Promoter hypermethylation of tumor-related genes in gastric intestinal metaplasia of patients with and without gastric cancer



Promoter hypermethylation is an alternative mechanism of gene silencing in human cancers including gastric cancer. While intestinal metaplasia (IM) is generally regarded as a precancerous lesion of the stomach, our study examines the presence of gene promoter hypermethylation in IM of patients with and without gastric cancer. We examined 31 samples of gastric cancer, 36 gastric IM (21 associated with gastric cancer and 15 from noncancer patients) and 10 normal gastric biopsies. Tissues containing foci of IM were carefully microdissected from paraffin-embedded section. Bisulfite-modifiedDNA was examined for gene promoter hypermethylation in DAP-kinase, E-cadherin, GSTP1, p14, p15, p16, RASSF1A and hMLH1 by methylation-specific-PCR. None of the control gastric tissues had hypermethylation detected, but gene promoter hypermethylation was frequently detected in gastric cancer and IM. The mean number of methylated genes in cancer and IM was 3.0 and 1.4, respectively (p < 0.0001). Methylation in IM from cancer patients was all associated with concurrent methylation in the corresponding tumor samples. The numbers of methylated genes were similar in IM obtained from cancer and noncancer patients. By examining the methylation patterns of these genes, 3 differential methylation patterns were recognized: hypermethylation was more frequent in cancer than in IM (DAP-kinase, p14, p15 and p16); comparable frequencies of methylation in cancer and IM (E-cadherin and hMLH1); and no methylation (GSTP1). Aberrant methylation in tumor-related genes is frequently detected in gastric IM of both cancer and noncancer patients, suggesting their early involvement in the multistep progression of gastric carcinogenesis. © 2002 Wiley-Liss, Inc.

Intestinal metaplasia (IM) is considered to be a precancerous lesion according to the Correa's proposed cascade of gastric carcinogenesis.1 Individuals with IM are estimated to have a 10-fold increase in the risk of developing gastric malignancy.2, 3, 4 However, the molecular pathways underlying progression of IM into cancer remains elusive. Past studies have identified a number of molecular alterations in IM. These alterations include p53 mutation,5, 6 overexpression of transforming growth factor alpha and epidermal growth factor receptor,7 microsatellite instability8, 9 and overexpression of cyclooxygenase-2.10, 11 However, these changes are only present in a subset of IM.

On the other hand, epigenetic changes have recently emerged as an important cause of tumorigenesis. Of particular interest is that hypermethylation of the CpG island of tumor-related genes can result in transcriptional silencing of the gene with subsequent loss of protein expression. Many studies have previously shown that CpG island hypermethylation is frequently detected in gastric cancer.12, 13, 14, 15, 16, 17, 18 Additionally, we have recently demonstrated the presence of concurrent hypermethylation of multiple tumor-related genes in gastric cancer.19 Herein, we examined the role of promoter hypermethylation in gastric intestinal metaplasia, a preneoplastic gastric lesion, in patients with and without gastric cancer in order to elucidate the chronology of the development of gene promoter hypermethylation in the multistep progression of the gastric carcinogenesis pathway. A panel of 8 tumor-related genes was used to determine the methylation pattern found in gastric cancer and its precursor lesions.



Archival gastrectomy specimens were randomly retrieved from 31 gastric cancer patients. Among these 31 cases, paired cancer and IM were available in 21 cases. Additionally, paraffin-embedded slides containing IM were randomly selected from the endoscopic biopsies of 15 noncancer patients with chronic gastritis. As control, gastric biopsies from 10 normal control subjects without IM were examined. All patients gave informed consent for obtaining the specimens.

Gastric carcinomas were classified into diffuse and intestinal types as defined by Lauren.20 Clinicopathologic characteristics of the tumors including demographic data, the site of the tumor and the presence of lymphatic and distant metastasis were recorded.

