- Email: [email protected]
Quantitative analysis of multiple methylated genes in plasma for the diagnosis and prognosis of hepatocellular carcinoma
Experimental and Molecular Pathology 91 (2011) 702–707
Contents lists available at SciVerse ScienceDirect
Experimental and Molecular Pathology journal homepage: www.elsevier.com/locate/yexmp
Quantitative analysis of multiple methylated genes in plasma for the diagnosis and prognosis of hepatocellular carcinoma Zhao-Hui Huang ⁎, Yu Hu, Dong Hua, Yu-Yu Wu, Ming-Xu Song, Zhi-Hong Cheng Oncology Institute, The Fourth Affiliated Hospital of Suzhou University, Wuxi, 214062 China
a r t i c l e
i n f o
Article history: Received 23 May 2011 Available online 22 August 2011 Keywords: Hepatocellular carcinoma Diagnosis Plasma DNA methylation Restriction enzyme
a b s t r a c t This study was aimed to evaluate the clinical value of plasma methylation analysis of a panel of four genes (APC, GSTP1, RASSF1A, and SFRP1), which was identified by our previous work, for the noninvasive diagnosis of hepatocellular carcinoma (HCC). The methylation status of these four genes in 150 plasma samples from 72 patients with HCC, 37 benign live diseases and 41 normal controls was detected with methylation-sensitive restriction enzymes-based quantitative PCR (MSRE-qPCR) method. The plasma methylation levels of APC, GSTP1, RASSF1A, and SFRP1 were significantly higher in HCCs than those in normal or benign controls (P b 0.05). Although the area under the receiver-operation characteristic curve (AUC-ROC) for individual gene was moderate (range, from 0.800 to 0.881), the combination analysis of these four genes resulted in an increased AUC of 0.933 with 92.7% sensitivity, 81.9% specificity, 90.5% positive predictive value (PPV), and 87.2% negative predictive value (NPV) in discriminating HCC from normal control. The combination analysis also indicated an increased AUC of 0.877 when compared with individual gene (from 0.666 to 0.850) in discriminating HCC from benign control, and the consultant sensitivity, specificity, PPV, and NPV was 84.7%, 81.1%, 89.7%, and 73.2%, respectively. Patients with elevated plasma methylation levels of APC or RASSF1A showed poorer overall survival than those with low levels (P b 0.05). Cox multivariate analysis demonstrated methylated RASSF1A in plasma to be an independent prognostic factor for overall survival (hazard ratio= 3.262, 95% CI: 1.476–7.209, P = 0.003). These data showed that quantitative analysis of multiple methylated genes in plasma may be a promising tool for noninvasive diagnosis of HCC; and methylated plasma RASSF1A appears to be a prognostic marker of HCC. © 2011 Elsevier Inc. All rights reserved.
Introduction Hepatocellular carcinoma (HCC) is one of the most common malignant tumors worldwide. Despite recent advances in the detection and therapy of HCC, the 5-year survival rate remains low, and late presentation remains the most important obstacle to successful treatment. Most HCC cases have already developed locally advanced disease or distant metastasis by the time of diagnosis. At present, α-fetoprotein (AFP) is the most common used clinically for the detection of HCC; however, the diagnostic sensitivity of AFP is relatively low. Therefore, the development of new biomarkers for early HCC detection is urgently needed. Aberrant hypermethylation of CpG islands, a well-known epigenetic event, is a hallmark of cancer (Levenson, 2010; McCabe et al., 2009). Silencing tumor suppressor gene (TSG) by promoter hypermethylation Abbreviations: HCC, hepatocellular carcinoma; MSRE, methylation-sensitive restriction enzymes; qPCR, quantitative polymerase chain reaction; AFP, α-fetoprotein; TSG, tumor suppressor gene; ROC, receiver operator characteristic curve; AUC, area under the curve; MP, methylation percentage; PPV, positive predictive value; NPV, negative predictive value; OS, overall survival; HR, hazard ratios; CI, confidence intervals. ⁎ Corresponding author at: Wuxi Oncology Institute, The Fourth Affiliated Hospital of Suzhou University, 200 Huihe Road, Wuxi, 214062, Jiangsu Province, China. Fax: +86 510 88682900. E-mail address: [email protected] (Z.-H. Huang). 0014-4800/$ – see front matter © 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.yexmp.2011.08.004
has been proved to be an early event in carcinogenesis and is present in the precursor lesions of a variety of cancers, such as HCC. Therefore, it has been intensively attempted to exploit the potential value of tumor associated pattern of DNA methylation as tumor biomarker (Qureshi et al., 2010; Shivapurkar and Gazdar, 2010). These studies revealed that DNA methylation appears to be an emerging tumor biomarker, and frequent promoter methylation of TSGs, such as CDKN2A, RASSF1A, and GSTP1, has been observed in HCC (Lee et al., 2003; Yang et al., 2003). Recent studies clearly demonstrated the advantages of multiple gene hypermethylation analysis in tissue and serum samples regarding diagnostic and prognostic information (Harder et al., 2008; Moribe et al., 2009; Zhang et al., 2007). However, little was known about the value of DNA methylation analysis at multiple gene sites for the detection of HCC in the Chinese population. Recently, we developed a simple method for the quantitative methylation analysis, which combines the use of methylation-sensitive restriction enzyme digestion (MSRE) and quantitative PCR (qPCR) (Huang et al., 2011). Furthermore, we identified that a combination analysis of four methylated genes (APC, GSTP1,RASSF1A, and SFRP1) in the tumor tissues may be suitable for the diagnosis of HCC (Hua et al., 2011). However, the application of tissue-based method is very limited for clinical diagnostic purposes; and a noninvasive test is desired for tumor early detection and for monitoring disease progression.
Z.-H. Huang et al. / Experimental and Molecular Pathology 91 (2011) 702–707
The aim of this study was to evaluate the potential application of this test for HCC detection and prognosis by measuring these methylated genes in plasma samples. Material and methods Collection of plasma specimens
703
pDC316 plasmid DNA was added to each plasma DNA sample (24 μL). Then, each specimen was divided into two equal aliquots. The first one was digested with 10 U HhaI at 37 °C for 16 h in a final reaction volume of 20 μL; and the second aliquot was used as a sham-treated control. After incubation, each digested sample was diluted 5-fold in sterile water and incubated at 65 °C for 15 min to inactivate HhaI. The methylation-sensitive enzyme degrades unmethylated DNA sequences, whereas methylated DNA sequences remain intact and detectable by PCR.
