The E-cadherin Gene Polymorphism -160C→A and Cancer Risk: A HuGE
Review and Meta-Analysis of 26 Case-Control Studies

¹ Laboratory of Population and Quantitative Genetics, Institute of Genetics and Biostatistics, School
University, Shanghai, People’s Republic of China.
² School of Biosciences, University of Birmingham, Edgbaston, Birmingham, United Kingdom.

Correspondence to Professor Z. W. Luo, School of Biosciences, University of Birmingham, Edgbaston, Birmingham B15 2TT, United Kingdom (e-mail: z.luo@bham.ac.uk or zwluo@fudan.edu.cn ).

Received for publication April 9, 2007; accepted for publication August 13, 2007.

A single nucleotide polymorphism, –160C→A, has been identified in the promoter region of the E-cadherin gene and has been shown to alter its transcriptional activity. To assess susceptibility of –160A allele carriers to seven types of cancers, the authors conducted a comprehensive meta-analysis, up to November 2006, of 26 case-control studies comprising 7,042 cases and 7,011 controls. Pooled odds ratios and 95% confidence intervals were calculated by using the random-effects model. Publication bias, subgroup, and sensitivity analyses were also performed, which showed that –160A allele carriers, compared with noncarriers, had about a 17–19% increased risk of several invasive/metastatic tumors. Analyses of various types of cancers revealed that, in Europeans, the –160AA homozygote was associated with an increased risk of urothelial cancer, carriers of –160A were at increased risk of lung and prostate cancers, and carriers of –160A with gastric cancer were found to suffer a significantly increased risk, whereas their Asian counterparts seemed to be tolerant. No evidence was found that the –160A allele predisposed its carriers to breast, colorectal, or esophageal cancers. These findings indicate that –160A of the E-cadherin gene is emerging as a low-penetrance tumor susceptibility allele for the development of gastric, lung, prostate, and urothelial cancers.

Keywords: E-cadherin; epidemiology; meta-analysis; mutation; neoplasms
Abbreviations: CI, confidence interval; OR, odds ratio.

The E-cadherin gene, located on chromosome 16 (16q22.1), encodes a transmembrane glycoprotein that mediates intercellular adhesion as well as cell signaling in conjunction with cytoplasmic catenin proteins. E-cadherin as a calcium-dependent intercellular adhesion molecule is highly expressed in normal epithelial cells and well-differentiated cancer cells, but its expression is largely reduced in undifferentiated cancers. The E-cadherin/catenin complex is important for cellular polarity and maintenance of normal tissue morphology and cellular differentiation (1). As a tumor suppressor, defective E-cadherin expression governs transition to an invasive phenotype in human epithelial cancers (2, 3). In addition, disruption of E-cadherin with rare mutations may also be involved in carcinogenesis through a modified Wnt signaling pathway (4).

Recently, a C/A single nucleotide polymorphism has been identified in the promoter region of the E-cadherin gene –160 base pairs upstream of the transcriptional start site (5). It has been shown that the A allele decreased transcriptional efficiency by about 10–68 percent compared with the wild-type C allele (5–7). It was also observed that the C allele showed much higher binding affinity to transcriptional factor than the mutant allele, indicating that the –160C/A variant may alter transcriptional activity of the E-cadherin gene and be responsible for decreased E-cadherin expression and increased susceptibility to epithelial cancers.

The frequency of the –160A allele varies in different geographic areas and ethnic populations. In Europe, the variant allele frequency ranges from 43.4 percent in Italy to 23.3 percent in the United Kingdom and, in Asia, from a high allele frequency in China (61.0 percent) to a lower frequency in Korea (14.3 percent); there is also marked variation in the frequency of the –160AA homozygote from 18.9 percent in Italy to 3.4 percent in the United Kingdom and from 44.0 percent in China to 0 percent in Japan (this information is summarized in Web table 1, the first of two supplementary tables; each is referred to as "Web table" in the text and is posted on the website of the Human Genome Epidemiology Network as well as on the Journal's website . (This paper also includes three supplementary figures, each referred to as "Web figure" in the text and posted on the same two websites.) Moreover, genetic effects of the polymorphism have been shown to vary from one type of cancer to the other. Recent epidemiologic studies revealed an increased risk of prostate cancer (8), a protective role for gastric cancer (9), but no significant association with breast cancer (6) or colorectal cancer (10). Otherwise, –160C/A has been associated with an increased susceptibility to sporadic diffuse-type gastric cancer (11), but no association with risk of gastric cancer was found (12). Thus, these observations raised quite a controversial question regarding the significance of –160C/A in cancer pathogenesis. Obviously, statistical power of an individual study could be very limited for efficient assessment of the –160A allele. Integration of these data sets may provide improved statistical power to detect the significance.

