Innate Immunity and Non-Hodgkin’s Lymphoma (NHL) Related Genes in a Nested Case-Control Study for Gastric Cancer Risk

Objective Genetic variants regulating the host immune system may contribute to the susceptibility for the development of gastric cancer. Little is known about the role of the innate immunity- and non-Hodgkin’s lymphoma (NHL)-related genes for gastric cancer risk. This nested case-control study was conducted to identify candidate genes for gastric cancer risk for future studies. Methods In the Discovery phase, 3,072 SNPs in 203 innate immunity- and 264 NHL-related genes using the Illumine GoldenGateTM OPA Panel were analyzed in 42 matched case-control sets selected from the Korean Multi-center Cancer Cohort (KMCC). Six significant SNPs in four innate immunity (DEFA6, DEFB1, JAK3, and ACAA1) and 11 SNPs in nine NHL-related genes (INSL3, CHMP7, BCL2L11, TNFRSF8, RAD50, CASP7, CHUK, CD79B, and CLDN9) with a permutated p-value <0.01 were re-genotyped in the Replication phase among 386 cases and 348 controls. Odds ratios (ORs) for gastric cancer risk were estimated adjusting for age, smoking status, and H. pylori and CagA sero-positivity. Summarized ORs in the total study population (428 cases and 390 controls) are presented using pooled- and meta-analyses. Results Four SNPs had no heterogeneity across the phases: in the meta-analysis, DEFA6 rs13275170 and DEFB1 rs2738169 had both a 1.3-fold increased odds ratio (OR) for gastric cancer (95% CIs = 1.1–1.6; and 1.1–1.5, respectively). INSL3 rs10421916 and rs11088680 had both a 0.8-fold decreased OR for gastric cancer (95% CIs = 0.7–0.97; and 0.7–0.9, respectively). Conclusions Our findings suggest that certain variants in the innate immunity and NHL-related genes affect the gastric cancer risk, perhaps by modulating infection-inflammation-immunity mechanisms that remain to be defined.


Introduction
Gastric cancer is the second most common cause for cancer death in the world [1] with a marked geographic and ethnic variation in incidence and mortality [2]. Helicobacter pylori (H. pylori) infection, identified by the International Agency for Research on Cancer (IARC) as a gastric carcinogen, is a proven causal factor for gastric cancer [1,2]. Nevertheless, some epidemiological studies have reported despite a high H. pylori prevalence rate, the incidence of gastric cancer is low in some Asian and African populations, known as the Asian/African enigma [3,4]. This enigma may be due to other risk factors related to differences in individual susceptibility to gastric cancer.
In H. pylori infections, chronic gastric inflammation occurs and leads to an immune response that includes normal (i.e., the innate immunity) and abnormal immune responses which commonly occur in Non-Hodgkin's lymphoma (NHL) [5][6][7][8]; and these processes lead to atrophic gastritis followed by metaplasia and dysplasia, and eventually connect to the development of gastric cancer [5][6][7][8]. Thus, it is assumed that normal immune responses triggered by H. pylori infection might be connected to abnormal immune responses in a direct and/or indirect manner and finally, might determine the final outcome such as gastric adenocarcinoma, lymphoma, and other malignancies.
Given that 1) a person's genome is a factor for individual susceptibility; and 2) H. pylori infection in gastric tissue leads to inflammation that in turn induces normal/abnormal immune responses, genes coding proteins related to the host normal and/or abnormal immunity may be of importance for individual susceptibility to gastric cancer risk. Several epidemiologic studies have demonstrated that the genes related to the host immune response play a critical role in gastric carcinogenesis: 1) proinflammatory host genotypes induced hypochlorhydric/atrophic responses to gastric H. pylori infections, which progressing to noncardia gastric adenocarcinoma [6]; 2) a genetic variant of TLR9, which plays an important role in the innate immune system, was significantly associated with H. pylori-induced premalignant gastric changes [7]; 3) TLR4 Asp299Gly might be a genetic susceptible factor for gastric mucosa-associated lymphoid tissue (MALT) lymphoma [8]; and 4) the CD14 genetic variants appear to be involved in the development of gastric MALT lymphoma [9].
Based on the evidence, we hypothesized that genetic variants in the normal immune response (i.e., the innate immunity related genes) and normal immune response (i.e., the NHL related genes) could influence gastric cancer risk by modulating individual susceptibility differences. To evaluate our hypothesis, we conducted a two-stage genetic analysis including: 1) the Discovery phase to explore the 1,536 SNPs in 203 innate immunity genes and 1,536 SNPs in 264 NHL related genes using the GoldenGate (Illumina) oligonucleotide pool assay (OPA) and 2) the Replication phase that further analyzed the most significant SNPs in the Discovery phase.

