help button home button Endocrine Society JCEM JCEM Call for Nominations for EIC
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Submit a related Letter to the Editor
Right arrow Purchase Article
Right arrow View Shopping Cart
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow Request Copyright Permission
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Nguyen, T. V.
Right arrow Articles by Eisman, J. A.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Nguyen, T. V.
Right arrow Articles by Eisman, J. A.
Right arrowPubmed/NCBI databases
*OMIM
The Journal of Clinical Endocrinology & Metabolism Vol. 88, No. 8 3614-3620
Copyright © 2003 by The Endocrine Society

Genetic Determination of Bone Mineral Density: Evidence for a Major Gene

Tuan V. Nguyen, Gregory Livshits, Jacqueline R. Center, Konstantin Yakovenko and John A. Eisman

Garvan Institute of Medical Research (T.V.N., J.R.C., J.A.E.), Sydney, New South Wales 2010, Australia; and Department of Anatomy and Anthropology, Tel Aviv University (G.L., K.Y.), Tel Aviv 69978, Israel

Address all correspondence and requests for reprints to: Dr Tuan V. Nguyen, Bone and Mineral Research Program, Garvan Institute of Medical Research, 384 Victoria Street, Darlinghurst, New South Wales 2010, Australia. E-mail: t.nguyen{at}garvan.org.au.


   Abstract
Top
 Abstract
Introduction
 Subjects and Methods
 Results
 Discussion
 References
 
This study was designed to test the hypothesis of a major gene influence on the variation in bone mineral density (BMD). BMD and bone mineral content at the lumbar spine and femoral neck were measured in 330 men and 413 women, aged 18–90 yr, from 107 nuclear and complex families (including 5 large pedigrees with 194 individuals who were identified through an index case with moderately high BMD at the femoral neck (z-score, >=1.28)). After adjusting for age and body weight, familial factors accounted for up to 72% of the total variation in BMD. In complex segregation analysis, for all variables examined the best-fitting most parsimonious model consistently suggested the Mendelian transmission of a major gene locus with significant residual correlations among siblings. This genetic model suggested that the proportion of a total variance (adjusted for significant covariates) attributable to a putative major gene effect ranged between 0.30 ± 0.09 for femoral neck BMD and 0.53 ± 0.07 for the principal component obtained on BMD and corresponding bone mineral content measures. These findings clearly support the hypothesis that a large component of the variance in BMD is under genetic control, with strong evidence for a major gene locus influencing BMD transmission.


   Introduction
Top
 Abstract
 Introduction
Subjects and Methods
 Results
 Discussion
 References
 
OSTEOPOROSIS, WITH ITS most serious sequel of low trauma fractures, is a major public health problem in the world because it incurs significant costs, morbidity and disability, and mortality (1, 2). In the absence of molecular insights into the cause of the disease, the most important phenotype of special clinical significance is fracture. Resistance to fracture is, in turn, determined largely by bone strength and possibly bone quality. Bone strength can be measured by bone mineral density (BMD). Epidemiological studies have consistently shown that BMD is a primary predictor of osteoporotic fractures, with each SD decrease in BMD being associated with a 1.5- to 2.5-fold increase in fracture risk (3, 4, 5) for both women and men. Therefore, BMD is often used as a surrogate end point for osteoporosis studies.

The variation in BMD is determined by a complex network of hormonal, nutritional, physical, lifestyle, and, most importantly, genetic factors. Based on twin and family data, it has been estimated that between 60–80% of the variance in BMD measured at the lumbar spine (LS) and femoral neck (FN) is attributable to genetic factors (6, 7, 8, 9, 10, 11). However, these types of studies, in particular twin designs, can lead to an overestimation of the heritability component that can occur as a result of any greater sharing of environmental factors in monozygotic than dizygotic twins. Although familial studies are less prone to this type of error, they rarely examined the specific pattern of BMD inheritance. Yet, the identification of the true mode of a complex trait inheritance has a crucial significance for further mapping of the respective genetic sources of the trait variation.

General consideration from the viewpoint of bone physiology and BMD complexity as a primary phenotype suggests a multifactorial nature of this trait and likely involvement of numerous independent genetic sources. It has been speculated that the number of genes that may be contributing to the regulation of bone mass is approximately 70 (12). Indeed, association studies indicated at least several potential candidate genes, each with a small effect (13). However, these data as well as a few genome scans (14, 15) and selected chromosomal scans (16, 17) are largely inconsistent, and molecular data are as yet unable to identify specific major genes (MG) involved in BMD determination.