DNA preparation

Fifteen paraffin sections, 5 μm thick, were retrieved from each case for manual microdissection to select areas containing tumors or intestinal metaplasia as described previously.8 All sections were lightly stained with hematoxylin for improved visualization. In cancer patients, the tumor and the corresponding area containing IM only without cancer, if available, were identified. Areas containing IM only were also selected from noncancer patients. None of the IM specimens were contaminated by cancer or dysplasia. The microdissected tissues were retrieved by using xylene, alcohol and digested with proteinase K solution. DNA was then extracted by using the DNA extraction kit as described by the manufacturer (Boehringer Mannheim, Indianapolis, IN).

Bisulfite conversion

Treatment of DNA samples with bisulfite converts all unmethylated cytosines to uracils, while leaving methylcytosines unaltered. This allows the subsequent differentiation of methylated from unmethylated sequences by primers that target specifically to either the methylated or unmethylated sequences (methylation-specific PCR or MSP). Bisulfite modification of DNA (1 μg) was carried out by the CpGenome DNA Modification Kit (Intergen, Purchase, NY) according to the manufacturer's instruction. Modified DNA samples were precipitated with ethanol and resuspended in TE buffer (10 mM Tris, 1 mM EDTA, pH 7.5).

Methylation-specific PCR (MSP)

The bisulfite-modified DNA samples were amplified by primers specific for either the methylated or unmethylated sequences. Primer sequences were based on previous published reports (Table I).21, 22, 23, 24, 25 PCR was performed in 25 μl reaction volumes containing 1× PCR buffer, 0.25 mM each of the deoxynucleoside triphosphates, 1 μM of each primer and 1 unit of Taq polymerase (AmpliTaq Gold; PE Applied Biosystems, Foster City, CA). The temperature profiles for the amplification were: 12 min at 95°C, 40 cycles of denaturing at 95°C for 30 sec, 45 sec of annealing at specific temperature (Table I), 60 sec of extension at 72°C and a final extension step of 5 min at 72°C. Positive control (universal methylated DNA; Intergen) and negative control (distilled water without DNA template) were included in each amplification process. PCR products were analyzed in 2% agarose gel or 10% acrylamide gel stained with ethidium bromide (Fig. 1). All reactions were repeated to ensure consistent, reproducible results were obtained.

Table I. Primer Sequences used in Methylation-Specific PCR
Genes SequenceTmRef
DAP-kinaseUnmethylated5′-GGAGGATAGTTGGATTGAGTTAATGTT-3′ (sense)5823
  5′-CAAATCCCTCCCAAACACCAA-3′ (antisense)  
 Methylated5′-GGATAGTCGGATCGAGTTAACGTC-3′ (sense)58 
  5′-CCCTCCCAAACGCCGA-3′ (antisense)  
GSTP1Unmethylated5′-GATGTTTGGGGTGTAGTGGTTGTT-3′ (sense)6023
 Methylated5′-TTCGGGGTGTAGCGCTCGTC-3′ (sense)60 
  5′-GCCCCAATACTAAATCACGACG-3′ (antisense)  
E-cadherinUnmethylated5′-TAATTTTAGGTTAGAGGGTTATTGT-3′ (sense)5321
  5′-CACAACCAATCAACAACACA-3′ (antisense)  
 Methylated5′-TTAGGTTAGAGGGTTATCGCGT-3′ (sense)57 
p14Unmethylated5′-TTTTTGGTGTTAAAGGGTGGTGTAGT-3′ (sense)6024
 Methylated5′-GTGTTAAAGGGCGGCGTAGC-3′ (sense)60 
  5′-AAAACCCTCACTCGCGACGA-3′ (antisense)  
p15Unmethylated5′-TGTGATGTGTTTGTATTTTGTGGTT-3′ (sense)6021
 Methylated5′-GCGTTCGTATTTTGCGGTT-3′ (sense)60 
  5′-CGTACAATAACCGAACGACCGA-3′ (antisense)  
p16Unmethylated5′-TTATTAGAGGGTGGGGTGGATTGT-3′ (sense)6021
  5′-CAACCCCAAACCACAACCATAA-3′ (antisense)  
 Methylated5′-TTATTAGAGGGTGGGGCGGATCGC-3′ (sense)60 
  5′-GACCCCGAACCGCGACCGTAA-3′ (antisense)  
 Methylated5′-ACGTAGACGTTTTATTAGGGTCGC-3′ (sense)60 
  5′-CCTCATCGTAACTACCCGCG-3′ (antisense)  
RASSF1AUnmethylated5′-TTTGGTTGGAGTGTGTTAATGTG-3′ (sense)6025
 Methylated5′-GTGTTAACGCGTTGCGTATC-3′ (sense)60 
  5′-AACCCCGCGAACTAAAAACGA-3′ (antisense)  
Figure 1.

