SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

This study was designed to assess the relative contributions of genetic and environmental factors to the variation and covariation of quantitative ultrasound (QUS) measurements and their relationships to bone mineral density (BMD). Forty-nine monozygotic (MZ) and 44 dizygotic (DZ) female twins between 20 and 83 years of age (53 ± 13 years, mean ± SD) were studied. Digital (phalangeal) QUS (speed of sound [SOS]) and calcaneal QUS (broadband ultrasound attenuation [BUA] and velocity of sound [VOS]) were measured using a DBM Sonic 1200 ultrasound densitometer and a CUBA ultrasound densitometer, respectively. Femoral neck (FN), lumbar spine (LS), and total body (TB) BMD were measured using dual-energy X-ray absorptiometry. Familial resemblance and hence heritability (proportion of variance of a trait attributable to genetic factors) were assessed by analysis of variance, univariate, and multivariate model-fitting genetic analyses. In both QUS and BMD parameters, MZ twins were more alike than DZ pairs. Estimates of heritability for age- and weight-adjusted BUA, VOS, and SOS were 0.74, 0.55, and 0.82, respectively. Corresponding indices of heritability for LS, FN, and TB BMD were 0.79, 0.77, and 0.82, respectively. In cross-sectional analysis, both BUA and SOS, but not VOS, were independently associated with BMD measurements. However, analysis based on intrapair differences suggested that only BUA was related to BMD. Bivariate genetic analysis indicated that the genetic correlations between BUA and BMD ranged between 0.43 and 0.51 (p < 0.001), whereas the environmental correlations ranged between 0.20 and 0.28 (p < 0.01). While the genetic correlations within QUS and BMD measurements were significant, factor analysis indicates that common genes affect BMD at different sites. Also, individual QUS measurements appear to be influenced by some common sets of genes rather than by environmental factors. Significant environmental correlations were only found for BMD measurements and ranged between 0.50 and 0.65 (p < 0.001). These data suggest that QUS and BMD measurements are highly heritable traits. While it appears that there is a common set of genes influencing both QUS and BMD measurements, specific genes yet to be identified appear to have greater effects than that of shared genes in each trait.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

OSTEOPOROSIS IS A multifactorial disease with both genetic and environmental determinants, characterized by inadequate amounts and deterioration of mechanical structure of bone. Much of the research on the pathophysiology of osteoporosis has concentrated on bone mineral density (BMD) as one of the principal determinants of osteoporotic fracture risk.1–4 However, BMD (a measure of bone strength) alone does not accurately identify all subjects prone to fracture, and recent attention has addressed the role of bone quality (or structure). Quantitative ultrasound measurements (QUS), including broadband ultrasound attenuation (BUA) and speed of sound (SOS), have been proposed as a measure of the component of bone structure. This is based on the concept that sound waves travel with greater speed and less attenuation through more dense and elastic bone.5 Indeed, cross-sectional studies have shown that subjects with fractures either at the hip or lumbar spine (LS) had ∼15–30% (or 0.5 SD) lower BUA than nonfracture subjects.6–8 Furthermore, longitudinal epidemiological studies have also shown that subjects with lower BUA at baseline tend to have a higher risk of subsequent hip and vertebral fractures.9–11 In these studies, each SD lower of BUA was associated with a doubling of hip fracture risk (95% confidence intervals [CI] 1.3–4.3) independent of BMD measurement. Another study found that BUA in combination with BMD provided essentially equivalent risk information to BMD alone.6 Nevertheless, if the effects of QUS on fracture risk are indeed independent of BMD, then a combination of QUS and BMD measurements could improve fracture prediction. Identification of factors that determine QUS and, for that matter, BMD is therefore a priority for preventive strategies in osteoporosis.

There is considerable evidence that genetic factors play an important role in the determination of bone mass throughout life.12–17 Twin studies have estimated that up to 80% of intersubject variance in bone density is attributable to genetic factors. Because BMD is moderately associated with QUS, with correlations between calcaneal QUS and BMD of 0.62–0.90,18–20 it is possible that QUS is also heritable. Indeed, a recent study estimated that genetic factors were responsible for 30–40% of the variation in QUS measurements.13 However, it is unclear whether common genetic factors affect bone parameters measured by both BMD and QUS.

