Quantitative promoter methylation analysis of hepatocellular carcinoma, cirrhotic and normal liver

Authors


Abstract

Hepatocellular carcinoma (HCC) is one of the most common cancers worldwide. Little is known about its molecular pathogenesis and the relevance of DNA methylation for disease initiation and progression. Nevertheless, promoter methylation of some genes has been implicated as potential marker for HCC. Thirty-four HCC, 34 matching non-malignant, cirrhotic livers and 16 normal livers were analyzed for the methylation status of the genes p16INK4a, GSTP1, MGMT, DAP-K and APC by quantitative methylation-specific PCR. DNA promoter methylation frequencies in HCC and matching non-malignant cirrhotic liver, respectively, were as follows: p16INK4a (76% vs. 24%), GSTP1 (53% vs. 32%), MGMT (6 vs. 12%), DAP-K (68 vs. 100%) and APC (100 vs. 100%). GSTP1 and/or p16INK4a promoter methylation was observed in 88% of the HCC samples. In normal liver tissue, the p16INK4a, GSTP1 and MGMT promoter were not methylated. DAP-K was methylated in 31% and APC even in 100% of normal liver samples. Quantitative levels of methylated promoter DNA of all genes were significantly different in the various tissue types except for MGMT. Our results suggest that promoter methylation of tumor-associated genes is a common event in hepatocarcinogenesis. Significantly, higher levels and frequencies of promoter methylation in HCC were found for p16INK4a and GSTP1 compared to non-malignant cirrhotic liver. This indicates that these epigenetic events may serve as a good marker for HCC. These data also demonstrate the importance of the quantification of methylated promoter DNA within a given sample and the use of normal tissue as controls. Quantitative analyses of methylated GSTP1 and p16INK4a promoter may serve as a powerful molecular marker in detecting HCC in biopsies. © 2008 Wiley-Liss, Inc.

Hepatocellular carcinoma (HCC) is the third most deadly cancer worldwide with increasing incidence in many countries including Japan, USA and Europe.1, 2 HCC is one of the few human cancers for which the etiology of the underlying disease is known at least in most cases, i.e., chronic viral hepatitis B or C, alcohol use or hereditary disorders. Nevertheless, it is not clear why the HCC risk is different for different etiologies. Furthermore, little is known about the molecular pathogenesis of HCC development. Phenotypically, liver inflammation leads to fibrosis, which progresses to cirrhosis. In cirrhosis, hepatocytes form dysplastic nodules and finally HCC lesions. The molecular mechanisms underlying disease progression vary with respect to etiology or race and even between different HCC nodules within the same liver.3, 4 Mutations of numerous genes, loss of heterozygosity, insertion of viral DNA and epigenetic changes have been described in HCC development.5 CpG island methylation, the most frequent epigenetic alteration, is often associated with a transcriptional silencing and loss of expression of the respective protein. This mechanism of inactivating tumor suppressor genes is common in human cancer, as shown for colon, lung, breast and prostate cancer among others including HCC. Genes silenced by this mechanism are p16INK4a, p15INK4b, p14ARF, GSTP1, APC, MGMT, hMLH1, SOCS-1, E- cadherin and RASSF1A among many others.6–9

In HCC, it has been demonstrated that p16INK4a gene expression is downregulated, associated with CpG island methylation of the gene's promoter region.10, 11 Aberrant DNA methylation has already been described in preneoplastic liver lesions, suggesting that this is an early event in hepatocarcinogenesis.12 While the significance of the epigenetic changes is not fully understood, in addition to functional implications of gene inactivation in tumor development, these aberrant methylation patterns are excellent potential marker for diagnosis and surveillance of persons at high risk for HCC development. Nevertheless, data about the methylation status of normal liver are limited since most studies only investigate neoplastic and cirrhotic tissue (CT) or tissue from patients with viral hepatitis. However, hypermethylation can also develop during normal cell aging, even without impact on malignant transformation.13–17 To study the relevance of gene-promoter hypermethylation of the p16INK4a, GSTP1, MGMT, DAP-K and APC genes as molecular marker, we examined their promoter methylation status in HCC and non-malignant liver tissue from 34 patients, respectively, as well as in 16 normal livers by real-time, quantitative methylation-specific PCR (MSP).

