Tests of Linkage and/or Association of Genes for Vitamin D Receptor, Osteocalcin, and Parathyroid Hormone With Bone Mineral Density†
The authors have no conflict of interest.
Bone mineral density (BMD) is a major determinant of osteoporotic fractures (OFs). The heritability of BMD ranges from 50% to 90% in human populations. Extensive molecular genetic analyses have been performed through traditional linkage or association approaches to test and identify genes or genomic regions underlying BMD variation. The results, particularly those concerning the vitamin D receptor (VDR) gene, have been inconsistent and controversial. In this study, we simultaneously test linkage and/or association of the genes for VDR, osteocalcin (also known as bone Gla protein [BGP]), and parathyroid hormone (PTH) with BMD in 630 subjects from 53 human pedigrees. Each of these pedigrees was ascertained through a proband with an extreme BMD value at the hip or spine (Z score ≤ −1.28). For the raw BMD values, adjusting for significant covariate effects of age, sex, and weight, we performed tests for linkage alone, association alone, and then both linkage and association. For the spine BMD, at the two markers (ApaI and FokI) inside the VDR gene we found evidence for linkage (p < 0.05) and for both linkage and association by the transmission disequilibrium test (TDT; p < 0.05); association was detected (p < 0.07) with regular statistical testing by analyses of variance (ANOVA). In addition, significant results were found for association alone (p < 0.05), linkage alone (p = 0.0005), and for linkage and association (p = 0.0019) for the intragenic marker HindIII of the BGP gene for the hip BMD. Through testing for association, linkage, and linkage and association simultaneously, our data support the VDR gene as a quantitative trait locus (QTL) underlying spine BMD variation and the BGP gene as a QTL underlying hip BMD variation. However, our data do not support the PTH gene as a QTL underlying hip or spine BMD variation. This is the first study in the broad field of bone genetics that tests candidate genes as QTLs for BMD by testing simultaneously for association alone, for linkage alone, and for association and linkage (via the TDT).
LOW BONE mineral density (BMD) is an important risk factor for osteoporotic fractures (OFs), and osteoporosis is characterized mainly by low BMD.(1–3) Osteoporosis results in more than 1.3 million OFs a year, with an estimated direct cost of ∼14 billion dollars in the United States alone in 1995.(4) Extensive data have been established that BMD variation is under strong genetic control with heritability (h2) estimates ranging from 0.5 to 0.9.(5–10) Recently, extensive molecular genetic studies have been launched to search for genes underlying BMD variation.(11–18) The results so far from either association or linkage approaches have largely been inconsistent.(13,19–23) The results concerning several candidate genes, for example, the vitamin D receptor (VDR) gene, are particularly controversial.(18,20,21,23,24)
It is well known that the association study approach may yield spurious associations between a complex trait and candidate genes because of population admixture that is difficult to detect.(25) In addition, we recently found that in the presence of population admixture, significant effects of a gene underlying a complex trait may disappear in population association studies.(26) Hence, studies using the population association approach, although valuable, are limited in that the results from this approach alone may not be conclusive.(27,28) On the other hand, the linkage study approach for a complex trait often is of limited statistical power.(29–31) This, together with the polygenic nature of complex traits, may lead to difficulty in replicating linkage results in different studies as well as false identification of genes in whole genome scans.(16) An alternative approach, the transmission disequilibrium test (TDT), has been developed for complex diseases(32) and extended to continuously distributed quantitative traits such as BMD.(30,31,33–38) When testing specific genes (e.g., candidate genes) such as putative quantitative trait loci (QTL) or disease susceptibility loci (DSL), the TDT is much more powerful than the linkage approach.(29–31,39) The TDT has been used increasingly widely to identify genes underlying complex diseases or to detect linkage and/or linkage disequilibrium (association) between markers and such genes.(40) In testing candidate genes for association with complex diseases in the presence of evidence of linkage, the TDT is not plagued by the problem of population admixture or stratification.(41,42) In the presence of association between genotypes and phenotypes, the TDT can be used to test linkage of candidate genes with complex traits.(40,42) Only when both linkage disequilibrium (association) and linkage of a marker with a QTL exist can TDT results be significant when testing candidate genes as putative QTLs. However, the TDT approach has been used seldom in bone genetics to identify genes for osteoporosis.(43) This is interesting given that a number of candidate genes and genomic regions have been suggested but the results have been inconsistent across various studies as mentioned earlier.