This study collected 109 plasma samples from 72 patients with HCC and 37 with benign liver diseases (including 25 patients with cirrhosis and 12 with chronic inactive hepatitis) in accordance with the institutional ethical guidelines. A total of 41 healthy volunteers, who were apparently healthy based on clinical and laboratory examination, were served as healthy controls. Specimens were selected on the basis of sufficient plasma samples and appropriate clinical data. The clinicopathologic data of these HCC patients at initial diagnosis were listed in Table 1. Blood specimens were collected before surgery or therapy. Five mL sample of peripheral blood was collected in blood collection tube containing EDTA and was centrifuged (2000 g; 10 min at 4 °C) within 2 h after venipuncture. The supernatants were carefully collected and centrifuged again (12,000 g; 10 min at 4 °C) to prevent potential cellular DNA contamination. The plasma was distributed into aliquots and stored at −80 °C until use.
Quantitative PCR was done in duplicate on a DNA Engineer Opticon II (Bio-Rad Laboratories, Hercules, CA, USA) as described (Hua et al., 2011). Melting curve analysis was performed to confirm the specificity of PCR products. Each run included negative controls, blank controls, and 5-fold dilutions of genomic DNA or pDC316 to construct an external standard. Methylation percentage (MP) was used to represent the methylation level of target CpG sites at a specific location in a promoter. MP was calculated using the following equation: MP = (ValueDigested / ValueSham-digested) × 100% where ValueDigested was the quantitative result of qPCR for digested DNA sample, and ValueSham-digested was the quantitative result of the qPCR for the sham-digested sample (Hua et al., 2011).
DNA purification
Statistical analysis
Genomic DNA was isolated from 600 μL plasma using TIANamp Micro DNA Kit (Tiangen, Beijing, China) and eluted in 30 μL sterile water following the manufacturers' protocol. To improve the extraction efficiency, carrier RNA was added after the proteinase K digestion. Two μL plasma DNA was used for the measurement of DNA concentration using qPCR as described by our previous work (Huang et al., 2010). If a plasma specimen had a DNA concentration less than 10 ng/mL, DNA was purified from 1 to 3 mL of plasma, and was concentrated in a final volume of 30 μL by using Eppendorf Concentrator Plus 5301 (Eppendorf, Germany). DNA samples were finally stored at − 20 °C until use.
The difference of DNA methylation status between different groups was analyzed using the Mann–Whitney U or the chi-square test where appropriate. Optimum cutoff values for the four genes were separately determined by simultaneously maximizing both sensitivity and specificity for the detection of HCC for all values of MP using receiver-operating characteristics (ROC) curves. Overall survival (OS) was defined as the time between diagnosis and either death or the time of the last followup. Survival curves were generated by the Kaplan–Meier method and the log-rank test was adopted to compare survival time between patients with different plasma methylation levels. Cox's proportional hazards model was used to estimate Hazard Ratios (HRs) and their 95% confidence intervals (CIs), representing the overall relative risk of death associated with plasma methylation. A p value of less than 0.05 was considered statistically significant. All statistical analyses were conducted with SPSS 13.0 software for windows (SPSS Inc., Chicago, USA).
DNA digestion with MSRE Digestions were performed with HhaI (Takara, Dalian, China) according to our previous work with some modifications (Hua et al., 2011; Huang et al., 2011). Briefly, twenty picograms of unmethylated Table 1 The clinical pathological parameters of patients with hepatocellular carcinomas. Characteristics
N (%)
Total case number Age ≥55 b 55 Gender Male Female UICC stage I–II III–IV Tumor size ≥5 cm b 5 cm HBV infection Yes No Serum AFP levels ≥400 μg/L b 400 μg/L
72 40(55.6) 32(44.4) 61(84.7) 11(15.3) 17(23.6) 55(76.4) 48(66.7) 24(33.3) 61(84.7) 11(15.3) 33(45.8) 39(54.2)
Quantitative PCR
Results The reliability of MSRE-qPCR for analysis of plasma methylation To evaluate the enzyme digestion efficiency of HhaI for plasma DNA, all 150 plasma DNA samples were detected in a separate PCR reaction using pDC316-specific primers. Results of this PCR served as a quality control for the digestion procedure. In this study, the digestion efficiencies for all plasma DNA samples ranged from 99.0% to 99.9%, suggesting the high efficiency of enzyme digestion; and any DNA sample with less than 1% input plasmid sequence after digestion was regarded as qualified sample. Methylation levels of four genes in plasma The methylation status of these four genes was evaluated using MSRE-qPCR in 150 plasma samples, including 72 patients with HCC, 37 with benign live diseases and 41 normal controls, and revealed that the MPs of these four genes (APC, GSTP1, RASSF1A, and SFRP1) were higher in HCCs than in benign controls or healthy controls (Mann–Whitney U Test, P b 0.05, Fig. 1). The diagnostic ability of these four methylated genes and their combination was evaluated using ROC analysis. The AUC for individual gene
704
Z.-H. Huang et al. / Experimental and Molecular Pathology 91 (2011) 702–707
Fig. 1. Quantitative methylation results of MSRE-qPCR on plasma samples. HCC: hepatocelluar carcenoma; Benign: benign liver diseases; Normal: normal control. The line represents the median value. Mann–Whitney U test was used to determine statistical significance.
in discriminating HCC from normal control was moderate (APC: 0.871; RASSF1A: 0.840;GSTP1: 0.800;SFRP1:0.801, Supplementary Fig. 1). However, the combination analysis of these four genes resulted in an increased AUC of 0.933 with 84.7% sensitivity, 87.8% specificity, 92.4% positive predictive value (PPV), and 76.6% negative predictive value (NPV) , respectively (Table 2 and Fig. 2). It is interesting that at least one of the four genes was detected by MSRE-qPCR in 100% (17/17) plasma samples from patients with stages I–II cancer. In addition, the four genes also show low to moderate diagnostic ability in discriminating HCC from benign control with AUCs range from 0.666 to 0.850 (Supplementary Fig. 2), and the combination analysis resulted in an increased AUC of 0.877 with 84.7% sensitivity, 81.1% specificity, 89.7% PPV, and 73.2% NPV, respectively (Table 3 and Fig. 2).
Correlation with clinicopathologic parameters For the analysis of the correlation between the methylation status of a single gene in plasma and clinicpathologic features, the methylation levels were converted into categorical data (i.e. a sample with a methylation value above the optimum cutoff value was considered as methylated). The results revealed that the plasma methylation ratio of RASSF1A was positively correlated with tumor size (P = 0.003), while GSTP1 hypermethylation correlated with elevated serum AFP levels (P = 0.026). In addition, plasma SFRP1 hypermethylation was more common in female HCC patients (P = 0.018). No significant correlation was observed between the methylation status of any one of these genes and other parameters, such as patient age,
Table 2 Diagnostic ability of plasma methylation analysis between HCC patients and normal controls.
