Liver Cancer
Copyright ©The Author(s) 2004. Published by Baishideng Publishing Group Inc. All rights reserved.
World J Gastroenterol. Feb 1, 2004; 10(3): 376-380
Published online Feb 1, 2004. doi: 10.3748/wjg.v10.i3.376
Hemizygous deletion and hypermethylation of RUNX3 gene in hepatocellular carcinoma
Wen-Hua Xiao, Wei-Wen Liu
Wen-Hua Xiao, Department of Oncology, 304th Hospital of PLA, Beijing 100037, China
Wei-Wen Liu, Department of Gastroenterology, Southwest Hospital, Third Military Medical University, Chongqing 400038, China
Author contributions: All authors contributed equally to the work.
Correspondence to: Wen-Hua Xiao, PhD, Associate Professor, Department of Oncology, 304th Hospital of PLA, Beijing 100037, China.
Telephone: +86-10-66867324 Fax: +86-10-66867672
Received: May 11, 2003
Revised: May 25, 2003
Accepted: June 2, 2003
Published online: February 1, 2004


AIM: To analyze the genetic and epigenetic alterations of RUNX3 gene, a potential putative tumor suppressor gene, in hepatocellular carcinoma (HCC).

METHODS: PCR-based loss of heterozygosity (LOH) detection, analysis of mutation with PCR-single strand conformational polymorphism (SSCP) and sequencing, and methylation study with methylation specific PCR (MSP) were performed on RUNX3 gene in a series of 62 HCCs along with their matched normal tissues.

RESULTS: Mutation of RUNX3 gene was not found, but one single nucleotide polymorphism with T to A transversion at the second nucleotide of the 18th condon was found. Nine of 26 informative cases (34.6%) showed allelic loss on the polymorphic site and 30 cases (48.4%) revealed hypermethylation of RUNX3 gene in promoter CpG islands. Furthermore, of the 9 cases with LOH, 8 (88.9%) also had hypermethylation.

CONCLUSION: Our findings indicate that inactivation of RUNX3 gene through allelic loss and promoter hypermethylation might be one of the major mechanisms in hepatocellualr carcinogenesis.


Transforming growth factor-β (TGF-β) is a multifunctional cytokine known to be a potent growth inhibitor for most epithelial cells[1,2]. TGF-β signaling pathway is composed of TGF-β type I, type II receptors and Smad proteins, and is transducted by forming heteromeric complex with its type I and type II transmembrane Ser/Thr kinase receptors. Activated type I receptors then activate the cytoplasmic Smad 2 and Smad 3 by phosphorylation, allowing them to form a hetermeric complex with Smad 4. This Smad complex can activate TGF-beta responsive gene transcription only after it is translocated to nucleus and bound to the specific target nuclear matrix site[2]. However, the key process of nuclear translocation and subnuclear distribution for regulating transcription of TGF-β-responsive gene needs a broad range of nuclear proteins[3]. Recently, RUNX proteins, including RUNX3 gene were proved to interact through their C-terminal segment with Smads and recruit Smads to subnuclear sites of active transcription, thus exerting their biological control[4]. The function of RUNX proteins has been considered as the subnuclear acceptor proteins for signal transduction. On the contrary, Smads cannot be directed to the nuclear matrix in the absence of RUNX proteins[4]. Therefore, TGF-β-Smad signal pathway would be disrupted. RUNX3, one member of the RUNT domain family, was recently found with a loss of 40% - 60% of expression due to a highly frequency of hemizygous deletion and hypermethylation in gastric cancer[5]. Also, the gastric mucosa of RUNX3 knocked out mouse exhibited hyperplasia and suppressed apoptosis and growth-inhibition induced by TGF-β in epithelial cells[5]. Taken together, it is strongly suggested that RUNX3 gene be a novel tumor-suppressor gene.