Understanding common and distinct genes that affect these bone parameters could help in our understanding of the precise bone characteristics assessed by these two modalities. This study was undertaken to explore mechanisms by which genetic and environmental factors may influence the covariation between QUS and BMD. The aim was to examine the influence of genetic and environmental factors on the variation in QUS measurements at the calcaneus and the phalanges and the extent to which such factors also influence BMD.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Subjects

Ninety-three female adult twin pairs were recruited from media campaigns in the Sydney Metropolitan region and through the Australian National Health and Medical Research Council Twin Registry. The study group was comprised of 49 monozygotic (MZ) and 44 dizygotic (DZ) female twin pairs; of these, 36 MZ (73%) and 30 DZ (68%) pairs were postmenopausal. One MZ twin pair and one DZ twin pair discordant for menopausal status were excluded from further analyses.

Subjects were interviewed by a registered nurse or a doctor using a structured medical questionnaire, which provides information on anthropometry, medical history, lifestyle factors, and dominant handedness (the side used for the majority of tasks). Twins who used medications or had medical conditions that could interfere with bone metabolism were excluded from the analysis. Zygosity was determined from the twin's self report, previously found to be accurate to within 5% and comparable with classification by more extensive investigation.21 The study was approved by St. Vincent's Research and Ethics Committee, and all patients gave informed consent.

Measurements of QUS and BMD

Bone densitometry was performed at the LS, femoral neck (FN), and total body (TB) in grams per square centimeter with dual-energy X-ray absorptiometry (DPX-L, Lunar Corp., Madison, WI, U.S.A.). The coefficient of reliability of BMD measurements at our institution is 0.98, 0.95, and 0.96 at the LS, FN, and TB, respectively.22 QUS parameters were measured on the nondominant calcaneus, as previously described,23 using a CUBA ultrasound densitometer (McCue Ultrasonics, London, U.K.), which measures BUA (dB/MHz) and velocity of sound (VOS, m/s). Repeated measurements on 10 different days in 10 subjects indicated that the coefficient of variation was 5.1% and 2.0% for BUA and VOS, respectively.23 Digital ultrasound measurement was measured on the distal metaphysis of the proximal phalanx of the medial four fingers of the nondominant hand using a DBM Sonic 1200 ultrasound densitometer (Igea Ultrasonics, Carpi, Italy). Reproducibility of digital ultrasound SOS based on eight subjects having three different measurements was 1.8%.

Analysis of association between QUS and BMD measurements

To assess the association between QUS and BMD measurements, linear regression analysis was performed between individuals and matched pairs. In the unmatched analysis, each twin within a pair was treated as an individual. Each BMD measurement of the twin was then modeled as a linear function of their BUA, SOS, and VOS measurements in a multiple regression analysis. In the matched pair analysis, intrapair differences in BMD and BUA, SOS and VOS (denoted by dBMD, dBUA, dSOS, and dVOS, respectively) were obtained by subtracting the value of one twin from that of the other. The intrapair differences in each BMD measurement were then expressed as a linear function of intrapair differences in each QUS measurement, i.e., dBMD = α(dBUA) + β(dSOS) + γ(dVOS) + ϵ, where α, β. and γ are regression coefficients associated with dBUA, dSOS, and dVOS, respectively, and ϵ is the residual error term. The iteratively reweighted least squares method24 was used to estimate the model parameters. Because the twins are not independent, the error terms of estimated regression parameters, although unbiased, tend to be correlated within pairs leading to underestimation of standard errors and hence overestimation of statistical significance. To avoid this problem, the generalized least square method24 with iterative adjustment for the correlation of errors within pairs was used. In each analysis, backward and stepwise algorithms were used to search for the most parsimonious model. Assessment of model adequacy was based on residual analysis.

Estimation of familial resemblance

In twin analysis, the extent to which MZ twins are more alike than DZ twins is taken to reflect genetic influences. This was assessed in MZ and DZ pairs separately by the intraclass correlation coefficient. One-way analysis of variance was used to partition the total variation of each variable trait into two parts, namely between-pair (B) and within-pair (W) variations. It can be shown that the intraclass correlation coefficient is given by the difference between B and W over their sum, i.e., (B – W)/(B + W).25

Univariate genetic analysis

To assess the relative contribution of genetic and environmental factors to the determination of QUS, the data were analyzed according to the classical twin model26 by partitioning the total phenotypic variance into genetic and environmental components. The genetic variance may be due to additive (A) or dominant (D) genetic influences. The environmental variance may be due to common (C) environmental factors shared by twins reared and living in the same environment or to nonshared (E) environmental factors (Fig. 1). Shared environmental effects and dominant genetic effects cannot be distinguished in the classical twin analysis. Additive genetic factors are the effects of genes taken singly and added over multiple loci, whereas dominant genetic factors represent genetic interaction within loci. The classical twin model assumes that additive genetic factors and dominant genetic factors are perfectly correlated in MZ pairs, while DZ pairs, like ordinary siblings, share only half of the additive genetic effects and one quarter of the dominant genetic effects. The model also assumes negligible effects of assortive mating, epistasis, genotype-environmental interaction, and/or correlation and that shared environmental influences are similar for MZ and DZ twins.FIG. 2

thumbnail image

Figure FIG. 1. The classical twin model. Latent variables are circled; observed variables are depicted in squares. A, additive genetic factors; D, dominant genetic factors; C, shared environmental factors; and E, nonshared environmental factors (including measurement error). The correlation between A1 and A2 is 1 for MZ pairs and 0.5 for DZ pairs; between D1 and D2 it is 1 for MZ and 0.25 for DZ pairs. The correlation between shared environmental factors (C1 and C2) is assumed to be unity for both zygosities.