APC GSTP1 RASSF1A SFRP1 APC, GSTP1, RASSF1A, or SFRP1
AUC (95% CI)
Sensitivity (%)
Specificity (%)
PPV (%)
NPV (%)
0.871 0.800 0.840 0.801 0.933
68.1 55.6 72.2 55.6 84.7
97.6 90.2 95.1 87.8 87.8
98.0 90.9 96.4 88.9 92.4
63.5 53.6 68.4 52.9 76.6
(0.808–0.935) (0.717–0.883) (0.768–0.912) (0.717–0.886) (0.890–0.975)
AUC: area under curve; CI: confidence interval; PPV: positive predictive value; NPV: negative predictive value.
Z.-H. Huang et al. / Experimental and Molecular Pathology 91 (2011) 702–707
705
the methylation level of RASSF1A in plasma was an independent prognostic factor for OS (HR = 3.262, 95% CI:1.476–7.209. P = 0.003). Discussion
Fig. 2. Receiver-operating characteristics curves for the combined analysis of DNA methylation levels (APC, GSTP1, RASSF1A, and SFRP1) in discriminating HCC from normal controls (A) or benign controls (B). AUC: area under the ROC curve.
tumor differentiation, and HBV infection (P N 0.05, Supplementary Table 1). Correlation with OS To evaluate the potential association between plasma DNA methylation and OS, patients were divided into two groups according to the individual median MPs of these four genes, respectively. HCC patients with elevated plasma methylation levels of APC or RASSF1A showed significantly poorer OS (Log-rank test, P b 0.05, Fig. 3 ) than those with lower concentrations, while no significant association was found between plasma GSTP1 or SFRP1 methylation and OS(Log-rank test, P N 0.05 ). After adjusting for age, gender, tumor size, TNM stage, and serum AFP levels, Cox multivariate analysis demonstrated that
In our previous work (Hua et al., 2011), we have identified that the six methylated genes (APC, GSTP1, RASSF1A, CDKN2A, RUNX3 and SFRP1) having the discriminatory power for HCC; and 100% cases of HCC had at least one promoter methylated in a panel of four targets (APC, GSTP1, RASSF1A, and SFRP1) (Hua et al., 2011). In this study, we evaluated the methylation level of these four TSGs in plasma using modified MSRE-qPCR method (Huang et al., 2011). The results revealed that the methylation levels of these genes were also significantly higher in plasma samples of HCC when compared to benign or normal control, and the combination analysis of these targets showed significant diagnostic value for HCC. In addition, the methylated RASSF1A in plasma appears to be an independent prognostic factor for survival time of HCC patients. For a definitive diagnosis, a tumor biopsy is required. Most studies about DNA methylation focused on the analysis of tumor tissue. However, tumor tissue is not always available and not suitable for screening or early detection of HCC. In contrary, serum or plasma sample is easily obtained. Circulating cell-free DNA shed from the primary tumor tissue, can be retrieved and tested for genetic and epigenetic alterations. Although the underlying mechanism of circulating DNA is unclear, accumulating evidences suggest that circulating DNA is a promising biomarker for human diseases, including cancer. For example, several studies have demonstrated a high concordance between DNA methylation alterations found in primary tumor specimens and in matched plasma or serum samples, suggesting the potential utility of these alterations as surrogate tumor markers (Iyer et al., 2010; Usadel et al., 2002; Wang et al., 2006; Yeo et al., 2005). Several groups have reported that analysis of circulating methylated TSGs could be used for the noninvasive detection of human tumors, including HCC (Chang et al., 2008; Iyer et al., 2010; Wang et al., 2006; Wong et al., 1999; Yeo et al., 2005; Zhang et al., 2007). However, most of these studies evaluated the methylation status of single target, which cannot provide enough diagnostic sensitivity. In this study, a combination analysis of four methylated genes in plasma revealed an increased diagnostic power both for discrimination HCC from normal or from benign control when compared to individual gene. At least one of the four genes was detected in 84.7% (61/72) plasma samples from HCC, 18.9% (7/37) benign controls and 12.2% (5/41) normal controls. The stepwise increasing tread of methylation levels in these genes from normal, benign to tumor plasma samples was similar to our finding in live tissues (normal liver, cirrhotic live, and HCC tissues) (Hua et al., 2011). Interestingly, four of five positive normal controls had HBV infection and/or alcohol drinking history that appear to be the etiological risk factors of HCC and have been reported to correlate with DNA methylation (Chang et al., 2008; Feng et al., 2010; Hernandez-Vargas et al., 2010; Nishida et al., 2008; Su et al., 2007; Yang et al., 2003). In addition to normal controls, a small percentage of plasma samples in patients with benign liver diseases (cirrhosis or chronic inactive hepatitis) showed low to moderate methylation levels, which have also been reported by other studies (Chan et al., 2008;
Table 3 Diagnostic ability of plasma methylation analysis between patients with HCC and benign live diseases.
APC GSTP1 RASSF1A SFRP1 APC,GSTP1,RASSF1A, or SFRP1
AUC (95% CI)
Sensitivity (%)
Specificity (%)
PPV (%)
NPV (%)
0.850 0.774 0.800 0.666 0.877
84.7 62.5 63.9 40.3 84.7
73.0 83.8 94.6 83.8 81.1
85.9 88.2 95.8 82.9 89.7
71.1 53.4 57.4 41.9 73.2
(0.776–0.923) (0.685–0.863) (0.720–0.881) (0.560–0.772) (0.813–0.940)
AUC: area under curve; CI: confidence interval; PPV: positive predictive value; NPV: negative predictive value.
706
Z.-H. Huang et al. / Experimental and Molecular Pathology 91 (2011) 702–707
Fig. 3. Overall survival probabilities in HCC patients according to the plasma methylation levels of four genes at diagnoss. Relation between overall survival (OS) and the plasma methylation levels of APC (A), RASSF1A (B), GSTP1 (C) and SFRP1 (D) in HCC patients. Patients with a high methylation levels of APC or RASSF1A had significantly shorter OS than did those with low levels (P = 0.025 and P = 0.010 by log-rank test, respectively). By contrast, plasma methylation levels of GSTP1 and SFRP1 were not related to OS.
