• Bone mineral density;
  • Chinese population;
  • Familial correlation;
  • Segregation analysis;
  • Major gene


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

China has the largest population in the world; approximately 7% of the total population suffers from primary osteoporosis. Osteoporosis is mainly characterized by low bone mineral density (BMD). In the present study, familial correlation and segregation analyses for spine and hip BMDs have been undertaken for the first time in a Chinese sample composed of 401 nuclear families with a total of 1,260 individuals. The results indicate a major gene of additive inheritance for hip BMD, whereas there is no evidence of a major gene influencing spine BMD. Significant familial residual effects are found for both traits, and heritability estimates (±SE) for spine and hip BMDs are 0.807(0.099) and 0.897(0.101), respectively. Sex and age differences in genotype-specific average BMD are also observed. This study provides the first evidence quantifying the high degree of genetic determination of BMD variation in the Chinese.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Osteoporosis is a complex disease involving fragile bones and high susceptibility to low-trauma fractures (Consensus Development Conference, 1993). It is a serious health problem, especially in elderly women (Cummings & Melton, 2002). Low bone mineral density (BMD) is a major risk factor for osteoporotic fracture, and osteoporosis is mainly characterized by low BMD (Melton et al. 1989; Cummings & Melton, 2002).

Twin and family studies indicate BMD variation is under strong genetic control, with heritability estimates ranging from 0.5–0.9 (Dequeker et al. 1987; Slemeda et al. 1991; Sowers et al. 1992; Gueguen et al. 1995; Livshits et al. 1998; Deng et al. 1999a, 2000). Several segregation analyses have yielded evidence consistent with major gene influences on BMD for the hand phalange (Livshits et al. 1996, 1999, 2002), the lumbar spine (Cardon et al. 2000) and a composite measure of the spine, hip and wrist, (Deng et al. 2002a), while another study has rejected models of a major gene effect on total body BMD (Gueguen et al. 1995). Most recently, extensive molecular genetic studies have been performed to search for genes underlying BMD variation (Morrison et al. 1994; Johnson et al. 1997; Deng et al. 1998, 1999b, 2001; Koller et al. 1998, 2000; Gong et al. 1999). The results so far, from extensive population association studies and linkage studies, have largely been inconsistent (Eisman, 1995; Peacock, 1995; Econs & Speer, 1996; Devoto et al. 1998; Duncan et al. 1999; Gong et al. 1999; Niu et al. 1999; Koller et al. 2000; Deng et al. 2001, 2002b,c). Whether BMD variation occurs with an oligogenic or polygenic mode of inheritance, or as a mixture of the two, is still under debate (Gueguen et al. 1995; Livshits et al. 1996, 1999, 2002; Cooper, 1999; Cardon et al. 2000; Uitterlinden, 2001; Deng et al. 2002a).

Numerous previous studies have shown that significant ethnic variation, along with genetic factors, plays an important role in the determination of bone mass (Pollitzer & Anderson, 1989; Davis et al. 1994, 1999; Plato et al. 1994; Wang et al. 1997; Bachrach et al. 1999). Differences in BMD between blacks and whites remain even after adjustment for body mass (Pollitzer & Anderson, 1989). These findings are naturally interpreted as showing strong evidence for population heterogeneity in BMD variation. A few studies have also revealed that potential sex- and age-genotype specific interactions are strongly related to the rate of bone loss, due possibly to different expression of genes at certain ages and/or in different sexes (Karasik et al. 2000; Deng et al. 2002a; Livshits et al. 2002).

Most segregation analyses have been performed to explore the pattern of familial aggregation for BMD in Caucasian populations. No such study has been performed in Chinese populations. To the best of our knowledge, even the heritability of BMD variation in Chinese populations has not yet been quantitatively determined. Thus, the objective of the present study is to 1) determine the heritability of BMD adjusted for several significant factors such as age, height and weight; and 2) examine the mode of inheritance of BMD variation in Chinese nuclear families by incorporating sex and age effects directly into a major gene penetrance function.

Materials and Methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References


The study was approved by the Research Administration Departments of Shanghai's Sixth People's Hospital and the Hunan Normal University. We recruited 401 nuclear families composed of both parents and at least one healthy female child, whose age was generally between 20–40 years, with 1,260 total individuals; all the children are daughters. The average family size is 3.14, and 348, 50, 2, and 1 families had 1, 2, 3, and 4 children, respectively. The mother of each family was recruited randomly from the patients visiting the clinic of the Center for Preventing and Treating Osteoporosis in Shanghai's Sixth Hospital. All the subjects involved in the study came from a local population of Shanghai City, located on the Mid-East coast of P. R. China, and belong to the Han ethnic group, the majority group in China that composes more than 90% of the total Chinese population of 1.3 billion (Liu et al. 2002). All subjects signed informed-consent documents before entering the project. For each study subject we also collected information on age, sex, medical history, family history, female history, physical activity, alcohol use, diet habits, smoking history, etc. Only families with healthy children (defined by the exclusion criteria below) were included in the analyses. The exclusion criteria (Deng et al. 2002b) for the study subjects of healthy children 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 non-genetic cause of low bone mass; 14) any disease, treatment, or condition that would be a non-genetic cause for low bone mass. The exclusion criteria were assessed by nurse-administered questionnaires and/or medical records.


