Prostate stem cell antigen (PSCA), a member of the LY-6/Thy-1 family of glycosylphosphatidylinositol-anchored cell surface proteins, is considered to be involved in the cell-proliferation inhibition and/or cell-death induction activity. Two single nucleotide polymorphisms (SNPs) (rs2976392 and rs2294008) in the PSCA gene were recently identified as the susceptibility loci of gastric cancer, especially in diffuse type. Therefore, this study was to investigate whether these 2 SNPs were associated with the risk of gastric cancer in Chinese population. We genotyped rs2976392 and rs2294008 in PSCA in a case–control study including 1,053 incident gastric cancer patients and 1,100 cancer-free controls in a high-risk Chinese population. We found that variant genotypes of rs2976392 (GA/AA) were associated with a significantly 37% increased risk of gastric cancer (adjusted OR =1.37, 95% CI = 1.15–1.62), compared with variant homozygote GG, and the associations were all consistently significant in both intestinal and diffuse subtypes, and among different subgroups stratified by age, sex, drinking or smoking status. Interestingly, a significant multiplicative interaction between rs2976392 (GA/AA) and alcohol drinking was detected on the development of intestinal-type gastric cancer (p = 0.009). However, rs2294008 variant genotypes (CT/TT) were associated with a nonsignificant increased risk of gastric cancer (adjusted OR = 1.14, 95% CI = 0.96–1.36). A small meta-analysis including 5 case–control studies showed undoubtedly associations between PSCA rs2294008 and rs2976392 and gastric cancer risk (OR = 1.83, 95% CI: 1.29–2.60 and OR = 1.84, 95% CI: 1.33–2.56, respectively). These findings provide further evidence supporting that the genetic variants of PSCA gene may contribute to the gastric carcinogenesis.
Approximately 40% gastric cancer cases occur in China,1 which remains a huge burden for Chinese and is one of the key public health issues in cancer prevention and control for Chinese government. Gastric cancer, predominantly adenocarcinoma, could be further classified as intestinal and diffuse subtypes.2 Intestinal-type tumors are characterized by a corpus-dominated gastritis with gastric atrophy and intestinal metaplasia, whereas diffuse-type tumors are characterized by gastritis throughout the stomach.3 Intestinal-type tumors are found predominantly in geographic areas with a high incidence of gastric cancer, whereas diffuse-type tumors are found more uniformly throughout the world.
The exact mechanism of gastric cancer development remains unclear. Although some exogenous factors, such as diet, tobacco smoke and Helicobacter pylori, have been proposed to play important role in gastric carcinogenesis, the genetic susceptibility factors may also contributed to gastric cancer development, especially to diffuse-type gastric cancer.4, 5 The prostate stem cell antigen (PSCA) gene encodes a 123-amino acid glycoprotein, which is a cell surface antigen. PSCA is highly expressed by a large proportion of human prostate tumors, including metastatic and hormone-refractory cancers, but has limited expression in normal tissues, which made it a potential utility in the diagnosis and treatment of prostate cancer.6, 7
In a recent 2-stage genome-wide association study (GWAS) of gastric cancer in Japanese population, the authors identified 2 single nucleotide polymorphisms (SNPs) in PSCA, rs2976392 and rs2294008, that were significantly associated with gastric cancer in Japanese, especially for diffuse-type gastric cancer.8 They also found that PSCA expressed in differentiating gastric epithelial cells exerting a cell-proliferation inhibition activity in vitro and was frequently silenced in gastric cancer cells. These 2 SNPs were in strong linkage disequilibrium (LD) with each other, and significant association between these 2 SNPs in PSCA and gastric cancer was further replicated in an independent Korean case–control study of gastric cancer.8
Interestingly, the frequency of risk alleles of rs2976392A and rs2294008T in populations of Japanese (0.616 and 0.617, respectively) and Korean (0.463 and 0.462, respectively)8 was more common than those in Chinese from HapMap data (0.244 and 0.256, respectively), suggesting the potential genetic heterogeneity among different populations. In this study, we hypothesized that genetic polymorphisms in PSCA identified in Japanese and Korean populations were also associated with gastric cancer in Chinese. To test this hypothesis, we genotyped SNPs of rs2976392 and rs2294008 in PSCA gene in a case–control study of 1,053 incident gastric cancer cases and 1,100 cancer-free controls in a high-risk Chinese population. In addition, a meta-analysis was also conducted to summarize the previous reports.
