Association of TNFAIP8 gene polymorphisms with endometrial cancer in northern Chinese women

This case–control study included 226 cases with EC and 248 cancer-free control samples. All subjects were ethnically homogenous Chinese and resided in the Heilongjiang province of China. Cases primarily diagnosed with EC were treated with hysterectomy, bilateral salpingooophorectomy, pelvic and/or paraaortic lymphadenectomy, partial omentectomy and peritoneal washing for cytology at the Department of Gynecology, Harbin Medical University Cancer Hospital. All the patients did not have radiotherapy or chemotherapy history before surgical operations. The participants were genetically not related in three generations. After providing informed consent, each participant was interviewed to collect detailed information on demographic characteristics and provided 5 mL venous blood from September 2015 to February 2017. This study was approved by the Medical Ethics Committee of Harbin Medical University, Harbin, China.

We performed a combined analysis of functional significance and Tag SNP strategies to select three potential functional SNPs of the TNFAIP8 gene from the dbSNP and HapMap databases. The three SNPs were rs11064, rs1045241, and rs1045242, respectively. Genomic DNA was obtained from the whole blood, and was isolated from EDTA anti-coagulated whole blood using the AxyPrep Blood Genomic DNA Miniprep Kit (Axygen Biotechnology, Union City, CA, USA). The SNaPshot SNP assay was carried out to detect the dimorphism at the seven SNP loci. The resulting data were analyzed with GeneMapperTM 4.0 Software (Applied Biosystems, Foster City, CA, USA). To ensure quality-control, genotyping was done without knowledge of case/control status of the subjects, and a 5% random sample of cases and controls was genotyped twice by different persons; the reproducibility was 100%.

All 226 primary EC tissues were stained by IHC. IHC staining for TNFAIP8, estrogen receptor (ER), progesterone receptor (PR), P53 and Ki67 were performed using the Two-Step IHC Detection Reagent (PV-6001) kit (Zhong Shan Golden Bridge Biological Technology Inc., Beijing, China). The antibodies dilutions and sources were as follows: rabbit polyclonal antibodies for TNFAIP8 (1:100; Abcam), monoclonal antibodies for ER (1:100; Ventana), PR (1:70; Dako), P53 (1:600; Dako) and Ki67 (1:250; Dako). TNFAIP8 status was scored as our previous research [7]. ER and PR status were scored by the current American Society of Clinical Oncology (ASCO)/College of American Pathologists (CAP) guidelines [19]. All the samples were considered to be positive for ER or PR when at least 1% of the tumor cell nuclei were stained. The samples with nuclear staining in at least 10% of tumor cells were considered positive for P53 [20]. And the positivity threshold for Ki67 was more than 14% of tumor cells with stained nuclei [21].

The genotype frequencies were tested for Hardy–Weinberg equilibrium using the Chi-square test among the controls. Differences between cases and controls in demographic characteristics were evaluated by the Chi-square test or Fisher’s text (for categorical variables) or Student’s t-test (for continuous variables). The association between TNFAIP8 gene polymorphisms and protein expression was evaluated by the Chi-square test or Fisher’s text. Associations between genotypes and EC risk were estimated by computing odds ratios (ORs) and 95% confidence intervals (CIs) from logistic regression with adjustment for age, smoking history, BMI, and menopausal status.

The 226 cases and 248 controls were similar with regard to age at interview, parity, diabetes and smoking history (Table 1). However, there were significant differences between cases and controls in the BMI (P < 0.001), age at menarche (P = 0.003), menopausal status (P < 0.001), and hypertension (P < 0.001).

The allele and genotype distributions for all SNPs in cases and controls were shown in Table 2. The observed genotype frequencies of three SNPs followed Hardy–Weinberg equilibrium among the controls (P > 0.05 for all three SNPs).

