ARTICLES
npg
© 2012 Nature America, Inc. All rights reserved.
Genome-wide meta-analysis identifies 56 bone mineral
density loci and reveals 14 loci associated with risk
of fracture
Bone mineral density (BMD) is the most widely used predictor of fracture risk. We performed the largest meta-analysis to date
on lumbar spine and femoral neck BMD, including 17 genome-wide association studies and 32,961 individuals of European and
east Asian ancestry. We tested the top BMD-associated markers for replication in 50,933 independent subjects and for association
with risk of low-trauma fracture in 31,016 individuals with a history of fracture (cases) and 102,444 controls. We identified 56 loci
(32 new) associated with BMD at genome-wide significance (P < 5 × 10−8). Several of these factors cluster within the RANKRANKL-OPG, mesenchymal stem cell differentiation, endochondral ossification and Wnt signaling pathways. However, we also
discovered loci that were localized to genes not known to have a role in bone biology. Fourteen BMD-associated loci were also
associated with fracture risk (P < 5 × 10−4, Bonferroni corrected), of which six reached P < 5 × 10−8, including at 18p11.21
(FAM210A), 7q21.3 (SLC25A13), 11q13.2 (LRP5), 4q22.1 (MEPE), 2p16.2 (SPTBN1) and 10q21.1 (DKK1). These findings shed
light on the genetic architecture and pathophysiological mechanisms underlying BMD variation and fracture susceptibility.
Osteoporosis is a disease characterized by low bone mass and microarchitectural deterioration of bone tissue leading to increased risk
of fracture. The disease accounts for approximately 1.5 million new
fracture cases each year, representing a huge economic burden on
health care systems, with annual costs estimated to be $17 billion in
the United States alone and expected to rise 50% by the year 2025
(ref. 1). Osteoporosis is defined clinically through the measurement
of BMD, which remains the single best predictor of fracture2,3.
Twin and family studies have shown that 50–85% of the variance
in BMD is genetically determined4. Osteoporotic fractures are also
heritable by mechanisms that are partly independent of BMD5. Over
the past 5 years, genome-wide association studies (GWAS) have revolutionized the understanding of the genetic architecture of common,
complex diseases6. This approach is providing key insights into the
mechanisms of disease, with prospects for the design of effective strategies for risk assessment and the development of new interventions7.
Previous GWAS have identified 24 loci that influence BMD variation8–14. Whereas several variants in these BMD-associated loci have
also been nominally associated with fracture risk15,16, none have shown
robust association with genome-wide significance (P < 5 × 10−8). We
report here the results of the largest effort to date searching for BMDassociated loci in >80,000 subjects and testing them for association
with fracture in >130,000 cases and controls. In addition, we employed
bioinformatics tools and gene expression analyses to place the identified variants in the context of pathways relevant to bone biology.
RESULTS
This study was performed across three main stages (Fig. 1): (i) the
discovery of BMD loci, (ii) follow-up replication and (iii) association
of the BMD-associated loci with fracture.
Discovery of BMD loci (stage 1)
We first performed a meta-analysis of multiple GWAS for BMD of
the femoral neck (FN-BMD; n = 32,961) and lumbar spine (LS-BMD;
n = 31,800 cases), including ~2.5 million genotyped or imputed autosomal SNPs from 17 studies of populations across North America,
Europe, East Asia and Australia, with a variety of epidemiological designs and subject characteristics (Online Methods). We also
performed meta-analysis in men and women separately to identify
sex-specific associations. The quantile-quantile plots of the discovery meta-analysis showed strong (and not early) deviation of the
observed statistics from the null distribution of no association for
both BMD traits (Supplementary Fig. 1). After double genomic control correction of the overall (λFN-BMD pooled = 1.112; λLS-BMD pooled =
1.127) and sex-stratified (λFN-BMD women = 1.091; λFN-BMD men = 1.059;
λLS-BMD women = 1.086; λLS-BMD men = 1.061) analyses, SNPs in 34 loci
surpassed genome-wide significance, whereas a total of 82 loci were
associated at P < 5 × 10−6 (Supplementary Figs. 2 and 3). Thirty-eight
loci were associated with FN-BMD, 25 with LS-BMD and 19 with both.
The overlap reflects correlation between the femoral neck and lumbar
spine measurements (Pearson’s correlation = 0.53). Of these 82 loci,
59, 18 and 5 were prioritized from analyses in the sex-combined,
female and male sample sets, respectively (Supplementary Table 1).
The meta-analysis was extended to include the evaluation of 76,253
markers on the X chromosome imputed across 14 of the discovery
GWAS, for a total of 31,801 participants (Online Methods). Five loci
on the X chromosome were associated at P < 5 × 10−5, with four of
these derived from the sex-combined analysis and one identified in the
analysis of men only (Supplementary Table 1). We further performed
genome-wide conditional analyses in all sex-combined stage 1 studies.
Each study repeated the GWAS analysis but also adjusted for 82 SNPs
A full list of authors and affiliations appears at the end of the paper.
Received 7 October 2011; accepted 16 March 2012; published online 15 April 2012; doi:10.1038/ng.2249
NATURE GENETICS
VOLUME 44 | NUMBER 5 | MAY 2012
491
ARTICLES
BMD discovery (stage 1)
meta-analysis of 17 BMD
genome-wide association studies
(n = 32,961)
BMD replication (stage 2)
96 SNPs in 34 studies
(de novo and in silico)
(n = 50,933)
Association of the BMD loci
with fracture (stage 3)
96 SNPs in 50 studies
(de novo and in silico)
(n = 31,016 cases and 102,444
controls)
npg
© 2012 Nature America, Inc. All rights reserved.
Figure 1 Description of study design. Stage 1: meta-analysis of 17
genome-wide association studies for BMD. Stage 2: 96 top independent
SNPs (82 autosomal SNPs with P < 5 × 10−6, 5 SNPs on the X chromosome
and 9 SNPs from conditional analysis) were followed up in de novo and
in silico replication of the BMD association in 34 studies. Stage 3: the
same 96 SNPs were tested for association with fracture in 50 studies with
de novo and in silico data.
representing the autosomal loci associated at P < 5 × 10−6 (Online
Methods). We then performed meta-analysis on these studies in the
same way as in the primary GWAS meta-analysis. Nine loci showed
at least two independent association signals in this conditional analysis (Supplementary Fig. 4 and Supplementary Table 2), suggesting
that allelic heterogeneity underlies BMD variation. We also assessed
all possible pairwise interactions of the 82 SNPs, but none were
significant after adjusting for the number of tests (Supplementary
Fig. 5 and Supplementary Table 3). A total of 96 independent SNPs
(82 autosomal SNPs with P < 5 × 10−6, 9 autosomal SNPs from conditional analysis and 5 SNPs on the X chromosome) from 87 genomic
loci were selected for further replication (Fig. 1).
Follow-up replication (stage 2)
We performed de novo genotyping of these 96 SNPs and tested them
for association with BMD in up to 50,933 additional participants from
34 studies (Online Methods). Meta-analysis of the 96 SNPs in the discovery and replication studies (n = 83,894) yielded 64 replicating SNPs
from 56 associated loci. Of these loci, 32 were newly found to show association (Table 1 and Supplementary Table 4a), and 24 were reported
previously8–14 (Supplementary Table 4b). Thirty-two SNPs did not
reach genome-wide significance after replication (Supplementary
Table 4c), including 10 markers that remained associated at a suggestive level. Of all the SNPs analyzed, only one (rs9533090 mapping
to 13q14.11 near TNFSF11 (also known as RANKL)) showed a high
degree of heterogeneity of effects (I2 > 50%) across studies, despite
being the marker that associated with highest significance (P = 4.82 ×
10−68) in the fixed-effect meta-analysis (Supplementary Table 4b).
After applying random-effects meta-analysis, this marker was still
associated with genome-wide significance (P = 3.98 × 10−13).
Two of the newly identified loci were discovered in the sex-stratified
meta-analysis: 8q13.3 in women and Xp22.31 in men; however, only
the association at Xp22.31 showed significant evidence for sex specificity, as reflected by significant heterogeneity of effects across sex
strata (Phet = 1.62 × 10−8). Yet, we acknowledge that the association at
8q13.3 in women may have been driven by a lower number of men in
the discovery and replication data sets (Table 1 and Supplementary
Table 5). Furthermore, evidence for BMD site specificity (Phet < 5 ×
10−4) was observed in a proportion of the loci, including 6 of the
32 new and 4 of the 24 known loci (Table 1 and Supplementary
492
Fig. 6). Among the newly identified loci, 2q14 (INSIG2), 12p11.22
(PTHLH) and 16q12.1 (CYLD) showed site specificity with FN-BMD,
and 8q13.3 (LACTB2), 10p11.23 (MPP7) and 10q22.3 (KCNMA1)
showed site-specificity with LS-BMD.
After replication, the conditional analysis provided significant evidence of association (P < 5 × 10−8) in eight of the nine loci containing secondary signals (Supplementary Fig. 4 and Supplementary
Table 2). Three loci had variants located less than 40 kb from the initial
main signal, suggesting allelic heterogeneity, including at 1p31.3
(represented by rs17482952 near WLS), 6q25.1 (rs7751941 near
ESR1) and 16q12.1 (rs1564981 near CYLD). The secondary signal
at 16q12.1 (rs1564981) showed a strong association with LS-BMD,
whereas the main signal in this locus (rs1566045) was only associated
with FN-BMD. The other five secondary signals were represented
by variants localized more than 180 kb from the initial main signal
and were located in different candidate genes, including at 1p36.12
(rs7521902 near WNT4), 7p14.1 (rs10226308 near SFRP4), 7q31.31
(rs13245690 near C7orf58), 12q13.13 (rs736825 near HOXC6) and
17q21.31 (rs4792909 near SOST). The secondary signal mapping to
the 13q14.11 locus (rs7326472) did not achieve genome-wide significance after replication.
Association of the BMD loci with fracture (stage 3)
We tested the 96 markers for association with fracture in 31,016
cases and 102,444 controls from 50 studies with fracture information. This collection included 5,411 cases and 21,909 controls tested
in the BMD GWAS discovery samples, 9,187 cases and 45,057 controls
tested by in silico replication and 16,418 cases and 35,478 controls
tested by de novo genotyping (Fig. 1 and Online Methods). In this
fracture meta-analysis, 14 loci were significantly associated with any
type of fracture at Bonferroni-corrected significance (P = 5 × 10−4),
of which five were new BMD-associated loci. None of the markers
showed large estimates of heterogeneity (Table 2, Supplementary
Fig. 7 and Supplementary Table 6). Markers at six of these loci
reached P < 5 × 10−8, including at 18p11.21 (FAM210A; also known
as C18orf19), 7q21.3 (SLC25A13), 11q13.2 (LRP5), 4q22.1 (MEPE),
2p16.2 (SPTBN1) and 10q21.1 (DKK1). The proportion of the overall
fracture risk explained by BMD ranged between 0.09 and 0.40 across
markers (Supplementary Table 7) and was estimated in a subset of
stage 2 samples (including n = 8,594 cases and 23,218 controls) by
modeling the effect of BMD-associated SNPs on fracture risk, with
and without the inclusion of BMD as a covariate. In general, the effect
of these SNPs on BMD was larger than on fracture risk (Fig. 2a),
except for the most significantly associated locus for fracture at
18p11.21 (Fig. 2b). SNPs in genes of the RANK-RANKL-OPG pathway (TNFRSF11A, TNFSF11 and TNFRSF11B, respectively), despite
being the strongest loci associated with BMD, were not significantly
associated with fracture. All 31 BMD-associated loci that had nominal
association with fracture risk (P < 0.05) showed consistent direction (the allele associated with decreasing BMD was associated with
increased risk of fracture). When we performed subgroup analyses
using cleaner phenotype definitions generated by limiting subjects
to those with clinically validated fractures and stratifying by anatomical site (for example, non-vertebral and vertebral fractures), we
did not identify any additional signals (Supplementary Table 8). At
a nominally significant level (P < 0.05), only 3 loci were associated
with vertebral fracture, and all 14 BMD-associated loci were associated with non-vertebral fracture, although the difference in effect
between fracture sites was not significant. Therefore, the power of
our study did not benefit from improving phenotype definitions at
the cost of lower sample size.
VOLUME 44 | NUMBER 5 | MAY 2012
NATURE GENETICS
npg
© 2012 Nature America, Inc. All rights reserved.
NATURE GENETICS
Table 1 Estimated effects of new genome-wide significant SNPs on FN-BMD and LS-BMD across stages
FN-BMD
Functional evidence
SNP
Locus
VOLUME 44 | NUMBER 5 | MAY 2012
Closest
Knockout
Tags
gene/candidate eQTL mouse OMIM function GRAIL Pathway A
Freq.
