Aberrant methylation of multiple tumor suppressor genes in aging liver, chronic hepatitis, and hepatocellular carcinoma†
Article first published online: 26 DEC 2007
Copyright © 2007 American Association for the Study of Liver Diseases
Volume 47, Issue 3, pages 908–918, March 2008
How to Cite
Nishida, N., Nagasaka, T., Nishimura, T., Ikai, I., Boland, C. R. and Goel, A. (2008), Aberrant methylation of multiple tumor suppressor genes in aging liver, chronic hepatitis, and hepatocellular carcinoma. Hepatology, 47: 908–918. doi: 10.1002/hep.22110
Potential conflict of interest: Nothing to report.
- Issue published online: 26 FEB 2008
- Article first published online: 26 DEC 2007
- Manuscript Accepted: 24 OCT 2007
- Manuscript Received: 6 AUG 2007
- National Cancer Institute, National Institutes of Health. Grant Numbers: R01 CA72851, R01 CA98572
- Baylor Research Institute
Aberrant DNA methylation is an important epigenetic alteration in hepatocellular carcinoma (HCC). However, the molecular processes underlying the methylator phenotype and the contribution of hepatitis viruses are poorly understood. The current study is a comprehensive methylation analysis of human liver tissue specimens. A total of 176 liver tissues, including 77 pairs of HCCs and matching noncancerous liver and 22 normal livers, were analyzed for methylation. Methylation of 19 epigenetic markers was quantified, and the results were correlated with different disease states and the presence or absence of hepatitis B virus (HBV) and hepatitis C virus (HCV) infections. Based on methylation profiles, the 19 loci were categorized into 3 groups. Normal liver tissues showed methylation primarily in group 1 loci (HIC-1, CASP8, GSTP1, SOCS1, RASSF1A, p16, APC), which was significantly higher than group 2 (CDH1, RUNX3, RIZ1, SFRP2, MINT31) and group 3 markers (COX2, MINT1, CACNA1G, RASSF2, MINT2, Reprimo, DCC) (P < 0.0001). Noncancerous livers demonstrated increased methylation in both group 1 and group 2 loci. Methylation was significantly more abundant in HCV-positive livers compared with normal liver tissues. Conversely, HCC showed frequent methylation at each locus investigated in all 3 groups. However, the group 3 loci showed more dense and frequent methylation in HCV-positive cancers compared with both HBV-positive cancers and virus-negative cancers (P < 0.0001). Conclusion: Methylation in HCC is frequent but occurs in a gene-specific and disease-specific manner. Methylation profiling allowed us to determine that aberrant methylation is commonly present in normal aging livers, and sequentially progresses with advancing stages of chronic viral infection. Finally, our data provide evidence that HCV infection may accelerate the methylation process and suggests a continuum of increasing methylation with persistent viral infection and carcinogenesis in the liver. (HEPATOLOGY 2008.)
Aberrant hypermethylation in the promoter regions of tumor suppressor genes is a crucial epigenetic alteration that involves the deregulation of many cellular processes that lead to the initiation and progression of human cancers.1–3 Several studies have suggested that methylation of multiple tumor suppressor genes in hepatocellular carcinomas (HCCs) may contribute to the pathogenesis of this disease.4–6 Such epigenetic defects also have been observed in noncancerous liver tissues of HCC patients, supporting the concept that methylation-induced silencing may play a role in the early stages of HCC.7 Because the non-neoplastic liver tissue in HCC patients is usually accompanied by chronic inflammation, it is conceivable that aberrant methylation seen in the surrounding liver tissue corresponds to the “field defect” that has been reported for colon and lung cancers.8, 9
Unlike genetic alterations such as mutations and deletions, epigenetic changes are potentially reversible. Exploiting this epigenetic characteristic, several clinical trials are underway evaluating the potential for cancer prevention and therapy through the reversal of methylation-induced alterations.10–12 Our current understanding suggests that some of the aberrant methylation observed in human cancer may be a consequence of the normal aging process, persistent viral infections, and chronic inflammation.2, 13, 14 Therefore, it is important to have a detailed understanding of the status of the promoters of genes known to be methylated in association with aging and chronic viral infection.
In this study, we quantified methylation densities at 19 CpG loci using combined bisulfite restriction analysis (COBRA) in the liver tissues of patients at various ages and compared this with the degrees of methylation seen in HCC and the corresponding nonneoplastic liver tissues. Herein, we report that aberrant methylation of a limited number of loci is commonly seen in the normal aging liver and that these epigenetic alterations gradually progress and expand to a larger panel of methylation markers in HCC. Additionally, we observed that persistent viral infection, particularly hepatitis C, accelerates age-related methylation in the liver, suggesting that this may play an important role in the pathogenesis of HCC.
