Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
JAMA. Author manuscript; available in PMC 2012 February 19.
Published in final edited form as:
PMCID: PMC3282142

Large-Scale Analysis of Association Between LRP5 and LRP6 Variants and Osteoporosis



Mutations in the low-density lipoprotein receptor-related protein 5 (LRP5) gene cause rare syndromes characterized by altered bone mineral density (BMD). More common LRP5 variants may affect osteoporosis risk in the general population.


To generate large-scale evidence on whether 2 common variants of LRP5 (Val667Met, Ala1330Val) and 1 variant of LRP6 (Ile1062Val) are associated with BMD and fracture risk.

Design and Setting

Prospective, multicenter, collaborative study of individual-level data on 37 534 individuals from 18 participating teams in Europe and North America. Data were collected between September 2004 and January 2007; analysis of the collected data was performed between February and May 2007. Bone mineral density was assessed by dual-energy x-ray absorptiometry. Fractures were identified via questionnaire, medical records, or radiographic documentation; incident fracture data were available for some cohorts, ascertained via routine surveillance methods, including radiographic examination for vertebral fractures.

Main Outcome Measures

Bone mineral density of the lumbar spine and femoral neck; prevalence of all fractures and vertebral fractures.


The Met667 allele of LRP5 was associated with reduced lumbar spine BMD (n =25 052 [number of participants with available data]; 20-mg/cm2 lower BMD per Met667 allele copy; P=3.3 × 10−8), as was the Val1330 allele (n = 24 812; 14-mg/cm2 lower BMD per Val1330 copy; P=2.6 × 10−9). Similar effects were observed for femoral neck BMD, with a decrease of 11 mg/cm2 (P =3.8 × 10−5) and 8 mg/cm2 (P=5.0×10−6) for the Met667 and Val1330 alleles, respectively (n=25 193). Findings were consistent across studies for both LRP5 alleles. Both alleles were associated with vertebral fractures (odds ratio [OR], 1.26; 95% confidence interval [CI], 1.08–1.47 for Met667 [2001 fractures among 20 488 individuals] and OR, 1.12; 95% CI, 1.01–1.24 for Val1330 [1988 fractures among 20 096 individuals]). Risk of all fractures was also increased with Met667 (OR, 1.14; 95% CI, 1.05–1.24 per allele [7876 fractures among 31 435 individuals)]) and Val1330 (OR, 1.06; 95% CI, 1.01–1.12 per allele [7802 fractures among 31 199 individuals]). Effects were similar when adjustments were made for age, weight, height, menopausal status, and use of hormone therapy. Fracture risks were partly attenuated by adjustment for BMD. Haplotype analysis indicated that Met667 and Val1330 variants both independently affected BMD. The LRP6 Ile1062Val polymorphism was not associated with any osteoporosis phenotype. All aforementioned associations except that between Val1330 and all fractures and vertebral fractures remained significant after multiple-comparison adjustments.


Common LRP5 variants are consistently associated with BMD and fracture risk across different white populations. The magnitude of the effect is modest. LRP5 may be the first gene to reach a genome-wide significance level (a conservative level of significance [herein, unadjusted P<10−7] that accounts for the many possible comparisons in the human genome) for a phenotype related to osteoporosis.

Osteoporosis is distinguished by low bone mineral density (BMD), worsening bone micro-architecture, and increased risk for fractures. Heritability data show that genetic factors determine up to 80% of the variance in BMD,1,2 which is a major predictor of osteoporotic fractures. While the genes that contribute to differences in risk for osteoporosis and osteoporotic fractures are for the most part unknown, it is thought that the risk of developing osteoporosis is dependent on several common gene variants, each with modest effects.3,4

During recent years, variation in the gene coding for low-density lipoprotein receptor–related protein 5 (LRP5) has been implicated in bone mass accrual and susceptibility to osteoporosis. LRP5, and its closely related homologue, LRP6, function as cell-membrane coreceptors for Wnt proteins in the canonical Wnt signaling pathway.5,6 Several lines of evidence suggest that LRP5 may be a key determinant of bone mass. Loss-of-function mutations in the LRP5 gene cause osteoporosis-pseudoglioma syndrome,6 characterized by severe osteoporosis and blindness. Conversely, activating point mutations in this same gene result in high bone mass.7,8 Other LRP5 missense mutations have been described in patients with bone mass disorders, including endosteal hyperostosis, osteopetrosis, and osteosclerosis.9 Various mouse models have also replicated the bone phenotype of mutated LRP5.6,10

Common genetic variations in LRP5 have been proposed as candidates for influencing bone phenotypes at the population level. Some reports have suggested that LRP5 polymorphisms contribute to variation in BMD in the general population,1123 but results are inconclusive. This inconsistency can be explained in part by variations in the examined polymorphisms, the analytical approaches used, and the examined phenotypes. Data on fracture risk are limited, with only 2 reports published so far.14,22 The most frequently studied polymorphisms in this gene are 2 amino acid substitutions (Val667Met and Ala1330Val),13,15,17,18,2125 and there is some additional in vitro evidence that the Ala1330Val variant results in a functional difference of the LRP5 protein.25

Mouse studies have shown that point mutations in the LRP6 gene lead to a low bone mass phenotype.26 While LRP6-deficient mice have early developmental problems that are not compatible with life, mice that carry (heterozygous) mutations in both LRP5 and LRP6 have decreased BMD and limb deformities, which indicates that LRP5 and LRP6 interact in limb development and BMD acquisition.27 A recent report has identified an inherited mutation in LRP6 to be linked to coronary heart disease but also to low-trauma fractures and low BMD.28 In addition, a common protein variant of LRP6 (Ile1062Val) has been found to contribute to fracture risk in elderly men.22 This same variant was recently shown to have functional consequences in vitro29.

The objective of the current study was to examine the contribution of 2 common amino acid substitutions in the LRP5 protein and of 1 amino acid substitution in the LRP6 protein to BMD and risk of fracture using large-scale evidence.

Some scattered studies1223 have tested this association, but results have not been conclusive due to limited sample size. The current collaborative study has the potential to answer this question more definitively because of its large sample size and therefore large power to observe the expected modest associations. In addition, its prospective design, consistent genotyping, and combined analysis of individual-level data diminish bias and the respective noise and heterogeneity that bias might introduce in the outcomes.

We report here on the combined analysis of individual-level data from the full Genetic Markers for Osteoporosis (GENOMOS) consortium, including data from 37 534 individuals. GENOMOS collected standardized data and performed prospective genotyping for these polymorphisms across a large number of teams, only a few of which had previously addressed some of these specific polymorphisms.22,23,25,30,31


Organizational Issues

The GENOMOS project is a large-scale study of candidate gene polymorphisms for osteoporosis outcomes.32 This report includes the 12 study populations included in previous collaborative analyses of other gene polymorphisms.3234 The decision to study the LRP5 and LRP6 polymorphisms in the GENOMOS consortium occurred on June 6, 2004, when the consortium consisted of these 12 European populations.3544 At that time, results were available for 1 study (ERGO [Rotterdam, the Netherlands]), so all other populations were genotyped prospectively. During the course of this study, participants from 6 other teams (4 from Europe,4548 1 from the United States,25 and 1 from Canada30) joined the consortium; these teams genotyped all polymorphisms after they joined the consortium, except for FOS, which had already genotyped the LRP5 polymorphisms.