T
T
A
0.74
0.23
0.76
rs479336
1q24.3
rs7584262 2p21
rs17040773 2q13
DNM3
PKDCC
ANAPC1
rs1878526
2q14.2
INSIG2
rs1026364
rs344081
rs3755955
rs11755164
3q13.2
3q25.31
4p16.3
6p21.1
KIAA2018
LEKR1
IDUA
SUPT3H/RUNX2
rs9466056
rs3801387
rs13245690e
rs7812088
rs7017914c
6p22.3
7q31.31
7q31.31
7q36.1
8q13.3
CDKAL1/SOX4
WNT16
C7orf58
ABCF2
XKR9/LACTB2
rs7851693
rs3905706
9q34.11 FUBP3
10p11.23 MPP7
rs1373004
rs7071206
10q21.1
10q22.3
12p13.33
12q13.12
12q23.3
14q32.12
16p13.11
16p13.3
16p13.3
•
rs1566045
16q12.1
ERC1/WNT5B
DHH
C12orf23
RPS6KA5
NTAN1
AXIN1
C16orf38/
CLCN7
SALL1/CYLD
rs1564981e
rs4790881
rs7217932
16q12.1
17p13.3
17q24.3
CYLD
SMG6
SOX9
Stages 1 and 2
(83,894)
P
P
P
βc
−0.04 1.1 × 10−7 1.3 × 10−8
0.03 1.4 × 10−7 3.4 × 10−4
0.03 4.3 × 10−6 6.1 × 10−5
0.22
T
T
A
T
0.37
0.87
0.16
0.40
0.03
0.03
−0.05
−0.01
2.0 × 10−6
1.1 × 10−4
3.9 × 10−7
0.23
A
A
A
A
A
0.38
0.74
0.62
0.13
0.49
−0.03
−0.08
0.00
0.04
0.02
1.8
4.2
8.6
1.2
4.7
C
T
0.64
0.22
0.05 3.1 × 10−8
−0.02 0.63
T
T
0.13
0.78
T
T
A
•
A
T
T
T
A
T
A
•
•
•
A
T
A
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
0.00 0.70
×
×
×
×
×
10−8
10−14
10−5
10−6
10−8
Stage 1
(31,800)
Stage 2
(45,708)
Stages 1 and 2
(77,508)
P
P
P
Phet siteb
5.0 × 10−4
0.28
0.21
2.1 × 10−5
0.07
0.19
0.05
0.01
5.2 × 10−3
3.4 × 10−6
1.2 × 10−10 8.6 × 10−5
βa
8.5 × 10−15 −0.03 0.01
1.3 × 10−9
0.01 0.13
1.5 × 10−9
0.01 0.61
0.79
4.1 × 10−10 0.02 0.04
2.2 × 10−6
0.06 2.8 × 10−5
1.5 × 10−14 −0.05 1.4 × 10−7
0.05
−0.03 3.5 × 10−7
1.6 ×
2.0 ×
0.69
4.4 ×
7.1 ×
10−6
10−27
10−4
10−3
2.7
5.0
8.2
7.3
1.9
×
×
×
×
×
0.04
7.3 × 10−6
0.97
2.5 × 10−5
2.5 × 10−3
6.1 × 10−9
0.12
10−13 −0.03 6.5 × 10−5
10−40 −0.10 1.4 × 10−16
10−4
0.03 1.1 × 10−9
10−9
0.04 2.9 × 10−5
10−8 −0.01 0.35
7.3
3.5
5.5
9.2
×
×
×
×
1.1 ×
1.5 ×
1.3 ×
1.1 ×
0.41
10−3
10−8
10−9
10−6
7.6
4.5
5.2
5.6
×
×
×
×
10−4
10−12
10−15
10−11
0.11
0.12
0.80
2.1 × 10−3
10−4
10−36
10−3
10−3
3.6 ×
3.2 ×
6.0 ×
2.2 ×
0.98
10−8
10−51
10−11
10−7
0.34
0.09
0.05
0.86
9.1 × 10−5
3.4 × 10−22
0.03
0.04 0.06
0.05 2.9 × 10−9
6.7 × 10−8
6.7 × 10−9
6.1 × 10−8 0.02
2.4 × 10−16 5.8 × 10−11
−0.04 1.4 × 10−5 1.5 × 10−4
0.01 0.29
0.26
1.5 × 10−8
0.81
−0.05 5.4 × 10−8
-0.05 1.5 × 10−12
2.2 × 10−6
6.2 × 10−9
1.6 × 10−12 0.28
5.0 × 10−19 5.9 × 10−9
0.39
0.55
0.18
0.03 1.4 × 10−4 1.6 × 10−6
−0.01 0.17
0.08
0.04 5.8 × 10−8 2.4 × 10−6
9.0 × 10−10 0.03 0.01
0.03
−0.02 3.0 × 10−5
1.9 × 10−12 −0.02 0.94
1.9 × 10−5
2.4 × 10−4
0.05
9.2 × 10−7
4.9 × 10−8
0.13
0.76
0.39
0.52
0.81
0.67
0.48
0.43
−0.03
0.03
−0.03
−0.05
−0.03
−0.03
−0.04
T
0.80
−0.06 5.0 × 10−12 3.0 × 10−12
1.9 × 10−22 −0.01
0.55
0.04
A
A
A
0.47
0.69
0.46
−0.02 1.1 × 10−3 0.01
0.05 1.7 × 10−8 1.2 × 10−11
0.03 3.7 × 10−8 2.7 × 10−5
4.4 × 10−5 −0.03 6.2 × 10−8
9.8 × 10−19 0.04 6.0 × 10−4
1.9 × 10−11 0.01 0.31
5.4 × 10−4
1.7 × 10−6
0.15
2.0 × 10−10 0.50
3.4 × 10−9 0.13
0.08
3.8 × 10−3
0.63
0.73
0.74
0.02 3.2 × 10−6 1.1 × 10−3
−0.02 5.7 × 10−6 7.1 × 10−4
−0.08 0.01
8.3 × 10−4
4.9 × 10−8
0.01 5.2 × 10−4
−8
5.5 × 10
−0.03 9.2 × 10−9
1.6 × 10−4 −0.09 5.7 × 10−6
0.11
1.2 × 10−4
3.2 × 10−4
6.7 × 10−4 0.29
6.6 × 10−11 0.38
1.2 × 10−8 0.17
1.1
1.9
1.4
2.9
3.5
2.5
2.9
×
×
×
×
×
×
×
10−4
10−4
10−5
10−8
10−4
10−7
10−7
1.4 × 10−15
1.7 × 10−3
1.6
5.2
1.8
9.3
1.4
2.4
1.1
×
×
×
×
×
×
×
10−5
10−4
10−5
10−9
10−7
10−6
10−10
6.5
3.3
9.6
2.0
1.7
5.2
1.5
×
×
×
×
×
×
×
10−9
10−7
10−10
10−15
10−10
10−12
10−16
−0.04
0.05
−0.02
−0.04
−0.03
−0.04
−0.04
2.2
1.5
2.5
1.7
8.7
2.2
5.9
×
×
×
×
×
×
×
10−7
10−7
10−6
10−11
10−7
10−8
10−5
7.8 × 10−3
2.9
1.9
2.4
7.1
1.8
8.3
5.8
×
×
×
×
×
×
×
10−6
10−9
10−3
10−6
10−4
10−10
10−10
5.6
1.2
7.9
1.8
2.2
1.0
1.7
×
×
×
×
×
×
×
10−12
10−15
10−8
10−14
10−9
10−16
10−13
0.58
0.03
2.3 × 10−7
0.37
0.06
0.76
0.92
0.98
0.26
0.80
7.5 × 10−6
Boldface indicates P < 5 × 10−8 or site-speciic P < 5 × 10−4. A, allele; β, effect estimates; freq., allele frequency of A. Effect estimates are expressed as standardized values per copy of the SNP allele from ixed-effects meta-analysis.
Black dots in the six functional evidence columns indicate, respectively, that the SNP is an eQTL, there is a knockout mouse with skeletal phenotypes (Mouse Genome Informatics (MGI) database 2011), the candidate gene is involved in
a monogenic syndrome with skeletal phenotypes (OMIM 2011), the most signiicant SNP tags a SNP predicted to have impact on function of the candidate gene, the gene is the best candidate in GRAIL analysis, and the candidate gene
is part of a bone-active pathway. Candidate genes from GRAIL and/or the literature are shown if different from the closest gene.
493
aEffect estimates were calculated in the stage 2 samples. bSite speciicity null hypothesis, β
c
LS-BMD = βFN-BMD. rs7017914 was discovered in the meta-analysis of women only. The effects and P value for this marker are for the meta-analysis of
women samples. drs5934507 was discovered in the meta-analysis of men only. The effects and P value for this marker are for the meta-analysis of men samples. ers13245690 and rs1564981 were independently associated to their main signals in
conditional analysis.
ARTICLES
rs4796995 18p11.21 FAM210A
rs10416218 19q13.11 GPATCH1
rs5934507d Xp22.31 FAM9B/KAL1
Stage 2
(50,933)
A
•
MBL2/DKK1
KCNMA1
rs7084921 10q24.2 CPN1
rs10835187 11p14.1 LIN7C
rs7953528 12p11.22 KLHDC5/PTHLH
rs2887571
rs12821008
rs1053051
rs1286083
rs4985155
rs9921222
rs13336428
•
LS-BMD
Stage 1
(32,961)
ARTICLES
Table 2 Association of identified BMD-associated loci with risk for any type of low-trauma fracture
Functional evidence
SNP
Locus
© 2012 Nature America, Inc. All rights reserved.
Combined meta-analysis results
25,605 cases,
80,535 controls
31,016 cases, 102,444 controls
Closest
Knockout
Tags
Risk
OR
gene/candidate eQTL mouse OMIM function GRAIL Pathway allele Freq.b (95% CI)
Loci signiicantly associated with fracture risk at P < 5 ×
npg
Meta-analysis without
studies included in
BMD discovery
rs4233949 2p16.2
SPTBN1
rs6532023 4q22.1
MEPE/SPP1
rs4727338 7q21.3
SLC25A13
rs1373004 10q21.1
MBL2/DKK1
•
rs3736228 11q13.2
LRP5
•
rs4796995 18p11.21
FAM210A
OR (95%
CI)
P
Q het P I 2
10−8
G
0.63
1.07
1.4 × 10−7
1.06
2.6 × 10−8
(1.04–1.09)
(1.04–1.08)
G
0.67
G
0.32
•
T
0.13
•
T
0.15
G
0.39
1.06
(1.04–1.09)
1.08
(1.05–1.10)
1.09
(1.06–1.13)
1.09
(1.05–1.12)
1.06
(1.04–1.09)
G
0.83
•
•
P
•
•
•
•
•
•
8.8 × 10−7
1.0 × 10−8
7.2 × 10−7
2.1 × 10−6
6.4 × 10−7
1.06
(1.04–1.09)
1.08
(1.05–1.10)
1.10
(1.06–1.13)
1.09
(1.06–1.13)
1.08
(1.06–1.10)
1.7 × 10−8
0.36
6
1.00
0
5.9 × 10−11 0.03 31
9.0 × 10−9
0.64
0
1.4 × 10−8
0.78
0
8.8 × 10−13 0.12 20
Other signiicant loci associated with fracture risk at P < 5 × 10−4 (Bonferroni)
rs6426749 1p36.12
ZBTB40
rs7521902 1p36.12a
WNT4
rs430727
CTNNB1
3p22.1
•
•
•
•
A
0.27
•
•
T
0.47
T
0.33
A
0.74
rs6959212 7p14.1
STARD3NL
rs3801387 7q31.31
WNT16
rs7851693 9q34.11
FUBP3
G
0.37
rs163879
DCDC5
T
0.66
rs1286083 14q32.12
RPS6KA5
T
0.81
rs4792909 17q21.31a
SOST
G
0.62
rs227584
C17orf53
A
0.67
11p14.1
17q21.31
•
•
•
•
•
•
•
1.06
(1.03–1.09)
1.10
(1.06–1.14)
1.05
(1.03–1.08)
1.04
(1.02–1.07)
1.08
(1.05–1.11)
1.04
(1.01–1.06)
1.06
(1.03–1.09)
1.05
(1.02–1.08)
1.07
(1.04–1.11)
1.05
(1.02–1.08)
2.4 × 10−4
3.5 × 10−6
2.4 × 10−5
1.0 × 10−3
4.9 × 10−9
1.9 × 10−3
6.4 × 10−6
9.8 × 10−4
4.0 × 10−5
2.2 × 10−4
1.07
(1.04–1.10)
1.09
(1.06–1.13)
1.06
(1.03–1.08)
1.05
(1.02–1.07)
1.06
(1.04–1.08)
1.05
(1.02–1.07)
1.05
(1.03–1.07)
1.05
(1.03–1.08)
1.07
(1.04–1.10)
1.05
(1.03–1.07)
3.6 × 10−6
0.07 24
1.4 × 10−7
0.87
0
2.9 × 10−7
0.93
0
7.2 × 10−5
0.43
2
2.7 × 10−7
0.69
0
3.5 × 10−5
0.65
0
3.3 × 10−5
0.05 28
7.2 × 10−5
0.01 34
6.9 × 10−6
0.31 10
4.1 × 10−5
0.49
0
Odds ratios (ORs) estimated per risk allele copy for any low-trauma fracture among cases compared with controls. Qhet is the Cochran’s Q statistic, and I2 is the measure of
heterogeneity. Boldface indicates gene names from new loci and/or those associated with P < 5 × 10−8. Black dots in the six functional evidence columns indicate that,
respectively, the SNP is an eQTL, there is a knockout mouse with skeletal phenotypes (MGI database 2011), the candidate gene is involved in a monogenic syndrome with skeletal
phenotypes (OMIM 2011), the most signiicant SNP tags a SNP predicted to have impact on function of the candidate gene, the gene is the best candidate in GRAIL analysis, and
the candidate gene is part of a bone-active pathway.
ars7521902
and rs4792909 are secondary independent signals. bFreq. is the frequency of the risk allele.