Material and Methods
All the subjects were genetically unrelated ethnic Han Chinese, and this case–control study was approved by the institutional review board of Nanjing Medical University. All gastric cancer cases were recruited in the areas with a high incidence of gastric cancer in Jiangsu province of eastern China, including cities of Yangzhong, Yangzhou, Yixing and Nanjing, between January 2003 and December 2006, as we previously reported.9 All the gastric cancer cases were histopathologically diagnosed, and the exclusion criteria of cases included having a history of any cancer or any metastasized cancer (carcinomas were not originally from stomach) and having undergone radiotherapy or chemotherapy. The population controls were selected from cancer-free individuals living in the same residential areas as the cases and were frequency matched to cases on age and sex. Subjects who consented to participate in the study and donate blood sample were included in this study. Each subject was personally face-to-face interviewed by trained interviewers, with a pretested questionnaire to obtain information on demographic data and related risk factors, including tobacco smoking and alcohol drinking. After the interview, ∼5-ml venous blood sample was collected from each subject, and the blood samples from the cases were taken as soon as the patients were diagnosed. Individuals that smoked once a day for more than 1 year were defined as smokers, and those who consumed 3 or more alcoholic drinks per week for more than 6 months were considered as drinkers.
We designed the PCR-restriction fragment length polymorphism assay to detect the SNP rs2294008 C>T with the primers as 5′-AGGTGGAAAGAAGGACAAAGGG-3′ (forward) and 5′-GGCCAAGCCTGCCATCAA-3′ (reverse). The 233-bp PCR products were digested overnight with the restriction enzyme NlaIII (New England BioLabs, Beverly, MA) and then separated on 3% agarose gel. The T allele resulted in 2 fragments of 193 and 40 bp, and the C allele produced 1 fragment of 233 bp. For another SNP rs2976392 G>A, we performed the primer-introduced restriction analysis-PCR assay.10 A mismatched G was introduced to the sense primer to replace A at −3 bp from the polymorphic site to create a PvuII restriction site (sense-5′-CTGGCCATCTGTCCGCAGCT-3′ and antisense-5′-CAGATGGAGGAGGATGGCTGGA-3′). The 117-bp PCR product was then digested by PvuII (New England Bio Labs, Beverly, MA). The G allele produced 2 fragments of 99 and 18 bp, whereas the A allele resulted in a single 117-bp fragment.
Genotyping was performed without knowing the subject's case or control status, and approximately equal number of the case and the control samples were assayed in each 96-well PCR plate. When genotyping was performed, 2 research assistants independently read the gel pictures. If they did not reach a consensus on the tested genotypes (<2%), they repeated the genotyping independently to reach the consensus. In addition, 10% of the samples were also randomly selected to perform the repeated assays, and the results were 100% concordant.
We searched all epidemiological studies involved in association between gastric cancer risk and PSCA rs2294008 and rs2976392 polymorphisms. Eligible studies published in English or Chinese language up to January 2010 were identified by using the search terms “PSCA and polymorphism and gastric cancer” in PubMed. Four eligible case–control studies8, 11, 12 (the study8 consisting Japanese and Korean population were divided into 2 ones) and this study were finally pooled by the meta-analysis, including 5,142 gastric cancer cases and 4,558 controls for rs2294008 and 5,158 cases and 4,578 controls for rs2976392.
The genotype data for both 2 SNPs were extracted from each study to estimate the effect values (ORs) and their 95% CIs for separate studies and pooling dataset in the dominant manner (CT/TT vs. CC for rs2294008 and GA/AA vs. GG for rs2976392) by using Review Manager 4.2.7. Heterogeneity between studies was assessed using the χ2-based Q test.13 Fix-effect model was used when the heterogeneity was absent (p > 0.05), otherwise, random-effect model was chosen.
Student's t-test or χ2 test was used to evaluate differences in the distributions of demographic characteristics, selected variables and genotypes of the 2 SNPs of PSCA (rs2976392 and rs2294008) between the cases and controls. The Hardy-Weinberg equilibrium was tested by a goodness-of-fit χ2 test to compare the observed genotype frequencies with the expected ones among the control subjects. The associations between the 2 genotypes and risk of gastric cancer were estimated by computing the odds ratios (ORs) and their 95% confidence intervals (CIs) with unconditional logistic regression analyses. Heterogeneity test among different strata according to selected variables was assessed with the χ2-based Q test. We used the PHASE 2.1 program to infer haplotype frequencies based on the observed genotypes. All the above statistical analyses were performed with Statistical Analysis System software (v.9.1.3; SAS Institute, Cary, NC).
