The study population consisted of 479 patients with primary GC undergoing total gastrectomy and 458 cancer-free individuals. The 479 patients with GC were recruited from Fujian Cancer Hospital (Fujian, China) between January 2007 and December 2009. All diagnoses were histopathologically confirmed as primary gastric cancer. The exclusion criteria were as follows: i) Patients who had previously been diagnosed with cancer; ii) any metastasis; and iii) patients who had received radiotherapy, chemotherapy or immunotherapy prior to surgery. The 458 cancer-free controls were randomly selected from a pool of healthy volunteers following a routine medical check during the same period at the Fujian cancer hospital; all volunteers had no history of personal or familial malignancy or other serious diseases. All participants provided written informed consent to participate in the study and the study was approved by the Ethics Committee of Fujian Cancer Hospital.

The DNA of the patients with GC was isolated from paraffin-embedded normal stomach tissue adjacent to the tumor site (distance, >5 cm) using a QIAamp DNA FFPE Tissue kit (Qiagen GmbH, Hilden, Germany). The DNA of cancer-free individuals was extracted from venous blood samples (2 ml) using a Whole Blood Genome DNA Isolation kit (BioTeke Corporation, Beijing, China). The concentration of DNA was determined using a NanoDrop-1000 ultraviolet spectrophotometer, with a A260/A280 ratio between 1.8 and 2.0.

Genotyping analyses for SNPs were performed using Matrix-Assisted Laser Desorption/Ionization Time of Flight Mass Spectrometry (MALDI-TOF MS) as previously described (23). Briefly, polymerase chain reaction (PCR) was conducted to amplify the target sequence using HotStarTaq DNA Polymerase (Qiagen GmbH, Hilden, Germany). Primer sequences used for rs4644 detection were as follows: 1st-PCR primer, ACG TTG GAT GTC CGG GAT AAG CTC CAG GT; 2nd-PCR primer, ACG TTG GAT GGC TTA TCC TGG ACA GGC AC; UEP_SEQ reverse primer, CTC CAG GCG CCT ACC. For rs4652 detection, 1st-PCR primer, ACG TTG GAT GAA GGA ATG CCA TCT CAC CAG; 2nd-PCR primer, ACG TTG GAT GCT ACC CAT CTT CTG GAC AGC; UEP_SEQ reverse primer, GGA CAG CCA AGT GCC.

The excess dNTPs were removed and base extension reactions were conducted using ThermoSequenase (GE Healthcare, Chicago, IL, USA). Extended products were treated with SpectroCLEAN resin (Sequenom, San Diego, CA, USA) to remove salts. Subsequently, 10 µl of the reaction was dispensed into a SpectroCHIP microarray (Sequenom). A MALDI-TOF mass spectrometer was used to acquire the SNP profile data, which was analyzed by TYPER 4.0 software (Sequenom).

A total of 3 control types were included: Negative control, internal standard control and repeat control. The number of negative controls and internal standard controls was 1% of the total sample number, whereas the number of repeat controls was 5% of the total sample number. For the negative control, sample DNA was replaced with water to monitor the contamination of product in the experiment. ‘Chinese number one’ cell line DNA was used as the internal standard control template to detect the primer design reliability of each SNP site. The ‘Chinese number one’ cell line is cultivated by Beijing Genomics Institute (Shenzhen, China) and the SNP site data can be accessed from the ‘Chinese number one’ database (24). For the repeat control, randomly selected samples were used as repeat controls to validate the stability of the experiment.

Immunohistochemical analyses were performed for 207 samples using the BenchMark XT platform (Ventana Medical Systems, Tucson, AZ, USA). Primary antibodies included P53 (Kit-0010-2, clone: DO-7), Ki67 (Kit-0005-2, clone: MIB-1), glutathione S-transferase (GST; MAB-0583, clone: LW29), P-glycoprotein (Pgp; MAB-0237, clone: C494), topoisomerase II (TOPII; MAB-0588, clone: 3F6) and thymidylate synthase (TS; MAB-0171, clone: TS106), (all ready-to-use dilution; Fuzhou Maixin Biotech, Fuzhou, Fujian, China). Each slide was subjected to heat treatment in CC1 retrieval solution (Ventana Medical Systems, Tucson, AZ, USA) for 30 min and subsequently incubated with all six antibodies at 37°C for 30 min. Immune complexes were detected with the use of an UltraView 3′3-diaminobenzidine detection kit (Ventana Medical Systems, Tucson, AZ, USA) according to the manufacturer's protocol.

