Abstract
Fragility fracture is a serious public health problem in the world. The risk of fracture is determined by genetic and nongenetic clinical risk factors. This study sought to quantify the contribution of genetic profiling to fracture prognosis. The study was built on the ongoing Dubbo Osteoporosis Epidemiology Study, in which fracture and risk factors of 858 men and 1358 women had been monitored continuously from 1989 and 2008. Fragility fracture was ascertained by radiologic reports. Bone mineral density at the femoral neck was measured by dualβenergy Xβray absorptiometry (DXA). Fifty independent genes with allele frequencies ranging from 0.01 to 0.60 and relative risks (RRs) ranging from 1.01 to 3.0 were simulated. Three predictive models were fitted to the data in which fracture was a function of (1) clinical risk factors only, (2) genes only, and (3) clinical risk factors and 50 genes. The area under the curve (AUC) for model 1 was 0.77, which was lower than that of model II (AUCβ=β0.82). Adding genes into the clinical risk factors model (model 3) increased the AUC to 0.88 and improved the accuracy of fracture classification by 45%, with most (41%) improvement in specificity. In the presence of clinical risk factors, the number of genes required to achieve an AUC of 0.85 was around 25. These results suggest that genetic profiling could enhance the predictive accuracy of fracture prognosis and help to identify highβrisk individuals for appropriate management of osteoporosis or intervention. Β© 2011 American Society for Bone and Mineral Research.