Chang et al., 2008; Lee et al., 2003). These data suggested that the hypermethylation of some TSGs is early event of liver carcinogenesis and present in the precursor lesions of HCC. Chan et al. revealed that hypermethylated RASSF1A was found in the sera of 93% HCC patients, 58% HBV carriers, and 8% of normal volunteers (Chan et al., 2008). Their positive ratio of RASSF1A was also higher both than ours and other reports (Wang et al., 2006; Yeo et al., 2005). The different target CpG cites may be the major reason for the difference in detection ratio. Zhang et al. reported that the analysis of hypermethylaiton of RASSF1A, p16, and p15 in serum DNA is a valuable biomarkers for early detection of HCC (Zhang et al., 2007). In contrast to these studies, Chang et al. showed that the methylation analysis of E-cadherin, GSTP1, p16, and RASSF1A in the plasma samples might have limited usage for HCC diagnosis according to their results on a study of small sample size using methylation-specific PCR method (Chang et al., 2008). These differing results may due to the lack of standardized processing of blood samples or analysis methods; and the relatively small sample size and diversity in the clinical courses of patients may also contribute to the variation. We also evaluated whether hypermethylation in plasma samples of HCC patients is of any prognostic value, and found that elevated plasma methylation levels of APC or RASSF1A were associated with poorer OS. In addition, Cox multivariate analysis further identified that plasma RASSF1A methylation is an independent prognostic factor of survival. Interestingly, the levels of circulating methylated RASSF1A were positively correlated with tumor size, suggesting that
the methylated RASSF1A sequences could reflect the tumor load. Consisted with our findings, other studies also revealed that methylation analysis of circulating DNA is potential prognostic factors for HCC (Chan et al., 2008; Hernandez-Vargas et al., 2010; Tangkijvanich et al., 2007). For example, Chan et al. (Chan et al., 2008) revealed that patients with elevated methylated RASSF1A in plasma showed significantly poorer disease-free survival than patients with low concentrations. These results suggested that plasma methylation detection may not only be potential diagnostic biomarker for HCC, but also show important prognostic value in HCC. Detection of circulating DNA methylation depends on the ability of the method to detect methylated sequences in a high background of wild-type DNA. Most studies used bisulfite-based, nonquantitative methods for methylation analysis in plasma/serum, resulting in low sensitivity or high false positive ratio. These methods require the tedious step of bisulfite conversion and subsequent DNA recover that can result in incomplete conversion and DNA loss in the sample. In this study, quantitative MSRE-qPCR, which show superiority for clinical analysis when compared to traditional bisulfite-based assays, was used to detect the levels of four methyalted genes in plasma (Gagnon et al., 2010; Hua et al., 2011; Huang et al., 2011). Our data suggested that MSRE-qPCR is particularly suitable for quantitative analysis of DNA methylation in clinical samples with limited amounts of DNA. In conclusion, we examined the methyaltion levels of four TSG promoters in plasma samples of HCCs and controls by a modified MSRE-qPCR, and found that the detection of these methylated genes
Z.-H. Huang et al. / Experimental and Molecular Pathology 91 (2011) 702–707
in plasma appears to be a valuable diagnostic and prognostic tool for HCC. Given that the small case size in this study, the value of these methylated genes on diagnosis or prognosis of HCC needs to be further validated. Supplementary materials related to this article can be found online at doi:10.1016/j.yexmp.2011.08.004. Acknowledgment This study supported by a grant from the Natural Science Foundation of Jiangsu Province (Grant no. BK2008114). References Chan, K.C., et al., 2008. Quantitative analysis of circulating methylated DNA as a biomarker for hepatocellular carcinoma. Clinical Chemistry 54, 1528–1536. Chang, H., et al., 2008. Methylation of tumor associated genes in tissue and plasma samples from liver disease patients. Experimental and Molecular Pathology 85, 96–100. Feng, Q., et al., 2010. DNA methylation changes in normal liver tissues and hepatocellular carcinoma with different viral infection. Experimental and Molecular Pathology 88, 287–292. Gagnon, J.F., et al., 2010. Quantitative DNA methylation analysis of laser capture microdissected formalin-fixed and paraffin-embedded tissues. Experimental and Molecular Pathology 88, 184–189. Harder, J., et al., 2008. Quantitative promoter methylation analysis of hepatocellular carcinoma, cirrhotic and normal liver. International Journal of Cancer 122, 2800–2804. Hernandez-Vargas, H., et al., 2010. Hepatocellular carcinoma displays distinct DNA methylation signatures with potential as clinical predictors. PloS One 5, e9749. Hua, H., et al., 2011. Quantitative methylation analysis of multiple genes using methylation-sensitive restriction enzymes-based quantitative PCR for the detection of hepatocellular carcinoma. Experimental and Molecular Pathology 91, 455–460. Huang, Z.H., et al., 2010. Quantitation of plasma circulating DNA in hepatocellular carcinoma using quantitative PCR and its potential clinical value. China Oncology 20, 663–667.
707
Huang, Z.H., et al., 2011. The evaluation and application of methylation-sensitive restriction enzymes-based quantitative PCR for DNA methylation detection. Zhonghua Bing Li Xue Za Zhi 40, 263–265. Iyer, P., et al., 2010. Concordance of DNA methylation pattern in plasma and tumor DNA of Egyptian hepatocellular carcinoma patients. Experimental and Molecular Pathology 88, 107–111. Lee, S., et al., 2003. Aberrant CpG island hypermethylation along multistep hepatocarcinogenesis. American Journal of Pathology 163, 1371–1378. Levenson, V.V., 2010. DNA methylation as a universal biomarker. Expert Review of Molecular Diagnostics 10, 481–488. McCabe, M.T., et al., 2009. Cancer DNA methylation: molecular mechanisms and clinical implications. Clinical Cancer Research 15, 3927–3937. Moribe, T., et al., 2009. Methylation of multiple genes as molecular markers for diagnosis of a small, well-differentiated hepatocellular carcinoma. International Journal of Cancer 125, 388–397. Nishida, N., et al., 2008. Aberrant methylation of multiple tumor suppressor genes in aging liver, chronic hepatitis, and hepatocellular carcinoma. Hepatology 47, 908–918. Qureshi, S.A., et al., 2010. Utility of DNA methylation markers for diagnosing cancer. International Journal of Surgery 8, 194–198. Shivapurkar, N., Gazdar, A.F., 2010. DNA methylation based biomarkers in non-invasive cancer screening. Current Molecular Medicine 10, 123–132. Su, P.F., et al., 2007. Differential DNA methylation associated with hepatitis B virus infection in hepatocellular carcinoma. International Journal of Cancer 121, 1257–1264. Tangkijvanich, P., et al., 2007. Serum LINE-1 hypomethylation as a potential prognostic marker for hepatocellular carcinoma. Clinica Chimica Acta 379, 127–133. Usadel, H., et al., 2002. Quantitative adenomatous polyposis coli promoter methylation analysis in tumor tissue, serum, and plasma DNA of patients with lung cancer. Cancer Research 62, 371–375. Wang, J., et al., 2006. Detection of aberrant promoter methylation of GSTP1 in the tumor and serum of Chinese human primary hepatocellular carcinoma patients. Clinical Biochemistry 39, 344–348. Wong, I.H., et al., 1999. Detection of aberrant p16 methylation in the plasma and serum of liver cancer patients. Cancer Research 59, 71–73. Yang, B., et al., 2003. Aberrant promoter methylation profiles of tumor suppressor genes in hepatocellular carcinoma. American Journal of Pathology 163, 1101–1107. Yeo, W., et al., 2005. High frequency of promoter hypermethylation of RASSF1A in tumor and plasma of patients with hepatocellular carcinoma. Liver International 25, 266–272. Zhang, Y.J., et al., 2007. Predicting hepatocellular carcinoma by detection of aberrant promoter methylation in serum DNA. Clinical Cancer Research 13, 2378–2384.
