Association between polymorphisms in segregation genes BUB1B and TTK and gastric cancer risk

The study population (n = 284) consisted of 108 Slovenian patients with gastric cancer and 176 control subjects who at the time of peripheral blood extraction did not have cancer. Tumour and corresponding non-tumour tissues at least 7 cm away from the edge of the adenocarcinoma were collected from patients who were admitted to the Clinical Department for Abdominal Surgery at the University Medical Centre Ljubljana and Department for Pathology at the Institute of Oncology Ljubljana during the years 2000-2008. Samples were macrodissected by pathologist, frozen in liquid nitrogen and stored at −70°C. Comprehensive medical data were obtained from registries and pathologist’s evaluation. The following clinico-histopathological parameters were recorded: tumour differentiation (grade), location, blood and lymphatic vessel invasion (vascular invasion, perineural invasion), occurrence of tumour cells in the lymphatic vessels (lymphatic invasion), depth of invasion (pT), lymph node involvement (pN), and presence of distant metastases (pM). The gastric cancer cases were classified into diffuse type (n = 46) and intestinal (n = 58) according to Lauren classification. The mean age ± standard deviation (SD) of patients was 66.12 ± 12.02 (range, 33-87 years), and the percentage of men was 63.0%. Cases lost to follow-up (n = 6) and those, who died within 30 days after surgery (n = 2), were excluded from survival analyses. The control population was randomly selected during the years 1999-2007 and shared the ethnic and geographic background of the gastric cancer patients. The research was approved by the National Medical Ethics Committee of the Republic of Slovenia and confidentiality of personal medical data as well as other data relating to individual identification has been assured in accordance with the Helsinki Declaration.

Genomic DNA from gastric tumour and nontumour tissues was extracted using a Wizard® Genomic DNA Purification Kit (Promega, Madison, WI, USA) and QuickGene™ DNA Tissue Kit S (Fujifilm Corporation, Tokyo, Japan) on QuickGene-810 DNA isolation system (Fujifilm) according to manufacturer’s protocol. Genomic DNA from control population was extracted from peripheral blood samples using Wizard® Genomic DNA Purification Kit (Promega) following the manufacturer’s protocol. The DNA was quantified using a NanoDrop spectrophotometer (Thermo Fisher Scientific Inc.). Genotyping for polymorphism rs151658 (C>G) in TTK gene, and polymorphisms rs1031963 (C>T) and rs1801376 (A>G) in BUB1B gene was performed using TaqMan-based allele-specific polymerase chain reaction assays on the ABI Prism 7000 Sequencing Detection System apparatus (Applied Biosystems, Foster City, CA, USA) according to the procedure recommended by Applied Biosystems. The 10 μL reaction volume contained 100 ng of DNA. Assay IDs were: C_3181603_10, C_1237153_10, and C_3052718_1. In order to confirm the veracity of the results, the polymorphisms were re-genotyped by direct sequencing on a randomly selected smaller batch of samples.

A total of 21 paired gastric adenocarcinoma (GA) and adjacent control tissue samples were ground with a mortar and pestle in liquid nitrogen and lysed with 7 mol/L urea, 2 mol/L thiourea, 40 g/L CHAPS, with a protease inhibitor cocktail (Sigma-Aldrich, St. Louis, MO, USA). For every 10 mg tissue, 50 μl lysis buffer was added. After sonication on ice (3 × 10 s), the samples were incubated for 1 h on ice with occasional vortexing, and then centrifuged at 20,000 × g for 1 h at 4°C. The supernatants were collected and the protein concentrations were determined using the commercial Bradford reagent (Thermo Fisher Scientific, Waltham, MA, USA) with BSA used as the standard. Immunoblot analysis was performed on 42 samples. A total of 30 μg protein per sample was loaded onto 10% gels, separated using SDS-PAGE, and transferred onto PDVF membranes (Millipore, Billerica, MA, USA), which were then blocked in 50 g/L skimmed milk 1 h. The primary antibody was used in the following dilution: anti-Mpsl (anti-TTK) antibody, 1 pg/ml (ab11108, Abcam, Cambridge, UK). Horseradish peroxidase-conjugated secondary antibody was used in the following dilution: goat anti-mouse antibody, 1:5000 (115-035-062, Jackson ImmunoResearch, Newmarket, Suffolk, UK). The proteins were revealed by chemiluminescence using LAS-4000 CCD camera (Fujifilm, Tokyo, Japan). The blots were then quantified with Multi Gauge software (Fujifilm) and the intensities were normalized to Ponceau-S-stained membranes, to allow for loading and transfer variations.