BMDs of spine and hip were measured by a Hologic QDR 2000+ dual-energy X-ray absorptiometry (DXA) scanner (Hologic Inc., Bedford, MA, USA). For the spine, the quantitative phenotype was combined BMD of L1-L4. For the hip, it was combined BMD of the femoral neck, trochanter, and intertrochanteric region. The machine is calibrated daily, and the coefficients of variability (CV) values, which are obtained from 7 individuals repeatedly measured five times, of the DXA measurements at the spine and hip are 0.9% and 0.8%, respectively. At the same visit as the BMD measurement, weight was measured without shoes and in light indoor clothing, using a calibrated digital or balance beam scale; height was measured using a calibrated stadiometer.

Data Adjustments

Before the genetic analyses, the data were examined for dependence on age, height and weight. Univariate analyses showed that all covariates are potentially significantly correlated with BMDs of the spine and hip. Weight was the most important predictor; for example, it accounted for 11% and 24% of the BMD variation at the hip in women and men, respectively. For familial correlation analysis, the data were adjusted for age, age2, weight and height effects by a multiple stepwise regression procedure (carried out separately for each sex). The covariates used to generate residuals were those retained in stepwise regression using a significance level of 0.1 for both inclusion and retention in the model. For segregation analysis, the data were only adjusted for weight and height. The gender specific residuals were used as a phenotypic variable in both familial correlations and segregation analyses.

Familial Correlation Analyses

Familial correlations (spousal, parent-offspring and sibling) and their equivalent pair counts were calculated using the FCOR program in Statistical Analysis for Genetic Epidemiology (S.A.G.E. 4.1, 2001). The 95% confidence intervals for these correlations were constructed using Fisher's z transformation (DeStefano et al. 1996). Heritability was therefore estimated as,

  • image

where rsib, rp -o and rsp are correlation coefficients for sibling pairs, parent-offspring pairs and spouse pairs, respectively. An approximation of the standard error of heritability can be obtained using exactly the same equation as above, but replacing the correlations with their corresponding standard errors (Rice et al. 1997). Note that we here first assessed whether the familial correlations were significantly different from zero for each trait. Then, the model chosen on the basis of evaluating familial correlations was used to test different hypotheses.

Segregation Analyses

Segregation analysis was performed using the program SEGREG, as implemented in the S.A.G.E. 4.1 package (2001). The class D regressive model was employed for continuous traits, assuming that the sibling correlations are equal (Bonney, 1984). Sex and age accounting for interaction with the putative major gene were incorporated directly into the genetic model. The phenotypic data were transformed with Box-Cox transformation (S.A.G.E. 2001), either to correct for non-normality and/or a heteroscedastic variance structure.

The segregation of a possible major locus is allowed for by letting the mean (and variance also) depend on an unobserved qualitative (genetic or non-genetic) factor g (g = AA, AB and BB). The parameter g is an individual's type (Go et al. 1978), which allows for a major effect; either a major gene locus or a random major environmental effect. Genotypes are the special case of types that transmit to offspring in Mendelian fashion (S.A.G.E. 2001). Parameters usually estimated in these models are listed below (for more details to see the S.A.G.E. 4.1 User Manual, 2001); 1) qA the population frequency of the allele A; allele A is assumed to be a cause for low value of the quantitative trait; 2) μg the mean of type g; where type g in a Mendelian or major gene inheritance model corresponds to genotypes AA, AB and BB, respectively; 3) σ2g, the trait variance in individuals having the same type g; 4) τg, is the transmission parameter, estimates the probability that a parent of type g transmits allele A to an offspring; τAA, τAB and τBB take the expectation 1, 0.5 and 0 respectively in Mendelian fashion; 5) βs, is the covariate coefficient of sex, which determines the extent of sex influence on the overall mean of all types; it measures the differences of the average BMD between sexes (female and male); 6) βsg, is the sex-type-specific covariate coefficient, which determines the extent of sex influence on mean of the specific type g; it measures the differences of the average BMD within a specific type g between sexes, i.e., sex by type interactions. A natural interpretation of sex–genotype interaction is that a specific genotype may play a different role in different sexes; 7) βa, is the covariate coefficient of age; it measures the differences of the average BMD among ages; 8) βag, is the age-type-specific covariate coefficient; it measures the differences of the average BMD within a specific type g among ages, i.e., age by type interactions. Age –genotype interaction demonstrates that a specific genotype may play different role during different stages of life; 9) ρsib, ρpo and ρsp are correlations between the trait residuals in sibs, parents/offspring and spouses, respectively; these correlations measure the magnitude of covariations between relatives which are attributable to polygenic and/or environmental effects other than major effects. The parameters described above are also referred to in several previous studies (Karasik et al. 2000; Deng et al. 2002a; Livshits et al. 2002).