This meta-analysis was conducted on seven types of cancers—breast, colorectal, esophageal, gastric, prostate, urothelial, and lung—on the basis of available data sets on the E-cadherin –160C/A polymorphism and cancer risk.

Breast cancer is the second leading cause of cancer deaths among women. Established risk factors include age at menarche and first full-term pregnancy, which play a role via hormonal mechanisms (13). In addition to variants in the high-penetrance genes BRCA1 and BRCA2, frequent variants of the E-cadherin gene have also been found to raise breast cancer risk (14).

Colorectal cancer is the third most common cause of cancer deaths. A recent study indicates that about 35 percent of all colorectal cancer can be ascribed to inherited genetic susceptibility (15). Expression of E-cadherin has been found to be significantly reduced in colorectal cancer but apparently not as a consequence of allele loss or somatic mutation (16).

Esophageal cancer is one of the most common malignancies. Risk factors include various environmental triggers such as cigarette smoking, alcohol drinking, and malnutrition (17). Genetic susceptibility is also suggested as an important factor in esophageal cancer risk, such as a polymorphism in the promoter region of E-cadherin (18).

Gastric cancer is another leading cause of cancer deaths. It is well established that environmental factors such as dietary habits and Helicobacter pylori infection are associated with gastric cancer risk (19). Most genetic factors such as single nucleotide polymorphisms in the E-cadherin promoter may be critical in gastric carcinogenesis (5, 20).

Prostate cancer is one of the most common cancers among men in developed countries. Established risk factors include age, ethnicity, and family history. In addition, a few low-penetrance susceptibility genes including E-cadherin with a higher population frequency may be relevant to prostate cancer risk in combination with environmental factors (21).

Urothelial cancer is the fourth most frequent cancer in men in the United States; exogenous carcinogens are widely recognized as the major cause. Increasing evidence suggests that genetic susceptibility should also be considered as a significant risk factor (22, 23). Furthermore, abnormal expression of E-cadherin might be related to –160C/A in urothelial cancer (5).

Lung cancer has remained the leading cause of cancer mortality in the Western world. Although environmental exposures such as cigarette smoking account for most cases, genetic variants as well as gene-environment interactions and epigenetic processes are likely to play a significant role in determining disease susceptibility (24).

To date, there have been 26 known case-control studies in the literature on associating the E-cadherin –160C/A polymorphism with different types of cancer in different ethnic populations. The aim of this meta-analysis was to assess the evidence supporting E-cadherin –160A as a tumor susceptibility allele by integrating all these case-control studies.

We conducted a systematic review and meta-analysis in accordance with the guidelines provided by the Human Genome Epidemiology Network (25).

Search strategy
First, we conducted a systematic literature search using the databases MEDLINE (US National Library of Medicine, Bethesda, Maryland) and PubMed (National Center for Biotechnology, National Library of Medicine) before November 2006 with keywords "–160C/A," "rs16260," "polymorphism of the E-cadherin gene," or "–160C/A" in combination with "cancer" or "neoplasm" or "carcinoma." The full texts of the candidate articles were examined carefully to determine whether they contained sufficient information on the –160C/A polymorphism and cancer risk. Furthermore, reference lists were also reviewed to trace further relevant studies.

Inclusion criteria
The constituent studies for the present meta-analysis were case-control studies with sufficient published data for estimating an odds ratio and corresponding 95 percent confidence interval.

Data extraction
The following information from each study was extracted by two investigators independently: 1) publication data, first author, year of publication, and country of origin; 2) cancer types; 3) study design (hospital based, population based, or nested) and genotyping method; and 4) number of cases and controls with the CC, CA, and AA genotypes. This information is summarized in Web tables 1 and 2.