Allele risk modeling for osteoporosis and fracture
The combined effect of all significant autosomal SNPs on BMD,
osteoporosis and any type of fracture was modeled in the Prospective
Epidemiological Risk Factor (PERF) study (n = 2,836), a prospective
study in postmenopausal Danish women aged 55–86 years17. This
study represents an independent validation setting, as it was excluded
from the overall meta-analysis for this purpose (Supplementary
Note). Risk alleles in the score (for example, BMD-decreasing alleles)
were weighted by their individual effects on BMD and grouped into
five bins (Supplementary Table 9). The difference in mean FN-BMD
between individuals in the highest bin of risk score (9% of the population; n = 244) and those in the middle bin (34% of the population;
n = 978) was −0.33 s.d. (Fig. 3a). This analysis was based on data at 63
SNPs and explained 5.8% (95% confidence interval (CI) = 4.0%–7.6%)
of the total genetic variance in FN-BMD.
494
The ability of this genetic score to predict the risk for osteoporosis
(defined by a T score of ≤−2.5) and for fracture was modeled in the
PERF study using the middle bin as reference (odds ratio (OR) = 1).
Women in the highest bin had 1.56 (95% CI = 1.12–2.18) increased
odds for osteoporosis (Fig. 3b), whereas women in the lowest bin
were protected from osteoporosis (OR = 0.38, 95% CI = 0.23–0.63).
A model based on the 16 BMD-associated SNPs that were also associated with fracture risk showed that women in the highest bin had 1.60
(95% CI = 1.15–2.24) increased odds for fracture, whereas women
in the lowest bin had a decreased risk for fracture (OR = 0.54, 95%
CI = 0.36–0.83) (Fig. 3c). Despite serving as robust proof of the
relationship between BMD-decreasing alleles and the risk of osteoporosis and fracture, prediction ability was modest. Receiver operating characteristics (ROC) analysis showed a significant but relatively
small discrimination ability of the genetic score alone, with an area
VOLUME 44 | NUMBER 5 | MAY 2012
NATURE GENETICS
ARTICLES
a
b
rs7851693 (9q34)
18p11
rs227584 (17q21)
r
rs163879 (11p14)
10
rs6959212 (7p14)
–log10 (P value)
rs1286083 (14q32)
rs4233949 (2p16)
rs6532023 (4q22)
rs3801387 (7q31)
rs4792909* (7q21)
0.8
0.6
0.4
0.2
100
–8
8
PBMD 1+2 = 4 × 10
6
PBMD 1 = 3 × 10
–6
80
60
4
40
2
20
rs6426749 (1p36)
Recombination rate (cM/Mb)
rs430727 (3p22)
rs4796995
–13
Pfracture = 8.8 × 10
2
rs4727338 (7q21)
rs3736228 (11q13)
rs7521902* (1p36)
0
0
rs1373004 (10q21)
C18orf1
ZNF519
RNMT
rs4796995 (18p11)
C18orf19
rs2062377 (OPG)**
MC2R
MC5R
rs884205 (RANK)**
rs9533090 (RANKL)**
13.5
npg
© 2012 Nature America, Inc. All rights reserved.
1.00
1.05
1.10
1.15
1.20
14.0
Position on chr. 18 (Mb)
Effect estimate
Figure 2 Association of BMD loci with fracture risk. (a) Phenotype-wide effects for the BMD loci associated with fracture and those that are
part of the RANK-RANKL-OPG pathway. Genetic effect estimates are shown for fracture (blue circles), LS-BMD (yellow squares) and FN-BMD
(green diamonds) for the 14 loci associated with fracture risk. Horizontal lines represent 95% confidence limits. Effect estimates are shown after
transformation of the standardized mean difference (SMD) in the BMD effect to odds ratio equivalents 34 (for example, a 0.02 SMD in the BMD
effect corresponds to an OR of 1.04). Secondary signals for rs227584 and rs6426749 are marked with an asterisk and the signals mapping to the
TNFRSF11B (also known as OPG; rs2062377), TNFRSF11A (also known as RANK; rs884205) and TNFSF11 (also known as RANKL; rs9533090)
genes are marked with a double asterisk. (b) Regional association plot for the 18p11.21 locus showing the P value for the top SNP associated with
fracture (rs4796995) together with P values from the BMD discovery set (stage 1) and combined with the BMD replication (stage 1 + 2). SNPs are
plotted by position in a 500-kb window of chromosome 18 against association with FN-BMD (−log 10 (P value)). Estimated recombination rates
(from HapMap) are plotted in cyan to reflect the local LD structure. SNPs surrounding the most significant SNP are color-coded according to
LD between these markers (pairwise r 2). Genes, exons and transcription direction are derived from the UCSC Genome Browser.
under the curve (AUC) of 0.59 (95% CI = 0.56–0.62) for osteoporosis
(Supplementary Fig. 8). Adding this score to a model with age and
weight alone (AUC = 0.75, 95% CI = 0.73–0.77) did not substantially
increase discrimination (AUC = 0.76, 95% CI = 0.74–0.78). A similar
pattern was observed for fracture discrimination, with AUCs of 0.57
(95% CI = 0.55–0.59) in a model with the score alone and 0.62 (95%
CI = 0.60–0.64) in a model with age, weight and height. A model
considering all 63 SNPs did not change the AUC for fracture risk
prediction (0.57, 95% CI = 0.54–0.59).
Functional annotations and pathway analyses
For the purpose of fine mapping and identifying additional SNPs with
putative functional implication using linkage disequilibrium (LD),
a subset of nine discovery studies (FN-BMD, n = 21,699; LS-BMD,
n = 20,835) used 1000 Genomes Project data (Release June 2010) to
re-impute genotypes at the 55 autosomal BMD loci (Supplementary
Note). In 13 of the 55 BMD-associated loci (the SNP on the X chromosome was not included), we identified markers in the surrounding
1-Mb region that were imputed from 1000 Genomes Project data
and that were more significant than the original HapMap signals
(Supplementary Tables 10 and 11), highlighting the benefit of using
a denser reference panel of markers. All HapMap markers in LD with
variants with functional annotation and showing higher significance in
the 1000 Genomes Project meta-analysis are shown (Supplementary
Table 12). In 14 of the 56 identified BMD-associated loci, a marker
from HapMap imputation was highly correlated (r 2 > 0.8) with at
least one putative functional variant annotated in the 1000 Genomes
Project reference. Three of the 14 BMD-associated loci that also associated with fracture contained putative functional variants tagged by
NATURE GENETICS
VOLUME 44 | NUMBER 5 | MAY 2012
the top SNPs of the BMD meta-analysis. These included the known
rs3736228 functional marker in LRP5 (encoding p.Ala1330Val)16,18,
the intronic marker rs3779381 within a promoter and/or regulatory
region of WNT16 and one intronic marker (rs4305309) within a promoter and/or regulatory region of SPTBN1.
Expression profiles at the BMD loci associated with genome-wide
significance were analyzed within four data sets (Supplementary
Note). In transiliac bone biopsies, expression of five genes correlated
with LS-BMD and/or FN-BMD of the donors with P < 0.001, including PSME4 (2p16.2), DKK1 (10q21.1), MIR22HG (also known as
C17orf91; 17p13.3), SOST (17q21.31_1) and DUSP3 (17q21.31_1)
(Supplementary Table 13). Among these loci, the SNP at DKK1
(10q21.1) was the most significantly correlated with FN-BMD
(P = 1.3 × 10−5) and LS-BMD (P = 3.2 × 10−4). Variants in all these
BMD-associated loci (with the exception of MIR22HG at 17p13.3)
were also associated with fractures.
SNP expression quantitative trait locus (eQTL) analyses were performed across diverse tissues, examining the correlation between
marker alleles and transcript levels at the associated BMD loci.
Fourteen of the BMD-associated SNPs correlated with the expression
of one or more of the nearby genes with P < 5 × 10−5 and were either the
strongest cis variants or were good surrogates of these for the affected
genes (Supplementary Tables 14 and 15). The most significant BMDassociated SNP eQTL was observed for rs10835187[T], resulting in
reduced expression of the LIN7C gene at the 11p14.1 locus (P = 2.8 ×
10−39 in adipose tissue). Of particular interest were BMD-associated
SNP cis variants at three loci that were also associated with fracture,
including 1p36.12, 4q22.1 and 17q21.31. At 1p36.12, rs6426749[G]
correlated with reduced WNT4 expression in fibroblasts, osteoblasts
495
ARTICLES
a
b
1,200
c
2.2
1,200
0.8
600
0
400
Number of individuals
Number of individuals
FN-BMD Z score
0.2
–0.2
0
1.6
800
1.4
1.2
600
1.0
0.8
400
0.6
0.4
200
–0.4
1,000
1.8
Odds ratio for osteoporosis
0.4
800
200
800
600
400
200
0.2
0
0
–0.6
0
<44
44–48
48–52
52–56
≥56
(n = 247) (n = 672) (n = 978) (n = 695) (n = 244)
<44
44–48
48–52
52–56
≥56
(n = 247) (n = 672) (n = 978) (n = 695) (n = 244)
<36
36–45
45–54
54–63
≥63
(n = 160) (n = 649) (n = 1,190) (n = 670) (n = 167)
Genetic score
Genetic score
Genetic score
Figure 3 Combined effect of BMD-decreasing alleles and fracture risk–increasing risk alleles modeled in the population-based PERF study (n = 2,836
women). (a–c) Effects are shown for baseline FN-BMD standardized residuals (Z scores) (a), risk for osteoporosis (b) and risk for any type of fracture (c).
The genetic score of each individual in a and b was based on the 63 SNPs showing genome-wide significant association with BMD (55 main and
8 secondary signals) and in c was based on the 16 BMD SNPs associated with fracture. Both genetic scores are weighted for relative effect sizes
estimated without the PERF study. Weighted allele counts summed for each individual were divided by the mean effect size, making them equivalent to
the percent of alleles carried by each individual, and sorted into five bins. Histograms show the numbers of individuals in each genetic score category
(left y axis). Diamonds (right y axis) represent mean FN-BMD standardized levels in a, risk estimates in the form of odds ratios and osteoporosis
(defined as NHANES T score of ≤ –2.5) in b and any type of fracture in c, using the middle category as reference (OR = 1). Vertical lines represent 95%
confidence limits.
320
rs65
2
PE
AM
HR
64
51
17
5
1
H
PT
LH
rs
npg
We applied the Gene Relationships Across Implicated Loci
(GRAIL) text-mining algorithm19 to investigate connections between
genes in the 55 autosomal BMD-associated loci. This analysis
revealed significant (P < 0.01) connections between genes in 18 of
the 55 input loci (Fig. 4 and Supplementary
Table 16). The strongest connections were
seen for members of three key biological
pathways: the RANK-RANKL-OPG pathway (encoded by TNFRSF11A, TNFSF11 and
TNFRSF11B, respectively); mesenchymal
stem cell differentiation (RUNX2, SP7 and
SOX9); and Wnt signaling (LRP5, CTNNB1,
SFRP4, WNT3, WNT4, WNT5B, WNT16
6
and
AXIN1), with the ten most frequently
26
6
01
rs2
connecting terms being bone, catenin, signaling, differentiation, rank, osteoblast,
diacylglycerol, kappab, development and
osteoclast. To assess the significance of this
biological gene connection enrichment, we
applied GRAIL to 2,000 randomly matched
sets of 55 SNPs (Supplementary Note) and
did not observe any set with 15 or more loci
SP7
with significantly enriched connectivity
SOX
6
23
228
rs3736
ME
TNFRSF11B
TNFSF
LRP5
rs884205
11
TNFRSF11A
3090
rs953
28
35
95
rs2062377
and adipose tissue; at 4q22.1, rs6532023[G] correlated with reduced
SPP1 (encoding osteopontin) expression in adipose tissue; and, at
17q21.31, rs227584[A] correlated with increased C17orf65 expression
in monocytes, adipose tissue, whole blood and lymphoblasts.
rs7
© 2012 Nature America, Inc. All rights reserved.
