Purpose: The objective of this study was to analyze the hypermethylation of tumor-related gene promoters for an association with therapy response and clinicopathologic features of neoadjuvant-treated gastric cancer patients. Furthermore, we analyzed the relationship of promoter hypermethylation with microsatellite instability and loss of heterozygosity (LOH) of the tumors.

Experimental Design: Pretherapeutic biopsies of 61 patients, subsequently treated with cisplatin and 5-fluorouracil, were studied. Methylation analysis of six gene promoters was done using MethyLight technology. Microsatellite analysis was mainly done in previous studies.

Results: The methylation frequencies for the analyzed genes were MGMT, 44%; LOX, 53%; p16, 46%, E-cadherin, 30%; 14-3-3σ, 69%; and HPP1, 82%. Concordant methylation of more than three genes was found in 46% of the tumors and was inversely correlated with the LOH rate (P = 9 × 10−5) and associated with female gender (P = 0.049), nonintestinal type tumors (P = 0.04), and a nonproximal tumor location (P = 0.003). No statistically significant association between the methylation of a single gene or the concordant methylation of multiple genes was found with response or survival. However, patients with none or only one methylated gene showed a trend for an increase in survival (5-year survival rate, 83% versus 35%; P = 0.067).

Conclusion: The highly significant inverse correlation of promoter methylation and LOH rate reflects major alternative molecular pathways in gastric carcinogenesis. Methylation was not statistically significantly associated with the response to cisplatin/5-fluorouracil–based therapy. However, a concordant methylation of more than three genes defines subgroups of gastric cancer with distinct biological and genetic characteristics.

Epigenetic alterations of DNA have been shown to be of high significance in the tumorigenesis of different tumor types (1, 2). Methylation of cytosine residues in the CpG-rich regions, the “CpG-islands,” of gene promoters leads to transcriptional repression of tumor suppressor genes, DNA repair genes, and genes involved in metastasis and invasion. Gene silencing by this promoter hypermethylation is observed in a considerable number of gastric carcinomas and has been detected in both early stages and premalignant lesions. This points to a critical role for epigenetic inactivation mechanisms in the carcinogenic process in the stomach (36). The concordant hypermethylation of multiple genes, termed the CpG island methylator phenotype (CIMP), has been described in various tumor types, including both colorectal and gastric carcinomas (710). The exact nature of this methylator phenotype, which may reflect a specific type of epigenetic instability, is poorly understood and controversially discussed (11). Several studies have shown that CIMP-positive tumors show a specific genetic profile and are associated with particular clinicopathologic variables (710). A prognostic relevance for CIMP after standard therapy or a 5-fluorouracil (5-FU)–based chemotherapy has been reported for different tumor entities, again including gastric carcinoma (1216).

Gastric carcinoma is characterized by a high mortality rate that is mainly due to the initial diagnosis being made at late/advanced stages. A neoadjuvant chemotherapy, based on 5-FU and cisplatin, is frequently used for the treatment of advanced gastric carcinomas. In a randomized trial, the perioperative treatment resulted in a significantly improved survival compared with surgery alone (17). However, only 20% to 30% of patients respond to this therapy, with the majority undergoing several months of toxic and costly therapy without a survival benefit (18). Thus, there is a pressing need for molecular markers that can be used to predict the individual response to therapy.

The goal of our study was to investigate the occurrence of the hypermethylation of tumor-related genes and to determine if they may be predictors of response and survival of advanced gastric cancer patients treated with a neoadjuvant, cisplatin/5-FU–based therapy. We have investigated if hypermethylation of single tumor and/or therapy-related genes or concordant hypermethylation of multiple genes is associated with patient outcome or other clinical features.

Finally, we analyzed the methylation status for an association between microsatellite instability (MSI) and loss of heterozygosity (LOH), which for the majority of tumors had previously been studied (19, 20). In these previous studies, we had shown a significant association of a high LOH rate with the response to a cisplatin/5-FU–based neoadjuvant chemotherapy (19, 20). Thus, the characterization of epigenetic changes in relation to these other genetic alterations might not only provide some deeper insight into their role for the carcinogenic process in the stomach but might especially be important in the light of a potential application of new epigenetic-based chemotherapeutic strategies for gastric cancer patients.

Patients. Pretherapeutic biopsies of 61 patients with locally advanced gastric cancer (tumor category cT3 or cT4) were analyzed. All patients were treated at the Department of Surgery between 1993 and 2003 with a uniform, combined preoperative chemotherapy containing cisplatin, leucovorin, and 5-FU (PLF-regimen). Procedures for staging, response evaluation, and surgery were according to institutional standard operating procedures, which have been described in detail (2123) and which did not change during the time of the recruitment of the patients. Selection criteria for the study were the availability of pretherapeutic biopsies and the suitability of material for the isolation of DNA from tumor areas containing at least 50% tumor cells by manual microdissection. In addition, each selected patient must have received 50% of the projected dose of the chemotherapeutic regimen.

Of the 61 patients, 48 (79%) had a complete tumor resection (R0), 10 (16%) had a resection with tumor involved resection margins (R1) after chemotherapy, and 3 (5%) were not resected due to tumor progression. All 61 patients were evaluated clinically for response and all were included in this study (none were lost to follow-up). The median follow-up of the patients was 58.7 months (range, 26.1-106.7 months). The median survival of the patients was 40.6 months (range, 2.4-106.7 months). The median event-free survival (time to progression or time to recurrence) was 27.2 months (range, 3.3-106.7 months). The patient characteristics are included in Table 1.

Table 1.

