Background

Gastric cancer (GC) is the fifth most common neoplasia and the third leading cause of cancer-related death worldwide [1]. The histological heterogeneity classified GC into different subtypes as cardia carcinoma and noncardia carcinoma, which show distinct clinical, epidemiological and molecular features among [2], and seriously hinder the diagnosis and treatment [3]. Moreover, the incidence of GC is highly determined by multiple environmental factors as: microbial infections, the host genetic background (age, gender, lifestyle, dietary regime) [4, 5], alcohol, processed meat, and obesity [6]. Notably, effective diagnostic markers and targeted therapy against GC are still lacking. Currently, GC remains to be a serious fatal disease with poor prognosis throughout Asia, especially in China [7].

DAL-1 (differentially expressed in adenocarcinoma of the lung-1) gene is located on human chromosome 18p 11.3 and belongs to the 4.1 protein superfamily, which is isolated and detected in lung adenocarcinoma for the first time by Tran et al. [8]. The expression of DAL-1 is significantly reduced or lost in various tumors such as breast [9], hepatocellular [10], colon [11], and ovarian [12]. Several studies have identified that DAL-1 is significantly associated with cancer cell differentiation, lymph node metastasis, disease progression, and TNM stage [13, 14]. Our previous study first identified the loss of heterozygosity (LOH) at 18p11.3 region (DAL-1 loci) in 45 sporadic GCs, suggesting DAL-1 might be a candidate tumor suppressor gene [15]. Afterwards, we found promoter methylation-mediated down-regulation of DAL-1 in four GC cell lines and 94.6% (35 of 37) of surgically resected primary GCs. We also demonstrated that DAL-1 effectively inhibited the malignant transformation of GC cells [16], and was significantly associated with cancer progression and poor survival of GC patients [17].

Previous studies indicated that genomic variations may have great impact on the liability for GC [18]. Patients who carry the GG genotype of rs1049174 of NKG2D gene have a higher incidence of GC in Fujian Province of China [2. Polymerase chain reactions (PCRs) were performed in a 20 μl reaction solution containing 1 µl of template DNA, 0.5 µl of each primer, 0.2 μl Taq enzyme (Qiagen, Germany) and 10 μl PCR reaction buffer (Takara, Japan). The original data were collected by ABI3730XL sequencer (Applied Biosystems, USA) and analyzed by GeneMapper 4.1 software (Applied Biosystems, USA). Genoty** data were confirmed with 10% randomly-selected samples which showed 100% concordance in repeated tests.

Table 2 The primer sequence for the six SNPs
Full size table

Statistical analysis

Continuous variables with a normal distribution were described as mean ± SD and compared with the Student’s t-test [21]. Discrete variables we re described as frequency (percentage) and compared using the Chi-square (χ2) test [21]. Genotype frequencies for each polymorphism in the control group were tested by Hardy–Weinberg equilibrium using the χ2 test. Associations between the genotypic or allelic frequency and the risk of GC were estimated by odds ratios (ORs) and 95% confidence intervals (CIs). The age of the two groups at the stratified analysis was dichotomized according to the median age (58 years) of the control group [22]. Linkage disequilibrium (LD) and haplotype analyses were performed with Haploview 4.2 software (http://sourceforge.net/projects/haploview/). P values and ORs with 95% CIs were calculated using multiple regression analysis adjusted for age, gender, smoking status, pack-years, and drinking status. All statistical analyses were conducted using the SAS 9.3 software. All P values were two-sided, and P < 0.05 was considered statistically significant.

Results

Population characteristics

The data in Table 3 (at the end of the manuscript) showed the general information of all subjects of 505 GCs and 544 HCs. The mean age was 59.08 (59.08 ± 10.55 years) in GC group and 58.33 (58.33 ± 11.55 years) in the HCs. According to the 7th Edition of the American Joint Committee on Cancer (AJCC) [23], 87 cases (17.2%), 156 cases (30.9%), 128 cases (25.3%), 65 cases (12.9%) and 69 cases (13.7%) were classified as TNM I, II, III, IV and other stages, respectively. The number of subjects with a history of GC, no family history of cancer, and other cancers were 398 (78.8%), 42 (8.3%) and 65 (12.9%), respectively. There was no significant difference in age or gender between the two groups (P = 0.105 and P = 0.404), indicating no sample matching bias between groups. However, there was a significant difference in smoking status, pack-years and drinking status between the two groups (P < 0.0001).

Table 3 Clinical and demographic characteristics of cases and controls
Full size table

Distribution of the genotypic and allelic frequencies of DAL-1 polymorphisms and their association with GC susceptibility

The genotypic frequencies of six SNPs except rs7240736 of the controls were all in accordance with Hardy–Weinberg equilibrium (P > 0.05). Hence the rs7240736 was eliminated in the next analysis. The genotypic distributions of the other five SNPs among the cases and controls and their associations with GC risk are summarized in Additional file 1: Table S1. All the allelic frequencies were not significantly different between the GCs and HCs. There was no evident association between the five SNPs with GC risk in the homozygotes, heterozygotes or two genetic models (dominant genetic model and recessive genetic model) after adjusting for age, gender, smoking status, pack-years, and drinking status (P > 0.05). The above comparisons were all not statistically significant by using multiple test correction.