In this study, for a large sample of human pedigrees, we will use simultaneously the classical approaches (association or linkage) and the more contemporary approach (the TDT) to test three candidate genes (the VDR; the osteocalcin, also known as bone Gla protein [BGP]; and parathyroid hormone [PTH] genes) as putative QTLs for BMD variation. The study will provide evidence supporting the VDR and BGP genes as QTLs for spine and hip BMD variation, respectively. This study represents our effort to use various approaches (association, linkage, and the TDT) simultaneously in resolving controversies in identification of genes underlying BMD variation.
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 study subjects came from an expanding database being created for a whole genome linkage study aimed at searching for genes underlying BMD variation and OF risk that is underway in the Osteoporosis Research Center (ORC) of Creighton University. Only healthy people (defined by the exclusion criteria to be detailed later) were included in the analyses. All the study subjects were whites of European origin; 53 pedigrees with 630 subjects (248 male subjects and 382 female subjects) from two to four generations were analyzed. The pedigrees vary in size from 3 to 99 individuals, with a mean of 11.7 individuals (±SE = 2.4). Each pedigree was identified through a proband having BMD Z scores ≤ −1.28 at the hip or spine so that the probands were selected from the bottom 10% of the population BMD variation with the purpose of achieving higher statistical power than random sampling.(39,44) BMD values are expressed as Z scores that adjust for age, gender, and ethnic difference in general referent healthy populations. The exclusion criteria for the study subjects were a history of (1) serious residuals from cerebral vascular disease; (2) diabetes mellitus, except for easily controlled, noninsulin-dependent diabetes mellitus; (3) chronic renal disease; (4) chronic liver disease or alcoholism; (5) significant chronic lung disease; (6) corticosteroid therapy at pharmacologic levels for >6 months duration; (7) treatment with anticonvulsant therapy for >6 months duration; (8) evidence of other metabolic or inherited bone disease such as hyper- or hypoparathyroidism, Paget's disease, osteomalacia, osteogenesis imperfecta, or others; (9) rheumatoid arthritis or collagen disease; (10) recent major gastrointestinal disease (within the past year) such as peptic ulcer, malabsorption, chronic ulcerative colitis, regional enteritis, or any significant chronic diarrhea state; (11) significant disease of any endocrine organ that would affect bone mass; (12) hyperthyroidism; (13) any neurological or musculoskeletal condition that would be a nongenetic cause of low bone mass; (14) any disease, treatment, or condition that would be a nongenetic cause for low bone mass. The exclusion criteria were assessed by nurse-administered questionnaires and/or medical records.
For each subject, blood (20 ml) was drawn into lavender cap (EDTA containing) tubes by certified phlebotomists and stored chilled (∼4°C) until DNA extraction that was normally completed within the next 5 calendar days. DNA was extracted by using a kit (Puregene DNA Isolation Kit, D-5000; Gentra Systems, Inc., Minneapolis, MN, USA) following the procedures detailed in the kit. DNA was genotyped for the restriction fragment length polymorphism (RFLP) at the following markers, respectively: the ApaI RFLP (in intron 8) and FokI RFLP (in exon 2) inside the VDR gene,(18) the HindIII RFLP (in the promoter region) of the BGP gene,(45) and the BstBI RFLP (in intron 2) of the PTH gene.(46)