Association of highly methylated tumors and clinicopathologic or molecular features

Total no. patients
MET-H ≥4/6 genes*
MET-M/L <4/6 genes*
P
n (%)n (%)n (%)
Total 61 (100) 28 (46) 33 (54)  
Gender     
    Female 12 (20) 9 (32) 3 (9) 0.049 
    Male 49 (80) 19 (68) 30 (92)  
Lauren type     
    Intestinal 29 (48) 9 (32) 20 (61) 0.040 
    Nonintestinal 32 (52) 19 (68) 13 (39)  
Grading     
    G1 + G2 10 (16) 4 (14) 6 (18) 0.741 
    G3 51 (84) 24 (86) 27 (82)  
Location     
    Proximal third 46 (75) 16 (57) 30 (91) 0.003 
    Middle third 10 (16) 7 (25) 3 (9)  
    Distal third 2 (3) 2 (7) 0 (0)  
    Total 3 (5) 3 (11) 0 (0)  
MSI     
    MSI-H 6 (10) 1 (3) 5 (15) 0.143 
    MSI-L 12 (20) 8 (29) 4 (12)  
    MSS 43 (70) 19 (68) 24 (73)  
FAL 55 (100) 27 (49) 28 (51)  
    FAL-H 11 (20) 1 (4) 10 (36) 9 × 10−5 
    FAL-M 22 (40) 8 (30) 14 (50)  
    FAL-L 22 (40) 18 (66) 4 (14)  
Mean age (range), y 55.13 (29.8-72.4) 55.10 (40.3-71.5) 55.15 (29.8-72.4) 0.64 
Total no. patients
MET-H ≥4/6 genes*
MET-M/L <4/6 genes*
P
n (%)n (%)n (%)
Total 61 (100) 28 (46) 33 (54)  
Gender     
    Female 12 (20) 9 (32) 3 (9) 0.049 
    Male 49 (80) 19 (68) 30 (92)  
Lauren type     
    Intestinal 29 (48) 9 (32) 20 (61) 0.040 
    Nonintestinal 32 (52) 19 (68) 13 (39)  
Grading     
    G1 + G2 10 (16) 4 (14) 6 (18) 0.741 
    G3 51 (84) 24 (86) 27 (82)  
Location     
    Proximal third 46 (75) 16 (57) 30 (91) 0.003 
    Middle third 10 (16) 7 (25) 3 (9)  
    Distal third 2 (3) 2 (7) 0 (0)  
    Total 3 (5) 3 (11) 0 (0)  
MSI     
    MSI-H 6 (10) 1 (3) 5 (15) 0.143 
    MSI-L 12 (20) 8 (29) 4 (12)  
    MSS 43 (70) 19 (68) 24 (73)  
FAL 55 (100) 27 (49) 28 (51)  
    FAL-H 11 (20) 1 (4) 10 (36) 9 × 10−5 
    FAL-M 22 (40) 8 (30) 14 (50)  
    FAL-L 22 (40) 18 (66) 4 (14)  
Mean age (range), y 55.13 (29.8-72.4) 55.10 (40.3-71.5) 55.15 (29.8-72.4) 0.64 
*

Number of methylated genes / analyzed genes per tumor.

Fisher's exact test (two sided) or Mann-Whitney test.

MSI-H tumors (n = 6) were not evaluated for FAL.

Preoperative chemotherapy. Preoperative therapy consisted of two cycles of combination chemotherapy, each of 49-day duration. On day 1, cisplatin at a dose of 50 mg/m2 body surface area was given by i.v. infusion over a period of 1 h. Thereafter, patients received leucovorin (500 mg/m2 body surface area) over a period of 2 h, followed by 5-FU (2 g/m2 body surface area) over a period of 24 h. Treatment with cisplatin was repeated on days 15 and 29. Infusion of leucovorin and 5-FU was repeated on days 8, 15, 22, 29, and 36 (21, 24). The inclusion and exclusion criteria for chemotherapy were as previously published (19, 24). No adjuvant chemotherapy or radiochemotherapy was applied. Surgical resection of the tumor according to local standards was scheduled 3 to 4 weeks after completion of the chemotherapy (21, 24).

Response evaluation. Response evaluation was done histopathologically and clinically as previously described (22, 24, 25). For histopathologic response evaluation, the macroscopically identifiable tumor bed of resected tumors was histologically completely examined. Those patients who had <10% residual tumor in the tumor bed area were classified as responders. All other patients, including patients with clinical progression, were classified as nonresponders (24, 25).

Clinical response evaluation was done by measuring the size of the primary tumor by computed tomography scan, endoluminal ultrasound, and endoscopy and was defined as previously published (24, 25).

For the analysis of the association between methylation status and response, only patients with a congruent evaluation by both methods (n = 54) were used as previously described (25). This included 15 (28%) responders and 39 (72%) nonresponders.

Patient survival was significantly associated with both clinical and histopathologic responses (P < 0.001). Analysis of the methylation status in association with survival was done for all 61 patients.

The study protocol was approved by the local ethics committee and informed consent was obtained according to institutional regulations.

Gene selection. The following six genes involved in different molecular pathways were analyzed: p16, 14-3-3σ (cell cycle), O6-methylguanine-DNA methyltransferase (MGMT; DNA repair), E-cadherin, lysyl oxidase (LOX; cell adhesion), and HPP1. A prerequisite for the selection of these genes was that promotor hypermethylation has previously been established as a mechanism for the transcriptional inactivation of the genes in gastric cancer. Furthermore, the reported methylation frequencies were required to be preferably between 25% and 75% because genes with a very low or a very high methylation frequency are not useful in distinguishing between groups (2631). Finally, functional aspects were considered. Thus, the genes involved in cell cycle control (i.e., p16 and 14-3-3σ) may also play a direct role for chemotherapeutic efficacy, and a relation to the effect of cisplatin has been reported for the E-cadherin and MGMT genes, making them additionally attractive candidates to test for a potential role in therapy response (32, 33).

DNA extraction from archival tissues. DNA from gastric cancer specimens was extracted from formalin-fixed, paraffin-embedded tissue. Defined areas having >50% of tumor cells were manually microdissected and DNA was isolated by proteinase K digestion and phenol-chloroform extraction according to standard procedures (34).