Participants are still being followed up for all cohorts with incident fracture data, with the exception of some teams in the EPOS multicenter study. The cutoff dates for fracture data in these cohorts are 2002 for APOSS, 2001–2004 for the EPOS centers, 2001 for ERGO, 2002 for LASA, and 2003 for UFO. Data were collected between September 2004 and January 2007, while analysis of the data occurred between February and May 2007. All studied individuals were white, and race/ethnicity was self-reported by study participants.

Details on the design of the 18 studies25,30,3540,4252 are provided in Table 1 and further details in Table 2, eTable 1, and eTable 2. Participants were unrelated in all studies except FAMOS, for which we selected 1 participant per pedigree using random-number selection. Participating teams contributed information on LRP5 and LRP6 genotypes, sex, age, height, weight, menopausal status, use of hormone therapy, activity and ability data (when available), BMD at lumbar spine and femoral neck (in mg/cm2), and fractures. Bisphosphonate use was very rare and thus was not believed to warrant a separate analysis (although available data are reported herein). Smoking status and exercise were not collected in the same format across cohorts. Nevertheless, exercise and ability-adjusted estimates of effect in single studies were obtained whenever possible. The coding of smoking was heterogeneous; thus, as opposed to exercise and ability data, for which the scales were simply different, the smoking categories in each cohort may be overlapping or inconsistent. Therefore, it was believed that adjustment for smoking could not provide meaningful results (although available data are reported herein). For all analyses, participants with missing relevant data were excluded.

Figure 2
Differences in Bone Mineral Density at the Femoral Neck Between Individuals, per Copy of the Risk Allele
Table 1
General Study Characteristics Among the 18 Participating GENOMOS Teams

This study was approved by the institutional review boards of each local institution, and all individuals provided written informed consent to participate in clinical and genetic studies.

BMD Measurements

Bone mineral density was measured by dual-energy x-ray absorptiometry with different devices (Table 1). Measurements used the same reference device within each population. We interpreted results of the analysis of individual-level data for BMD by comparing within-population absolute differences in the mean values of BMD across genotypes. We do not focus on absolute BMD values, because these values may depend on the measuring device.

Fracture Assessment

Fractures were identified either by questionnaire, medical records, or radiographic documentation. Details of fracture assessment and exclusion of age at fracture, fracture type, and trauma type for each cohort are given in Table 2. Longitudinal studies also had data available on incident fractures that had occurred during the follow-up period. Information on incident vertebral fractures included in the analysis was collected with routine surveillance methods using radiographic examination.

Table 2
Assessment Methods and Exclusion Criteria Among the 18 Participating GENOMOS Teams


We genotyped LRP5 Val667Met (dbSNP [] ID rs4988321), LRP5 Ala1330Val (rs3736228), and LRP6 Ile1062Val (rs2302685) single-nucleotide polymorphisms (SNPs) prospectively. These 3 SNPs were the only ones examined in this study. LRP5 Val667Met (rs4988321), LRP5 Ala1330Val (rs3736228), and LRP6 Ile1062Val (rs2302685) polymorphisms were assessed by Taqman, except for the AOS study, for which fluorescence polarization was used for assessment of the LRP5 Val667Met and LRP6 Ile1062Val polymorphisms. We cross-validated genotypes from different laboratories by blinded genotyping of 50 reference samples by all genotyping teams. The coordinating team in Rotterdam evaluated the results and reported any discrepancies in the reference samples in general terms to improve calling of genotypes by failing teams. We repeated genotyping of the reference samples, and teams had to switch genotyping techniques if they were still generating more than 5% errors in the reference samples. In addition, each team checked its own cohort genotyping afterward by reanalyzing at least 5% of their samples selected at random. Genotyping was performed after all prospective radiographic measurements had been performed and had been entered into the databases, so assessment of whether or not a fracture existed would not have been affected by knowledge of genotype.


The main outcomes included BMD of the lumbar spine and femoral neck; all prevalent fractures; and prevalent vertebral fractures by clinical or morphometric criteria.53 We also conducted sensitivity analyses for incident fractures; incident vertebral fractures; and low- and no-trauma fractures. The latter exclude fractures occurring with high trauma, as assessed by the circumstances in which they had occurred, their location, or both. Information on high-and low-trauma fractures was available for 6 of the 18 studies.


Hardy-Weinberg Equilibrium and Haplotype Reconstruction

We performed exact tests for Hardy-Weinberg equilibrium proportions54 using GENEPOP version 4.0.55 We reconstructed haplotypes of the 2 LRP5 polymorphisms using PHASE version 2.0.56

Evaluation of Genetic Effects

All analyses were stratified per study and sex (29 study-sex population strata). For single-SNP analyses we obtained summary estimates using inverse-variance random-effects meta-analysis. For haplotype-based analyses we used mixed models, as described below.

Inverse-Variance Random-Effects Analyses of Individual-Level Data (Single-SNP-Based Analyses)

This is a 2-step approach. Separate regression models were performed in each study-sex population stratum (genetic information was coded using dummy variables, depending on the genetic model assessed). We calculated summary genetic effect as the weighted average of regression coefficients across the different strata using the DerSimonian and Laird random-effects method.57 This method allows for between-strata heterogeneity (dissimilarity) and incorporates it in the calculations. We tested for heterogeneity using the Cochran Q statistic (traditionally considered statistically significant at P<.10)58 and quantified its extent using the I2 statistic (large heterogeneity for values ≥50%).59 Results of single-SNP-based analyses with mixed models were identical and thus not shown.

Mixed Models (Haplotype-Based Analyses)

Linear mixed models were used for continuous outcomes (ie, BMD measurements), and the corresponding generalized linear mixed models were used for binary outcomes (eg, fractures). Population stratum was treated as a random factor and genetic information (haplotypes) as fixed. All models were fitted using maximum likelihood. We relied on a likelihood ratio test to assess whether a model taking into account the genetic information provided better fit (ie, explained data better) than a similar model without the genetic information (eg, a constant-only model).

Choice of Genetic Model and Adjustments

Since there is no strong evidence in favor of a specific genetic model, main analyses used allele-based contrasts. Additional analyses assumed a dominant model for continuous as well as binary outcomes and a “model-free” approach that considers the 3 genotypes as independent factors. For analysis of incident fractures, the binary variable “fracture: yes/ no” was used, and odds ratios (ORs) were calculated and translated into risk ratios (RRs) as described below. There would be no rationale for longitudinal time-to-event analyses (eg, a vertebral fracture identified on a radiograph may have occurred at any point in the period between enrollment and follow-up radiography).

