Molecular Epidemiology Study On The Associations Of Genetic Polymorphisms In TGFβ Signaling Pathway With Colorectal Cancer Development

Background:Colorectal cancer (CRC) is the second common cancer and the fourth leading cause of cancer-related death in the world. In China, there have been huge changes in the incidence and mortality of CRC, which has been one of the cancers with most rapidly in increasing incidence and mortality during the past decades. The rapid increase in incidence and mortality of CRC highlight the importance of prevention against this disease. Converging epidemiological studies have revealed several environmental factors, including obesity, diet, and smoking, were involved in the development of CRC. However, only a fraction of exposed individuals actually develop CRC during their lifespan, indicating that genetic susceptibility factors may also play importance roles in this cancer. Recently, the twelve low-penetrance variants contributing to CRC risk identified by multiple genome-wide association (GWA) studies yield the potential to lead to novel insight into the genetic etiology of CRC. Six of the twelve variants tag linkage disequilibrium blocks that include or are near to the genes of transforming growth factor-beta (TGFβ) signaling pathway. Furthermore, TGFβpathway has been previously implicated in tumor biology given their pivotal role in regulation of extensive cell processes, including cell proliferation, differentiation, migration, and apoptosis. Therefore, a hypothesis relied on the overrepresentation of TGFP components in GWAS that the genetic polymorphisms altering the expression or function of individual component of this pathway contributed to CRC development. Furthermore, since CRC is complex trait involving multiple genetic and environmental factors and their interactions, another hypothesis regarding TGFP and CRC is that complex gene-gene or gene-environment interaction have more important role than the single polymorphism in development of CRC.Objectives:1. To explore the association of the genetic polymorphisms in TGFP signaling pathway with the susceptibility to colorectal cancer.2. To analysis the association between the genetic polymorphisms in TGFβsignaling pathway and the risk of colorectal cancer progression.3. To explore the potential gene-gene or gene-environment interactions in TGFβsignaling pathway in relationship to colorectal cancer development.Methods:We applied a two-stage analysis strategy and conducted two case-control studies in Wuhan and Beijing population independently to investigated the association of genetic polymorphisms in TGFβsignaling pathway with CRC genetic susceptibility, and assessed the role of candidate polymorphisms in metastasis of CRC in Wuhan CRC patients. Candidate polymorphisms were genotyped by Sequenom MassARRAY and Taqman OpenArray methods,χ2 test or t test were used to examined differences between cases and controls in distribution of sex, age, smoking, drinking, and genotypes. The Cochran-Armitage trend test was used to examine association between CRC and SNP genotype. False discovery rate (FDR) method was applied for multiple comparison correction. Additionally, the CRC risk associated with genotype was estimated as odds ratios (ORs) and 95% confidence interval (95% CI) calculated by logistic regression under dominant, recessive and additive genetic models after adjusted for sex, age, smoking, and drinking. We also applied a comprehensive strategy to systematically investigate gene-gene or gene-environment interaction by combining logistic regression (LR) analysis, multifactor dimensionality reduction (MDR), and classification and regression tree (CART) approaches both in two stages. Results:1. Smoking was shown to be risk factor for colorectal cancer development, with an OR of 1.83 (95% CI=1.40-2.38) for smokers compared with non-smokers in combined analysis of two stages.2. Of the 77 SNPs analyzed in stage 1 study,5 SNPs, including rs10988706, rs6478972, rs11129420, rs948588, and rs11874392 were significantly association with CRC risk (Ptrend=0.025,0.039,0.047,0.015, and 3.859×10-4). Agreement with the results of stage 1 study, rs11129420 and rs11874392 in stage 2 study exhibited significant association with CRC risk (FDR-P=1.110×10-4,0.014), even after FDR adjustment for multiple comparison (Ptrend=4.211×10-6,0.001). In combined analysis across the two stage studies, the associations of rs11129420 or rs11874392 and CRC risk were highly statistically significant (FDR-Ptrend=3.434×10-5 and 7.274×10-6, respectively). The multivariate logistic regression analyses showed that the rs11129420 CT genotype had an OR of 0.50 (95% CI=0.37-0.67) compared with the CC