Patients and Methods
We analyzed a total of 176 liver tissues, including 77 pairs of HCC and the matched corresponding noncancerous liver tissues, and 22 normal liver tissues. The HCC and noncancerous liver tissues were frozen at −80°C immediately after resection. The hepatitis virus status of patients with HCC were as follows. Thirteen patients were positive for hepatitis B virus surface antigen, 44 were hepatitis C virus (HCV) antibody positive, 2 were positive for both hepatitis B surface antigen and HCV antibody, and 18 were negative for both. The details of the clinical data from the patients, classified by virus status, are listed in Table 1. Among the 22 normal liver tissues, 19 specimens came from patients who had colon cancer with hepatic metastasis. The remaining normal liver tissues were from focal nodular hyperplasia, a hepatic hemangioma, and a hepatic adenoma. All normal liver tissues were confirmed to be free of serum hepatitis B surface antigen and HCV antibody, to have normal serum alanine aminotransferase levels, and to have normal blood platelet counts. Histological data were available for normal livers and demonstrated no evidence of fibrosis or inflammation. Nine of the 22 normal livers were fresh-frozen tissues that were stored at −80°C after surgery. The remaining 13 specimens were formalin-fixed paraffin-embedded samples. Written informed consent was obtained from each patient, and the study was approved by the institutional review boards of all the involved institutions. Although the ages of the patients with hepatitis B virus (HBV)-related HCC tended to be younger compared with the other groups, there were no statistical differences in the distributions of any clinical data among the HCV-related, HBV-related, virus-negative HCC groups and the normal liver group (Table 1).
|Feature||HCC and Their Corresponding Noncancerous Liver||Normal Liver Cases (n = 22)|
|HCV* (n = 46)||HBV* (n = 15)||NBNC* (n = 18)|
|Mean (95% CI)||64.2 (61.8–66.7)||53.8 (47.2–60.4)||61.5 (55.1–67.9)||56.5 (50.0–63.0)|
Methylation Quantification Using COBRA.
Genomic DNA was extracted using QIAamp DNA Mini Kit (Qiagen Inc., Valencia, CA) and TaKaRa DEXPAT kits (Takara Bio Inc., Otsu, Japan) for frozen tissues and paraffin-embedded samples, respectively. Approximately 2 μg DNA was subjected to bisulfite modification treatment to convert all the unmethylated cytosines to thymines. We have previously reported that the 19 methylation loci selected for this study exhibited significantly higher methylation levels in HCC compared with corresponding noncancerous liver tissues.15 The methylation loci included 16 gene promoters: HIC-1, CASP8, GSTP1, SOCS1, RASSF1A, p16, APC, RUNX3, RIZ1, SFRP2, CDH1, COX2, CACNA1G, RASSF2, Reprimo, DCC, and 3 MINT loci: MINT1, MINT2, MINT31. Primer sequences, polymerase chain reaction (PCR) conditions, and restriction enzymes used for methylation analysis of p16, CACNA1G, DCC, MINT1, MINT2 and MINT31, were reported previously.16, 17 The remaining assays were designed in our laboratory, and this information is available on request. Each PCR reaction was performed in a total volume of 25 μL, which contained 12.5 μL HotStarTaq Master Mix (Qiagen), 40 ng bisulfite-treated DNA template, and 0.2 μM of each primer pair. After PCR amplification, 5 μL of the amplified products were subjected to 5 units of restriction enzyme digestion for the determination of the degree of methylation at each CpG locus. Digested PCR products were electrophoresed on 2.5% agarose gels and subsequently visualized by ethidium bromide staining. Each assay included a positive control DNA sample that was treated with CpG methylase (CpGenome Universal Methylated DNA; Chemicon International Inc., Temecula, CA), as well as a negative control comprising normal lymphocytic and fibroblast DNA. Band intensities of digested or undigested PCR products were determined using a Kodak Gel-Logic 200 Imaging System (Eastman Kodak Co., Rochester, NY). The band intensities of restriction enzyme digested PCR products (implying a methylated product) were divided by the total of both band intensities, and quantity of methylation was represented as percentage density as described.18 To account for differences attributable to incomplete digestion by the restriction enzyme, the percentage methylation of each tumor DNA sample was normalized to that of CpG methylase-treated DNA, which should theoretically result in 100% methylation.15 The sensitivity of COBRA to detect methylated alleles in our study was as low as 2%, because we could not distinguish band intensities less than 2% from the background staining. Therefore, percent methylation level of greater than 2% was defined as “positive for methylation” to determine frequencies of methylation of each locus. To validate and ensure the sensitivity and specificity of our COBRA analyses, we designed 2 independent primer pairs (around the same CpG site) for multiple genes/loci. We compared methylation levels obtained from these 2 assays and discovered that in each instance, both primers yielded concordant results regardless of their primer sequences and PCR condition, suggesting the specificity and sensitivity of our COBRA analyses (Supplementary Fig. 1). Additionally, we compared our COBRA data with pyrosequencing analyses for some of the genes and found concordant methylation results by these 2 independent methodologies.
