Associations between SLC39A6 rs1050631 genotype and clinico-demographic characteristics. Among the 512 tissue samples from patients resected for GA, 3.9% (20 cases), 27.1% (139 cases), and 69.0% (353 cases) had the TT, CT, or CC genotype, respectively (Table 1). Genotype frequencies did not differ significantly (based on the chi-squared test) with age (<60 vs. ≥60 years), degree of differentiation, gross findings (apophysis vs. invasion), Lauren classification (intestinal type, diffuse type, uncertain type), tumor location (upper third, middle third, lower third), tumor size (≥5 vs. <5 cm), distant metastasis status or chemotherapy status (Table 1). Potential association was observed between genotype and lymph node metastasis status (p <0.2). Genotype showed a significant association with post-resection survival (p = 0.002) and recurrence (p = 0.004) (Table 1).
Associations between SLC39A6 rs1050631 genotype and GA recurrence. Of all 512 patients, 334 (65.2%) experienced recurrence. Of the 353 patients with the CC genotype, 214 (60.6%) experienced recurrence (Tables 1 and 2). A significantly higher proportion of patients with the CT (104 of 139, 74.8%) or TT genotype experienced recurrence (16 of 20, 80.0%).
Univariable analysis showed risk of recurrence to be markedly higher in patients with the CT genotype (HR 1.813, p = 0.022) or the CT+TT genotype (HR 1.440, p = 0.001) (Table 2). Similar results were obtained with multivariable analysis: CT genotype, HR 1.379, p = 0.007; CT+TT genotype, HR 1.387, p = 0.004. In order to isolate factors that may influence post-resection outcomes in patients with GA, we stratified our patients by sex, age, tumor size, histologic grade,postoperative chemotherapy status and lymph node metastasis.Multivariable analysis showed that patients with the CT genotype were at increased recurrence risk if they were male (HR 1.564, p = 0.001), aged ≥60 years (HR 1.512, p = 0.009), with a tumor ≥5 cm (HR 1.807, p = 0.001) or with a moderately differentiated tumor (HR 1.830, p < 0.001) than those patients with CC genotype had (Table S1). Similar results were obtained for patients with the or CT+TT genotype: male, HR 1.545, p = 0.001; age, HR 1.529, p = 0.005; tumor ≥ 5 cm, HR 1.789, p = 0.001; and moderately differentiated tumor, HR 1.780, p < 0.001 (Table S1). In terms of the associations among the genotypes and recurrence risk involved chemotherapy status, Multivariable analysis exhibited that patients with CT, TT or CT+TT had higher recurrence risk only appeared in the group of patients without performing postoperative chemotherapy (Table S1).
Kaplan–Meier and log-rank analyses showed that median recurrence-free survival time was only 20 months in patients with the CT+TT genotype, significantly shorter than the 36 months in patients with the CC genotype (p = 0.001, Fig. 1A). Stratification analyses based on sex, age, differentiation grade, tumor size, histologic grade and postoperative chemotherapy status showed that patients with the CT+TT genotype who were male, aged ≥60 years or who had a tumor ≥ 5 cm or had a moderately differentiated tumor or no matter patients who whether performed postoperative chemotherapy had significantly shorter median recurrence-free survival time than patients with the CC genotype had (sex: 18 vs. 38 months, p < 0.001; age, 13 vs. 35 months, p = 0.001; tumor size, 9 vs. 35 months, p < 0.001; moderately differentiated tumor, 22 vs. 56 months, p < 0.001; had performed postoperative chemotherapy: 23 vs. 49months, p=0.031; or had not perform postoperative chemotherapy: 15 vs. 24 months, p=0.004. Fig. 1B, 1C and Fig. 2).
Associations between SLC39A6 rs1050631 genotype and overall survival. In the entire cohort of 512 patients, 330 (64.5%) died, similar to the proportion of patients with the CC genotype who died (210 of 353, 59.5%; p = 0.125). A significantly higher rate of death occurred among patients with the CT (104 of 139, 74.8%) or TT genotype (16 of 20, 80%), based on the chi-squared test and chi-square partitioning (Table 1).
