ANTICANCER RESEARCH 33: 3765-3774 (2013)

Frequent Promoter Methylation of CDH1 in Non-neoplastic Mucosa of Sporadic Diffuse Gastric Cancer

KYUNG HWA LEE1*, DAVID HWANG2*, KI YOUNG KANG2, SOONG LEE3, DONG YI KIM4, YOUNG EUN JOO5 and JAE HYUK LEE1

Departments of 1Pathology, 4Surgery, and 5Internal Medicine, Chonnam National University Medical School, Gwangju, Republic of Korea; Departments of 2Anatomy and 3Internal Medicine, College of Medicine, Seonam University, Namwon, Republic of Korea

Abstract. Background/Aim: To identify promoter observed in recent decades (1, 2). Diffuse gastric cancer methylation as a major silencing mechanism in potential (DGC) accounts for approximately 30% of all gastric precursor lesions of sporadic diffuse gastric cancer (DGC), carcinomas, and the prognosis is poor particularly for young we investigated promoter methylation of CDH1 (E-Cadherin patients (3, 4). It has long been known that DGCs show gene) in a series of DGCs and matched normal mucosa. diminished homophilic cell-to-cell cohesion (5). Inactivating Materials and Methods: The extent of CDH1 gene promoter germline CDH1 (E-Cadherin gene) mutation has been methylation was explored using methylation-specific described in the families with hereditary DGC, an polymerase chain reaction (MSP) and pyrosequencing (PS) autosomal-dominant disease characterized by clustering of in 72 DGCs with a matched pair of normal mucosa. Results: early-onset DGC (6, 7). The diminished or lack of E- MSP and PS revealed CDH1 promoter methylation in 73.6% Cadherin immunoreactivity observed in hereditary DGC cells (53/72) and 77.8% (56/72) of DGC samples, respectively. PS harboring CDH1 mutations is consistent with bi-allelic detected CDH1 methylation in 70.8% (51/72) and 72.2% CDH1 inactivation by a second-hit mechanism that leads to (52/72) of matched normal mucosa from adjacent and remote E-Cadherin loss and determines diffuse cancer development foci, respectively. In comparison, CDH1 promoter (8, 9). Epigenetic silencing of CDH1 through promoter methylation was detected in 12/24 normal gastric mucosa methylation plays an important role as a second-hit samples without DGC, and this difference was statistically mechanism in patients with hereditary DGC and patients significant (p=0.016). Conclusion: The current study showed with germline CDH1 mutations. Moreover, CDH1 promoter that the majority of DGC cases demonstrated concurrent methylation is also frequently found in sporadic DGC, and CDH1 promoter methylation in tumor tissues and in non- is responsible for inactivation of the second allele following neoplastic mucosa. somatic mutations of the first allele (10, 11). In this study, we hypothesized that CDH1 promoter Gastric carcinoma is one of the most common malignancies methylation would be detected more frequently in the non- and remains the second leading cause of cancer death neoplastic gastric mucosa from patients with DGCs than in worldwide although a universally declining trend has been the normal gastric mucosa from patients without DGCs. By performing a comparison of qualitative methylation determined by methylation-specific polymerase chain This article is freely accessible online. reaction (MSP) with quantitative methylation determined by pyrosequencing (PS), CpG methylation in the CDH1 *These Authors contributed equally to this work. promoter was investigated in 72 cases of DGCs with matched pairs of tumor-adjacent and tumor-remote gastric Correspondence to: Jae Hyuk Lee, MD, Ph.D., Department of mucosa and in 24 control samples of normal gastric mucosa Pathology, Chonnam National University Medical School, #8, Hak- along with E-Cadherin protein expression determined by dong, Dong-gu, Gwangju, 501-757, Republic of Korea. Tel: +82 immuno-histochemistry. Additionally, in order to determine 613797073, Fax: +82 613797099, e-mail: [email protected] which method has greater potential for clinical application, Key Words: Methylation, human CDH1 protein; stomach neoplasm, the correlation between methylation status and diffuse gastric cancer, methylation-specific polymerase chain clinicopathological variables, as well as survival rates was reaction, pyrosequencing. also investigated.

0250-7005/2013 $2.00+.40 3765 ANTICANCER RESEARCH 33: 3765-3774 (2013)

Materials and Methods Table I. Clinicopathological features of selected patients.