Hepatocelluar carcinoma (HCC) is one of the most common causes of cancer death in the world, especially in Asia and Africa[6,7]. HCC, like many other kinds of human malignancy, has been reported to overexpress TGF-β[8]. The serum concentration of TGF-β is also elevated with tumor progression[9]. Therefore, HCC cells resistant to the anti-proliferative function of TGF-β may be a critical step in the development of HCC[10]. However, until the present no molecular event has been found to contribute to the impairment of TGF-β signal pathway in HCC[11]. It is well documented that aberrance of molecules of the pathway including TGF-β receptor, Smads 2, 3, 4, 6 and 7 was very rare in HCC[12-14]. The exact mechanism of HCC with loss of TGF-β responsiveness still remains unknown. A growing body of evidence showed that chromosome 1p36 was a common deletion region where just loci of RUNX3 gene exist[15,16]. Several putative tumor suppressor genes are believed to be in this region. But, different types of tumor have different regions of consensus deletion. For example, the consensus deletion of neuroblastoma has been mapped to 1p36.2-36.3, a region distal to the deleted region in HCC. While in HCC, a minimally deleted region of about 4 Mb on chromosome 1p36 was well defined[17]. Within the common deletion region, another candidate tumor suppressor gene, retinoblastoma protein (Rb)-interacting zinc finger gene (RIZ) was also identified[18]. Unfortunately, mutation of RIZ gene was not found in HCC[15]. Its role in hepatocarcinogenesis has not been clarified yet. Most notably, LOH encompassing RUNX3 gene occurs in early stage of HCC, even in precancerous condition[19,20]. In the current paper, we studied the genetic and epigenetic alterations of RUNX3 gene in HCC in order to find out new clues to the development of HCC.

Tissue samples

Sixty-two frozen HCC specimens and their adjacent normal liver tissue specimens were obtained from Southwest Hospital, Third Military Medical University, Chongqing, China. Informed consents were obtained from every patient. The patients’ age ranged 29-72 years with an average of 48.6 years. The male to female ratio was 52:10. The background liver showed cirrhosis in 53 (85.4%) cases, chronic persistent hepatitis in 6 cases (9.7%), and non-specific change in 3 (3.2%) cases. HBV was detected in 49 cases (79.0%), HCV was detected in 5 (8.1%) and non-virus hepatitis in 8 (12.8%). The number of cases with histological grades I, II and III was 8, 26, 28, respectively. Three pathologists reviewed independently one 5 μm thick section stained with hematoxylin and eosin.

DNA extraction

Frozen tissue samples were ground into very fine powder in liquid nitrogen, suspended in lysis buffer and treated with proteinase K. DNA was extracted by phenol-chloroform-isoamyl alcohol and ethanol precipitation[21]. Adjacent normal liver tissues were used as corresponding normal controls.

Single strand conformational polymorphism (SSCP) and DNA sequencing

A total of 6 exons of RUNX3 gene were screened for inactivation mutations with PCR-SSCP, cyclic sequencing on genomic DNA templates. The primers were designed with OLIGO software program (version 5.0; National Bioscience Inc., Plymouth. MN) using the genomic sequences obtained from GenBank (accession No. NT_004391). PCR primer pairs for amplification of RUNX3 gene are described in Table 1. Each PCR reaction except for exon 2 was performed under standard conditions in a 10 μl reaction mixture containing 1 μl of template DNA, 0.5 μM of each primer, 0.2 mM of each dNTP, 1.5 mM MgCI2, 0.5 unit of Taq polymerase (Ampli Taq GoldTM containing antibody to Taq, Roche), 0.5 μCi of 32P-dCTP (Amersham, Buckinghamshire, UK), and 1 μl of 10 × buffer. AdvantageR-GC genomic PCR kit (Clontech Laboratories, Inc., CA, USA) was used to amplify exon 2 containing CpG-rich sequence according to the user manual. The reaction mixture was denatured for 5 min at 95 °C and incubated for 35 cycles (denaturing for 30 s at 95 °C, annealing for 30 s at 51-67 °C, and extending for 30 s at 72 °C). A final extension was continued for 5 min at 72 °C in a thermal cycler (PE 480, USA). After amplification, the PCR products were denatured for 5 min at 95 °C at 1:1 dilution of sample buffer containing 98% formamide/5 mmol/L NaOH and loaded onto a SSCP gel (FMC mutation detection enhancement system, Intermountain Scientific, Kaysville, UT) with 10% glycerol. After electrophoresis, the gels were transferred to 3-mm Whatman paper and dried, and autoradiography was performed with Kodak X-OMAT film (Eastman Kodak, Rochester, NY). For the detection of mutations, DNAs showing mobility shifts were cut out from the dried gel, and reamplified for 30 cycles using the same primer set. Sequencing kit (Perkin-Elmer, Foster City, CA) was used according to the manufacturer’s recommendations. Cycling sequencing products were resolved on a 6% denatured sequencing gel (USBTM, Cleveland, USA).

