Implementation of personalized medicine for fracture risk assessment in osteoporosis

The present review summarizes our efforts to identify genetic polymorphisms associated with osteoporotic fractures and to establish a genetic risk score (GRS) to predict fracture risk in consecutive Japanese autopsy cases carried out at Tokyo Metropolitan Geriatric Hospital between 1995 and 2011. Three single nucleotide polymorphisms in transforming growth factor β-1, rs1800470; thrombospondin, type 1, domain-containing 7A, rs12673692; and formiminotransferase N-terminal subdomain-containing gene, rs7605378, showed a significant association with vertebral fracture prevalence, whereas five (α-l-iduronidase, rs3755955; C7orf58, rs190543052; homeobox C4, rs75256744; G patch domain-containing gene 1, rs2287679; and Werner syndrome, rs2230009) were significantly associated with femoral fracture. GRS values were calculated as the sum of risk allele counts (unweighted GRS) or of weighted scores estimated from logistic regression coefficients (weighted GRS). Both GRS values using the five single nucleotide polymorphisms adequately predicted femoral fracture prevalence for 924 male subjects; the areas under receiver-operating characteristic curves were 0.750 (95% confidence interval [CI] 0.660-0.840) and 0.770 (95% CI 0.681-0.859), respectively. Logistic regression analysis showed that the odds ratio for the association between fracture prevalence and unweighted GRS ≥3 (n = 124) was 8.39 (95% CI 4.22-16.69, P < 0.001) relative to GRS < 3 (n = 797). Likewise, the odds ratio for a weighted GRS of 6-15 (n = 135) was 7.73 (95% CI 3.89-15.36, P < 0.001) relative to GRS 0-5 (n = 786). Therefore, the GRS based on the risk allele profiles of these five single nucleotide polymorphisms could help identify at-risk individuals and enable implementation of preventive measures for femoral fracture.