Meta-analysis
As the first step in the meta-analysis, we tested significance of deviation of genotype distribution at the polymorphic site from that expected from Hardy-Weinberg equilibrium in the control sample for each of the selected case-control data sets. The random-effects model was used to calculate the pooled odds ratio estimates for genotype AA, genotype CA, and A-allele carriers (AA + CA) against the CC genotype by using review manager version 4.2 software (RevMan; The Cochrane Collaboration, Oxford, England). To evaluate whether results of the data sets were homogeneous, we used Cochran's Q test (26). We also calculated the quantity I2 that represents the percentage of total variation across studies that is a result of heterogeneity rather than chance (27). As a guide, I2 values of less than 25 percent may be considered "low," values of about 50 percent may be considered "moderate," and values of more than 75 percent may be considered "high" (27). A value of 0 percent indicates no observed heterogeneity, and larger values show increasing heterogeneity. In the absence of heterogeneity, the random-effects and fixed-effects models will provide similar results. When heterogeneity is significant, both models may be biased (28).

Publication bias was evaluated by using the Begg and Mazumdar adjusted rank correlation test (29) and the Egger regression asymmetry test (30). To evaluate the stability of the results, we performed a one-way sensitivity analysis. The scope of this analysis reflects the influence of an individual data set by estimating the average odds ratio in the absence of each data set (31). Statistical tests performed in the present analysis were considered significant whenever the corresponding null-hypothesis probability was p < 0.05.

A total of 31 published studies relevant to the current study were carefully examined, and six of these candidate studies were removed from consideration, the one that reported discovery of the –160C/A polymorphism of the E-cadherin gene (5) and another five that were not case-control studies (32–36). Web figure 1 illustrates the search strategy for constituent studies in the present meta-analysis. Finally, we used 26 independent case-control studies (6–12, 37–54), including our unpublished data collected on non-small-cell lung cancer. The present study involved a total of 7,042 cases and 7,011 controls.

Overall data analysis
For all 31 data sets extracted from the 26 case-control studies, –160A allelic frequency in the pooled cases was significantly higher than that in the corresponding controls (Z = 4.4733, p = 0.0000077). The pooled odds ratio estimate was 1.19 (95 percent confidence interval (CI): 1.06, 1.33) (table 1), suggesting that carriers of the –160A allele would have a significantly higher risk of being predisposed to developing cancer (p = 0.002, Web figure 2). However, genotype distribution in controls from four studies (39, 41, 50, 53) and one group of controls from the study by Jonsson et al. (48) significantly deviated from the expected Hardy-Weinberg equilibrium (p < 0.05). After these data sets were excluded, the odds ratio estimate for carriers of the –160A allele was 1.22 (95 percent CI: 1.10, 1.35) and p = 0.0001, indicating that the risk estimate for the –160A carriers to be predisposed to cancer remained almost unchanged before and after excluding the data sets.

click to view

Table 1  Estimates of odds ratios and the corresponding 95% confidence intervals for AA, CA genotypes and A allele carriers versus the CC genotype for seven types of cancers analyzed by a random-effects model, up to November 2006

 
Both Cochran's Q test (Q = 66.06 with 30 df, p = 0.0002) and the estimate of I2 (54.6 percent) revealed a significant heterogeneity among the constituent studies. To assess the cause of heterogeneity, we carried out subgroup analyses for each type of cancer and ethnic group. Table 1 summarizes the odds ratio estimates for three genotypes at the –160C/A site and the corresponding 95 percent confidence intervals for seven types of cancer, and for ethnic groups whenever there were at least two data sets for either of the ethnic groups. The p values for the Begg and Mazumdar test and the Egger test were 0.27 and 0.13, respectively, for the –160A allele carriers (table 2), both suggesting a negligible publication bias. Table 2 also illustrates the significance tests for genetic heterogeneity in addition to the publication bias analysis. Furthermore, we performed a one-way sensitivity analysis by removing one data set at a time, and stability of the odds ratio estimates was confirmed (figure 1 and Web figure 3). The analyses for each type of cancer are detailed in the following seven sections of the text.

click to view

Table 2  Heterogeneity test for studies of each genotype in different cancer types (up to November 2006) with Cochran's Q test and the quantity I2, and publication bias test with the Begg and Mazumdar adjusted rank correlation test and the Egger regression asymmetry test

 

click to view

Figure 1: One-way sensitivity analysis of the pooled odds ratios and 95% confidence intervals for carriers of the –160A allele versus the CC genotype, omitting each data set in the meta-analysis (up to November 2006). The pooled odds ratios were calculated by means of a random-effects model. The numbers on the x-axis refer to the data sets extracted from referred studies. For instance, 6a indicates the first data set from reference 6. *, the authors' unpublished data set; total, no data set omitted.