For comparing age differences between groups of HCV-related, HBV-related, virus-negative patients with HCC and normal liver cases, both Student t test and the Wilcoxon rank-sum test were used. For categorical comparisons of clinical data, the chi-squared test or Fisher's exact test were used. The Wilcoxon rank-sum test was also applied to compare the methylation level of each locus between any 2 categorical variables. To examine relationships between age and methylation levels of each locus in normal liver, Pearson's correlation and Spearman correlation tests were applied. Spearman's rank correlation test was also used to examine the relationship between methylation densities at different loci. Because hierarchical clustering analysis is most appropriate to statistically discriminate HCC according to methylation levels of 19 loci, we used this approach to classify all HCC into 2 distinct groups (group A and group B). Afterwards, the chi-squared test was used to compare the virus status in the 2 HCC groups. To compare the differences in methylation levels among 3 categorical variables, each of which contained methylation levels of different loci, we calculated Z-scores for normalization. Subsequently, 1-way factorial analysis of variance (ANOVA) and post hoc comparisons (Fisher's PLSD) as well as Kruskal-Wallis tests were performed based on these Z-scores. All P-values were 2-sided, and P < 0.05 was considered statistically significant. All statistical analyses were calculated using JMP version 4.05J software (SAS Institute Inc., Cary, NC).
Gene-Specific Methylation Patterns in Liver Tissues from Normal, Noncancerous, and HCC Patients Allows Classification of Methylation Markers into Distinct Groups.
In this study, we analyzed methylation frequencies and densities of 19 methylation targets in liver tissue DNA obtained from a series of normal and HCC patients. Methylation analysis of normal liver tissues that were negative for the presence of both HBV and HCV, and showed neither inflammation nor fibrosis, demonstrated low levels of methylation at some of the loci. To determine whether differences in sample processing for frozen and paraffin-embedded tissues may have affected our results, we obtained both types of tissues from a subset of the patients and performed COBRA analyses on some of the Group 1 loci/markers. We noted that the methylation levels at each marker for a given patient were comparable in DNA obtained from either type of tissue, suggesting that sample processing did not affect our methylation results (data not shown). The overall methylation frequencies and densities in normal liver tissues were significantly lower in comparison with HCC as well as noncancerous liver tissues from patients with HCC. Analyzing 22 normal livers for methylation at 19 loci, we found that 7 of 19 markers (HIC-1, CASP8, GSTP1, SOCS1, RASSF1A, p16, and APC) showed methylation, and the frequencies of methylation ranged between 27.3% and 72.7%, and the mean percent methylation ranged between 2.7 [95% confidence interval (CI), 0.6–4.7 for p16] and 10.5 (95% CI, 4.1–16.9 for SOCS1) (Table 2). Conversely, the remaining 12 methylation loci were either completely unmethylated or showed a negligible degree of methylation. Based on the clear distinction between frequently methylated and unmethylated loci in normal liver tissues, initially, we categorized the 7 loci with methylation as group 1 methylation markers. The distributions of percent methylation of group 1 markers were clearly different from those of the remaining 12 loci (Fig. 1A). Calculating the Z-scores for methylation (Fig. 1D), we found that the methylation of group 1 loci in normal livers was significantly higher compared with the other 2 groups [analysis of variance (ANOVA); F (2, 415) = 46.7, P < 0.0001, Kruskal-Wallis test; P < 0.0001].