Patients and specimens. DGC specimens were obtained from 72 Characteristics Total n Proportion (%) patients who underwent surgical resection at the Chonnam National University Hospital between 2004 and 2008. The cases were identified Age, years <55 33 45.8 retrospectively based on clinicopathological data. Patients with a ≥55 39 54.2 history of familial gastric cancer were excluded. Samples of gastric Gender M 37 51.4 F 35 48.6 carcinoma; non-neoplastic mucosa adjacent to the tumor, but at least Location Upper third 9 12.5 1.5 cm away from the tumor; and non-neoplastic mucosa remote from Middle third 13 18.1 the tumor, at least 5 cm away from the tumor, respectively were Lower third 49 68.1 obtained. To exclude the possibility of a field defect, a control group of Whole 1 1.4 normal gastric mucosas from 24 patients with benign gastric pathology Tumor size, cm <2.5 35 48.6 was included. The diagnoses of patients in the control group were as ≥2.5 37 51.4 follows: 11 cases of gastric peptic ulcer, five cases of ectopic pancreas, Lauren classification Diffuse 66 91.7 four cases of leiomyoma, two cases of schwannoma, and two cases of Mixed 6 8.3 hyperplastic polyps. None of the control patients had a clinical history Differentiation Poor 52 72.2 of gastric dysplasia or carcinoma. The tumors were divided into two Signet 20 27.8 histological subgroups: poorly-differentiated adenocarcinomas and Depth of invasion T1 45 62.5 signet ring carcinomas. Of the 72 gastric tumors, 40 tumors were of T2 20 27.8 the pure diffuse type and 32 tumors were of the mixed type, according T3 7 9.7 to the Lauren classification. The extent of the tumor was determined Lymph node metastasis Absent 49 68.1 using the American Joint Committee on Cancer (AJCC) staging Present 23 31.9 criteria (12). Patients were followed-up for an average duration of 92.4 Distant metastasis Absent 69 95.8 months (range 3 to 108 months) to determine the clinical outcomes. Present 3 4.2 This study was approved by the Institutional Review Board of TNM stage I 52 72.2 Chonnam National University Hospital (#2013-78). II 11 15.3 III 6 8.3 IV 3 4.2 Microdissection and DNA extraction. Tissues were carefully microdissected on H&E-stained slides, as described previously (13). Genomic DNA was extracted from the microdissected tissue according to the manufacturer’s instruction (Boehringer Mannheim, Germany). In brief, the microdissected tissue was treated with Pyrosequencing. PS of the PCR products from the bisulfite-treated 50 μL of buffer containing 0.5% Tween 20 (Boehringer Mannheim, DNA was used to determine the percentage of methylation of the Germany), 20 μg of proteinase K (Boehringer Mannheim), eight CpG sites within the CDH1 gene promoter. DNA (30-50 ng) 50 mmol/L Trizma base at pH 8.9, and 2 mmol/L EDTA, followed was amplified by PCR in a 50-μl reaction mixture containing forward by incubation at 56˚C for 12-18 h. Proteinase K was inactivated by primer (5’-TTTGATTTTAGGTTTTAGTGAGT-3’) and biotinylated incubating the samples at 100˚C for 10 min. reverse primer (5’-biotin-ACCACAACCAATC AACAAC-3’). PS was performed according to the manufacturer’s instructions using the Bisulfite treatment of DNA and MSP. The methylation status of sequencing primer (5’ TGTGTGTT TGAGGATTTG-3’) on a PSQ 96 CDH1 exon 1 was determined by MSP after bisulfite treatment of MA System (Biotage AB, Uppsala, Sweden). Amplification was the DNA, as described previously but with some modifications (14). performed under the following conditions: 95˚C for 15 min, 35 cycles Methprimer (http://www.urogene.org/methprimer/index1.html) was of 95˚C for 30 s, 54˚C for 40 s, 72˚C for 1 min, and finally 72˚C for used to design MSP primer sets for methylated DNA (forward, 5’- 10 min. The methylation status determined by the PS assay was TTAGGTTAGAGGGTTATCGCG-3’, and reverse, 5’- TAACTAA analyzed as both a continuous variable (methylation level) and a AAATTCACCTACCGAC-3’, yielding a 116-bp product) and categorical variable (methylation negative: methylation level <10%; unmethylated DNA (forward 5’-TAATTTTAGGTTAGAGGG methylation positive: methylation level ≥10%). An empirical cut-off TTATTGT-3’, and reverse 5’ CACAACCAATCAACACA-3’, of ≥10% was selected, according to the limit of the unmethylated yielding a 98-bp product) in CDH1 exon 1 (Figure 1). Briefly, for control. The technical controls for PS revealed median methylation MSP analysis, 2 μg of genomic DNA were denatured by NaOH and of 4% (unmethylated control) and 90% (methylated control) in the modified by bisulfite. The modified DNA was purified using a DNA eight sequential CpG sites, defining the detection limits of the assay cleanup kit (Promega, Madison, WI, USA), treated with NaOH to (Figure 2A and B). disulfonate, precipitated with ethanol and resuspended in water. PCR amplification was carried out in a GeneAmp PCR System 9700 Immunohistochemical staining and assessment. Immunohisto- thermocycler (Perkin Elmer, Santa Clara, California, USA). PCR chemical staining for E-Cadherin protein was performed on 4-μm products were visualized on 2% agarose gels stained with ethidium thick paraffin-embedded tissue sections. Immunohistochemistry was bromide. Genomic DNA after bisulfite modification and after performed using automated standard procedures (Flex Plus detection treatment with Sss1 methylase (New England Biolabs, Ipswich, MA, system; DAKO, Carpinteria, CA, USA) and monoclonal antibody USA) was used as a positive control for methylated DNA. A sample to E-Cadherin (Clone M3612; DAKO Carpinteria, CA, USA) at was deemed positive for methylation if a band was seen in the DNA dilution 1:100. The specific standardized protocol provided by the that were amplified using the methylated reaction primers (Figure 2). manufacturer was followed. Either the fundic or pyloric gland tissue

3766 Lee et al: CDH1 Methylation in Gastric Cancer

Table II. Status of E-Cadherin gene (CDH1) promoter methylation in gastric carcinoma and normal mucosa by methylation-specific polymerase chain reaction (MSP) and pyrosequencing (PS).

