Polygenic Risk Scores for Prediction of Gastric Cancer Based on Bioinformatics Screening and Validation of Functional lncRNA SNPs

Abstract

INTRODUCTION:Single-nucleotide polymorphisms (SNPs) are used to stratify the risk of gastric cancer. However, no study included gastric cancer-related long noncoding RNA (lncRNA) SNPs into the risk model for evaluation. This study aimed to replicate the associations of 21 lncRNA SNPs and to construct an individual risk prediction model for gastric cancer.METHODS:The bioinformatics method was used to screen gastric cancer-related lncRNA functional SNPs and verified in population. Gastric cancer risk prediction models were constructed using verified SNPs based on polygenic risk scores (PRSs).RESULTS:Twenty-one SNPs were screened, and the multivariate unconditional logistic regression analysis showed that 14 lncRNA SNPs were significantly associated with gastric cancer. In the distribution of genetic risk score in cases and controls, the mean value of PRS in cases was higher than that in controls. Approximately 20.1% of the cases was caused by genetic variation (P = 1.9 × 10-34) in optimal PRS model. The individual risk of gastric cancer in the lowest 10% of PRS was 82.1% (95% confidence interval [CI]: 0.102, 0.314) lower than that of the general population. The risk of gastric cancer in the highest 10% of PRS was 5.75-fold that of the general population (95% CI: 3.09, 10.70). The introduction of family history of tumor (area under the curve, 95% CI: 0.752, 0.69-0.814) and Helicobacter pylori infection (area under the curve, 95% CI: 0.773, 0.702-0.843) on the basis of PRS could significantly improve the recognition ability of the model.DISCUSSION:PRSs based on lncRNA SNPs could identify individuals with high risk of gastric cancer and combined with risk factors could improve the stratification.