Our genotyping procedures for these markers were modified from those of Tokita et al.(47) and Uitterlinden et al.(48) for the ApaI RFLP, Harris et al.(49) for the FokI RFLP, Dohi et al.(45) for the HindIII RFLP, and Mullersman et al.(46) for the BstBI RFLP. Briefly, for the ApaI RFLP inside the VDR gene, the forward primer in intron 8 (5′-CAG AGC ATG GAC AGG GAG CAA G-3′), and the reverse primer in exon 9 (5′-GCA ACT CCT CAT GGC TGA GGT CTC A-3′) were used in polymerase chain reaction (PCR) to produce a 745-base pair (bp) DNA fragment. For the FokI RFLP inside the VDR gene, the forward primer in intron 1 (5′-AGC TGG CCC TGG CAC TGA CTC TGC TCT-3′) and the reverse primer in intron 2 (5′-ATG GAA ACA CCT TGC TTC TTC TCC CTC-3′) were used in PCR to generate a DNA fragment of 265 bp. For the HindIII RFLP inside the BGP gene, the forward primer in the promoter region (5′-CCG CAG CTC CCA ACC ACA ATA AGC T-3′) and the reverse primer in exon 1 (5′-CAA TAG GGC GAG GAG T-3′) were used in the PCR to generate a DNA fragment of 253 bp. For the BstBI RFLP inside the PTH gene, the forward primer in intron 1 (5′-CAT TCT GTG TAC TAT AGT TTG-3′) and the reverse primer in the 3′ flank region (5′-GAG CTT TGA ATT AGC AGC ATG-3′) were used in the PCR to generate a DNA fragment of 600 bp. The PCR amplification was all conducted in reaction mixtures each containing 10.08 μl of ddH2O, 2 μl of 10× PCR buffer, 1.5 mM of MgCl2, deoxynucleoside triphosphate (dNTP; 200 μM each), 0.6 U of Taq Polymerase (AmpliTaq Gold with GeneAmp; Applied Biosystems, Foster City, CA, USA), 0.4 μM each of the two primers for each marker, and 1 μl (∼50 ng/μl) of genomic DNA. The PCR was performed on PE 9700 thermocyclers (GeneAmp PCR System 9700; Applied Biosystems). PCR cycling conditions for the four markers are as follows: for the VDR ApaI marker, they were 94°C for 1 minute, 60°C for 1 minute, and 72°C for 1 minute each, 35 cycles. For the VDR FokI marker, they were 94°C for 30 s, 60°C for 30 s, and 72°C for 30 s each, 35 cycles. For the PTH BstBI marker, they were 94°C for 1 minute, 50°C for 1 minute, and 72°C for 2 minutes each, 35 cycles. For the PTH HindIII marker, they were 94°C for 30 s, 53°C for 30 s, and 72°C for 1 minute each, 35 cycles. These cycling conditions were preceded by 10 minutes at 94°C for denaturing and followed by 10 minutes at 72°C for extension. After PCR amplification, 8 μl of the respective PCR products were removed and digested with the following restriction endonucleases, respectively, at their respective temperatures: 5 U of ApaI at 25°C, 4 U of FokI at 37°C, 5 U of BstBI at 65°C, and 5 U of HindIII at 37°C (Life Technologies, Grand Island, NY, USA) all for 3 h. Uncut and digested samples were electrophoresed in 2% Metaphor agarose gels (FMC Inc., Rockland, ME, USA) in 1× Tris-borate-EDTA (TBE) buffer and 0.3 μg/ml of ethidium bromide. Gels were then visualized on a transilluminator under UV light and photographed. The absence and presence of the ApaI, FokI, BstBI, and HindIII restriction sites in the three candidate genes were designated as A and a alleles, F and f alleles, B and b alleles, and H and h
respectively. The program PedCheck (available at http://watson.hgen.pitt.edu/register/soft_doc.html) was used for verifying Mendelian inheritance of all the marker alleles.(50)
BMDs of spine and hip were measured by a Hologic, Inc. 1000, 2000+, or 4500 scanner (Hologic, Inc., Waltham, MA, USA). All machines are calibrated daily, and long-term precision is monitored with external spine and hip phantoms. Hip and spine were chosen because they are the most common OF sites.(1) Short-term precision in humans is 0.7% for spine BMD and 1.0% for hip BMD. We maintain constant quality assurance procedures that track potential confounding events such as X-ray tube replacement, arm realignments, collimator changes, and software version updates. Technicians maintain scan-by-scan surveillance for quality control. We have chosen BMD rather than bone mineral content as our bone mass phenotype because BMD is the measure most closely correlated with fracture risk.(51) For the spine, our quantitative phenotype was combined BMD of L1-L4. For the hip, it was combined BMD of the femoral neck, trochanter, and intertrochanteric region. Weight was measured at the same visit when the BMD measurements were taken. Data obtained from different machines are transformed to a compatible measurement by an algorithm developed by us (R. R. Recker and K. M. Davies, unpublished data, 2001) and members of the same pedigree usually are measured on the same type of machine.