Abbreviations

ACTB, actin-beta; APC, adenomatous polyposis coli; DAP-K, death-associated protein kinase; GSTP1, glutathione S-transferase 1; MGMT, O6-methylguanine DNA methyltransferase; MSP, methylation-specific PCR; PCR, polymerase chain reaction.

Materials and methods

Tissue samples and DNA preparation

We obtained HCC tissue samples from 34 patients who underwent partial liver resection for HCC, including non-malignant cirrhotic liver (CT). In addition, 16 normal liver tissue (NLT) samples were obtained from patients who underwent cholecystectomy. Informed consent was obtained from all patients. Eight patients (23.5%) were women and 26 (76.5%) men with a mean age of 58.7 (range 40–72) years. In 91.2% liver cirrhosis was due to hepatitis B or C virus infection. Seventeen (50%) patients had a single HCC lesion and 17 (50%) had multifocal disease. The mean tumor diameter was 9.9 cm (1–15 cm) with mainly moderate differentiation grade (79.4%). Among the patients 5.9% had UICC stage I, 26.5% UICC stage II, 58.8% UICC stage III and 8.8% UICC stage IV disease.

The tissue specimens were snap frozen in liquid nitrogen and stored at −80°C. H&E stained sections were histologically examined every 20 sections for the presence or absence of tumor cells as well as for tumor cell density. Only HCC tissue that showed solely tumor cells was used for DNA extraction after microdissection. The tissues were digested with proteinase K. DNA was extracted by standard procedures and ethanol precipitated. Before MSP, chemical modification of unmethylated but not of methylated cytosines to uracil within the CpG islands was performed as described previously.18 After bisulfite modification, the DNA was eluted in 30 μl H2O and 3 μl was used for each sample for PCR.

Methylation-specific, quantitative real-time PCR

Quantitative real-time PCR was performed using the ABI Prism 7700 PCR cycler (Applied Biosystems, Foster City, USA).19, 20 The sequences of the primers and probes were as follows: (a) APC Promoter 1 A 5′-GAA CCA AAA CGC TCC CCA T-3′, 5′-TTA TAT GTC GGT TAC GTG CGT TTA TAT-3′, 6FAM5′-CCC GTC GAA AAC CCG CCG ATT A-3′TAMRA; (b) DAP-kinase 5′-GAA CCA AAA CGC TCC CCA T-3′, 5′-TTA TAT GTC GGT TAC GTG CGT TTA TAT-3′, 6FAM 5′-CCC GTC GAA AAC CCG CCG ATT A-3′TAMRA; (c) GSTP1 5′-AGT TGC GCG GCG ATT TC-3′, 5′-GCC CCA ATA CTA AAT CAC GAC G-3′, 6FAM 5′-CGG TCG ACG TTC GGG GTG TAG CG-3′TAMRA; (d) p16INK4a 5′-TTA TTA GAG GGT GGG GCG GAT CGC-3′, 5′-GAC CCC GAA CCG CGA CCG TAA-3′, 6FAM5′-AGT AGT ATG GAG TCG GCG GCG GG-3′TAMRA; (e) MGMT 5′-CGA ATA TAC TAA AAC AAC CCG CG-3′, 5′-GTA TTT TTT CGG GAG CGA GGC-3′, 6FAM5′-AAT CCT CGC GAT ACG CAC CGT TTA CG-3′TAMRA; (f) ACTB 5′-TGG TGA TGG AGG AGG TTT AGT AAG T-3′, 5′-AAC CAA TAA AAC CTA CTC CTC CCT TAA-3′, 6FAM5′-ACC ACC ACC CAA CAC ACA ATA ACA AAC ACA-3′TAMRA.

Fluorogenic PCR was carried out in a reaction volume of 25 μl using components of the TaqMan® PCR Buffer A Pack (PE, Applied Biosystems). Fluorogenic probes were custom synthesized by PE Applied Biosystems. PCR primers were synthesized by Genescan, Freiburg, Germany. The PCR reaction mixture consisted of 600 nM of each primer, 200 nM probe, 5 U AmpliTaq Gold® polymerase, 200 μM each dATP, dCTP, dGTP, 400 μM dUTP, 5.5 mM MgCl2 and 1 × TaqMan Buffer A. Thermal cycling was initiated with a first denaturation step of 95°C for 10 min. The thermal profile for the PCR was 95°C for 15 sec, 60°C for 30 sec and 72°C for 1 min for a total of 50 cycles.