Methylation-specific PCR (MSP). Representative MSP results of intestinal metaplasia (IM) from cancer and noncancer patients. In each panel, the upper panel demonstrates the MSP results of IM from cancer patients and the lower panel shows the MSP results from noncancer IM. U, results of unmethylated primers; M, methylated primers. IVD is the universally methylated positive control, and water (H2O) is used as negative control for each amplification.

Statistical analysis

All statistical calculations were made by the SPSS software (version 10.0, SPSS, Chicago, IL). The frequency of promoter hypermethylation between tumor and IM was compared by χ2 test. The difference in the number of methylated genes between tumor and IM was compared by the t-test. The associations between clinicopathologic parameters and methylation of individual marker were analyzed by χ2 test (categorical data) or t-test (age).


Methylation in gastric cancer and IM

Eight tumor-related genes previously reported to have promoter hypermethylation in human cancer were examined.12, 13, 14, 15, 16, 17, 18, 19 These include p16 (CDKN2A), p15 (INK4b), p14 (ARF), death-associated protein kinase (DAP-kinase), hMLH1, E-Cadherin, Ras association domain family 1A (RASSF1A) and GSTP1. p16 and p14 are closely related genes located on chromosome region 9p21. These genes share a common exon in different reading frames, and they all function as growth suppressors. The p16 inhibits G1 cyclin-dependent kinases, CDK4 and CDK6, and induces a G1-phase arrest. The p15 gene is also an inhibitor of cyclin-dependent kinase 4, which is an important mediator of cell cycle control, especially in the pathway stimulated by transforming growth factor β. DAP-kinase, a pro-apoptotic calcium-regulated serine/threonine kinase, is involved in interferon-γ-induced apoptosis. hMLH1 is a DNA mismatch-repair gene, and mutation or promoter hypermethylation frequently results in the microsatellite instability phenotype. RASSF1A, a new family of genes encoding RAS-binding proteins located at 3p21.3, was recently identified to be a tumor suppressor gene. E-cadherin, a cell adhesion molecule and a potential invasion/metastasis suppressor, frequently undergoes hypermethylation in human gastric cancers, particularly those of the undifferentiated-scattered histologic subtype.17, 19 Promoter hypermethylation of these 7 genes has been previously reported in gastric carcinoma.12, 13, 14, 15, 16, 17, 18, 19 We also included GSTP1, which is a detoxification gene, but promoter hypermethylation is rarely detected in human gastric cancer.19

The presence of methylated allele was indicated by the presence of a positive PCR product with the methylated-specific primers (Fig. 1). Notably, even with the use of microdissection for DNA extraction, it is common to find the presence of both methylated and unmethylated alleles in tumors and IM. This is most likely due to the inclusion of normal tissue within the tumor or IM as well as the extreme sensitivity of MSP.21

Among the 10 control nonmetaplastic gastric biopsies, none had methylation as detected by MSP. On the other hand, 29 (93.5%) tumors and 28 (77.8%) IM samples had methylation detected in at least 1 of the genes examined. The overall frequencies of methylation in individual loci are summarized in Table II. Nine (29.0%) tumors demonstrated hypermethylation in ≥4 (50%) of the genes examined. On the other hand, only 1 (2.8%) IM sample had methylation detected in ≥4 genes. The number of methylated genes was significantly higher in primary cancer than in IM (mean 3.0 vs. 1.4, p < 0.0001; Fig. 2). For intestinal metaplasia, the mean number of methylated genes in cancer and noncancer patients were both 1.4. The frequencies of methylation in individual loci were comparable for IM obtained from cancer and noncancer patients (p > 0.3). However, the maximum number of methylated genes in IM obtained from cancer and noncancer patients were 4 and 2, respectively (Fig. 2).