Contents lists available at SciVerse ScienceDirect
Experimental and Molecular Pathology journal homepage: www.elsevier.com/locate/yexmp
Quantitative analysis of multiple methylated genes in plasma for the diagnosis and prognosis of hepatocellular carcinoma Zhao-Hui Huang ⁎, Yu Hu, Dong Hua, Yu-Yu Wu, Ming-Xu Song, Zhi-Hong Cheng Oncology Institute, The Fourth Affiliated Hospital of Suzhou University, Wuxi, 214062 China
a r t i c l e
i n f o
Article history: Received 23 May 2011 Available online 22 August 2011 Keywords: Hepatocellular carcinoma Diagnosis Plasma DNA methylation Restriction enzyme
a b s t r a c t This study was aimed to evaluate the clinical value of plasma methylation analysis of a panel of four genes (APC, GSTP1, RASSF1A, and SFRP1), which was identified by our previous work, for the noninvasive diagnosis of hepatocellular carcinoma (HCC). The methylation status of these four genes in 150 plasma samples from 72 patients with HCC, 37 benign live diseases and 41 normal controls was detected with methylation-sensitive restriction enzymes-based quantitative PCR (MSRE-qPCR) method. The plasma methylation levels of APC, GSTP1, RASSF1A, and SFRP1 were significantly higher in HCCs than those in normal or benign controls (P b 0.05). Although the area under the receiver-operation characteristic curve (AUC-ROC) for individual gene was moderate (range, from 0.800 to 0.881), the combination analysis of these four genes resulted in an increased AUC of 0.933 with 92.7% sensitivity, 81.9% specificity, 90.5% positive predictive value (PPV), and 87.2% negative predictive value (NPV) in discriminating HCC from normal control. The combination analysis also indicated an increased AUC of 0.877 when compared with individual gene (from 0.666 to 0.850) in discriminating HCC from benign control, and the consultant sensitivity, specificity, PPV, and NPV was 84.7%, 81.1%, 89.7%, and 73.2%, respectively. Patients with elevated plasma methylation levels of APC or RASSF1A showed poorer overall survival than those with low levels (P b 0.05). Cox multivariate analysis demonstrated methylated RASSF1A in plasma to be an independent prognostic factor for overall survival (hazard ratio= 3.262, 95% CI: 1.476–7.209, P = 0.003). These data showed that quantitative analysis of multiple methylated genes in plasma may be a promising tool for noninvasive diagnosis of HCC; and methylated plasma RASSF1A appears to be a prognostic marker of HCC. © 2011 Elsevier Inc. All rights reserved.
Introduction Hepatocellular carcinoma (HCC) is one of the most common malignant tumors worldwide. Despite recent advances in the detection and therapy of HCC, the 5-year survival rate remains low, and late presentation remains the most important obstacle to successful treatment. Most HCC cases have already developed locally advanced disease or distant metastasis by the time of diagnosis. At present, α-fetoprotein (AFP) is the most common used clinically for the detection of HCC; however, the diagnostic sensitivity of AFP is relatively low. Therefore, the development of new biomarkers for early HCC detection is urgently needed. Aberrant hypermethylation of CpG islands, a well-known epigenetic event, is a hallmark of cancer (Levenson, 2010; McCabe et al., 2009). Silencing tumor suppressor gene (TSG) by promoter hypermethylation Abbreviations: HCC, hepatocellular carcinoma; MSRE, methylation-sensitive restriction enzymes; qPCR, quantitative polymerase chain reaction; AFP, α-fetoprotein; TSG, tumor suppressor gene; ROC, receiver operator characteristic curve; AUC, area under the curve; MP, methylation percentage; PPV, positive predictive value; NPV, negative predictive value; OS, overall survival; HR, hazard ratios; CI, confidence intervals. ⁎ Corresponding author at: Wuxi Oncology Institute, The Fourth Affiliated Hospital of Suzhou University, 200 Huihe Road, Wuxi, 214062, Jiangsu Province, China. Fax: +86 510 88682900. E-mail address: [email protected] (Z.-H. Huang). 0014-4800/$ – see front matter © 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.yexmp.2011.08.004
has been proved to be an early event in carcinogenesis and is present in the precursor lesions of a variety of cancers, such as HCC. Therefore, it has been intensively attempted to exploit the potential value of tumor associated pattern of DNA methylation as tumor biomarker (Qureshi et al., 2010; Shivapurkar and Gazdar, 2010). These studies revealed that DNA methylation appears to be an emerging tumor biomarker, and frequent promoter methylation of TSGs, such as CDKN2A, RASSF1A, and GSTP1, has been observed in HCC (Lee et al., 2003; Yang et al., 2003). Recent studies clearly demonstrated the advantages of multiple gene hypermethylation analysis in tissue and serum samples regarding diagnostic and prognostic information (Harder et al., 2008; Moribe et al., 2009; Zhang et al., 2007). However, little was known about the value of DNA methylation analysis at multiple gene sites for the detection of HCC in the Chinese population. Recently, we developed a simple method for the quantitative methylation analysis, which combines the use of methylation-sensitive restriction enzyme digestion (MSRE) and quantitative PCR (qPCR) (Huang et al., 2011). Furthermore, we identified that a combination analysis of four methylated genes (APC, GSTP1,RASSF1A, and SFRP1) in the tumor tissues may be suitable for the diagnosis of HCC (Hua et al., 2011). However, the application of tissue-based method is very limited for clinical diagnostic purposes; and a noninvasive test is desired for tumor early detection and for monitoring disease progression.
Z.-H. Huang et al. / Experimental and Molecular Pathology 91 (2011) 702–707
The aim of this study was to evaluate the potential application of this test for HCC detection and prognosis by measuring these methylated genes in plasma samples. Material and methods Collection of plasma specimens
703
pDC316 plasmid DNA was added to each plasma DNA sample (24 μL). Then, each specimen was divided into two equal aliquots. The first one was digested with 10 U HhaI at 37 °C for 16 h in a final reaction volume of 20 μL; and the second aliquot was used as a sham-treated control. After incubation, each digested sample was diluted 5-fold in sterile water and incubated at 65 °C for 15 min to inactivate HhaI. The methylation-sensitive enzyme degrades unmethylated DNA sequences, whereas methylated DNA sequences remain intact and detectable by PCR.
This study collected 109 plasma samples from 72 patients with HCC and 37 with benign liver diseases (including 25 patients with cirrhosis and 12 with chronic inactive hepatitis) in accordance with the institutional ethical guidelines. A total of 41 healthy volunteers, who were apparently healthy based on clinical and laboratory examination, were served as healthy controls. Specimens were selected on the basis of sufficient plasma samples and appropriate clinical data. The clinicopathologic data of these HCC patients at initial diagnosis were listed in Table 1. Blood specimens were collected before surgery or therapy. Five mL sample of peripheral blood was collected in blood collection tube containing EDTA and was centrifuged (2000 g; 10 min at 4 °C) within 2 h after venipuncture. The supernatants were carefully collected and centrifuged again (12,000 g; 10 min at 4 °C) to prevent potential cellular DNA contamination. The plasma was distributed into aliquots and stored at −80 °C until use.