 
Breast cancer
Three data sets from two case-control studies (6, 37) were for breast cancer, including 1,043 patients and 817 controls (Web table 1). The heterogeneity test for the pooled data sets was not significant (Q = 0.38 with 2 df, p = 0.83; I2 = 0 percent), suggesting robustness of the meta-analysis for this cancer type (table 2). In addition, genotypic distributions among all controls were in agreement with the Hardy-Weinberg equilibrium, and we found no significant difference between different ethnic populations. The pooled odds ratio estimate for the –160A carriers was 1.12 (95 percent CI: 0.93, 1.35) (Web figure 2). There was no significant difference in either allelic frequency or genotypic frequency between the controls and the breast cancer cases, suggesting that the –160A allele conferred no detectable risk of breast cancer.

Colorectal cancer
Three studies investigated susceptibility of –160A carriers to colorectal cancer, comprising 646 patients and 465 normal controls (7, 10, 38). When these data sets were analyzed together, there was no significant heterogeneity (Q = 0.65 with 2 df, p = 0.72; I2 = 0 percent). The odds ratio for the risk of cancer for the –160A carriers was estimated as 1.15 (95 percent CI: 0.89, 1.50), which showed no significant association of the –160A allele with progression of colorectal cancer.

Esophageal cancer
Two studies involving 407 patients and 490 controls detected the –160C/A polymorphism in esophageal cancer patients in Japan (7) and China (39). The heterogeneity tests were not statistically significant (Q = 0.46 with 1 df, p = 0.50; I2 = 0 percent (table 2)). The odds ratio estimate from the pooled data sets for the –160A carriers was 1.22 (95 percent CI: 0.93, 1.61) (Web figure 2). No significant difference was detected in allelic and genotypic frequency distribution between patients and controls, implying that the –160A allele carriers had no detectable risk of esophageal cancer.

Gastric cancer
Eleven data sets from nine studies were on the –160C/A polymorphism and gastric cancer (9, 11, 12, 39–44), which involved 1,660 cases and 1,815 controls. The odds ratio estimate for the –160A allele carriers was not significant (odds ratio (OR) = 0.98, 95 percent CI: 0.78, 1.24). However, heterogeneity among these data sets was significant (Q = 23.217 with 10 df, p = 0.01; I2 = 56.9 percent). Thus, we analyzed Asian and European subgroups separately. In the subgroups, the heterogeneity decreased effectively (table 2). Interestingly, the odds ratio estimates for the –160A carriers were less than 1.0 for Asians (OR = 0.81, 95 percent CI: 0.67, 0.99) but significantly greater than 1.0 for Europeans (OR = 1.45, 95 percent CI: 1.06, 1.97) (table 1). This finding demonstrates that the –160A allele is an ethnicity-dependent risk factor for gastric cancer.

Prostate cancer
Eight studies focused on the association between the –160A allele and prostate cancer, with 2,633 patients and 2,634 controls (8, 45–51). Overall meta-analysis showed that the –160A allele carriers had a significantly increased risk of developing prostate cancer (OR = 1.31, 95 percent CI: 1.08, 1.60). The pooled odds ratio estimates for –160A homozygotes and heterozygotes were 1.43 (95 percent CI: 0.94, 2.19) and 1.30 (95 percent CI: 1.07, 1.59), respectively (table 1). However, statistical tests showed a significant heterogeneity among these studies (Q = 17.93 with 7 df, p = 0.01; I2 = 61.0 percent). To avoid influence of the genetic heterogeneity on the association analysis, we separated the meta-analysis according to Asian and European groups (table 1). The odds ratio estimates for –160A allele carriers were 1.36 (95 percent CI: 1.05, 1.77) for Asians and 1.32 (95 percent CI: 1.02, 1.70) for Europeans. However, heterogeneity in the European group was still highly significant (p < 0.01). This finding might be mainly attributed to the data set from Verhage et al. (8), in which subjects with benign prostatic hyperplasia, vasectomy patients, and visitors were treated as normal controls. The heterogeneity was effectively removed after exclusion of this data set (Q = 4.25 with 6 df, p = 0.64; I2 = 0 percent). Sensitivity analysis did not detect significant change in the pooled odds ratio estimates before or after excluding this data set. Overall analysis suggests that E-cadherin –160A carriers have an increased risk of prostate cancer.