The main analyses were unadjusted for other variables. We also performed secondary adjusted analyses by accounting for age, weight, and height (as continuous variables) in the models. Whenever statistically significant genetic effects were identified, additional adjustments for postmenopausal status and use of hormone therapy among women were undertaken. Fracture-risk analyses were also adjusted for BMD (lumbar spine BMD or femoral neck BMD in separate analyses). The proportion of the fracture risk explained by BMD was calculated from the regression coefficients as (βunadjusted–βadjusted)/βunadjusted. In additional analyses, we tested for interactions of the 2 LRP5 SNPs with age (among all individuals) and postmenopausal status (among women).

Effects at the Population Level

For an indicative population-level estimate, the per-allele OR for the significant associations with fractures and vertebral fractures was also converted into an RR60 considering the median fracture prevalence across the included cohort studies. We calculated the population-attributable fraction using allele frequencies from the median cohort study

Adjustments for Multiple Comparisons

Adjustment for multiple comparisons is generally not favored for hypothesis-validating studies as opposed to discovery studies. Nevertheless, we have illustratively also adjusted the main estimates for the main analyses for 3 polymorphisms × 4 main outcomes (lumbar spine BMD, femoral neck BMD, all fractures, and vertebral fractures) using the Boole-Bonferroni inequality.61 We emphasize that because the 4 main outcomes are correlated and 2 of 3 polymorphisms are also in linkage disequilibrium, Bonferroni adjustments are overly conservative. Conventional statistical significance is claimed for P < .05 adjusted for multiple comparisons. Genome-wide significance is claimed for unadjusted P<10−7.6264 Genome-wide significance accounts for the very large number of polymorphisms and associations thereof that can be tested across the human genome, regardless of whether all or some of them are tested in a study.

Our study is more than 90% powered to detect effect sizes of 0.1 SD in BMD and ORs of 1.20 for fractures and vertebral fractures, if the associations are consistent across different populations. Power would be eroded in the presence of large between-population heterogeneity.65

All statistical analyses were performed using Intercooled Stata 8.2 (StataCorp, College Station, Texas) and R 2.4.1 (the R Foundation for Statistical Computing []). All reported P values are 2-tailed.



Data were collected between September 2004 and January 2007. Analysis of the collected data was performed between February and May 2007. Among the 37 534 participants (24 177 women) analyzed, data on lumbar spine BMD, femoral neck BMD, all fractures, and vertebral fractures existed for 28 073,28 022,35 762, and 22 580 participants, respectively. There were 8932 participants with any fracture and 2146 with vertebral fractures. Basic characteristics and further details of the cohorts are shown in Tables 1 and and22 and eTables 1 and 2. Genotypic information on LRP5 Val667Met, LRP5 Ala1330Val, and LRP6 Ile1062Val was available for 32 720,32423, and 33 038 individuals, respectively. Information on all 3 SNPs was available for 30 989 individuals. The eFigure shows the position of the SNPs in the gene with the haplotypes and its frequencies in the total population studied. The frequency of the Met667 allele ranged from 2% to 8%, of Val1330 from 10% to 19%, and of LRP6 Val1062 from 15% to 23% (for details see eTable 3).

Genotype frequencies were similar across the participating populations (eTable 3). No data set deviated significantly from Hardy-Weinberg equilibrium (P>.05), except for LRP6Ile1062Val (in AOS and APOSS) and LRP5 Ala1330Val (in FLOS and LASA). Exclusion of these data did not affect summary estimates or conclusions (not shown). Linkage disequilibrium between the LRP5 polymorphisms was consistently high across all studies (D’ > 0.85), which allowed inference of haplotypes with high confidence for all cohorts. We consistently identified 3 major haplotypes, and haplotype frequencies were similar across cohorts (eTable 3).

BMD Analyses

Effects of LRP5 Met667 and Val1330

For the LRP5Val667Met and Ala1330Val polymorphisms, highly significant effects on the lumbar spine and femoral neck BMD were observed (Table 3). The BMD effects tended to be larger for Val667Met than for Ala1330Val. The largest effects were found for lumbar spine BMD, which decreased by 20 mg/cm2 (n=25 052 [number of participants with available data]; P=3.3 × 10−8) per copy of Met667 allele and 14mg/cm2 (n=24 812;P=2.6 × 10−9) per copy of Val1330 allele. For the femoral neck, the effects were 11 mg/cm2 (n=25 193;P=3.8 × 10−5) and 8 mg/cm2 (n = 25 026; P=5.0 × 10−6), respectively The aforementioned results remained significant after adjusting for multiple comparisons (the adjusted P values were 4.0 × 10−7, 3.1 × 10−8, 4.6 × 10−4, and 6.0 × 10−5, respectively).

Table 3
Unadjusted Difference in Bone Mineral Density (BMD) for LRP5 Val667Met, LRP5 Ala1330Val, LRP6 Ile1062Val, and LRP5 Haplotypes in Allele-based and Genotype-based Contrastsa

Findings were highly consistent across studies for both LRP5 variants (Figure 1 and Figure 2), and no heterogeneity was detected (P for heterogeneity >.90 for all analyses). Adjustment of the estimates for age, height, and weight and further adjustment for postmenopausal status and use of hormone therapy in women had no major effect on the associations (eTable 4). Teams used very different scales to measure activity or ability as shown in eTable 1, but stratum-specific adjustments using mean-centered scores did not appreciably alter the within-strata estimates of the genetic effects (P > .10 by likelihood ratio test compared with corresponding models without the exercise and ability information). We could not detect a sex difference in the association between LRP5 variants and BMD, but modest sex-specific associations cannot be excluded.

Figure 1
Differences in Bone Mineral Density at the Lumbar Spine Between Individuals, per Copy of the Risk Allele

LRP5 haplotypes were highly significantly associated with lumbar spine BMD and femoral neck BMD overall (P = 9.3 × 10−10 and P=8.4 × 10−6, likelihood ratio tests vs similar models without the LRP5 haplotypes). Using haplotype 1 (Val667-Ala1330) as a reference, each copy of haplotype 2 (Val667-Val1330) and haplotype 3 (Met667-Val1330) was associated with a lower lumbar spine BMD of 10 mg/ cm2 (P = 3.6 × 10−4) and 21 mg/cm2 (P = 1.7 × 10−8), respectively (Table 3). The corresponding decreases in femoral neck BMD were 6 mg/cm2 (P=.003) and 13 mg/cm2 (P = 5.8 × 10−6).

Effects of LRP6 Val1062

The Ile1062Val polymorphism of LRP6 did not show a significant association with BMD (Table 3 and Figures 1 and and2).2). No significant between-study heterogeneity was detected (P> .57 for all analyses). Adjusted analyses showed similar results (eTable 4). There was no significant interaction between the LRP5 haplotypes and the LRP6 Ile1062Val polymorphism on BMD based on likelihood ratio tests vs similar mixed models without the interaction terms.

Fracture Analyses

Effects of LRP5 Met667 and Val1330

Both LRP5 variants were significantly associated with fracture risk (Figure 3, Figure 4, and Table 4). For each Met667 allele, the odds for any prevalent fracture increased by 14% (7876 fractures among 31 435 individuals: OR, 1.14; 95% CI, 1.05–1.24; P=.002), and for prevalent vertebral fractures by 26% (2001 fractures among 20 488 individuals: OR, 1.26; 95% CI, 1.08–1.47; P=.004). The increased risk for prevalent vertebral fractures was found mainly in women (OR, 1.29; 95% CI, 1.08–1.54; P=.004). After adjusting for multiple comparisons, the P values for the association of Met667 with all fractures and vertebral fractures became .02 and .048, respectively.