Meanwhile, a few recent publications reported a MG pattern of BMD inheritance in several ethnic populations. Thus, in pedigree samples randomly drawn from two Caucasian populations (Chuvasha, East Europe; and Turkmenians, Central Asia) complex segregation analysis suggested that a MG accounted for approximately 35% of the radiographic phalange BMD variation in each sample (18, 19, 20). To date, similar genetic analysis of BMD measures obtained by dual energy x-ray absorptiometry methods of lumbar vertebrae and femur were conducted only in three studies. Of these, the results reported by Gueguen et al. (21), obtained on a modest sample of nuclear pedigrees, were not unequivocally interpretable. Two other studies provided clear evidence in support of MG involvement in BMD transmission. However, they were obtained on a rather nongeneral population, but on pedigrees selected via probands with extreme BMD (22) or even suffering from early idiopathic osteoporosis (23). Obviously, more data are needed and from different cultural and environmental backgrounds to clearly establish the pattern of BMD inheritance.

It is known that the measurements of BMD obtained at different skeletal sites are highly correlated (6, 7). For example, covariation of BMD measured at the LS and FN appeared to be under the effects of a common set of genetic factors (6, 7, 24), i.e. pleiotropic effects. Previous studies [with the exception of one study (22)] have been primarily based on univariate analysis, in which each phenotype was analyzed separately. Although there are good biological reasons for separate analyses, when pleiotropy is present, a multivariate analysis based on the linear combination of many traits into factor scores is more powerful than the use of univariate or mean phenotypic data (25, 26).

The primary aims of this study were, therefore, to estimate the heritability and the possible pattern of inheritance of each BMD measure and a BMD/bone mineral content (BMC) common principal component. We attempted to determine whether a MG model is consistent with the inheritance of each phenotype in the studied pedigrees and to obtain estimates for the parameters of the most parsimonious model.


   Subjects and Methods
Top
 Abstract
 Introduction
 Subjects and Methods
Results
 Discussion
 References
 
Subjects

The Dubbo Osteoporosis Epidemiology Study (DOGS) was initiated in 1997, with the principal aims being 1) to investigate the contributions of genes and environments to the development of osteoporosis, and 2) ultimately to search for genes underlying BMD variation and osteoporotic fracture risk. Dubbo is a semirural city, situated about 400 km (250 miles) northwest of Sydney, Australia. Dubbo was selected for the study because of its size and the similarity of the age and sex distributions of its population to the Australian population. Moreover, it is relatively isolated in terms of medical care, so that for certain health events (such as fractures) almost all patients are likely to be observed within the local public hospital and/or the single private radiological practice. The study was approved by the St. Vincent’s ethics committee, and written informed consent was obtained from all individuals.

The DOGS study was designed as a population-based, genetic epidemiological investigation, using data collected under the auspices of the Dubbo Osteoporosis Epidemiology Study (DOES), a longitudinal epidemiology of risk factors for osteoporotic fractures (3, 27). Based on the DOES database, families were identified and invited to participate in the DOGS study. To optimize information on genetic factors, families were identified through an index case with the following criteria: 1) moderately high bone density at the femoral (z-score greater than +1.28, or top 10% of the age-specific BMD distribution), and 2) availability and accessibility of extended families of adult members, identified through these older (age >60 yr) parent index cases and their siblings and families. The probands were identified from the database of the DOES. The justification for recruiting large pedigrees (rather than nuclear families) is mainly based on the consideration that there could be increased power to detect linkage. All subjects were Caucasians and were descendants of Anglo-Saxon background.

Exclusion criteria were applied to subjects with Paget’s disease or hyperparathyroidism, Cushing’s syndrome, diabetes mellitus, complete or partial gastrectomy, chronic malabsorptive syndromes and jejuno ileal bypass, chronic renal disease, chronic liver disease, prolonged immobilization, recuperation from stroke or other serious illness, or metastatic malignancies; to subjects who had used medications known to affect bone metabolism, such as corticosteroids (in excess of 7.5 mg prednisolone/d or 1600 µg/d inhaled beclomethasone or equivalents), T4, anticonvulsants, or bone-active agents such as calcitonins, bisphosphonates, fluorides, and calcitriol, in the past 5 yr for greater than 6 months; and to subjects who were unable to give informed consent.

Data collection and measurements

Data collection was via direct interview by trained nurses, who administered a structured questionnaire. The following anthropometric data were recorded: age, height, and weight. Age was calculated from the date of birth to the date of interview. Height without shoes (in centimeters) was measured to the nearest 0.1 cm by a wall-mounted stadiometer. Peak height (around the age of 30 yr) was based on actual records or personal recall. Weight, without shoes or clothing, was measured (to the nearest 0.1 kg) on an electronic scale. Body mass index was derived as the ratio of weight (kilograms) over squared height (meters).

BMD (grams per square centimeter) and BMC (grams) were measured in the LS and FN by dual energy x-ray absorptiometry using a DPX densitometer (Lunar Corp., Madison, WI). BMD was derived as BMC over the skeletal area of interest. The radiation dose with this method is less than 0.1 µGy. The coefficient of variation with this method at our institution in normal subjects for BMD was 1.5% for the LS and 1.3% for the FN.