With different restrictions to the above parameters, it is possible to model various genetic and non-genetic models. In the present study, the following series of competing models were examined. (1) Mendelian model, which is characterized by fixing τAA= 1.0, τAB= 0.5 and τBB= 0.0, corresponding to additive, codominant, dominant and recessive modes. (2) Environmental model, which is undertaken by setting qAAAABBB, allowing the presence of a random unmeasured environmental major effect. (3) No major gene model, which shows only one type of distribution, and the parameters qA and τg are ignored. (4) Unrestricted (general) model, which adjusts all parameters to the empirical data without restrictions, thereby providing the best fit to the data. This model provides a “baseline” against which the other hypotheses can be tested.

The likelihood ratio test (LRT) was used to test each specified reduced model against the unrestricted (general) model, and was calculated as the difference in negative two times the natural log likelihoods (–2lnL) between the two models. This difference is asymptotically distributed as a χ2 distribution with degrees of freedom equal to the difference in the number of parameters estimated in the two models. The significance level was set to 0.05. Akaike's Information Criteria (AIC), which is defined as AIC =−2lnL+2 (number of parameters estimated), was also used to compare any two competing models. The most parsimonious model has the minimum AIC value (Akaike, 1974).


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Descriptive Statistics and Familial Correlations

Descriptive statistics for BMDs at the spine and hip, and the covariates of age, height and weight are summarized for fathers, mothers and daughters separately (Table 1). The mean BMDs unadjusted for any covariates at the spine and hip are significantly different among fathers, mothers and daughters (P < 0.01). Daughters (at the ages of peak bone mass) have the highest mean BMD at the spine, whereas fathers have the highest mean BMD at the hip. Mothers generally have the lowest mean BMDs at both the spine and hip.

Table 1.  Descriptive statistics for the 1,260 subjects in 401 Chinese families (mean ± standard deviation)a
 Fathers (n = 386)Mothers (n = 384)Daughters (n = 458)
  1. aThe reported BMDs are for unadjusted BMD values at the spine and hip. Some observations are not available for a few subjects.

Age (years) 62.41 ± 6.63 59.18 ± 6.53 31.37± 5.77
Height (cm)166.16 ± 6.06154.64 ± 5.59159.84 ± 5.16
Weight (kg) 68.41 ± 9.96 59.18 ± 8.58 55.08 ± 8.01
Spine BMD (g/cm2) 0.928 ±0.145 0.812 ± 0.151 0.960 ± 0.102
Hip BMD (g/cm2) 0.874 ± 0.122 0.749 ± 0.134 0.855 ± 0.108

The familial correlation coefficients and 95% confidence intervals are shown in Table 2. All parent-offspring and sib-sib correlations for BMDs at the spine and hip are positive and highly significant (P < 0.05), suggesting a familial aggregation of the studied traits. However, the spouse correlations are not significantly different from zero (χ21= 0.002, P= 0.964 and χ21= 2.072, P= 0.150 for BMDs at the spine and hip, respectively). This is consistent with the results of Deng et al. (1999a, 2002a), indicating that common familial living environment may contribute little to the covariation of BMD between relatives. Thus, for segregation analysis, the model with spouse correlations fixed to zero, allowing parent-offspring and sib-sib correlations to be incorporated, was used throughout in estimation. The heritability estimates (±SE) for BMDs at the spine and hip were determined to be 0.807 (0.099) and 0.897 (0.101) respectively, which fall into the upper range of the heritability estimates reported elsewhere (Recker & Deng, 2002).

Table 2.  Correlation coefficients (95% confidence interval) and heritability estimates (mean ± SE) adjusted for age, age2, height and weight
Familial PairsSpine BMDHip BMD
Spouse−0.002 (−0.104, 0.100)−0.075 (−0.176, 0.027)
Parent-offspring0.352 (0.290, 0.411)0.340 (0.277, 0.399)
Sib-sib0.454 (0.206, 0.647)0.508 (0.272, 0.685)
Heritability0.807 ± 0.0990.897 ± 0.101

Segregation Analyses

Incorporation of the covariates of sex (denoting male and female as 0 and 1, respectively) and age (year), accounting for interaction with a major gene, results in significant improvement of the likelihood estimates. For example, for the spine BMD the maximum log-likelihood values (−2lnL) for the unrestricted model involving interaction with the covariate of age, the covariate of sex, and with both age and sex, are 3741.64, 3769.31, and 3730.26, respectively. However, the –2lnL is 3783.25 for the unrestricted model without interactions with covariates of age and sex (here the difference between any two nested models is greater than X2df=3= 7.81 and X2df=6= 12.59). Similar results are observed for the hip BMD. Compared with non-transformed data, the data with Box-Cox transformation also produced marginally better likelihoods in our segregation analyses. As mentioned above, spouse correlations are not significantly different from zero. Thus the final model presented here is for incorporating sex and age as covariates, with Box-Cox transformation, and with spouse correlations fixed to zero.