Haplotype analysis and GC risk

No strong LD for the selected SNPs of the DAL-1 were identified by the Haploview software (Fig. 1). One LD block was found in the DAL-1 gene, which contained five haplotypes in the block. The associations between haplotype frequencies and GC risk were shown in Additional file 1: Table S2. The most common haplotypes within this block were determined as GTTC (0.777), ACTC (0.074), GTCC (0.068), followed by GTTG (0.044), and ATTC (0.020). No apparent association between each haplotype and GC risk within this block was observed.

Fig. 1
figure 1

Linkage relationship and haplotype block in DAL-1 gene

Full size image

Stratified analysis between DAL-1 polymorphisms and GC risk

We conducted stratified analyses for all candidate SNPs according to age, gender, smoking status, pack-years, and drinking status, which were reported to have potential influences on GC occurence. As was shown in Table 4 (at the end of the manuscript), the TA + AA genotypes of rs9953490 were associated with a significantly higher GC risk in smoker than nonsmoker before adjustment by age, gender, smoking status, pack-years, and drinking status (TA + AA vs TT, OR = 2.33, 95% CI = 1.11–4.87, P = 0.025). However, this significant association was compromised after adjustment by age, gender, smoking status, pack-years, and drinking status (TA + AA vs TT, adjusted OR = 1.62, 95% CI = 0.61–4.27, P = 0.333). Meanwhile, for other SNPs, no significant association was found between genotypes and GC risk in the stratified analyses (Additional file 1: Tables S3–S7).

Table 4 Stratified analyses for the rs9953490 genotypes of DAL-1 gene in cases and controls
Full size table

We further conducted stratified analyses in 274 patients with available clinicopathological information such as family history of cancer, tumor size, neoplasia location, depth of invasion, lymph metastasis, TNM stage, and Lauren’s classification. As shown in Table 5 (at the end of the manuscript), the TA + AA genotypes of rs9953490 were significantly associated with increased GC risk in N3 compared with N0 (TA + AA vs TT adjusted OR = 4.56, 95% CI = 1.49–13.98, P = 0.008), and with a obviously increased GC risk in TNM stage III compared with stage I-II (TA + AA vs TT adjusted OR = 2.33, 95% CI = 1.16–4.67, P = 0.017). No associations were found between other SNPs and the clinicopathological characteristics of GC (Additional file 1: Tables S8–S12). The above comparisons showed no statistical significance using multiple test correction analysis.

Table 5 Association between DAL-1 (rs9953490) genotypes and clinicopathologic features of GC
Full size table

Discussion

In the present study, we investigated the associations between six SNP polymorphisms of the DAL-1 gene and GC risk in the Han population in Northeast China. Stratification analyses based on smoking revealed that the TA + AA genotypes of rs9953490 were significantly associated with a significantly higher GC risk in smoker than nonsmoker. However, this association was abolished after adjustment by age, gender, smoking status, pack-years, and drinking status.

The reason might be as the following: First, although many studies have shown that smoking can increase the risk of GC [24, 25], the relationship between genomic polymorphisms and susceptibility to smoking-related GC has not yet been defined. A Meta-analysis about CYP1A1 polymorphisms with GC susceptibility pointed out that the m1 genotypes (CC + CT) decreased the susceptibility of GC among ever-smokers, but there wasn’t any association between the m2 genotypes (GG + AG) and GC risk among the smokers [26]. Second, the mechanism by which tobacco smoke facilitates cancer development is not completely elucidated. Moreover, most smoke-derived procarcinogens and their bioactivation are not organ-specific. When all tobacco-related cancers are considered together, smoking tends to increase the risk for develo** cancers which are cancer-prone [24, 26, 27]. Third, the reason may also be due to the smaller sample size caused by subgrou** in stratified analysis, and the great discrepancy of sample number between subgroups. Last but not the least, GC is a heterogeneous disease involved with multiple etiological factors including age, gender, geography, lifestyle, dietary regime [4, 5, 28]. To confirm the results of this study, larger sample size will facilitate the statistic power so as to rule out the possibility of population stratification and “observation bias” [26] in future study.

Interesting, we found that the TA + AA genotypes of the rs9953490 were associated with the significant increased GC risk in N3 and TNM stage III, representing its possible correlation with lymph node metastases and poor prognosis of GC. We speculated that the TA + AA genotype, located in the 3’UTR region of the DAL-1 gene, might be able to regulate DAL-1 expression or its promoter methylation status, which may inhibit its original tumor suppressive function. In this case, the association of DAL-1 gene polymorphism with GC would further consolidate our previous study, which has demonstrated that DAL-1 gene is involved in anti-proliferation, inhibition of metastasis, associated with poor survival in GC [15,16,17].

Altogether, we have identified the association between the DAL-1 gene polymorphisms and GC susceptibility. The study has provided new evidence for DAL-1 gene as a potential target in GC clinical practice.

Conclusion

This study demonstrated that the rs9953490 TA + AA genotypes of DAL-1 gene is significantly associated with the occurrence and development GC in the Han population of Northeast China.