Statistical genetic analyses

Before segregation analysis, an initial analysis of the data indicated that there was no significant difference in heritabilities of BMD between males and females; hence, the data from the two sexes were combined. Subsequently, BMD was adjusted for sex, age, and weight, the three most powerful determinants of BMD, in a multiple linear regression model. Age was modeled as both a linear term and a quadratic term. The residual scores from the regression model, which were independent of age, sex, and weight, were then used for testing the hypothesis of inheritance.

As BMD/BMC at the FN and LS are correlated, and that covariation at the two sites is under genetic influence (7), a principal component analysis was used to extract the components, and this was followed by an orthogonal rotation. Based on the factor loading scores for BMD and BMC, a combined score representing both LS and FN BMD and BMC were derived. This score as well as each of the two BMD measurements separately were then subject to complex segregation analysis (28). Complex segregation analysis, with the MAN statistical program (29), was used to model the patterns in heritance. This analysis models that BMD/BMC (Y) being independently and additively determined by a MG effect (MG), a transmitted multifactorial (possibly polygenic) effect (M), and a unique environmental effect (E), e.g. schematically Y = µ + MG + M + E, where µ is the overall mean. The MG effect is assumed to result from a single autosomal locus with two alleles, A and a, whose distribution of the three hypothetical genotypes (AA, Aa, and aa) is in Hardy-Weinberg equilibrium from the allele frequency P. In this analysis, allele A was assumed to be responsible for low values of BMD. Apart from the parameter of allele frequency, P, the following parameters were also considered: genotypic means (µg) in all individuals with genotype g at the MG locus, where g = AA, Aa, or aa; genotypic variance ({sigma}2g); the transmission probabilities of the MG from parent to offspring ({tau}g, so that {tau}AA is the probability an AA individual transmits allele A to the offspring, {tau}Aa is the probability an Aa individual transmits allele A, and {tau}aa is the probability an aa individual transmits allele A); and residual familial correlations (e.g. multifactorial effects) adjusted for MG effect in spouses ({rho}), parent/offspring (ß), and siblings ({epsilon}). The pairwise correlations between residuals in any pair of pedigree members [as in program packages SAGE (30) and PAP (31)] are expressed through these parameters and depend additionally on the pedigree structure and the particular position of each relative pair. Under Mendelian transmission, {tau}AA = 1; {tau}Aa = 0.5, and {tau}aa = 0. When the three {tau}AA = {tau}Aa = {tau}aa, no transmission of the MG is obtained. The following three conditions are required to infer a MG: 1) rejection of the no MG effect hypothesis, 2) nonrejection of the Mendelian transmission hypothesis, and 3) rejection of the environmental (no transmission) hypothesis.

With the above parameters, a number of classes of models were considered in each segregation analysis: S1) the general mixed model, which assumes that all components were required to explain the observed pattern of BMD data on the pedigrees, it does not assume a particular mode of transmission, and therefore the parameters {tau}g were estimated together with other parameters; S2) the sporadic model assumes no intergenerational transmission of BMD values, i.e. constraining the MG effect and all residual familial correlations to zero; S3) the multifactorial model in which it was assumed that the familial correlations ({rho}, ß, and {epsilon}) can explain the familial aggregation of the trait considered, i.e. excluding the parameters representing the MG, and constraining P = 1, and µAA = µAa = µaa; S4) the mixed multifactorial-environmental model, in which there was environmental, and multifactorial (possibly polygenic) effects, but no transmission of MGs (e.g. so that all transmission probabilities were set equal to a constant, e.g. {tau}AA = {tau}Aa = {tau}aa and/or P = {tau}AA = {tau}Aa = {tau}aa); and S5) the Mendelian model, in which the {tau}g parameters were constrained as expected by the Mendelian transmission as follows: {tau}AA = 1, {tau}Aa = 0.5, and {tau}aa = 0.

Likelihood ratio tests were used to compare and choose among nested models. For two models, 1 and 2, with likelihoods L1 and L2, respectively, where model 1 is nested within model 2, -2Ln(L1/L2) is asymptotically distributed as a {chi}2 distribution with k degrees of freedom, where k is the difference, in number of estimated parameters, between the two models. The best-fitting and most parsimonious model (MP) was the model that included the minimal number of parameters estimated and did not significantly differ from the respective general model. To test the reliability of the most parsimonious model, a number of additional models were fitted to the data as shown in Results.

The variance of the variable adjusted for significant covariates, attributable to the putative MG, was estimated as VMG = {sum} fg (µg - µ)2/VTot.