Statistical evaluation of the genotyping data was carried out using the χ² or Fischer’s exact tests to compare the groups regarding genotype frequencies. Hardy-Weinberg (HW) equilibrium was calculated with an online program (http://www.genes.org.uk/software/hardy-weinberg.shtml).31 Survival was assessed by the Kaplan-Meier method and differences between groups were evaluated using the log-rank test. Multivariate survival analyses were further performed using the Cox proportional-hazards regression model. In the Cox multivariate analyses, forced entry procedure was used to determine the predictor variables. Only the variables that resulted in p-values < 0.05 in the Kaplan-Meier test were entered into the Cox proportional hazard model for the determination of independent prognostic factors for gastric cancer. The postoperative period was measured from the date of surgery to the date of the last follow-up or death. Statistical software used for calculations was IBM® SPSS® Statistics Version 20. For all statistical tests, a probability level (p-value) of less than 0.05 was considered significant.

To assess the statistical significance of altered protein abundance in the immunoblotting (as the tumour vs. non-tumour paired samples), non-parametric Wilcoxon signed-rank test was used. The tests were double-sided and the values with p < 0.05 with a confidence level of 95% were considered to be statistically significant. To assess the correlation of the altered protein abundance from the immunoblotting with the histopathological parameters, repeated measures ANOVA was used. The values with p < 0.05 were considered to be statistically significant. Bonferroni post-tests were used to determine where the differences were significant. All of the analyses were performed using Microsoft Office Excel 2007 (Microsoft Corporation, Washington, USA) and GraphPad Prism 5 (GraphPad Software, Inc., California, USA).

In order to functionally evaluate intronic polymorphisms we identified their effect in the context of polymorphic biological sequences on protein binding motifs. We used web-based software PROMO, which is part of the ALGGENE web-server.32,33 The search for putative binding sites was performed using the following parameters: human species, all motifs, and all factors. The data for comparisons of genotype frequencies in European populations of examined SNPs in this study was extracted from the 1000 Genomes Project data platform using a specific version of the Ensembl browser (http://browser.1000genomes.org).34

The clinical information and demographic characteristics of selected patients with gastric cancer in this study are summarized in Table 1. At the end of a period of up to 11 years of follow-up, a total of 69 patients out of 100 have died.

The overall 5-year survival was 33.5%. No statistically significant association between tested genetic variations and survival was observed (p > 0.05). Univariate survival analysis showed that only Lauren’s classification and lymph node involvement (pN) were significant prognostic factors. Diffuse type predicted shorter 5-year survival (logrank test, χ² = 5.516, p = 0.019) with overall mean estimate of survival for patients with intestinal type 64.67 months ± 7.74 (SE) (CI = 49.50-79.84) and 39.73 months ± 6.83 (SE) (CI = 26.75-53.11) for patients with diffuse type of gastric cancer. Regarding the parameter pN, patients with 7 or more positive lymph nodes had shorter survival time of 21.93 months ± 4.30 (SE) with CI = 13.5030-36 (log-rank test, χ² = 34.169, p = 0.000). Multivariate analysis was performed for the same set of patients with complete clinical data sets. Cox regression model included both significant variables, pN and tumour classification. The enter method showed significant improvement (p < 0.05) if both parameters were entered into the model (Table 2).

