Association of prostate stem cell antigen gene polymorphisms with the risk of stomach cancer in Japanese



A recent whole-genome association study identified a strong association between polymorphisms in the prostate stem cell antigen (PSCA) gene and stomach cancer risk. In this case-control study, we aimed to validate this association, and further to explore environmental factors possibly interacting with PSCA polymorphisms in 708 incident stomach cancer cases and 708 age–sex matched controls. The association between PSCA polymorphisms and Helicobacter pylori infection was also examined. We found that rs2294008 and rs2976392, which were strongly linked to each other (D′ = 1.00), were significantly associated with stomach cancer risk. Per allele odds ratio for rs2994008 was 1.40 (95% confidence interval: 1.19–1.65; p = 3.7 × 10−5). We found significant interaction with a family history of stomach cancer in first-degree relatives (p-heterogeneity = 0.009). Similar to originally reported association, we found significant heterogeneity between diffuse and intestinal type (p-heterogeneity = 0.007). No association was seen between PSCA polymorphisms and H. pylori infection. In conclusion, PSCA polymorphisms are associated with stomach cancer risk in Japanese. A possible interaction with family history warrants further evaluation. © 2009 UICC

Although the age-standardized incidence of stomach in Japan is decreasing, the stomach remains one of the most common sites of cancer.1 Worldwide, the health burden of stomach cancer in terms of incidence and mortality is similarly still high.2 Although Helicobacter pylori infection is a well-established causative agent,3 epidemiologic studies exploring other risk/protective factors are still ongoing, including genetic factors.

A recent genome-wide association study (GWAS) conducted in Japanese and Korean populations reported a strong association between prostate stem cell antigen (PSCA) gene polymorphisms and the risk of stomach cancer. Findings showed that one of the polymorphisms examined, rs2294008 in the first exon, may be the locus responsible for changes in the transcriptional activity of an upstream fragment of PSCA.4 Despite the importance of this finding, the independence of this putative effect of PSCA loci from H. pylori infection in stomach carcinogenesis remains to be clarified. This study also did not investigate potential interactions with confounders.

Here, we conducted a case-control study with 3 goals: First, to validate the association between these PSCA polymorphisms and stomach cancer risk; Second, to explore the interaction between PSCA loci and confounders; and Third, to examine the association between the PSCA loci and H. pylori infection.

Material and methods

Study population

The present subjects were aged 20–79 years, and were enrolled between January 2001 and November 2005, in the framework of the Hospital-based Epidemiologic Research Program at Aichi Cancer Center (HERPACC). Details of the HERPACC have been described elsewhere.5, 6 In brief, the first version of the HERPACC (HERPACC-I) was initiated at Aichi Cancer Center Hospital, Nagoya, Japan, in 1988, with information on lifestyle factors collected from all first-visit outpatients, including cancer and non-cancer patients. The second version (HERPACC-II) was launched in 2001, and asked all first-visit outpatients to provide 7 ml of blood as well as information on lifestyle factors. Patients were asked about their lifestyle when healthy or before the current symptoms developed. Information from the HERPACC-II questionnaire was systematically collected and checked by trained interviewers. Complete responses were obtained from 96.7% of 29,538 eligible subjects, of whom 50.7% donated a blood sample. Questionnaire data were loaded into the HERPACC database and periodically linked with the hospital cancer registry system to update information on cancer incidence. All participants gave written informed consent, and the study was approved by the Ethics Committee of Aichi Cancer Center.

Cases and controls

A total of 708 patients who provided completed questionnaires and donated blood samples within the framework of HERPACC-II and were newly diagnosed with stomach cancer were deemed potential cases. Among cases, 71.6% were available for histologic information (274 cases: Intestinal type; 304 cases: Diffuse type and 130 cases: unknown).

Control subjects were randomly selected from first-visit outpatients who visited our hospital in the same period and enrolled in HERPACC-II. A total of 9,497 individuals who completed the questionnaire and donated blood samples and were confirmed to have no cancer according to the cancer registry and medical records were deemed potential controls. We excluded 642 patients with a history of cancer, leaving 8,855 controls eligible for analysis. We used non-cancer patients at our hospital as controls, given the likelihood that our cases arose within this population base. Eventually, 706 controls were individually matched for age (±3years) and sex to cases with a 1:1 case-control ratio. On examination at our hospital, 29% of controls had no abnormal finding, 32% had digestive diseases (e.g., gastritis and gastric or colon polyp), 11% had respiratory diseases (e.g., non-specific chest shadow and benign tumors), 2% had benign breast diseases (e.g., fibroadenoma), 4% had gynecologic diseases (e.g., cervix dysplasia), 8% had head and neck diseases (e.g., benign tumors), and 14% had miscellaneous other conditions (e.g., skin and orthopedic disorders).7 Our previous study demonstrated that the general lifestyles of non-cancer subjects in HERPACC were accordant with those of the general population in the same area,8 warranting the external validity of the results of HERPACC-based studies. All subjects in the present study were Japanese living in and around Aichi Prefecture, Central Japan.