Table 1 Primers used for PCR amplification of RUNX3 gene.
NameExonPrimer sequenceAnnealing Tm(°C)Size (bp)
Loss of heterozygosity (LOH) analysis

We found 1 polymorphic site during SSCP and sequencing analysis using the primer sets covering 6 exons. The polymorphic site had highly frequent information of heterozygote in HCC patients. This made it feasible as an intragenic polymorphic marker for LOH analysis of RUNX3 gene. PCR and SSCP conditions for LOH analysis were exactly the same as described above. PCR products from the corresponding normal and tumor DNAs were run on SSCP gel. Allelic loss was scored when the band intensity of one allelic marker was significantly decreased (more than 70% reduction) in tumor DNA as compared with that in normal DNA.

DNA methylation analysis of RUNX3 gene by methylation specific PCR (MSP)

The methylation status of RUNX3 gene was determined by sodium bisulfate treatment of DNA followed by methylation-specific polymerase chain reaction (MSP), as described with modification[22,23]. In brief, about 100 ng DNA was incubated in 0.2 M NaOH at 42 °C for 30 minutes in a total volume of 50 μl. After the addition of 350 μl of 3.6 M sodium bisulfate (Sigma) containing 1 mM hydroquinone at pH5, the samples were incubated for 4-5 hours at 55 °C in the dark. The modified DNA was recovered with 5 μl of glassmilk (BIO 101, Inc., CA, USA) and 800 μl of 6 M NaI. The glassmilk catching the modified DNA was washed three times with 70% ethanol at room temperature, and then treated with 0.3 M NaOH/90% ethanol once, washed twice again with 90% ethanol. The DNA was finally eluted from the dried pellet with 30 μl of 1 mM Tris-HCI (pH8.0) for 15 minutes at 55 °C. Five μL of bisulfate-modified DNA was subjected to MSP using two sets of primer specific for methylation detection and unmethylation detection as reported previously[5]. PCR was performed in a total volume of 30 μl containing 5 μl template DNA, 0.5 μM of each primer, 0.2 mM of each dNTP, 1.5 mM MgCI2, 0.5 unit of Taq polymerase (Ampli Taq GoldTM containing antibody to Taq, Roche) and 3 μl of 10 × buffer. The reaction solution was initially denatured at 95 °C for 1 minute. Amplification was carried out for 40 cycles at 95 °C for 30 s, at 63 °C for 30 s and at 72 °C for 30 s, followed by a final extension at 72 °C for 5 min. Controls without DNA were performed for each set for PCRs. Ten μl of PCR products was directly loaded onto 2% agarose gel containing ethidium bromide, and directly visualized under UV illumination, and photographed. The size of PCR products was 234 bp.

Frequency of LOH in RUNX3 gene and its clinical significance

We failed to detect a mutation in all six exons and partial intron adjacent to exon in 62 HCCs by PCR-SSCP and sequencing. But the polymorphic site with T to A transition at the second nucleotide of codon 18 was found at exon 1, a relatively high frequency of heterozygotes (26/62) was used as an intragenic marker to examine LOH of RUNX3 gene, and 34.6% (9/26) of informative cases showed allelic loss (Figure 1).