Group No. MSP p-Value PS p-Value PS mean p-Value

MUM M UM

T 72 53 (73.6%) 19 (26.4%) 0.032 56 (77.8%) 16 (22.2%) 0.010 17.1% 0.016 N1 72 46 (63.9%) 26 (36.1%) 0.228 51 (70.8%) 21 (29.2%) 0.063 18.7% 0.002 N2 72 49 (68.1%) 23 (31.9%) 0.111 52 (72.2%) 20 (27.8%) 0.046 15.1% 0.053 N3 24 12 (50.0%) 12 (50.0%) 12 (50.0%) 12 (50.0%) 11.5%

T, Diffuse gastric cancer; N1, tumor-adjacent normal mucosa; N2, tumor-remote normal mucosa; N3, normal mucosa with benign gastric pathology.

was used as a positive control for primary antibody binding. Positive Representative results are shown in Figure 2. Of the 72 DGCs, immunoreactions for the E-Cadherin protein appeared as a strong 53 (73.6%) exhibited methylation of CpG islands of the CDH1 membranous pattern along the cytoplasmic border (Figure 3). gene, and 19 (26.4%) did not exhibit promoter methylation, as Evaluation of the immunostaining results was performed as amplification signals were obtained exclusively with primers previously described (15). The membranous expression of E- Cadherin was defined when E-Cadherin was solely localized in the specific to the unmethylated DNA. Of the 72 paired samples of cell membrane. Abnormal expression of E-Cadherin was defined tumor-adjacent and tumor-remote gastric mucosas, 46 (63.9%) when there was a lack of expression of E-Cadherin or the protein and 49 (68.1%) samples exhibited CDH1 promoter methylation, was expressed in the cytoplasm. The slides were coded without respectively (Table II). Tumor samples also exhibited knowledge of the molecular analyses, and used to evaluate the unmethylated promoter sequences, possibly due to staining in the tumor cell cytoplasm. contamination with small amounts of stromal and endothelial cells. In the 24 normal tissues with benign gastric pathology, 12 Statistical analysis. Statistical analyses were completed using SPSS version 20.0 (Chicago, IL, USA). A nominal logistic regression model (50.0%) exhibited gene methylation. Interestingly, 69 out of 72 analyzed the methylation status changes of CDH1 in gastric cases (90.3%) had CDH1 promoter methylation in either of the carcinoma and paired tumor-adjacent and tumor-remote normal- paired non-neoplastic mucosa. The difference in the frequency appearing mucosa. Comparison of CDH1 methylation status by MSP of CDH1 promoter methylation between normal mucosa and and PS, and the associations of promoter gene methylation status with DGC was statistically significant (50.0% vs. 73.6%, clinicopathological variables were analyzed using the χ2 test or respectively, p=0.032). Although CDH1 promoter methylation Fisher’s exact probability test, as appropriate. Survival curves were occurred more frequently in tumor-adjacent and tumor-remote estimated using the Kaplan–Meier method and were compared using the log-rank test. For multivariate analysis, the independent prognostic mucosa than in normal mucosa, this difference was not factors were determined using Cox’s proportional hazard model. All statistically significant (p=0.228 and 0.112, respectively). statistical analyses were performed at a significance level of p<0.05. CpG methylation status of the CDH1 promoter using PS and Results correlation with MSP data. To quantitatively measure the extent of CDH1 promoter methylation in DGCs, eight CpG Clinicopathological findings. The average age of the patients sites in the CDH1 promoter region were assessed by was 54.9 years (range 22-81 years); the cohort included 37 quantitative PS. PS analysis of the DGC samples revealed males and 35 females. The tumor extent was T1 in most cases low levels of CDH1 methylation, with wide variation among (45/72). Tumor metastasis to lymph nodes was observed in 23 the methylated tumors (Figure 2B). The overall mean out of 72 cases and distant metastasis to other organs was methylation densities of eight CpG sites in the CDH1 observed in three cases. Disease in the majority (52/72) of promoter in 72 DGCs, 72 tumor-adjacent mucosa, 72 tumor- cases was TNM stage 1. Relevant data are shown in Table I. remote mucosa, and 24 normal mucosa were 17.1%, 18.7%, 15.1%, and 11.5%, respectively (Table II). The mean Frequency of CDH1 promoter methylation in DGC and normal percentage methylation of CDH1 was significantly higher in gastric mucosa using MSP. In order to determine the proportion DGCs than in normal gastric mucosa (p=0.016). of DGC cases with CDH1 promoter methylation, 72 tumor By applying an empirical cut-off of ≥10% to discriminate samples with matched pairs of non-neoplastic gastric mucosas between CDH1 methylation and non-methylation, the were analyzed by MSP. Additionally, 24 samples of qualitative results for CDH1 methylation in DGCs obtained histologically -normal gastric epithelium from patients with by MSP were compared with the quantitative PS results. PS benign gastric pathology were analyzed in parallel. analysis revealed a high frequency of CDH1 methylation in

3767 ANTICANCER RESEARCH 33: 3765-3774 (2013)

Figure 1. Methylation analysis of human E-Cadherin gene (CDH1) in gastric carcinoma. A: Schematic representation of the CDH1 structure on chromosome 16q22.1, derived from the UCSC (University of California, Santa Cruz) genome browser (http://genome.ucsc.edu). The regions of CpG dinucleotides in exon 1 selected for methylation-specific polymerase chain reaction (MSP) and pyrosequencing (PS) analyses are indicated below the CpG map. B, C: CpG islands in the promoter region of CDH1 and the detailed nucleotide sequence (ENSG00000039068). The MSP and PS primer sets were designed using Methprimer (http://www.urogene.org/methprimer/index1.html).