Sodium bisulfite treatment of DNA. Treatment of the DNA with sodium bisulfite was done essentially as described (35). Genomic DNA (up to 1 μg) in a volume of 50 μL Tris-EDTA buffer was denatured at 95°C for 10 min, followed by incubation with 5.5 μL of 3 mol/L NaOH for 20 min at 40°C and 3 min at 95°C. Samples were put on ice and treated with 500 μL of freshly prepared 2.8 mol/L sodium bisulfite/2.5 mmol/L hydrochinon solution for 3 h at 55°C. DNA was purified using the Wizzard DNA clean-up kit (Promega Corp.) according to the manufacturer's instructions. DNA was eluted with 100 μL of Tris-EDTA buffer, desulfonated with 11 μL of 3 mol/L NaOH for 20 min at 40°C, precipitated with ethanol, and resuspended in 75 μL of distilled water.

MethyLight analysis. After sodium bisulfite conversion, genomic DNA was analyzed by the MethyLight technique as described (36) using the ABI PRISM 7700 Sequence Detection System instrument and software (Applied Biosystems, Inc.). Published primer and probe systems were used or were designed specifically for fully methylated bisulfite-converted DNA using the primer express software (Applied Biosystems). Primer and probe sequences, as well as reaction conditions, are summarized in Table 2. To normalize for the input of DNA, a region of the MYOD1 gene lacking bisulfite-sensitive CpG islands was used (36). SssI-treated human lymphocyte DNA was used as a fully methylated positive control. PCR products were cloned and sequenced, and DNA of positive clones was used as internal positive control. Human sperm DNA or DNA from lymphocytes was used as the unmethylated negative control.

Table 2.

Primer and probe sequences of the MethyLight systems

Gene primer and probeSequence 5′-3′Size of PCR product (bp)Reference
MYOD1    
    Forward CCAACTCCAAATCCCCTCTCTAT 107 (36) 
    Reverse GGTTTTTTTAGGGAGTAAGTTTGTTAGG   
    Probe TCCCTTCCTATTCCTAAATCCAACCTAAATACCTCC   
14-3-3σ    
    Forward GAAGGTTAAGTTGGTAGAGTAGGTCGAAC 117 (47) 
    Reverse AACTACTAAAAACAAATTTCGCTCTTCG   
    Probe CTCGCCCTTCTCCACGACGCC   
E-Cadherin    
    Forward AATTTTAGGTTAGAGGGTTATCGCGT 70 (48) 
    Reverse TCCCCAAAACGAAACTAACGAC   
    Probe CGCCCACCCGACCTCGCAT   
HPP1    
    Forward GTTATCGTCGTCGTTTTTGTTGTC 87 (49) 
    Reverse GACTTCCGAAAAACACAAAATCG   
    Probe CCGAACAACGAACTACTAAACATCCCGCG   
LOX    
    Forward GAATAAATAGTTGAGGGGCGGTC 122 (5) 
    Reverse GCGACAATCCCGAAAAACG   
    Probe CACGTTTACAAAATTACACAAACCGTTCTAACCC   
MGMT    
    Forward CGAATATACTAAAACAACCCGCG 122 (50) 
    Reverse GTATTTTTTCGGGAGCGAGGC   
    Probe AATCCTCGCGATACGCACCGTTTACG   
p16    
    Forward GGGGAGAGTAGGTAGCGGGC 84 (47) 
    Reverse AACCAATCAACCGAAAACTCCATA   
    Probe TACTCCCCGCCGCCGACTCCAT   
Gene primer and probeSequence 5′-3′Size of PCR product (bp)Reference
MYOD1    
    Forward CCAACTCCAAATCCCCTCTCTAT 107 (36) 
    Reverse GGTTTTTTTAGGGAGTAAGTTTGTTAGG   
    Probe TCCCTTCCTATTCCTAAATCCAACCTAAATACCTCC   
14-3-3σ    
    Forward GAAGGTTAAGTTGGTAGAGTAGGTCGAAC 117 (47) 
    Reverse AACTACTAAAAACAAATTTCGCTCTTCG   
    Probe CTCGCCCTTCTCCACGACGCC   
E-Cadherin    
    Forward AATTTTAGGTTAGAGGGTTATCGCGT 70 (48) 
    Reverse TCCCCAAAACGAAACTAACGAC   
    Probe CGCCCACCCGACCTCGCAT   
HPP1    
    Forward GTTATCGTCGTCGTTTTTGTTGTC 87 (49) 
    Reverse GACTTCCGAAAAACACAAAATCG   
    Probe CCGAACAACGAACTACTAAACATCCCGCG   
LOX    
    Forward GAATAAATAGTTGAGGGGCGGTC 122 (5) 
    Reverse GCGACAATCCCGAAAAACG   
    Probe CACGTTTACAAAATTACACAAACCGTTCTAACCC   
MGMT    
    Forward CGAATATACTAAAACAACCCGCG 122 (50) 
    Reverse GTATTTTTTCGGGAGCGAGGC   
    Probe AATCCTCGCGATACGCACCGTTTACG   
p16    
    Forward GGGGAGAGTAGGTAGCGGGC 84 (47) 
    Reverse AACCAATCAACCGAAAACTCCATA   
    Probe TACTCCCCGCCGCCGACTCCAT   

NOTE: Primer and probe concentrations were for MyoD1, 300 nmol/L primer/50 nmol/L probe; 14-3-3σ, 600 nmol/L primer/200 nmol/L probe; E-cadherin, 600 nmol/L primer/200 nmol/L probe; HPP1, 600 nmol/L primer/100 nmol/L probe; LOX, 600 mol/L primer/100 nmol/L probe; MGMT, 600 nmol/L primer/100 nmol/L probe; and p16, 600 nmol/L primer/75 nmol/L probe.