Associations between clinicopathological parameters and genotypes of SNPs are presented in Table 3. Statistical analysis revealed a weakly significant association for rs151658 genotypes C/G and risk of developing diffuse type of gastric cancer, and genotype G/G and risk of developing intestinal type of cancer (p = 0.049). Similar results were obtained for both male and female populations of patients (p = 0.047 and p = 0.024, respectively). Genotype A/A of rs1801376 polymorphism was significantly associated with higher risk of developing the diffuse type of gastric cancer in total and female populations of patients (p = 0.007). Interestingly, A/G genotype was under-represented in populations with diffuse type of gastric cancer. Genotype A/A of this polymorphism was also associated with the invasion of tumour cells into subserosa layer of stomach in male population (p = 0.009). A/A genotype was also associated with tumour location, namely, A/A frequencies were increased in patients with tumours disseminated across the whole stomach (p = 0.035).

Genotype T/T of rs1031963 was associated with well differentiated tumours in total population (p = 0.035); however, when we stratified it into female and male populations, we observed a significant association of this genotype with moderately differentiated tumours in the female population (p = 0.004). Clinico-pathological features lymph node involvement (pN), depth of invasion (pT), vascular invasion, perineural invasion and lymphatic invasion did not show significant associations with investigated polymorphisms.

The analyses of genotype frequencies in selected SNPs between cases and controls are shown in Tables 4 and 5. The frequencies of all genotypes in cases and control groups were in Hardy-Weinberg equilibrium.

The tested polymorphisms did not show significant differences between gastric cancer patients and control group. In contrast, when we stratified the population for gender, we found significant association between BUB1B rs1801376 genotypes and higher risk for developing gastric tumours (p = 0.029). Similarly, dominant model combining genotypes A/G and G/G showed comparable results (p = 0.010; p [Yates correction] = 0.017). Furthermore, tests for association showed analogous results and confirmed significantly higher frequency of G allele in female population of patients with gastric cancer (0.41 vs. 0.28 in control group). We also observed allele frequency difference in male patient population for BUB1B rs1031963. The dominant model, combining genotypes TT+CT versus CC, showed that patients with C/C homozygous allele had significantly higher risk for developing gastric cancer.

Comparisons of the SNPs’ genotype frequencies between our test groups and European populations are presented in Figure 3. Genotype frequencies of rs151658, rs1031963 and rs1801376 in our groups of populations showed significant differences from European population (Table 6). The frequency of rs151658 C/C genotype was higher than expected in the Slovenian population of patients compared to total European population (p = 0.015). Similarly, we observed more rs1031963 C/C genotypes in the male population of Slovenian patients with gastric cancer (p = 0.042) compared with total European population and European population stratified for males. The rs1801376 A/G genotype was higher and A/A genotype was under-represented in female population of patients with gastric cancer compared to the total European population (p = 0.034) and female European population (p = 0.014).

Immunobloting data on individual samples (Figure 1A) demonstrated statistical significance for the increased abundance of TTK (p = 0.03) in the tumour tissues. No statistically significant correlation of TTK abundance with clinical his topathological parameters or rs151658 genotypes was observed. However, some trends were observed (Figures 1B and C) for lymph node involvement (pN) and antral tumour location: TTK abundance was higher in normal tissues compared to tumour tissues when no regional nodes were invaded with tumour cells (pN = 0) and when the tumours were located at the bottom of the stomach (antrum).

Figure 1

To determine if different intronic polymorphisms could affect binding of transcription factors, we performed in silico analysis of conserved human motifs using polymorphic sequences as templates (Figure 2). We identified distinct recognition sites for different proteins for both TTK rs151658 and BUB1B rs1031963. TFII-I, c-Myb, NFI/CTF, and HNF-4 alpha binding motifs were recognized if rs151658 polymorphic site contained C allele. In contrast, if G allele was present, GR, TFII-I, NCFI/CTF, and HNF-4 alpha were identified. Comparison of rs1031963 alleles showed that if allele T was present, several binding motifs were predicted, whereas in the case of allele C, there were no recognizable binding patterns.

Figure 2

Figure 3