Genetic risk and gastric cancer: polygenic risk scores in population-based case-control study.

OBJECTIVE: This study aimed to screen and identify common variants and long noncoding RNA (lncRNA) single nucleotide polymorphisms (SNPs) associated with gastric cancer risk, and construct prediction models based on polygenic risk score (PRS).

METHODS: The risk factors associated with gastric cancer were screened following meta-analysis and bioinformatics, verified by population-based case-control study. We constructed PRS and weighted genetic risk scores (wGRS) derived from the validation data set. Net reclassification improvement (NRI), integrated discrimination improvement (IDI), Akaike information criterion (AIC) and Bayesian information criterion (BIC) were used to evaluate model.

RESULTS: The PRS was divided into 10 quantiles, with the 40-60% quantile as a reference. A risk gradient was revealed across quantile of the PRS, the risk of gastric cancer in the highest 10 quantile of PRS was 3.24 folds higher than that in control population ( OR =3.24, 95%CI: 2.07, 5.06). For NRI and IDI, PRS combinations were significantly improved compared to wGRS model combinations ( P <0.001). The model of PRS combined with lncRNA SNPs, smoking, drinking and Helicobacter pylori infection was the best fitting model (AIC=117.23, BIC=122.31).

CONCLUSION: The model based on PRS combined with lncRNA SNPs, H. pylori infection, smoking and drinking had the optimal predictive ability for gastric cancer risk, which was helpful to distinguish high-risk groups.