Results of segregation analyses of BMDs at the spine and hip are presented in Tables 3 and 4, respectively. For the spine BMD, the environmental model and the no major gene model are both significantly rejected against the unrestricted model (P < 0.001). However, the additive gene model is the second parsimonious model according to AIC score. It is interesting that all models show substantial residual familial correlations between parents and offspring, and in particular between sibs (ranging from 0.352-0.598). This is presumably because the polygenes make a substantial contribution to the BMD variation. For the hip BMD, the no major gene model is strongly rejected against the unrestricted model 212= 27.31, P= 0.007). The dominant and recessive gene models are also significantly rejected against the unrestricted model (χ25= 26.50, P < 0.001 and χ25= 14.40, P= 0.013). The additive and codominant gene models, and environmental model, are not significantly different from the unrestricted model (P > 0.05). According to AIC, the additive gene model, which has the minimum value of AIC, provides the most parsimonious fit to the data.

Table 3.  Segregation analysis for BMD at spine, adjusted for height and weight
 Mendelian major gene model 
  EnvironmentalNo MajorUnrestricted
  1. aValue fixed

  2. bValue fixed by maximization procedure MAXFUN

  3. cMultipled by 100

  4. dNumber of parameters estimated

−2 ln  L3800.003807.883855.093812.323841.043911.993730.26
P value<0.001<0.001<0.001<0.001<0.001<0.001
Table 4.  Segregation analysis for BMD at hip adjusted for height and weight
 Mendelian major gene model 
  EnvironmentalNo MajorUnrestricted
−2 ln  L3441.863439.773459.843447.743439.143460.653433.34
χ2(df)8.52(5)6.63 (3)26.50 (5)14.40 (5)5.80 (3)27.31 (12)
P value0.1300.097<0.0010.0130.1220.007

When the additive gene model is specified, the frequency of allele A is 0.693, and the means μAA and μBB converge at –0.216 and 0.507 respectively, indicating that the frequent allele A is associated with a lower BMD at the hip. This model also shows strong co-effects of sex and age on BMD. The mean BMD is decreased with increasing age in women. Sex and age differences in genotypic-specific average BMD are found (−2lnL are 3770.32 and 3494.51 for the additive gene model without interaction with age or sex, respectively). Women have lower BMD for genotypes AA and AB, and higher BMD for genotype BB. The opposite is observed in men. Furthermore, women and men with the genotypes AA and AB generally lose bone mass faster with increasing age than those with the genotype BB.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

China has the largest population in the world, and approximately 7% of this population suffers from primary osteoporosis (Liu et al. 2002). Osteoporosis is mainly characterized by low BMD (Cummings et al. 1985; Melton et al. 1989). In this study, familial correlations and segregation analyses for the spine and hip BMDs have been undertaken for the first time in a Chinese sample composed of 401 nuclear families with a total of 1,260 individuals. The results indicate a major gene of additive inheritance for the hip BMD, whereas evidence of a major gene influencing the spine BMD is not strong. Significant familial residual effects are found for both traits, and heritability (±SE) estimates for the spine and hip BMD are 0.807 (0.099) and 0.897 (0.101), respectively. Sex and age differences in genotype-specific average BMD are observed.

Our results support the existence of a major gene with additive effects on BMD variation at the hip, which is consistent with several segregation analyses performed earlier in Caucasian populations (Livshits et al. 1996, 1999, 2002; Cardon et al. 2000; Deng et al. 2002a). Livshits et al. (1996, 1999, 2002) reported a major gene with additive/codominant effects on BMD at the hand phalange, while Cardon et al. (2000) showed a major gene with codominant effects on BMD variation of the lumbar spine. More recently, Deng et al. (2002a) analyzed BMD and BMC at the spine, hip and wrist jointly, by employing factor scores (FS) of the principle component as the study phenotype, and indicated a major gene with codominant effect responsible for ∼16% of the FS variations.

Unlike hip BMD, our analyses do not provide strong evidence of a major gene influencing spine BMD in our Chinese sample, while a major gene for spine BMD has been suggested in Caucasian populations (Cardon et al. 2000). This result suggests that BMD variations at the hip and spine may differ in their mode of inheritance in the Chinese, as reflected by our results here. Several recent studies have implicitly or explicitly revealed such a site-specific genetic determination in BMD variation (Jones & Nguyen, 2000; Boyanov, 2001; Ryan, 2001). In addition, measurement of L1-L4 lumbar spine BMD was suspected to reflect age-related artifacts (e.g., aortic calcification, ossification of the anterior spinal ligament, spinal osteoarthritis etc.) that make phenotypic characterization more difficult (Slosman et al. 1990; von der Recke et al. 1996; Liu et al. 1997; Schwartz et al. 2001). These artifacts may contribute to the above discrepancies as well, since both parents in each nuclear family are not necessarily healthy in our studied sample.