Study Subjects
Subjects were selected from the Korean Multi-center Cancer Cohort (KMCC) in this population-based nested case-control study. The design and sampling strategy of the KMCC have been described in detail elsewhere [10][11][12]. Briefly, from 1993 to 2004, a total of 19,688 participants were recruited from four urban and rural areas in Korea. All participants 1) signed an informed consent form before entering the cohort; 2) completed detailed standardized questionnaires by personal interview; 3) donated blood and urine samples; and 4) were passively followed-up through record linkages to the national death certificate, health insurance medical records databases and the national cancer registry.
By December 2001, a total of 100 incident gastric cancer cases were newly identified according to the International Statistical Classification of Diseases and Related Health Problems 10th Revision (ICD-10, C16). Of them, 42 cases and 42 matched controls by age, sex, enrollment area, and year were included in the Discovery phase and successfully genotyped with the Gold-enGate (Illumina) oligonucleotide pool assay (OPA) panels. In the Replication phase, the study populations were selected as follows: 1) according to the same methodology for case ascertainments mentioned above, 199 gastric cancer cases were additionally defined from the KMCC in December 2008 and matched 1:1 to controls according to age (65 years), sex, and enrollment year; 2) at the Chungnam University Hospital and Hanyang University GURI Hospital, 189 newly diagnosed gastric cancer cases were recruited from March 2002 to September 2006. They provided written informed consent and the blood samples at the time of diagnosis or prior to gastric cancer surgery; and 3) a total of 189 community-based controls matched by age (65 years), sex and enrollment year were randomly selected from the KMCC. Of the 388 case-control matches, two cases and forty controls were eliminated due to poor genotyping rate and genomic DNA level.
Finally, 386 cases and 348 controls were analyzed in the Replication phase.
Additionally, based on serology-based detection, H. pylori infection and CagA/VacA seropositivity were evaluated among all the study subjects included in both phases. Using the immunoblot assay Helico Blot 2.1 TM (MP Biomedicals Asia Pacific, Singapore) which assures high sensitivity (99% for both H.pylori and CagA seropositivity) as well as specificity (98% for H. pylori and 90% for CagA seropositivity) among the Korean population [13], serum samples were analyzed and interrelated in accordance with the manufacturer's instructions. Given that CagA-producing H. pylori are significantly associated with an increased risk of gastric cancer, H. pylori and CagA seropositivity were treated as potent covariates for gastric cancer and adjusted for in the statistical models.

Ethics Statement
The study protocols of the KMCC and the genetic study were approved by the institutional review boards of Seoul National University Hospital (H-0110-084-002, and C-0603-161-170, respectively). In addition, the hospital-based gastric cancer study was approved by the Institutional Review Board of Hanyang University Hospital (IRB no.2003-4). All study participants signed an informed consent form before entering the studies.