|Locus||Normal Liver (n = 22)||Noncancerous Liver From HCC Patients (n = 77)||P*|
|No. of Methylated Samples (%)||Mean % Level (95% CI)||No. of Methylated Samples (%)||Mean % Level (95% CI)|
|HIC-1||16 (72.7)||7.2 (4.1–10.3)||49 (63.6)||6.9 (5.4–8.4)||NS|
|CASP8||13 (53.1)||5.8 (2.9–8.8)||26 (33.8)||7.0 (4.4–9.6)||NS|
|GSTP1||13 (53.1)||5.2 (2.8–7.5)||27 (35.1)||7.8 (4.7–10.9)||NS|
|SOCS1||10 (45.5)||10.5 (4.1–16.9)||53 (68.9)||16.9 (13.5–20.3)||NS|
|RASSF1A||10 (45.5)||6.3 (2.1–10.4)||43 (55.8)||11.9 (8.2–15.7)||NS|
|P16||7 (31.8)||2.7 (0.6–4.7)||25 (32.5)||3.2 (1.8–4.6)||NS|
|APC||6 (27.3)||3.3 (0.7–5.9)||23 (30.0)||7.1 (4.3–9.8)||NS|
|CDH1||4 (18.2)||1.2 (−0.4–2.8)||21 (27.8)||1.8 (1.0–2.5)||NS|
|RUNX3||0||0||14 (18.1)||1.3 (0.4–2.2)||.0324|
|RIZ1||1 (4.5)||0.2 (−0.2–0.6)||12 (15,6)||2.4 (1.0–3.7)||NS|
|SFRP2||3 (13.6)||1.5 (−0.3–3.3)||10 (13.0)||2.2 (0.7–3.7)||NS|
|MINT31||2 (9.1)||1.7 (−1.1–4.5)||9 (11.7)||2.3 (0.6–3.9)||NS|
|COX2||2 (9.1)||0.4 (−0.2–0.9)||0||0||NS|
|MINT1||2 (9.1)||0.8 (−0.3–1.9)||0||0||NS|
|CACNA1G||1 (4.5)||0.2 (−0.2–0.5)||0||0||NS|
|RASSF2||0||0||2 (2.6)||0.3 (−0.1–0.8)||NS|
|Reprimo||0||0||2 (2.6)||0.4 (−0.2–0.9)||NS|
|DCC||0||0||1 (1.3)||0.1 (−0.1–0.4)||NS|
We next categorized the remaining 12 methylation loci into 2 groups based on methylation frequencies of the noncancerous liver tissues from HCC patients. Methylation analysis of 77 noncancerous liver tissues revealed that only 5 of 12 loci (CDH1, RUNX3, RIZ1, SFRP2, and MINT31) were more frequently methylated in these tissues, with the frequencies ranging from 11.7% to 27.8%, and the mean methylation ranging from 2.4 (95% CI, 1.0–3.7 for RIZ1) to 1.3 (95% CI, 0.4–2.2 for RUNX3) (Table 2, Fig. 1B). Conversely, limited evidence of methylation was observed at the remaining 7 of 12 markers (COX2, MINT1, CACNA1G, RASSF2, MINT2, Reprimo and DCC) in these tissues. Comparing Z-scores, it was clear that the methylation frequencies of these 5 markers segregated with noncancerous liver tissues, because the methylation levels of these markers were significantly higher than those of the remaining 7 markers (ANOVA; F (2, 1460) = 153.5; P < 0.0001, Kruskal-Wallis test; P < 0.0001) (Fig. 1E). We classified these 5 loci as group 2 markers and the remaining 7 markers as group 3.
Analysis of HCC showed frequent and dense methylation at all loci. Of note, only HCC showed methylation of the 7 loci categorized as group 3. However, collective analysis of the methylation data from all 19 markers revealed that the methylation frequencies were highest in group 1 and least in group 3 loci (ANOVA; F (2, 1536) = 207.4, P < 0.0001, Kruskal-Wallis test; P < 0.0001; the difference was significant in every pair by post hoc comparisons) (Fig. 1C, F).
Age-Related Methylation Is Present in Normal Liver.
As discussed previously, only group 1 loci demonstrated methylation in pathologically and clinically normal liver tissues. These data supported the hypothesis that low levels of methylation accrue as a function of age in the normal liver. We therefore analyzed the data using Pearson's as well as Spearman's correlation tests and calculated the ‘γ’ and ‘ρ’ values respectively. Interestingly, we observed that the percent methylation positively correlated with age at all 7 group 1 loci (r values ranged from 0.65 to 0.44; ρ values ranging from 0.73 to 0.28) (Fig. 2). In addition, we observed that methylation levels of all group 1 loci in patients ages 65 or older were higher compared with those who were younger than 65 years of age in normal liver samples (statistically significant for all markers but GSTP1; Supplementary Table 1). These data demonstrated that the methylation of tumor suppressor genes that are reportedly responsible for the development of HCC also takes place in the normal liver as it ages.