1,000
Number of individuals
0.6
1,000
Odds ratio for fracture
3.0
2.8
2.6
2.4
2.2
2.0
1.8
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0
1,200
2.0
rs710
8738
rs72
179
32
2
NX
RU
49
267
rs64
4
WNT
rs3801387
WNT16
WNT3
SO
X9
rs6
SF
95
RP
4
rs1
86
43
25
IN1
AX
rs9
FOXL1
12
92
CTNNB1
WNT5B
22
0481
rs10
46
rs430727
7571
rs288
496
92
12
Figure 4 Graphic representation of GRAIL
connections between SNPs and corresponding
genes for the 18 SNPs, as determined with
GRAIL P < 0.01. The top ten keywords linking
the genes were bone, catenin, signaling,
differentiation, rank, osteoblast, diacylglycerol,
kappab, development and osteoclast. Thicker
redder lines imply stronger literature-based
connectivity. Blue and black boxes depict loci
boundaries represented for each top-associated
marker (outer circle) and for each gene in the
region (inner circle).
VOLUME 44 | NUMBER 5 | MAY 2012
NATURE GENETICS
ARTICLES
npg
© 2012 Nature America, Inc. All rights reserved.
(Supplementary Fig. 9), providing strong statistical evidence of the
significant clustering of our BMD-associated loci (P < 0.0005).
DISCUSSION
In this report of the largest GWAS for osteoporosis traits to date, we identified 32 new genomic loci, bringing the total number of loci robustly
associated with BMD variation to 56. Furthermore, we report that six of
these BMD-associated loci are also associated with low-trauma fractures
at P < 5 × 10−8, an association that has not previously been detected.
In terms of other complex traits, our results indicate that hundreds of
variants with small effects may contribute to the genetic architecture of
BMD and fracture risk20. Our hypothesis-free assessment of common
variants of the genome provides new insights into biology, implicating
several factors that cluster in bone-active pathways.
Our results highlight the highly polygenic nature of BMD variation
and the critical role of several biological pathways influencing osteoporosis and fracture susceptibility (Supplementary Fig. 10). In addition to the Wnt factors known to be associated with BMD (CTNNB1,
SOST, LRP4, LRP5, WLS, WNT4 and MEF2C), several of the newly
discovered loci implicate additional Wnt signaling factors (including WNT5B, WNT16, DKK1, PTHLH, SFRP4 and AXIN1). Another
clearly delineated pathway is that involved in mesenchymal stem cell
differentiation, including the newly identified RUNX2, SOX4 and SOX9
BMD-associated loci along with the previously known SP7. Another
bone-relevant pathway includes that of endochondral ossification, which
involves essential processes during fetal development of the mammalian
skeleton and in which several of our identified BMD-associated loci
are implicated, including SPP1, MEF2C, RUNX2, SOX6, PTHLH, SP7
and SOX9. In addition, the biological relevance of our associations is
accentuated by the identification of genes underlying rare monogenetic
forms of osteoporosis and/or high bone mass, such as SOST, CLCN7
and LRP5 (refs. 21–23) (Supplementary Table 17), which also contain
common variants involved in normal BMD variation at the population
level11,14,16. This is supportive of a genetic architecture where both common and rare genetic variation may reside in the same locus24. Other
genes have not been reported to be associated with monogenic forms of
osteoporosis but have clear involvement in bone development in animal
models. For example, SNPs in the BMD-associated locus at 16q12.1
map near CYLD. Human mutations in this gene have been described
to cause familial cylindromatosis, a condition without phenotypic
skeletal manifestations. However, it has been shown that Cyld knockout
mice have significant bone loss, leading to a severe osteoporosis phenotype25 and also that CYLD regulates osteoclastogenesis26. Moreover,
evidence from the GWAS and eQTL analyses also suggests that some
loci contain more than one common variant with independent effects
on BMD and fracture risk. On the other hand, when no correlation is
observed between gene expression and a particular SNP, it is difficult
to draw conclusions. A correlation might be missed if the expression of
the transcript was not measured in a relevant tissue or if the expression
of a particular splice variant was not measured27.
BMD and fracture genetic effects correlate to some extent, but
some important risk variants for fracture may have minimal impact
on BMD and vice versa. This is the case for the signal at 18p11.21
(Fig. 2b), which, despite a modest effect on BMD (0.02% variance
explained), showed the most significant association with fracture
risk (OR = 1.08, 95% CI = 1.06–1.10; P = 8.8 × 10−13). This is in
contrast to variants that are known to have stronger effects on BMD
that were not significantly associated with fracture risk. For example,
variants affecting the RANK-RANKL-OPG pathway that has a critical
role in osteoclastogenesis had clear associations with BMD but not
with fracture risk (Fig. 2a). Even though loci discovery was based
NATURE GENETICS
VOLUME 44 | NUMBER 5 | MAY 2012
on the BMD phenotype, these findings reflect the heterogeneous
and complex nature of the mechanistic pathways leading to fracture.
Therefore, given our study design, we cannot rule out the possibility
that unidentified genetic loci influence risk for fracture independently
of BMD. Future well-powered GWAS meta-analyses on fracture risk
will address this question, while corroborating the associations with
fracture that we report for some of the BMD-associated loci (particularly those not associated with fracture at P < 5 × 10−8).
Our study also provides indication that there is sex and site specificity underlying BMD variation. One of the GWAS signals (Xp22.31)
was only significant in the sex-stratified analysis in men and showed
significant sex heterogeneity (Phet = 1.62 × 10−8). This is expected,
considering the sexual dimorphism of bone28,29. In fact, in a recent
GWAS, the rs5934507 SNP mapping to Xp22.31, which is associated
with BMD in the current study, was previously associated with male
serum testosterone levels30. Thus, it is likely that rs5934507 affects
serum testosterone, which in turn regulates BMD. In line with the
different types of bone composition at different skeletal sites (predominantly trabecular at the lumbar spine and cortical at the femoral
neck), we observed some indication of site specificity in 10 of the 56
BMD loci, suggesting differential genetic influences on BMD determination across skeletal sites. As has been previously shown 31, we
did not find in our results major differences in effect sizes between
individuals of European and east Asian ancestry (Supplementary
Fig. 7). However, this may be due to reduced power, given the smaller
number of individuals of east Asian ancestry. We tested a genetic risk
score to identify individuals at risk for osteoporosis and fracture and
showed that, cumulatively, the identified variants generate a gradient
of risk. These gradients reach ORs of 1.56 for osteoporosis and 1.60 for
fractures, when comparing participants with the highest risk scores to
those having the mean score. Yet, at present, there is limited clinical
usefulness for this score, as evidenced by its non-significant contribution to case discrimination when considering clinical risk factors with
strong effects on osteoporosis and fracture risk (like age and weight).
This is not unexpected, given the small fraction of genetic risk for
either BMD or fracture that has been identified thus far.
Our study has limitations. The identified SNPs are probably not
the causal variants; it is more likely that these markers are in LD
with the underlying causal variants. Additional analyses on potential
functional SNPs identified in this study will be required to determine
whether they are causal in these relationships with BMD. Moreover,
the causal genes underlying the GWAS signals may be different from
the candidate genes we describe, considering that our understanding
of the role of these candidate genes in bone biology is limited. Further
exploration of these loci with more detailed sequencing, gene expression and translational studies will be required. Such studies can also
disentangle the diverse types of complex relationships we currently
cannot distinguish in the BMD-associated loci with secondary signals
to determine whether these are the result of true allelic heterogeneity
or if they are driven by a second gene in the same region32. Similarly,
despite our large sample size, power limitations still influence the
detection of additional associations with smaller effect sizes and/or
those arising from rarer variants. Finally, given the different levels of
data availability and the difficulty of standardization across studies, we
did not evaluate the effect of additional risk factors for osteoporosis,
such as menopausal status and smoking, which can influence genetic
associations with BMD. Nonetheless, despite these limitations, we
have identified many new and previously unsuspected associations
with BMD variation and fracture risk.
Finally, the relatively weak effects of the variants discovered by GWAS
do not undermine the biological relevance of the genes identified,
497
npg
© 2012 Nature America, Inc. All rights reserved.
ARTICLES
as exemplified by the identification of genetic signals at genes coding
for proteins currently targeted by new osteoporosis treatments
(Supplementary Fig. 10). The new genes identified in our study may
represent new candidates to target for osteoporosis drug discovery.
Most established treatments for osteoporosis focus on curtailing bone
resorption (for example, bisphosphonates and RANKL inhibitors),
whereas only a few anabolic treatments are currently approved for the
treatment of osteoporosis (recombinant truncated or altered PTH).
Other anabolic compounds undergoing Phase 2 development include
PTHrP fragments and Wnt signaling enhancers, such as antibodies to
sclerostin33. Several of the variants robustly associated with BMD map
in or close to genes that encode proteins involved in these pharmacologic pathways, namely TNFRSF11B (encoding osteoprotegerin),
TNFRSF11A (encoding RANK), TNFSF11 (encoding RANKL),
PTHLH (encoding PTHrP), LRP5 (encoding low-density lipoprotein
receptor–related protein 5), SOST (encoding sclerostin) and DKK1
(encoding Dickkopf-1).
In conclusion, these findings highlight the highly polygenic and
complex nature of BMD variation, shed light on the pathophysiological mechanisms underlying fracture susceptibility and may contribute to the identification of future drug targets for the treatment
of osteoporosis.
URLs. GEFOS Consortium, http://www.gefos.org/; GENOMOS
Consortium, http://www.genomos.eu/; HapMap Project, http://hapmap.
ncbi.nlm.nih.gov/; 1000 Genomes Project, http://www.1000genomes.
org/; LocusZoom, http://csg.sph.umich.edu/locuszoom/; METAL,
http://www.sph.umich.edu/csg/abecasis/Metal/.
METHODS
Methods and any associated references are available in the online
version of the paper at http://www.nature.com/naturegenetics/.
Note: Supplementary information is available on the Nature Genetics website.
ACKNOWLEDGMENTS
We thank all study participants for making this work possible. This research and
the Genetic Factors for Osteoporosis (GEFOS) consortium have been funded by
the European Commission (HEALTH-F2-2008-201865-GEFOS). We acknowledge
funding from the following organizations: the US National Institutes of Health
(NIH; R01 AG18728, R01 HL088119, R01AR046838, U01 HL084756, P30
DK072488, T32 AG000262, F32 AR059469, P01 AG-18397, R01 AG041517,
M01 RR-00750 and N01-AG-12100), the NIA Intramural Research Program
(AG-023629, AG-15928, AG-20098 and AG-027058), Hjartavernd (the Icelandic
Heart Association), the Althingi (the Icelandic Parliament), the Australian National
Health and Medical Research Council (511132), the Australian Cancer Research
Foundation and the Rebecca Cooper Foundation, the Australian National Health
and Medical Research Council Career Development Award (569807 to E.L.D.),
an MRC New Investigator Award (MRC G0800582 to D.M.E.), the Health Research
Council of New Zealand, Sanofi-Aventis, Eli Lilly, Pfizer, Proctor & Gamble
Pharmaceuticals, Roche, the Medical Benefits Fund (MBF) Living Well
Foundation, the Ernst Heine Family Foundation, Arthritis Research UK (17539
and 15389), The Victorian Health Promotion Foundation, Geelong Region Medical
Research Foundation, Australia (628582), Action Research UK, the European
Commission (QLRT-2001-02629), the UK Food Standards Agency, BioPersMed
(COMET K-project 825329), the Austrian Federal Ministry of Transport,
Innovation and Technology (BMVIT), the Austrian Federal Ministry of Economics
and Labour (BMWA), the Austrian Federal Ministry of Economy, Family and
Youth (BMWFJ), the Styrian Business Promotion Agency (SFG), the Red de
Envejecimiento y Fragilidad (RETICEF), Instituto Carlos III, the Spanish Ministry
of Education and Science (SAF2010-15707), the Government of Catalonia
(2009SGR971 and 2009SGR818), Instituto de Salud Carlos III–Fondo de
Investigaciones Sanitarias (PI 06/0034 and PI08/0183), Healthway Health
Promotion Foundation of Western Australia, Australasian Menopause Society and
the Australian National Health and MRC Project (254627, 303169 and 572604),
the Finnish Ministry of Education, Merck Frosst Canada, Eli Lilly Canada, Novartis
Pharmaceuticals, Procter & Gamble Pharmaceuticals Canada, Servier Canada,
498
Amgen Canada, The Dairy Farmers of Canada, The Arthritis Society, the US
National Heart, Lung, and Blood Institute (NHLBI; N01-HC-85239, N01-HC85079 through N01-HC-85086; N01-HC-35129, N01 HC-15103, N01 HC-55222,
N01-HC-75150, N01-HC-45133, HL080295, HL075366, HL087652, HL105756
NINDS, HL 043851 and HL69757, CA 047988, and the Framingham Heart Study
(N01-HC-25195) and its contract with Affymetrix, Inc, for genotyping services
(N02-HL-6-4278)). Untied Educational Grants were provided by Amgen, Eli Lilly
International, GE-Lunar, Merck Australia, Sanofi-Aventis Australia and Servier.