Table II. Frequencies of Hypermethylation in Cancer and Intestinal Metaplasia
 Tumor (n = 31)All IM (n = 36)p*IM (CA) (n = 21)IM (Non-CA) (n = 15)p**
  • CA, cancer patients; IM, intestinal metaplasia; NA, not applicable.–

  • *

    Tumor vs. all IM.–

  • **

    Tumor vs. IM (CA) vs. IM (non-CA).–IM (CA) vs. IM (non-CA), p > 0.3.

DAP-kinase22 (71%)14 (39%)0.0149 (43%)5 (33%)0.027
E-cadherin14 (45%)13 (36%)0.476 (29%)7 (47%)0.42
p1410 (32%)3 (8%)0.0272 (10%)1 (7%)0.046
p1515 (48%)4 (11%)0.0013 (14%)1 (7%)0.003
p1614 (45%)5 (14%)0.0073 (14%)2 (13%)0.018
hMLH19 (29%)8 (22%)0.584 (19%)4 (27%)0.71
RASSF1A8 (26%)4 (11%)0.203 (14%)1 (7%)0.25
Figure 2.

Frequency distribution of genes that showed methylation in gastric cancer and in intestinal metaplasia (IM) from cancer (CA) and noncancer (non-CA) patients.

In patients with gastric cancer, promoter hypermethylation in IM was all associated with hypermethylation in the corresponding cancer tissues (Table III). It is also interesting to note that concordant methylation patterns were observed in tumor and their corresponding IM.

Table III. Methylation Patterns in Primary Gastric Tumor and Intestinal Metaplasia from Cancer and Noncancer Patients
  1. CA 1–21, cancer patients with tumor and corresponding IM samples; T 1–10, cancer patients with tumor samples only; G 1–15, IM from noncancer patients.–DAP, DAP-kinase; E-cad, E-Cadherin; GST, GSTP1; RAS, RASSF1A; T, tumor samples; IM, intestinal metaplasia.–Black boxes indicate methylated sequences, whereas blank boxes indicate unmethylated sequences. There were no corresponding samples in boxes with slashes.

inline image

Differential methylation patterns between cancer and IM

As shown in Table II, gene promoter hypermethylation was frequently detected in all tumor-related genes except for GSTP1. Of the 8 tumor-related genes, DAP-kinase had the highest frequency of methylation for both primary gastric cancer (71%) and IM (39%). The frequencies of methylation in all genes were comparable in IM obtained from cancer and noncancer patients. However, based on the differential methylation frequencies of cancer and IM, 3 different patterns were recognized. The first pattern was observed in DAP-kinase, p14, p15 and p16 where tumor tissues had significantly more frequent methylation than in IM (Table II). Methylation in RASSF1A also tended to be more frequent in cancer than in IM (26% and 11%, respectively) but the difference did not reach statistical significance. The frequency of hypermethylation was comparable in both cancer and metaplasia for E-Cadherin and hMLH1. None of the tumor and IM samples tested showed methylation in GSTP1.

Association between methylation and clinicopathologic parameters of cancer

The association between promoter hypermethylation and relevant clinicopathologic parameters including gender, age, site, histologic type and presence of lymphatic metastasis were shown in Table IV. There was no significant correlation between promoter hypermethylation in all 8 tumor-related genes and clinicopathologic characteristics of the tumors.

Table IV. Association between Methylation and Clinicopathologic Features of Gastric Cancer
  1. Early CA, Early T1 cancer; LN, lymph node; M, methylated; U, unmethylated; p, p-value.