Simultaneous HCV Infection Enhances Age-Related Methylation.
We next determined whether hepatitis viruses have any influence on the epigenetic alterations in hepatocarcinogenesis. Figure 3 illustrates the status at all methylation loci within the 3 groups in each of the noncancerous livers, as well as HCC tissues. These tissues have further been classified according to presence or absence of the hepatitis viruses, HCV and HBV. We observed that HCV-positive noncancerous liver tissues were more highly methylated at group 1 and group 2 loci in comparison with HBV-positive or virus-negative cases. In addition, methylation of group 3 loci was rare in noncancerous liver tissues regardless of the viral status. In HCC tissues, although methylation of group 1 loci was more frequent and dense regardless of viral status, HCV-related HCC tended to have higher methylation levels at both group 2 and group 3 loci compared with virus-negative HCC.
To determine whether the presence of viral hepatitis might enhance age-related methylation detected in normal liver, we compared methylation of group 1 and group 2 loci between normal liver and noncancerous liver tissues classified according to viral status using the Wilcoxon rank-sum test. Only 1 locus showed significant differences in methylation levels between normal liver and HBV-positive noncancerous livers (SOCS1; P = 0.0193, Supplementary Table 2), whereas none of the loci showed a significantly higher methylation level in virus-negative noncancerous liver when compared with normal liver. Conversely, 3 of 7 group 1 loci (SOCS1, RASSF1A, and APC) and 3 of 5 group 2 loci (CDH1, RUNX3, and RIZ1) showed significantly higher levels of methylation in HCV-positive noncancerous tissues than in normal liver (Table 3). Although it did not reach significance, the distribution of methylation levels (mean, median, and maximum percent methylation) at the remaining 4 group 1 and 2 group 2 loci was higher in HCV-positive noncancerous liver than in normal liver (data not shown). Next, to determine whether HCV infection dominates the age effect in terms of progression of methylation, we made 2 separate comparisons; first, we examined the relationship between age and total number of methylation events in group 1 loci for HCV-positive noncancerous tissues; secondly, we analyzed the relationship between blood platelet counts and methylation events in group 1 loci. We did not find any significant relationship between the number of group 1 methylated loci and age in HCV-positive cases (categorized as ≥65 years of age versus < 65 years of age; P = 0.2937 by Kruskal-Wallis test). However, interestingly enough, patients with lower platelet counts tended to carry increased numbers of methylated loci in their HCV-positive noncancerous livers (platelet count of more than 18 × 104/mL versus less than 18 × 104/mL; P = 0.0858 by Kruskal-Wallis test). Because platelet count is known to correlate inversely with increased fibrosis stage in patients with chronic hepatitis C, it would seem that methylation events may be influenced as a function of HCV infection rather than aging in HCV-positive cases.
|Locus||Normal Liver (n = 22)||HCV-Positive Noncancerous Liver (n = 46)||P*|
|No. of Methylated Samples (%)||Mean % Level (95% CI)||No. of Methylated Samples (%)||Mean % Level (95% CI)|
|HIC-1||16 (72.7)||7.2 (4.1–10.3)||36 (78.3)||9.1 (7.1–11.0)||0.1741|
|CASP8||13 (53.1)||5.8 (2.9–8.8)||20 (43.5)||10.0 (6.2–13.9)||0.7670|
|GSTP1||13 (53.1)||5.2 (2.8–7.5)||19 (41.3)||10.8 (6.1–15.6)||0.6545|
|SOCS1||10 (45.5)||10.5 (4.1–16.9)||34 (73.9)||20.2 (15.7–24.6)||0.0242|
|RASSF1A||10 (45.5)||6.3 (2.1–10.4)||28 (60.9)||16.1 (10.4–21.9)||0.0353|
|P16||7 (31.8)||2.7 (0.6–4.7)||21 (45.7)||4.9 (2.7–7.1)||0.2800|
|APC||6 (27.3)||3.3 (0.7–5.9)||21 (45.7)||11.5 (7.3–15.7)||0.0303|
|CDH1||4 (18.2)||1.2 (−0.4–2.8)||20 (43.5)||2.9 (1.8–4.1)||0.0354|
|RUNX3||0||0||11 (23.9)||1.9 (0.4–3.4)||0.0134|
|RIZ1||1 (4.5)||0.2 (−0.2–0.6)||11 (23.9)||3.8 (1.6–6.0)||0.0411|
|SFRP2||3 (13.6)||1.5 (−0.3–3.3)||9 (19.6)||3.4 (1.0–5.9)||0.4838|
|MINT31||2 (9.1)||1.7 (−1.1–4.5)||7 (15.2)||2.7 (0.7–4.6)||0.5043|
Increased Methylation Associates With HCV Infection in HCC.