Figure 1
Figure 1 Allelic loss of RUNX3 gene in HCC. N: normal, T: tumor, Arrow indicate allelic loss.
Frequent hypermethylation of RUNX3 gene in HCC

On the basis of the presence of a CpG island in the 5’ region of RUNX3 gene, we examined the promoter 2 hypermethylation using two sets of primers specific for MSP reported by Li[5], and 48.4% (30/62) HCCs were found to have hypermethylation (Figure 2). Notably, we found the degree of hypermethylation was quite different among individual tumors by comparing with the intensity between unmethylation and methylation bands under the same PCR conditions. Although our methods could not be used to quantitate methylation, hypermethylation was not found in the matching normal liver tissues.

Figure 2
Figure 2 Methylation state of RUNX3 gene in HCC. M: Mo-lecular weight of 50 bp DNA ladder, u: unmethylation, m: methylation, N: normal, T: tumor.
Biallelic aberrant of RUNX3 gene in HCC

LOH and hypermethylation are two distinct ways to inactivate tumor suppressor gene. It is widely known that both of them were often involved in complete loss of gene function through cooperation. We, here, found 8 cases had hypermethylation in the 9 cases with LOH of RUNX3 gene.


RUNX3 gene has been found belonging to the runt domain family of transcription factors acting as master regulators of gene expression in major developmental pathway[3]. At present, three RUNX genes, RUNX1, RUNX2 and RUNX3, have been identified. All the three genes have been found to share a highly conserved region, called runt domain[24]. They have been shown to interact with Smads 1, 2, 3 and 5, and are indispensable in mediating Smads compound nuclear distribution and Smads specific binding to target DNA[25,4]. Therefore, RUNX proteins are important targets of TGF-beta-Smads signaling pathway. It is well documented that mutation of RUNX1 gene was associated with the development of acute myelogenous leukemia, while mutation of RUNX2 contributed to celeidocranial dysplasia (CCD)[25]. Recently, RUNX3 gene was found to play an important role during the development of gastric cancer. Its absence would lead to abnormal proliferation of gastric epithelial cells, lack of responsiveness to apoptosis and growth-inhibitory effect induced by TGF-beta in knock out mouse. Moreover, wild type RUNX3 gene significantly reduced the tumorigenesis ability of tumor cells, whereas mutant type RUNX3 gene would abolish the tumor suppressor action of RUNX3 in nude mice and drive tumor cells to grow much faster. In human primary gastric tumor, 60% cases do not significantly express RUNX3 gene because of hemizygous deletion and hypermethylation. Correlation between hypermethylation and under-expression or no expression was further confirmed in in vitro experiment[5]. These evidences strongly suggest that RUNX3 gene be a tumor suppressor gene.

It is commonly known that TGF-beta-Smad signal pathway is disrupted in HCC, but the exact mechanism of disruption of the signal pathway has still remained to be worked out[8]. Furthermore, LOH of 1p36 encompassing RUNX3 gene was a common event in pathogenesis of HCC[19]. So, it is reasonable to consider RUNX3 gene as a most possible target gene in the development of HCC. In this study, we found 34.6% of HCC showed LOH of RUNX3. This result is in concordance with the previously reported 30% more or less frequency of LOH at 1p36, an early event in the development of HCC[17,20]. Unfortunately, no mutation was discovered in 62 HCCs.