genotype. Analysis with pooled rs11129420 CT and TT genotypes had an OR of 0.50 (95%CI=0.37-0.66) compared with the CC genotype, suggesting a dominant effect of rs11129420. Further, in additive model, the rs11129420 was associated with decreased risk of CRC in additive model (OR=0.53,95% CI=0.40-0.69). For rs11874392, the AT or AA genotype had an OR of 1.33 (95% CI=1.05-1.69) or 2.01 (95% CI=1.49-2.71) compared with the TT genotype, rs11874392 was significant associated with increased risk of CRC in dominant and additive modes (OR=1.50, 95% CI=1.20-1.88; OR=1.41; 95% CI=1.21-1.63).3.In stage 1 study, we firstly assessed pair-wise interaction between among 5 promising SNPs, smoking, and drinking by LR. Significant pair-wise multiplicative and additive gene-gene interactions were observed between rs11874392 and rs6478972 (FDR-Pmult=0.044, FDR-Padd=0.028). Additionally, rs6478972, rs11129420, and rs11874392 showed multiplicative interaction, but not additive interaction, with smoking in modulation of CRC risk (Pmult=0.017,0.013, and 3.476×10-5), whereas no interaction effect was found between these 5 SNPs and drinking. Consistently with the LR analysis, the MDR analysis in stage 1 also showed a significant interaction between rs6478972 and rs11874392, which comprised the best model in MDR with the highest testing accuracy and cross-validation consistency among all factor models (TA=0.5685, CVC=10/10, permutation P=0.004). Additionally, when environment factors were brought into MDR analysis, a significant gene-environment interaction was observed between smoking, rs6478972, and rsl 1874392 (TA=0.583, CVC=10/10, permutation P=0.001). The final pruned decision tree included rs6478972, rs11874392, rs1528734, and rs3821671, among which, rs6478972 and rs11874392 were selected as the first and second split on the tree, respectively. Conceivably, these results collectively raised the best gene-gene interaction model comprising rs6478972 and rs11874392 in the stage 1 study. Interestingly, this significant interaction model including rs6478972 and rs11874392 was validated in stage 2 study, and then was further confirmed in the combined analysis of these two stage studies. Firstly, LR analysis in stage 2 and combined study consistently showed the multiplicative and additive interactions between rs6478972 and rs11874392 (Pmult=8.817×10-5; Padd=0.006; Pmult=5.722×10-8, Padd=1.000×10-4). Secondly, the best MDR model in stage 2 study was the three-factor model including rs11129420, rs6478972, and rs11874392 (TA=0.5919, CVC=10/10, permutation P<0.001). Further MDR analysis in whole subjects showed the best interaction model included rs6478972 and rs11874392 (TA=0.5693, CVC=10/10, permutation P<0.001). Finally, in the CART in stage 2 study and the pooled study, the both pruned decision trees included rs11129420, rs11874392, and rs6478972, consistent with results from LR and MDR analyses indicated the significant interaction between rs6478972 and rs11874392. Collectively, the strong interaction existed between rs11874392 and rs6478972 in relation to CRC risk in this current study.4. Additionally, the LR, MDR, and CART consistently suggested that rs11129420 was the most significant variant in modulation of the risk of CRC progression, with an OR of 1.8 for the rs11129420 T allele versus rs11129420A allele (95% CI=1.01-3.21). The MDR analysis further suggested that rs1800469 and rs235764 might cooperatively contribute to CRC progression, by showing the best interaction model comprising rs1800469 and rs235764 (TA=0.5842, CVC=10/10, permutation P=0.015). Conclusions:1. Smoking was the main risk factor for colorectal cancer in this current study;2. rs11129420 in TGFβR2 and rs1187432 in SMAD7 were the most important genetic polymorphisms in TGFβsignaling pathway for susceptibility to colorectal cancer; additionally, rs11129420 was also invovled in progression of colorectal cancer.3. Interaction between rs6478972 in TGFβR1 and rs11874392 in SMAD7 contributed significantly to colorectal cancer incidence; furthermore, rs1800469 in TGFβ1 and rs235764 cooperatively affect progression of colorectal cancer, suggesting the gene-gene interaction among genetic polymorphisms in the TGFβsignaling pathway plays a more importance role than the single polymorphism in development of CRC.Innovations:This current study firstly revealed the significant contributions of genetic polymorphisms in TGFβsignaling pathway to development of colorectal cancer in Chinese population. Further, a comprehensive statistical strategy combining LR, MDR, and CART was applied to explored potential complex gene-gene or gene-environment interactions in modulating colorectal cancer incidence and progression.