Additional support was provided by the US National Center for Research
Resources (M01-RR00425 to the Cedars-Sinai General Clinical Research Center
Genotyping Core), the US National Institute of Diabetes and Digestive and Kidney
Diseases (DK063491 to the Southern California Diabetes Endocrinology Research
Center), deCODE Genetics, The UK National Institute for Medical Research
(NIMR) Biomedical Research Centre, the Cancer Research Campaign, the Stroke
Association, the British Heart Foundation, the UK Department of Health,
the Europe Against Cancer Programme Commission of the European Union,
the Ministry of Agriculture, Fisheries and Food, EU Biomed 1 (BMHICT920182,
CIPDCT925012, ERBC1PDCT 940229 and ERBC1PDCT930105), the UK MRC
(G9321536 and G9800062), the Wellcome Trust Collaborative Research Initiative
1995, MAFF AN0523, EU Framework Programme 5 (FP5; QLK6-CT-2002-02629),
the Food Standards Agency (N05046), the Netherlands Organization for Scientific
Research (NWO), Erasmus University Medical Center, the Centre for Medical
Systems Biology (CMSB1 and CMSB2) of the Netherlands Genomics Initiative
(NGI), the F.I.R.M.O. Fondazione Raffaella Becagli, the National Institute for
Arthritis, Musculoskeletal and Skin Diseases, the National Institute on Aging
(R01 AR/AG 41398, N01AG62101, N01AG62103, N01AG62106, 1R01AG032098
and R01 AR 050066), the Canadian Institutes for Health Research (86748), Federal
Program of the Ministry of Education and Science of the Russian Federation
Scientific and Pedagogical Staff of Innovative Russia in 2009–2013 (P-601), the
Federal Program Research and Development of Prior Directions of ScientificTechnological Complex of Russia in 2007–2012 (16.512.11.2032), the Swedish
Research Council (K2010-54X-09894-19-3, 2006-3832, K2010-52X-20229-05-3
and K20006-72X-20155013) the Swedish Foundation for Strategic Research, the
ALF/LUA research grant in Gothenburg, the Lundberg Foundation, the Torsten
and Ragnar Söderberg’s Foundation, the Västra Götaland Foundation, the Göteborg
Medical Society, the Novo Nordisk foundation, University of Athens, Greece
(Kapodistrias 2009), the UK NIHR Musculoskeletal BRU Oxford, the UK NIHR
Nutrition BRU Southampton, The Center for Inherited Disease Research (CIDR),
National Institutes of Health (HHSN268200782096C), the Hong Kong Research
Grant Council (HKU 768610M), The Bone Health Fund of the HKU Foundation,
The KC Wong Education Foundation, Small Project Funding (201007176237),
Matching Grant, Committee of Research and Conference Grants (CRCG) Grant,
the Osteoporosis and Endocrine Research Fund, the Genomics Strategic Research
Theme of The University of Hong Kong, Chinese University of Hong Kong,
the Korea Health 21 Research & Development Project, the Korean Ministry of
Health & Welfare, Republic of Korea (A010252), the Korea Healthcare Technology
Research & Development Project, the Ministry for Health, Welfare and Family
Affairs (A110536), The Netherlands Ministry of Health, Welfare and Sports
Directorate of Long-Term Care, the World Anti-Doping Agency, the Danish
Ministry of Culture, the Institute of Clinical Research of the University of Southern
Denmark, the Chief Scientists Office of the Scottish Government (CZB/4/276),
a Royal Society University Research Fellowship (to J.F.W.), the European Union
Framework Program 6 EUROSPAN project (LSHG-CT-2006-018947), the
European Union’s Seventh Framework Programme (FP7/2007-2013; HEALTH-F22009-223004 PHASE), the Netherlands Organization of Scientific Research NWO
Investments (175.010.2005.011 and 911-03-012), the Research Institute for
Diseases in the Elderly (RIDE2; 014-93-015), the Netherlands Genomics
Initiative/Netherlands Consortium for Healthy Aging (050-060-810), the German
Bundesministerium fuer Forschung und Technology (01 AK 803 A-H and 01 IG
07015 G), the NIHR Biomedical Research Centre (grant to Guys’ and St. Thomas’
Hospitals and King’s College London), the Chronic Disease Research Foundation,
the Canadian Institutes of Health Research, the Canadian Foundation for
Innovation, the Fonds de la Recherche en Santé Québec, The Lady Davis Institute,
the Jewish General Hospital, the Ministère du Développement Economique,
de l’Innovation et de l’Exportation du Quebec, the Swedish Sports Research
Council (87/06), the Swedish Society of Medicine, the Kempe Foundation
(JCK-1021), the Medical Faculty of Umeå University (ALFVLL:968:22-2005,
ALFVL:-937-2006, ALFVLL:223:11-2007 and ALFVLL:78151-2009), the County
Council of Västerbotten (Spjutspetsanslag; VLL:159:33-2007), the US National
Cancer Institute, the Donald W. Reynolds Foundation, the Fondation Leducq, the
Academy of Finland (126925, 121584, 124282, 129378 (Salve), 117787 (Gendi)
and 41071 (Skidi)), the Social Insurance Institution of Finland, Kuopio, Tampere
and Turku University Hospital Medical Funds (9M048 for TeLeht), the Juho
Vainio Foundation, the Paavo Nurmi Foundation, the Finnish Foundation of
VOLUME 44 | NUMBER 5 | MAY 2012
NATURE GENETICS
ARTICLES
Cardiovascular Research, the Finnish Cultural Foundation, the Tampere Tuberculosis
Foundation and the Emil Aaltonen Foundation (K08AR055688 to T.L.). A detailed
list of acknowledgments by study is given in the Supplementary Note. The members
of the GEFOS Consortium mourn the passing of co-author Philip Neil Sambrook,
a good friend, respected colleague and outstanding research scientist in the
prevention, treatment, epidemiology and genetics of osteoporosis.
npg
© 2012 Nature America, Inc. All rights reserved.
AUTHOR CONTRIBUTIONS
This work was done under the auspices of the European Commission–sponsored
Genetic Factors for Osteoporosis (GEFOS) consortium.
Study-specific design and management were performed by U.S., M.A., L.M., J.P.,
S.B., M.L.B., B.M.B., C. Christiansen, C. Cooper, G.D., I.F., M.F., D.G., J.G.-M.,
M. Kähönen, M. Karlsson, J.-M.K., P.K., B.L.L., W.D.L., P.L., Ö.L., R.S.L., J.M.,
D.M., J.M.O., U.P.-K., J.A.R., P.M.R., F. Rousseau, P.E.S., N.L.S.T., R.U., W.V.H., J.V.,
M.T.Z., K.M.G., T.P., D.I.C., S.R.C., R.E., J.A.E., V.G., A.H., R.D.J., G.J., J.W.J.,
K.-T.K., T.L., M. Lorentzon, E.M., B.D.M., G.C.N., M.P., H.A.P.P., R.L.P., O.R., I.R.R.,
P.N.S., P.C.S., A.R.S., F.A.T., C.M.v.D., N.J.W., L.A.C., M.J.E., T.B.H., A.W.C.K.,
B.M.P., J. Reeve, T.D.S., E.A.S., M.C.Z., U.T., C.O., J.B.R., M.A.B., K. Stefansson,
A.G.U., S.H.R., J.P.A.I., D.P.K. and F. Rivadeneira. Study-specific genotyping was
performed by K.E., U.S., E.L.D., L.O., L.V., S.-M.X., A.K.A., D.J.D., S.G., R.K.,
C.K., A.Z.L., J.R.L., S.M., S.M.-B., S.S., S.T., O.T., S.C., E.K., J.M., B.O.-P., Y.S.A.,
E.G., L.H., H.J., T. Kwan, R. Luben, C.M.-G., S.T.P., S. Reppe, J.I.R., J.B.J.v.M.,
D.V., K.M.G., D.I.C., G.R.C., P.D., R.D.J., T.L., Y.L., M. Lorentzon, R.L.P., N.J.W.,
L.A.C., C.O., M.A.B., A.G.U. and F. Rivadeneira. Study-specific phenotyping was
performed by U.S., E.L.D., O.M.E.A., A.M., S.-M.X., N. Alonso, S.K.K., S.G.W.,
A.K.A., T.A., J.R.C., Z.D., N.G.-G., S.G., G.H., L.B.H., K.A.J., G.K., G.S.K., C.K.,
T. Koromila, M. Kruk, M. Laaksonen, A.Z.L., S.H.L., P.C.L., L.M., X.N., J.P.,
L.M.R., K. Siggeirsdottir, O.S., N.M.v.S., J.W., K.Z., M.L.B., C. Christiansen, M.F.,
M. Kähönen, M. Karlsson, J.-M.K., Ö.L., J.M., D.M., B.O.-P., J.M.O., U.P.-K.,
D.M.R., J.A.R., P.M.R., F. Rousseau, W.V.H., J.V., M.C.-B., E.G., T.I., R. Luben,
S. Reppe, G.S., J.B.J.v.M., D.V., F.M.K.W., K.M.G., J.A.C., D.I.C., E.M.D., R.E., J.A.E.,
V.G., A.H., R.D.J., G.J., Y.L., M. Lorentzon, E.M., G.C.N., B.A.O., M.P., H.A.P.P.,
R.L.P., O.R., I.R.R., J. Robbins, P.N.S., C.M.v.D., M.J.E., J. Reeve, E.A.S., M.C.Z.,
C.O., M.A.B., A.G.U., D.P.K. and F. Rivadeneira. Study-specific data analysis were
performed by K.E., U.S., E.E., Y.-H.H., E.L.D., E.E.N., L.O., O.M.E.A., N. Amin,
J.P.K., D.L.K., G.L., C.L., R.L.M., A.M., L.V., D.W., S.-M.X., L.M.Y.-A., H.-F.Z., J.E.,
C.M.K., S.K.K., P.J.L., G.T., J.F.W., V.A., A.K.A., T.A., J.R.C., G.H., L.J.H., C.K.,
T. Koromila, A.Z.L., S.M.-B., T.V.N., M.S.P., J.P., L.M.R., A.V.S., O.S., S.T., S.C., J.M.,
B.O.-P., U.P.-K., R. Li, R. Luben, S. Reppe, J.I.R., A.R.W., Y.Z., S. Raychaudhuri,
D.I.C., J.A.E., R.D.J., T.L., K.N., O.R., D.M.E., D.K., J.B.R., M.A.B., J.P.A.I., D.P.K.
and F. Rivadeneira. Analysis plan design was performed by K.E., E.E., U.S., D.K.,
D.P.K., J.P.A.I. and F. Rivadeneira. K.E., E.E., Y.-H.H. and E.E.N. carried out
meta-analyses. K.E., E.E. and A.R.W. determined gene-by-gene interaction.
Risk modeling and analysis of secondary signals were performed by K.E. and
F. Rivadeneira. Expression QTLs were analyzed by U.S., G.T., E.G., S. Reppe,
K.M.G. and T.P. Y.-H.H. performed functional SNP prediction. GRAIL was carried
out by K.E., E.L.D., D.W. and S. Raychaudhuri. Standardization of phenotype and
genotype replication data sets was performed by K.E., U.S., E.E., E.L.D., L.O.,
G.T., L.H. and C.M.-G. Interpretation of results was carried out by K.E., U.S.,
E.E., Y.-H.H., E.L.D., E.E.N., L.O., O.M.E.A., N. Amin, D.L.K., C.-T.L., R.L.M.,
A.M., L.V., D.W., S.-M.X., L.M.Y.-A., J.E., C.M.K., S.K.K., A.W.C.K., J. Reeve, M.C.Z.,
C.O., D.K., J.B.R., M.A.B., A.G.U., S.H.R., J.P.A.I., D.P.K. and F. Rivadeneira.
The manuscript draft was prepared by K.E., U.S., E.E., Y.-H.H., E.L.D., E.E.N.,
L.O., O.M.E.A., A.M., C.O., D.K., J.B.R., M.A.B., A.G.U., S.H.R., J.P.A.I., D.P.K.
and F. Rivadeneira. The steering committee for GEFOS includes U.S., E.E., U.T.,
A.G.U., S.H.R., J.P.A.I. and F. Rivadeneira.
COMPETING FINANCIAL INTERESTS
The authors declare competing financial interests: details accompany the full-text
HTML version of the paper at http://www.nature.com/naturegenetics/.
Published online at http://www.nature.com/naturegenetics/.
Reprints and permissions information is available online at http://www.nature.com/
reprints/index.html.
1. Burge, R. et al. Incidence and economic burden of osteoporosis-related fractures
in the United States, 2005–2025. J. Bone Miner. Res. 22, 465–475 (2007).
2. Johnell, O. et al. Predictive value of BMD for hip and other fractures. J. Bone
Miner. Res. 20, 1185–1194 (2005).
NATURE GENETICS
VOLUME 44 | NUMBER 5 | MAY 2012
3. Kanis, J.A. et al. The use of clinical risk factors enhances the performance of BMD
in the prediction of hip and osteoporotic fractures in men and women. Osteoporos.