Amplifications were carried out in 96 well plates in a 7700 Sequence detector (Applied Biosystems) in triplicates. Each plate consisted of patient samples and multiple water blanks as well as a positive and a negative control. To validate methylation-specific binding to the respective promoter, different cell lines were used, which all had been characterized extensively regarding the methylation status of the respective promoter, mainly by the Sidransky Laboratory.15, 21–24 The lung cancer cell line H157 is considered a and served as positive control for the genes MGMT and APC, the non-small cell lung cancer cell line H1299 for DAP-K and the prostate cancer cell line LNCaP for GSTP1. The myeloma cell line HS-Sultan was used as positive control for p16INK4a. Negative controls were the fetal human epithelial lung cell line L132 for DAP-K and APC, H157 for GSTP1, H1299 for MGMT and the lung cancer cell line A549 for p16INK4a, respectively.15, 21–24 Serial dilutions of the positive controls were used for generating a calibration curve for each single analysis. To determine the relative levels of methylated gene-promoter DNA in each sample the values of each ‘gene of interest’ was compared to the values of the internal reference, the housekeeping gene ACTB ([gene of interest/ACTB] × 100).

Statistical analysis

Real-time PCR analyses yielded values that are expressed as ratios between 2 absolute measurements ([gene of interest/internal reference] × 100). Medians and ranges were calculated for the methylation values of each sample. Associations between variables were tested by using the Wilcoxon signed rank test or the Mann-Whitney's U test. The significance of rank ordering between variables was tested by using the Kruskal–Wallis analysis of variance for ordinal data. To estimate the correlation between selected variables Spearman's rank correlation coefficient was used. Survival was estimated according to Kaplan and Meier. Multivariate analysis was performed with the Cox proportional hazard regression model. The level of significance was set to p < 0.05. Analyses were carried out using the SPSS software package (Chicago, USA).

Results

Frequencies of methylation in HCC, cirrhotic and normal liver

Aberrant promoter methylation of 4 tumor suppressor genes and 1 detoxification gene was determined by quantitative real-time PCR in 34 HCC and matched cirrhotic liver tissue samples. The mean age of the study population was 58.7 years (range, 40–72). Liver cirrhosis and HCC was mainly due to hepatitis B and C virus infection (29.5% and 61.8%). The clinicopathological characteristics of the HCC patients are summarized in Table I. The 34 HCC and non-tumorous cirrhotic liver tissue samples (CT) were additionally compared to normal non-cirrhotic liver tissue samples (NLT) from 16 healthy individuals.

Table I. Clinicopathological Characteristics of Patients with HCC
CharacteristicsHCC/cirrhosis, N (%)
  1. AFP, alpha fetoprotein.

Patients34
Female8 (23.5)
Male27 (76.5)
Mean age (range)58.7 (40–72)
Cause of liver cirrhosis
 Hepatitis B positive10 (29.4)
 Hepatitis C positive21 (61.8)
 Unknown3 (8.8)
Tumor differentiation
 Well5 (14.7)
 Moderately27 (79.4)
 Poorly2 (5.9)
AFP ng/ml median (range)950 (3–55,000)
UICC stage
 I2 (5.9)
 II9 (26.5)
 III20 (58.8)
 IV3 (8.8)
Unilocular17 (50)
Multilocular17 (50)
Portal vein invasion6 (17.6)
Mean tumor size [cm (range)]9.9 (1–15)

The p16INK4A and GSTP1 gene showed a stepwise increase of aberrant promoter methylation from normal to neoplastic tissue: p16INK4a: NLT 0%, CT 24%, HCC 76%; GSTP1: NLT 0%, CT 32% and HCC 53%, whereas promoter methylation in MGMT was infrequent in all tissues examined NLT 0%, CT 12% and HCC 6%. For the 3 genes, for which there was no methylated promoter DNA in NLT, p16INK4a, GSTP1 and MGMT, a robust amplification of the ACTB control gene was documented. Two genes were aberrantly methylated in NLT: DAP-K in NLT 31%, CT 100% and HCC 68%; APC in NLT 100%, CT 100% and HCC 100%, respectively (Table II). GSTP1 and/or p16INK4a methylation was observed in 88% of HCC. There was no significant difference in methylation frequency between HBV and HCV-positive HCC.