Figure 3
Odds of Any Fracture, per Copy of the Risk Allele
Figure 4
Odds of Vertebral Fracture, per Copy of the Risk Allele
Table 4
Unadjusted Odds for Fracture Risk per Minor Allele for LRP5 Val667Met, LRP5 Ala1330Val, LRP6 Ile1062Val, and LRP5 Haplotype Allelesa

A borderline significant association was found between the Val1330 variant and overall fracture risk. Participants carrying the Val1330 allele had 6% higher odds for any prevalent fracture (95% CI, 1.01–1.12; P= .02) (analysis of 7802 fractures among 31 199 individuals). Again, a larger effect was seen for vertebral fracture risk (1988 fractures among 20 096 individuals): carriers of the Val1330 allele had an OR of 1.12 (95% CI, 1.01–1.24; P=.03). The effects were no longer significant after adjustments for multiple comparisons (adjusted P values became .30 and .31, respectively).

The median prevalence of all fractures and vertebral fractures among cohort studies was 27% and 2.7%, respectively. The calculated RRs for each copy of the Met667 allele on the population level were 1.10 and 1.25 for all fractures and vertebral fractures, respectively. The corresponding RRs for the Val1330 allele were 1.12 and 1.04. The population-attributable risk for both Val1330 and Met667 was approximately 1% for fractures and 3% for vertebral fractures.

Excluding patients with vertebral fractures, the per-allele ORs for non-vertebral fractures were found to be 1.12 (95% CI, 1.02–1.23) and 1.05 (95% CI, 0.99–1.12) for the LRP5 Met667 and LRP5 Val1330 alleles, respectively. Effects on fractures were unaltered when adjustments were made for age, weight, height, and postmenopausal status; no between-study heterogeneity was detected (P> .33 for all analyses). When adjustments for each individual’s BMD were performed at either the lumbar spine or femoral neck, the formal significance of the overall effects on fracture was lost for most of the associations. The effect of the Val1330 allele on all fractures was unaltered by adjustment with lumbar spine BMD, while approximately 30% of the increased risk for vertebral fractures conferred by the LRP5 Met667 and LRP5 Val1330 alleles was explained by the lumbar spine BMD. Similarly lumbar spine BMD explained approximately one-third of the effect of the Met667 allele on risk of all fractures and vertebral fractures (see also eTable 5).

Overall, LRP5 haplotypes were marginally associated with the risk for all fractures (P=.05; likelihood ratio test) and with the risk for vertebral fractures (P= .02; likelihood ratio test) (Table 4). Using the most common haplotype (haplotype 1, Val667-Ala1330) as reference, carriage of each copy of haplotype 3 (Met667-Val1330) was associated with an increase in the odds for vertebral fractures of 28% (95% CI, 1.08–1.55; P=.006). Associations with any prevalent fracture were not beyond what would be expected by chance.

Effects of LRP6 Val1062

The LRP6 Ile1062Val polymorphism was not associated with fractures overall (Table 4). The CIs excluded 6% differences in the OR for any prevalent fracture between alleles. Adjustment for age (as well as sex, weight, and height) did not appreciably change any of the summary estimates for fracture risk. There was no significant heterogeneity between studies in any analysis.

eTable 6 depicts analyses for incident and low-energy fractures. Results were not conclusive, given the availability of much more limited data.

Sensitivity and Interaction Analyses

There was no evidence for a statistically significant interaction of the LRP5 variants with age for lumbar spine BMD, femoral neck BMD, any prevalent fracture, and any prevalent vertebral fracture. The same was true when interactions with menopausal status were assessed, with a single exception: each copy of the Val1330 allele was associated with an approximately 18-mg/cm2 decrease in femoral neck BMD among premenopausal women but with only a 4-mg/cm2 decrease among postmenopausal women (P=.007 for the Val1330 by menopausal status interaction).


In this large-scale multicenter collaborative study, we obtained evidence that genetic variation of the LRP5 gene is associated with both BMD and fracture risk. The magnitude of the effects was modest but very consistent across studies. The effect size was 14 to 20 mg/ cm2 for lumbar spine and 8 to 11 mg/ cm2 at the femoral neck, which approximately corresponds to a 0.15-SD difference at both sites. Based on the general acceptance that a 1-SD reduction in bone mass doubles the fracture rate,66 an increase of fracture risk of about 15% to 20% is expected. This is similar to the observed effects on fracture, although adjustment for BMD only partly reduced the increase in fracture risk. This could raise the possibility of effects on bone quality, bone dimension, or other nonskeletal determinants of fracture, but also could be due to error in measurement of BMD. Further work will be required to address this point.

Several previous reports have suggested that the association between genetic variation of the LRP5 gene and BMD might be stronger in men compared with women.22,25 We could not find such a sex difference. In fact, for fractures we found a slightly stronger effect for women as compared with men, although power was lower to detect effects for men.

LRP5 may be involved in the establishment of peak bone mass6 and to a lesser extent involved in bone loss. Bone mineral density is substantially affected by age-related bone loss at older ages, so differences in BMD between LRP5 genotype groups might become smaller with age.25 In our study there was no clear influence of age on the magnitude of the association between LRP5 variants and BMD or fracture. For femoral neck BMD, differences between the Ala1330Val genotypes were larger in premenopausal women compared with postmenopausal women, which could indicate that the effect of LRP5 variants is largely seen on peak bone mass. However, this was not observed for lumbar spine BMD and the Ala1330Val variant or with the Val667Met polymorphism for any of the outcomes. Even with such large-scale evidence, the presence or absence of interaction effects should be interpreted very cautiously.

The 2 polymorphisms in LRP5 are each strongly associated with BMD. Although these polymorphisms are in strong linkage disequilibrium, the risk alleles were separated in 2 haplotypes: haplotype 2, carrying the common Val667 and the Val1330 risk allele, and haplotype 3, carrying risk alleles for both Met667 and Val1330. Haplotypes 2 and 3 were both associated with BMD while haplotype 3 was more strongly associated, which suggests that both variants have distinct effects. However, we cannot exclude that the polymorphisms are in linkage disequilibrium with 1 or more other causative polymorphisms rather than having an effect themselves.

The 2 studied LRP5 variants are situated in different domains of the protein. The Val667Met polymorphism is localized at the top of the third propeller module in the receptor extracellular domain. This domain is thought to be involved in binding of the Wnt-inhibitor Dkk1, so perhaps binding efficacy of this inhibitor is changed in the Met667 variant. The Ala1330Val polymorphism lies within a second low-density lipoprotein (LDL) receptor domain of LRP5. The function of this region in LRP5 is unknown, but similar domains in the LDL receptor domain interact with the propeller domains.67 Therefore, variations in the LDL receptor domains, such as Ala1330Val, may still alter protein function. Indeed, a recent report showed in vitro that Wnt-signaling capacity of the LRP5 Val1330 variant was decreased compared to the Ala1330 variant.25

The strengths of our consortium analysis include the very large sample size, consistency across cohorts, lack of publication bias within the consortium due to its prospective design, and analysis of individual-level data, which allows standardized statistical analyses across participating teams.