The collected array of pedigrees violated the random sampling assumption due to identification of the first five large pedigrees via proband with BMD greater than 1.28 SD. To test and correct for this deviation, we undertook three different approaches. First, we evaluated to what extent the distribution of the putative MG genotypes in the parental generation violated the expectation of a random mating. One way to do this is to estimate parameter {varphi} ({phi}), which measures the correlation between the putative MG genotypes in parents. No correlation is expected under the random mating assumption (31, 32). Another approach is to maximize for parameters of P(AA) and P(AB), reflecting the frequencies of the corresponding putative genotypes independently, and not P(A), the parameter of allele A frequency (see above). Second, using the data from the five large pedigrees, the corresponding general model was constructed under the assumptions of no ascertainment bias and introducing the ascertainment correction following the method of Thompson and Canning (33) as implemented in the PAP package. Third, the whole sample was subjected to ascertainment correction following a general method of Ewens and Shute (34, 35) as implemented in the MAN-5 package. In this method the correction is achieved by conditioning of the pedigree likelihood on the part of the pedigree data included into a pedigree substructure file, defined as the pedigree substructure where probands can be found. The ascertainment correction was introduced differentially. That is, only the likelihood of the five pedigrees was corrected for the corresponding probability to be ascertained.


   Results
Top
 Abstract
 Introduction
 Subjects and Methods
 Results
Discussion
 References
 
Data from 128 pedigrees, with 256 men and 374 women, were used in the analysis. The individuals included 543 parent-offspring, 250 sibling, and 131 husband-wife pairs. The two genders were not significantly different in terms of age (56.2 ± 20.5 yr for men and 54.3 ± 17.9 yr for women; mean ± SD). However, as expected, weight, height, and BMD were significantly higher in men than in women (Table 1Go). In multiple linear regression analysis, age, body weight, height, and sex were each significantly related to BMD and BMC. The four independent predictors collectively accounted for 23%, 47%, 48%, and 57% of the total variation in LS BMD, LS BMC, FN BMD, and FN BMC, respectively. The study variables that were adjusted for significant covariates were then considered as phenotypes of interest in the subsequent segregation analysis.


View this table:
[in this window]
[in a new window]
  		TABLE 1. Characteristics of study subjects

 
BMD and BMC at the LS and FN were strongly interrelated; therefore, principal component analysis was used to combine the four variables into a single variable. Only the first component showed a value greater than 1; therefore, the first component was retained. It accounted for 76% of the covariation in the four variables and was, therefore, used as an additional phenotype in the segregation analysis (Table 2Go). This phenotype is referred to as PC1-BMD/BMC in subsequent analyses and reports.


View this table:
[in this window]
[in a new window]
  		TABLE 2. Principal component analysis of BMD and BMC

 
There were positive and significant correlations in all BMD and BMC measurements considered for parent-offspring and siblings, and as expected, none of the correlations was statistically significant for spouses (Table 3Go). Using the formula developed by Rice et al. formula (33), it was estimated that the maximal proportion of variance in BMD/BMC attributable to genetic factors ranged from about 31–72%.


View this table:
[in this window]
[in a new window]
  		TABLE 3. Familial correlations in pedigrees

 
Before conducting the complex segregation analysis reported below, the effect of ascertainment of five large pedigrees on the parameter estimates was examined. The results of all three types of analysis indicated that introduction of the additional parameters into a most parsimonious model to correct for non-Hardy-Weinberg distribution of genotypes in the parental generation did not significantly improve (P = 0.095–0.560) the model’s likelihood. Testing separately only five large pedigrees with and without introduction of the ascertainment correction resulted in very similar parameter estimates. Moreover, in both instances the models assuming no MG effect were clearly rejected (P < 0.003 to P < 0.001), whereas the models including the Mendelian transmission probabilities were accepted with P > 0.30. Finally, the results (not shown) of the analyses of the total sample that included Ewens and Shute (33, 34) ascertainment correction and those without correction were remarkably similar. Again for all three phenotypes the corresponding parameter estimates showed only minor difference. Thus, all three types of analysis unequivocally and consistently indicated a negligible effect of the five pedigrees ascertainment of the parameter estimates for all studied BMD/BMC phenotypes.

In the complex segregation analysis for PC1 (Table 4Go), the sporadic model (S2), multifactorial model (S3), and mixed environmental models (S4) did not fit the data adequately and were all rejected by the likelihood ratio test (with P ranging from 0.048 to <0.001). However, the model with a MG characterized by the Mendelian transmission (S5) fitted the data well ({chi}2 = 2.02; df = 3; P = 0.57). The estimated {tau} parameters (1.0, 0.46, and 0.06) from the general transmission model strongly favor those predicted under Mendelian segregation. Therefore, the magnitude of heritability and mode of inheritance under the assumption of Mendelian transmission were further tested under the formulation of the parsimonious model. As the estimated residual correlations for spouses ({rho}) and parent-offspring (ß) were low, the two parameters were then constrained to zero, and the fit of the new model was still adequate ({chi}2 = 4.20; df = 7; P = 0.76). Therefore, the best-fitting and the most parsimonious model was the one with the expected Mendelian transmission probabilities and residual siblings correlation (column MP). Further testing of the MP vs. the corresponding model with arbitrary estimates of the parameters {tau}g (MP1) and the model denying the possible MG effect (MP2) suggested the likelihood of the MG effect. The amount of genetic variance attributable to the MG effect was 53 ± 7% (h2 ± SE).