Table II. Frequency of Promoter Methylation in HCC, Liver Cirrhosis and Normal Liver Tissue
GeneHCC (n = 34) N (%)Cirrhosis (n = 34) N (%)Normal liver (n = 16) N (%)
p16INK4a26 (76)8 (24)0
GSTP118 (53)11 (32)0
MGMT2 (6)4 (12)0
DAP-K23 (68)34 (100)5 (31)
APC34 (100)34 (100)16 (100)

Level of methylation in HCC, cirrhotic and normal liver

In addition to determining the pure frequency of gene-promoter methylation, the methylation levels were quantified using real-time PCR (see also Table III and Fig. 1). Well-characterized cell lines were used to validate real-time, quantitative MSP for each individual gene (Table IV).

Figure 1.

GSTP1, p16INK4a, APC and DAP-K/actin-beta ratios × 100 on a log scale. HCC and non-tumorous cirrhotic liver samples from 34 patients and NLT samples from 16 healthy controls. GSTP1 and p16INK4a/ACTB ratios of liver from healthy control individuals are negative. Boxes indicate the fluorescence emission intensity values of the gene of interest, positive for the methylated gene-promoter DNA. Bars indicate the median methylation level within a sample type. Values of 0.001 (APC), 0.0001 (GSTP1, DAP-K) and 0.00001 (p16) are zero values because they cannot be plotted correctly on a log scale. *p = <0.001, other statistical correlations between the 3 tissue types is given in Table III.

Table III. Statistical Correlation Between Promoter Methylation and Tissue Type (p Values)
GeneHCC versus cirrhosisHCC versus normalCirrhosis versus normal
  1. ns, not significant.

p16INK4a<0.001<0.0010.025
GSTP10.015<0.0010.012
MGMTnsnsns
DAP-K<0.0010.004<0.001
APC<0.0010.9<0.001
Table IV. Gene-Specific Positive and Negative Control Cell Lines Used for Validation of the Gene-Specific Quantitative Real-Time MSP
GenePositive controlNegative control
  1. The cell lines were used for which promoter methylation of the respective gene and its functional consequences were characterized extensively.15, 21–24

p16INK4aCRL-1484 (HS-Sultan)A549
GSTP1LNCaPH157
MGMTH157H1299
DAP-KH1299L132
APCH157L132

Real-time MSP analysis revealed that the APC gene was significantly higher methylated in NLT [APC/ACTB] × 100: median 88.0 (range, 1.4–209) and HCC: 97.5 (0.8–412), NLT versus HCC p = 0.9, than in CT 6.6 (0.2–60). By contrast, the methylation level of DAP-K was significantly higher in CT [DAP-K/ACTB] × 100: 0.7 (0.3–10.3) than in HCC 0.1 (0.01–3.3) and NLT 0.001(0.001–0.6). There was no statistical difference of the level of methylated promoter DNA for MGMT [MGMT/ACTB] × 100 between HCC: 0.001 (0.001–0.17), CT: 0.001 (0.001–0.004) and NLT: 0 (p = ns).

The methylation levels of p16INK4a and GSTP1 increased significantly from NLT to HCC. There was no methylated p16INK4a promoter DNA in NLT; in CT, the median level was [p16INK4a/ACTB] × 100: 0.01 (range, 0.01–14.9); and in HCC: 23.6 (0.01–110.1). The p values for NLT versus CT were 0.025, for NLT versus HCC 0.001 and for CT versus HCC 0.001, respectively.

Furthermore, for the GSTP1 promoter, there was no methylation detectable in NLT. The median methylation level for CT was [GSTP1/ACTB] × 100: 0.01 (range, 0.01–178) and for HCC 1.4 (0.01–210). The p values for NLT versus CT were 0.012, for NLT versus HCC 0.001 and for CT versus HCC 0.015.