In our results, significant familial correlations are found in parents/offspring pairs and sibling pairs. Similar results were also observed by Deng et al. (2002a), who reported that ∼56% FS variation is explained by residual polygenic effects. This is due to the fact that the major gene does not account for the majority of the similarity among family members, and that residual polygenic effects are present in determining BMD variation. It is naturally hypothesized that, based on our results and previous results (Gueguen et al. 1995; Livshits et al. 1996, 1999, 2002; Cardon et al. 2000; Deng et al. 2002a), at some skeletal sites. BMD variation may be attributed to at least one major gene acting in concert with minor polygenes simultaneously, which could react differently to environmental exposures. As such, it has been a challenge to dissect genetic factors underlying osteoporosis, which is generally investigated through studies of BMD as a surrogate phenotype.

Potential sex- and age-genotype specific interactions are of great interest in the determination of bone health status (Pocock et al. 1987; Sowers et al. 1992; Gueguen et al. 1995; Karasik et al. 2000). These interactions may encompass several possible underlying biological phenomena, including parent-of-origin effect (such as genomic imprinting) and differential gene expression during a life cycle. In our results, significant sex- and age-genotype interactions are found. This agrees well with the recent segregation analyses performed in randomly ascertained pedigrees from two ethnically different populations, the Chuvasha and Turkmenians (Livshits et al. 2002), and in proband ascertained pedigrees from the Midwestern United States of America which were mainly of western or northern European origin (Deng et al. 2002a). An important implication of these results is that the inheritance pattern of bone mass restoration and loss may be different between sexes and during various stages of life cycle.

In our analyses, heritability estimates for spine and hip BMDs are generally high, and are within the range, but near the upper limits, of previous family and twin studies (Dequeker et al. 1987; Slemeda et al. 1991; Sowers et al. 1992; Gueguen et al. 1995; Livshits et al. 1998; Deng et al. 1999a, 2000). The heritability estimated by the approach of Rice et al. (1997) is a maximal heritability estimate, precisely because there may be both genetic and familial environmental effects included in the estimate. However, in our samples spouse correlations are not significant at either the spine or the hip, suggesting that common environmental effects may not contribute significantly to the familial correlation between relatives. Therefore, the high heritability estimates in our study are due primarily to genetic factors. Jones & Nguyen (2000) found consistently and markedly higher heritability for bone density in mother-daughter pairs than in mother-son pairs. A related problem was also investigated by McKay et al. (1994), who reported consistently higher correlation coefficients for bone mass in mother-daughter pairs as compared with mother-son pairs. In our sample, only female children were recruited as nuclear families with their parents. Thus, the sample scheme may result in high estimates of heritability in our study.

In our samples, the nuclear families we generally small in size, with an average size of 3.14, which may limit the transmission information provided by family structure (Jarvik, 1998). Additional analyses were performed separately in two subsamples, trio families (n = 1044) and multi-sib families (n = 216). The hypothesis of the major gene model was clearly confirmed in the two subsamples (P= 0.10 to 0.31). In addition, our samples were recruited from a local population of Shanghai City. Since Shanghai is a metropolitan area and home to tens of millions of people from many Chinese ethnic groups from different geographical areas, there may exist population heterogeneity, even though all subjects in our study were from the Chinese Han ethnic group. However, a Hardy-Weinburg test for several candidate genes (such as the BsaHI polymorphism in the calcium-sensing receptor gene, SacI in the alpha 2HS-glycoprotein gene, PvuII in the estrogen receptor gene, ApaI in the vitamin D receptor gene, and BstBI in the parathyroid hormone gene) underlying BMD in our sample showed that our populations are basically in equilibrium (Deng et al. unpublished data).