In particular, we focused on validation of genotyping to minimize genotyping errors and aimed at standardized definitions for the outcomes. Limitations arise due to ascertainment of fractures, which differed across participating studies. This could introduce some unavoidable heterogeneity in the analyses. Another potential limitation is due to missing data in some cohorts. In addition, our results might not pertain to Asian and/or African populations, since we only examined white populations.

Our findings demonstrate that the modest effects of common genetic variations in complex diseases can be effectively addressed through large consortia and coordinated, standardized analysis. Such effects might be missed by smaller and potentially underpowered individual studies. This prospective collaborative study with individual level-data of 37 534 participants shows an effect of LRP5 genetic variation on both BMD and risk of fracture. While some other common variants have been associated previously with osteoporosis phenotypes with large-scale evidence,1719 this may be the first time that an association in this field crosses the threshold of genome-wide statistical significance (P<10−7). Given the large number of polymorphisms that can be tested in the human genome, it has been argued that to fully account for all these possible comparisons (regardless of whether all of them are made), a very conservative threshold is needed.6264 Although the magnitude of the effect was modest, the effect was very consistent in different populations and independent of sex or age. This suggests a role for LRP5 in determining BMD and fracture risk throughout life in the general population. Although any single marker explains only a small portion of the phenotype risk, identification of several such osteoporosis risk variants may eventually help in improving clinical prediction. Single genetic risk variants such as LRP5 variants may also offer useful insights about mechanisms and pathways that may be useful in drug development.


Funding/Support: The European Commission supported this study under grant QLK6-CT-2002-02629. The AOS study was supported by WADA (World Anti Doping Agency), Novo Nordisk A/S, Pfizer A/S, Danish Ministry of Culture, Overlægerådets Legatudvalg–Odense University Hospital, the Fonds voor wetenschappelijk onderzoek (FWO), and Institute for the Promotion of Innovation through Science and Technology in Flanders (IWT-Vlaanderen). The EPOLOS study is supported by Ministry of Health and State Committee for Scientific Research grant no 2483/C. P05-4/99 and Ministry of Scientific Research and Information Technology grant no 2PO5A14228. The EPOS study was financially supported by a European Union Concerted Action Grant under Biomed-1 (BMH1CT920182) and also European Union grants C1PDCT925102, ERBC1PDCT 930105 & 940229. The ERGO study (Rotterdam Study) is supported by the Erasmus Medical Center and Erasmus University, Rotterdam; the Netherlands organization for scientific research (NWO); the Netherlands Organization for the Health Research and Development (ZonMw); the Research Institute for Diseases in the Elderly (RIDE); the Ministry of Education, Culture and Science; the Ministry for Health, Welfare and Sports; the European Commision (DG XII); and the Municipality of Rotterdam. The Framingham Osteoporosis Study (FOS) is supported by the Framingham Heart Study, which is funded by the National Heart, Lung, and Blood Institute of the National Institutes of Health (N01-HC-25195) as well as by grants from the National Institute of Arthritis and Musculoskeletal and Skin Diseases and the National Institute on Aging (AR/AG41398, and R01-AR050066). The Longitudinal Aging Study Amsterdam (LASA) is funded by the Ministry of Health, Welfare and Sports of the Netherlands and by ZonMW. The Umeå Fracture and Osteoporosis study (UFO) is supported by the Swedish Research Council (K2006-72X-20155013), the Swedish Sports Research Council (87/06), the Swedish Society of Medicine, and by grants from the Medical Faculty of Umeå University and from the County Council of Vasterbotten.

Role of the Sponsor: The European Commission had no role in design and conduct of the study; the collection, management, analysis, and interpretation of the data; or the preparation, review, or approval of the manuscript.

Additional Investigators and Study Group Members: GENOMOS: University of Ioannina School of Medicine: Despina G. Contopoulos-Ioannidis; Department of Internal Medicine, Erasmus MC, Rotterdam: Pascal Arp, Mila Jhamai, Hans van Leeuwen; Institute of Medical Sciences, University of Aberdeen Medical School, Aberdeen, UK: Omar M.E. Albagha, Helen Macdonald, Alison Stewart, Amelia Bassiti; Strangeways Research Laboratory, Cambridge University, Cambridge, UK: Alison M. Dunning; Aarhus University Hospital, Denmark: Mette Carstens, Liselotte Stenkjaer, Nuria Gonzalez Bofill; University of Florence Medical School, Florence, Italy: Annalisa Tanini, Alberto Falchetti; University of Barcelona, CIBERER, IBUB, Barcelona, Spain: Daniel Grinberg and Mariona Bustamante; Hospital del Mar, Barcelona, Spain: Adolfo Diez-Perez, Leonardo Mellibovsky, Susana Jurado; Department of Internal Medicine, Medical University, Graz, Austria: Daniela Walter, Ursula Hartl, Markus Gugatschka, Christine Bonelli, Harald Dobnig, Astrid Fahrleitner-Pammer; Department of Biochemistry and Experimental Medicine, The Children’s Memorial Health Institute, Warsaw, Poland: Elzbieta Karczmarewicz, Pawel Pludowski; Department of Medical Genetics, University of Antwerp, Antwerp, Belgium: Sigri Beckers, Armand Peeters, Elke Piters, Wendy Balemans; Umeå University Hospital, Umeå, Sweden: Olle Svensson, Peter Nordström. AOS Study Group: Torben L. Nielsen, Kristian Wraae, Lise Bathum, Claire Brasen, Claus Hagen, Marianne Andersen, Bo Abrahamsen (Odense, Denmark). APOSS Study Group: Claire Parsons, Stuart Bear, Rosie Farmer (Aberdeen, UK). DOPS Study Group: Jens-Erik Beck Jensen (Hvidovre, Denmark), Pia Eiken (Hilleroed, Denmark). EPOLOS Study Group: Jacek Lukaszkiewicz, Piotr Bilinski, Edward Czerwinski, Andrzej Lewinski, Ewa Marcinowska-Suchowierska, Andrzej Milewicz, Marek Spaczynski, Maciej Jaworski (Poland). EPOS Study Group: Raniero Nuti (Siena, Italy), Simeon Grazio (Zagreb, Croatia), Thomas Miazgowski (Szczecin, Poland), Steven R. Boonen (Leuven, Belgium), Pavol Masaryk (Piestany, Slovakia), Jan J. Stepan (Prague, Czech Republic), Antonio Lopes Vaz (Porto, Portugal), Jacome Brughes Armas (Azores, Portugal), Jorge Cannata (Oviedo, Spain), Roman Perez Cano (Sevilla, Spain), Christopher Todd and Kay-Tee Khaw (Norfolk, Cambridge and Harrow, UK), Jose A. da Silva (Coimbra, Portugal), Ashok Bhalla (Bath, UK), Gyula Poor (Budapest, Hungary), George Lyritis (Athens, Greece), Terrence W. O’Neill, Mark Lunt (Cambridge and Manchester UK, Coordination). ERGO/ Rotterdam Study Group: Cornelia M. van Duijn, Paulus J. de Jong, Monique M. Breteler, Bruno H. Stricker, Jacqueline C. Witteman (Rotterdam, the Netherlands). FAMOS Study Group: Juliet Compston (University of Cambridge, Cambridge, UK), Cyrus Cooper (University of Southampton, Southampton, UK), Emma Duncan (Nuffield Orthopaedic Center, Oxford, UK), Richard Keen, (University College, London, UK), Alastair McLellan (University of Glasgow, Glasgow, UK), John Wass (Nuffield Orthopaedic Center, Oxford, UK). Framingham Osteoporosis Study Group: L. Adrienne Cupples and Serkalem Demissie (Department of Biostatistics, Boston University School of Public Health, Boston, Massachusetts), Alma Imamovic (Department of Neurology and Framingham Heart Study Genetics Laboratory, Boston, Massachusetts). LASA Study Group: Ebbo Dekema, Huib van Essen, Saskia Pluijm, Dorly Deeg. MrOS-Sweden Study Group: Hans Mallmin, Elin Grundberg (Uppsala University, Uppsala, Sweden), Anna Holmberg (Lund University, Malmö , Sweden), Eric Orwoll (Oregon Health & Science University, Portland, Oregon). UFO Study Group: Åsa Ågren, Hubert Sjödin, Kerstin Enquist, Ingvar Bergdahl, Ulrica Bergström (Umeå University, Umeå, Sweden).