Methylation profile and clinicopathological features

Statistical analyses of the data revealed no significant correlations between the methylation levels of the respective promoter and clinicopathological parameters, such as histologic subtype of HCC, HCC grade or stage, patient age or gender, HBV and HCV status, tumor size and lymph node involvement. The sensitivity and specificity for the methylation levels was 55.9% and 70.6% for GSTP1 and 73.5% and 76.5% for p16INK4a, respectively. If either p16INK4a or GSTP1 promoter methylation status was combined with AFP (with a cut off of AFP > 200 ng/ml), the sensitivity was improved to 65.6% and 87.5% for GSTP1 and p16INK4a, respectively. For the combination of both genes and AFP, the sensitivity was also 87.5%. There was a significant correlation for AFP levels with GSTP1 methylation levels (r = 0.432, p = 0.013) but not with p16INK4a (r = −0.065, p = 0.724).

To determine whether different methylation levels in HCC have prognostic significance, in addition to a diagnostic value, we analyzed survival data, available for 33 of the 34 patients. Seventeen patients died during a mean follow-up of 4.3 years (0–9.8). Two patients (6%) died from surgery and were excluded, and 11 patients (32.3%) died from HCC recurrence with a median time to recurrence of 350 days (range, 25–1204). Two patients (6%) died from liver failure and 2 patients (6%) from other causes. For the 5 genes, no statistically significant association was found between methylation levels analyzed and overall patient survival.

Discussion

HCC development is a multistep process associated with multiple genetic alterations. Among others, loss of function mutations, for example, of the tumor suppressor genes p53 and beta-catenin as well as gene overexpression, e.g., c-myc and cyclin D1 have been reported. Gene inactivation, mostly affecting tumor suppressor genes, can be due to loss of heterozygosity as well as epigenetic events such as aberrant methylation of CpG islands of the promoter region of respective genes. The promoter methylation of numerous genes, e.g., p16INK4a, p15INK4b, p14ARF, GSTP1, APC, MGMT, hMLH1, DLC-1, SOCS-1, HLTF, E-cadherin, RAR-β, RASSF1A, HIN-1 and RIZ-1, has been associated with hepatocarcinogenesis.3, 5, 12, 25–27 The spectrum of methylated genes detected to date is possibly related to geographic or etiologic factors inducing different mechanisms of hepatocarcinogenesis. p16INK4a, GSTP1, MGMT, DAP-K and APC have been frequently observed at least in qualitative analyses. In only very few studies, normal liver and paired samples of HCC and non-tumor liver were compared and analyzed quantitatively.

Here, we show that the tumor suppressor genes APC and DAP-K are highly methylated in normal liver and therefore are unlikely to contribute to hepatocarcinogenesis or to serve as molecular marker. Furthermore, the tumor suppressor gene APC was found methylated in numerous tissues. In previous studies, aberrant APC methylation was found in 0–82% of HCC.3, 12, 25, 26 In our study, APC promoter methylation was found in 100% even in normal liver. These results may in part be explained by the different ethnic background of the patients examined. Furthermore, methodological advances, e.g., increased sensitivity of the fluorogenic real-time MSP used in our study, when compared to conventional MSP, may also have resulted in a higher detection rate of aberrant promoter methylation. Along the same line, the extended analysis of CpG dinucleotides in the APC promoter region (GenBank accession number: U02509) may have contributed to a higher rate of detection of aberrant promoter methylation.15, 16 Our findings clearly indicate that APC promoter methylation is common in normal liver and therefore is not a useful marker for HCC.15, 17

Previously, in one study, 9 genes were analyzed in HCC, liver adenomas and normal liver for promoter methylation by quantitative MSP.26 A threshold was defined for each gene and each tissue, leading to significant differences in the methylation frequencies of the genes examined, even though promoter methylation in normal, non-neoplastic tissues has been demonstrated previously.13–15, 17, 28, 29 Therefore, taking NLT for threshold calculation in real-time MSP may lead to false negative results for promoter methylation.26, 30 In our study, lower APC methylation in CT compared to HCC and NLT (97.5 in HCC vs. 6.6 in CT, both p < 0.001) could be explained by the relative lower amount of hepatocytes compared to connective tissue in cirrhosis. Similarly, the higher methylation level of DAP-K in CT when compared to HCC and NLT could be explained by aberrant promoter methylation in fibroblasts and Ito-cells (stromal cells) in connective tissue as described for GSTP1 in prostate cancer.7, 31, 32