Int. 18, 1033–1046 (2007).
4. Peacock, M., Turner, C.H., Econs, M.J. & Foroud, T. Genetics of osteoporosis.
Endocr. Rev. 23, 303–326 (2002).
5. Ralston, S.H. & Uitterlinden, A.G. Genetics of osteoporosis. Endocr. Rev. 31,
629–662 (2010).
6. Hardy, J. & Singleton, A. Genomewide association studies and human disease.
N. Engl. J. Med. 360, 1759–1768 (2009).
7. Manolio, T.A. Genomewide association studies and assessment of the risk of disease.
N. Engl. J. Med. 363, 166–176 (2010).
8. Richards, J.B. et al. Bone mineral density, osteoporosis, and osteoporotic fractures:
a genome-wide association study. Lancet 371, 1505–1512 (2008).
9. Styrkarsdottir, U. et al. Multiple genetic loci for bone mineral density and fractures.
N. Engl. J. Med. 358, 2355–2365 (2008).
10. Rivadeneira, F. et al. Twenty bone-mineral-density loci identiied by large-scale
meta-analysis of genome-wide association studies. Nat. Genet. 41, 1199–1206
(2009).
11. Styrkarsdottir, U. et al. New sequence variants associated with bone mineral density.
Nat. Genet. 41, 15–17 (2009).
12. Hsu, Y.H. et al. An integration of genome-wide association study and gene expression
proiling to prioritize the discovery of novel susceptibility loci for osteoporosis-related
traits. PLoS Genet. 6, e1000977 (2010).
13. Kung, A.W. et al. Association of JAG1 with bone mineral density and osteoporotic
fractures: a genome-wide association study and follow-up replication studies.
Am. J. Hum. Genet. 86, 229–239 (2010).
14. Duncan, E.L. et al. Genome-wide association study using extreme truncate selection
identiies novel genes affecting bone mineral density and fracture risk. PLoS Genet.
7, e1001372 (2011).
15. Richards, J.B. et al. Collaborative meta-analysis: associations of 150 candidate
genes with osteoporosis and osteoporotic fracture. Ann. Intern. Med. 151, 528–537
(2009).
16. van Meurs, J.B. et al. Large-scale analysis of association between LRP5 and LRP6
variants and osteoporosis. J. Am. Med. Assoc. 299, 1277–1290 (2008).
17. Bagger, Y.Z. et al. Links between cardiovascular disease and osteoporosis in
postmenopausal women: serum lipids or atherosclerosis per se? Osteoporos. Int.
18, 505–512 (2007).
18. Kiel, D.P. et al. Genetic variation at the low-density lipoprotein receptor–related
protein 5 (LRP5) locus modulates Wnt signaling and the relationship of physical
activity with bone mineral density in men. Bone 40, 587–596 (2007).
19. Raychaudhuri, S. et al. Identifying relationships among genomic disease regions:
predicting genes at pathogenic SNP associations and rare deletions. PLoS Genet.
5, e1000534 (2009).
20. Lango Allen, H. et al. Hundreds of variants clustered in genomic loci and biological
pathways affect human height. Nature 467, 832–838 (2010).
21. Brunkow, M.E. et al. Bone dysplasia sclerosteosis results from loss of the SOST
gene product, a novel cystine knot–containing protein. Am. J. Hum. Genet. 68,
577–589 (2001).
22. Guo, Y.F. et al. Polymorphisms of the low-density lipoprotein receptor–related protein 5
(LRP5) gene are associated with obesity phenotypes in a large family-based
association study. J. Med. Genet. 43, 798–803 (2006).
23. Kornak, U. et al. Loss of the ClC-7 chloride channel leads to osteopetrosis in mice
and man. Cell 104, 205–215 (2001).
24. Panagiotou, O.A., Evangelou, E. & Ioannidis, J.P. Genome-wide signiicant
associations for variants with minor allele frequency of 5% or less—an overview: a
HuGE review. Am. J. Epidemiol. 172, 869–889 (2010).
25. Jin, W. et al. Deubiquitinating enzyme CYLD negatively regulates RANK signaling
and osteoclastogenesis in mice. J. Clin. Invest. 118, 1858–1866 (2008).
26. Sundaram, K., Shanmugarajan, S., Rao, D.S. & Reddy, S.V. Mutant p62P392L
stimulation of osteoclast differentiation in Paget’s disease of bone. Endocrinology
152, 4180–4189 (2011).
27. Grundberg, E. et al. Population genomics in a disease targeted primary cell model.
Genome Res. 19, 1942–1952 (2009).
28. Duan, Y., Beck, T.J., Wang, X.F. & Seeman, E. Structural and biomechanical basis
of sexual dimorphism in femoral neck fragility has its origins in growth and aging.
J. Bone Miner. Res. 18, 1766–1774 (2003).
29. Karasik, D. & Ferrari, S.L. Contribution of gender-speciic genetic factors to
osteoporosis risk. Ann. Hum. Genet. 72, 696–714 (2008).
30. Ohlsson, C. et al. Genetic determinants of serum testosterone concentrations in
men. PLoS Genet. 7, e1002313 (2011).
31. Styrkarsdottir, U. et al. European bone mineral density loci are also associated with
BMD in East-Asian populations. PLoS ONE 5, e13217 (2010).
32. Wood, A.R. et al. Allelic heterogeneity and more detailed analyses of known loci
explain additional phenotypic variation and reveal complex patterns of association.
Hum. Mol. Genet. 20, 4082–4092 (2011).
33. Mosekilde, L., Torring, O. & Rejnmark, L. Emerging anabolic treatments in
osteoporosis. Curr. Drug Saf. 6, 62–74 (2011).
34. Hasselblad, V. & Hedges, L.V. Meta-analysis of screening and diagnostic tests.
Psychol. Bull. 117, 167–178 (1995).
499
npg
© 2012 Nature America, Inc. All rights reserved.
ARTICLES
Karol Estrada1–3,142, Unnur Styrkarsdottir4,142, Evangelos Evangelou5,142, Yi-Hsiang Hsu6,7,142,
Emma L Duncan8,9,142, Evangelia E Ntzani5,142, Ling Oei1–3,142, Omar M E Albagha10, Najaf Amin2,
John P Kemp11, Daniel L Koller12, Guo Li13, Ching-Ti Liu14, Ryan L Minster15, Alireza Moayyeri16,17,
Liesbeth Vandenput18, Dana Willner8,19, Su-Mei Xiao20,21, Laura M Yerges-Armstrong22, Hou-Feng Zheng23,
Nerea Alonso10, Joel Eriksson18, Candace M Kammerer15, Stephen K Kaptoge16, Paul J Leo8, Gudmar Thorleifsson4,
Scott G Wilson17,24,25, James F Wilson26,27, Ville Aalto28,29, Markku Alen30, Aaron K Aragaki31, Thor Aspelund32,33,
Jacqueline R Center34–36, Zoe Dailiana37, David J Duggan38, Melissa Garcia39, Natàlia Garcia-Giralt40,
Sylvie Giroux41, Göran Hallmans42, Lynne J Hocking43, Lise Bjerre Husted44, Karen A Jameson45, Rita Khusainova46,47,
Ghi Su Kim48, Charles Kooperberg31, Theodora Koromila49, Marcin Kruk50, Marika Laaksonen51,
Andrea Z Lacroix31, Seung Hun Lee48, Ping C Leung52, Joshua R Lewis24,25, Laura Masi53, Simona Mencej-Bedrac54,
Tuan V Nguyen34,35, Xavier Nogues40, Millan S Patel55, Janez Prezelj56, Lynda M Rose57, Serena Scollen58,
Kristin Siggeirsdottir32, Albert V Smith32,33, Olle Svensson59, Stella Trompet60,61, Olivia Trummer62,
Natasja M van Schoor63, Jean Woo64, Kun Zhu24,25, Susana Balcells65, Maria Luisa Brandi53, Brendan M Buckley66,
Sulin Cheng67,68, Claus Christiansen69, Cyrus Cooper45, George Dedoussis70, Ian Ford71, Morten Frost72,73,
David Goltzman74, Jesús González-Macías75,76, Mika Kähönen77,78, Magnus Karlsson79,80, Elza Khusnutdinova46,47,
Jung-Min Koh48, Panagoula Kollia49, Bente Lomholt Langdahl44, William D Leslie81, Paul Lips82,83,
Östen Ljunggren84, Roman S Lorenc50, Janja Marc54, Dan Mellström18, Barbara Obermayer-Pietsch62,
José M Olmos75,76, Ulrika Pettersson-Kymmer85, David M Reid43, José A Riancho75,76, Paul M Ridker57,86,
François Rousseau41,87,88, P Eline Slagboom3,89, Nelson L S Tang90,91, Roser Urreizti65, Wim Van Hul92,
Jorma Viikari93, María T Zarrabeitia94, Yurii S Aulchenko2, Martha Castano-Betancourt1–3, Elin Grundberg95–97,
Lizbeth Herrera1, Thorvaldur Ingvarsson33,98,99, Hrefna Johannsdottir4, Tony Kwan95,96, Rui Li100,
Robert Luben16, Carolina Medina-Gómez1,2, Stefan Th Palsson4, Sjur Reppe101, Jerome I Rotter102,
Gunnar Sigurdsson33,103, Joyce B J van Meurs1–3, Dominique Verlaan95,96, Frances M K Williams17,
Andrew R Wood104, Yanhua Zhou14, Kaare M Gautvik101,105,106, Tomi Pastinen95,96,107, Soumya Raychaudhuri108,109,
Jane A Cauley110, Daniel I Chasman57,86, Graeme R Clark8, Steven R Cummings111, Patrick Danoy8,
Elaine M Dennison45, Richard Eastell112, John A Eisman34–36, Vilmundur Gudnason32,33, Albert Hofman2,3,
Rebecca D Jackson113,114, Graeme Jones115, J Wouter Jukema60,116,117, Kay-Tee Khaw16, Terho Lehtimäki118–120,
Yongmei Liu121, Mattias Lorentzon18, Eugene McCloskey112,122, Braxton D Mitchell22, Kannabiran Nandakumar6,7,
Geoffrey C Nicholson123, Ben A Oostra124, Munro Peacock125, Huibert A P Pols1,2, Richard L Prince24,25,
Olli Raitakari28,29, Ian R Reid126, John Robbins127, Philip N Sambrook128, Pak Chung Sham129,130,
Alan R Shuldiner22,131, Frances A Tylavsky132, Cornelia M van Duijn2, Nick J Wareham133, L Adrienne Cupples14,134,
Michael J Econs12,125, David M Evans11, Tamara B Harris39, Annie Wai Chee Kung20,21, Bruce M Psaty135–138,
Jonathan Reeve139, Timothy D Spector17, Elizabeth A Streeten22,131, M Carola Zillikens1, Unnur Thorsteinsdottir4,33,143,
Claes Ohlsson18,143, David Karasik6,7,143, J Brent Richards17,23,100,140,143, Matthew A Brown8,143,
Kari Stefansson4,33,143, André G Uitterlinden1–3,143, Stuart H Ralston10,143, John P A Ioannidis5,141,143,
Douglas P Kiel6,7,143 & Fernando Rivadeneira1–3,143
1Department
of Internal Medicine, Erasmus Medical Center, Rotterdam, The Netherlands. 2Department of Epidemiology, Erasmus Medical Center, Rotterdam,
The Netherlands. 3Netherlands Genomics Initiative (NGI)-sponsored Netherlands Consortium for Healthy Aging (NCHA), Leiden, The Netherlands. 4deCODE Genetics,
Reykjavik, Iceland. 5Department of Hygiene and Epidemiology, University of Ioannina School of Medicine, Ioannina, Greece. 6Institute for Aging Research, Hebrew
SeniorLife, Boston, Massachusetts, USA. 7Department of Medicine, Harvard Medical School, Boston, Massachusetts, USA. 8Human Genetics Group, University of
Queensland Diamantina Institute, Brisbane, Queensland, Australia. 9Department of Endocrinology, Royal Brisbane and Women’s Hospital, Brisbane, Queensland,
Australia. 10Rheumatic Diseases Unit, Institute of Genetics and Molecular Medicine, University of Edinburgh, Edinburgh, UK. 11Medical Research Council (MRC)
Centre for Causal Analyses in Translational Epidemiology, University of Bristol, Bristol, UK. 12Department of Medical and Molecular Genetics, Indiana University
School of Medicine, Indianapolis, Indiana, USA. 13Cardiovascular Health Research Unit, University of Washington, Seattle, Washington, USA. 14Department of
Biostatistics, Boston University School of Public Health, Boston, Massachusetts, USA. 15Department of Human Genetics, University of Pittsburgh, Pittsburgh,
Pennsylvania, USA. 16Department of Public Health and Primary Care, University of Cambridge, Cambridge, UK. 17Department of Twin Research and Genetic
Epidemiology, King’s College London, London, UK. 18Centre for Bone and Arthritis Research, Institute of Medicine, Sahlgrenska Academy, University of Gothenburg,
Gothenburg, Sweden. 19Australian Centre for Ecogenomics, University of Queensland, Brisbane, Queensland, Australia. 20Department of Medicine, The University of
Hong Kong, Hong Kong, China. 21Research Centre of Heart, Brain, Hormone and Healthy Aging, The University of Hong Kong, Hong Kong, China. 22Department of
Medicine, Division of Endocrinology, Diabetes and Nutrition, University of Maryland School of Medicine, Baltimore, Maryland, USA. 23Department of Human
Genetics, Lady Davis Institute, McGill University, Montreal, Quebec, Canada. 24School of Medicine and Pharmacology, University of Western Australia, Perth, Western
Australia, Australia. 25Department of Endocrinology and Diabetes, Sir Charles Gairdner Hospital, Perth, Western Australia, Australia. 26Centre for Population Health
Sciences, University of Edinburgh, Edinburgh, UK. 27MRC Human Genetics Unit, MRC Institute of Genetics and Molecular Medicine, University of Edinburgh,
Edinburgh, UK. 28Department of Clinical Physiology and Nuclear Medicine, Turku University Hospital, Turku, Finland. 29Research Centre of Applied and Preventive
Cardiovascular Medicine, University of Turku, Turku, Finland. 30Department of Medical Rehabilitation, Oulu University Hospital and Institute of Health Sciences,
Oulu, Finland. 31Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, Seattle, Washington, USA. 32Icelandic Heart Association, Kopavogur,
Iceland. 33Faculty of Medicine, University of Iceland, Reykjavik, Iceland. 34Osteoporosis and Bone Biology Program, Garvan Institute of Medical Research, Sydney,
500
VOLUME 44 | NUMBER 5 | MAY 2012
NATURE GENETICS
npg
© 2012 Nature America, Inc. All rights reserved.