View this table:
[in this window]
[in a new window]
  		TABLE 4. Complex segregation analysis of PCI-BMD/BMC adjusted for significant covariates

 
For LS BMD (Table 5Go) and FN BMD (Table 6Go), the results were similar to those obtained for PC1. In both analyses the sporadic model (S2) was strongly rejected (P < 0.001 for both LSBMD and FNBMD). On the other hand, although the simplest multifactorial model (S3; without MG effect) provided a better fit to the data than did the sporadic model, the fit was still not adequate ({chi}2 = 36.7; df = 8; P < 0.001 for LSBMD; and {chi}2 = 38.8; df = 8; P < 0.001 for FNBMD). A comparison of the environmental model (S4) and the mixed Mendelian model (S5) provided strong evidence that the Mendelian model gave the better fit, as did a more restricted Mendelian model (MP) compared with the environmental model. Therefore, a Mendelian component appeared to be important. A mixed model containing both a multifactorial component and a MG component fit the data significantly better than the submodel model containing only the multifactorial component or the MG component alone (P ranged between 0.03–0.009 for LSBMD and FNBMD). Further evaluation among Mendelian submodels suggested that the MG model with the additive alleles interaction was superior to the recessive and dominant MG models. The most parsimonious model also contained significant residual sibling correlations. The contribution of the MG locus explained between 30 ± 9% and 51 ± 7% of the variance in LS and FN BMD, respectively.


View this table:
[in this window]
[in a new window]
  		TABLE 5. Complex segregation analysis of LS-BMD adjusted for significant covariates

 

View this table:
[in this window]
[in a new window]
  		TABLE 6. Complex segregation analysis of FN-BMD adjusted for significant covariates

 

   Discussion
Top
 Abstract
 Introduction
 Subjects and Methods
 Results
 Discussion
References
 
Epidemiological and genetic studies have consistently shown that BMD, a measure of bone strength, is a primary determinant of fracture risk in the general population (1, 2, 3), that a large component of the variance of BMD (up to 80%) is determined by genetic factors (6, 7, 8), and that a family history of fracture confers a significant increase in the risk of fracture for a relative (36, 37, 38, 39). However, there are still several issues concerning the genetics of BMD to be resolved. Among these are: what is the genetic architecture underlying the observed variation in bone density; what are the modes of intergenerational transmission; and how large is the contribution of MGs, if any? Molecular genetic data obtained to date are still controversial and inconsistent, although a number of genome-wide scans for genes associated with BMD variation in humans have been carried out over the past years (14, 15, 16, 17). Two main problems can be indicated in this regard. First, the revealed chromosomal areas varied largely between studies. Second, within each study there were practically no chromosomal segments (DNA markers) common for BMD measurements at different skeletal sites.

On the other hand, Devoto and colleagues (40) recently obtained molecular data that strongly supported the hypothesis that a MG (located on chromosome lp36.2–36.3) controls a large proportion of the genetic variation in FN-BMD in Canadian families.

In this study complex segregation analysis provided evidence that a putative MG codominant model best explains the inheritance pattern for BMD. Furthermore, complex segregation analysis suggests that a codominant MG model best explains the inheritance pattern for BMD/BMC. The results are also consistent with earlier analyses that there are genetic components shared by LS and FN BMD, with additional genes contributing to LS and not FN BMD.

The finding of a MG effect is consistent with previous studies in two ethnic populations from Russia (18, 19, 20) and in Caucasian populations in the United States (22) and New Zealand (23), but it is at variance with the findings reported by Guegen et al. (21). Differences in sampling strategies, clinical characteristics of study subjects, and analytical methods may contribute to these different conclusions. Although the sampling strategy and study subjects in the study by Deng et al. (22) are similar to ours, others are not. For example, in the study by Cardon et al. (23) the pointers for selected families had idiopathic osteoporosis, and in this study the researchers found only MG effects, but no polygenic component, in the genetic determination of BMD. On the other hand, it is difficult to compare the present results with those obtained by Guegen et al. (21) because of a contradiction between the likelihoods of the polygenic model vs. the MG model in their analysis (19).

The search for genetic loci affecting BMD/BMC is intensive, but there are few data on segregation analyses. To date, a few genes have been identified that affect BMD, but none with a major effect. The chance of identifying a QTL genomic regions through a linkage or transmission linkage disequilibrium approach depends more critically on the degree of genetic determination by MGs than on the overall heritability of a complex trait (14). The present analysis suggests that the magnitude of a MG effect (30–50% of the total genetic variance) is quite significant. If there is indeed a single gene accounting for this variance, that gene should be fairly easy to localize. On the other hand, the analyses also suggest that there are probably many other genes with probably small effects, and the localization of these genes may prove to be a major, if not intractable, challenge.