77.8% (56/72) of the tumors, 70.8% (51/72) of tumor- between non-neoplastic mucosa according to the distance adjacent mucosa, 72.2% (52/72) of tumor-remote mucosa, from the tumor. This suggests that the methylation pattern whereas CDH1 was less frequently methylated in normal based on location in the stomach is somewhat random. The gastric mucosa (50.0%, 12/24; Table II). There were no differences in the frequencies of CDH1 promoter methylation significant differences in the frequency CDH1 of methylation in DGCs and tumor-remote mucosas versus normal gastric

3768 Lee et al: CDH1 Methylation in Gastric Cancer

Figure 2. Comparative examples of methylation analyses of the E- Cadherin gene (CDH1). A: Genomic DNA treated with Sss1 methylase as a positive methylated control shows high methylation levels by the pyrosequencing (PS) assay. B: Genomic DNA as a positive unmethylated control shows low methylation levels by the PS assay. The three sets of panels show representative methylation-specific polymerase chain reaction (MSP) electrophoresis products (middle panel) and typical pyrograms (upper and lower panels). Lane U, Unmethylated CDH1; lane M, methylated CDH1; each lane number correlates with a sample number from a patient. C: MSP and PS analyses of a diffuse gastric carcinoma Figure 3. Microphotograph of immunohistochemical stains for E- sample (T, lane 3). D: MSP and PS analyses of a normal gastric mucosa Cadherin. The microphotograph shows strong membranous expression sample adjacent to diffuse gastric cancer (DGC) (N1, lane 5). E: MSP of E-Cadherin in normal gastric mucosa (A), lack of expression in and PS analyses of a normal gastric mucosa sample remote from DGC diffuse gastric cancer (DGC) (B), and partial absence of expression in (N2, lane 2). F: MSP and PS analyses of a normal gastric mucosa sample tumor-adjacent mucosa (C) (immunohistochemistry, original from a patient with benign gastric pathology (N3, lanes 5). magnification ×200).

3769 ANTICANCER RESEARCH 33: 3765-3774 (2013)

Immunohistochemical assessment of CDH1 and correlation with promoter methylation using MSP and PS. Immunohisto-chemical analyses were conducted to investigate CDH1 protein expression in 72 tumors and paired adjacent mucosa tissues. Lack of expression of CDH1 protein was detectable in 90.3% (65/72) of the DGC samples with varying degrees of expression loss, and only seven cases out of the DGCs exhibited relatively strong membrane staining (Table IV, Figure 3). Nonetheless, neither methylation detected by MSP nor by PS showed statistically significant concordance with the lack of CDH1 protein. Out of the 72 tumor-adjacent mucosa samples, 56.9% (41/72) had variable-sized areas showing lack of CDH1 protein expression. Likewise, methylation detection neither by MSP nor by PS was statistically significantly related to CDH1 protein expression (p=0.371 and p=0.983, respectively).

Association between CDH1 promoter methylation and clinicopathological variables. The relationships between CDH1 promoter methylation and clinicopathological features of patients with DGC are shown in Table V. Except for greater tumor size (p=0.044), CDH1 promoter methylation as determined by MSP was not associated with any of the factors that were examined, including age, gender, Lauren classification, histological type, depth of invasion, metastasis, and TNM stages (p>0.05 each). However, CDH1 promoter methylation as determined by PS was closely-associated with lower depth of invasion (p=0.003), and TNM stages (p=0.024). CDH1 was more frequently methylated in larger tumors (≥2.5 cm) than in smaller tumors. CDH1 promoter methylation was more common in early gastric cancer (T1) than in advanced gastric cancer (T2+T3), and similarly in low TNM stages (stage I) than in high TNM stages (stages II and III). In this study, the median follow-up period after surgery was 92.4 months (range 3 to 108 months). Overall survival analysis was performed by log-rank analysis of Kaplan–Meier survival curves (Figure 4). The results of Figure 4. A: Survival curve for patients with gastric carcinoma, analyses of the variables that could be correlated with the according to methylation of the E-Cadherin gene (CDH1) as determined survival time are shown in Table VI. Although the patient by PS (p=0.343). B: Survival curve for patients with gastric carcinoma, group with CDH1 promoter methylation detected by PS had according to tumor stage (p<0.001). a longer median survival period than the patient group without CDH1 promoter methylation, the survival difference was not statistically significant (p=0.343). In mucosas were statistically significant (77.8% and 72.2% vs. univariate analysis, tumor size and TNM stage were 50.0%, p=0.010 and 0.046, respectively), whereas the statistically significant (Table VI). The group with differences in the frequencies of CDH1 promoter methylation smaller–sized tumors (<2.5 cm) survived longer than the between tumor-adjacent mucosas and normal gastric mucosa group with larger–sized tumors, and the survival benefit were marginally significant (70.8% vs. 50.0%, p=0.063). was marginally significant in multivariate analysis MSP and PS analyses of methylation of CDH1 produced (p=0.061, hazard ratio 1:5.109). Not surprisingly, only relatively similar results (Table III). The concordance of the TNM stage was the most powerful prognostic factor both results by two techniques, however, did not reach statistical in univariate and multivariate analyses (p<0.001 and significance (p>0.05). So they did not concur. p=0.039, respectively).

3770 Lee et al: CDH1 Methylation in Gastric Cancer

Table III. Comparison of E-Cadherin gene (CDH1) promoter methylation status as determined by methylation-specific polymerase chain reaction (MSP) and pyrosequencing (PS).