Author Contributions: Drs Ioannidis and Trikalinos had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Drs van Meurs and Trikalinos contributed equally to this article.

Study concept and design: van Meurs, Ralston, Balcells, Langdahl, Lorenc, Hofman, Kaptoge, Lorentzon, Mellström, Mosekilde, Nogues, Pols, Reeve, Weber, Ioannidis, Uitterlinden.

Acquisition of data: van Meurs, Ralston, Balcells, Brandi, Brixen, Kiel, Langdahl, Lips, Ljunggren, Obermayer-Pietsch, Ohlsson, Pettersson, Reid, Rousseau, Scollen, Van Hul, Agueda, Åkesson, Benevolenskaya, Ferrari, Hallmans, Husted, Kruk, Kaptoge, Karasik, Karlsson, Lorentzon, Masi, McGuigan, Mellström, Mosekilde, Nogues, Reeve, Renner, Rivadeneira, van Schoor, Weber, Ioannidis, Uitterlinden.

Analysis and interpretation of data: van Meurs, Trikalinos, Ralston, Brixen, Pettersson, Scollen, Agueda, Kruk, Kaptoge, Rivadeneira, Weber, Ioannidis, Uitterlinden.

Drafting of the manuscript: van Meurs, Ralston, Lorenc, Karlsson, Lorentzon, Nogues, Ioannidis, Uitterlinden.

Critical revision of the manuscript for important intellectual content: Trikalinos, Ralston, Balcells, Brandi, Brixen, Kiel, Langdahl, Lips, Ljunggren, Obermayer-Pietsch, Ohlsson, Pettersson, Reid, Rousseau, Scollen, Van Hul, Agueda, Åkesson, Benevolenskaya, Ferrari, Hallmans, Hofman, Husted, Kruk, Kaptoge, Karasik, Lorentzon, Masi, McGuigan, Mellström, Mosekilde, Nogues, Pols, Reeve, Renner, Rivadeneira, van Schoor, Weber, Ioannidis, Uitterlinden.

Statistical analysis: van Meurs, Trikalinos, Rivadeneira, Ioannidis.

Obtained funding: Ralston, Brandi, Brixen, Kiel, Langdahl, Lips, Ohlsson, Pettersson, Rousseau, Van Hul, Åkesson, Hofman, Karlsson, Mellström, Mosekilde, Reeve, Weber, Uitterlinden.

Administrative, technical, or material support: van Meurs, Ralston, Kiel, Ljunggren, Obermayer-Pietsch, Pettersson, Reid, Scollen, Van Hul, Agueda, Åkesson, Benevolenskaya, Hallmans, Husted, Kruk, Kaptoge, Karasik, Karlsson, Lorentzon, Masi, McGuigan, Mellström, Mosekilde, Reeve, Renner, Rivadeneira, van Schoor, Weber, Uitterlinden.

Study supervision: Ralston, Balcells, Lips, Reid, Ferrari, Hofman, Mellström, Nogues, Pols, Reeve, Weber, Uitterlinden.

Financial Disclosures: Dr Ralston reported receiving consultancies from Novartis, Proctor & Gamble, and Sanofi-Aventis; serving as a lecturer at meetings sponsored by Merck, GlaxoSmithKline, Roche, Novartis, Proctor & Gamble, Sanofi-Aventis, and Servier; and receiving research grants from Novartis, Proctor & Gamble, and Wyeth. Dr Brixen reported serving on the advisory boards of Eli Lilly, Nycomed, Servier, Novartis, and Amgen and on the speakers bureau of Servier; and conducting clinical trials for Eli Lilly, Novartis, and Osteologix. Dr Langdahl reported serving on the advisory boards of Eli Lilly, Nycomed, Servier, Novartis, and Amgen and on the speakers bureau of Roche; and conducting clinical trials for Pfizer. No other disclosures were reported. GENOMOS is supported by the reported European Union grant and the specific teams that participated have additional grants as specified herein; these grants have been used for funding largely personnel salaries and laboratory expenses. None of the additional investigators or study group members listed below received compensation for services involving preparation or writing of the manuscript.

Previous Presentations: The material presented in this article was presented as an abstract at The American Society for Bone and Mineral Research 29th Annual Meeting; September 19, 2007; Honolulu, Hawaii.

Additional Information: The eFigure and eTables 1–6 are available at

Additional Contributions: We especially thank Professor Olof Johnell, who initiated the participation of MrOs in GENOMOS but who unfortunately passed away during this study.