The sections were scored semi-quantitatively, using light microscopy, by two pathologists without any knowledge of the clinical history of the patients. The intensity of staining was evaluated according to the following scale: 0, no staining; 1, weak staining; 2, moderate staining; and 3, strong staining. The extent of staining was scored as follows: 0, <10%; 1, ≥10% and <25%; 2, ≥25% and <50%; 3, ≥50% and <75%; 4, ≥75%. The final results were the product of the intensity and extent of staining. The grading system was divided into four parts according to the product of the intensity and the extent of staining: 0–3, negative; 4–6, +; 7–9, 2+; and 10–12, 3+.

Univariate analysis was performed with a χ² test. The computation of odds ratio (OR) and 95% confidence intervals (95% CI) were conducted using SPSS software (version 17.0; SPSS, Inc., Chicago, IL, USA). The tests included an evaluation of the association between genotypic distributions and the development of GC, clinicopathological features (including tumor size, depth of invasion, differentiation and the status of the lymph node metastasis) and the protein expression levels from the immunohistochemistry data. Haplotypes were estimated using the Expectation Maximization algorithm (25). Haplotype association with GC was determined using SHEsis, as previously described (26).

The fundamental background data for patient cases and patient controls are summarized in Table I, and demonstrate they were adequately matched for age and sex. In the control patients, the frequency of the genotypes of the two polymorphisms were determined using the Hardy-Weinberg test, which revealed that the P-values for rs4644 and rs4652 were 0.81 and 0.50, respectively. These results demonstrated that the genotype distribution conforms to a Hardy-Weinberg equilibrium.

The present study was based on a large number of clinical samples. A total of 10 negative controls were used and analyzed to validate that the results were not false positives. The data of the 10 internal standard controls were all matched in the ‘Chinese number one’ database. In addition, the conformity of 47 repeat controls was >98%. Therefore, the results of the genotyping were deemed reliable.

The genotyping results of 479 patients and 458 controls are presented in Table II, and the MALDI-TOF MS spectra are presented in Fig. 1. Compared with the controls, the patients' rs4652 allelic genotype and its genotypic distribution frequency differed significantly; the rs4652 CA/AA carriers exhibited a significant increased risk of GC (adjusted-OR, 1.51; 95% CI, 1.05–2.18; adjusted-P=0.03), which indicated that the patients with rs4652 CA/AA were more likely to develop GC. There was no significant difference identified in the genotypic distribution of rs4644 between patients and controls, indicating that this polymorphism was not associated with the development of GC.

The present study aimed at identifying associations between the genotypic distributions of galectin-3 SNPs and the overall clinicopathological features of the disease, including tumor size, depth of invasion, differentiation and the status of lymph node metastasis. No associations were determined between rs4644 and rs4652 genotypic distribution with the overall clinicopathological features of patients with GC.

Samples from 207 patients with GC were analyzed by immunohistochemistry. Representative hematoxylin and eosin staining images of GC is presented in Fig. 2. The Pgp protein expression in the tumor tissues was predominantly homogenous and was primarily identified in the cytoplasm or at the cell membrane. With the exclusion of 4 patients who failed SNPs genotyping analyses due to low sample quality, the Pgp protein expression in the remaining 203 patients was as follows: Negative, 41.9% (85/203); +, 15.8% (32/203); 2+, 35.0% (71/203); 3+, 7.4% (15/203; Fig. 3). The association between the rs4652 genotypic distribution and Pgp protein expression is presented in Table III. The results indicated that there were significant associations between rs4652 genotypic distributions and Pgp expression level (χ²=9.063; P=0.028), although the association was not linear. There was no significant association identified between the genotypic distribution of galectin-3 SNPs and other proteins including P53, Ki67, GST, Pgp, TOPII and TS.

The haplotype of galectin-3 rs4644 and rs4652 were analyzed using SHEsis software and three haplotypes were identified in patients and controls. No association was revealed between the galectin-3 haplotype and the risk of gastric cancer.