Urothelial cancer
Three studies investigated predisposition to urothelial cancer for carriers of the –160C/A polymorphism, involving 558 patients and 705 controls (52–54). Although no significant association was detected between this polymorphism and various developmental stages of cancer (poorly differentiated and invasiveness), carriers of the –160AA genotype had a 2.57-fold increased cancer risk (95 percent CI: 1.55, 4.24) compared with those with the wild-type homozygote. Moreover, a much larger and significant odds ratio estimate (OR = 4.16, 95 percent CI: 1.74, 9.94) was observed for the –160A allele carriers in the Chinese study (53). Frequency of the –160A allele was significantly higher among invasive than superficial cancer patients, suggesting that –160C/A may serve as a prognostic marker for transitional cell carcinoma of the bladder.

Lung cancer
We recently detected the –160C/A polymorphism in 95 hospital-based tissue samples from non-small-cell lung cancer cases and 85 randomly selected, normal controls. No occurrence of the –160A homozygote was detected in the controls, but five cases were found to carry the –160AA genotype. The odds ratio estimate for the –160A allele carriers was 2.81 (95 percent CI: 1.36, 5.83), supporting it as a significant susceptibility factor for lung cancer (unpublished data).

To our knowledge, the present article is the first general overview of associations between the E-cadherin –160C/A polymorphism and susceptibility to seven different types of cancer based on 26 up-to-date case-control studies for which information was available. In the future, this meta-analysis will be updated as evidence emerges.

Our meta-analysis showed that carriers of the –160C/A polymorphism had an increased risk of prostate cancer, but there was no obvious difference between Asians and Europeans. In particular, compared with that for the C allele in prostate cell lines, transcriptional activity in the –160A allele was found to be decreased by 68 percent (5, 45). This evidence is probably the most direct supporting gene functional alteration attributable to the –160A carried promoter in prostate cancer.

Although there was no significant association of the –160A allele with gastric cancer risk, it conferred a significantly higher risk for Europeans (OR = 1.45) and a tolerant contribution to susceptibility for Asians (OR = 0.81). When compared with their Chinese neighbors, Japanese carriers of the mutant allele even had a decreased risk of gastric cancer (41). In addition, no significant impact of the –160A allele on tumor stage and lymph node metastasis was observed. These findings suggest that the –160A allele may be a marker for genetic susceptibility rather than a prognostic marker for gastric cancer.

Furthermore, the –160AA homozygote conferred a significantly increased urothelial cancer risk (OR = 2.57), and our unpublished data support the –160A allele as a significant susceptibility factor for lung cancer. Otherwise, Lin et al. (36) reported a significant 32 percent reduction in recurrence risk for the subjects carrying at least one –160A allele, supporting the –160C/A polymorphism as a reference molecular marker for urothelial cancer recurrence.

The –160A mutant allele was not found to be a statistically significant risk factor for breast, colorectal, or esophageal cancer in this meta-analysis. For breast cancer, the explanation may be that transcriptional activity of the –160A allele was reduced by only about 10 percent (6). Regarding colorectal cancer, the –160C/A polymorphism showed neither a detectable effect on expression of E-cadherin (7) nor any significant association with localization (10) and microsatellite instability (38), one of the most distinct characteristics of colorectal cancer (55). Otherwise, the study carried out in China did not reveal any association of this polymorphism with lymph node metastasis status (39), but 71 percent of Japanese esophageal cancer patients with the –160AA genotype had an invasive phenotype (7). Thus, further research is needed to clarify this inconsistency.

One of the major concerns in a robust meta-analysis is potential bias due to selected publication of the constituent studies, as pointed out by Kaklamani et al. (56). To assess this problem, we presented the relation between the odds ratio estimates in a logarithmic scale and their corresponding standard errors across all constituent data sets and showed that the likelihood of important publication bias in the present analysis was negligible. In fact, both the Begg and Mazumdar adjusted rank correlation analysis and the Egger regression asymmetry test revealed no correlation between the estimate of odds ratio and sample size (table 2).