Number229 1417 1021 1516 1417 921 822 
Site distal2081.013151.09191.014141.013151.09190.547211.0
Intestinal type1461.09111.07131.07130.078120.485150.685151.0
LN metastasis1740.1111100.287141.011100.707140.126151.05161.0
Early CA300.54210.58030.53210.6210.58121.0030.55
Distant metastasis201.0020.49021.0020.48020.48110.50021.0


Although IM is generally regarded as a premalignant gastric lesion, the role and significance of various genetic alterations found in gastric intestinal metaplasia is poorly defined. Many of these genetic alterations are detected in a subset of IM only.5, 6, 7, 8, 9, 10, 11 In our study, we elucidated the role of promoter hypermethylation in gastric IM. Promoter hypermethylation results in transcriptional silencing of genes and is increasingly recognized to play an instrumental role in the gastric carcinogenesis process.12, 13, 14, 15, 16, 17, 18, 19 These epigenetic changes have been demonstrated in a number of tumor-related genes and in various forms of human cancer. Recently, a study had reported the detection of promoter methylation in p16, hMLH1 and DAP-kinase in gastric intestinal metaplasia.26 However, all metaplastic tissues were obtained from noncancer patients only. In contrast, we have examined metaplastic tissues from both cancer and noncancer patients. By comparing the methylation patterns in (i) IM and their corresponding cancer and (ii) IM from cancer and noncancer patients, we could better delineate the chronology of development of epigenetic changes in the multistep gastric cancer development cascade.

Our results showed that hypermethylation was frequently detected in both cancer and IM. The number of genes with aberrant methylation was significantly higher in primary tumor than in IM. Accordingly, hypermethylation in ≥4 genes was detected in 9 (29%) tumors but only in 1 (2.7%) IM sample (Fig. 2). Hypermethylation in IM also appears to be a specific event since aberrant methylation in IM was all associated with similar changes in the corresponding tumor. On the other hand, IM from both cancer and noncancer patients showed comparable frequencies of methylation. It thus appears that epigenetic alterations are already present in the preneoplastic stage even in patients without cancer. Thus, the continuous accumulation of aberrant methylation may be an important pathogenetic mechanism for gastric cancer development.

In addition to the demonstration of promoter hypermethylation in gastric cancer and IM, our study illustrates the differential methylation patterns of gastric cancer and their precursor lesions. The differential increase in methylation frequencies from premalignant to malignant tissues in DAP-kinase, p14, p15 and p16 suggests a continuous accumulation of hypermethylation in these loci during the progression from metaplasia to cancer. Intuitively, they are involved in the early initiation as well as in the subsequent transformation process. In contrast, E-cadherin and hMLH1 had comparable methylation frequency in cancer and IM, suggesting that these loci are more likely to be involved in the early initiation process of cancer development. The absence of GSTP1 methylation illustrates the gene specificity of methylation in gastric cancer. Accordingly, the study of methylation patterns may help to elucidate the relative importance and contributions of different tumor-related genes in different stages of gastric cancer development. Since not all IM will progress to gastric cancer and promoter hypermethylation is not detected in all metaplastic tissues, follow-up study will be helpful to elucidate the prognostic value of detecting promoter hypermethylation in IM. In this regard, detection of hypermethylation may be developed into a diagnostic tool to identify patients at risk for further histologic progression. In fact, 1 of the noncancer IM patients was detected to have early gastric cancer 2 years after the initial examination. However, not all non-IM patients in our study had repeated endoscopic examination and the actual percentage of progression could not be determined in our study.

In summary, promoter hypermethylation is frequently detected in IM, a premalignant gastric lesion, irrespective of the presence of tumor or not. The differential methylation frequencies in cancer and IM may help to elucidate the chronology of involvement of different tumor-related genes in gastric cancer development. Further studies are necessary to determine whether the presence of promoter hypermethylation in gastric IM is associated with higher risk of subsequent cancer development.


The authors express their gratitude to Dr. R. Leong for proofreading the manuscript and Ms. J. Ching for statistical advice.