The authors state that they have no conflicts of interest.
Article first published online: 12 MAR 2007
Copyright © 2007 ASBMR
Journal of Bone and Mineral Research
Volume 22, Issue 6, pages 808–816, June 2007
How to Cite
Deng, F.-Y., Xiao, P., Lei, S.-F., Zhang, L., Yang, F., Tang, Z.-H., Liu, P.-Y., Liu, Y.-J., Recker, R. R. and Deng, H.-W. (2007), Bivariate Whole Genome Linkage Analysis for Femoral Neck Geometric Parameters and Total Body Lean Mass. J Bone Miner Res, 22: 808–816. doi: 10.1359/jbmr.070303
Published online on March 12, 2007;
- Issue published online: 4 DEC 2009
- Article first published online: 12 MAR 2007
- Manuscript Accepted: 5 MAR 2007
- Manuscript Revised: 12 FEB 2007
- Manuscript Received: 1 AUG 2006
- bivariate analysis;
- whole genome scan;
- geometric parameters;
- lean mass;
A genome-wide bivariate analysis was conducted for femoral neck GPs and TBLM in a large white sample. We found QTLs shared by GPs and TBLM in the total sample and the sex-specific samples. QTLs with potential pleiotropy were also disclosed.
Introduction: Previous studies have suggested that femoral neck cross-section geometric parameters (FNCS-GPs), including periosteal diameter (W), cross-sectional area (CSA), cortical thickness (CT), buckling ratio (BR), and section modulus (Z), are genetically correlated with total body lean mass (TBLM). However, the shared genetic factors between them are unknown.
Materials and Methods: To identify the specific QTLs shared by FNCS-GPs and TBLM, we performed bivariate whole genome linkage analysis (WGLA) in a large sample of 451 white families made up of 4498 subjects.
Results: Multipoint bivariate linkage analyses for 22 autosomes showed evidence of suggestive or significant linkages (thresholds of LOD = 2.3 and 3.7, respectively) to chromosomes 3q12 and 20q13 in the entire sample, 6p25 and 10q24 in women, and 4p15, 5q34–35 and 7q21 in men. Two-point linkage analyses for chromosome X showed strong linkage to Xp22.13, Xp11.4, Xq22.3, Xq23–24, and Xq25. Complete pleiotropy was identified on 10q24 and 5q35 for TBLM and BR in women and for TBLM and CT in men, respectively. Furthermore, chromosomes 5q34–35, 7q21, 10q24, 20q13, Xp22.13, Xp11.4, and Xq25 are also of importance because of their linkage to multiple trait pairs. For example, linkage to chromosome 10q24 was found for TBLM × W (LOD = 2.31), TBLM × CT (LOD = 2.51), TBLM × CSA (LOD = 2.51), TBLM × BR (LOD = 2.64), and TBLM × Z (LOD = 2.55) in women.
Conclusions: In this study, we identified several genomic regions (e.g., 3q12 and 20q13) that seem to be linked to both FNCS–GPs and TBLM. These regions are of interesting because they may harbor genes that may contribute to variation in both FNCS-GPs and TBLM.
Bone strength is a direct determinant of osteoporotic fracture. Recent studies have shown that BMD accounts for, at most, 50–70% of total bone strength.(1,2) Other factors, such as bone structural properties (shape, size, microarchitecture) and bone remodeling status, also play important roles, independent of BMD, in determining bone strength and the associated osteoporosis fracture.(1,3,4) Geometric parameters (GPs) at the hip can be used to enhance the accuracy of identification of people at high risk of hip fracture,(4–6) the most serious and disabling type of osteoporotic facture.(7) Melton et al.(8) suggested that some femoral neck cross-section GPs (FNCS-GPs), such as section modulus (Z) and buckling ratio (BR), are as good as BMD measurement for fracture risk prediction.
Bone and muscle are two closely related organs. Evidence showed that healthy bone can dynamically adapt to the mechanical loads,(9) and muscle, which is mainly composed by total body lean mass (TBLM), is responsible for the major mechanical load to bone.(10–15) The adaptive response of bone to altered mechanical loading may be better represented by the indices of bone geometry rather than BMD.(16) However, to date, the underlying mechanism accounting for the phenotypic correlation between TBLM and bone geometry remains unclear.
Evidence has shown that both FNCS-GPs and TBLM are controlled by genetic factors, with the heritability ranging from 0.37 to 0.62.(17–19) Sun et al.(19) recently reported significant phenotypic and genetic correlations between FNCS-GPs and TBLM and suggested that these traits may share some common genetic factors, with genetic correlation coefficients ranging from 0.28 to 0.69.