ARTICLES
New South Wales, Australia. 35Department of Medicine, University of New South Wales, Sydney, New South Wales, Australia. 36Department of Endocrinology, St.
Vincent’s Hospital, Sydney, New South Wales, Australia. 37Department of Orthopaedic Surgery, Medical School University of Thessalia, Larissa, Greece.
38Translational Genomics Research Institute, Phoenix, Arizona, USA. 39Laboratory of Epidemiology, Demography, and Biometry, National Institute on Aging,
Bethesda, Maryland, USA. 40Department of Internal Medicine, Hospital del Mar, Instituto Municipal de Investigación Médica (IMIM), Red Temática de Investigación
Cooperativa en Envejecimiento y Fragilidad (RETICEF), Universitat Autònoma de Barcelona (UAB), Barcelona, Spain. 41Unité de Recherche en Génétique Humaine et
Moléculaire, Centre de Recherche du Centre Hospitalier Universitaire de Québec–Hôpital St-François-d’Assise (CHUQ-HSFA), Québec City, Quebec, Canada.
42Department of Public Health and Clinical Medicine, Umeå Unviersity, Umeå, Sweden. 43Musculoskeletal Research Programme, Division of Applied Medicine,
University of Aberdeen, Aberdeen, UK. 44Department of Endocrinology and Internal Medicine, Aarhus University Hospital, Aarhus C, Denmark. 45MRC Lifecourse
Epidemiology Unit, University of Southampton, Southampton, UK. 46Ufa Scientific Centre of the Russian Academy of Sciences, Institute of Biochemistry and
Genetics, Ufa, Russia. 47Biological Department, Bashkir State University, Ufa, Russia. 48Division of Endocrinology and Metabolism, Asan Medical Center, University
of Ulsan College of Medicine, Seoul, South Korea. 49Department of Genetics and Biotechnology, Faculty of Biology, University of Athens, Athens, Greece.
50Department of Biochemistry and Experimental Medicine, The Children’s Memorial Health Institute, Warsaw, Poland. 51Department of Food and Environmental
Sciences, University of Helsinki, Helsinki, Finland. 52Jockey Club Centre for Osteoporosis Care and Control, The Chinese University of Hong Kong, Hong Kong, China.
53Department of Internal Medicine, University of Florence, Florence, Italy. 54Department of Clinical Biochemistry, University of Ljubljana, Ljubljana, Slovenia.
55Department of Medical Genetics, University of British Columbia, Vancouver, British Columbia, Canada. 56Department of Endocrinology, University Medical Center,
Ljubljana, Slovenia. 57Division of Preventive Medicine, Brigham and Women’s Hospital, Boston, Massachusetts, USA. 58Department of Medicine, University of
Cambridge, Cambridge, UK. 59Department of Surgical and Perioperative Sciences, Umeå Unviersity, Umeå, Sweden. 60Department of Cardiology, Leiden University
Medical Center, Leiden, The Netherlands. 61Department of Gerontology and Geriatrics, Leiden University Medical Center, Leiden, The Netherlands. 62Department of
Internal Medicine, Division of Endocrinology and Metabolism, Medical University Graz, Graz, Austria. 63Department of Epidemiology and Biostatistics, Extramuraal
Geneeskundig Onderzoek (EMGO) Institute for Health and Care Research, Vrije Universiteit (VU) University Medical Center, Amsterdam, The Netherlands.
64Department of Medicine and Therapeutics, The Chinese University of Hong Kong, Hong Kong, China. 65Department of Genetics, University of Barcelona, Centro de
Investigación Biomédica en Red de Enfermedades Raras (CIBERER), Institut de Biomedicina de la Universitat de Barcelona (IBUB), Barcelona, Spain. 66Department
of Pharmacology and Therapeutics, University College Cork, Cork, Ireland. 67Department of Health Sciences, University of Jyväskylä, Jyväskylä, Finland.
68Department of Orthopaedics and Traumatology, Kuopio University Hospital, Kuopio, Finland. 69Center for Clinical and Basic Research (CCBR)-Synarc, Ballerup,
Denmark. 70Department of Nutrition and Dietetics, Harokopio University, Athens, Greece. 71Robertson Center for Biostatistics, University of Glasgow, Glasgow, UK.
72Department of Endocrinology, Odense University Hospital, Odense, Denmark. 73Clinical Institute, University of Southern Denmark, Odense, Denmark. 74Department
of Medicine, McGill University, Montreal, Quebec, Canada. 75Department of Medicine, University of Cantabria, Santander, Spain. 76Department of Internal Medicine,
Hospital Universitario Marqués de Valdecilla and Instituto de Formación e Investigación Marqués de Valdecilla (IFIMAV), Santander, Spain. 77Department of Clinical
Physiology, Tampere University Hospital, Tampere, Finland. 78Department of Clinical Physiology, University of Tampere School of Medicine, Tampere, Finland.
79Clinical and Molecular Osteoporosis Research Unit, Department of Clinical Sciences, Lund University, Malmö, Sweden. 80Department of Orthopaedics, Lund
University, Malmö, Sweden. 81Department of Internal Medicine, University of Manitoba, Winnipeg, Manitoba, Canada. 82Department of Endocrinology, VU University
Medical Center, Amsterdam, The Netherlands. 83Extramuraal Geneeskundig Onderzoek (EMGO) Institute for Health and Care Research, VU University Medical Center,
Amsterdam, The Netherlands. 84Department of Medical Sciences, University of Uppsala, Uppsala, Sweden. 85Department of Pharmacology and Neuroscience, Umeå
University, Umeå, Sweden. 86Harvard Medical School, Boston, Massachusetts, USA. 87Department of Molecular Biology, Medical Biochemistry and Pathology,
Université Laval, Québec City, Quebec, Canada. 88The APOGEE-Net/CanGèneTest Network on Genetic Health Services and Policy, Université Laval, Québec City,
Quebec, Canada. 89Department of Molecular Epidemiology, Leiden University Medical Center, Leiden, The Netherlands. 90Department of Chemical Pathology, The
Chinese University of Hong Kong, Hong Kong, China. 91Li Ka Shing Institute of Health Sciences, The Chinese University of Hong Kong, Hong Kong, China.
92Department of Medical Genetics, University of Antwerp, Antwerp, Belgium. 93Department of Medicine, Turku University Hospital, Turku, Finland. 94Department of
Legal Medicine, University of Cantabria, Santander, Spain. 95Department of Human Genetics, McGill University, Montreal, Quebec, Canada. 96McGill University and
Genome Québec Innovation Centre, Montreal, Quebec, Canada. 97Wellcome Trust Sanger Institute, Hinxton, Cambridge, UK. 98Department of Orthopedic Surgery,
Akureyri Hospital, Akureyri, Iceland. 99Institution of Health Science, University of Akureyri, Akureyri, Iceland. 100Department of Epidemiology and Biostatistics, Lady
Davis Institute, McGill University, Montreal, Quebec, Canada. 101Department of Medical Biochemistry, Oslo University Hospital, Oslo, Norway. 102Medical Genetics
Institute, Cedars-Sinai Medical Center, Los Angeles, California, USA. 103Department of Endocrinology and Metabolism, University Hospital, Reykjavik, Iceland.
104Genetics of Complex Traits, Peninsula College of Medicine and Dentistry, University of Exeter, Exeter, UK. 105Department of Clinical Biochemistry, Lovisenberg
Deacon Hospital, Oslo, Norway. 106Institute of Basic Medical Sciences, University of Oslo, Oslo, Norway. 107Department of Medical Genetics, McGill University
Health Centre, Montreal, Quebec, Canada. 108Division of Genetics and Rheumatology, Brigham and Women’s Hospital, Harvard Medical School, Boston,
Massachusetts, USA. 109Program in Medical and Population Genetics, Broad Institute, Cambridge, Massachusetts, USA. 110Department of Epidemiology, University
of Pittsburgh, Pittsburgh, Pennsylvania, USA. 111California Pacific Medical Center, San Francisco, California, USA. 112National Institute for Health and Research
(NIHR), Musculoskeletal Biomedical Research Unit, University of Sheffield, Sheffield, UK. 113Department of Internal Medicine, The Ohio State University, Columbus,
Ohio, USA. 114Center for Clinical and Translational Science, The Ohio State University, Columbus, Ohio, USA. 115Menzies Research Institute, University of Tasmania,
Hobart, Tasmania, Australia. 116Durrer Center for Cardiogenetic Research, Amsterdam, The Netherlands. 117Interuniversity Cardiology Institute of the Netherlands,
Utrecht, The Netherlands. 118Department of Clinical Chemistry, Tampere University Hospital, Tampere, Finland. 119Department of Clinical Chemistry, University of
Tampere School of Medicine, Tampere, Finland. 120Fimlab, Tampere, Finland. 121Center for Human Genomics, Wake Forest University School of Medicine, WinstonSalem, North Carolina, USA. 122Academic Unit of Bone Metabolism, Metabolic Bone Centre, University of Sheffield, Sheffield, UK. 123Rural Clinical School, The
University of Queensland, Toowoomba, Queensland, Australia. 124Department of Clinical Genetics, Erasmus Medical Center, Rotterdam, The Netherlands.
125Department of Medicine, Indiana University School of Medicine, Indianapolis, Indiana, USA. 126Department of Medicine, University of Auckland, Auckland, New
Zealand. 127Department of Medicine, University of California, Davis, Sacramento, California, USA. 128Kolling Institute, Royal North Shore Hospital, University of
Sydney, Sydney, New South Wales, Australia. 129Department of Psychiatry, The University of Hong Kong, Hong Kong, China. 130Centre for Reproduction, Development
and Growth, The University of Hong Kong, Hong Kong, China. 131Geriatric Research and Education Clinical Center (GRECC), Veterans Administration Medical Center,
Baltimore, Maryland, USA. 132Department of Preventive Medicine, University of Tennessee Health Science Center, Memphis, Tennessee, USA. 133MRC Epidemiology
Unit Box 285, MRC, Cambridge, UK. 134Framingham Heart Study, Framingham, Massachusetts, USA. 135Department of Medicine, University of Washington, Seattle,
Washington, USA. 136Department of Epidemiology, University of Washington, Seattle, Washington, USA. 137Deparment of Health Services, University of Washington,
Seattle, Washington, USA. 138Group Health Research Institute, Group Health Cooperative, Seattle, Washington, USA. 139Medicine Box 157, University of Cambridge,
Cambridge, UK. 140Department of Medicine, Lady Davis Institute, McGill University, Montreal, Quebec, Canada. 141Stanford Prevention Research Center, Stanford
University, Stanford, California, USA. 142These authors contributed equally to this work. 143These authors jointly directed this work. Correspondence should be
addressed to F. Rivadeneira (f.rivadeneira@erasmusmc.nl).
NATURE GENETICS
VOLUME 44 | NUMBER 5 | MAY 2012
501
ONLINE METHODS
npg
© 2012 Nature America, Inc. All rights reserved.
Study design. This study was conducted as part of the GEFOS Consortium, a
coalition of teams of investigators dedicated to identifying the genetic determinants of osteoporosis. The discovery samples comprised 17 GWAS (n = 32,961)
from populations across North America, Europe, East Asia and Australia, with
a variety of epidemiological designs (Supplementary Table 18a) and clinical
characteristics of individuals (Supplementary Table 18b); a subset of these
had fracture information available (Supplementary Table 18c). Subjects from
34 additional studies with BMD data (n = 50,933) were used for replication,
and association with fracture was tested across 50 studies with fracture information, most of which were also used for the BMD analysis (n = 31,016 cases
and 102,444 controls) (Fig. 1 and Supplementary Tables 19a–c and 20a–c).