It was critical to understand the impact of hepatitis virus infection on the progression of methylation from a noncancerous stage to cancer. To answer this question, we calculated the difference of methylation between the noncancerous liver and HCC in the same patient at every locus. Using hierarchical clustering analysis, we were able to classify all HCC cases according to differences in methylation. As shown in Fig. 4, 39 of 75 HCCs were classified as group A, and 36 were segregated into group B. We excluded the 2 patients from analysis that were positive for the simultaneous presence of both HCV and HBV. We then analyzed differences in methylation between the HCC and noncancerous tissue at each locus in group A and group B (Supplementary Table 3). All but 5 loci showed significantly more methylation in the group B cancers in comparison with group A neoplasms (P value ranged from <0.0001 to 0.0169), indicating that group B cancers represented a subset of HCC with enhanced progression of methylation from surrounding noncancerous liver.
Among 36 group B HCCs, 29 (81%) were HCV positive, 5 (14%) were HBV positive, and 2 (5%) were virus-negative. Among the 39 group A HCCs, 15 (38%) were HCV positive, 8 were HBV positive (21%), and 16 (41%) were virus negative (P = 0.0002 by chi-squared test; Table 4). In addition, the proportion of HCV-related HCC was significantly higher in group B compared with group A cancers (HCV-related versus HBV-related or virus-negative; P = 0.0002 by chi-squared test). Group B HCC was predominantly virus positive compared with group A neoplasms. In group B, 34 of 36 (94%) were either HCV-positive or HBV-positive, and only 2 of 36 (6%) were virus-negative, whereas in group A, 23 of 39 HCCs (59%) were virus positive, and 16 of 39 (41%) were virus negative (P = 0.0001 by chi-squared test).
|Group A (n = 39)||15 (38%; 15/39)||8 (21%; 8/39)||16 (41%; 16/39)|
|Group B (n = 36)||29 (81%; 29/36)||5 (14%; 5/36)||2 (5%; 2/36)|
Table 5 summarizes comparisons of methylation status between HCV-related and virus-negative HCCs in all 19 loci. Among the 19 loci, 1 of 7 group 1 loci (GSTP1), 3 of 5 group 2 loci (RIZ1, RUNX3, MINT31), and 6 of 7 group 3 loci (CACNA1G, COX2, RASSF2, MINT1, MINT2, and Reprimo) showed significantly higher levels of methylation in HCV-related HCCs than virus-negative HCC. Conversely, 1 group 1 locus (GSTP1) and 1 group 2 locus (RIZ1), but no group 3 loci represented higher methylation levels in HBV-related HCCs compared with virus-negative tumors (Supplementary Table 4). Next, we compared the overall distribution of methylation level at the 3 groups of methylation loci expressed as Z-scores in HCV, HBV, and virus-negative HCC. Using ANOVA and post-hoc comparisons (Fisher's PLSD), methylation of group 1 loci was significantly higher in HCV-related HCC than in virus-negative HCC (Fig. 5A) (ANOVA; F(2, 522) = 4.34, P = 0.0134, Kruskal-Wallis test; P = 0.0168). Similarly, for group 2 loci, methylation was more prominent in HCV-related or HBV-related HCC compared with non-B, non-C–related HCC (Fig. 5B) (ANOVA; F(2, 372) = 12.17, P < 0.0001, Kruskal-Wallis test; P < 0.0001). Methylation of group 3 was more prominent in HCV-related HCC compared with HBV-related or virus-negative HCC (Fig. 5C) (ANOVA; F(2, 522) = 18.31, P < 0.0001, Kruskal-Wallis test, P < 0.0001).