Quantitative PCR was done in duplicate on a DNA Engineer Opticon II (Bio-Rad Laboratories, Hercules, CA, USA) as described (Hua et al., 2011). Melting curve analysis was performed to confirm the specificity of PCR products. Each run included negative controls, blank controls, and 5-fold dilutions of genomic DNA or pDC316 to construct an external standard. Methylation percentage (MP) was used to represent the methylation level of target CpG sites at a specific location in a promoter. MP was calculated using the following equation: MP = (ValueDigested / ValueSham-digested) × 100% where ValueDigested was the quantitative result of qPCR for digested DNA sample, and ValueSham-digested was the quantitative result of the qPCR for the sham-digested sample (Hua et al., 2011).
DNA purification
Statistical analysis
Genomic DNA was isolated from 600 μL plasma using TIANamp Micro DNA Kit (Tiangen, Beijing, China) and eluted in 30 μL sterile water following the manufacturers' protocol. To improve the extraction efficiency, carrier RNA was added after the proteinase K digestion. Two μL plasma DNA was used for the measurement of DNA concentration using qPCR as described by our previous work (Huang et al., 2010). If a plasma specimen had a DNA concentration less than 10 ng/mL, DNA was purified from 1 to 3 mL of plasma, and was concentrated in a final volume of 30 μL by using Eppendorf Concentrator Plus 5301 (Eppendorf, Germany). DNA samples were finally stored at − 20 °C until use.
The difference of DNA methylation status between different groups was analyzed using the Mann–Whitney U or the chi-square test where appropriate. Optimum cutoff values for the four genes were separately determined by simultaneously maximizing both sensitivity and specificity for the detection of HCC for all values of MP using receiver-operating characteristics (ROC) curves. Overall survival (OS) was defined as the time between diagnosis and either death or the time of the last followup. Survival curves were generated by the Kaplan–Meier method and the log-rank test was adopted to compare survival time between patients with different plasma methylation levels. Cox's proportional hazards model was used to estimate Hazard Ratios (HRs) and their 95% confidence intervals (CIs), representing the overall relative risk of death associated with plasma methylation. A p value of less than 0.05 was considered statistically significant. All statistical analyses were conducted with SPSS 13.0 software for windows (SPSS Inc., Chicago, USA).
DNA digestion with MSRE Digestions were performed with HhaI (Takara, Dalian, China) according to our previous work with some modifications (Hua et al., 2011; Huang et al., 2011). Briefly, twenty picograms of unmethylated Table 1 The clinical pathological parameters of patients with hepatocellular carcinomas. Characteristics
N (%)
Total case number Age ≥55 b 55 Gender Male Female UICC stage I–II III–IV Tumor size ≥5 cm b 5 cm HBV infection Yes No Serum AFP levels ≥400 μg/L b 400 μg/L
72 40(55.6) 32(44.4) 61(84.7) 11(15.3) 17(23.6) 55(76.4) 48(66.7) 24(33.3) 61(84.7) 11(15.3) 33(45.8) 39(54.2)
Quantitative PCR
Results The reliability of MSRE-qPCR for analysis of plasma methylation To evaluate the enzyme digestion efficiency of HhaI for plasma DNA, all 150 plasma DNA samples were detected in a separate PCR reaction using pDC316-specific primers. Results of this PCR served as a quality control for the digestion procedure. In this study, the digestion efficiencies for all plasma DNA samples ranged from 99.0% to 99.9%, suggesting the high efficiency of enzyme digestion; and any DNA sample with less than 1% input plasmid sequence after digestion was regarded as qualified sample. Methylation levels of four genes in plasma The methylation status of these four genes was evaluated using MSRE-qPCR in 150 plasma samples, including 72 patients with HCC, 37 with benign live diseases and 41 normal controls, and revealed that the MPs of these four genes (APC, GSTP1, RASSF1A, and SFRP1) were higher in HCCs than in benign controls or healthy controls (Mann–Whitney U Test, P b 0.05, Fig. 1). The diagnostic ability of these four methylated genes and their combination was evaluated using ROC analysis. The AUC for individual gene
704
Z.-H. Huang et al. / Experimental and Molecular Pathology 91 (2011) 702–707
Fig. 1. Quantitative methylation results of MSRE-qPCR on plasma samples. HCC: hepatocelluar carcenoma; Benign: benign liver diseases; Normal: normal control. The line represents the median value. Mann–Whitney U test was used to determine statistical significance.
in discriminating HCC from normal control was moderate (APC: 0.871; RASSF1A: 0.840;GSTP1: 0.800;SFRP1:0.801, Supplementary Fig. 1). However, the combination analysis of these four genes resulted in an increased AUC of 0.933 with 84.7% sensitivity, 87.8% specificity, 92.4% positive predictive value (PPV), and 76.6% negative predictive value (NPV) , respectively (Table 2 and Fig. 2). It is interesting that at least one of the four genes was detected by MSRE-qPCR in 100% (17/17) plasma samples from patients with stages I–II cancer. In addition, the four genes also show low to moderate diagnostic ability in discriminating HCC from benign control with AUCs range from 0.666 to 0.850 (Supplementary Fig. 2), and the combination analysis resulted in an increased AUC of 0.877 with 84.7% sensitivity, 81.1% specificity, 89.7% PPV, and 73.2% NPV, respectively (Table 3 and Fig. 2).
Correlation with clinicopathologic parameters For the analysis of the correlation between the methylation status of a single gene in plasma and clinicpathologic features, the methylation levels were converted into categorical data (i.e. a sample with a methylation value above the optimum cutoff value was considered as methylated). The results revealed that the plasma methylation ratio of RASSF1A was positively correlated with tumor size (P = 0.003), while GSTP1 hypermethylation correlated with elevated serum AFP levels (P = 0.026). In addition, plasma SFRP1 hypermethylation was more common in female HCC patients (P = 0.018). No significant correlation was observed between the methylation status of any one of these genes and other parameters, such as patient age,
Table 2 Diagnostic ability of plasma methylation analysis between HCC patients and normal controls.
APC GSTP1 RASSF1A SFRP1 APC, GSTP1, RASSF1A, or SFRP1
AUC (95% CI)
Sensitivity (%)
Specificity (%)
PPV (%)
NPV (%)
0.871 0.800 0.840 0.801 0.933
68.1 55.6 72.2 55.6 84.7
97.6 90.2 95.1 87.8 87.8
98.0 90.9 96.4 88.9 92.4
63.5 53.6 68.4 52.9 76.6
(0.808–0.935) (0.717–0.883) (0.768–0.912) (0.717–0.886) (0.890–0.975)
AUC: area under curve; CI: confidence interval; PPV: positive predictive value; NPV: negative predictive value.