Genotyping of PSCA polymorphisms

DNA of each subject was extracted from the buffy coat fraction using BioRobot EZ1 and an EZ1 DNA Blood 350 ml Kit (Qiagen, Tokyo, Japan) or DNA Blood mini kit (Qiagen). Genotyping for the 2 loci in the PSCA gene (dbSNP ID: rs2294008, rs2976392 and rs2976391) was based on TaqMan assays by Applied Biosystems (Foster City, CA). In our laboratory, the quality of genotyping was routinely assessed statistically using the Hardy-Weinberg test. When allelic distributions for controls departed from the Hardy-Weinberg frequency, genotyping was assessed using direct sequencing.

Helicobacter assessment

Among 708 controls, 644 controls were examined for plasma IgG levels for H. pylori using a commercially available direct enzyme-linked immunosorbent assay (ELISA) kit (“E Plate ‘Eiken’ H. Pylori Antibody” from Eiken Kagaku, Tokyo, Japan). This ELISA kit was developed in Japan using the antigen extracted from the domestic strain in Japan and is commonly used in medical studies.9, 10 A positive status for H. pylori infection was defined as an H. pylori IgG antibody level greater than 10 U/ml in serum. Serum pepsinogens (PG) were measured by chemiluminescence enzyme immunoassay (CLEIA). Severe gastric mucosal atrophy was defined as those with PG I ≤ 30 ng/ml and PG I/PG II ≤ 2.

Assessment of smoking exposures

Cumulative smoking dose was evaluated as pack-years, the product of the number of packs consumed per day and years of smoking. Smoking habits were entered under the 4 categories of never, former and current smoker of <40 and ≥40 pack-years. Former drinkers or smokers were defined as those who had quit drinking or smoking at least 1 year before the survey, respectively.

Statistical analysis

To assess the strength of the associations between PSCA polymorphisms and risk of stomach cancer, odd ratios (ORs) with 95% confidence intervals (CIs) were estimated using conditional logistic models adjusted for potential confounders. Potential confounders considered in the multivariate analyses were age, sex, smoking habit (never smokers, former smokers, current smokers of <40, or ≥40 pack-years), and family history of stomach cancer in a first-degree relative (yes or no). Differences in categorized demographic variables between the cases and controls were tested by the chi-squared test. Accordance with the Hardy-Weinberg, equilibrium was checked for controls using the chi-squared test and used to assess any discrepancies between genotype and allele frequencies. Under the conditions of this study of 708 cases and 708 matched controls, and assuming a risk allele frequency 0.66 (according to rs2294008 in HapMap database for Japanese in Tokyo) with a per allele OR = 1.624 and α-error of 0.001, statistical power was more than 0.99. To explore possible gene-environment interactions, we conducted stratified unconditional logistic regression analyses according to the adjusted factors. To avoid the dropping of subjects from models by stratification, a conditional logistic model was not applied. We used the Mantel-Haenzel test to assess the homogeneity of association for PSCA by stratifying factors. The association between PSCA and H. pylori infection among controls was assessed by unconditional logistic regression for age, sex and smoking habit. We defined p-values less than 0.05 as statistically significant for the homogeneity test and H. pylori analysis.

All analyses were performed using STATA version 10 (Stata Corp., College Station, TX). Power calculations for sample sizes in gene-environment interactions were performed using QUANTO.11


Demographic features of cases and controls are shown in Table I. Age and sex were appropriately matched. Heavy smokers was more frequent among cases than controls. The proportion of 20 pack-years or more current smokers was significantly higher among cases than controls. Those with a family history of stomach cancer in a first-degree relative were more common among cases.

Table I. Characteristics of Cases and Controls
  • 1

    Moderate and heavy smoker was defined as eversmoker with pack-years less than 20 and more or equal to 20, respectively.

Sex    1.00
Age (years)    0.446
 Mean age (SD)59.4 (10.4)59.8 (10.4) 
Smoking status1    <0.001
Family history of stomach cancer in the first degree relatives    0.032

Table II shows genotype distributions for rs2294008, rs2976392 and rs2976391. Genotype frequencies for all polymorphisms were in accordance with the Hardy-Weinberg law in controls: rs2294008 (p = 0.64), rs2976392 (p = 0.64) and rs2976391 (p = 0.36). Allele frequencies in each locus were consistent with those in the HapMap database. rs2294008 and rs2976392 showed linkage disequilibrium (LD) (D′ = 1.00 and R2 = 0.99) while, rs2976391 did not show LD with the other two. rs2294008 showed a statistically significant association in all models. Table III shows stratified analysis according to sex, smoking, family history of stomach cancer and histologic subtypes. A significant interaction was seen for a family history of stomach cancer (p-homogeneity = 0.009), with those with a history showing a higher per-allele OR (2.98: 1.47–6.02) than those without (1.34: 1.14–1.59). Similar to the original report,4 we found significant impact of rs2294008 in diffuse type.

Table II. Genotypes Distribution of PSCA Polymorphisms and Their Odds Ratios for Stomach Cancer Risk
PSCA locus and genotypePerallele modelDominant model1Recessive model2
Model 11Model 22OR95% CIp-valueOR95% CIp-value
OR95% CIp-valueOR95% CIp-value
  • 1

    Crude conditional logistic regression model.