The characteristics of the 1,053 gastric cancer cases and 1,100 cancer-free controls are summarized in Table 1. The frequency matching on age and sex between the cases and the controls was adequate as suggested by the χ2 tests (p = 0.451 for age and p = 0.209 for sex, respectively). There was not significant difference for the distribution of cigarette smoking and alcohol drinking between the cases and controls (p = 0.328 and p = 0.286, respectively). Of the 1,053 patients, 735 were intestinal, 188 diffuse, 55 mixed and 75 unknown-type gastric cancer.
Table 1. Distributions of selected characteristics in gastric cancer cases and controls
Ten (0.9%) cases and 18 (1.6%) controls for rs2976392 and 30 (2.8%) cases and 31 (2.8%) controls for rs2294008 were failed to genotype owning to poor DNA quantity and/or quality. As shown in Table 2, the observed genotype frequencies for rs2976392 and rs2294008 polymorphisms in the controls were all in Hardy-Weinberg equilibrium (p = 0.336 and p = 0.166, respectively). The allele frequencies for rs2976392-A and rs2294008-T of the PSCA gene were 29.8 and 26.8% among gastric cancer cases, compared with 25.8 and 25.3% among controls, respectively.
Table 2. Genotype and allele frequencies of PSCA polymorphism in gastric cancer cases and controls and their associations with of gastric cancer
Unconditional logistic regression analysis revealed that the rs2976392 GA heterozygote was associated with a significantly increased risk of gastric cancer (adjusted OR = 1.40, 95% CI: 1.17–1.67), and the rs2976392 AA homozygote was associated with a nonsignificantly increased risk (adjusted OR = 1.23, 95% CI: 0.88–1.72), when compared with the wild-type homozygote (GG). In the dominant model, combined rs2976392 genotypes (GA/AA) were associated with a significantly 37% increased risk of gastric cancer (adjusted OR = 1.37, 95% CI: 1.15–1.62). For the rs2294008 SNP, a similar but nonsignificant association was observed between rs2294008CT/TT genotypes and gastric cancer risk (adjusted OR = 1.14, 95% CI: 0.96–1.36). We further evaluated the contributions of both SNPs to different histological types of gastric cancer. The dominant models showed that rs2976392, but not rs2294008, had a consistently increased risk on both intestinal- and diffuse-type gastric cancer (adjusted OR = 1.31, 95% CI: 1.09–1.58 for intestinal type and adjusted OR = 1.43, 95% CI: 1.04–1.96 for diffuse type).
LD analyses showed that rs2976392 and rs2294008 were in high LD (D′ = 0.928, r2 = 0.848), suggesting that there might be haplotype effects for these 2 SNPs. Four haplotypes (G-C, A-T, A-C and G-T) were inferred according to observed genotypes, and G-C and A-T are 2 common haplotypes, which represented 94.5% of the chromosomes for cases and 97.1% of that for controls (Table 3). Compared with the most common haplotype G-C (consisting of the common alleles from each polymorphic site), the A-T haplotype, which contains both risk alleles of the 2 SNPs, was associated with a significantly increased gastric cancer risk (adjusted OR = 1.14, 95% CI: 1.00–1.32), especially in diffuse-type cases (adjusted OR = 1.31, 95% CI: 1.02–1.68), whereas the A-C haplotype was associated with a 2.77-fold significantly increased risk of gastric cancer (OR = 2.77, 95% CI = 1.86–4.12), despite of any subtypes (intestinal type: OR = 2.51, 95% CI: 1.64–3.86 and diffuse type: OR = 2.20, 95% CI: 1.12–4.32).