Another crucial question for any meta-analysis is heterogeneity between the component studies. Lack of proper consideration of this commonly occurring problem may cause a misleading statistical inference. To test the significance of heterogeneity, we carried out Cochran's Q test and calculated the quantity I2 that describes the magnitude of heterogeneity across the constituent studies. These analyses are the most popularly recommended in many meta-analyses but could be questionable because statistical power depends heavily on the sample sizes of the component data sets to be pooled (57). Given that the sample sizes used in the gastric cancer case-control studies were moderate in comparison to those in the case and control data sets for other types of cancers, and that significant heterogeneity was detected properly, the statistical tests must be recognized efficiently enough so that the confounder effect of heterogeneity is appropriately detected and assessed for other types of cancer.

The case and control samples in the constituent studies included in the present meta-analysis may have different degrees of selection bias regarding their representativeness of the corresponding natural populations, which could prevent inference from the sample-based analyses to the general population. To minimize potential artifacts in determining entry of research subjects, we conducted the meta-analysis after testing the genotype distribution at the polymorphic site in the control samples. Although there were four data sets in which the genotype distribution did not follow Hardy-Weinberg equilibrium, the corresponding meta-analysis was qualitatively similar with or without excluding them. This finding may indicate that the samples effectively maintained the most important inherent nature of population genetic structure and thus largely improves the predictability and reliability of the meta-analysis. On the other hand, stability of the analysis was confirmed by the sensitivity analysis.

In summary, the present analysis supports growing evidence that –160A in the promoter region of the E-cadherin gene is emerging as a low-penetrance tumor susceptibility allele in the development of certain forms of cancers such as gastric, lung, prostate, and urothelial. Therefore, this evidence raises the need for exploration of the molecular etiology by which development/progression of cancers can be explained by functional alteration of the E-cadherin gene and also the need for investigating this polymorphism as a potential genetic factor for different types of cancer in different ethnic populations.

This work was supported in part by grants from China's Key Basic Research Program (2004CB518605) and China's National Natural Science Foundation (30430380). Z. W. L. is also supported by research grants from the Medical Research Committee of the University of Birmingham, Biotechnology and Biological Sciences Research Council (BBSRC) and Natural Environmental Research Council (NERC) of the United Kingdom.

Conflict of interest: none declared.