PCR was done in duplicate in a final volume of 30 μL with the TaqMan Universal PCR Master Mix without uracil-N-glycosylase (Applied Biosystems) using 5 μL of bisulfite-treated DNA as template. Cycling conditions were 95°C for 10 min, followed by 45 cycles at 95°C for 15 s and 60°C for 1 min. Samples with no detectable methylation signal using 45 cycles but with successful amplification of MYOD1 were considered to be nonmethylated. Results with SD >30% were repeated. Relative promoter methylation levels were determined by the standard curve method, defining SssI-treated human lymphocytic DNA as 100% methylation ratio.

The frequency of methylation of the genes was determined by choosing a specific cutoff value for the respective genes. As the determination of cutoff values is not standardized and the detection of methylation may depend on the amplification efficiency of the reaction of the individual genes, and may for some cases be influenced by patient specific characteristics, we defined methylation relative to the methylation signal detected in nontumorous gastric epithelial tissues. Thus, methylation was defined as positive if the percent methylation ratio was above the median percent methylation ratio of the respective gene in an analysis of 10 nontumorous gastric epithelial tissues from randomly selected patients. The cutoff values were as follows: 0.01 for LOX, p16, MGMT; 2.23 for HPP1; 3.45 for E-cadherin; and 23.49 for 14-3-3σ.

Tumors were grouped according to the number of methylated genes per six analyzed genes found per tumor. There is no standardized definition of highly methylated tumors that most likely represent the CIMP phenotype. We applied the following classification systems: we categorized the tumors into a high methylation group (MET-H) with tumors showing ≥4 methylated genes, a medium methylation group (MET-M) with tumors showing 2 to 3 methylated genes, and a low methylation group (MET-L) with tumors showing 0 to 1 methylated genes. We analyzed (a) the MET-H versus the MET-M/L group and (b) the MET-H/M group versus the MET-L group for an association with the clinicopathologic features, therapy response, and molecular variables.

Microsatellite analysis. Of the 61 tumors analyzed for promoter hypermethylation in the present study, 34 had previously been characterized by microsatellite analysis (19, 20). To allow for a comparison of promoter hypermethylation with MSI and LOH in the whole study group, microsatellite analysis was done for the remaining 27 cases with the same 11 microsatellite markers and protocol, using fluorescence labeled primers and separation and detection by an automated sequencing system (ABI 377, Perkin-Elmer) as published (19, 20).

Instability in ≥40% of the analyzed markers was defined as MSI-H, instability in <40% as MSI-L, and no instability as microsatellite stable (MSS).

LOH was defined if the allele peak ratio was <60%, representing a signal reduction of one allele of at least 40% (19, 20). Fractional allelic loss (FAL) was defined as the ratio of the number of chromosomal sites showing LOH divided by the number of informative chromosomal sites for each case. The tumors were categorized as low (0-0.25), medium (>0.25-0.5), and high FAL (>0.5) as described (19). Tumors with MSI-H were excluded from evaluation for FAL.

Statistical analysis. Fisher's exact test (two sided) was used to compare for the relative frequencies between different groups. Differences between survival were evaluated using the log-rank test. All survival data were calculated from the start of chemotherapy to the date of death or most recent follow-up. The Mann-Whitney test was used to compare metric variables in two groups. P < 0.05 was considered to be statistically significant and P < 0.1 was considered as trend. Statistical analysis was done using the SPSS software (SPSS, Inc.).

Methylation frequencies. Among the 61 gastric carinomas, promoter methylation of the six genes analyzed was detected at the following frequencies: 18 (30%) for E-cadherin, 27 (44%) for MGMT, 28 (46%) for p16, 32 (53%) for LOX, 42 (69%) for 14-3-3σ, and 50 (82%) for HPP1.

Partition of the tumors into a high methylation group (MET-H; ≥4/6 methylated genes) and a medium/low methylation group (MET-M/L; <4/6 methylated genes) revealed that 28 of 61 (46%) tumors belonged to the MET-H group and 33 of the 61 (54%) belonged to the MET-M/L group.

Categorization of the tumors into a high/medium methylation group (MET-H/M; ≥2/6 methylated genes) and a low methylation group (MET-L; 0-1/6 methylated genes) revealed that, overall, 52 of 61 (85%) tumors belonged to the MET-H/M group and 9 of 61 (15%) tumors belonged to the MET-L group.

Concordant methylation and correlation with MSI and LOH. Among the 61 tumors included in this study, 6 (10%) tumors were MSI-H, 12 (20%) were MSI-L, and 43 (70%) were MSS. The degree of MSI was not correlated with the methylation of multiple genes, comparing the MET-H group versus the MET-M/L group or comparing the MET-H/M group versus the MET-L group (Table 1 and data not shown).

Among the 55 tumors that were evaluated for LOH, 11 (20%) showed LOH at >50% of the informative chromosomal sites and were classified as FAL-H tumors; 22 (40%) showed LOH between 25% and 50% of the informative chromosomal sites and were classified as FAL-M tumors; and 22 (40%) showed LOH at <25% of the chromosomal sites and were classified as FAL-L. Analyzing the FAL rate for an association with the methylation rate of the tumors revealed a highly statistically significant association of MET-H tumors and low FAL levels and vice versa (P = 9 × 10−5). The data are included in Table 1 and shown in Fig. 1.

Fig. 1.

Comparison of the FAL value in relation to the methylation rate. The proportion of highly methylated tumors with at least 4/6 methylated genes and the proportion of medium/low methylated tumors with <4/6 methylated genes analyzed, in relation to the FAL value of the tumors.

Fig. 1.

Comparison of the FAL value in relation to the methylation rate. The proportion of highly methylated tumors with at least 4/6 methylated genes and the proportion of medium/low methylated tumors with <4/6 methylated genes analyzed, in relation to the FAL value of the tumors.