SNP Selection and Genotyping
In the Discovery phase, the Illumina GoldenGate TM OPA Panel was designed with 203 innate immunity and 264 NHL related candidate genes using SNPs in the SNP500Cancer project (http://snp500cancer.nci.nih.gov) with known re-sequence data and gene related carcinogenic mechanisms [14]. Focusing on a region that was 20 kb 59 to the start of transcription (exon 1) and 10 kb 39 to the end of the last exon (N) for each candidate gene, candidate SNPs were preliminarily screened. After this, Tag SNPs were selected from the designable set of SNPs that were part of the International HapMap Project (http://www.hapmap.org/index. html) and in later iterations, we used TagZilla (http://tagzilla.nci. nih.gov) according to the following criteria: 1) minor allele frequency (MAF) .0.05; 2) design score = 1.1; and 3) r 2 .0.8 cut-off using dbSNP from HapMap Caucasian (CEU) (data not shown). About 55% of the SNPs were located in the introns, 22% in the promoters (flanking region, UTR), 15% 39 to the stop codon (STP), and 9% in the exons; among the SNPs that were located in exon, 73% were synonymous and 27% were nonsynonymous changes [15].
Genotyping was performed using genomic DNA extracted from peripheral blood with the Gentra Puregene Blood Kit (Gentra, Minneapolis, USA) at the Core Genotyping Facility (CGF) of the Division of Cancer Epidemiology and Genetics, National Cancer Institute. In this study, genotype completion on genomic DNA samples exceeded 95%.
For the Replication phase, six SNPs of four genes in the innate immunity gene panel (DEFA6 rs2738120, DEFB1 rs2702829, DEFA6 rs13275170, JAK3 rs2286662, DEFB1 rs2738169, and ACAA1 rs2239621) and 11 SNPs of nine genes in the NHL panel rs11862306, and TNFRSF8 rs641941) were selected in accordance with the following criteria: the SNPs with 1) a permutated pvalue,0.01; and/or 2) Benjamini-Hochberg false discovery rate (BH-FDR) corrected p-values ,0.2 on the most significant SNPs of all the genes in the Discovery phase. Genotyping was performed using the Illumina VeraCode GoldenGate Assay with BeadXpress according to the manufacturer's instructions (Illumina, San Diego, CA, USA) [16]. To ensure the reliability of the genotyping methods in the two phases, 188 samples were genotyped twice by each method. The concordance rate was .98.4%. Two cases and forty controls with insufficient DNA or a genotyping call rate ,90% were excluded.

Statistical Analysis
Hardy-Weinberg equilibrium (HWE) in the control group was evaluated using the chi-square test or Fisher's exact test with a cutoff level of HWE ,0.0001. Selected characteristics were evaluated using x 2 for categorical variables and t tests for continuous variables. We evaluated the association between individual SNPs and gastric cancer risk based on both the raw and permutated pvalues that was computed using the Likelihood ratio test (LRT) with 1 degree of freedom in the trend (additive) model. The trend test assumes a dose response effect (i.e., linear effect) with an increasing number of variant alleles. Permutated p-values were estimated by 10,000 permutation tests. Odds ratios (ORs) and 95% confidence intervals (95% CIs) were estimated using logistic regression model adjusting for age, cigarette smoking status (yes vs. no), H. pylori infectivity, and CagA virulent factors (positive vs. negative) which are alleged risk factors for gastric cancer although the variables were not significant in our data. When data are sparse and asymptotic approximations based on logistic regression models are unreliable, exact inference should be considered [17]. Thus, if the genotype frequency in either the cases or controls was less than 10% (N = 4), exact logistic regression analysis was conducted to calculate the ORs and 95% CIs.
To avoid spurious associations with false positive outcomes, we selected all SNPs with the most significant p-value for each gene, and the Benjamini-Hochberg false discovery rate (BH-FDR) corrected p-values of each SNP was computed according to the number of genes [18]. We primarily selected significant SNPs with a permutated p-value,0.01 and a BH-FDR p-value,0.2.
In the Replication phase, the most significant SNPs identified in the Discovery phase were re-evaluated. Based on the additive or recessive models, gastric cancer risk was estimated as ORs and 95% CIs using logistic regression model adjusting for the same covariates mentioned above. To summarize the results from the Discovery and the Replication phases, pooled-and meta-analyses were conducted. Using the fixed effect model, the summarized ORs and 95% CIs were computed. In addition, heterogeneity across the studies was evaluated by the Cochran Q statistics [19].