Hypermethylation is a regional event that occurs frequently in GC-rich sequences, called CpG islands, often located within the 5’ regulatory regions of non-transcribed genes. In contrast, actively transcribed genes are always in unmethylation status. Inactivation of genes by hypermethylation of their CpG islands has been well clarified[21,26,27]. Now, it has been recognized that hypermethylation of CpG islands in the promoter region is an alternative way to silence some cancer-associated gene as effectively as inactivation by mutation or deletion[28,29]. To date, genes involved in regulation of cell cycle[21,30], DNA repair[31], angiogenesis[32], and apoptosis[33,34] have been shown to be inactivated by hypermethylation which is also a frequent event in many human cancers including hepatocellular carcinoma[35-37]. For RUNX3 gene, transcription is regulated by two distinct promoters, P1 and P2. The major RUNX3 mRNA is transcribed from P2. The genomic region surrounding the P2 promoter constituted a large (4.2 kb) CpG island with a GC content of 64%[24]. These features showed P2 possessed the hallmark characteristics of GC-rich promoters. So, it is rational that transcription from P2 should be regulated by DNA methylation in theory. In practice, Li et al[5] confirmed the presumption in in vitro experiment. Our analysis of RUNX3 gene in HCCs provided evidence of promoter hypermethylation, a common alteration as well as an early event. Thus, methylation of the promoter region appears to be the dominant mode of inactivation of RUNX3 gene in human HCC, just the same as in human gastric cancer. Unfortunately, we did not detect the expression of RUNX3 gene due to unavailability of the sample and antibody. However, methylation changes are considered as a surrogate for altered expression of the gene product, thus, the detection of any abnormally methylated site is a strong indication that this mechanism could alter the expression levels of target genes. It was reported that hypermethylation correlated with LOH and often occurred before the allelic loss[38]. We found in our study 48.3%(30/62) of HCCs showed hypermethylation, which was higher than LOH (34.6%), and 88.9%(8/9) of HCCs with LOH had hypermethylation in the promoter of RUNX3 gene, hence being in line with Knudson’s two hit hypothesis and consistent with previous reports.

In summary, we have demonstrated a high frequency of RUNX3 gene aberration - allelic loss together with hypermethylation of the remaining alleles in HCC. These observations can provide the evidence that promoter hypermethylation and allelic loss are the major mechanisms for inactivation of RUNX3 gene in HCC. RUNX3 gene may be one of the key tumor suppressor genes at 1p36 which is the common deletion site of RUNX3 gene in HCC. Inactivation of RUNX3 gene functions, resulting in impairment of TGF-beta-Smads signal pathway and other tumor suppressor function, may be closely associated with the development of HCC.