Close modal

Methylation and correlation with clinicopathologic variables. Correlation of concordant methylation of multiple genes with clinicopathologic variables, such as age, gender, Lauren classification, tumor grading, and tumor location, revealed a statistically significant association between MET-H tumors and female gender (P = 0.049), nonintestinal type tumors (P = 0.040), and nonproximal tumor location (P = 0.003). These data are summarized in Table 1.

Considering methlyation of the single genes, the following statistically significant correlations were found: methylation of the LOX gene was associated with nonintestinal type tumors (P = 0.011), worse tumor differentiation (P = 0.037), and nonproximal tumor location (P = 0.010); methylation of the 14-3-3σ gene was associated with female gender (P = 0.012); and methylation of the HPP1 gene was associated with age (P = 0.03).

Methylation and correlation with response and survival. The analysis of the number of methylated genes per tumor in relation to therapy response revealed a higher methylation frequency among the nonresponding patients (Fig. 2). Comparing the patients with tumors of the MET-H group versus the MET-M/L group showed that 19 of 39 (49%) of the nonresponding and 4 of 15 (27%) of the responding patients were MET-H tumors (P = 0.22). Comparing the patients with tumors of the MET-L group versus the MET-H/M group showed that 4 of 39 (10%) among the nonresponding and 4 of 15 (27%) of the responding patients were MET-L (P = 0.20). These differences, however, were not statistically significant (Table 3).

Fig. 2.

Comparison of the number of methylated genes in relation to response to neoadjuvant chemotherapy.

Fig. 2.

Comparison of the number of methylated genes in relation to response to neoadjuvant chemotherapy.

Close modal
Table 3.

Methylation status and association with therapy response and survival

Methylation statusNo. methylated tumors
Overall survival
5-y survival
Event-free survival
Responder, n = 15 (%)Nonresponder, n = 39 (%)P*Median (mo)P(% of patients)P*Median (mo)P
MET-H: ≥4/6 4 (27) 19 (45) 0.2 Not reached 0.933 35 0.54 31.50 0.75 
MET-M/L: <4/6 11 (73) 20 (51)  87.87  48  22.90  
MET-H/M: ≥2/6 11 (73) 35 (90) 0.2 44.00 0.191 35 0.067 22.20 0.097 
MET-L: <2/6 4 (27) 4 (10)  87.87  83  Not reached  
Methylation statusNo. methylated tumors
Overall survival
5-y survival
Event-free survival
Responder, n = 15 (%)Nonresponder, n = 39 (%)P*Median (mo)P(% of patients)P*Median (mo)P
MET-H: ≥4/6 4 (27) 19 (45) 0.2 Not reached 0.933 35 0.54 31.50 0.75 
MET-M/L: <4/6 11 (73) 20 (51)  87.87  48  22.90  
MET-H/M: ≥2/6 11 (73) 35 (90) 0.2 44.00 0.191 35 0.067 22.20 0.097 
MET-L: <2/6 4 (27) 4 (10)  87.87  83  Not reached  
*

Fisher's exact test (two sided).

Log-rank test.

Number of methylated genes / analyzed genes per tumor.

A trend toward longer survival was observed for patients with MET-L tumors. The median overall and the median event-free survival time in the MET-L group were 87.87 months or were not reached, respectively, compared with 44.0 and 22.2 months, respectively, in the MET-H/M group (P = 0.191 and P = 0.097, log-rank test). Using the 5-year survival rate as an end point, a statistical trend for an increase in survival was also found, as 83% of the patients with MET-L tumors were still alive compared with 35% with MET-H/M tumors (P = 0.067). All data are summarized in Table 3.

Evaluation of methylation of the single genes revealed a trend for an association with therapy response for the LOX gene, with 4 of 15 (27%) responders and 23 of 39 (59%) nonresponders showing methylation in their tumors (P = 0.067). No associations with survival were found (Table 4).

Table 4.

Methylation of tumor-related genes and association with therapy response and survival

GeneNo. methylated tumors / analyzed tumors (n = 54)
Overall survival
Event-free survival
Responder (n = 15)
Nonresponder (n = 39)
P*PP
n (%)n (%)
LOX 4 (27) 23 (59) 0.066 >0.1 >0.1 
14-3-3σ 7 (47) 29 (74) >0.1 >0.1 >0.1 
p16 5 (33) 18 (46) >0.1 >0.1 >0.1 
MGMT 6 (49) 18 (59) >0.1 >0.1 >0.1 
E-Cadherin 5 (33) 13 (33) >0.1 >0.1 >0.1 
HPP1 11 (73) 33 (85) >0.1 >0.1 >0.1 
GeneNo. methylated tumors / analyzed tumors (n = 54)
Overall survival
Event-free survival
Responder (n = 15)
Nonresponder (n = 39)
P*PP
n (%)n (%)
LOX 4 (27) 23 (59) 0.066 >0.1 >0.1 
14-3-3σ 7 (47) 29 (74) >0.1 >0.1 >0.1 
p16 5 (33) 18 (46) >0.1 >0.1 >0.1 
MGMT 6 (49) 18 (59) >0.1 >0.1 >0.1 
E-Cadherin 5 (33) 13 (33) >0.1 >0.1 >0.1 
HPP1 11 (73) 33 (85) >0.1 >0.1 >0.1 
*

Fisher's exact test (two sided).

Log-rank test.

In this study, we have analyzed promoter hypermethylation of six gastric cancer related–genes in 61 pretherapeutic biopsies of patients, who subsequently received a neoadjuvant treatment based on cisplatin and 5-FU. As recent studies emphasized the existence of a so-called CIMP as a distinct molecular subgroup of gastrointestinal tumors (7, 10, 13, 16, 27, 37), we were particularly interested if the concordant methylation of multiple genes was associated with response and/or survival of the patients.

For colorectal cancer, an optimal marker panel for the characterization of CIMP has recently been determined (7). Because there is no standardized criteria for the definition of CIMP in gastric cancer, we analyzed the concordant methylation of multiple genes using two different classification systems.