Results
Cases and controls were comparable in age, H. pylori infection, CagA/VacA seropositivity, drinking status, and gastric ulcer history. A marginally significant and significant different distribution for cigarette smoking between cases and controls was observed in the Replication and Pooled analyses (Table 1).

Discussion
A two-stage genetic association study with 1,536 SNPs in 203 innate immunity related (i.e., the normal immune response related) genes and 1,536 SNPs in 264 NHL-related (i.e., the abnormal immune response related) genes was conducted to identify variants in innate immunity and NHL-related candidate genes associated with gastric cancer. In the Discovery phase, six SNPs in four innate immunity genes (DEFA6, DEFB1, JAK3, and ACAA1) and 11 SNPs in nine NHL-related genes (INSL3, CHMP7, CASP7, RAD50, CHUK, TNFRSF8, CD79B, BCL2L11, and CLDN9) were significantly associated with gastric cancer risk (permutated p,0.01); and after the BH-FDR test for the genes in 42 sets of cases and controls, DEFA6 and DEFB1 in the innate immunity gene panel and INSL3 in the NHL-related gene panel were significantly associated with gastric cancer risk (BH-FDR p,0.1). In the Replication genotyping phase with more case-control sets and in the meta-and pooled analyses, DEFA6, DEFB1, and INSL3 were still significantly associated with gastric cancer risk without heterogeneity across the phases.
DEFA6 (defensin alpha 6, Paneth cell-specific) and DEFB1 (defensin beta 1), which belong to the defensin family members, consists of 6 human beta-defensins (hBD1-hBD6) and two human alpha-defensins (hD5-hD6) and play a crucial role in the host defense mechanism, especially in relation to antimicrobial activity and cytotoxicity of neutrophils [20,21]. Defensins inducible by inflammation or infections can resist bacterial colonization at epithelial surfaces including gastrointestinal epithelium [22][23][24]. In particular, defensin binds lipopolysaccharide on the H. pylori cell wall and the defensin-H. pylori complex activates the NF-kB signal transduction pathway for gastric carcinogenesis [25]. Recent studies indicated defensins involved in cancer development were expressed in various human tumors including gastric cancer [26][27][28]. As one of the potential agents for gastric carcinogenesis, defensins may interact with H. pylori infection for both resistance and protection, and their abnormal activities can induce the first step toward gastric cancer development. Thus, it is possible that some variants in DEFA6 and DEFB1 affect the susceptibility to gastric cancer by modifying the inflammatory response through changes in the expression or function of those proteins. Our findings suggest that DEFA6 rs2738120 (located in the intron) and DEFB1 rs2702829 (located in the 39 of STP) have potential genetic effects on the development of gastric cancer. Despite the heterogeneity of the discovery and replication phase, the other locations of DEFA6 (rs2738120) and DEFB1 (rs2702829) were also significant in our results. Further studies should be conducted to Table 1. General characteristics of gastric cancer cases and controls selected for the screening of 203 innate immunity-and 264 NHL-related candidate genes in the discovery, replication and pooled analyses. explore the exact biological functions of the SNPs related to the defensin activity and/or cellular capacity. The INSL3 gene encodes the protein INSL3, which is an insulin-like hormone. The INSL3 protein is produced mainly in gonadal tissues in the human body; however, it has been identified in tumor tissues of the gastrointestinal tract [29]. Little is known about how INSL3 gene interferes in human carcinogenesis. In addition, it is not well known whether the genetic variants of the INSL3 gene are linked to gastric carcinogenesis, but the results of this study indicate that at least two SNPs (rs10421916 located in the intron and rs11086080 located in the 39 of STP) in strong linkage disequilibrium (D9 = 0.86, LOD = 13.42, and r 2 = 0.56) appear to have a strong association with a reduced risk for gastric cancer. Although experimental evidence indicating a functional effect of the SNPs on a molecular level is insufficient to date, it is speculated that INSL3 rs10421916 and rs11086080 may affect individual susceptibility in gastric carcinogenesis. Further research including in-depth in vivo and animal studies should be done to clarify the causal mechanisms and the INSL3 genetic effects involved in human gastric carcinogenesis.
