Replication of previous genome-wide association studies of bone mineral density in premenopausal American women

Abstract

Bone mineral density (BMD) achieved during young adulthood (peak BMD) is one of the major determinants of osteoporotic fracture in later life. Genetic variants associated with BMD have been identified by three recent genome-wide association studies. The most significant single-nucleotide polymorphisms (SNPs) from these studies were genotyped to test whether they were associated with peak BMD in premenopausal American women. Femoral neck and lumbar spine BMD were determined by dual-energy X-ray absorptiometry in two groups of premenopausal women: 1524 white women and 512 black women. In premenopausal white women, two SNPs in the C6orf97/ESR1 region were significantly associated with BMD (p < 4.8 x 10(-4)), with suggestive evidence for CTNNBL1 and LRP5 (p < .01). Evidence of association with one of the two SNPs in the C6orf97/ESR1 region also was observed in premenopausal black women. Furthermore, SNPs in SP7 and a chromosome 4 intergenic region showed suggestive association with BMD in black women. Detailed analyses of additional SNPs in the C6orf97/ESR1 region revealed multiple genomic blocks independently associated with femoral neck and lumbar spine BMD. Findings in the three published genome-wide association studies were replicated in independent samples of premenopausal American women, suggesting that genetic variants in these genes or regions contribute to peak BMD in healthy women in various populations.

Fig. 1

Evidence of association in the CTNNBL1 region. Arrows indicate the size and location of the genes. Circles (femoral neck) and triangles (lumbar spine) denote SNPs tested in white women in our replication study (black) and genome-wide association study (gray). Association results (top panel) are indicated by chromosomal location on the x axis and –log of p values on the y axis. Threshold of p = .01 is marked by the dashed line. The extent of linkage disequilibrium (r²) structure (bottom panel) is represented by the degree of shading. Darker shades indicate greater linkage disequilibrium.

Fig. 2

Evidence of association in the LRP5 region. Arrows indicate the size and location of the genes. Circles (femoral neck) and triangles (lumbar spine) denote SNPs tested in white women in our replication study (black) and genome-wide association study (gray). Association results (top panel) are indicated by chromosomal location on the x axis and –log of p values on the y axis. Threshold of p = .01 is marked by the dashed line. The extent of linkage disequilibrium (r²) structure (bottom panel) is represented by the degree of shading. Darker shades indicate greater linkage disequilibrium.

Fig. 3

Evidence of association in the C6orf97/ESR1 region. Arrows indicate the size and location of the genes. Circles (femoral neck) and triangles (lumbar spine) denote SNPs tested in white women in our replication study (black) and genome-wide association study (gray). Association results (top panel) are indicated by chromosomal location on the x axis and –log of p values on the left y axis. Threshold of p = .01 is marked by the dashed line. Peaks indicate historic recombination in this region. Recombination rates are shown on the right y axis. The extent of linkage disequilibrium structures (middle, D' and bottom, r²) is represented by the degree of shading. Darker shading indicates stronger correlation between SNPs. Blocks based on linkage disequilibrium are represented by inverted triangles. The chromosomal regions corresponding to these blocks are marked by colored horizontal lines (top panel).

Fig. 4

Comparison of mean BMD as a function of SNP genotypes. The most significant SNP for lumbar spine and femoral neck BMD was selected from each of the four different blocks in Fig. 3. Genotypes and corresponding sample numbers (in parentheses) are on the x axis; mean BMD ± SEM (g/cm²) is on the y axis.