Overall, we found that 46% of the tumors had concurrent hypermethylation of at least four of the six analyzed genes. This is similar to previous studies reporting a concurrent hypermethylation of multiple genes in 31% to 41% of gastric carcinomas (3, 13, 37). In our study, these tumors showed distinct molecular and clinicopathologic features. The most interesting of which was the highly significant inverse correlation between the methylation rate and the rate of LOH (FAL value; P = 9 × 10−5). This suggests that epigenetic instability, reflected by the concordant promoter hypermethylation of multiple genes, and chromosomal instability, reflected by a high FAL value or high LOH rate, represent pathways of genetic alterations driving distinct carcinogenic pathways in gastric cancer. Similar results have recently been reported for colorectal carcinomas (38). In gastric cancer, analyzing the inactivation mechanisms of the E-cadherin gene, a mutual exclusivity between LOH and hypermethylation has been reported, which is essentially in line with our results (39). Altogether, this strongly supports the existence of two independent mechanisms of genetic and epigenetic instability in gastrointestinal cancer.

Considering the methylation rate in relation to response to therapy, we observed tumors with concurrent methylation of multiple genes more frequently among nonresponding patients, although the differences did not reach statistical significance in either of the two classification systems tested (P = 0.2 for both).

In previous studies, we have shown that there is a statistically significant relationship between response to a cisplatin/5-FU–based chemotherapy and the FAL rate because patients showing high FAL values in their tumors were more frequently found among responding patients (19, 20). This finding holds true also for the tumors included in the present study (P = 2 × 10−4; data not shown). Taken together with the highly inverse correlation observed between the FAL and methylation rates, this raises the important question if an epigenetic-based chemotherapy may represent a promising alternative treatment for a subgroup of gastric cancer patients. Various agents targeting epigenetic alterations are currently under clinical investigation and promising results have been reported for inhibitors of DNA methylation and histone deacetylases, mainly in myeloid neoplasms and some solid tumors (4043). According to the results of our study, one could hypothesize that, for the group of patients with a high FAL value and low methylation, the cisplatin/5-FU treatment might be appropriate, but for patients showing higher methylation and lower FAL rates in their tumors, a combination therapy with epigenetic drugs might represent an alternative treatment.

With respect to survival, a trend for an increase in the event-free survival time and in the 5-year survival rate was found for the group of patients showing none or one methylated gene in their tumors (P = 0.097 and P = 0.067, respectively). At first, this contradicts other studies reporting an association between the concordant methylation of multiple genes and better survival after surgery for gastric cancer (13, 16). However, it has to be stressed that the survival rates for treatment by surgery alone are not comparable with those for neoadjuvant treatment and surgery used in our study. Furthermore, in the study by An et al. (13), methylation of the MLH1 gene was included in the analysis. This favors the inclusion of microsatellite unstable tumors in the group of highly methylated tumors. Because patients with microsatellite unstable tumors have been shown to have a better prognosis in gastric and colorectal cancers (44, 45), this may lead to confounding results about the prognostic significance of methylation of multiple genes. Furthermore, as the number of patients included in our study is relatively small, the prognostic significance of promoter hypermethylation and survival has to be considered with care and an extended analysis in a larger number of cases is needed to draw firm conclusions.

In our study, we did not observe a correlation between concordant methylation of multiple genes and MSI, which has previously been described for gastric cancer by others (29). This controversial result may also be related to the fact that we have not included in our study the MLH1 gene, which is frequently inactivated by promoter hypermethylation in sporadic gastric and colorectal carcinomas showing high MSI (3, 7, 29).

With respect to the association between concurrent methylation of multiple genes and clinicopathologic variables, we found a significant association of the high methylation group and nonintestinal type tumors (P = 0.04). This result is consistent with recent findings that indicate different pathways of gastric carcinogenesis. The so-called “methylator” phenotype is preferentially associated with diffuse type tumors, whereas the more mutational and genomic unstable phenotype is preferentially associated with intestinal type gastric tumors (27, 46). Similar to results described for colorectal carcinomas (7), we found an association of highly methylated tumors with female gender (P = 0.049) and tumor location (P = 0.003), but the reasons are poorly understood.

Considering methylation of the single genes, the observed methylation frequencies ranged from 30% to 82% and were essentially in line with previously published data for gastric cancer (2630). Interestingly, for the LOX gene, significant associations were found between promoter methylation and nonintestinal type tumors (P = 0.011), worse tumor differentiation (P = 0.037), and nonproximal tumor location (P = 0.012), and a trend was observed for an association with nonresponse (P = 0.067). A correlation of methylation of the LOX gene with the histopathologic type may be related to one of the functions of LOX as an extracellular enzyme that initiates covalent cross-linking of collagens and elastins. Loss of this function might facilitate spreading of tumor cells, which is characteristic of diffuse type gastric cancer. LOX has been described as a tumor suppressor gene and inactivation by methylation and LOH has been shown in gastric cancer (30). Our finding of a preferential methylation of the LOX gene in diffuse type gastric cancer is in line with a previous report (5) and supports a specific role of LOX as a diffuse type gastric cancer tumor suppressor gene.

In summary, we have shown a relative high frequency of concordant methylation of multiple genes in pretherapeutic gastric cancer biopsies. Although methylation of the analyzed genes was not statistically significantly associated with a response to a cisplatin/5-FU–based therapy in this study, we showed that tumors with more than three methylated genes define subgroups with distinct biological and genetic characteristics. In particular, the highly significant inverse correlation between the methylation and LOH rate indicates major alternative molecular pathways in gastric carcinogenesis. This may be of clinical relevance for a potential application of an epigenetic-based chemotherapy in gastric cancer.

Grant support: Deutsche Forschungsgemeinschaft grant KE 498/1-3.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Mike Atkinson for critically reading the manuscript.