Table 3. Haplotype frequencies in gastric cancer cases and controls and their association with risk of gastric cancer
In addition, we examined the effects of rs2976392 and rs2294008 in subgroups according to age, sex, smoking and alcohol drinking status. As shown in Table 4, the variant genotypes of rs2976392 (GA/AA) were consistently associated with a significantly increased risk of gastric cancer in all subgroups. There were no significantly differences for the effects of rs2976392 variant genotypes on the risk of gastric cancer in strata stratified by age, sex, smoking and drinking status (p > 0.10 for all heterogeneity tests). However, the effect of rs2976392 (GA/AA) associated with increased risk of intestinal-type gastric cancer was more evident among drinkers (adjusted OR = 1.98, 95% CI: 1.36–2.88) than that in nondrinkers (adjusted OR = 1.13, 95% CI: 0.90–1.40) (p = 0.012 for heterogeneity test). Interestingly, a significant multiplicative interaction between the variant genotypes of rs2976392 and drinking was detected on the development of intestinal-type gastric cancer (p = 0.009). Compared with never-drinking subjects carrying rs2976392 GG genotype, those drinkers with rs2976392 GA or AA genotype had a significantly 1.42-fold (95% CI: 1.04–1.95) increased risk to develop intestinal-type gastric cancer (Table 5). For rs2294008, the variant genotypes rs2294008 (CT/TT) was only associated with significantly increased risk of intestinal-type gastric cancer in drinkers (adjusted OR = 1.57, 95% CI: 1.08–2.29).
Table 4. Stratified analyses of rs2976392 genotypes of prostate stem cell antigen gene polymorphisms and gastric cancer risk
Table 5. Joint effects of rs2976392 variant genotypes and drinking on risk of intestinal-type gastric cancer
A small meta-analysis based on 4 published case–control studies and this study from different Asians (2 from Japanese, 2 from Chinese and 1 from Korean) was conducted to clarify the association between the 2 SNPs of PSCA and risk of gastric cancer. As substantial heterogeneities among the 5 studies were observed for both 2 SNPs (both p < 0.001), random-effect models were used to quantitatively synthesize extracted data. As shown in Figure 1, variant genotypes of rs2294008 and rs2976392 were both undoubtedly associated with a significant increased risk of gastric cancer in dominant models (OR = 1.83, 95% CI: 1.29–2.60 for rs2294008 CT/TT vs. CC; and OR = 1.84, 95% CI: 1.33–2.56 for rs2976392 GA/AA vs. GG).
In this case–control study, we investigated the roles of PSCA rs2976392 and rs2294008 polymorphisms in gastric cancer susceptibility in a Chinese population. We found that the variant genotypes of rs2976392 (GA/AA) were associated with a significantly increased risk of both intestinal- and diffuse-type gastric cancer in this high-risk Chinese population. A significant interaction between rs2976392 GA/AA and alcohol drinking was also observed in the development of intestinal gastric cancer. Meta-analyses incorporating 4 published case–control studies and our data indicated that variant genotypes of both SNPs in PSCA were associated with an increased risk of gastric cancer. These findings support that variants in PSCA gene may contribute to the etiology of gastric carcinogenesis.
PSCA is a member of the LY-6/Thy-1 family of glycosylphosphatidylinositol-anchored cell surface proteins and is a human cancer marker closely related to stem cell antigen-2.6, 14 PSCA has been implicated to be used in the diagnosis and treatment of prostate,15 bladder,16 gastric16 and pancreatic cancer.17 Overexpression of PSCA was observed in tumors of prostate15 and pancreas17 and was also associated with tumor stage and grade,15 whereas the decreased expression of PSCA was observed in tumors from head-and-neck squamous cell,18 urothelium, esophagus and stomach.16 It was considered that PSCA involved in the cell-proliferation inhibition and/or cell-death induction activity.19, 20 PSCA protein was undetectable in the normal intestinal epithelium and was downregulated in the gastric tissue with intestinal metaplasia, and the expression of PSCA was shown frequently suppressed in both the intestinal- and diffuse-type gastric cancer, suggesting the important role of PSCA in gastric carcinogenesis.8
The PSCA gene is located on chromosome 8q24.2. Recently, in a GWAS, the Study Group of Millennium Genome Project identified 2 SNPs in PSCA contributing to gastric cancer susceptibility in Japanese and Korean populations.8 Among them, rs2294008 SNP located in the translation starting site can modulate PSCA promoter activity. In this study, we found that the rs2294008T allele was nonsignificantly associated with risk of diffuse-type gastric cancer, and the association with intestinal-type gastric cancer was only restricted in drinkers. In contrast, rs2976392, located in intron 2, was significantly associated with the risk of both diffuse- and intestinal-type gastric cancer in a Chinese population, and the result was consistent with previous findings reported in Japanese and Korean populations.8 In addition, the results from another Japanese case–control study also suggested that PSCA rs2976392 and rs2294008 polymorphisms were associated with gastric cancer risk in Japanese, especially in those with a family history of gastric cancer.12 To quantitatively synthesize data from different studies and provide favorable evidence on the association of PSCA polymorphisms with gastric cancer susceptibility, we performed a meta-analysis incorporating other 4 published case–control study8, 11, 12 and this study. Indeed, we observed an undoubtedly association between both SNPs (rs2976392 and rs2294008) and gastric cancer with a similar effect strength (OR = 1.83 for rs2976392 and OR = 1.84 for rs2294008), which further highlighted the important role of PSCA polymorphisms on gastric cancer development.