|Locus||HCV-Related HCC (n = 44)*||Virus-Negative HCC (n = 18)||P†|
|No. of Methylated Samples (%)||Mean % Level (95% CI)||No. of Methylated Samples (%)||Mean % Level (95% CI)|
|HIC-1||38 (86.4)||34.0 (27.2–40.9)||14 (77.8)||25.7 (15.4–36.1)||0.2109|
|CASP8||19 (43.2)||17.8 (10.2–25.3)||9 (50.0)||15.5 (6.1–24.9)||0.9526|
|GSTP1||38 (86.4)||59.0 (49.1–68.9)||13 (72.2)||34.8 (18.7–50.9)||0.0155|
|SOCS1||29 (65.9)||40.1 (29.6–50.6)||12 (66.7)||37.6 (21.2–54.1)||0.7758|
|RASSF1A||38 (86.4)||33.4 (26.1–40.7)||16 (88.9)||28.3 (18.5–38.1)||0.5814|
|P16||35 (79.5)||32.2 (24.6–39.7)||12 (66.7)||22.1 (10.6–33.6)||0.2030|
|APC||40 (90.9)||52.5 (45.6–59.5)||14 (77.8)||41.6 (28.2–55.1)||0.1947|
|CDH1||20 (45.5)||4.5 (2.5–6.5)||7 (38.9)||2.8 (0.8–4.9)||0.5604|
|RUNX3||36 (81.8)||23.7 (16.7–30.8)||7 (38.9)||8.1 (1.5–14.6)||0.0029|
|RIZ1||36 (4.5)||33.4 (25.9–40.9)||6 (33.3)||8.8 (0.9–16.7)||0.0002|
|SFRP2||13 (29.5)||9.8 (3.8–15.8)||2 (11.1)||2.4 (−1.4–6.2)||0.1167|
|MINT31||33 (75.0)||28.3 (19.9–36.7)||7 (38.9)||9.9 (1.9–18.0)||0.0063|
|COX2||20 (45.5)||16.0 (8.9–23.2)||3 (16.7)||4.1 (−0.8–9.0)||0.0355|
|MINT1||20 (45.5)||12.9 (7.3–18.5)||1 (5.6)||0.7 (−0.7–2.1)||0.0024|
|CACNA1G||30 (88.6)||18.8 (11.2–26.3)||6 (33.3)||0.7 (−0.7–6.6)||0.0061|
|RASSF2||21 (47.7)||13.4 (8.0–18.8)||1 (5.6)||1.8 (−2.0–5.5)||0.0029|
|MINT2||17 (38.6)||9.4 (4.9–14.0)||0||0||0.0025|
|Reprimo||13 (29.5)||6.4 (2.6–10.2)||1 (5.6)||0.4 (−0.5–1.4)||0.0315|
|DCC||4 (9.1)||1.6 (−0.1–3.2)||1 (5.6)||1.3 (−1.5–4.1)||0.6694|
Epigenetic instability characterized by methylation of multiple cancer-related genes is gaining recognition as a key mechanism of tumor suppressor gene silencing in many human cancers, including HCC.19, 20 In this study, we performed a detailed quantitative methylation analysis of a large number of methylation loci in normal aging liver, noncancerous liver tissues from HCC patients, and neoplastic HCC tissues. The results presented herein clearly demonstrate that some degree of methylation occurs in the context of the normal aging process in the liver, and it is likely that some of these methylation events may sequentially progress and participate in the development of hepatic neoplasia. Another important observation is that the presence of hepatitis viruses, especially HCV, could play a role in accelerating the methylation process that is involved in HCC development.
In this study, we first categorized all CpG methylation loci into 3 groups according to the frequencies of methylation in various subsets of liver tissues to better understand gene-specific characteristics of individual CpG loci. This categorization allowed us to appreciate that group 1 loci demonstrate methylation in the normal liver in association with increasing age, suggesting that these loci may get methylated as a function of age, because these may be weakly protected in the aging liver against methylation alterations.1 Conversely, we noted hypermethylation of both group 1 and group 2 loci in noncancerous liver. HCC tissues carried much more frequent and dense methylation at all methylation markers, but the group 3 markers were solely methylated in HCC. The differences in frequency and pattern of methylation among various stages of liver disease suggest that methylation progresses sequentially from group 1 to group 3 loci with the progression of disease.
We observed significantly higher levels of methylation in HCC as well as the corresponding non-cancerous liver tissues compared to normal tissue. We also found that the background liver tissues in HCC patients, where chronic viral infection and inflammation are common, also carried concordant methylation of multiple genes. Because HCV is a common cause of chronic liver disease and HCC worldwide, we speculated that chronic infection by HCV might accelerate the methylation process during hepatocarcinogenesis. Comparison of methylation levels between normal liver and noncancerous livers of HCV-positive patients demonstrated that several markers showed significantly higher methylation in HCV-positive cases, which contrasts with HBV-positive and virus-negative cases. More frequent and denser methylation of all 19 loci was a characteristic feature of HCC; however, clear differences in methylation were also evident between HCV-positive and virus-negative HCC in all 3 subgroups of methylation loci.