Reports regarding the expression and methylation status of the DNA repair gene MGMT in HCC and the importance for hepatocarcinogenesis are inconsistent to date. Transcriptional silencing of the MGMT gene can be caused by promoter methylation but also by exposure to certain carcinogens.33 Our findings indicate that MGMT promoter methylation is probably not involved in HCC development and is not a useful molecular marker. This is consistent with previous results by Esteller et al.6 By contrast, Zhang et al. found MGMT gene methylation in 39% of HCC and a correlation with aflatoxin B1-exposure.33 These discrepant results might be explained by the influence of aflatoxin B1-exposure in the study population examined. Furthermore, for MGMT gene-promoter methylation analysis, it is important to define the CpG islands to be analyzed as demonstrated by Matsukura et al.34

In contrast to MGMT, APC and DAP-K, the methylation status of the promoter of p16INK4a and GSTP1, which encodes for the gluthatione S-transferase, show a strong correlation with hepatocarcinogenesis. We demonstrate p16INK4a promoter methylation in 76% of HCC and in 24% of CT samples. Promoter methylation of p16INK4a has been previously found in 16–73% of HCC and 0–30% of CT.10–12, 25 Furthermore, there is previous strong evidence that the p16INK4a gene-promoter methylation leads to gene silencing and loss of protein expression, which was not the scope of this study, suggesting its important role in hepatocarcinogenesis. Nevertheless, this would be supported by our data, showing that promoter methylation was only detected in HCC and CT, but not at all in normal liver. Additionally, the p16INK4a promoter methylation in HCC was correlated with HBV and HCV infection in Japan.10, 11, 12 Some studies from Europe and the USA show a significant correlation between hepatitis B or C virus infection and p16INK4a hypermethylation while others do not.10–12, 25, 26 In our study, quantitative MSP did not identify an association of p16INK4a methylation frequencies or methylation levels with hepatitis B and C virus infection. This fact raises the question whether the host-virus interaction could lead to this different methylation profiles in different ethnic groups.35

The expression of the glutathione S-transferase gene is transcriptionally silenced by GSTP1 promoter methylation as shown in many previous studies.36, 37 For GSTP1, there is a significantly increased methylation level in HCC compared to NLT and CT (NLT 0%, CT 32%, HCC 53%). Furthermore, for GSTP1, quantitative MSP revealed significantly different methylation levels for all 3 tissue types (HCC vs. CT, p = 0.015; HCC vs. NLT, p < 0.001; CT vs. NLT, p = 0.012), similar to the levels shown for p16INK4a (Fig. 1, Table III). Therefore, GSTP1 promoter methylation is also likely to contribute to hepatocarcinogenesis and may be used as a molecular marker for HCC.

Dependent on the cut-off used, AFP is reported to have a sensitivity of about 60% for the diagnosis of HCC.38 Combining elevated AFP levels with methylation levels for GSTP1 and p16INK4a could increase the sensitivity to 87.5% showing the additional value of a sensitive quantitative real-time MSP. Non-standardized MSP analyses are difficult to compare. Quantitative real-time MSP, however, allows to detect aberrant promoter methylation in relation to the housekeeping gene ACTB and makes the analyses, e.g., independent from tissue processing (snap frozen or formalin fixed), DNA contend and MSP conditions.

In summary, our analyses clearly indicate that APC and DAP-K promoter methylation do not contribute to hepatocarcinogenesis and cannot be used as marker for HCC surveillance or detection. In contrast, significant higher levels and frequencies of methylated p16INK4a and GSTP1 promoter in HCC compared to non-malignant cirrhotic liver or normal liver indicate that these epigenetic events may serve as good marker for HCC. Since the methylation of both genes can also be detected by quantitative MSP in plasma/serum, p16INK4a and GSTP1 promoter methylation, besides AFP levels could even be used as marker for identifying patients with liver cirrhosis at risk for developing HCC in the future.31, 39 This issue needs to be addressed in a prospective setting in patients with liver cirrhosis and then compared to the generally practiced surveillance strategy for these patients.

Ancillary