The results showed that the minor allele of rs11064 in the TNFAIP8 gene was strongly associated with EC in patients (crude OR: 1.593, 95% CI 1.153–2.201, P = 0.005 and adjust OR: 1.719, 95% CI 1.180–2.506, P = 0.005). There was no association between the minor allele of rs1045242 in the TNFAIP8 gene and EC risk (crude OR: 1.353, 95% CI 0.968–1.892, P = 0.077). After adjusting for age, BMI, menopausal status, and smoking history, however, we observed that the minor allele of rs1045242 in the TNFAIP8 gene related significantly with EC risk (adjust OR: 1.636, 95% CI 1.107–2.417, P = 0.014).

We further analyzed the effect of the genotypes of these SNPs under three different genetic models. In the logistic regression models, compared with AA genotype of rs11064, GG genotypes was associated with an increased risk of EC (adjusted OR = 2.582, 95% CI 1.034–6.445, P = 0.042). This SNP was also related with an increased risk of EC under a dominant model (GG + AG vs. AA, adjust OR = 1.778, 95% CI 1.134–2.782, P = 0.012). Compared with the rs1045242 AA genotype, the AG and GG genotype possibly conferred increased risk for EC in the dominant model (adjust OR = 1.651, 95% CI 1.049–2.599, P = 0.030). However, no significant association with EC risk was observed for rs1045241 in the TNFAIP8 gene.

The results of stratified analyses are shown in Additional file 1: Table S1, Additional file 2: Table S2, Additional file 3: Table S3, and Additional file 4: Table S4. For the patients whose age was greater than 54, in the dominant model, combined genotypes (AG + GG) of rs11064 had a 1.899-fold increase EC risk compared with the genotype AA (OR = 1.899, 95% CI 1.051–3.430, P = 0.034). However, for patients whose age was no more than 54, we did not observe the association between genotypes and EC risk. For patients who have no smoking history, the minor allele of rs1045242 significantly increased EC risk under co-dominant and dominant models (P < 0.05). Furthermore, for patients whose BMI was greater than 25, in the dominant model, combined genotypes (AG + GG) of rs11064 had a 2.358-fold increase EC risk compared with the genotype AA (OR = 2.358, 95% CI 1.133–4.906, P = 0.022). Interestingly, when evaluating menopausal status, we found that the minor allele of rs1045242 strongly increased EC risk under co-dominant and dominant models (P < 0.05) in patients who were at pre-menopausal. In addition, we observed that for patients who were at pre-menopausal, combined genotypes (AG + GG) of rs11064 had a 1.747-fold increase EC risk compared with the genotype AA (OR = 1.747, 95% CI 1.027–2.972, P = 0.040).

The TNFAIP8 protein expression in EC tissue was shown in Fig. 1, and the immunostaining were localized within the cytoplasm of tumor cells. SNP rs11064 was significantly associated with TNFAIP8 protein expression under the codominant model (P = 0.005, Table 3). Moreover, patients with genotypes AG and GG were significantly associated with increased TNFAIP8 protein expression under the dominant model (P = 0.015, Table 2).

SNP rs1045242 was correlated with TNFAIP8 protein expression under the dominant model (P = 0.020, Table 3). However, the correlation with TNFAIP8 protein expression was not found in the codominant model (P = 0.060, Table 3).

No significant association was observed between SNP rs1045241 and TNFAIP8 protein expression under the codominant model and the dominant model (P = 0.229; P = 0.099, respectively, Table 3).

The association of rs11064, rs1045241 and rs1045242 polymorphisms with clinicopathological characteristics is shown in Table 4. SNP rs11064 had a significant association with FIGO stage (P = 0.011), depth of myometrial invasion (P = 0.047), and lymph node metastasis (P = 0.048) under the codominant model. However, these correlations were not found under dominant model. For SNP rs1045242, it was found that the patients with genotypes AG and GG were more likely to have advanced FIGO stage compared to the patients with genotype AA (P = 0.041). Furthermore, we observed the patients with genotypes AG and GG were more likely to have deeper myometrial invasion (P = 0.034). Yet, these correlations were not found under codominant model. No significant associations could be found between these two SNPs and EC patients’ histologic grade, histological type, LVSI, CA-125 level, P53, ER, PR, and Ki67 expression. Additionally, in this study, there were no significant associations between rs1045241 and all disease characters.