The aim of this study was to identify specific genomic regions that may harbor genes contributing to both FNCS-GPs and TBLM through bivariate whole genome linkage analysis (WGLA). The results from this study may provide helpful information for further positional cloning and functional studies of specific genes important to regulate FNCS-GPs and TBLM.
MATERIALS AND METHODS
The study was approved by the Creighton University Institutional Review Board. All the study subjects signed informed consent documents before entering the project. The sample contains a total of 4498 subjects from 451 pedigrees, and the pedigree size varied from 4 to 416 individuals, with a mean of 11.6 ± 28.5 (SD), providing an exceedingly large number of relative pairs (>150,000) informative for linkage analysis.(20) All the study subjects were whites of European origin. The sampling scheme and exclusion criteria have been detailed in our previous publication.(21) Briefly, patients with chronic diseases and conditions that might potentially affect bone mass, structure, or metabolism were excluded.
The traits of TBLM (g), areal BMD (g/cm2), and bone size (cm2) at the femoral neck (FN) were measured by Hologic 1000, 2000+, or 4500 DXA scanners (Hologic, Bedford, MA, USA). All scanners are calibrated daily, and long-term precision is monitored with external phantoms. The CVs of FN BMD, bone size, and TBLM measurement obtained on the Hologic 2000+ scanner were 1.87%, 1.94%, and 1.0%, respectively. Similar CVs were obtained on the other scanners.(21,22) The members of the same pedigree were usually measured on the same type of machine.
Using the DXA-derived FN BMD and bone size, we estimated five FNCS-GPs based on the method detailed elsewhere(23–26) and used in our earlier studies.(18,27) The five estimated FNCS-GPs were as follows: periosteal diameter (W), cross-sectional area (CSA), cortical thickness (CT), buckling ratio (BR), and section modulus (Z). BR is an index of cortical instability, which indicates the risk of fracture by buckling. Z is an index of bone bending strength, which indicates the bending resistance of a tube. CSA is an indicator of bone axial compression strength. More details of the explanations about the FNCS-GPs were provided by Beck.(23)
For each subject, DNA was extracted from peripheral blood using the Puregene DNA isolation kit (Gentra Systems, Minneapolis, MN, USA). All the subjects were genotyped for 410 microsatellite markers (including 393 markers for 22 autosomes and 17 markers for the X chromosome) from the Marshfield screening set 14 by Marshfield Center for Medical Genetics. The markers had an average population heterozygosity of 0.75 ± 0.06 and were spaced on average 8.9 cM apart. The detailed genotyping protocol is available at http://research.marshfieldclinic.org/genetics/Lab_Methods/methods.html. A genetic database management system GenoDB(28) was used to manage the phenotype and genotype data for linkage analyses. PedCheck was performed to ensure that the genotype data conform to Mendelian inheritance pattern at all the marker loci.(29) We used MERLIN(30) to detect genotyping errors of unlikely recombination (e.g., double recombination) in our sample. The genotyping error rate of ∼0.03% was determined.
The general statistical analyses were carried out using SAS version 9 (SAS Institute, Cary, NC, USA). Bivariate WGLA was performed using SOLAR v3.0.4 (Sequential Oligogenic Linkage Analysis Routines, available at http://www.sfbr.org/solar/).(31–33) The variance component analysis implemented in SOLAR makes an assumption of normal distribution of the quantitative traits studied. We checked the normality of distribution of the six studied traits (raw and transformed data) using the index kurtosis, which ranged from 0.36 to 1.0, generally conform to what SOLAR requires for robust linkage results (i.e., kurtosis < 0.8).(34) Bivariate multipoint linkage analyses on 22 autosomes and two-point linkage analysis on the X chromosome were performed between the five FNCS-GPs and TBLM. In linkage analysis, age and sex were incorporated in the models as covariates.
Using the variance component model in SOLAR, we tested the null hypothesis of no linkage by comparing the likelihood of the restricted model (i.e., σ2m = 0, here σ2m is the additive genetic variance caused by a major locus) with the likelihood of the model of which σ2m was estimated. The LOD score is calculated as the difference between the two log10 likelihoods, which indexes the probability of existence of a QTL. In bivariate linkage analysis, the bivariate test statistic (2 × Ln10 × LOD2) follows a mixture distribution of 1/2 χ12, 1/4 χ32, and 1/4 χ02,(35) and the corresponding two-trait LOD score can be calculated using the command “-clod-raw.” In univariate analysis, the statistic follows a χ2 distribution with 1 df. To compare with univariate LOD scores, the “1 df” LOD scores (denote LODeq, i.e., with equal p values of the two-trait LODs) were also calculated and are presented in the Results section and Table 2.