  • 2

    Adjusted for smoking (non, moderate, heavy, unknwon), and family history of stomach cancer in conditional logistic regression model.

rs2294008: risk allele (A) frequency in controls subjects = 0.626
rs2976392: risk allele (A) frequency in controls subjects = 0.626
rs2976391: risk allele (C) frequency in 708 controls subjects = 0.804
Table III. Stratified Analysis According to Potential Confounding Factors for PSCA rs2294008 Genotype
Exposurers2294008 genep-homogeneity2
TTCTCCAllele model1
Cases/controlsCases/controlsCases/controlsOR95% CIp-value
  • 1

    Odds ratios were adjusted for age and sex in unconditional logistic regression models.

  • 2

    Mantel-Haenzel homogeneity test.

  • 3

    Moderate and heavy smoker was defined as eversmoker with pack-years less than 20 and more or equal to 20, respectively.

  • 4

    130 cases were excluded from analysis because of lack of histology information. Corresponding 130 controls were also excluded.

Smoking status3
Family history of stomach cancer in the first degree relatives

Among 642 controls (492 males and 150 females; mean age 60.8 years) examined for H. pylori, prevalence was 64.2% (95% confidence interval: 60.3–67.9). No significant association was seen between any PSCA polymorphism and H. pylori. The per allele OR adjusted for age, sex and smoking habit for rs2294008 was 0.97 (95% CI: 0.76–1.24, p = 0.817), indicating a lack of association between PSCA polymorphism and H. pylori infection among this Japanese population. This association was not changed if we excluded subjects with sever gastric atrophy. Further, no associations for the other polymorphisms were seen.


In this study, which had sufficient statistical power, we found a significant association between the PSCA polymorphisms rs2294008 and rs2976392 and stomach cancer risk. This association was consistent regardless of age, sex and smoking habit. Consistent with the former report, we found significant association with diffuse type gastric cancer. Interaction with a family history of stomach cancer in first-degree relatives was observed. The lack of association between these PSCA polymorphisms and H. pylori infection strengthen the independent impact of PSCA polymorphisms on stomach cancer risk.

To date, the background biological mechanism of PSCA and its transcript in stomach cancer has not been clarified. Sakamoto et al. suggested a role in signal transduction via a glycosylphosphatidylinositol anchor domain in PSCA, especially specific to diffuse type gastric cancer,12 while several other reports have suggested an involvement in cell growth regulation in various systems.13–15 In an in vitro evaluation, HSC57 cells stably expressing PSCA grew more slowly than not expressing PSCA,4 supporting a potential role for PSCA in signal transduction. On the basis of their suppression of PSCA expression in epithelial cells in several tissues, including stomach, Sakamoto et al. also suggested a tumor suppressor-like function in certain types of cancer.4 Our present study does not clarify the significance of rs2294008 and rs2976392. rs2994008 is non-synonymous polymorphism in which the first methionine is changed to threonine, possibly leading to a difference between each allele. It was suggested that rs2294008 might modulate transcriptional activity of the upstream region of PSCA.4

We found that PSCA polymorphisms had a higher impact in those with a family history of stomach cancer than in those without. To our knowledge, this is the first report of this association. Several explanations can be considered. First, this phenomenon may simply reflect dominant nature of this polymorphism. Second, the prevalence of unknown loci in the PSCA gene or neighboring regions, which confer additional functional significance to the present subject loci and links with them, is high among those with a family history of stomach cancer. These loci might be difficult to detect in GWAS owing to their lower prevalence in those with a family history. Given this small ratio (8.8% in cases and 5.8% in controls), a third explanation may be chance. Further examination of this question is warranted.

As with other hospital-based case-control studies, our controls may have differed from the general population. However, the equivalence of genotype distributions for the PSCA polymorphism between our controls and those in the HapMap database for Japanese indicates a lack of bias in the selection of controls, and the strong association even after adjustment for potential confounders therefore confirms the solidness of the association. We did not assess the H. pylori infection status of cases, but rather assessed H. pylori status and PSCA polymorphisms. This is because the comparison of plasma/serum measurement between cases and controls in a case-control study is not necessarily an appropriate way to assess a causal relation. Given several findings that advanced gastric atrophy induces the elimination of H. pylori from the gastric mucosa, measurement of H. pylori does not necessarily reflect current/former infection status.16 To our knowledge, this is the first study to report a lack of association between PSCA polymorphisms and H. pylori infection.

In conclusion, our case-control study confirms a strong association between the PSCA polymorphisms rs2294008 and rs297639s and stomach caner risk in the Japanese population. This association was independent of age, sex, smoking and drinking habits and family history of stomach cancer. The lack of association between these PSCA polymorphisms and H. pylori warrants the independence of their effect in stomach carcinogenesis. Further studies examining the biological mechanism behind this association are warranted.


The authors are grateful to many doctors, nurses and technical and administration staff of Aichi Cancer Center Hospital for the daily administration of the HERPACC study.