Download figure to PowerPoint

thumbnail image

Figure FIG. 2. Cholesky model for the multivariables. Path diagrams depict the common and unique factors for genetic and environmental sources of variance and covariance for BUA, VOS, and BMD at the LS, FN, and TB. There are six genetic factors (G1, G2, G3, G4, G5, and G6) and six nonshared environmental factors (E1, E2, E3, E4, E5, and E6). The paths from G1 to BUA, VOS, LS BMD, FN BMD, and TB BMD are denoted by ga1, gb1, gc1, gd1, ge1, and gf1, respectively. The paths from G2 to VOS, DUS, LS BMD, FN BMD, and TB BMD are denoted by gb2, gc2, ge2, and gf2, respectively. The paths from G3 to DUS, LS BMD, FN BMD, and TB BMD are denoted by gc3, gd3, ge3, and gf2, respectively. The paths from G4 to LS BMD, FN BMD and TB BMD are denoted by gd4, ge4, and gf4, respectively. The paths from G5 to FN BMD and TB BMD are denoted by ge5 and gf5. The paths of environmental factors, ea1, eb1, ec1, ee6, are denoted similarly. The figure illustrates only one twin in a pair.

Download figure to PowerPoint

To estimate variance due to A, C, and E, the observed variance-covariance matrices between twin pairs were compared across four different models. Model I incorporates the effects of A, C, and E; model II, A and E; model III only C and E; and model IV only E. The most parsimonious model was selected based on nonsignificant χ2 goodness-of-fit statistic and the minimum value of the Akaike Information Criterion, which is equal to χ2 – 2df (df = degrees of freedom). The Chi-square measure is distributed asymptotically as a Chi-square distribution. The 95% confidence interval of each variance component was constructed using a method described by Neale et al.27 The index of heritability was computed as the ratio of genetic variance over total phenotypic variance from the most parsimonious model.

Multivariate genetic analysis

The above univariate analysis allows the assessment of the contribution of genetic and environmental factors to a single variable trait such as BUA, SOS, or BMD taken separately. However, this study also aimed to address whether genetic or environmental factors that affect one variable trait are the same as those that affect another, or indeed, a correlated trait. If correlations between QUS and BMD were attributable to common genetic factors, then it would be expected that the correlation between QUS of twin 1 and BMD of twin 2 and between QUS of twin 2 and BMD of twin 1 would be higher in MZ than in DZ pairs. This statistic is referred to as “cross-trait” correlation. However, this analysis is only exploratory and does not allow testing of hypotheses about the structure of genetic and environmental effects on QUS and BMD traits. Thus, in a second analysis, three multivariate genetic models, namely Cholesky decomposition, the common pathway model, and the independent pathway model,28 were considered. Preliminary analyses indicated that only the Cholesky model provided adequate fit, therefore the observed 12 × 12 variance-covariance matrices (Cov) between twins and variables of BUA, VOS, SOS, and BMD measurements were decomposed into different components, such as Cov = A + C + E or Cov = A + E, where A, C, and E are matrices of estimated parameters with elements aij, cij, and eij (i,j = 1, 2, 3, …, 6), respectively. Thus, the diagonal elements of A, for example, represent the genetic factors that are specific to the variable traits, while off-diagonal elements represent shared genetic factors. The extent to which two variables, i and j, share genetic effects can be measured by the genetic correlation

  • equation image

where gij is the genetic covariance between variables i and j, and gii and gjj are genetic variances of the variables i and j, respectively. Similarly, the environmental correlation between the two traits is obtained by

  • equation image

where eij is the environmental covariance between variables i and j, and eii and ejj are environmental variances of the variables i and j, respectively. Preliminary univariate analyses suggested that a model with additive genetic (A) and nonshared environmental (E) factors fitted the data adequately. Therefore, a Cholesky model of decomposition including these parameters was fitted to the variance-covariance matrices.28