• Table 1
• Table 2
• Figure 1

1. Grunwald GB. The structural and functional analysis of cadherin calcium-dependent cell adhesion molecules. Curr Opin Cell Biol (1993) 5:797–805.
2. Hirohashi S. Inactivation of the E-cadherin-mediated cell adhesion system in human cancers. Am J Pathol (1998) 153:333–9.
3. Christofori G, Semb H. The role of the cell-adhesion molecule E-cadherin as a tumor-suppressor gene. Trends Biochem Sci (1999) 24:73–6.
4. Polakis P. Wnt signaling and cancer. Genes Dev (2000) 14:1837–51.
5. Li LC, Chui RM, Sasaki M, et al. A single nucleotide polymorphism in the E-cadherin gene promoter alters transcriptional activities. Cancer Res (2000) 60:873–6.
6. Lei H, Sjoberg-Margolin S, Salahshor S, et al. CDH1 mutations are present in both ductal and lobular breast cancer but promoter allelic variants show no detectable breast cancer risk. Int J Cancer (2002) 98:199–204.
7. Nakamura A, Shimazaki T, Kaneko K, et al. Characterization of DNA polymorphisms in the E-cadherin gene (CDH1) promoter region. Mutat Res (2002) 502:19–24.
8. Verhage BA, Houwelingen KV, Ruijter TE, et al. Single-nucleotide polymorphism in the E-cadherin gene promoter modifies the risk of prostate cancer. Int J Cancer (2002) 100:683–5.
9. Wu MS, Huang SP, Chang YT, et al. Association of the –160 C→A promoter polymorphism of E-cadherin gene with gastric carcinoma risk. Cancer (2002) 94:1443–8.
10. Porter TR, Richards FM, Houlston RS, et al. Contribution of cyclin d1 (CCND1) and E-cadherin (CDH1) polymorphisms to familial and sporadic colorectal cancer. Oncogene (2002) 21:1928–33.
11. Humar B, Graziano F, Cascinu S, et al. Association of CDH1 haplotypes with susceptibility to sporadic diffuse gastric cancer. Oncogene (2002) 21:8192–5.
12. Pharoah PD, Oliveira C, Machado JC, et al. CDH1 c-160a promoter polymorphism is not associated with risk of stomach cancer. Int J Cancer (2002) 101:196–7.
13. Hulka BS, Stark AT. Breast cancer: cause and prevention. Lancet (1995) 346:883–7.
14. Berx G, Cleton-Jansen AM, Nollet F, et al. E-cadherin is a tumor/invasion suppressor gene mutated in human lobular breast cancers. EMBO J (1995) 14:6107–15.
15. Lichtenstein P, Holm NV, Verkasalo PK, et al. Environmental and heritable factors in the causation of cancer. N Engl J Med (2000) 343:78–85.
16. Garinis GA, Menounos PG, Spanakis NE, et al. Hypermethylation-associated transcriptional silencing of E-cadherin in primary sporadic colorectal carcinomas. J Pathol (2002) 198:442–9.
17. Yu HP, Zhang XY, Wang XL, et al. DNA repair gene XRCC1 polymorphisms, smoking, and esophageal cancer risk. Cancer Detect Prev (2004) 28:194–9.
18. Shi QL, Xu DH, Sun CS, et al. Study on family aggregation of esophageal cancer in Linzhou city. Chin J Prev Med (2000) 34:269–70.
19. Forman D, Newell DG, Fullerton F, et al. Association between infection with Helicobacter pylori and risk of gastric cancer. BMJ (1991) 302:1302–5.
20. El-Omar EM, Carrington M, Chow WH, et al. Interleukin-1 polymorphisms associated with increased risk of gastric cancer. Nature (2000) 404:398–402.
21. Gsur A, Feik E, Madersbacher S. Genetic polymorphisms and prostate cancer risk. World J Urol (2004) 21:414–23.
22. Lerner SP, Schoenberg M, Sternberg C. Kiemeney LALM. Familial bladder cancer Textbook of bladder cancer. (2005) London, United Kingdom: Taylor & Francis–Informa.
23. Kirkali Z, Chan T, Manoharan M, et al. Bladder cancer: epidemiology, staging and grading, and diagnosis. Urology (2005) 66(suppl 6A):4–34.
24. Christiani DC. Genetic susceptibility to lung cancer. J Clin Oncol (2006) 24:1651–2.
25. Little J, Higgins JPT, eds. The HuGENetTM HuGE review handbook, version 1.0. (http://www.hugenet.ca). (Accessed February28, 2006).
26. Cochran WG. The combination of estimates from different experiments. Biometrics (1954) 10:101–29.
27. Higgins JPT, Thompson SG, Deeks JJ, et al. Measuring inconsistency in meta-analyses. BMJ (2003) 327:557–60.
28. Petitti DB. Statistical methods in meta-analysis. In: Meta-analysis, decision analysis and cost-effectiveness analysis—Petitti DB, ed. (1994) New York, NY: Oxford University Press. 90–114.
29. Begg CB, Mazumdar M. Operating characteristics of a rank correlation test for publication bias. Biometrics (1994) 50:1088–101.
30. Egger M, Smith GD, Schneider M, et al. Bias in meta-analysis detected by a simple graphical test. BMJ (1997) 315:629–34.
31. Tobias A. Assessing the influence of a single study in the meta-analysis estimate. Stata Tech Bull (1999) 8:15–17.