Z.-H. Huang et al. / Experimental and Molecular Pathology 91 (2011) 702–707
705
the methylation level of RASSF1A in plasma was an independent prognostic factor for OS (HR = 3.262, 95% CI:1.476–7.209. P = 0.003). Discussion
Fig. 2. Receiver-operating characteristics curves for the combined analysis of DNA methylation levels (APC, GSTP1, RASSF1A, and SFRP1) in discriminating HCC from normal controls (A) or benign controls (B). AUC: area under the ROC curve.
tumor differentiation, and HBV infection (P N 0.05, Supplementary Table 1). Correlation with OS To evaluate the potential association between plasma DNA methylation and OS, patients were divided into two groups according to the individual median MPs of these four genes, respectively. HCC patients with elevated plasma methylation levels of APC or RASSF1A showed significantly poorer OS (Log-rank test, P b 0.05, Fig. 3 ) than those with lower concentrations, while no significant association was found between plasma GSTP1 or SFRP1 methylation and OS(Log-rank test, P N 0.05 ). After adjusting for age, gender, tumor size, TNM stage, and serum AFP levels, Cox multivariate analysis demonstrated that
In our previous work (Hua et al., 2011), we have identified that the six methylated genes (APC, GSTP1, RASSF1A, CDKN2A, RUNX3 and SFRP1) having the discriminatory power for HCC; and 100% cases of HCC had at least one promoter methylated in a panel of four targets (APC, GSTP1, RASSF1A, and SFRP1) (Hua et al., 2011). In this study, we evaluated the methylation level of these four TSGs in plasma using modified MSRE-qPCR method (Huang et al., 2011). The results revealed that the methylation levels of these genes were also significantly higher in plasma samples of HCC when compared to benign or normal control, and the combination analysis of these targets showed significant diagnostic value for HCC. In addition, the methylated RASSF1A in plasma appears to be an independent prognostic factor for survival time of HCC patients. For a definitive diagnosis, a tumor biopsy is required. Most studies about DNA methylation focused on the analysis of tumor tissue. However, tumor tissue is not always available and not suitable for screening or early detection of HCC. In contrary, serum or plasma sample is easily obtained. Circulating cell-free DNA shed from the primary tumor tissue, can be retrieved and tested for genetic and epigenetic alterations. Although the underlying mechanism of circulating DNA is unclear, accumulating evidences suggest that circulating DNA is a promising biomarker for human diseases, including cancer. For example, several studies have demonstrated a high concordance between DNA methylation alterations found in primary tumor specimens and in matched plasma or serum samples, suggesting the potential utility of these alterations as surrogate tumor markers (Iyer et al., 2010; Usadel et al., 2002; Wang et al., 2006; Yeo et al., 2005). Several groups have reported that analysis of circulating methylated TSGs could be used for the noninvasive detection of human tumors, including HCC (Chang et al., 2008; Iyer et al., 2010; Wang et al., 2006; Wong et al., 1999; Yeo et al., 2005; Zhang et al., 2007). However, most of these studies evaluated the methylation status of single target, which cannot provide enough diagnostic sensitivity. In this study, a combination analysis of four methylated genes in plasma revealed an increased diagnostic power both for discrimination HCC from normal or from benign control when compared to individual gene. At least one of the four genes was detected in 84.7% (61/72) plasma samples from HCC, 18.9% (7/37) benign controls and 12.2% (5/41) normal controls. The stepwise increasing tread of methylation levels in these genes from normal, benign to tumor plasma samples was similar to our finding in live tissues (normal liver, cirrhotic live, and HCC tissues) (Hua et al., 2011). Interestingly, four of five positive normal controls had HBV infection and/or alcohol drinking history that appear to be the etiological risk factors of HCC and have been reported to correlate with DNA methylation (Chang et al., 2008; Feng et al., 2010; Hernandez-Vargas et al., 2010; Nishida et al., 2008; Su et al., 2007; Yang et al., 2003). In addition to normal controls, a small percentage of plasma samples in patients with benign liver diseases (cirrhosis or chronic inactive hepatitis) showed low to moderate methylation levels, which have also been reported by other studies (Chan et al., 2008;
Table 3 Diagnostic ability of plasma methylation analysis between patients with HCC and benign live diseases.
APC GSTP1 RASSF1A SFRP1 APC,GSTP1,RASSF1A, or SFRP1
AUC (95% CI)
Sensitivity (%)
Specificity (%)
PPV (%)
NPV (%)
0.850 0.774 0.800 0.666 0.877
84.7 62.5 63.9 40.3 84.7
73.0 83.8 94.6 83.8 81.1
85.9 88.2 95.8 82.9 89.7
71.1 53.4 57.4 41.9 73.2
(0.776–0.923) (0.685–0.863) (0.720–0.881) (0.560–0.772) (0.813–0.940)
AUC: area under curve; CI: confidence interval; PPV: positive predictive value; NPV: negative predictive value.
706
Z.-H. Huang et al. / Experimental and Molecular Pathology 91 (2011) 702–707
Fig. 3. Overall survival probabilities in HCC patients according to the plasma methylation levels of four genes at diagnoss. Relation between overall survival (OS) and the plasma methylation levels of APC (A), RASSF1A (B), GSTP1 (C) and SFRP1 (D) in HCC patients. Patients with a high methylation levels of APC or RASSF1A had significantly shorter OS than did those with low levels (P = 0.025 and P = 0.010 by log-rank test, respectively). By contrast, plasma methylation levels of GSTP1 and SFRP1 were not related to OS.