Successful dissection of the genetic components of osteoporosis will require careful consideration of the evidence for underlying heterogeneity of the bone phenotype. The approach taken here is a step in this direction, identifying evidence for a genetic basis of BMD and BMC phenotypes that are primary predictors of fracture. In any skeletal region, these two traits are highly correlated, but past studies have been primarily based on univariate analysis, in which each phenotype (mostly BMD) has been analyzed separately. However, as in the case of multiple comparison, multiple univariate statistical analyses ultimately increase the type I error rate under the complete null hypothesis of no segregation. Procedures for correction, such as Bonferroni adjustment, can be applied; however, the results may be too conservative, because the phenotypes are correlated. Therefore, a multivariate analysis may be preferable. It has been estimated that linear combination of BMD and BMC into a factor score, as we have done here, is more powerful than the use of multivariate or mean phenotypic data (22). Indeed, the MG effect suggested by the present analysis was the greatest in this combined phenotype compared with BMD or BMC in the FN or LS alone. These data suggest that future study of genetic linkage should be based on a combined phenotype rather than individual phenotypes.

It is interesting to observe that the effect of common familial environments on the covariation of BMD and BMC is not significant, which is consistent with a number of previous twin (6, 7, 8) and family (13, 14) studies. This suggests that although many environmental factors, such as lifestyle, nutrition, and physical activity, are important contributors to BMD variation, their influences are not sufficient to influence the resemblance among family members due to shared environment.

However, there are a number of issues to consider in interpreting the results. The study used a large number of families, some of which were very large, which increased the power to detect a Mendelian locus. Complex segregation analysis involves fitting an inherently single Mendelian locus, or major locus, model with possibly a polygenic component, to absorb the remaining familial effects. The power to detect the polygenic component, especially in nuclear families, is not high (41, 42). No tools exist for easy estimation of power in this context, so it is not possible to state exactly what this would be in the current dataset; however, we could discriminate among several possible genetic models. Some families used in this analysis were selected through a moderately high FN BMD proband, so they may be segregating component phenotypes that are specifically related to FN, and FN may provide stronger evidence for an MG effect than does LS. These results from subjects of Caucasian background may not be generalizable to other populations.

In summary, these findings support the hypothesis that a large component of the variance in BMD is under genetic control, and that there exists a MG locus influencing BMD and BMC. These results also provide evidence that the study of large pedigrees identified via selected probands may increase the probability of finding a MG for variation in bone phenotype.


   Acknowledgments
 
We gratefully acknowledge the expert assistance of Janet Watters and Donna Reeves with the interview, data collection, and measurement bone densitometry, and the invaluable help of the staff of Dubbo Base Hospital. We thank Dr. I. Malkin for discussion and advice concerning the ascertainment correction matter. We gratefully acknowledge Mr. J. McBride for the development of and Mrs. N. Ivankovic for the management and maintenance of the complex genetic database system.


   Footnotes
 
This work was supported by the National Health and Medical Research Council of Australia.

Abbreviations: BMC, Bone mineral content; BMD, bone mineral density; DOES, Dubbo Osteoporosis Epidemiology Study; DOGS, Dubbo Osteoporosis Epidemiology Study; FN, femoral neck; LS, lumbar spine; MG, major gene.

Received December 31, 2002.

Accepted April 28, 2003.


   References
Top
 Abstract
 Introduction
 Subjects and Methods
 Results
 Discussion
 References
 