Edited by Zhu LH and Wang XL

1.  Kloos DU, Choi C, Wingender E. The TGF-beta--Smad network: introducing bioinformatic tools. Trends Genet. 2002;18:96-103.  [PubMed]  [DOI]
2.  Moustakas A, Pardali K, Gaal A, Heldin CH. Mechanisms of TGF-beta signaling in regulation of cell growth and differentiation. Immunol Lett. 2002;82:85-91.  [PubMed]  [DOI]
3.  Leboy P, Grasso-Knight G, D'Angelo M, Volk SW, Lian JV, Drissi H, Stein GS, Adams SL. Smad-Runx interactions during chondrocyte maturation. J Bone Joint Surg Am. 2001;83-A Suppl 1:S15-S22.  [PubMed]  [DOI]
4.  Zaidi SK, Sullivan AJ, van Wijnen AJ, Stein JL, Stein GS, Lian JB. Integration of Runx and Smad regulatory signals at transcriptionally active subnuclear sites. Proc Natl Acad Sci USA. 2002;99:8048-8053.  [PubMed]  [DOI]
5.  Li QL, Ito K, Sakakura C, Fukamachi H, Inoue Ki, Chi XZ, Lee KY, Nomura S, Lee CW, Han SB. Causal relationship between the loss of RUNX3 expression and gastric cancer. Cell. 2002;109:113-124.  [PubMed]  [DOI]
6.  Korn WM. Moving toward an understanding of the metastatic process in hepatocellular carcinoma. World J Gastroenterol. 2001;7:777-778.  [PubMed]  [DOI]
7.  Qin LX, Tang ZY. The prognostic significance of clinical and pathological features in hepatocellular carcinoma. World J Gastroenterol. 2002;8:193-199.  [PubMed]  [DOI]
8.  Bedossa P, Peltier E, Terris B, Franco D, Poynard T. Transforming growth factor-beta 1 (TGF-beta 1) and TGF-beta 1 receptors in normal, cirrhotic, and neoplastic human livers. Hepatology. 1995;21:760-766.  [PubMed]  [DOI]
9.  Shirai Y, Kawata S, Tamura S, Ito N, Tsushima H, Takaishi K, Kiso S, Matsuzawa Y. Plasma transforming growth factor-beta 1 in patients with hepatocellular carcinoma. Comparison with chronic liver diseases. Cancer. 1994;73:2275-2279.  [PubMed]  [DOI]
10.  Song BC, Chung YH, Kim JA, Choi WB, Suh DD, Pyo SI, Shin JW, Lee HC, Lee YS, Suh DJ. Transforming growth factor-beta1 as a useful serologic marker of small hepatocellular carcinoma. Cancer. 2002;94:175-180.  [PubMed]  [DOI]
11.  Matsuzaki K, Date M, Furukawa F, Tahashi Y, Matsushita M, Sugano Y, Yamashiki N, Nakagawa T, Seki T, Nishizawa M. Regulatory mechanisms for transforming growth factor beta as an autocrine inhibitor in human hepatocellular carcinoma: implications for roles of smads in its growth. Hepatology. 2000;32:218-227.  [PubMed]  [DOI]
12.  Kawate S, Takenoshita S, Ohwada S, Mogi A, Fukusato T, Makita F, Kuwano H, Morishita Y. Mutation analysis of transforming growth factor beta type II receptor, Smad2, and Smad4 in hepatocellular carcinoma. Int J Oncol. 1999;14:127-131.  [PubMed]  [DOI]
13.  Kawate S, Ohwada S, Hamada K, Koyama T, Takenoshita S, Morishita Y, Hagiwara K. Mutational analysis of the Smad6 and Smad7 genes in hepatocellular carcinoma. Int J Mol Med. 2001;8:49-52.  [PubMed]  [DOI]
14.  Schutte M, Hruban RH, Hedrick L, Cho KR, Nadasdy GM, Weinstein CL, Bova GS, Isaacs WB, Cairns P, Nawroz H. DPC4 gene in various tumor types. Cancer Res. 1996;56:2527-2530.  [PubMed]  [DOI]
15.  Simon D, Knowles BB, Weith A. Abnormalities of chromosome 1 and loss of heterozygosity on 1p in primary hepatomas. Oncogene. 1991;6:765-770.  [PubMed]  [DOI]
16.  Yeh SH, Chen PJ, Chen HL, Lai MY, Wang CC, Chen DS. Frequent genetic alterations at the distal region of chromosome 1p in human hepatocellular carcinomas. Cancer Res. 1994;54:4188-4192.  [PubMed]  [DOI]
17.  Fang W, Piao Z, Simon D, Sheu JC, Huang S. Mapping of a minimal deleted region in human hepatocellular carcinoma to 1p36.13-p36.23 and mutational analysis of the RIZ (PRDM2) gene localized to the region. Genes Chromosomes Cancer. 2000;28:269-275.  [PubMed]  [DOI]
18.  Huang S. The retinoblastoma protein-interacting zinc finger gene RIZ in 1p36-linked cancers. Front Biosci. 1999;4:D528-D532.  [PubMed]  [DOI]