Chang et al., 2008; Lee et al., 2003). These data suggested that the hypermethylation of some TSGs is early event of liver carcinogenesis and present in the precursor lesions of HCC. Chan et al. revealed that hypermethylated RASSF1A was found in the sera of 93% HCC patients, 58% HBV carriers, and 8% of normal volunteers (Chan et al., 2008). Their positive ratio of RASSF1A was also higher both than ours and other reports (Wang et al., 2006; Yeo et al., 2005). The different target CpG cites may be the major reason for the difference in detection ratio. Zhang et al. reported that the analysis of hypermethylaiton of RASSF1A, p16, and p15 in serum DNA is a valuable biomarkers for early detection of HCC (Zhang et al., 2007). In contrast to these studies, Chang et al. showed that the methylation analysis of E-cadherin, GSTP1, p16, and RASSF1A in the plasma samples might have limited usage for HCC diagnosis according to their results on a study of small sample size using methylation-specific PCR method (Chang et al., 2008). These differing results may due to the lack of standardized processing of blood samples or analysis methods; and the relatively small sample size and diversity in the clinical courses of patients may also contribute to the variation. We also evaluated whether hypermethylation in plasma samples of HCC patients is of any prognostic value, and found that elevated plasma methylation levels of APC or RASSF1A were associated with poorer OS. In addition, Cox multivariate analysis further identified that plasma RASSF1A methylation is an independent prognostic factor of survival. Interestingly, the levels of circulating methylated RASSF1A were positively correlated with tumor size, suggesting that
the methylated RASSF1A sequences could reflect the tumor load. Consisted with our findings, other studies also revealed that methylation analysis of circulating DNA is potential prognostic factors for HCC (Chan et al., 2008; Hernandez-Vargas et al., 2010; Tangkijvanich et al., 2007). For example, Chan et al. (Chan et al., 2008) revealed that patients with elevated methylated RASSF1A in plasma showed significantly poorer disease-free survival than patients with low concentrations. These results suggested that plasma methylation detection may not only be potential diagnostic biomarker for HCC, but also show important prognostic value in HCC. Detection of circulating DNA methylation depends on the ability of the method to detect methylated sequences in a high background of wild-type DNA. Most studies used bisulfite-based, nonquantitative methods for methylation analysis in plasma/serum, resulting in low sensitivity or high false positive ratio. These methods require the tedious step of bisulfite conversion and subsequent DNA recover that can result in incomplete conversion and DNA loss in the sample. In this study, quantitative MSRE-qPCR, which show superiority for clinical analysis when compared to traditional bisulfite-based assays, was used to detect the levels of four methyalted genes in plasma (Gagnon et al., 2010; Hua et al., 2011; Huang et al., 2011). Our data suggested that MSRE-qPCR is particularly suitable for quantitative analysis of DNA methylation in clinical samples with limited amounts of DNA. In conclusion, we examined the methyaltion levels of four TSG promoters in plasma samples of HCCs and controls by a modified MSRE-qPCR, and found that the detection of these methylated genes
Z.-H. Huang et al. / Experimental and Molecular Pathology 91 (2011) 702–707
in plasma appears to be a valuable diagnostic and prognostic tool for HCC. Given that the small case size in this study, the value of these methylated genes on diagnosis or prognosis of HCC needs to be further validated. Supplementary materials related to this article can be found online at doi:10.1016/j.yexmp.2011.08.004. Acknowledgment This study supported by a grant from the Natural Science Foundation of Jiangsu Province (Grant no. BK2008114). References Chan, K.C., et al., 2008. Quantitative analysis of circulating methylated DNA as a biomarker for hepatocellular carcinoma. Clinical Chemistry 54, 1528–1536. Chang, H., et al., 2008. Methylation of tumor associated genes in tissue and plasma samples from liver disease patients. Experimental and Molecular Pathology 85, 96–100. Feng, Q., et al., 2010. DNA methylation changes in normal liver tissues and hepatocellular carcinoma with different viral infection. Experimental and Molecular Pathology 88, 287–292. Gagnon, J.F., et al., 2010. Quantitative DNA methylation analysis of laser capture microdissected formalin-fixed and paraffin-embedded tissues. Experimental and Molecular Pathology 88, 184–189. Harder, J., et al., 2008. Quantitative promoter methylation analysis of hepatocellular carcinoma, cirrhotic and normal liver. International Journal of Cancer 122, 2800–2804. Hernandez-Vargas, H., et al., 2010. Hepatocellular carcinoma displays distinct DNA methylation signatures with potential as clinical predictors. PloS One 5, e9749. Hua, H., et al., 2011. Quantitative methylation analysis of multiple genes using methylation-sensitive restriction enzymes-based quantitative PCR for the detection of hepatocellular carcinoma. Experimental and Molecular Pathology 91, 455–460. Huang, Z.H., et al., 2010. Quantitation of plasma circulating DNA in hepatocellular carcinoma using quantitative PCR and its potential clinical value. China Oncology 20, 663–667.
707
Huang, Z.H., et al., 2011. The evaluation and application of methylation-sensitive restriction enzymes-based quantitative PCR for DNA methylation detection. Zhonghua Bing Li Xue Za Zhi 40, 263–265. Iyer, P., et al., 2010. Concordance of DNA methylation pattern in plasma and tumor DNA of Egyptian hepatocellular carcinoma patients. Experimental and Molecular Pathology 88, 107–111. Lee, S., et al., 2003. Aberrant CpG island hypermethylation along multistep hepatocarcinogenesis. American Journal of Pathology 163, 1371–1378. Levenson, V.V., 2010. DNA methylation as a universal biomarker. Expert Review of Molecular Diagnostics 10, 481–488. McCabe, M.T., et al., 2009. Cancer DNA methylation: molecular mechanisms and clinical implications. Clinical Cancer Research 15, 3927–3937. Moribe, T., et al., 2009. Methylation of multiple genes as molecular markers for diagnosis of a small, well-differentiated hepatocellular carcinoma. International Journal of Cancer 125, 388–397. Nishida, N., et al., 2008. Aberrant methylation of multiple tumor suppressor genes in aging liver, chronic hepatitis, and hepatocellular carcinoma. Hepatology 47, 908–918. Qureshi, S.A., et al., 2010. Utility of DNA methylation markers for diagnosing cancer. International Journal of Surgery 8, 194–198. Shivapurkar, N., Gazdar, A.F., 2010. DNA methylation based biomarkers in non-invasive cancer screening. Current Molecular Medicine 10, 123–132. Su, P.F., et al., 2007. Differential DNA methylation associated with hepatitis B virus infection in hepatocellular carcinoma. International Journal of Cancer 121, 1257–1264. Tangkijvanich, P., et al., 2007. Serum LINE-1 hypomethylation as a potential prognostic marker for hepatocellular carcinoma. Clinica Chimica Acta 379, 127–133. Usadel, H., et al., 2002. Quantitative adenomatous polyposis coli promoter methylation analysis in tumor tissue, serum, and plasma DNA of patients with lung cancer. Cancer Research 62, 371–375. Wang, J., et al., 2006. Detection of aberrant promoter methylation of GSTP1 in the tumor and serum of Chinese human primary hepatocellular carcinoma patients. Clinical Biochemistry 39, 344–348. Wong, I.H., et al., 1999. Detection of aberrant p16 methylation in the plasma and serum of liver cancer patients. Cancer Research 59, 71–73. Yang, B., et al., 2003. Aberrant promoter methylation profiles of tumor suppressor genes in hepatocellular carcinoma. American Journal of Pathology 163, 1101–1107. Yeo, W., et al., 2005. High frequency of promoter hypermethylation of RASSF1A in tumor and plasma of patients with hepatocellular carcinoma. Liver International 25, 266–272. Zhang, Y.J., et al., 2007. Predicting hepatocellular carcinoma by detection of aberrant promoter methylation in serum DNA. Clinical Cancer Research 13, 2378–2384.