Abstract
Osteoporosis (OST) is a complex multifactorial disease. Prior studies evaluated gene effects seperately on OST. We evaluated gene-gene interactions of polymorphisms in tumor necrosis factor-α (TNF-α) and vitamin D receptor (VDR) genes on OST in elders. A total of 472 elders were included from Taichung Community Health Study for Elders; polymorphisms (TNF-α: rs1799964, rs1800629, rs3093662; VDR: rs7975232, rs1544410, rs2239185, rs3782905) were genotyped. Bone mineral densities (BMD) of lumbar spine (LS), femoral neck (FN), and total hip (TH) were measured by DEXA. Overall and site-specific OSTs were defined as BMD T-score ≤ − 2.5 standard deviations. Predictive models’ ability to discriminate OST status was evaluated by areas under the receiver operating characteristics (AUROC) curve. After considering age, BMI, physical activity, smoking, and drinking, significant interactions of TNF-α rs1799964 and VDR rs2239185, and TNF-α rs1800629 and VDR rs3782905 on overall and LS OST; interaction of TNF-α rs1799964 and VDR rs3782905 on LS OST in women were observed (P < 0.05). AUROC (95% CI) for Model1 (traditional factors), Model2 (Model1 + rs1800629 and rs3782905), and Model3 (Model2 + gene-gene interaction) for overall OST were 0.77 (0.70, 0.84), 0.79 (0.72, 0.86), and 0.81 (0.75, 0.88) in women, respectively. There were significant differences in AUROC between Models3 and Model1 (P=0.028), and Models3 and Model2 (P=0.047), indicating gene-gene interaction improved OST discrimination. Adding gene-gene interaction into traditional factors model did improve OST risk prediction in Han Chinese elders, and help to identify high-risk individuals for OST appropriate management and intervention.