We performed statistical analyses to test for association, linkage, and association and linkage (TDT) between each of the markers and BMD at the spine and hip. An association test between a marker and a quantitative trait was developed by Abecasis et al.(37) for pedigree data and the TDT in pedigrees was developed by Abecasis et al.(38) Both of these types of analyses have been implemented in a program for quantitative trait locus transmission disequilibrium test (QTDT), which is available on the internet (http://www.well.ox.ac.uk/asthma/QTDT). These tests were all developed under a variance component framework. The permutation procedure built in the QTDT may yield significance levels (p values) of the tests that are not biased by (and thus robust to) the ascertainment schemes of pedigrees as in this study. In the QTDT program, population stratification may be tested also. Only when the result for population admixture is not significant is the test for population association of a marker with BMD variation warranted. The linkage test was performed also by the variance component linkage analyses for quantitative traits.(52–54) The variance component analysis is based on specifying the expected genetic covariances between arbitrary relatives as a function of the identity by descent at a given marker locus. The analysis considers the phenotypic and genetic information from all pedigree members simultaneously. The analysis assumed multivariate normality of phenotypic values, additive genetic effects, and no interaction between genes and the residual. The common familial environmental effects were assumed to be negligible, which is reasonable and supported by previous studies.(7–9) The QTDT also provides a module for linkage testing; however, this module does not have the capacity to account for ascertainment of pedigrees via extreme probands as in this study. Hence, we also tested linkage using the program Sequential Oligogenic Linkage Analysis Routines (SOLAR),(54) which is available on the internet (http://www.sfbr.org/sfbr/public/-software/solar/solar.html). The ascertainment scheme of pedigrees based on the low BMD values of probands was accounted for in analyses with SOLAR by identifying to the program the probands for each pedigree. The conditional likelihood models built in the program then will account for the ascertainment by the proband status and the cut-off BMD values for the probands.
In all the statistical analyses, age, sex, and weight were adjusted as covariates (if having significant effects in our sample) to adjust for raw BMD values (not the Z scores). These factors generally affect BMD variation significantly.(10,55) Analyses also were performed without adjusting for these covariates. Generally, adjustment for significant covariates in genetic analyses can increase the genetic signal to noise ratio (i.e., h2 estimates) by decreasing the proportion of the residual phenotypic variation attributable to random environmental factors.(9–10,15) This can improve statistical power in our association and/or linkage analyses. The BMD data were tested by graphical methods and found not to deviate from normal distributions.(56) Marker allele frequencies were obtained by maximum likelihood estimation in SOLAR. Hypothesis testing was conducted by the maximum likelihood method. The method compares the maximum likelihoods obtained in the full model (with association, linkage, or both association and linkage) and the nested null model (without association, linkage, or without either association or linkage). The QTDT program generates p values for various tests (asymptotic κ2 tests or permutation tests). The SOLAR program generates logarithm of odds (LOD) scores that can be converted to approximate p values through a κ2 distribution.(57)
The basic characteristics of the study subjects stratified by age and sex are summarized in Appendix 1. Some information about the family structure of the study pedigrees is summarized in Appendix 2. Appendix 2 shows that there are ∼1380 parent offspring pairs, 1249 sibling pairs, 1098 grandparent-grandchild pairs, 2589 first cousin pairs etc., reflecting the richness of the genetic information for linkage analyses in our sample. The frequencies of genotypes and alleles, together with the genotype data missing rate are summarized in Table 1. Missing genotype data includes those that failed to be amplified in PCR reaction even after three times of repeated experiments and those that did not pass a Mendelian inheritance check within pedigrees with the PedCheck program. The missing rate for genotype data generally are <1% except at the VDR ApaI marker where it is 7.7%. Except for the PTH BstBI marker, the genotypes do not generally deviate from the Hardy-Weinberg equilibrium, despite the fact that the data are from 53 pedigrees and not from random and unrelated population samples.
Table Table 1.. Allele and Genotype Frequencies and Genotype Data Missing Rates
We found significant (p < 0.05) or marginally significant (p < 0.07) associations when testing relationships of the marker genotypes with BMD variation by regular analyses of variance (ANOVA), which is used commonly in association studies and which ignores relatedness of the subjects in our sample (Table 2). None of the tests for population admixture is significant for the four markers (Table 3) in the study population. For association tests with the QTDT, BGP HindIII marker genotypes are significantly associated with spine BMD variation (Table 3).
Table Table 2.. Phenotype Distribution Among the Genotypes
Table Table 3.. Results of Tests of Population Stratification, Association, Linkage, and Linkage and Association
Interesting and compatible results emerge (Table 3) in tests for linkage with both the QTDT and SOLAR programs. It is notable that both the VDR ApaI and the FokI markers are linked to spine BMD variation but not to hip BMD variation. The BGP HindIII marker is linked to hip BMD variation but not to the spine BMD variation. More interestingly, when testing for both linkage and association by TDT with the QTDT, we find significant results for linkage and association between spine BMD variation and the VDR ApaI and FokI markers and between hip BMD variation and the BGP HindIII marker.