1
Esteller M. Epigenetics provides a new generation of oncogenes and tumour-suppressor genes.
Br J Cancer
2006
;
94
:
179
–83.
2
Issa JP. CpG island methylator phenotype in cancer.
Nat Rev Cancer
2004
;
4
:
988
–93.
3
Toyota M, Ahuja N, Suzuki H, et al. Aberrant methylation in gastric cancer associated with the CpG island methylator phenotype.
Cancer Res
1999
;
59
:
5438
–42.
4
Kang GH, Shim YH, Jung HY, Kim WH, Ro JY, Rhyu MG. CpG island methylation in premalignant stages of gastric carcinoma.
Cancer Res
2001
;
61
:
2847
–51.
5
Kaneda A, Kaminishi M, Yanagihara K, Sugimura T, Ushijima T. Identification of silencing of nine genes in human gastric cancers.
Cancer Res
2002
;
62
:
6645
–50.
6
Sato F, Meltzer SJ. CpG island hypermethylation in progression of esophageal and gastric cancer.
Cancer
2006
;
106
:
483
–93.
7
Weisenberger DJ, Siegmund KD, Campan M, et al. CpG island methylator phenotype underlies sporadic microsatellite instability and is tightly associated with BRAF mutation in colorectal cancer.
Nat Genet
2006
;
38
:
787
–93.
8
Toyota M, Ohe-Toyota M, Ahuja N, Issa JP. Distinct genetic profiles in colorectal tumors with or without the CpG island methylator phenotype.
Proc Natl Acad Sci U S A
2000
;
97
:
710
–5.
9
Kang GH, Lee S, Kim WH, et al. Epstein-Barr virus-positive gastric carcinoma demonstrates frequent aberrant methylation of multiple genes and constitutes CpG island methylator phenotype-positive gastric carcinoma.
Am J Pathol
2002
;
160
:
787
–94.
10
Samowitz WS, Albertsen H, Herrick J, et al. Evaluation of a large, population-based sample supports a CpG island methylator phenotype in colon cancer.
Gastroenterology
2005
;
129
:
837
–45.
11
Anacleto C, Leopoldino AM, Rossi B, et al. Colorectal cancer “methylator phenotype”: fact or artifact?
Neoplasia
2005
;
7
:
331
–5.
12
Van Rijnsoever M, Elsaleh H, Joseph D, McCaul K, Iacopetta B. CpG island methylator phenotype is an independent predictor of survival benefit from 5-fluorouracil in stage III colorectal cancer.
Clin Cancer Res
2003
;
9
:
2898
–903.
13
An C, Choi IS, Yao JC, et al. Prognostic significance of CpG island methylator phenotype and microsatellite instability in gastric carcinoma.
Clin Cancer Res
2005
;
11
:
656
–63.
14
Brock MV, Gou M, Akiyama Y, et al. Prognostic importance of promoter hypermethylation of multiple genes in esophageal adenocarcinoma.
Clin Cancer Res
2003
;
9
:
2912
–9.
15
Roman-Gomez J, Jimenez-Velasco A, Agirre X, Prosper F, Heiniger A, Torres A. Lack of CpG island methylator phenotype defines a clinical subtype of T-cell acute lymphoblastic leukemia associated with good prognosis.
J Clin Oncol
2005
;
23
:
7043
–9.
16
Kusano M, Toyota M, Suzuki H, et al. Genetic, epigenetic, and clinicopathologic features of gastric carcinomas with the CpG island methylator phenotype and an association with Epstein-Barr virus.
Cancer
2006
;
106
:
1467
–79.
17
Cunningham D, Allum WH, Stenning SP, et al. Perioperative chemotherapy versus surgery alone for resectable gastroesophageal cancer.
N Engl J Med
2006
;
355
:
11
–20.
18
Lordick F, Stein HJ, Peschel C, Siewert JR. Neoadjuvant therapy for oesophagogastric cancer.
Br J Surg
2004
;
91
:
540
–51.
19
Ott K, Vogelsang H, Mueller J, et al. Chromosomal instability rather than p53 mutation is associated with response to neoadjuvant cisplatin-based chemotherapy in gastric carcinoma.
Clin Cancer Res
2003
;
9
:
2307
–15.
20
Grundei T, Vogelsang H, Ott K, et al. Loss of heterozygosity and microsatellite instability as predictive markers for neoadjuvant treatment in gastric carcinoma.
Clin Cancer Res
2000
;
6
:
4782
–8.
21
Ott K, Fink U, Becker K, et al. Prediction of response to preoperative chemotherapy in gastric carcinoma by metabolic imaging: results of a prospective trial.
J Clin Oncol
2003
;
21
:
4604
–10.
22
Becker K, Mueller JD, Schulmacher C, et al. Histomorphology and grading of regression in gastric carcinoma treated with neoadjuvant chemotherapy.
Cancer
2003
;
98
:
1521
–30.
23
Fink U, Schuhmacher C, Stein HJ, et al. Preoperative chemotherapy for stage III-IV gastric carcinoma: feasibility, response and outcome after complete resection.
Br J Surg
1995
;
82
:
1248
–52.
24
Ott K, Sendler A, Becker K, et al. Neoadjuvant chemotherapy with cisplatin, 5-FU, and leucovorin (PLF) in locally advanced gastric cancer: a prospective phase II study.
Gastric Cancer
2003
;
6
:
159
–67.
25
Napieralski R, Ott K, Kremer M, et al. Combined GADD45A and thymidine phosphorylase expression levels predict response and survival of neoadjuvant-treated gastric cancer patients.
Clin Cancer Res
2005
;
11
:
3025
–31.
26
Geddert H, Kiel S, Iskender E, et al. Correlation of hMLH1 and HPP1 hypermethylation in gastric, but not in esophageal and cardiac adenocarcinoma.
Int J Cancer
2004