1. Pocock NA, Eisman JA, Hopper JL, Yeates MG, Sambrook PN, Eberl S. Genetic determinants of bone mass in adults: a twin study. J Clin Invest. 1987;80(3):706–710. [PMC free article] [PubMed]
2. Evans RA, Marel GM, Lancaster EK, Kos S, Evans M, Wong SY. Bone mass is low in relatives of osteoporotic patients. Ann Intern Med. 1988;109(11):870–873. [PubMed]
3. Lohmueller KE, Pearce CL, Pike M, Lander ES, Hirschhorn JN. Meta-analysis of genetic association studies supports a contribution of common variants to susceptibility to common disease. Nat Genet. 2003;33(2):177–182. [PubMed]
4. Uitterlinden AG, van Leeuwen JPTM, Pols HAP. Genetics and genomics of osteoporosis. In: Feldman D, Marcus R, Kelsey J, editors. Osteoporosis. San Diego, CA: Academic Press; 2001. pp. 639–667.
5. Tamai K, Semenov M, Kato Y, et al. LDL-receptor-related proteins in Wnt signal transduction. Nature. 2000;407(6803):530–535. [PubMed]
6. Gong Y, Slee RB, Fukai N, et al. LDL receptor-related protein 5 (LRP5) affects bone accrual and eye development. Cell. 2001;107(4):513–523. [PubMed]
7. Boyden LM, Mao J, Belsky J, et al. High bone density due to a mutation in LDL-receptor-related protein 5. N Engl J Med. 2002;346(20):1513–1521. [PubMed]
8. Little RD, Carulli JP, Del Mastro RG, et al. A mutation in the LDL receptor-related protein 5 gene results in the autosomal dominant high-bone-mass trait. Am J Hum Genet. 2002;70(1):11–19. [PubMed]
9. Van Wesenbeeck L, Cleiren E, Gram J, et al. Six novel missense mutations in the LDL receptor-related protein 5 (LRP5) gene in different conditions with an increased bone density. Am J Hum Genet. 2003;72 (3):763–771. [PubMed]
10. Babij P, Zhao W, Small C, et al. High bone mass in mice expressing a mutant LRP5 gene. J Bone Miner Res. 2003;18(6):960–974. [PubMed]
11. Ferrari SL, Deutsch S, Antonarakis SE. Pathogenic mutations and polymorphisms in the lipoprotein receptor-related protein 5 reveal a new biological pathway for the control of bone mass. Curr Opin Lipidol. 2005;16(2):207–214. [PubMed]
12. Urano T, Shiraki M, Ezura Y, et al. Association of a single-nucleotide polymorphism in low-density lipoprotein receptor-related protein 5 gene with bone mineral density. J Bone Miner Metab. 2004;22(4):341–345. [PubMed]
13. Koay MA, Woon PY, Miles L, et al. Influence of LRP5 polymorphisms on normal variation in BMD [published online ahead of print July 7, 2004] J Bone Miner Res. 2004;19(10):1619–1627. doi: 10.1359/JBMR.040704. [PubMed] [Cross Ref]
14. Bollerslev J, Wilson SG, Dick IM, et al. LRP5 gene polymorphisms predict bone mass and incident fractures in elderly Australian women. Bone. 2005;36 (4):599–606. [PubMed]
15. Ezura Y, Nakajima T, Urano T, et al. Association of a single-nucleotide variation (A1330V) in the low-density lipoprotein receptor-related protein 5 gene (LRP5) with bone mineral density in adult Japanese women. Bone. 2007;40(4):997–1005. [PubMed]
16. Koay MA, Tobias JH, Leary SD, Steer CD, Vilarino-Guell C, Brown MA. The effect of LRP5 polymorphisms on bone mineral density is apparent in childhood. Calcif Tissue Int. 2007;81(1):1–9. [PMC free article] [PubMed]
17. Koller DL, Ichikawa S, Johnson ML, et al. Contribution of the LRP5 gene to normal variation in peak BMD in women. J Bone Miner Res. 2005;20(1):75–80. [PubMed]
18. Saarinen A, Valimaki VV, Valimaki MJ, et al. The A1330V polymorphism of the low-density lipoprotein receptor-related protein 5 gene (LRP5) associates with low peak bone mass in young healthy men. Bone. 2007;40(4):1006–1012. [PubMed]
19. Urano T, Shiraki M, Narusawa K, et al. Q89R polymorphism in the LDL receptor-related protein 5 gene is associated with spinal osteoarthritis in postmeno-pausal Japanese women. Spine. 2007;32(1):25–29. [PubMed]
20. Xiong DH, Lei SF, Yang F, et al. Low-density li-poprotein receptor-related protein 5 (LRP5) gene polymorphisms are associated with bone mass in both Chinese and whites. J Bone Miner Res. 2007;22(3):385–393. [PubMed]
21. Zhang ZL, Qin YJ, He JW, et al. Association of polymorphisms in low-density lipoprotein receptor-related protein 5 gene with bone mineral density in postmenopausal Chinese women. Acta Pharmacol Sin. 2005;26(9):1111–1116. [PubMed]
22. van Meurs JB, Rivadeneira F, Jhamai M, et al. Common genetic variation of the low-density lipoprotein receptor-related protein 5 and 6 genes determines fracture risk in elderly white men. J Bone Miner Res. 2006;21(1):141–150. [PubMed]
23. Brixen K, Beckers S, Peeters A, et al. Polymorphisms in the low-density lipoprotein receptor-related protein 5 (LRP5) gene are associated with peak bone mass in non-sedentary men: results from the Odense Androgen Study. Calcif Tissue Int. 2007;81(6):421–429. [PMC free article] [PubMed]
24. Ferrari SL, Deutsch S, Choudhury U, et al. Poly­morphisms in the low-density lipoprotein receptor-related protein 5 (LRP5) gene are associated with variation in vertebral bone mass, vertebral bone size, and stature in whites. Am J Hum Genet. 2004;74(5):866–875. [PubMed]
25. Kiel DP, Ferrari SL, Cupples LA, 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. 2007;40(3):587–596. [PMC free article] [PubMed]
26. Kokubu C, Heinzmann U, Kokubu T, et al. Skeletal defects in ringelschwanz mutant mice reveal that Lrp6 is required for proper somitogenesis and osteogenesis. Development. 2004;131(21):5469–5480. [PubMed]
27. Holmen SL, Giambernardi TA, Zylstra CR, et al. Decreased BMD and limb deformities in mice carrying mutations in both Lrp5 and Lrp6. J Bone Miner Res. 2004;19(12):2033–2040. [PubMed]
28. Mani A, Radhakrishnan J, Wang H, et al. LRP6 mutation in a family with early coronary disease and metabolic risk factors. Science. 2007;315(5816):1278–1282. [PMC free article] [PubMed]
29. De Ferrari GV, Papassotiropoulos A, Biechele T, et al. Common genetic variation within the low-density lipoprotein receptor-related protein 6 and late-onset Alzheimer’s disease. Proc Natl Acad Sci USA. 2007;104(22):9434–9439. [PubMed]
30. Giroux S, Elfassihi L, Cardinal G, Laflamme N, Rousseau F. LRP5 coding polymorphisms influence the variation of peak bone mass in a normal population of French-Canadian women. Bone. 2007;40(5):1299–1307. [PubMed]
31. Grundberg E, Lau EM, Lorentzson M, et al. Large-scale association study between two coding LRP5 gene polymorphisms and bone phenotypes and fractures in men [published online ahead of print November 17, 2007] Osteoporos Int. doi: 10.1007/s00198–007-0512-z. [PubMed] [Cross Ref]
32. Ioannidis JP, Ralston SH, Bennett ST, et al. Differential genetic effects of ESR1 gene polymorphisms on osteoporosis outcomes. JAMA. 2004;292(17):2105–2114. [PubMed]