Data not shown indicate that analyses not adjusting for other covariates such as age, sex, and weight showed qualitatively the same results with significance strengthened in several cases. For example, the p values of linkage tests for the VDR ApaI and FokI markers with QTDT changed to 0.0070 and 0.0094, respectively, and those with SOLAR changed to 0.0014 and 0.040, respectively. The TDT for linkage and association with QTDT changed to 0.0038 and 0.0075, respectively, for the VDR ApaI and FokI markers. These results render the significance of our tests of linkage and TDT unchanged even after we apply the Bonferroni correction for the multiple markers tested in our study. Therefore, the linkage results and the TDT results for linkage and association seem to be robust.
For those markers with significant association (either by regular ANOVA or QTDT), linkage (by QTDT and SOLAR), and linkage and association (by TDT in the QTDT program), the allele a of the VDR ApaI marker and the allele F of the VDR FokI marker are associated with larger spine BMD values; the allele h of the BGP HindIII marker is associated with larger hip BMD values. The different effects of the VDR and BGP genes on spine and hip BMD values are noteworthy. The apparent discrepancy of the p values of various tests for linkage alone, association alone, or linkage and association (Tables 2 and 3) may reflect the different power associated with different tests and/or the different sensitivities of various tests to confounding effects such as potential population admixture.
BMD is a major determinant of osteoporosis. The heritability of BMD ranges from 50% to 90% in humans. In the recent past, extensive molecular genetic analyses have been performed by traditional linkage or association approaches to identify and confirm genes or genomic regions underlying BMD variation. The results, particularly those of the VDR gene, have been inconsistent and controversial. In this study, we test linkage and/or association of the genes for VDR, BGP, and PTH with spine and hip BMD in 630 subjects from 53 human pedigrees each ascertained through a proband with an extreme BMD value at the hip or spine. Although the tests for population admixture/stratification by the test in the QTDT are not significant, the power of such tests is not known and it is known that population admixture/stratification is notoriously difficult to detect.(25) Adjusting for significant covariate effects of age, sex, and weight, we perform tests for linkage alone, association alone, and for both linkage and association. For spine BMD, at the VDR ApaI and FokI markers we find evidence for association, linkage, and for both linkage and association by TDT. Significant results are found also for association alone, linkage alone, and for linkage and association of the BGP HindIII marker and hip BMD. Other unspecified results are generally not significant. By testing for association and linkage independently and for linkage and association simultaneously, our data support the VDR gene as a QTL underlying spine BMD variation and the BGP gene as a QTL underlying hip BMD variation. However, our data do not support the PTH as a QTL underlying hip or spine BMD variation. To the best of our knowledge, this is the first study in bone genetics to test candidate genes as QTLs for BMD by testing simultaneously, association, linkage, and association and linkage.
Decades of studies of various disciplines (e.g., molecular biology and cell biology) of bone have suggested a long list of genes that are of potential importance in skeletal biology and the list keeps growing.(58) Dozens of candidate genes have been tested in bone genetics as potential QTLs for BMD variation in association or linkage studies.(17–18,59) As pointed out by Duncan et al.,(17) candidate gene study uses an enormous body of knowledge regarding the pathophysiology of osteoporosis, which generally is ignored in whole genome screening. The importance of the three candidate genes has been shown in bone biology. The VDR gene is an interesting candidate gene because of the long-time association of vitamin D with the skeleton.(60) Vitamin D is a major regulator of calcium and bone metabolism,(59,61–62) because its action modulates intestinal calcium absorption, osteoclast and osteoblast activities, PTH production, and renal hydroxylation of 25-hydroxyvitamin D [25(OH)D]. Subtle variations in expression and/or function of VDR may contribute to major differences in the regulation of other target genes. The VDR gene, acting through target gene response elements, is a good candidate for modulating calcium and bone metabolism and consequently skeletal mineralization.(59,61,62) BGP is the major noncollagenous protein of bone. It is synthesized exclusively by osteoblasts under transcriptional regulation through a vitamin D-response element.(18) BGP plays a role in bone remodeling and is important for skeletal development.(63) PTH is involved intimately in the homeostasis of normal serum levels of calcium and phosphate, which, in turn, regulate the synthesis and secretion of PTH.(64,65) PTH is anabolic to bone when administered by intermittent injection and is an effective means of treating postmenopausal osteoporosis.(66,67) However, for the reasons outlined in the introduction, none of the earlier studies has convincingly and unambiguously shown the importance of these three candidate genes or any other candidate gene as QTLs underlying BMD variation.