Group No. PS/MSP, n (%) Kappa value p-Value for Kappa value (+)/(+) (+)/(–) (–)/(+) (–)/(–)

T 72 40 (55.6%) 16 (22.2%) 13 (18.1%) 3 (4.2%) –0.09 0.43 N1 72 34 (47.2%) 17 (23.6%) 12 (16.7%) 9 (12.5%) 0.09 0.44 N2 72 35 (48.6%) 17 (23.6%) 14 (19.4%) 6 (8.3%) –0.03 0.83 N3 24 8 (33.3%) 4 (16.7%) 4 (16.7%) 8 (33.3%) 0.33 0.10

T, Diffuse gastric cancer; N1, tumor-adjacent normal mucosa; N2, tumor-remote normal mucosa; N3, normal mucosa with benign gastric pathology.

Discussion Table IV. Comparison of lack of E-Cadherin protein expression and E- Cadherin gene (CDH1) promoter methylation status by methylation- In the current study, the majority of sporadic DGCs exhibitied specific polymerase chain reaction (MSP) and pyrosequencing (PS). CDH1 promoter methylation either by MSP or by PS (73.6% E-Cadherin No. (%) MSP p-Value PS p-Value and 77.8%, respectively). The frequency of CDH1 methylation protein loss in our study was similar to that previous studies conducted on MUM M UM sporadic DGCs (10, 11). Tumor-adjacent or tumor-remote mucosa also demonstrated a comparable prevalence of CDH1 Present 65 (90.3%) 48 17 1.000 50 15 1.000 Absent 7 (9.7%) 5 2 6 1 methylation. In addition, 90.3% (65 out of 72 cases) had more than one focus of CDH1 methylation in matched non- neoplastic gastric mucosa. It is probable that CDH1 promoter methylation in non-neoplastic tissues from DGC patients other studies. Chan et al. showed that CDH1 methylation was occurs in a random pattern with a high incidence comparable to present in 35% (11/31) of gastric mucosa samples from patients that in DGC tissues. A large number of sporadic DGC samples with chronic gastritis, and it was closely associated with and matched non-neoplastic normal mucosa samples were Helicobacter pylori infection. They also demonstrated that collected from each individual as well as non-matched normal CDH1 methylation was present in none of eight (0%) normal mucosa samples from patients with benign gastric pathology mucosa samples, 12 out of 21 (57%) intestinal metaplastic in the present study. It is believed that the inclusion of matched samples, 15 out of 26 (58%) primary cancer samples, and 21 samples in this study led to the detection of a high prevalence out of 32 (65%) metastatic cancer samples (18). However, the of CDH1 methylation in tumor-related mucosa. authors did not divide the gastric cancer cases into an intestinal E-Cadherin (encoded by the CDH1 gene) is a homophilic type and a diffuse type according to Lauren classification. Blok cell adhesion molecule that is expressed on most epithelial cells. et al. also reported a lower expression of E-Cadherin protein in It is a transmembrane glycoprotein with five tandemly repeated intestinal metaplasia in patients with early gastric cancer (21). extracellular domains and a cytoplasmic domain that binds to a Reduction of E-Cadherin expression has been reported from cytosolic complex with α-, β-, and γ-catenins, which are in turn intestinal metaplasia to gastric adenocarcinoma throughout the links to actins (16, 17). E-Cadherin is important for establishing histological sequence of chronic gastritis – atrophic gastritis – cell polarity and maintaining normal tissue morphology and intestinal metaplasia – adenocarcinoma (22). Chan et al. cellular differentiation (16). Defective cell adhesion may postulated that although methylation of CDH1 was reversed in contribute to loss of contact inhibition of growth, which is an chronic gastritis after H. pylori eradication, methylation of important early step in the neoplastic process. Furthermore, loss CDH1 persisted in intestinal metaplasia even after H. pylori of cell adhesion may account for the ability of cancer cells to eradication, which may subsequently lead to further molecular cross normal tissue boundaries and metastasize (18). The changes in gastric carcinogenesis (23). A great majority of the importance of E-Cadherin in maintaining cell adhesion implies DGC cases in our study also showed lack of E-Cadherin that its dysfunction may play an important part in tumorigenesis expression (90.3%, 65/72). Even in non-neoplastic mucosa, (19). In gastric carcinogenesis, the critical role of E-Cadherin is variable degrees of E-Cadherin expression loss were observed, underlined by the observation that familial gastric cancer is which corresponded to the findings of a previous study (22). related to germline mutations of the CDH1 gene (6, 20). In the present study, the ability of two techniques, quantitative Researchers have previously demonstrated that the CDH1 PS and qualitative MSP, to determine the methylation status of promoter was consistently unmethylated in normal mucosa from the CDH1 promoter in DGCs was compared. There are a patients with DGC (8). However, this claim was challenged by number of techniques for assessing promoter methylation status.

3771 ANTICANCER RESEARCH 33: 3765-3774 (2013)

Table V. Clinicopathological features and promoter methylation status of E-Cadherin gene (CDH1) in patients with gastric carcinoma.

Characteristic No. MSP PS

MUMp-Value M UM p-Value

Age, years <55 33 22 11 0.219 25 8 0.704 ≥55 39 31 8 31 8 Gender M 37 29 8 0.345 29 8 0.900 F352411278 Location Lower third 49 34 15 0.269 37 12 0.560 Others 23 19 4 19 4 Tumor size <2.5 35 22 13 0.044 29 6 0.313 ≥2.5 37 31 6 27 10 Lauren classification Diffuse 40 29 11 0.811 33 7 0.281 Mixed 32 24 8 23 9 Differentiation Poor 51 37 14 0.750 37 14 0.125 Signet 21 16 5 19 2 Depth of invasion T1 45 31 14 0.241 40 5 0.003 T2+T3 27 22 5 16 11 Lymph node metastasis Absent 49 34 15 0.269 40 9 0.251 Present 23 19 4 16 7 Distant metastasis Absent 69 51 18 1.000 54 15 0.535 Present 3 2 1 2 1 TNM stage I 52 37 15 0.558 44 8 0.024 ≥II 11 16 4 12 8

MSP: Methylation-specific polymerase chain reaction; PS: pyrosequencing.