Univariable Cox analysis revealed markedly increased risk of death in patients with the CT genotype (HR 1.440, p = 0.002) and also in patients with the CT+TT genotype (HR 1.492, p < 0.001). Similarly, multivariable Cox analysis revealed markedly increased risk of death in patients with the CT genotype (HR 1.416, p = 0.004) or the CT+TT genotype (HR 1.429, p=0.002), after adjusting for age, sex, and lymph node metastasis status (Table 2). Stratification by sex showed that risk of death was significantly higher among male patients with the CT genotype (HR 1.601, p = 0.001) or CT+TT genotype (HR 1.586, p < 0.001) than among the patients with CC genotype (Table S2). Similarly, stratification by age, tumor size, differentiation grade, postoperative chemotherapy status or lymph node metastasis revealed significantly increased risk of death among patients with the CT or CT+TT genotype when they were aged ≥60 years (HR 1.531, p = 0.007; HR 1.542, p = 0.004) or had a tumor ≥ 5 cm (HR 1.928, p < 0.001; HR 1.904, p < 0.001) or a moderately differentiated GA (HR 1.735, p = 0.001; HR 1.706, p = 0.001) (Table S2). In terms of the associations among the genotypes and death risk involved chemotherapy status, no matter patients had or had not performed postoperative chemotherapy, who carried CT or CT+TT genotypes showed higher death risk than those carried CC genotype (Table S2).
Patients with the CC genotype showed median overall survival time of 43 months, compared to 27 months for patients with the CT+TT genotype (p < 0.001; Fig. 3A). Stratification analyses showed that survival time was significantly shorter in patients with the CT+TT genotype if they had any of the following characteristics: male (26 vs. 45 months, p < 0.001), aged ≥60 years (20 vs. 42.0 months, p = 0.001), had a tumor ≥ 5 cm (16 vs. 42 months, p < 0.001), had moderately differentiated GA (31 vs. 59 months, p < 0.001) or no matter patients who whether performed postoperative chemotherapy: had performed postoperative chemotherapy: 31 vs. 57 months, p=0.05, which almost equal to have statistically significant difference ; or had not perform postoperative chemotherapy: 22 vs. 28 months, p=0.012.(Fig. 3B, 3C and Fig. 4).
SLC39A6 overexpression in GA. The results of the above mentioned experiments suggest an association between SLC39A6 rs1050631 and post-resection outcomes in patients with GA. This raised the question of whether SLC39A6 expression might be associated with GA. We used quantitative PCR and immunohistochemistry to assess expression levels in four GA cell lines (AGS, BGC-823, SGC-7901 and MGC-803) and our cohort of GA tissues. We found that the protein was overexpressed in all four cell lines relative to the normal gastric cell line GES-1, and that the protein was present (immunopositivity scores of ≥1+) in 150 of 198 (75.76%) GA tissues, compared to only 48 of 83 (57.83%) non-cancerous gastric tissues (p = 0.003; Fig.5A and 5B).
SLC39A6 knockdown inhibits proliferation, migration and invasion of GA cells. We designed two short interfering siRNAs to inhibit SLC39A6 expression in the GA cell lines BGC-823 and SGC-7901 (Fig. 6A and 6B), and this knockdown led to significantly less proliferation in both lines than in untransfected cells (Mock) or cells transfected with scrambled control siRNA (Fig. 6C and 6D). Knockdown also significantly reduced migration (Fig. 7A and 7B) and invasion by both cell lines (Fig. 7C and 7D). These results suggest that SLC39A6 may contribute to GA by functioning as a typical oncogene.
Association of different genotype at SLC39A6 rs1050631 with SLC39A6 expression in GA. Immunohistochemistry of 198 GA tissues from our patient cohort detected high SLC39A6 protein expression (≥2+) in 86 tissues (43.43%; Fig. 8); high expression was detected in 6 of 14 (42.86%) patients with TT, 50 of 92 (54.35%) patients with CT, 56 of 106 (52.83%) patients with CT+TT, and 31 of 92 (33.70%) patients with CC genotype. Though the high positive expression rate among the three different genotypes groups showed no statistically significantly different from that in the whole group (p>0.05) , the rate of high SLC39A6 expression was significantly greater among patients with CT genotype than among those with CC genotype (p=0.005) based on chi-square partitioning (Table 3). We speculate that the SLC39A6 rs1053631 genotype is associated with post-resection prognosis of patients with GA because the CT genotype leads to higher SLC39A6 expression.