  1. Cummings SR, Melton LJ 2002 Epidemiology and outcomes of osteoporotic fractures. Lancet 359:1761–1767[CrossRef][Medline]
  2. Center JR, Nguyen TV, Schneider D, Sambrook PN, Eisman JA 1999 Mortality after all major types of osteoporotic fracture in men and women: an observational study. Lancet 353:878–882[CrossRef][Medline]
  3. Nguyen TV, Sambrook P, Kelly P, Jones G, Lord S, Freund J, Eisman J 1993 Prediction of osteoporotic fractures by postural instability and bone density. Br Med J 307:1111–1115
  4. Melton III LJ, Atkinson EJ, O’Fallon WM, Wahner HW, Riggs BL 1993 Long term fracture prediction by bone mineral assessed at different skeletal sites. J Bone Miner Res 8:1227–1233[Medline]
  5. Marshall D, Johnell O, Wedel H 1996 Meta-analysis of how well measures of bone mineral density predict occurrence of osteoporotic fractures. BMJ 312:1254–1259[Abstract/Free Full Text]
  6. Pocock NA, Eisman JA, Hopper JL, Yeates MG, Sambrook PN, Eberl S 1987 Genetic determinants of bone mass in adults: a twin study. J Clin Invest 80:706–710
  7. Nguyen TV, Howard GM, Kelly PJ, Eisman JA 1998 Bone mass, lean mass and fat mass: same genes or same environmemts. Am J Epidemiol 147:3–16[Abstract/Free Full Text]
  8. Flicker L, Hopper JL, Rogers L, Kaymacki B, Green RM, Wark JD 1995 Bone mineral density determinants in elderly women: a twin study. J Bone Miner Res 10:1607–1613[Medline]
  9. Lutz J, Tesar R 1990 Mother-daughter pairs: spinal and femoral bone densities and dietary intakes. Am J Clin Nutr 52:872–877[Abstract/Free Full Text]
  10. Krall EA, Dawson-Hughes B 1993 Heritability and life-style determinants of bone mineral density. J Bone Miner Res 8:1–9[Medline]
  11. Deng HW, Chen WM, Conway T, Zhou Y, Davies KM, Stegman MR, Deng H, Recker RR 2000 Determination of bone mineral density of the hip and spine in human pedigrees by genetic and life-style factors. Genet Epidemiol 19:160–177[CrossRef][Medline]
  12. Khoury MJ 1997 Genetic epidemiology and the future of disease prevention and public health. Epidemiol Rev 19:175–180[Abstract/Free Full Text]
  13. Nguyen TV, Blangero J, Eisman JA 2000 Genetic epidemiological approaches to the search for osteoporosis genes. J Bone Miner Res 15:392–401[CrossRef][Medline]
  14. Deng HW, Xu FH, Huang QY, Shen H, Deng H, Conway T, Liu YJ, Liu YZ, Li JL, Zhang HT, Davies KM, Recker RR 2002 A whole-genome linkage scan suggests several genomic regions potentially containing quantitative trait Loci for osteoporosis. J Clin Endocrinol Metab 87:5151–5159[Abstract/Free Full Text]
  15. Devoto M, Shimoya K, Caminis J, Ott J, Tenenhouse A, Whyte MP, Sereda L, Hall S, Considine E, Williams CJ, Tromp G, Kuivaniemi H, Ala-Kokko L, Prockop DJ, Spotila LD 1998 First-stage autosomal genome screen in extended pedigrees suggests genes prediposing to low bone mineral density on chromosomes 1p, 2p, and 4q. Eur J Hum Genet 6:151–157[CrossRef][Medline]
  16. Duncan EL, Brown MA, Sinsheimer J, Bell J, Carr AJ, Wordsworth BP, Wass JA 1999 Suggestive linkage of the parathyroid receptor type 1 to osteoporosis. J Bone Miner Res 14:1993–1999[CrossRef][Medline]
  17. Koller DL, Rodriguez LA, Christian JC, Slemenda CW, Econs MJ, Hui SL, Morin P, Conneally PM, Joslyn G, Curran ME, Peacock M, Johnston CC, Foroud T 1998 Linkage of a QTL contributing to normal variation in bone mineral density to chromosome 11q12–13. J Bone Miner Res 13:1903–1908[CrossRef][Medline]
  18. Livshits G, Karasik D, Kobyliansky E 2002 Complex segregation analysis of the radiographic phalanges bone mineral density and their age-related changes. J Bone Miner Res 17:152–161[CrossRef][Medline]
  19. Livshits G, Karasik D, Pavlovsky O, Kobyliansky E 1999 Segregation analysis reveals a major gene effect in compact and cancellous bone mineral density in 2 populations. Hum Biol 71:155–172[Medline]
  20. Livshits G, Pavlovsky O, Kobyliansky E 1996 Population biology of human aging: segregation analysis of bone age characteristics. Hum Biol 68:540–554[Medline]
  21. Gueguen R, Jouanny P, Guillemin F, Kuntz C, Pourel J, Siest G 1995 Segregation analysis and variance components analysis of bone mineral density in healthy families. J Bone Miner Res 10:2017–2022[Medline]
  22. Deng HW, Livshits G, Yakovenko K, Xu FH, Conway T, Davies KM, Deng H, Recker RR 2002 Evidence for a major gene for bone mineral density/content in human pedigrees identified via probands with extreme bone mineral density. Ann Hum Genet 66:61–74[CrossRef][Medline]
  23. Cardon LR, Garner C, Bennett ST, Mackay IJ, Edwards RM, Cornish J, Hegde M, Murray MA, Reid IR, Cundy T 2000 Evidence for a major gene for bone mineral density in idiopathic osteoporotic families. J Bone Miner Res 15:1132–1137[CrossRef][Medline]
  24. Deng HW, Stegman MR, Davies KM, Conway T, Recker RR 1999 Genetic determination of variation and covariation of peak bone mass at the hip and spine. J Clin Densitom 2:251–263[CrossRef][Medline]
  25. Allison DB, Thiel B, St Jean P, Elston RC, Infante MC, Schork NJ 1998 Multiple phenotype modeling in gene-mapping studies of quantitative traits: power advantages. Am J Hum Genet 63:1190–1201[CrossRef][Medline]
  26. Boomsma DI, Dolan CV 1998 A comparison of power to detect a QTL in sib-pair data using multivariate phenotypes, mean phenotypes and factor scores. Behav Genet 28:329–340[CrossRef][Medline]
  27. Jones G, Nguyen T, Sambrook PN, Kelly PJ, Gilbert C, Eisman JA 1994 Symptomatic fracture incidence in elderly men and women: the Dubbo Osteoporosis Epidemiology Study (DOES). Osteoporos Int 4:277–282[CrossRef][Medline]
  28. Elston RC 1981 Segregation analysis. Adv Hum Genet 11:63–120[Medline]
  29. Ginsburg E 1997 Program package for Mendelian analysis of pedigree data (MAN, version 4). Technical Report. Tel Aviv: Sackler Faculty of Medicine, Tel Aviv University
  30. Elston RC 1995 S.A.G.E. statistical analysis for genetic epidemiology. Cleveland: Case Western Reserve University
  31. Hasstedt SJ 1994 Pedigree analysis package. Salt Lake City: University of Utah
  32. Ginsburg E, Livshits G 1999 Segregation analysis of quantitative traits. Ann Hum Biol 26:103–129[CrossRef][Medline]
  33. Cannings C, Thompson EA 1977 Ascertainment in sequential sampling of pedigrees. Clin Genet 12:208–212[Medline]
  34. Ewens WJ, Shute NCE 1986 A resolution of the ascertainment sampling problem. I. Theory. Theory Pop Biol 30:388–412
  35. Shute NCE, Ewens WJ 1988 A resolution of the ascertainment sampling problem. II. Generalization and numerical results. Am J Hum Genet 43:374–386[Medline]
  36. Keen RW, Hart DJ, Arden NK, Doyle DV, Spector TD 1999 Family history of appendicular fracture and risk of osteoporosis: a population-based study. Osteop Int 10:161–166[CrossRef][Medline]
  37. Fox KM, Cummings SR, Powell-Threets K, Stone K 1998 Family history and risk of osteoporotic fracture. Study of Osteoporotic Fractures Research Group. Osteop Int 8:557–562[CrossRef][Medline]
  38. Deng HW, Chen WM, Recker S, Stegman MR, Li JL, Davies KM, Zhou Y, Deng H, Heaney R, Recker RR 2000 Genetic determination of Colles’ fracture and differential bone mass in women with and without Colles’ fracture. J Bone Miner Res 15:1243–1252[CrossRef][Medline]
  39. Soroko SB, Barrett-Connor E, Edelstein SL, Kritz-Silverstein D 1994 Family history of osteoporosis and bone mineral density at the axial skeleton: the Rancho Bernardo Study. J Bone Miner Res 9:761–769[Medline]
  40. Devoto M, Specchia C, Hui-Hua Li, Caminis J, Tenenhouse A, Rodriguez H, Spotila LD 2001 Variance component linkage analysis indicates a QTL for femoral neck bone mineral density on chromosome 1p36. Hum Mol Genet 10:2447–2452[Abstract/Free Full Text]
  41. Allison DB, Heo M, Kaplan N, Martin ER 1999 Sibling-based tests of linkage and association for quantitative traits. Am J Hum Genet 64:1754–1763[CrossRef][Medline]
  42. MacLean CJ, Morton NE, Lew R 1975 Analysis of family resemblance. IV. Operational characteristics of segregation analysis. Am J Hum Genet 27:365–384[Medline]



