Lack of Association Between CDH1 C−160A Genetic Polymorphism and Gastric Cancer Risk Among Asian Population

To the editor:

W e read with interest the recent report by Cui et al. (2011) that investigated the association of E-cadherin (CDH1) C−160A single-nucleotide polymorphism with gastric cancer susceptibility by performing a meta-analysis including 14 case–control studies with 2698 cases and 3723 controls, and the results suggested that there is no apparent relationship between CDH1 C−160A polymorphism with the susceptibility to gastric cancer in overall population, whereas CDH1 C−160A polymorphism may contribute to gastric cancer risk among Asians. This study adds further evidence to the potential role of CDH1 C−160A polymorphism in the development of gastric cancer. However, there are some concerns worth commenting.

First, in the results, they stated that “One study (Kuraoka et al., 2003) deviated from HWE. For brevity and uniformity, that study was removed from our meta-analysis.” However, of the 14 studies, the genotype distributions of the control population in two studies were not in agreement with Hardy-Weinberg equilibrium (HWE) (Kuraoka et al., 2003; Zhang et al., 2008b). It has been suggested that sensitivity analyses including and excluding the HWE-violating studies may need to be routinely conducted in meta-analyses of genetic association studies (Mao et al., 2010). Second, the data reported by Cui et al. (2011) for the study by Humar et al. (2002) were not consistent with the data provided by Humar et al. (2002) in their original publication. In the study by Humar et al. (2002), the numbers for CC, CA, and AA genotypes in control populations was 40, 27, and 3, respectively. Additionally, the study by Zhang et al. (2008a) was conducted in two populations with high and low risk of sporadic gastric carcinoma in China, in which the low-risk area includes 572 gastric cancer cases and 625 nonmalignant controls, while the high-risk area includes 489 subjects with 196 chronic atrophic gastritis, 197 intestinal metaplasia/gastric dysplasia, and 96 gastric cancer. In this meta-analysis, Cui et al. (2011) only chose the 96 gastric cancer patients in high-risk area as cases; however, they omitted the study in the low-risk area. On the other hand, 196 subjects with chronic atrophic gastritis, which is associated with the development of gastric cancer (Correa, 1988), were erroneously considered as controls. One should merge the gastric cancer cases from these two areas and choose the controls in low-risk area as the control group.

Third, another often neglected factor worth consideration is subgroup analysis by ethnicity. Several studies suggested that Mexican and Brazilian population have experienced large degrees of recent admixture with ancestors from widely divergent regions (Collins-Schramm et al., 2004; Lins et al., 2011). In this meta-analysis, there are two studies performed in Mexico and Brazil, respectively, which were both treated separately. Given above all, it seems appropriate to treat them as mixed. In addition, the Omani population should be regarded as Caucasian rather than Arab.

Fourth, using the same search strategy and end-of-search date as those of Cui et al. (2011), we have located two other relevant case–control studies conducted by Cattaneo et al. (2006) and Shin et al. (2004) in PubMed, which were not included in this meta-analysis. Of the two studies, one was carried out by Cattaneo et al. (2006) with 107 cases and 246 controls in Italy, and the other study was by Shin et al. (2004) involving 38 cases and 24 controls in Korea. Available data could be obtained from the two studies that were both satisfied the search criteria and used to calculate the odds ratios (ORs) with confidence intervals (CIs). However, the authors omitted these two studies.

In light of the issues mentioned previously, we performed a new meta-analysis including the two aforementioned eligible studies. There was no evidence of associations for all genetic models in overall population when all studies were pooled into the meta-analysis (A vs. C: OR=1.05, 95% CI=0.93–1.19, P _h=0.001; AA vs. CC: OR=1.18, 95% CI=0.86–1.61, P _h=0.003; CA vs. CC: OR=0.99, 95% CI=0.86–1.15, P _h=0.017; Dominant model: OR=1.03, 95% CI=0.89–1.19, P _h=0.005; Recessive model: OR=1.17, 95% CI=0.88–1.57, P _h=0.08). In the subgroup analysis by ethnicity and source of controls, significant associations were found in mixed ethnicity (A vs. C: OR=2.00, 95% CI=1.32–3.02, P _h=0.709; AA vs. CC: OR=3.65, 95% CI=1.51–8.81, P _h=0.371; Dominant model: OR=2.00, 95% CI=1.16–3.46, P _h=0.903; Recessive model: OR=3.01, 95% CI=1.29–7.03, P _h=0.359) rather than in the Asian population, which was found by Cui et al. (2011).

Collectively, the results of the meta-analysis should be treated with caution. The association between CDH1 C−160A polymorphism and gastric cancer risk is complicated, and further studies on Asian populations are needed to demonstrate these results in larger sets of case–control studies.