28 (the lowest 10% of the population) at either LS or FN; high BMD subjects had BMD z-score ≥ +1.0 (the highest 15% of the population) at one or both skeletal sites [17, 18]. www.selleckchem.com/products/prt062607-p505-15-hcl.html Height was measured using a wall-mount stadiometer and weight with an electronic scale. HKOS prospective cohort (for

replication) This random population is also a part of the on-going HKSC database with BMD (n = 2,509) and KU-57788 vertebral fracture (n = 1,746) data. A total of 1,794 unrelated postmenopausal women (≥45 years) and 715 men (≥50 years), without receiving osteoporosis treatment or any drug known to influence bone metabolism, were included as described previously [19]. Vertebral fractures were assessed by digital measurements of morphologic changes on a lateral radiograph of the thoracolumbar spine. A vertebral body was considered fractured if there was a reduction of at least 3 SD in anterior, click here mid or posterior ratios compared with normative means [20]. The information on vertebral

fracture was available for a total of 1,746 subjects. All subjects gave informed consent. The study was approved by the institutional review board of the Hong Kong West Cluster Hospitals of the Hospital Authority and the University of Hong Kong and was conducted according to the Declaration of Helsinki. SNP genotyping A total of 10 SNPs in the POSTN gene were selected for genotyping: seven tSNPs with reported minor allele frequency (MAF) ≥0.05 in Chinese and three potentially functional SNPs located in exons. The tSNPs were identified using data from the phase II HapMap CHB (r 2 ≥ 0.8). SNPs for HKSC extreme cohort were genotyped using the high-throughput Sequenom MassARRAY platform (Sequenom, San Diego, CA, USA). DNA from high and low BMD subjects were randomly assigned to the 96-well plates and genotyping performed

with sample status blinded. Genotyping was repeated in 5% of the samples for verification: Data were confirmed to have an error rate <0.1%. The TaqMan system (Applied Biosystems, Foster City, CA, USA) was used for SNP genotyping in the verification and replication steps. Statistical methods Both single marker and haplotype association analyses were performed using the PLINK software [21]. Any SNP with call rate <90%, MAF <0.01 or Hardy–Weinberg CYTH4 equilibrium (HWE) P < 0.001 was excluded. The binary logistic regression was used to test the association between each SNP and BMD variation of the HKSC extreme cohort and vertebral fractures under the additive model. The association of SNP with BMD variation in the replication cohort was detected by the linear regression analysis. In the block-based haplotype association analysis, the haplotype global test is an omnibus test (if there are H haplotypes, a single test with H-1 degrees of freedom is conducted). The haplotype-specific test evaluates each specific haplotype versus all other haplotypes (i.e., tests with 1 degrees of freedom).