
    Nagashio R, Arai E, Ojima H, Kosuge T, Kondo Y, Kanai Y. Carcinogenetic risk estimation based on quantification of DNA methylation levels in liver tissue at the precancerous stage. Int J Cancer 2011; 129: 1170-1179.
  • 19
    Llovet JM, Di Bisceglie AM, Bruix J, Kramer BS, Lencioni R, Zhu AX, et al. Design and endpoints of clinical trials in hepatocellular carcinoma. J Natl Cancer Inst 2008; 100: 698-711.
  • 20
    Mormone E, Lu Y, Ge X, Fiel MI, Nieto N. Fibromodulin, an oxidative stress-sensitive proteoglycan, regulates the fibrogenic response to liver injury in mice. Gastroenterology 2012: 142: 612-621.
  • 21
    O'Hagan HM, Wang W, Sen S, Destefano Shields C, Lee SS, Zhang YW, et al., Oxidative damage targets complexes containing DNA methyltransferase, SIRT1, and polycomb members to promoter CpG islands. Cancer Cell 2011; 20: 606-619.
  • 22
    Shiratori Y, Imazeki F, Moriyama M, Yano M, Arakawa Y, Yokosuka O, et al., Histologic improvement of fibrosis in patients with chronic hepatitis C who have sustained response to interferon therapy. Ann Intern Med 2000; 132: 517-524.

Supporting Information

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
HEP_25706_sm_SuppFig1.doc637KSupporting Information Figure 1.
HEP_25706_sm_SuppFig2.doc103KSupporting Information Figure 2.
HEP_25706_sm_SuppFig3.doc54KSupporting Information Figure 3.
HEP_25706_sm_SuppTab1.doc28KSupporting Information Table 1.
HEP_25706_sm_SuppTab2.doc77KSupporting Information Table 2.
HEP_25706_sm_SuppTab3.doc23KSupporting Information Table 3.
HEP_25706_sm_SuppTab4.doc43KSupporting Information Table 4.
HEP_25706_sm_SuppTab5.doc30KSupporting Information Table 5.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.