In summary, our results suggest that there is a major gene of additive inheritance for hip BMD in Chinese populations. This study provides the first strong evidence for the high degree of genetic determination of BMD variation in the Chinese. These results may provide a challenge, and justification, as well as an opportunity, for geneticists to pursue gene identification for osteoporosis in the Chinese.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The study was partially supported by Hunan Province Special Professor Start-up Fund (25000612), Chinese National Science Foundation (CNSF) Outstanding Young Scientist Award (30025025), CNSF Grant (30170504), a key project grant from CNSF, a young investigator award from Huo Ying Dong Education Foundation, and a Seed Fund and a key project fund from the Ministry of Education of P. R. China (25000106). Some investigators were partially supported by grants from Health Future Foundation of the USA, grants from the National Institute of Health (K01 AR02170-01, R01 GM60402-01A1), grants from the State of Nebraska Cancer and Smoking Related Disease Research Program and the State of Nebraska Tobacco Settlement Fund, and US department of Energy Grant (DE-FG03-00ER63000/A00). Ms. Q. R. Huang coordinated sample recruitment for this study. We thank all the study subjects for volunteering to participate in the study. We thank F. H. XU, Y. Y. ZHANG and H. SHEN for their helpful discussions and comments on this paper, and thank Ms. M. HANSEN for careful reading of the manuscript. Two anonymous reviewers are acknowledged for their criticisms and comments, which have been very helpful in clarifying ambiguities and improving the overall presentation of the article.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • Akaike, H. 1974 A new look at the statistical model identification. IEEE Trans Automatic Control AC 19, 716723.
  • Bachrach, L. K., Hastie, T., Wang, M. C., Narasimhan, B. & Marcus, R. 1999 Bone mineral acquisition in healthy Asian, Hispanic, black, and Caucasian youth: a longitudinal study. J Clin Endocrinol Metab 84, 47024712.
  • Bonney, G. E. 1984 On the statistical determination of major gene mechanism in continuous human traits: Regressive models. Am J Med Genet 18, 731749.
  • Boyanov, M. 2001 Diagnostic discrepancies between two closely related forearm bone density measurement sites. J Clin Densitom 4, 6371.
  • Cardon, L. R., Garner, C., Bennett, S. T., Mackay, I. J., Edwards, R. M., Cornish, J., Hegde, M., Murray, M. A., Reid, I. R. & Cundy, T. 2000 Evidence for a major gene for bone mineral density in idiopathic osteoporotic families. J Bone Miner Res 15, 11321137.
  • Consensus Development Conference. 1993 Bone mass and fracture risk. Am J Med 95 (5A), 1S78S.
  • Cooper, G. S. 1999 Genetic studies of osteoporosis: What have we learned J Bone Miner Res 14, 16461648.
  • Cummings, S. R., Kelsey, J. L., Nevitt, M. C. & Dowd, K. J. 1985 Epidemiology of osteoporosis and osteoporotic fractures. Epidemiol Rev 7, 178208.
  • Cummings, S. R. & Melton, L. J. 2002 Epidemiology and outcomes of osteoporotic fractures. Lancet 359, 17611767.
  • Davis, J. W., Novotny, R., Ross, P. D. & Wasnich, R. D. 1994 The peak bone mass of Hawaiian, Filipino, Japanese, and white women living in Hawaii. Calcif Tissue Int 55, 249252.
  • Davis, J. W., Novotny, R., Wasnich, R. D. & Ross, P. D. 1999 Ethnic, anthropometric, and lifestyle associations with regional variations in peak bone mass. Calcif Tissue Int 65, 100105.
  • Deng, H. W., Li, J., Li, J. L., Johnson, M., Davies, M. & Recker, R. R. 1998 Change of bone mass in postmenopausal Caucasian women with & without hormone replacement therapy is associated with Vitamin D receptor & estrogen receptor genotypes. Hum Genet 103, 576585.
  • Deng, H. W., Stegman, M. R., Davies, M., Conway, T. & Recker, R. R. 1999a Genetic determination of peak bone mass (PBM) at hip and spine and common familiar environmental effects on bone qualities. J Clin Densitom 2, 251263.
  • Deng, H. W., Li, J., Li, J. L., Johnson, M., Gong, G. & Recker, R. R. 1999b Association of VDR and estrogen receptor genotypes with bone mass in postmenopausal Caucasian women: different conclusions with different analyses and the implications. Osteoporos Int 9, 499507.
  • Deng, H. W., Chen, W. M., Recker, S., Stegman, M. R., Li, J. L., Davies, K. M., Zhou, Y., Deng, H. Y., Heaney, R. R. & Recker, R. R. 2000 Genetic determination of Colles' fractures and differential bone mass in women with and without Colles' fractures. J Bone Miner Res 15, 12431252.