Interestingly, we found that the effects of both SNPs were weaker in Chinese population (OR < 1.5) indicated in this study and the study of Wu et al.11 than those reported in Japanese8, 12 and Korean (OR > 2.0).8 The possible explanation may be the different ethnic population with genetic heterogeneity. In this study, both the risk alleles were minor alleles (25.8 and 25.3%, respectively), which was similar to the data from another study in Chinese population (29.5 and 28.4%, respectively)11 and those from HapMap Chinese data (24.4 and 25.6%). However, these 2 risk allele frequencies were more common in Japanese (61.6 and 61.7%) and in Korean (46.3 and 46.2%).8 Furthermore, the LD strength in Chinese population (D′ = 0.92 and r2 = 0.848 in our study, and D′ = 0.93 northern Chinese11) was somewhat weaker than that in Japanese with almost complete LD (D′ = 0.999 and r2 = 0.9958; or D′ = 1.00 and r2 = 0.9912). In addition, we inferred haplotypes of the 2 SNPs rs2976392 and rs2294008 and showed that rs2976392A-rs2294008T and rs2976392A-rs2294008C haplotypes were significantly associated with increased risk of gastric cancer, indicating that rs2976392A, but not rs2294008T, may be more important in modulating the gastric cancer susceptibility. However, the functional significance of the SNP rs2976392 was remained unknown. It is also possible that rs2976392 is in high LD with other potentially functional or causal SNPs contributing to the development of gastric cancer.
Intestinal and diffuse type have different epidemiology and pathogenesis. The intestinal-type gastric cancer is international variation and predominates in high-risk geographic areas, especially in Japan, Korea and China, whereas the diffuse-type gastric cancer has uniform geographic distribution.21, 22 Environmental factors have been proposed to play an important role in the etiology of intestinal-type gastric cancer, and the genetic susceptibility factors may contributed more to diffuse-type gastric cancer development. Interestingly, in the stratified analyses, we found that the effect of rs2976392 variant genotypes associated with intestinal-type gastric cancer was more pronounced in drinkers, and the variant genotypes of rs2294008 was also associated with significantly increased risk of intestinal-type gastric cancer in drinkers. A significant multiplicative interaction between rs2976392A allele and alcohol drinking was observed on the development of intestinal-type gastric cancer, though the exact biological mechanism was unclear and need to be clarified in further studies.
The rs2294008 C>T change was shown to be functionally associated with downregulation of PSCA and to increase the risk of gastric cancer, especially for diffuse type, in Japanese and Korean populations,8 and we did observe a similar trend but weak effect in our Chinese population. When assuming the frequency of 43.4% for rs2294008 variant genotypes (CT/TT) and significant level of 0.05, the sample size (1,069 cases and 1,023 controls) in our study has statistic power of only 31.9% to detect the effect of 1.14 (OR value in this study for CT/TT vs. CC) (calculated by using DSTPLAN Version 4.2, Houston, TX). Hence, the limited statistic power and the low penetrance effect of the locus may result in the nonsignificant association in this study.
In summary, our study findings indicate that the genetic variants of PSCA may predispose to the susceptibility of both diffuse- and intestinal-type gastric cancer in Chinese population. rs2976392 variant genotypes may be a risk marker for gastric cancer and exert potential role in intestinal-type gastric cancer development by interacting with alcohol drinking. Large population-based prospective studies with ethnically diverse populations, as well as biologically functional studies, are warranted to verify these findings.