In Supplementary table 3, the MAF of PGF rs8019391 and rs2268615, and TNFAIP2 rs710100 between the case and control groups were listed. The genotype distribution of these SNPs in controls were in accordance with the Hardy-Weinberg equilibrium (p > 0.05). The call rate for rs8019391, rs2268615 and rs710100 were 100%, 99.7% and 99.1%, respectively. The MAFs distribution of PGF rs2268615-A allele and TNFAIP2 rs710100-A allele were higher in the case group, which increased the risk of CC (rs2268615, A vs C, OR = 1.27, 95% CI = 1.03-1.58, p = 0.029; and rs710100, A vs G, OR = 1.23, 95% CI = 1.01-1.50, p = 0.043).
The results of multiple genetic model adjusted by age revealed PGF rs2268615 and TNFAIP2 rs710100 conferred to the increased CC risk (Table 2). PGF rs2268615 was associated with an increased risk of CC under heterozygote (OR = 1.39, 95% CI = 1.04-1.86, p = 0.024), dominant (OR = 1.40, 95% CI = 1.06-1.84, p = 0.018) and log-additive (OR = 1.29, 95% CI =1.03-1.61, p = 0.027) models. For rs710100 in TNFAIP2, compared with GG genotype, GA genotype (OR = 1.44, 95% CI =1.07-1.95, p = 0.018) and GA+AA genotype (OR = 1.42, 95% CI = 1.07-1.89, p = 0.016) increased 1.44-fold and 1.42-fold CC risk, respectively. Moreover, the result of the additive model also showed an increased risk of CC (rs710100, OR = 1.23, 95% CI = 1.00-1.50, p = 0.046). However, there was no significant association between PGF rs8019391 and CC susceptibility.
Age stratification displayed that PGF rs2268615 and TNFAIP2 rs710100 increased the risk of CC among women at age £ 43 years (Table 3). After calculating the ORs for the allele (OR = 1.38, p = 0.041 and OR = 1.42, p = 0.018, respectively), genotype (CA vs CC, OR = 1.55, p = 0.039; and AA vs GG, OR = 1.97, p = 0.031, respectively), dominant (OR = 1.56, p = 0.030; and OR = 1.57, 95%, p = 0.034, respectively), and log-additive (OR = 1.40, p = 0.042; and OR = 1.42, p = 0.020) genetic models, they all displayed the genetic association of PGF rs2268615 and TNFAIP2 rs710100 with CC susceptibility.
Subsequently, stratification analysis by tumor stage showed that the risk effect for PGF rs8019391 appeared to be more prominent in the subset of patients with stage III+IV (Table 4). Compared with the C allele, rs8019391 T allele was highly represented in patients with stage III–IV as compared to patients with stage I–II under the allele (OR = 2.17, p = 4.58´10-4), heterozygote (OR = 2.34, p = 0.005), homozygote (OR = 5.76, p = 0.015), dominant (OR = 2.59, p = 0.001), recessive (OR = 4.13, p = 0.045), and log-additive models (OR = 2.36, p < 0.001).
Subsequently, MDR analysis was implemented to assess the impact of SNP-SNP interaction. Association of higher order interaction with CC risk was summarized in Figure 1. The result revealed the additive effect between TNFAIP2 rs710100-GA, PGF rs2268615-CA, and PGF rs8019391-CT on conferring risk towards the susceptibility to CC. The result of dendrogram and the Fruchterman-Reingold (Figure 2)showed that PGF rs2268615, TNFAIP2 rs710100, PGF rs8019391 exhibited a strong synergy effect on CC risk. Table 5 showed that TNFAIP2 rs710100 was the best single-locus model to predict the risk of CC (testing accuracy = 0.508, CVC = 6/10, p = 0.014). The best two-locus model was the combination of PGF rs2268615 and TNFAIP2 rs710100 (testing accuracy = 0.536, CVC = 9/10, p <0.0001. The three-locus model included TNFAIP2 rs710100, PGF rs2268615, and PGF rs8019391 (testing accuracy = 0.550, CVC = 10/10, p <0.0001).