The relationship between TLR2 polymorphisms and the progresses of tumors have been explored and reported by some researchers. Boraska Jelavić et al.  reported an association of TLR2 GT microsatelite alleles with 20 and 21 GT repeats with sporadic colorectal cancer among Croatians. In the study by Srivastava et al. , del allele carriers of TLR2 (Delta22) polymorphism were associated with a 1.54-fold increased risk for gallbladder cancer. In another study by Pandey et al. , TLR 2 gene polymorphisms (-196 to -174 del) showed significant association (OR 1.6, 95% CI 1.00-2.51) with cervical cancer susceptibility. Similarly, it was also confirmed in the study by Nischalke et al.  that the frequency of the TLR2 -196 to -174del allele was significantly higher in cases with HCV-associated HCC than in HCV-infected patients without HCC. And in carriers of the TLR2 -196 to -174del allele stimulation of monocytes resulted in significantly lower TLR2 expression levels and IL-8 induction than in individuals with the TLR2 -196 to -174ins/ins genotype. The current study was the first one to show that inherited variation in TLR2 influences the risk of HCC. We observed the positive association between two SNPs in exon (rs3804099 and rs3804100) and the risk of HCC. The risk of HCC was found to be significantly low in individuals carrying the heterozygous genotypes of these two SNPs by comparing with those carrying wild-type homozygous genotypes (OR, from 0.331 to 0.759, P < 0.001). Our data suggested that TLR2 gene variation may play an important protective role in the occurrence of HCC. Though the observed two-fold decrease in risk is modest, our finding is intriguing because genes in multiple pathways alter the risk for HCC, and each individual gene is likely to contribute only a modest risk.
rs3804099 and rs3804100 of TLR2 are synonymous SNPs. It has long been assumed that synonymous SNPs are inconsequential, as they do not lead to a change in primary polypeptide sequence. However, over the last decade, it has been confirmed in many studies that synonymous mutations are implicated in diseases. Several mechanisms by which synonymous mutations exert their impact on gene function are: (1) Perturbations of mRNA splicing. Ectopic mRNA splicing generated by synonymous SNPs can bring about distinct phenotypes, leading to human disease . Synonymous base pair changes in auxiliary elements within exons (exonic splicing enhancers and exonic splicing silencers) can directly change the splicing patterns of mRNA transcripts, or they can alter the penetrance of concurring mutations elsewhere in the gene [22, 23]. (2) The stability of mRNA. Synonymous SNPs can affect the stability of mRNA via cis factors which determine mRNA stability fall within the 3'-untranslated region of a transcript . (3) mRNA structure. Synonymous SNPs can modulate mRNA structure and have downstream effects on protein expression and phenotype . (4) Protein folding. It has been confirmed by some computational and experimental studies [26, 27] that synonymous SNPs have an influence on protein folding and ultimately protein function. Which mechanism by these synonymous SNPs impact should be investigated in future research.
In the present study, LD analysis showed that these two TLR2 synonymous SNPs were in high LD and located in one haplotype block. Individuals carrying haplotype TT was significantly associated with the decrease of the risk of HCC (OR 0.524, 95% CI 0.394 - 0.697, P = 0.000). Inversely, the risk of HCC significantly increased in patients carrying haplotype CC (OR 2.743, 95% CI 1.915 - 3.930, P = 0.000). In a recent study by Diatchenko et al. , it was shown that the low pain sensitivity (LPS) haplotype composed of synonymous SNPs produces much higher levels of the cathechol-O-methyltransferase (COMT) enzymatic activity when compared with the average pain sensitivity (APS) or high pain sensitivity (HPS) haplotypes. The variation in COMT expression levels among haplotypes was due to the differences in protein translation efficiency, which marks the final manner in which synonymous SNPs can exert their influence on protein translation and cotranslational protein folding. Hunt et al.  suggested that haplotypes composed of synonymous SNPs can have profound effects on gene function, and in some cases their effects can be stronger than those of their non-synonymous counterparts.
Few investigations have been reported about the relationship between TLR9 polymorphisms and the tumor susceptivity. In the study by Ashton et al. , haplotype analysis showed that the combination of the variant alleles of the two TLR9 polymorphisms, rs5743836 and rs187084, were protective in endometrial cancer (OR 0.11, 95% CI 0.03-0.44). In a case-control study by Etokebe et al. , TLR9 (c.1635A > G) polymorphisms were not likely to be a risk factor in developing breast cancer. Although Hold et al.  found TLR9 (TLR9-1237 T/C, rs574383) was associated with the development of premalignant gastric lesions in 2006, but their most recent study showed that this SNP didn't increase the risk of gastric cancer itself . Similarly, we didn't find any relationship between two TLR9 genetic variations and the occurrence of HCC.
Recently, Zhang and colleagues  conducted a genome-wide association study by genotyping 440,794 SNPs in 355 chronic HBV carriers with HCC and 360 chronic HBV carriers without HCC in Chinese population using Affymetrix Genome-Wide Human SNP Array 5.0. In this study, one intronic SNP (rs17401966) in KIF1B on chromosome 1p36.22 was highly associated with HBV-related HCC. However, only rs11938228 and rs10759930 of TLR2 and rs1927911 of TLR9 were enrolled in Affymetrix Genome-Wide Human SNP Array 5.0 chip and these three SNPs were not related to HCC. In our study, only TLR2 rs11938228 was determined and our result was consistent to them, that TLR2 rs11938228 was not associated with the risk of HCC.
Although SNP rs3804099 and rs3804100 were out of HWE (P = 0.01 - 0.02), we retained them in the analyses as these P values were marginal and may be chance findings and the internal blinded quality-control specimens did not show evidence of genotyping error. These two SNPs genotyped were deviated HWE in the controls but not in the cases. The association for these two polymorphisms showed a FPRP below 0.200. It suggested strongly that these associations are unlikely to represent a false-positive result. Considering these two SNPs showed any evidence of deviation in cases, it was not believed that the marginal HWE tests in controls suggest systematic genotyping errors.