Because of the potential sex-specific effects on bone geometry and lean mass,(36–38) we further conducted bivariate linkage analyses in women and men separately. For precision, in sex-specific analyses, we used the Marshfield sex-specific genetic maps instead of the sex-averaged map.
We adopted a method developed by Camp and Farnham(39) to correct for multiple testing caused by the multiple correlated traits studied in this study. This method, by calculating the correlation coefficients between LOD scores for each trait, determines the number of effectively independent tests (N). In this study, the independent test numbers were 2.65, 2.54, and 2.21 for the entire sample, the female subgroup, and the male subgroup, respectively. Based on the following formula,(39–41)
the genome-wide “suggestive” and “significant” LOD thresholds of linkage (corresponding p = 0.0005 and 0.000017, respectively) for this study were 2.34 and 3.75 for the entire sample, 2.32 and 3.73 for the female group, and 2.25 and 3.67 for the male group, respectively.
We further tested whether a significant linkage was caused by pleiotropic effects of a single locus or co-incidence of tightly linked loci within the region. We denote ρm as a measure of the shared major genetic effect near the genomic region that linkage is being assessed. Likelihoods for the linkage model in which ρm was estimated was compared with the likelihood of the linkage model in which ρm was constrained to 0 (complete co-incident linkage) or 1 (complete pleiotropy).(31,35) The possibility of co-incident linkage and complete pleiotropy was determined using a cut-off p value of 0.01 (p1 for co-incident linkage and p2 complete pleiotropy).
The study sample contained a total of 4498 subjects from 451 pedigrees. Among the 4152 genotyped subjects, 4096 and 3141 subjects were phenotyped for FNCS-GPs and TBLM, respectively, with 3126 subjects having both traits. The basic characteristics of the 3126 samples are summarized in Table 1. Generally, FNCS-GPs (especially CSA and Z) and TBLM of the men were larger than those of the age-matched women. For both men and women, W and BR increased with aging, whereas CT, CSA, and Z had a tendency to decrease gradually with aging. TBLM reached its peak value in the fourth decade for both sexes and decreased slowly with aging (<2.0% per decade).
Bivariate linkage analysis
Multipoint analyses in the entire sample on 22 autosomes:
Using the thresholds after correction for multiple testing, significant linkage was found on chromosome 20q13 with multipoint LOD score of 3.88 (p = 0.00001) for TBLM and CT (Table 2). Suggestive linkage was detected on the following three genomic regions: 20q13 (LOD = 3.32) for TBLM and BR; 3q12 (LOD = 2.49) for TBLM and BR; and 20q13 (LOD = 2.39) for TBLM and CSA. It is notable that chromosome 20q13 was linked to (significantly or suggestively) three pairs of traits (TBLM × BR, TBLM × CSA, and TBLM × CT; Table 2). To highlight the linked regions, we show in Fig. 1A the distribution of multipoint bivariate LOD scores for the five studied phenotype pairs, with LOD peak achieved around 20q13.
Sex-specific multipoint linkage analyses on 22 autosomes:
Female-specific linkage evidence was found on 10q24–25 for TBLM × CT (LOD = 2.51), TBLM × CSA (LOD = 2.51), TBLM × BR (LOD = 2.64), and TBLM × Z (LOD = 2.55), and on 6p25 for TBLM × CSA (LOD = 2.52). Figure 1B intuitively shows the distribution of the multipoint bivariate LOD scores for the five studied trait pairs, with LOD peak achieved around 10q24.
Male-specific linkage evidence was found on 5q34–35 (LOD = 3.01 for TBLM × CSA; LOD = 2.26 for TBLM × W; LOD = 2.28 for TBLM × BR; and LOD = 2.26 for TBLM × CT) and 7q21 (LOD = 3.6 for TBLM × CSA; LOD = 3.26 for TBLM × Z; and LOD = 2.57 for TBLM × CT). Figures 1C and 1D intuitively show the distributions of the multipoint bivariate LOD scores for the five studied trait pairs, with LOD peaks achieved around 5q34–35 and 7q21.