All studies were approved by their institutional ethics review committees, and
all participants provided written informed consent.
BMD measurements and fracture definition. LS-BMD and FN-BMD were
measured in all cohorts using dual-energy X-ray absorptiometry, following
standard protocols (Supplementary Tables 18b, 19b and 20b). Three clinically
distinct fracture definitions were used: (i) any type, consisting of low-trauma
fractures at any skeletal site (except fingers, toes and skull) occurring after
age 18 years, assessed by X-ray, radiographic report, clinical record, clinical interview and/or questionnaire, (ii) validated non-vertebral, consisting of
fractures occurring after age 50 years, with diagnosis confirmed by hospital
records and/or radiographs, and (iii) radiographic vertebral fractures, from
lateral morphometry scored on X-rays. The first definition is most-inclusive,
whereas the latter two are more stringent fracture definitions that are commonly used in randomized trials35,36. Controls were defined as individuals
without a history of fracture, using for each fracture type the same age limit
categories as for the cases.
Stage 1 genome-wide association analysis. Genotyping and imputation. GWAS
genotyping was performed by each study following standard protocols, and
imputation was then carried out on ~2.5 million SNPs from HapMap37 Phase
2 release 22 using Genome Build 36. Quality control was performed independently for each study. To facilitate meta-analysis, each group performed
genotype imputation with BIM-BAM38, IMPUTE39 or MACH40 software using
genotypes from HapMap Phase 2 release 22 (CEU or Han Chinese in Beijing
(CHB) and Japanese in Tokyo (JPT) as appropriate). HapMap release 21 was
used as a reference for SNPs residing on the X chromosome, and IMPUTE
software was used for imputation. Overall, imputation quality scores for
each SNP were obtained from IMPUTE (proper_info) and MACH (rsq_hat)
statistics. Details of the genotyping platform, genotype quality control
procedures and software for imputation that were used by each study are presented (Supplementary Tables 18d and 19d).
Association analysis with BMD. Each study performed genome-wide association analysis for FN-BMD and LS-BMD, using sex-specific and age-, weightand principal component–adjusted standardized residuals analyzed under an
additive (per allele) genetic model. Analyses of autosomal and X-chromosome
markers were performed separately. The analysis of imputed genotype data
accounted for uncertainty in each genotype prediction by using either the
dosage information from MACH or the genotype probabilities from IMPUTE
and BIM-BAM. Studies used MACH2QTL40 directly or via GRIMP41 (which
uses genotype dosage value as a predictor in a linear regression framework),
SNPTEST39, Merlin42, BIM-BAM or the linear mixed-effects model of the
Kinship and ProbABEL43 (Supplementary Tables 18d and 19d). For analysis of the X chromosome, either SNPTEST or R software was used in each
participating study. We coded ‘effect allele homozygous genotype’ as 2 and
‘other allele homozygous genotype’ as 0 in the genotyped SNPs in men on the
X chromosome. The imputed genotypes were coded as continuous variables
from 0 to 2 to take into account imputation uncertainty. The genomic control
method44 was used to correct the standard error (SE) by the square root of the
genomic inflation factor (λ): SEcorrected = s.e.m. × √λ.
Meta-analysis of the GWAS. Before performing meta-analysis on the
genome-wide association data, SNPs with poor imputation quality scores
(rsq_hat of <0.3 in MACH, proper_info of <0.4 in IMPUTE or a ratio of
NATURE GENETICS
observed-to-expected dosage variance of <0.3 in BIM-BAM) and markers
with a minor allele frequency (MAF) of <1% were excluded from each study.
All individual GWAS were genomic control corrected before meta-analysis44.
Individual study-specific genomic control values ranged from 0.98 to 1.08
(Supplementary Table 18d). A total of 2,483,766 autosomal SNPs were
included in meta-analysis across 17, 16 and 13 studies for FN-BMD (pooled,
women-only and men-only analyses, respectively) and 16, 13 and 12 studies for LS-BMD (pooled, women-only and men-only analyses, respectively).
A total of 76,253 X-linked SNPs were included in meta-analysis across 14, 13
and 10 studies for LS-BMD and FN-BMD (pooled, women-only and menonly analyses, respectively). In our discovery analysis, we chose to implement
a fixed-effects model, as it is generally preferable for the purposes of initial
discovery, where the aim is to screen and identify as many of the true variants
as possible45,46. SNPs present in less than three studies were removed from the
meta-analysis, yielding ~2.2 million SNPs in the final results. The genomic
inflation factors (λ) were 1.11, 1.09 and 1.06 for FN-BMD (pooled, womenonly and men-only analyses, respectively) and 1.13, 1.09 and 1.06 for LS-BMD
(pooled, women-only and men-only analyses, respectively). A second genomic
control correction was applied to the overall meta-analysis results, although
such a second correction is considered overly conservative47. Significance for
BMD association was set at P < 5 × 10−8, and a Bonferroni correction was used
for association with fracture48.
Selection of SNPs for replication. We took forward the most significant 96 SNPs
for replication. With respect to power estimations, after adding 30,000 samples
in stage 2, these variants had a priori power of ≥85% to reach P = 5 × 10−8 in
the meta-analysis. Loci were considered independent when separated by at
least 1 Mb from a top GWAS signal. The 96 variants included the 82 index
SNPs representing each of the 82 loci reaching P < 5 × 10−6 in stage 1, 9 SNPs
that were within the same 2-Mb windows as the 82, which were independent
from the main signals (secondary signals), and the top 5 most-associated SNPs
on the X chromosome (with P < 5 × 10−5).
Association analyses with fracture risk. Effect estimates (odds ratios) for association of allele dosage of the top signals with fracture risk were obtained
from logistic regression models adjusted for age, age2, weight, sex, height and
four principal components. The proportion of the fracture risk explained by
FN-BMD was calculated from the regression coefficients as (βunadjusted −
βBMDadjusted) / βunadjusted in a subset of replication samples for which both
FN-BMD and complete fracture information was available.
Stage 2 replication. Samples and genotyping. Fracture association results
were also obtained for the 82 most-significant SNPs from 54,244 individuals of European ancestry from 7 GWAS (in silico genotyping) that had
not been included in the stage 1 analyses (Supplementary Table 19a–c).
Subjects from 34 studies of the GENOMOS Consortium with BMD and/or
fracture information were studied in replication analysis (Supplementary
Table 3a–c). De novo replication genotyping was performed in the UK
(Kbiosciences), Iceland (deCODE Genetics), Australia (University of
Queensland Diamantina Institute) and the United States (WHI GeCHIP)
using KASPar, Centaurus, OpenArray and iSelect assays, respectively
(Supplementary Note). Minimum genotyping quality control criteria were
defined as sample call rate of >80%, SNP call rate of >90%, Hardy-Weinberg
Equilibrium P value of >1 × 10−4 and MAF of >1%.
Association analyses and meta-analysis. We tested the association between
the 96 SNPs and BMD and fracture risk in each in silico and de novo
stage 2 study separately, as described for the stage 1 studies. We subsequently
performed meta-analysis of effects and standard errors from the stage 2
studies and then carried out a meta-analysis of the summary statistics of
stages 1 and 2 combined using the inverse-variance method in METAL.
At the replication stage, where more than 30 studies were synthesized, we
chose to first assess the underlying heterogeneity, considering both the
Cochran’s Q statistic and the I 2 metric. If the heterogeneity was not
significant, fixed-effects model were applied. If the Cochran’s Q P value
was <0.0005 and I2 was >50%, we used the more conservative randomeffects model.
doi:10.1038/2249
35. Harris, S.T. et al. Effects of risedronate treatment on vertebral and nonvertebral
fractures in women with postmenopausal osteoporosis: a randomized controlled trial.
Vertebral Eficacy With Risedronate Therapy (VERT) Study Group. J. Am. Med.
Assoc. 282, 1344–1352 (1999).
36. Jackson, R.D. et al. Calcium plus vitamin D supplementation and the risk of
fractures. N. Engl. J. Med. 354, 669–683 (2006).
37. Frazer, K.A. et al. A second generation human haplotype map of over 3.1 million
SNPs. Nature 449, 851–861 (2007).
38. Servin, B. & Stephens, M. Imputation-based analysis of association studies:
candidate regions and quantitative traits. PLoS Genet. 3, e114 (2007).
39. Marchini, J., Howie, B., Myers, S., McVean, G. & Donnelly, P. A new multipoint
method for genome-wide association studies by imputation of genotypes. Nat. Genet.
39, 906–913 (2007).
40. Li, Y., Willer, C., Sanna, S. & Abecasis, G. Genotype imputation. Annu. Rev.
Genomics Hum. Genet. 10, 387–406 (2009).
41. Estrada, K. et al. GRIMP: a web- and grid-based tool for high-speed analysis of
large-scale genome-wide association using imputed data. Bioinformatics 25,
2750–2752 (2009).
42. Abecasis, G.R., Cherny, S.S., Cookson, W.O. & Cardon, L.R. Merlin—rapid analysis
of dense genetic maps using sparse gene low trees. Nat. Genet. 30, 97–101
(2002).
43. Aulchenko, Y.S., Struchalin, M.V. & van Duijn, C.M. ProbABEL package for genomewide association analysis of imputed data. BMC Bioinformatics 11, 134 (2010).
44. Devlin, B., Roeder, K. & Wasserman, L. Genomic control, a new approach to geneticbased association studies. Theor. Popul. Biol. 60, 155–166 (2001).
45. Ioannidis, J.P., Thomas, G. & Daly, M.J. Validating, augmenting and reining genomewide association signals. Nat. Rev. Genet. 10, 318–329 (2009).
46. Pereira, T.V., Patsopoulos, N.A., Salanti, G. & Ioannidis, J.P. Discovery properties
of genome-wide association signals from cumulatively combined data sets. Am. J.
Epidemiol. 170, 1197–1206 (2009).
47. Yang, J. et al. Genomic inlation factors under polygenic inheritance. Eur. J. Hum.
Genet. 19, 807–812 (2011).
48. Pe’er, I., Yelensky, R., Altshuler, D. & Daly, M.J. Estimation of the multiple testing
burden for genomewide association studies of nearly all common variants. Genet.
Epidemiol. 32, 381–385 (2008).
49. Reppe, S. et al. Eight genes are highly associated with BMD variation in
postmenopausal Caucasian women. Bone 46, 604–612 (2010).
50. Ge, B. et al. Global patterns of cis variation in human cells revealed by high-density
allelic expression analysis. Nat. Genet. 41, 1216–1222 (2009).
51. Montgomery, S.B. et al. Transcriptome genetics using second generation sequencing
in a Caucasian population. Nature 464, 773–777 (2010).
52. Stranger, B.E. et al. Population genomics of human gene expression. Nat. Genet.
39, 1217–1224 (2007).
53. Grundberg, E. et al. Global analysis of the impact of environmental perturbation
on cis-regulation of gene expression. PLoS Genet. 7, e1001279 (2011).
54. Emilsson, V. et al. Genetics of gene expression and its effect on disease. Nature
452, 423–428 (2008).
55. Zeller, T. et al. Genetics and beyond—the transcriptome of human monocytes and
disease susceptibility. PLoS ONE 5, e10693 (2010).
npg
© 2012 Nature America, Inc. All rights reserved.
Additional analyses. Further analyses were performed for the SNPs carried
forward for replication. Each of these analyses is described in detail in the
Supplementary Note.
In brief, we performed (i) a conditional genome-wide association analysis to
determine whether any of the 82 BMD loci harbored additional independent
signals, (ii) tested gene-by-gene pairwise interactions between these BMD loci,
(iii) assessed within the independent setting of the PERF study (for details on
study design see Supplementary Table 20a–c) the predictive ability derived
from the cumulative effect of the 63 autosomal SNPs associated with BMD with
genome-wide significance in relation to BMD levels and osteoporosis risk and
that of the 16 BMD SNPs also associated with fracture risk in relation to fracture
risk, (iv) identified SNPs with r2 of ≥0.80 with the lead SNP that were potentially
functional (for example, nonsense, non-conservative nonsynonymous, synonymous, exonic splicing, transcription factor binding), using regional imputation
with 1000 Genomes Project data (June 2010 release), (v) tested the relationship
between gene expression profiles from transiliac bone biopsies and BMD in
84 unrelated postmenopausal women49 and examined cis associations between
each of the 55 significant BMD SNPs and expression of nearby genes in different
tissues, including lymphoblastoid cell lines50–52, primary human fibroblasts and
osteoblasts53, adipose tissue54, whole blood54 and circulating monocytes55, and
(vi) evaluated the connectivity and relationships between identified loci using
literature-based annotation with the GRAIL19 statistical strategy.
doi:10.1038/2249
NATURE GENETICS