Table IV. Univariate and multivariate analysis for overall survival predictors in patients with diffuse gastric cancer.

Characteristic No. Survival, mean (weeks) Univariate p-Value Multivariate p-Value Hazard ratio

Age, years <55 33 91.2 0.812 0.407 1 ≥55 39 93.8 1.773 Gender M 37 91.2 0.677 0.870 1 F 35 91.3 0.899 Location Lower third 49 97.6 0.114 0.294 1 Others 23 75.7 2.052 Tumor size <2.5 35 107.0 0.002 0.061 1 ≥2.5 37 78.5 8.262 Lauren classification Diffuse 40 89.8 0.721 0.875 1 Mixed 32 95.7 1.118 Differentiation Poor 51 90.1 0.331 0.798 1 Signet 21 93.3 0.791 TNM stage Ⅰ 52 102.0 <0.001 0.039 1 ≥Ⅱ 11 62.5 5.109 Methylation by MSP Absent 19 92.1 0.937 0.294 1 Present 53 92.7 0.397 Methylation by PS Absent 16 84.8 0.343 0.912 1 Present 56 94.7 0.924

MSP: Methylation-specific polymerase chain reaction; PS: pyrosequencing.

Several techniques have been reported, and the most widely used positive signals at a high number of PCR cycles (24, 25). Recent technique is MSP analysis of DNA after bisulfite treatment (14). attempts to address these limitations have led to the development However, the popularity of MSP has fallen due to its well-known of an alternative high-throughput technique for methylation shortcomings, including its qualitative nature, the limited number analysis, which is known as PS (14, 26). PS is a sensitive real- of CpG sites targeted, lack of an internal control, and false- time sequencing-by-synthesis method based on the conversion

3772 Lee et al: CDH1 Methylation in Gastric Cancer of DNA by pyrophospate enzymatic treatment and stepwise Somatic mutations of CDH1 had been identified earlier in luminescence nucleotide incorporation for the quantitative sporadic DGCs rather than in hereditary DGCs (34, 35). measurement of CpG island methylation (27, 28). However, PS Somatic mutations of the E-Cadherin gene were identified in has several disadvantages. Firstly, the read length in a single 50% of the DGCs, in 14% of the mixed type, but they were reaction is currently limited to less than 100 bp and 20 CpG loci, absent from the intestinal type, whereas these alterations were and it cannot always be used to detect an entire CpG island. not seen in non-neoplastic gastric tissues (5). Although CDH1 Owing to difficulties in primer design, PS cannot always be used mutations are detected in approximately half of the sporadic to detect certain regions in a CpG island (25, 29). DGCs (5, 10, 36), they are rarely found in early DGCs (37). The rates of CDH1 methylation detected by PS were almost Conversely, methylation of CDH1 was observed in more than similar to those detected by MSP. Nonetheless, their 50% of early DGCs (37, 38). Considering the aforementioned implications in association with clinicopathological variables nature of CDH1 methylation, such as its significant reduction or survival rates were somewhat different. Although the rates after H. pylori eradication, frequent occurrence in early DGCs, of CDH1 methylation were similar for both MSP and PS, the and high incidence in non-neoplastic mucosa of DGCs as concordance between MSP and PS was not statistically shown in the current study, CDH1 methylation may be a significant. CDH1 methylation detected by MSP revealed a leading contributor to DGC development (39). statistically significant association with only tumor size. In In conclusion, the current study showed that the majority of comparison, CDH1 methylation detected by PS was closely DGC cases demonstrated concurrent CDH1 promoter associated with depth of invasion and TNM stage. In terms of methylation both in tumor tissues and in non-neoplastic gastric overall survival rates, patients with CDH1 methylation detected mucosa. This suggests that CDH1 promoter methylation may by PS had a greater survival rate than those with CDH1 precede carcinogenesis in wide areas of gastric tissues and methylation detected by MSP, although neither result was provide fertile 'soil' for carcinogenesis of sporadic DGCs. statistically significant (p=0.937 and p=0.343, respectively). Since MSP is a simple qualitative technique for screening and Acknowledgements detecting hypermethylated genes in tumors, it can be suggested This study was supported by a grant from the National R&D Program with great caution that there is a possibility that the methylation for Cancer Control, Ministry of Health & Welfare, Republic of Korea detected by PS, a more comprehensive and quantitative (0720570). method, reflects the biological behavior of the tumor. Germline CDH1 mutations were found in families with References inherited DGC, while no germline mutations were detected in 1 Ferlay J, Shin HR, Bray F, Forman D, Mathers C and Parkin DM: families with intestinal gastric cancer (6, 20). Promoter Estimates of worldwide burden of cancer in 2008: GLOBOCAN methylation of CDH1 played a role of a second genetic hit in 2008. Int J Cancer 127: 2893-2917, 2010. silencing CDH1 expression in half of the cases with abrogation 2 Khushalani N: Cancer of the esophagus and stomach. Mayo Clin of CDH1 expression in hereditary DGC, while the remaining Proc 83: 712-722, 2008. half harbored somatic CDH1 mutations in the unmethylated 3 Lauwers GY, Carneiro F, Graham DY, Curado MP, Franceschi S, In allele of the CDH1 promoter (8). Barber et al. showed that Tatematsu M and Hattori T: Gastric carcinoma. : WHO Classification of Tumours of the Digestive System, Bosman FT, individuals with germline mutations had somatic mutations in Carneiro F, Hruban RH, Theise ND (eds.). Lyon: International 37.5% (6/16) and promoter hypermethylation in 25% (4/16), as Agency for Research on Cancer, pp. 48-58, 2010. a second-hit mechanism. They also showed that there were 4 Oliveira C, Sousa S, Pinheiro H, Karam R, Bordeira-Carrico R, individuals with a germline mutation who did not appear to Senz J, Kaurah P, Carvalho J, Pereira R, Gusmao L, Wen X, exhibit any of the second-hit mechanisms (30). In another series Cipriano MA, Yokota J, Carneiro F, Huntsman D and Seruca R: of hereditary DGC, Oliveira et al. demonstrated that half of the Quantification of epigenetic and genetic second hits in CDH1 CDH1 second-hits were epigenetic modifications and a slightly during hereditary diffuse gastric cancer syndrome progression. Gastroenterology 136: 2137-2148, 2009. lower percentage of CDH1 second-hits were loss of 5 Becker KF, Atkinson MJ, Reich U, Becker I, Nekarda H, Siewert heterogeneity, and there were no alterations in 25% of lesions JR and Hofler H: E-Cadherin gene mutations provide clues to (4). Other mechanisms of E-Cadherin inactivation include diffuse type gastric carcinomas. Cancer Res 54: 3845-3852, 1994. intragenic deletions, transcriptional repressions or post- 6 Guilford P, Hopkins J, Harraway J, McLeod M, McLeod N, translational modifications, such as ectoderm shedding (30). Harawira P, Taite H, Scoular R, Miller A and Reeve AE: E- Alternatively, mutation of a single CDH1 allele is sufficient for Cadherin germline mutations in familial gastric cancer. Nature 392: tumor initiation, especially if it is acting in a dominant-negative 402-405, 1998. 7 Oliveira C, Bordin MC, Grehan N, Huntsman D, Suriano G, fashion (31). Other indirect mechanisms of silencing may also Machado JC, Kiviluoto T, Aaltonen L, Jackson CE, Seruca R and play a role, such as the miR-200 family of microRNAs, which Caldas C: Screening E-Cadherin in gastric cancer families reveals has been shown to indirectly regulate the expression of CDH1 germline mutations only in hereditary diffuse gastric cancer through the transcriptional repressor ZEB1 and ZEB2 (32, 33). kindred. Hum Mutat 19: 510-517, 2002.