To provide useful information and to avoid possible missing of potentially important linkages, we report here all the genomic regions achieving LOD ≥ 1.80 (Table 2). This is because LOD = 1.86 (corresponding p = 0.0017) is widely accepted and used as the threshold for “suggestive” linkage in linkage analysis for single trait.(40)
Two-point linkage analyses on chromosome X:
Table 3 summarizes the “suggestive” and “significant” linkage results from bivariate two-point analysis on X chromosome in the entire sample and the sex-stratified samples. In the entire sample, the highest LOD score of 5.4 was obtained at the marker ATCT003 (Xq25) for TBLM and CT. LOD scores of >4.0 were also found at the markers GATG011 (Xp11.4) and GATA172D05 (Xq22.3) for TBLM and CT (LOD = 4.51 and 4.9, respectively), and at the marker GATG011 (Xp11.4) for TBLM and CSA (LOD = 4.51). Female-specific linkage (LOD > 4.0) was found at the markers GATA175D03 (Xp22.13) and GATA165B12P (Xq23–24) for TBLM, W, and BR. No evidence of significant linkage was found in the male group.
Pleiotropy versus co-incident linkage
Using a p value of 0.01 as a cut-off point for rejection of complete pleiotropy or co-incident linkage, discrimination analysis suggested that the chromosome 5q35 may harbor genes with pleiotropic effect on TBLM and CT in men (129 cM, p1 = 0.0003, p2 = 0.039), and the chromosome 10q24 may harbor genes with pleiotropic effect on TBLM and BR in women (162 cM, p1 = 0.005, p2 = 0.012). Incomplete or partial pleiotropy was suggested for the other linked genomic regions (significant or suggestive), because complete pleiotropy was rejected (data not shown). For the tentative linkage signals with LOD ≥ 1.80, discrimination analysis suggested complete pleiotropy of 12q13 on the male TBLM and BR, although co-incident linkage was not rejected (50 cM, p1 = 0.045, p2 = 0.048).
Complete co-incident linkage was found for TBLM and W in the entire sample on 14q32 (119 cM, p1 = 0.02, p2 = 0.0002), on 17p13 (10 cM, p1 = 0.09, p2 = 0.0002) in men for TBLM and Z, on 15p13 (0 cM, p1 = 0.02, p2 = 0.0002) in women for TBLM and CT, and on 15p11 (12 cM, p1 = 0.045, p2 = 0.0002) in women for TBLM and Z.
This study screened the genomic regions that may harbor genes affecting both FNCS-GPs and TBLM. Evidence of linkage was found on chromosomes 3q12, 4p15, 5q34–35, 6p25, 7q21, 10q24, 20q13, Xp22.13, Xp11.4, Xq22.3, Xq23–24, and Xq25. Furthermore, our results suggested that certain genes located on 10q24 and 5q35, respectively, may have pleiotropic effects on TBLM × BR in women and TBLM × CT in men. This study supports the hypothesis that, in addition to mechanical stimuli, certain genetic factors may account for the phenotypic correlation between FNCS-GPs and TBLM.
Comparison with previous univariate WGLA studies
Generally, bivariate linkage analysis may provide greater power than univariate linkage analysis to identify QTLs underlying multiple correlated traits. Earlier univariate linkage studies were conducted for TBLM(21,42) or FNCS-GPs.(23) Some of the reported linkage regions, such as 5q23 (for male TBLM), 7q32 and 15q12 (for female TBLM), 20q12 (for BR, CSA, and CT), 5q33 and 7q21 (for male CSA), 7q21(for male CT), 3q26 (for female CSA and CT), and 10q26 (female CT), retained significant (or suggestive) (LOD > 2.0) in this bivariate linkage analysis, which confirms the significance of these genomic regions on TBLM and FNCS-GPs. However, some other earlier reported linkage regions, such as 12q14 and 17p12 (for TBLM),(21) 5q35 (for TBLM),(42) and 12p13 (for CT),(27) did not show any suggestive/significant linkage evidence in this study. Because only the loci having effects on both traits can be detected in bivariate analyses, we postulate that the effects of 12q14, 17p12, 5q35, and 12p13 could be trait specific.
On the other hand, many regions identified in this bivariate linkage study (i.e., 20q13, 3q12, 10q24, 6p25, 5q34–35, 7q21, and 4p15) were seldom significant for both FNCS-GPs and TBLM in earlier univariate linkage studies, further indicating that bivariate analysis may be more powerful than single-trait analysis in identifying QTLs underlying multiple correlated traits.
Sex-specific genomic regions
In our sample, the men had ∼40% higher TBLM than age-matched women. Our results support the sex-specific genetic influence for both TBLM and bone geometry.(36–38) Although some regions identified in the entire sample were not found in the sex-stratified samples, we cannot rule out potential sex-specific effects. This is because of the decreased power caused by smaller sample sizes in the sex-stratified sample.