32. Liu YC, Shen CY, Wu HS, et al. Helicobacter pylori infection in relation to E-cadherin gene promoter polymorphism and hypermethylation in sporadic gastric carcinomas. World J Gastroenterol (2005) 11:5174–9.
33. Ramos-de la Medina A, More H, Medina-Franco H, et al. Single nucleotide polymorphisms (SNPs) at CDH1 promoter region in familial gastric cancer. Rev Esp Enferm Dig (2006) 98:36–41.
34. 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.
35. Humar B, Toro T, Graziano F, et al. Novel germline CDH1 mutations in hereditary diffuse gastric cancer families. Hum Mutat (2002) 19:518–25.
36. Lin J, Dinney CP, Grossman HB, et al. E-cadherin promoter polymorphism (C-160A) and risk of recurrence in patients with superficial bladder cancer. Clin Genet (2006) 70:240–5.
37. Yu JC, Hsu HM, Chen ST, et al. Breast cancer risk associated with genotypic polymorphism of the genes involved in the estrogen-receptor-signaling pathway: a multigenic study on cancer susceptibility. J Biomed Sci (2006) 13:419–32.
38. Shin Y, Kim IJ, Kang HC, et al. A functional polymorphism (-347G→GA) in the E-cadherin gene is associated with colorectal cancer. Carcinogenesis (2004) 25:2173–6.
39. Zhang XF, Wang YM, Wang R, et al. Correlation of E-cadherin polymorphisms to esophageal squamous cell carcinoma and gastric cardiac adenocarcinoma. Chin J Cancer (2005) 24:513–19.
40. Shin Y, Kim IJ, Kang HC, et al. The E-cadherin -347G→GA promoter polymorphism and its effect on transcriptional regulation. Carcinogenesis (2004) 25:895–9.
41. Kuraoka K, Oue N, Yokozaki H, et al. Correlation of a single nucleotide polymorphism in the E-cadherin gene promoter with tumorigenesis and progression of gastric carcinoma in Japan. Int J Oncol (2003) 23:421–7.
42. Park WS, Cho YG, Park JY, et al. A single nucleotide polymorphism in the E-cadherin gene promoter -160 is not associated with risk of Korean gastric cancer. J Korean Med Sci (2003) 18:501–4.
43. Song CG, Huang CM, Liu X, et al. Association of -160(C→A) polymorphism in CDH1 gene with gastric cancer risk in Fujian Chinese population. Chin J Med Genet (2005) 22:557–9.
44. Lu Y, Xu YC, Shen J, et al. E-cadherin gene C-160A promoter polymorphism and risk of non-cardia gastric cancer in a Chinese population. World J Gastroenterol (2005) 11:56–60.
45. Kamoto T, Isogawa Y, Shimizu Y, et al. Association of a genetic polymorphism of the E-cadherin gene with prostate cancer in a Japanese population. Jpn J Clin Oncol (2005) 35:158–61.
46. Tsukino H, Kuroda Y, Imai H, et al. Lack of evidence for the association of E-cadherin gene polymorphism with increased risk or progression of prostate cancer. Urol Int (2004) 72:203–7.
47. Hajdinjak T, Toplak N. E-cadherin polymorphism -160 C/A and prostate cancer. Int J Cancer (2004) 109:480–1.
48. Jonsson BA, Adami HO, Hägglund M, et al. -160 C/A polymorphism in the E-cadherin gene promoter and risk of hereditary, familiar and sporadic prostate cancer. Int J Cancer (2004) 109:348–52.
49. Lindström S, Wiklund F, Jonsson BA, et al. Comprehensive genetic evaluation of common E-cadherin sequence variants and prostate cancer risk: strong confirmation of functional promoter SNP. Hum Genet (2005) 118:339–47.
50. Pookot D, Li LC, Tabatabai ZL, et al. The E-cadherin -160 C/A polymorphism and prostate cancer risk in white and black American men. J Urol (2006) 176:793–6.
51. Bonilla C, Mason T, Long L, et al. E-cadherin polymorphisms and haplotypes influence risk for prostate cancer. Prostate (2006) 66:546–56.
52. Tsukino H, Kuroda Y, Nakao H, et al. E-cadherin gene polymorphism and risk of urothelial cancer. Cancer Lett (2003) 195:53–8.
53. Zhang X, Ma X, Zhu QG, et al. Association between a C/A single nucleotide polymorphism of the E-cadherin gene promoter and transitional cell carcinoma of the bladder. J Urol (2003) 170:1379–82.
54. Kiemeney LA, van Houwelingen KP, Bogaerts M, et al. Polymorphisms in the E-cadherin (CDH1) gene promoter and the risk of bladder cancer. Eur J Cancer (2006) 42:3219–27.
55. Laiho P, Launonen V, Lahermo P, et al. Low-level microsatellite instability in most colorectal carcinomas. Cancer Res (2002) 62:1166–70.
56. Kaklamani VG, Hou N, Bian Y, et al. TGFBR1*6A and cancer risk: a meta-analysis of seven case-control studies. J Clin Oncol (2003) 21:3236–43.
57. Munafò MR, Flint J. Meta-analysis of genetic association studies. Trends Genet (2004) 20:439–44.