This article has been cited by other articles:


Home page
 J. Cell Sci.Home page
L. S. H. Nissen-Meyer, R. Jemtland, V. T. Gautvik, M. E. Pedersen, R. Paro, D. Fortunati, D. D. Pierroz, V. A. Stadelmann, S. Reppe, F. P. Reinholt, et al.
Osteopenia, decreased bone formation and impaired osteoblast development in Sox4 heterozygous mice
J. Cell Sci., August 15, 2007; 120(16): 2785 - 2795.
[Abstract] [Full Text] [PDF]

Home page
 J. Clin. Endocrinol. Metab.Home page
N. D. Nguyen, J. A. Eisman, J. R. Center, and T. V. Nguyen
Risk Factors for Fracture in Nonosteoporotic Men and Women
J. Clin. Endocrinol. Metab., March 1, 2007; 92(3): 955 - 962.
[Abstract] [Full Text] [PDF]

Home page
 J. Clin. Endocrinol. Metab.Home page
M. Peacock, D. L. Koller, T. Fishburn, S. Krishnan, D. Lai, S. Hui, C. C. Johnston, T. Foroud, and M. J. Econs
Sex-Specific and Non-Sex-Specific Quantitative Trait Loci Contribute to Normal Variation in Bone Mineral Density in Men
J. Clin. Endocrinol. Metab., May 1, 2005; 90(5): 3060 - 3066.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Submit a related Letter to the Editor
Right arrow Purchase Article
Right arrow View Shopping Cart
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow Request Copyright Permission
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Nguyen, T. V.
Right arrow Articles by Eisman, J. A.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Nguyen, T. V.
Right arrow Articles by Eisman, J. A.
Right arrowPubmed/NCBI databases
*OMIM

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Endocrinology Endocrine Reviews J. Clin. End. & Metab.
Molecular Endocrinology Recent Prog. Horm. Res. All Endocrine Journals