Previous reports have suggested that methylation of GSTP1 and p16 are frequent in HBV-related HCC.21, 22 In our study, significant differences in methylation density were detected at GSTP1 and RIZ1 between HBV-related and virus-negative HCC. In addition, overall methylation levels of group 2 loci were significantly higher in HBV-related HCC than in virus-negative HCC. Because HBV-related HCC patients tend to be younger than virus-negative HCC patients, and because methylation events could be affected by the aging process, HBV infection also may play a role in the progression of methylation in these patients. Conversely, the highest levels of methylation, which correlated with group 3 loci, appeared to be cancer-specific and were unique for HCV-related HCC only. Previous reports suggest that methylation of SOCS-1 and APC, p15 was more frequently observed in HCV-related HCCs than in virus-negative HCCs,6 whereas another report failed to show a clear relationship between methylation of specific loci and HCV in HCC tissues.5 However, most of the previous studies were reported based on nonquantitative methylation assays. Because the biological meaning of methylation may be attributed according to its density at a given locus, in the current study we performed quantitative methylation analyses for individual genes. Based on our data, we could clearly demonstrate methylation differences in HCV-positive, HBV-positive, and virus-negative cases. In addition, we classified all HCCs using hierarchical clustering analysis to account for the differences in methylation between HCCs and the noncancerous liver tissues. Even using this approach, it was clear that infection with hepatitis virus, especially HCV, strongly associated with the progression of methylation.
Our study showed that group 3 loci carried more methylation in HCV-related HCCs than HBV-related and virus-negative HCCs. This was not merely attributed to age, because even though HCV-positive cases were older than HBV-positive cases, there were no age differences between HCV-positive and virus-negative HCC cases. In addition, we could not find any association between age and methylation level of group 3 loci for HCC tissues in the HCV-positive group (data not shown). Similarly, increased methylation levels at group 1 and group 2 loci in HCV-positive noncancerous liver could not be attributed to age because there were no age differences between HCV-positive and virus-negative patients. As it is believed that disruption of the balance between methylation pressure and the protective mechanisms may be responsible for the induction of aberrant DNA methylation,1 we can speculate that protection against the spread of methylation is weaker at group 1 loci and strongest at group 3 loci. However, chronic HCV infection may act as a powerful epi-mutagen and may induce methylation even at group 3 loci, making the group 3 loci unique HCV-associated events in hepatic carcinogenesis.
The data presented here also suggest that methylation causes disruption of a variety of genes and pathways during hepatocarcinogenesis, such as RB-related (p16, RIZ1), p53-related (HIC-1, Reprimo), WNT/APC (APC, SFRP2, CDH1), receptor-tyrosine kinase-associated (RASSF1A, RASSF2, SOCS-1), transforming growth factor beta signaling (RUNX3), and apoptosis-related pathways (CASP8). These disruptions are supposed to act in concert and may play an active role in HCC. It is well known that tumor cells need to accumulate several rate-limiting mutations for cancer development. In this regard, we can speculate that continuous exposure to an epi-mutagen, such as HCV, will induce the disruption of multiple genes and pathways for cancer development.
In conclusion, we have provided data to suggest a potentially novel sequence of epigenetic changes that may conspire during HCC development. This process appears to be associated with viral infection and is more prominent in HCV than in HBV. Because HCV is a major cause of HCC, these data have clinical implications for the prevention of HCC, because epigenetic alterations are potentially reversible.12 We propose that the data presented here provide clues to develop improved risk assessment markers, and insight into potential prevention strategies for the subset of HCCs that develop through the epigenetic pathway.
- 7Genetic 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., , , , , .
Supplementary material for this article can be found on the H EPATOLOGY Web site ( http://interscience.wiley.com/jpages/0270-9139/suppmat/index.html ).
|hep22110-SupplTable1.pdf||9K||Supplemental Table 1:Comparison of methylation status between normal liver at age ≥65 yo and those <65 years of age|
|hep22110-SupplTable2.pdf||10K||Supplemental Table 2:Comparison of methylation status between normal liver and HBV-positive non-cancerous liver of HCC patients|
|hep22110-SupplTable3.pdf||11K||Supplemental Table 3:Differences in methylation levels at each locus between HCC and corresponding non-cancerous tissues in Group-A and Group-B HCC|
|hep22110-SupplTable4.pdf||13K||Supplemental table 4:Comparison of methylation status between HBV-related HCC and virus-negative HCC|
|hep22110-SupplementalFig.1.pdf||413K||Suppl Fig 1: Comparison of methylation densities|
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.