33. Uitterlinden AG, Ralston SH, Brandi ML, et al. The association between common vitamin D receptor gene variations and osteoporosis: a participant-level meta-analysis [published correction appears in Ann Intern Med. 2006;145(12):936] Ann Intern Med. 2006;145(4):255–264. [PubMed]
34. Ralston SH, Uitterlinden AG, Brandi ML, et al. Large-scale evidence for the effect of the COLIA1 Sp1 polymorphism on osteoporosis outcomes: the GENOMOS study. PLoS Med. 2006;3(4):e90. [PubMed]
35. Bustamante M, Nogues X, Enjuanes A, et al. COL1A1, ESR1, VDR and TGFB1 polymorphisms and haplotypes in relation to BMD in Spanish postmenopausal women. Osteoporos Int. 2007;18(2):235–243. [PubMed]
36. Gugatschka M, Dobnig H, Fahrleitner-Pammer A, et al. Molecularly-defined lactose malabsorption, milk consumption and anthropometric differences in adult males. QJM. 2005;98(12):857–863. [PubMed]
37. Langdahl BL, Carstens M, Stenkjaer L, Eriksen EF. Polymorphisms in the osteoprotegerin gene are associated with osteoporotic fractures. J Bone Miner Res. 2002;17(7):1245–1255. [PubMed]
38. McGuigan FE, Macdonald HM, Bassiti A, et al. Large-scale population-based study shows no association between common polymorphisms of the TGFB1 gene and BMD in women. J Bone Miner Res. 2007;22(2):195–202. [PubMed]
39. Mosekilde L, Hermann AP, Beck-Nielsen H, Charles P, Nielsen SP, Sorensen OH. The Danish Osteoporosis Prevention Study (DOPS): project design and inclusion of 2000 normal perimenopausal women. Maturitas. 1999;31(3):207–219. [PubMed]
40. Obermayer-Pietsch BM, Bonelli CM, Walter DE, et al. Genetic predisposition for adult lactose intolerance and relation to diet, bone density, and bone fractures. J Bone Miner Res. 2004;19(1):42–47. [PubMed]
41. Hofman A, Breteler MM, van Duijn CM, et al. The Rotterdam Study: objectives and design update. Eur J Epidemiol. 2007;22(11):819–829. [PMC free article] [PubMed]
42. EPOS. Incidence of vertebral fracture in Europe: results from the European Prospective Osteoporosis Study (EPOS) J Bone Miner Res. 2002;17(4):716–724. [PubMed]
43. Deeg DJ, van Tilburg T, Smit JH, de Leeuw ED. Attrition in the Longitudinal Aging Study Amsterdam: the effect of differential inclusion in side studies. J Clin Epidemiol. 2002;55(4):319–328. [PubMed]
44. Ralston SH, Galwey N, MacKay I, et al. Loci for regulation of bone mineral density in men and women identified by genome wide linkage scan: the FAMOS study. Hum Mol Genet. 2005;14(7):943–951. [PubMed]
45. Hallmans G, Agren A, Johansson G, et al. Cardiovascular disease and diabetes in the Northern Sweden Health and Disease Study Cohort—evaluation of risk factors and their interactions. Scand J Public Health Suppl. 2003;61:18–24. [PubMed]
46. Lorentzon M, Swanson C, Andersson N, Mellstrom D, Ohlsson C. Free testosterone is a positive, whereas free estradiol is a negative, predictor of cortical bone size in young Swedish men: the GOOD study. J Bone Miner Res. 2005;20(8):1334–1341. [PubMed]
47. Mellström D, Johnell O, Ljunggren O, et al. Free testosterone is an independent predictor of BMD and prevalent fractures in elderly men: MrOS Sweden. J Bone Miner Res. 2006;21(4):529–535. [PubMed]
48. Abrahamsen B, Jorgensen HL, Nielsen TL, et al. MTHFR c.677C>T polymorphism as an independent predictor of peak bone mass in Danish men—results from the Odense Androgen Study. Bone. 2006;38 (2):215–219. [PubMed]
49. Kruk M, Jaworsky M, Lukaszkiewicz J, et al. Effect of Pvull and Xbal polymorphisms of ESRI gene on body height in women depends on menarche—the EPO-LOS study. Calcif Tissue Int. 2007;80:S100.
50. Hofman A, Breteler MM, van Duijn CM, et al. The Rotterdam Study: objectives and design update. Eur J Epidemiol. 2007;22(11):819–829. [PMC free article] [PubMed]
51. Dobnig H, Piswanger-Sölkner JC, Roth M, et al. Type 2 diabetes mellitus in nursing home patients: effects on bone turnover, bone mass and fracture risk. J Clin Endocrinol Metab. 2006;91(9):3355–3363. [PubMed]
52. Masi L, Becherini L, Gennari L, et al. Polymorphism of the aromatase gene in postmenopausal Italian women: distribution and correlation with bone mass and fracture risk. J Clin Endocrinol Metab. 2001;86(5):2263–2269. [PubMed]
53. McCloskey EV, Spector TD, Eyres KS, et al. The assessment of vertebral deformity: a method for use in population studies and clinical trials. Osteoporos Int. 1993;3(3):138–147. [PubMed]
54. Rousset F, Raymond M. Testing heterozygote excess and deficiency. Genetics. 1995;140(4):1413–1419. [PubMed]
55. Raymond M, Rousset F. GENEPOP (version 1.2): population genetics software for exact tests and ecumenicism. J Hered. 1995;86:248–249.
56. Stephens M, Smith NJ, Donnelly P. A new statistical method for haplotype reconstruction from population data. Am J Hum Genet. 2001;68(4):978–989. [PubMed]
57. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials. 1986;7(3):177–188. [PubMed]
58. Cochran WG. Some methods for strengthening the common χ2 tests. Biometrics. 1954;10(4):417–451.
59. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ. 2003;327(7414):557–560. [PMC free article] [PubMed]
60. Zhang J, Yu KF. What’s the relative risk? a method of correcting the odds ratio in cohort studies of common outcomes. JAMA. 1998;280(19):1690–1691. [PubMed]
61. Brown BW, Russell K. Methods correcting for multiple testing: operating characteristics. Stat Med. 1997;16(22):2511–2528. [PubMed]
62. Risch N, Merikangas K. The future of genetic studies of complex human diseases. Science. 1996;273 (5281):1516–1517. [PubMed]
63. Wellcome Trust Case Control Consortium. Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls. Nature. 2007;447(7145):661–678. [PMC free article] [PubMed]
64. Hoggart CJ, Clark TG, De Iorio M, Whittaker JC, Balding DJ. Genome-wide significance for dense SNP and resquencing data. Genet Epidemiol. 2008;32 (2):179–185. [PubMed]
65. Moonesinghe R, Khoury MJ, Liu T, Ioannidis JP. Required sample size and non-replicability thresholds for heterogeneous genetic associations [published online ahead of print January 3, 2008] Proc Natl Acad Sci USA. 2008;105(2):617–633. doi: 10.1073/pnas.0705554105. [PubMed] [Cross Ref]
66. Melton LJ, III, Atkinson EJ, O’Fallon WM, Wahner HW, Riggs BL. Long-term fracture prediction by bone mineral assessed at different skeletal sites. J Bone Miner Res. 1993;8(10):1227–1233. [PubMed]
67. Rudenko G, Henry L, Henderson K, et al. Structure of the LDL receptor extracellular domain at en-dosomal pH. Science. 2002;298(5602):2353–2358. [PubMed]