19.  Kuroki T, Fujiwara Y, Tsuchiya E, Nakamori S, Imaoka S, Kanematsu T, Nakamura Y. Accumulation of genetic changes during development and progression of hepatocellular carcinoma: loss of heterozygosity of chromosome arm 1p occurs at an early stage of hepatocarcinogenesis. Genes Chromosomes Cancer. 1995;13:163-167.  [PubMed]  [DOI]
20.  Sun M, Eshleman JR, Ferrell LD, Jacobs G, Sudilovsky EC, Tuthill R, Hussein MR, Sudilovsky O. An early lesion in hepatic carcinogenesis: loss of heterozygosity in human cirrhotic livers and dysplastic nodules at the 1p36-p34 region. Hepatology. 2001;33:1415-1424.  [PubMed]  [DOI]
21.  Liu LH, Xiao WH, Liu WW. Effect of 5-Aza-2'-deoxycytidine on the P16 tumor suppressor gene in hepatocellular carcinoma cell line HepG2. World J Gastroenterol. 2001;7:131-135.  [PubMed]  [DOI]
22.  Lehmann U, Hasemeier B, Lilischkis R, Kreipe H. Quantitative analysis of promoter hypermethylation in laser-microdissected archival specimens. Lab Invest. 2001;81:635-638.  [PubMed]  [DOI]
23.  Grunau C, Clark SJ, Rosenthal A. Bisulfite genomic sequencing: systematic investigation of critical experimental parameters. Nucleic Acids Res. 2001;29:E65-E65.  [PubMed]  [DOI]
24.  Bangsow C, Rubins N, Glusman G, Bernstein Y, Negreanu V, Goldenberg D, Lotem J, Ben-Asher E, Lancet D, Levanon D. The RUNX3 gene--sequence, structure and regulated expression. Gene. 2001;279:221-232.  [PubMed]  [DOI]
25.  Cohen MM. RUNX genes, neoplasia, and cleidocranial dysplasia. Am J Med Genet. 2001;104:185-188.  [PubMed]  [DOI]
26.  Baylin SB, Herman JG, Graff JR, Vertino PM, Issa JP. Alterations in DNA methylation: a fundamental aspect of neoplasia. Adv Cancer Res. 1998;72:141-196.  [PubMed]  [DOI]
27.  Esteller M, Corn PG, Baylin SB, Herman JG. A gene hypermethylation profile of human cancer. Cancer Res. 2001;61:3225-3229.  [PubMed]  [DOI]
28.  Jones PA, Laird PW. Cancer epigenetics comes of age. Nat Genet. 1999;21:163-167.  [PubMed]  [DOI]
29.  Wajed SA, Laird PW, DeMeester TR. DNA methylation: an alternative pathway to cancer. Ann Surg. 2001;234:10-20.  [PubMed]  [DOI]
30.  Roncalli M, Bianchi P, Bruni B, Laghi L, Destro A, Di Gioia S, Gennari L, Tommasini M, Malesci A, Coggi G. Methylation framework of cell cycle gene inhibitors in cirrhosis and associated hepatocellular carcinoma. Hepatology. 2002;36:427-432.  [PubMed]  [DOI]
31.  Esteller M, Hamilton SR, Burger PC, Baylin SB, Herman JG. Inactivation of the DNA repair gene O6-methylguanine-DNA methyltransferase by promoter hypermethylation is a common event in primary human neoplasia. Cancer Res. 1999;59:793-797.  [PubMed]  [DOI]
32.  Li Q, Ahuja N, Burger PC, Issa JP. Methylation and silencing of the Thrombospondin-1 promoter in human cancer. Oncogene. 1999;18:3284-3289.  [PubMed]  [DOI]
33.  Teitz T, Lahti JM, Kidd VJ. Aggressive childhood neuroblastomas do not express caspase-8: an important component of programmed cell death. J Mol Med (Berl). 2001;79:428-436.  [PubMed]  [DOI]
34.  Jones PA. Cancer. Death and methylation. Nature. 2001;409:141, 143-144.  [PubMed]  [DOI]
35.  Shen L, Ahuja N, Shen Y, Habib NA, Toyota M, Rashid A, Issa JP. DNA methylation and environmental exposures in human hepatocellular carcinoma. J Natl Cancer Inst. 2002;94:755-761.  [PubMed]  [DOI]
36.  Zhong S, Tang MW, Yeo W, Liu C, Lo YM, Johnson PJ. Silencing of GSTP1 gene by CpG island DNA hypermethylation in HBV-associated hepatocellular carcinomas. Clin Cancer Res. 2002;8:1087-1092.  [PubMed]  [DOI]
37.  Yoshikawa H, Matsubara K, Qian GS, Jackson P, Groopman JD, Manning JE, Harris CC, Herman JG. SOCS-1, a negative regulator of the JAK/STAT pathway, is silenced by methylation in human hepatocellular carcinoma and shows growth-suppression activity. Nat Genet. 2001;28:29-35.  [PubMed]  [DOI]
38.  Makos M, Nelkin BD, Reiter RE, Gnarra JR, Brooks J, Isaacs W, Linehan M, Baylin SB. Regional DNA hypermethylation at D17S5 precedes 17p structural changes in the progression of renal tumors. Cancer Res. 1993;53:2719-2722.  [PubMed]  [DOI]