A number of genes (i.e., JAK3, CHMP7, CASP7, RAD50, CHUK, TNFRSF8, CD79B, BCL2L11, and CLDN9) showed a significant association with gastric cancer in the Discovery phase, although not significant in the BH-FDR tests. JAK3, TNFRSF8, and CHUK might affect chronic inflammation and immunity with H. pylori colonization toward gastric cancer, which is important in gastric carcinogenesis as follows: 1) JAK3 is involved in cytokine receptormediated intracellular signal transduction, and the cytokine receptor was commonly found in human gastric cancer tissue [30]; 2) TNFRSF8 mediated signal transduction leads to activation of NF-kB [31]; and 3) CHUK is involved in NF-kB signaling pathway activated by H. pylori in gastric cancer cells [32]. Given that the human gastric pathogen H. pylori exerts much of its pathogenicity by inducing DNA damage and apoptosis in host gastric epithelial cells [33], BCL2L11, CASP7, and CD79B reported to be linked to apoptosis [34][35][36][37][38] and RAD50 and ACAA1 reported to be linked to the DNA damage and repair system in relation to gastric cancer development [39,40] also showed a biological plausibility. CLDN9 acts as a tumor suppressor gene implicated in intestinal-type gastric cancer [41].
Although this study is a two-stage genetic association study, there are several limitations to the study. First, most genes did not overcome the heterogeneity across the Discovery and Replication phases. The heterogeneity is attributed to the different statistical power in each phase: A small sample size in the Discovery phase resulted in biased ORs (i.e., over-estimated), and the results were statistically distinguished from the normal-estimated values in the Replication phase. Although the distribution of the environmental factors between the cases ascertained from a community-based cohort study and hospital-based cases was slightly different from each other, the MAFs of most SNPs had little difference. Second, stratified analyses concerning the diverse characteristics of gastric cancer (i.e., histological subtypes, cardiac vs. non-cardiac; diffuse type vs. intestinal type; and TNM stage) and/or independent risk factors (i.e., age, sex, and H. pylori infection status) could not be conducted because of 1) the homogeneous nature of the study population (i.e., most patients showed non-cardiac types or unspecified types, .95%; and the intestinal and diffuse types Table 3. The results of the pooled-and meta-analyses including the discovery and replication phases: Association of most significant SNPs in the innate immunity and NHL-related gene variants and gastric cancer risk. were less than 3%); and 2) a lack of detail information on the histopathology data due to passive follow-up using data linkage methods. Third, since tag SNPs were selected based on data from a Caucasian population, the Asian-specific genetic markers might be disregarded in the design stage. Finally, we could not confirm a functional effect of the SNP at a molecular level. Thus, the results should be interpreted with caution. In spite of the limitations, our study is stable against several biases that are common in retrospective designs due to its population-based nested case-control design.
In conclusion, our findings suggest DEFA6, DEFB1, and INSL3 genetic variants are susceptibility factors for the development of gastric cancer. Further studies with a greater number of cases and a more substantial genomic coverage will allow us to elucidate the pathological mechanisms of gastric cancer. Moreover, further biological studies focused on these genes should clarify their roles in gastric carcinogenesis.

Supporting Information
Supporting Information S1 Detailed information on the selected SNPs in 203 innate immunity genes and 264 NHL related genes: In the discovery phase. (XLSX)