By simultaneously providing evidence for association, linkage, and linkage and association by various tests, our study provides strong support for the VDR and BGP genes as QTLs for spine and hip BMD, respectively. The differential effect of various genes on BMD at different skeletal sites should not come as a surprise,(68) because it has been shown that the genetic correlation between BMD at various skeletal sites, although significant, is not very high.(9) Although we cannot show the importance of the PTH gene via the BstBI marker, it should be noted that, in general, our analyses test the specific markers genotyped inside candidate genes. This is particularly true for association and TDT analyses, which depend crucially on the existence and extent of linkage disequilibrium between a marker and a functional mutation inside a gene. Because linkage disequilibrium generally only exists over a short distance in the genome (particularly in large populations), a marker that resides inside a candidate gene may not be detected as significant in association and TDT analyses even if there may be a functional mutation existing elsewhere inside the candidate gene. This may occur if the candidate gene has a long DNA sequence and if the marker is relatively distant from the functional mutation and is not in (or is in weak) linkage disequilibrium with the functional mutation.(31,69) Therefore, in both association and TDT analyses, only through thorough examination of polymorphisms densely located throughout the candidate gene (or even outside of the gene in its transcriptional regulatory regions) may we confidently conclude the lack of importance of a candidate gene per se.
Among the available markers inside the VDR gene, the ApaI marker locus used here is linked closely and in strong linkage disequilibrium with the BsmI marker locus that was used in the first report of the association of the VDR gene polymorphisms with BMD variation.(11,18) Both of these marker loci are located in intron VIII of the VDR gene and represent silent mutations that do not alter the protein sequence of the VDR.(18) The FokI marker locus in exon II of the VDR gene represents a translation initiation codon polymorphism (T-C transition), which results in variants of VDR that differ by three amino acids.(70) The shorter form of VDR was shown to have ∼1.7-fold greater transcriptional activity in transfected HeLa cells than the longer form.(70) However, whether the VDR FokI polymorphism represents a functional mutation for BMD variation awaits further examination. The HindIII marker used here for the BGP gene represents a C-T transition in the promoter region of the BGP gene and is of potential functional importance in the regulation of the BGP gene. The association with larger BMD values for alleles a and F at the ApaI and FokI marker loci in the VDR gene and h allele at the HindIII marker in the BGP gene is consistent with the majority of the results from population association studies.(18)
Most of the tests of candidate genes in bone genetics or even for the broad field of human genetics are through the association study approach in random population samples. Comparatively speaking, only a limited few have used the linkage or TDT approach in human nuclear families or pedigrees. An exception in the bone genetics field is the study by Duncan et al.(17) who tested linkage of 23 candidate genes using microsatellite markers that often are not inside the candidate genes (which reduces the power of the linkage test). The samples (614 individuals from England) of Duncan are comparable with ours in terms of sample size and ascertainment through extreme probands.(17) Their results suggest linkage (p < 0.05) of the genomic region of the VDR gene to BMD variation and do not support linkage of that of the PTH gene with BMD variation, which is consistent with our results here. The BGP gene was not examined by Duncan et al.(17)
In the bone genetics field, our study here has the largest sample among the few linkage studies in terms of the number of informative relative pairs contained in the sample. Our study is also the first in testing candidate genes via the TDT, plus traditional tests for linkage and for association. The study and the samples should be able to serve as a template for examinations of the large number of candidate genes in the bone field.(58) This approach of studying candidate genes, together with our ongoing whole genome screen in our expanding samples will likely capture important QTLs for BMD variation eventually.
Investigators of this work are partially supported by grants from the Health Future Foundation, National Institutes of Health grants (K01 AR02170-01, R01 AR45349-01, R01 GM60402-01A1, and P01 DC01813-07), grants from the State of Nebraska Cancer and Smoking-Related Disease Research Program, U.S. Department of Energy grant DE-FG03-00ER63000/A00, Creighton University, a grant (30025025) from National Science Foundation of China, and a grant from HuNan Normal University.
Table . Basic Characteristics of the Study Subjects Stratified by Age of Each Decade
Table . Relationships Used in Linkage Analyses by SOLAR