3773 ANTICANCER RESEARCH 33: 3765-3774 (2013)

8 Grady WM, Willis J, Guilford PJ, Dunbier AK, Toro TT, Lynch H, 24 An Q, Liu Y, Gao Y, Huang J, Fong X, Li L, Zhang D and Cheng Wiesner G, Ferguson K, Eng C, Park JG, Kim SJ and Markowitz S: Detection of p16 hypermethylation in circulating plasma DNA S: Methylation of the CDH1 promoter as the second genetic hit in of non-small cell lung cancer patients. Cancer Lett 188: 109-114, hereditary diffuse gastric cancer. Nat Genet 26: 16-17, 2000. 2002. 9 Oliveira C, de Bruin J, Nabais S, Ligtenberg M, Moutinho C, 25 Shaw RJ, Liloglou T, Rogers SN, Brown JS, Vaughan ED, Lowe Nagengast FM, Seruca R, van Krieken H and Carneiro F: D, Field JK and Risk JM: Promoter methylation of P16, RARβ, E- Intragenic deletion of CDH1 as the inactivating mechanism of the Cadherin, cyclin A1 and cytoglobin in oral cancer: Quantitative wild-type allele in an HDGC tumour. Oncogene 23: 2236-2240, evaluation using pyrosequencing. Br J Cancer 94: 561-568, 2006. 2004. 26 Tost J, El abdalaoui H and Gut IG: Serial pyrosequencing for 10 Liu YC, Shen CY, Wu HS, Hsieh TY, Chan DC, Chen CJ, Yu JC, quantitative DNA methylation analysis. Biotechniques 40: 721-722, Yu CP, Harn HJ, Chen PJ, Hsieh CB, Chen TW and Hsu HM: 724, 726, 2006. Mechanisms inactivating the gene for E-Cadherin in sporadic 27 Laird PW: The power and the promise of DNA methylation gastric carcinomas. World J Gastroenterol 12: 2168-2173, 2006. markers. Nat Rev Cancer 3: 253-266, 2003. 11 Machado JC, Oliveira C, Carvalho R, Soares P, Berx G, Caldas C, 28 Uhlmann K, Brinckmann A, Toliat MR, Ritter H and Nurnberg P: Seruca R, Carneiro F and Sobrinho-Simoes M: E-Cadherin gene Evaluation of a potential epigenetic biomarker by quantitative (CDH1) promoter methylation as the second hit in sporadic diffuse methyl-single nucleotide polymorphism analysis. Electrophoresis gastric carcinoma. Oncogene 20: 1525-1528, 2001. 23: 4072-4079, 2002. 12 Stomach. In: AJCC Cancer Staging Manual. Edge SB, Byrd DR, 29 Colella S, Shen L, Baggerly KA, Issa JP and Krahe R: Sensitive Compton CC, Fritz AG Greene FL and Trotti A (eds.). Springer and quantitative universal Pyrosequencing methylation analysis of Press, pp. 117-126, 2009. CpG sites. Biotechniques 35: 146-150, 2003. 13 Moskaluk CA and Kern SE: Microdissection and polymerase chain 30 Barber M, Murrell A, Ito Y, Maia AT, Hyland S, Oliveira C, Save reaction amplification of genomic DNA from histological tissue V, Carneiro F, Paterson AL, Grehan N, Dwerryhouse S, Lao-Sirieix sections. Am J Pathol 150: 1547-1552, 1997. P, Caldas C and Fitzgerald RC: Mechanisms and sequelae of E- 14 Herman JG, Graff JR, Myohanen S, Nelkin BD and Baylin SB: Cadherin silencing in hereditary diffuse gastric cancer. J Pathol Methylation-specific PCR: a novel PCR assay for methylation 216: 295-306, 2008. status of CpG islands. Proc Natl Acad Sci USA 93: 9821-9826, 31 Dong HM, Liu G, Hou YF, Wu J, Lu JS, Luo JM, Shen ZZ and 1996. Shao ZM: Dominant-negative E-Cadherin inhibits the invasiveness 15 Li Y, Chen CQ, He YL, Cai SR, Yang DJ, He WL, Xu JB and Zan of inflammatory breast cancer cells in vitro. J Cancer Res Clin WH: Abnormal expression of E-Cadherin in tumor