  • Deng, H. W., Xu, F. H., Conway, T., Deng, X. T., Li, J. L., Davies, K. M., Deng, H. Y., Johnson, M. & Recker, R. R. 2001 Is population BMD variation linked the marker D11S987 on chromosome 11q12–13. J Clinic Endocrinol Metab 86, 37353741.
  • Deng, H. W., Levshits, G., Yakovenko, K., Xu, F. H., Conway, T., Davies, K. M., Deng, H. & Recker, R. R. 2002a Evidence for a major gene for bone mineral density/content in human pedigrees idenfied via probands with extreme bone mineral density. Ann Hum Genet 66, 6174.
  • Deng, H. W., Shen, H., Xu, F. H., Deng, H. Y., Conway, T., Zhang, H. T. & Recker, R. R. 2002b Tests of linkage and/or association of genes for vitamin D receptor, osteocalcin, and parathyroid hormone with bone mineral density. J Bone Miner Res 17, 678686.
  • Deng, H. W., Xu, F. H., Huang, Q. Y., Shen, H., Deng, H., Conway, T., Liu, Y. J., Liu, Y. Z., Li, J. L., Zhang, H. T., Davies, K. M. & Recker R. R. 2002c A whole-genome linkage scan suggests several genomic regions potentially containing quantitative trait loci for osteoporosis. J Clin Endocrinol Metab 87, 51515159.
  • Dequeker, J., Nijs, J., Verstraeten, A., Geusens, P. & Gevers, G. 1987 Genetic determinants of bone mineral content at the spine and radius: A twin study. Bone8, 207209.
  • DeStefano, A. L., Cupples, L. A., Maciel, P., Gaspar, C., Radvany, J., Dawson, D. M., Sudarsky, L., Corwin, L., Coutinho, P. & MacLeod, P. 1996 A familial factor independent of CAG repeat length influences age at onset of Machado-Joseph disease. Am J Hum Genet 59, 119127.
  • Devoto, M., Shimoya, K., Caminis, J., Ott, J., Tenenhouse, A., Whyte, M. P., Sereda, L., Hall, S., Considine, E., Williams, C. J., Tromp, G., Kuivaniemi, H., Ala-Kokko, L., Prockop, D. J. & Spotila, L. D. 1998 First–stage autosomal genome screen in extended pedigrees suggests genes predisposing to low bone mineral density on chromosomes 1p, 2p and 4q. Eur J Hum Genet 6, 151157.
  • Duncan, E. L., Brown, M. A., Sinsheimer, J., Bell, J., Carr, A. J., Wordsworth, B. P. & Wass, J. A. 1999 Suggestive linkage of the parathyroid receptor type 1 to osteoporosis. J Bone Miner Res 14, 19931999.
  • Econs, M. J. & Speer, M. C. 1996 Genetic studies of complex diseases: Let the reader beware. J Bone Miner Res 11, 18351840.
  • Eisman, J. A. 1995 Vitamin D receptor gene alleles and osteoporosis: An affirmative view. J Bone Miner Res 10, 12891293.
  • Go, R. C. P., Elston, R. C. & Kaplan, E. B. 1978 Efficiency and robustness of pedigree segregation analysis. Am J Hum Genet 30, 2837.
  • Gong, G. D., Sterns, H. S., Cheng, S. C., Fong, N., Mordeson, J., Deng, H. W. & Recker, R. R. 1999 On the association of bone mass density and Vitamin–D receptor genotype polymorphisms. Osteoporos Int 9, 5564.
  • Gueguen, R., Jouanny, P., Guillemin, F., Kuntz, C., Pourel, J. & Siest, G. 1995 Segregation analysis and variance components analysis in bone mineral density in healthy families. J Bone Miner Res 10, 20172022.
  • Jarvik, C. P. 1998 Complex segregation analyses: use and limitations. Am J Hum Genet 63, 94246.
  • Johnson, M. L., Gong, G. D., Kimberling, W., Recker, S. M., Kimmel, D. B. & Recker, R. R. 1997 Linkage of a gene causing high bone mass to human chromosome 11 (11q12–13). Am J Hum Genet 60, 13261332.
  • Jones, G. & Nguyen, T. V. 2000 Associations between maternal peak bone mass and bone mass in prepubertal male and female children. J Bone Miner Res 15, 19982004.
  • Karasik, D., Ginsburg, E., Livshits, G., Pavlovsky, O. & Kobyliansky, E. 2000 Evidence of major gene control of cortical bone loss in humans. Genet Epidemiol 19, 410421.
  • Koller, D. L., Rodriguez, L. A., Christian, J. C., Slemenda, C. W., Econs, M. J., Hui, S. L., Morin, P., Conneally, P. M., Joslyn, G., Curran, M. E., Peacock, M., Johnston, C. C. & Foroud, T. 1998 Linkage of a QTL contributing to normal variation in bone mineral density to chromosome 11q12–13. J Bone Miner Res 13, 19031908.
  • Koller, D. L., Econs, M. J., Rodriguez, L. A., Christian, J. C., Hui, S. L., Morin, P., Conneally, P. M., Joslyn, G., Peacock, M. & Johnston, C. C. 2000 Genome screen for QTLs contributing to normal variation in bone mineral density and osteoporosis. J Clin Endocrinol Metab 85, 31163120.
  • Liu, G., Peacock, M., Eilam, O., Dorulla, G., Braunstein, E. & Johnston, C. C. 1997 Effect of osteoarthritis in the lumbar spine and hip on bone mineral density and diagnosis of osteoporosis in elderly men and women. Osteoporos Int 7, 564569