;
110
:
208
–11.
27
Oue N, Oshimo Y, Nakayama H, et al. DNA methylation of multiple genes in gastric carcinoma: association with histological type and CpG island methylator phenotype.
Cancer Sci
2003
;
94
:
901
–5.
28
Suzuki H, Itoh F, Toyota M, Kikuchi T, Kakiuchi H, Imai K. Inactivation of the 14-3-3σ gene is associated with 5′ CpG island hypermethylation in human cancers.
Cancer Res
2000
;
60
:
4353
–7.
29
Carvalho B, Pinto M, Cirnes L, et al. Concurrent hypermethylation of gene promoters is associated with a MSI-H phenotype and diploidy in gastric carcinomas.
Eur J Cancer
2003
;
39
:
1222
–7.
30
Kaneda A, Wakazono K, Tsukamoto T, et al. Lysyl oxidase is a tumor suppressor gene inactivated by methylation and loss of heterozygosity in human gastric cancers.
Cancer Res
2004
;
64
:
6410
–5.
31
Sabbioni S, Miotto E, Veronese A, et al. Multigene methylation analysis of gastrointestinal tumors: TPEF emerges as a frequent tumor-specific aberrantly methylated marker that can be detected in peripheral blood.
Mol Diagn
2003
;
7
:
201
–7.
32
Fricke E, Hermannstadter C, Keller G, et al. Effect of wild-type and mutant E-cadherin on cell proliferation and responsiveness to the chemotherapeutic agents cisplatin, etoposide, and 5-fluorouracil.
Oncology
2004
;
66
:
150
–9.
33
D'Atri S, Graziani G, Lacal PM, et al. Attenuation of O(6)-methylguanine-DNA methyltransferase activity and mRNA levels by cisplatin and temozolomide in Jurkat cells.
J Pharmacol Exp Ther
2000
;
294
:
664
–71.
34
Keller G, Vogelsang H, Becker I, et al. Diffuse type gastric and lobular breast carcinoma in a familial gastric cancer patient with an E-cadherin germline mutation.
Am J Pathol
1999
;
155
:
337
–42.
35
Herman JG, Graff JR, Myohanen S, Nelkin BD, Baylin SB. Methylation-specific PCR: a novel PCR assay for methylation status of CpG islands.
Proc Natl Acad Sci U S A
1996
;
93
:
9821
–6.
36
Eads CA, Danenberg KD, Kawakami K, et al. MethyLight: a high-throughput assay to measure DNA methylation.
Nucleic Acids Res
2000
;
28
:
E32
.
37
Oue N, Mitani Y, Motoshita J, et al. Accumulation of DNA methylation is associated with tumor stage in gastric cancer.
Cancer
2006
;
106
:
1250
–9.
38
Goel A, Nagasaka T, Arnold CN, et al. The CpG island methylator phenotype and chromosomal instability are inversely correlated in sporadic colorectal cancer.
Gastroenterology
2006
;
132
:
127
–38.
39
Liu YC, Shen CY, Wu HS, et al. Mechanisms inactivating the gene for E-cadherin in sporadic gastric carcinomas.
World J Gastroenterol
2006
;
12
:
2168
–73.
40
Gore SD, Baylin S, Sugar E, et al. Combined DNA methyltransferase and histone deacetylase inhibition in the treatment of myeloid neoplasms.
Cancer Res
2006
;
66
:
6361
–9.
41
Garcia-Manero G, Kantarjian HM, Sanchez-Gonzalez B, et al. Phase I/II study of the combination of 5-aza-2′-deoxycytidine with valproic acid in patients with leukemia.
Blood
2006
;
108
:
3271
–9.
42
Yang H, Hoshino K, Sanchez-Gonzalez B, Kantarjian H, Garcia-Manero G. Antileukemia activity of the combination of 5-aza-2′-deoxycytidine with valproic acid.
Leuk Res
2005
;
29
:
739
–48.
43
Kelly WK, O'Connor OA, Krug LM, et al. Phase I study of an oral histone deacetylase inhibitor, suberoylanilide hydroxamic acid, in patients with advanced cancer.
J Clin Oncol
2005
;
23
:
3923
–31.
44
dos Santos NR, Seruca R, Constancia M, Seixas M, Sobrinho-Simoes M. Microsatellite instability at multiple loci in gastric carcinoma: clinicopathologic implications and prognosis.
Gastroenterology
1996
;
110
:
38
–44.
45
Ward RL, Cheong K, Ku SL, Meagher A, O'Connor T, Hawkins NJ. Adverse prognostic effect of methylation in colorectal cancer is reversed by microsatellite instability.
J Clin Oncol
2003
;
21
:
3729
–36.
46
Etoh T, Kanai Y, Ushijima S, et al. Increased DNA methyltransferase 1 (DNMT1) protein expression correlates significantly with poorer tumor differentiation and frequent DNA hypermethylation of multiple CpG islands in gastric cancers.
Am J Pathol
2004
;
164
:
689
–99.
47
Lehmann U, Langer F, Feist H, Glockner S, Hasemeier B, Kreipe H. Quantitative assessment of promoter hypermethylation during breast cancer development.
Am J Pathol
2002
;
160
:
605
–12.
48
Toyooka KO, Toyooka S, Maitra A, et al. Establishment and validation of real-time polymerase chain reaction method for CDH1 promoter methylation.
Am J Pathol
2002
;
161
:
629
–34.
49
Sato F, Shibata D, Harpaz N, et al. Aberrant methylation of the HPP1 gene in ulcerative colitis-associated colorectal carcinoma.
Cancer Res
2002
;
62
:
6820
–2.
50
Brabender J, Usadel H, Metzger R, et al. Quantitative O(6)-methylguanine DNA methyltransferase methylation analysis in curatively resected non-small cell lung cancer: associations with clinical outcome.
Clin Cancer Res
2003
;
9
:
223
–7.