cells is Oncol 133: 83-92, 2007. associated with poor prognosis of gastric carcinoma. J Surg Oncol 32 Gregory PA, Bert AG, Paterson EL, Barry SC, Tsykin A, Farshid 106: 304-10, 2012. G, Vadas MA, Khew-Goodall Y and Goodall GJ: The miR-200 16 Grunwald GB: The structural and functional analysis of Cadherin family and miR-205 regulate epithelial to mesenchymal transition calcium-dependent cell adhesion molecules. Curr Opin Cell Biol by targeting ZEB1 and SIP1. Nat Cell Biol 10: 593-601, 2008. 5: 797-805, 1993. 33 Park SM, Gaur AB, Lengyel E and Peter ME: The miR-200 family 17 Takeichi M: Cadherins: A molecular family important in selective determines the epithelial phenotype of cancer cells by targeting the cell–cell adhesion. Annu Rev Biochem 59: 237-252, 1990. E-Cadherin repressors ZEB1 and ZEB2. Genes Dev 22: 894-907, 18 Chan AO, Lam SK, Wong BC, Wong WM, Yuen MF, Yeung YH, 2008. Hui WM, Rashid A and Kwong YL: Promoter methylation of E- 34 Berx G, Becker KF, Hofler H and van Roy F: Mutations of the Cadherin gene in gastric mucosa associated with Helicobacter human E-Cadherin (CDH1) gene. Hum Mutat 12: 226-237, 1998. pylori infection and in gastric cancer. Gut 52: 502-506, 2003. 35 Tamura G, Yin J, Wang S, Fleisher AS, Zou T, Abraham JM, Kong 19 Wijnhoven BP, Dinjens WN and Pignatelli M: E-Cadherin–catenin D, Smolinski KN, Wilson KT, James SP, Silverberg SG, Nishizuka cell–cell adhesion complex and human cancer. Br J Surg 87: 992- S, Terashima M, Motoyama T and Meltzer SJ: E-Cadherin gene 1005, 2000. promoter hypermethylation in primary human gastric carcinomas. 20 Gayther SA, Gorringe KL, Ramus SJ, Huntsman D, Roviello F, J Natl Cancer Inst 92: 569-573, 2000. Grehan N, Machado JC, Pinto E, Seruca R, Halling K, MacLeod P, 36 Muta H, Noguchi M, Kanai Y, Ochiai A, Nawata H and Hirohashi Powell SM, Jackson CE, Ponder BA and Caldas C: Identification S: E-Cadherin gene mutations in signet ring cell carcinoma of the of germ-line E-Cadherin mutations in gastric cancer families of stomach. Jpn J Cancer Res 87: 843-848, 1996. European origin. Cancer Res 58: 4086-4089, 1998. 37 Tamura G: Alterations of tumor suppressor and tumor-related genes 21 Blok P, Craanen ME, Dekker W and Tytgat GN: Loss of E- in the development and progression of gastric cancer. World J Cadherin expression in early gastric cancer. Histopathology 34: Gastroenterol 12: 192-198, 2006. 410-415, 1999. 38 Ushijima T and Sasako M: Focus on gastric cancer. Cancer cell 5: 22 Chan AO, Wong BC, Lan HY, Loke SL, Chan WK, Hui WM, Yuen 121-125, 2004. YH, Ng I, Hou L, Wong WM, Yuen MF, Luk JM and Lam SK: 39 Yamamoto E, Suzuki H, Takamaru H, Yamamoto H, Toyota M and Deregulation of E-Cadherin–catenin complex in precancerous Shinomura Y: Role of DNA methylation in the development of lesions of gastric adenocarcinoma. J Gastroenterol Hepatol 18: diffuse-type gastric cancer. Digestion 83: 241-249, 2011. 534-539, 2003. 23 Chan AO, Huang C, Hui WM, Cho CH, Yuen MF, Lam SK, Rashid A and Wong BC: Stability of E-Cadherin methylation status in Received July 12, 2013 gastric mucosa associated with histology changes. Aliment Revised July 23, 2013 Pharmacol Ther 24: 831-836, 2006. Accepted July 24, 2013

3774