Identification Of Gene Polymorphisms Loci Associated With Lung Cancer And Construction Of Prediction Models

Lung cancer is one of the most common types of malignant tumors in the world and it is caused by various factors. Although smoking is considered the most important factor, only10-15%of smokers finally develop into lung cancer, suggesting that individual genetic plays a significant role which should not be ignored in development of lung cancer.From research reports about gene polymorphisms associated with susceptibility of lung cancer, after a series of screening, we selected24single nucleotide polymorphisms (SNP) loci. Using SNPscan genotyping technology and Case-Control study method, we analysis susceptibility of lung cancer with tagSNPs or functional SNPs loci among1597cases and2040sex and age matched cancer-free controls in a Chinese Han population. Then we selected the significant SNPs loci associated with lung cancer and the smoking status, age, and sex as factors, and then use the stepwise backward regression for variable selection, in order to build risk prediction models for lung cancer. Interactions were tested and cumulative effects of risk factors were analyzed. We assessed models predictive performance by estimating area under the curve (AUC) based on receiver operating characteristic (ROC) curve. Meanwhile, the total data was averagely divided into separate training and validation sets, and the discriminatory ability of the model was assessed with accuracy, sensitivity and specificity. Interactions and reliability were analyzed and validated by Multifactor Dimensionality Reduction (MDR) method.The results showed that8factors including sex, smoking, and6SNPs loci (TERT-rs2736098and rs2853668, BAG6-rs2242656, MMP2-rs243865, CHEK2-rs2236141and XRCC6-s2267437) were selected by Logistic stepwise backward regression. TERT-rs2736098, BAG6-rs2242656, MMP2-rs243865, CHEK2-rs2236141and XRCC6-rs2267437were under the recessive genetic model; TERT-rs2853668was under the dominant genetic model. The mutant homozygous of TERT-rs2736098and CHEK2-rs2236141individuals significantly increased risk of lung cancer (rs2736098(TT):adjusted OR (95%CI)=1.25(1.01,1.54), P=0.038; rs2236141(TT):adjusted OR (95%CI)=2.03(1.31,3.16), P=0.002). The wild-type allele carriers on TERT-rs2853668and wild homozygous of BAG6-rs2242656, MMP2-rs243865and XRCC6-rs2267437individuals significantly increased risk of lung cancer (rs2853668(TT+GT):adjusted OR (95%CI)=1.20(1.04,1.40), P=0.017; rs2242656(AA):adjusted OR (95%CI)=1.22(1.04,1.43), P=0.017; rs243865(CC):adjusted OR (95%CI)=1.36(1.13,1.64), P=0.001; rs2267437(CC):adjusted OR (95%CI)=1.28(1.10,1.50), P=0.002).We found that the gene-gene and gene-environmental two-way interactions were not significant (P>0.05), and hence6SNPs were independently associated with lung cancer risk and statistically significant (P<0.05).The6SNPs plus smoking had a cumulative association with lung cancer. In person who had any five or more of these factors associated with lung cancer, the odds ratio for lung cancer was4.50(adjusted OR (95%CI)=4.50(2.93,6.91), P=5.8×10-12), as compared with person with one or no factor.Over all models, he Environmental-Genetic lung cancer risk prediction model containing sex, smoking, and6SNPs performed best (AUC (95%CI)=0.63(0.61,0.65), P=2.2×10-3), and was significantly better than Demographic model (AUC (95%CI)=0.51(0.49,0.53), P=0.307), the Genetic variant model (AUC (95%CI)=0.56(0.54,0.58), P=2.4×10-10) and Nongenetic model (AUC (95%CI)=0.60(0.58,0.62), P=4.9×10-25).The predictive performance for lung cancer using the8factors in training set:accuracy was60.97%, sensitivity was44.68%and specificity was73.73%; in the validation set:accuracy was61.77%, sensitivity was36.84%and specificity was81.27%; in the all data:accuracy was61.37%, sensitivity was40.76%and specificity was77.50%.MDR analysis also verified the reliability of our model.In this thesis, the Environmental-Genetic model for predicting lung cancer in the Chinese Han population has been initially established and has a certain capacity of prediction for risk of lung cancer, playing an assisting role in the prevention and diagnosis of lung cancer. Therefore, more gene polymorphisms associated with lung cancer need to be identified for improving the risk prediction models for lung cancer.