Single nucleotide polymorphisms have proven insightful as disease association and prediction markers. Two different disease, gastric cancer and osteoporosis were studied. In the gastric cancer study, 24 tag SNPs in gene regions of EZH2, JMJD3, and UTX were genotyped in 2,349 Korean gastric cancer patients and healthy controls and subsequently analyzed for association with gastric cancer. The genotyped SNPs were then calculated for interactions among different SNPs in different genes and their functional roles were predicted using web based tools. In the osteoporosis study, 30 SNPs in gene regions of 39 genes were genotyped in 1,229 Korean postmenopausal female patients and analyzed for association with osteoporosis related traits such as bone mineral density (BMD) or osteoporotic fractures (OF). 21 SNPs were then chosen to construct a genetic risk score (GRS) base on the genotypes of the SNPs, and the GRS was tested whether it could improve predictions of osteoporosis related traits.
EZH2 SNP rs6950683 (p = 0.0011), JMJD3 SNP rs78633955 (p = 0.00010), and UTX SNPs rs144974719 (p = 0.00024) and rs5952279 (p = 0.0011) were significantly associated with gastric cancer susceptibility after Bonferroni correction for multiple testing of 24 SNPs (P < 0.05/24 = 0.0021). Pairwise interaction of SNPs also showed marginally significant synergistic interaction among single pairs of EZH2, JMJD3, and UTX SNPs. (EZH2-JMJD3: p = 0.0003, EZH2-UTX: p = 0.020, JMJD3-UTX: p = 0.044).
GRS marginally associated with osteoporosis related traits such as BMD (p = 0.018) and nonvertebral fractures (NVF) (p = 0.045) after adjusting for clinical risk factors (CRFs). In terms of predicting NVF, adding GRS to the prediction model with only clinical risk factors marginally increased the area under the receptor-operator characteristics curve (AUC) from 0.65 to 0.67, and improved the accuracy of NVF classification by 11.5% (P = 0.014).