Pleiotropy versus co-incident linkage and candidate genes
In the study, two interesting regions (i.e., 5q35 and 10q24) were found to have pleiotropic effects on both FNCS-GPs and TBLM. On chromosome 5q35, there are some interesting genes, such as msh homeobox 2 (MSX2), sequestosome 1 (SQSTM1), stanniocalcin 2 (STC2), PDZ and LIM domain 7 (PDLIM7) that are related to bone and/or muscle metabolism. Among these, STC2 seems to be an attractive candidate gene. STC2 is a secreted glycoprotein hormone expressed in a wide variety of tissues,(43,44) with the highest expression in skeletal muscle and heart. STC2 is actively involved in phosphate homeostasis and is considered to play an important role in diseases related to phosphate homeostasis.(45) Transgenic mice model showed that STC2 can act as a potent growth inhibitor and reduce intramembranous and endochondral bone development and skeletal muscle growth.(46) However, the importance of the STC2 gene in humans awaits further exploration.
Among the seven promising autosomal regions disclosed in the study, 10q24 was found to have pleiotropic effect on TBLM and FNCS-GPs. Within the region 10q24, several interesting genes, such as plasminogen activator urokinase (PLAU), TNF receptor superfamily member 6 (FAS), cytochrome P450 family 17 subfamily A (CYP17A1), are related to bone or muscle metabolism. CYP17A1 has been considered as a strong candidate gene for osteoporosis in postmenopausal women.(47) The CC genotype at T-34C polymorphism was associated with increased BMD in postmenopausal Japanese women.(48) The A2 allele of the CYP17 T (27)-C polymorphism was associated with reduced bone mass and CSA in the spine and femoral neck in lean perimenopausal women.(49) The protein encoded by the CYP17A1 gene is a key enzyme in the steroidogenic pathway that produces progestins, mineralocorticoids, glucocorticoids, estrogens, and androgens.(50–53) It is well known that androgens are a powerful factor that may effectively improve muscle mass and muscle strength.(54–56) Hence, CYP17A1 represents a promising candidate gene with pleiotropic effects on FNCS-GPs and TBLM.
For the other five autosomal genomic regions with suggestive or significant linkage (3q12, 4p15, 6p25, 7q21, and 20q13), complete pleiotropy was rejected, suggesting incomplete or partial pleiotropy. Possible explanations are (1) multiple functional variants may underlie the linkage signal, with some variants influencing both traits and others influencing only one of them; (2) genotype-by-environment interactions may cause differential effect of the QTL; and (3) no pleiotropy but co-incident linkage with independent functional variants that are in linkage disequilibrium with each other (and therefore partially, but not completely, correlated).
The regions of mapped loci are still too large for gene identification. To fine map these QTLs, denser genetic markers around the disclosed chromosome regions need to be genotyped and analyzed further by linkage disequilibrium test, which can narrow the region down to a few dozens of kilobases in general. For conveniences of candidate gene studies in further exploration, we also list the bone-related and muscle-related genes on these interest regions in Table 4, which can be speculated as candidate genes of both traits and deserve attention. Certainly, we cannot exclude the possibility that there are other genes under cover that might contribute to the positive signals detected in the study.
In summary, this study identified several genomic regions that may harbor genes contributing to both FNCS-GPs and TBLM. Some regions with potential pleiotropic effects and sex-specific effects are also revealed. Their importance in other populations should be studied for replication/confirmation. Our results provided useful information for future candidate gene studies and QTL fine mapping for FNCS-GPs and TBLM, aiming at eventually unraveling the mechanisms underlying phenotypic dependence of FNCS-GPs and TBLM.
The study was partially supported by grants from NIH (K01 AR02170-01, R01 AR45349-01, and R01 GM60402-01 A1), a key project grant of National Science Foundation of China (30230210), and a project from Hunan Provincial Natural Science Foundation of China (04JJ1004). The authors thank all the study subjects for volunteering to participate in the study. The genotyping experiment was performed by Marshfield Center for Medical Genetics and supported by NHLBI Mammalian Genotyping Service (Contract Number HV48141).
- 91892 Das Gesetz der Transformation der Knochen. Kirschwald, Springer, Berlin, Germany.
- 492004 Two single nucleotide polymorphisms in the CYP17 and COMT Genes–relation to bone mass and longitudinal bone changes in postmenopausal women with or without hormone replacement therapy. The Danish Osteoporosis Prevention Study. Calcif Tissue Int 75: 123–132., , , , , , , , ,
- 511986 Molecular biology of steroid hormone synthesis. DNA 5: 61., , , ,