  • Liu, Z. H., Piao, J. H., Pang, L. P., Qing, X., Nan, S., Pan, Z. A., Guo, Y. C., Wang, X. H., Li, F. F., Liu, J. P. & Cheng, X. G. 2002 The diagnosis criteria for primary osteoporosis and the incidence of osteoporosis in China. J Bone Miner Metab 20, 181190.
  • Livshits, G., Pavlovsky, O. & Kobyliansky, E. 1996 Population biology of human aging: segregation analysis of bone age characteristics. Hum Biol 68, 540554.
  • Livshits, G., Karasik, D., Otremski, I. & Kobyliansky, E. 1998 Genes play an important role in bone aging. Am J Hum Biol 10, 421438.
  • Livshits, G., Karasik, D., Pavlovsky, O. & Kobyliansky, E. 1999 Segregation analysis reveals a major gene effect in compact and cancellous bone mineral density in two populations. Hum Biol 71, 155172.
  • 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, 152161.
  • McKay, H. A., Bailey, D. A., Wilkinson, A. A. Houston, C. S. 1994 Familial comparison of bone density at the proximal femur and spine. Bone Miner 24, 95107.
  • Melton, L. J., Kan, S. H., Frye, M. A., Wahner, H. W., O'Fallon, W. M. & Riggs, B. L. 1989 Epidemiology of vertebral fractures in women. Am J Epidemiol 129, 10001011.
  • Morrison, N. A., Qi, J. C., Tokita, A., Kelly, P. J., Crofts, L., Nguyen, T. V., Sambrook, P. N. & Eisman, J. A. 1994 Prediction of bone density from vitamin D receptor alleles. Nature 367, 284287.
  • Niu, T., Chen, C., Cordel, H., Yang, J., Wang, B., Wang, Z., Fang, Z., Schork, N. J., Rosen, C. J. & Xu, X. 1999 A genome–wide scan for loci linked to forearm bone mineral density. Hum Genet 104, 226233.
  • Peacock, M. 1995 Vitamin D receptor gene alleles and osteoporosis: A contrasting view. J Bone Miner Res 10, 12941297.
  • Plato, C. C., Fox, K. M. & Tobin, J. D. 1994 Skeletal changes in human aging. In: Biological anthropology and aging (eds. Crews, D. E. & Garruto, R. M.), pp. 272300. New York : Oxford University Press.
  • Pocock, N. A., Eisman, J. A., Hopper, J. L., Yeates, M. G., Sambrook, P. N. & Eberl, S. 1987 Genetic determinants of bone mass in adults. J Clin Invest 80, 706710.
  • Pollitzer, W. S. & Anderson, J. J. B. 1989 Ethnic and genetic differences in bone mass: a review with a hereditary vs environmental perspective. Am J Clin Nutr 50, 12441259.
  • Recker, R. R. & Deng, H. W. 2002 Role of genetics in osteoporosis. Endocrine 17, 112.
  • Rice, T., Despr'es, J. P., Daw, E. W., Gagnon, J., Borecki, I. B., P'erusse, L., Leon, A. S., Skinner, J. S., Wilmore, J. H., Rao, D. C. & Bouchard, C. 1997 Familial resemblance for abdominal visceral fat: the HERITAGE family study. Int J Obes Relat Metab Disord 21, 10241031.
  • Ryan, P. J. 2001 Bone densitometry in the management of Colles' fractures: which site to measure Br J Radiol 74, 11371141.
  • S.A.G.E. 2001 Statistical analysis for genetic epidemiology, Release 4.1. Computer program package available from the Department of Epidemiology and Biostatistics, Rammelkamp Center for Education and Research, MetroHealth Campus, Case Western Reserve University , Cleveland , OH .
  • Schwartz, J. B. & Rackson, M. 2001 Diffuse idiopathic skeletal hyperostosis causes artificially elevated lumbar bone mineral density measured by dual X-ray absorptiometry. J Clin Densitom 4, 385388.
  • Slemeda, S. W., Christian, J. C. C., Williams, C. J., Norton, J. A. & Johnston, Jr. C. C. 1991 Genetic determinants of bone mass in adult women: A reevaluation of the twin model and the potential importance of gene interaction on heritability estimates. J Bone Miner Res 6, 561567.
  • Slosman, D. O., Rizzoli, R., Donath, A. & Bonjour, J. P. 1990 Vertebral bone mineral density measured laterally by dual-energy X-ray absorptiometry. Osteoporos Int 1, 2329.
  • Sowers, M. R., Boehnke, M., Jannausch, M. L., Crutchfield, M., Corton, G. & Burns, T. L. 1992 Familiality and partitioning the variability of femoral bone mineral density in women of child–bearing age. Calcif Tissue Int 50, 110114.
  • Wang, M. C., Aguirre, M., Bhudhikanok, G. S., Kendall, C. G., Kirsch, S., Marcus, R. & Bachrach, L. K. 1997 Bone mass and hip axis length in healthy Asian, black, Hispanic, and white American youths. J Bone Miner Res 12, 19221935.
  • Uitterlinden, A. G. 2001 Genetics and genomics of osteoporosis review of the European Calcified Tissue Society.
  • Von Der Recke, P., Hansen, M. A., Overgaard, K. & Christiansen, C. 1996 The impact of degenerative conditions in the spine on bone mineral density and fracture risk prediction. Osteoporos Int 6, 4349.