However, while the Cholesky decomposition analysis allows tests for genetic and environmental correlations, it does not identify the number and nature of underlying factors that are responsible for the observed correlations. Factor analysis29 was thus performed to test hypotheses about the phenotypic, genetic, and environmental correlation matrices estimated from the Cholesky decomposition analysis. This analysis seeks to explain the pattern of correlations between these variables in terms of the linear additive effects on those variables of a smaller number of latent variables (or “factors”). Thus, the analysis allows for the reduction of the original six variables into a reduced multivariate space while maximizing the variation explained on the original data. In essence, such an analysis entails the finding of linear functions of the original data so that each new synthetic variable is orthogonal (uncorrelated) to the others. This is done by extracting the eigenvalues of a covariance matrix whose elements have been standardized to have means of 0 and variances of 1, in other words, a correlation matrix. The eigenvalues represent the variance explained by each of the principal components.

All genetic model parameters were estimated by using the maximum likelihood method via the Mx program.30 Principal component and multiple regression analyses were performed with the SAS Statistical Analysis System (SAS Institute, Cary, NC, U.S.A.).31

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Characteristics of study sample

Twins of the two zygosities were comparable with respect to anthropological characteristics, with mean ± SD age of 53 ± 13 years (range 20–83 years), weight 66 ± 11 kg, and body mass index 25.3 ± 4.2 kg/m2. Of the individuals, 132 twins (or 71%) were postmenopausal. Among the postmenopausal twins, there was no significant difference between MZ and DZ twins in years postmenopause (11.4 ± 9.1 vs. 9.3 ± 8.6 years, respectively, p = 0.2). Also, variances for BMD and QUS parameters in MZ or DZ pairs were not significantly different.

Association between QUS and BMD measurements

In cross-sectional analysis, lower weight and advancing age were associated with lower BMD at all sites. The combination of age and weight accounted for 27% of total variance of LS and 28% for FN BMD and 46% for TB BMD. Age was also negatively related to BUA (r = −0.45; p < 0.0001) and SOS (r = −0.62; p < 0.0001), but not with VOS measurements (r = −0.12; p = 0.11). However, greater weight was associated with higher BUA (r = 0.24; p = 0.001), but lower SOS (r = −0.16; p = 0.04) and VOS (r = −0.14; p = 0.05).

In interindividual analysis, both BUA and SOS were positively associated with all BMD measurements (Table 1). These two variables accounted for about 30% (LS), 35% (FN), and 37% (TB) of variation of BMD. However, in intratwin analysis, BUA was the only significant predictor of BMD. The proportion of variance of intrapair differences in BMD attributable to intrapair differences in BUA was 15% at the FN, 28% at the LS, and 32% at the TB BMD.

Table TABLE 1. ASSOCIATION BETWEEN BMD AND QUS MEASUREMENTS: STANDARDIZED REGRESSION COEFFICIENT RELATING BMD TO BUA AND SOS
Thumbnail image of

Univariate genetic analysis

For all QUS measurements, intraclass correlations for MZ pairs were consistently greater than for DZ pairs, although the magnitude of the effect varied (diagonals in Table 2). Univariate genetic model-fitting analyses indicated that for all QUS and BMD parameters, the models incorporating the additive genetic factors and random environmental factors with or without shared environmental factors (AE or ACE models) fitted the data equally adequately. Both of these models fitted the data better than models lacking additional genetic effects (CE or E models). Estimates of heritability and variance components (and their 95% confidence intervals) were based on the more parsimonious AE model (Table 3). In unadjusted analysis, the indices of heritability for BUA, VOS, and SOS were 0.68, 0.59, and 0.74, respectively. After adjusting for age and weight, these indices changed minimally to 0.74, 0.55, and 0.82 for BUA, VOS, and SOS, respectively. The indices for heritability of the BMD measurements between 0.70 and 0.85 were minimally effected by adjustment for age and weight.

Table TABLE 2. CROSS-TRAIT AND INTRACLASS CORRELATIONS OF AGE-ADJUSTED QUS AND BMD MEASUREMENTS
Thumbnail image of
Table TABLE 3. GENETIC AND ENVIRONMENTAL CONTRIBUTIONS TO VARIANCES OF QUS AND BMD MEASUREMENTS
Thumbnail image of

Multivariate genetic analyses

QUS and BMD measurements were significantly correlated: BUA (r = 0.45, 0.46, and 0.53 with LS, FN, and TB BMD, respectively), followed by SOS and BMD (r = 0.34, 0.36, and 0.38) and VOS and BMD (r = 0.18, 0.20, and 0.16). The cross-trait correlations between QUS and BMD measurements were higher in MZ than in DZ pairs (Table 2). However, the cross-trait correlations were somewhat greater within sets of QUS or BMD measurements than between QUS and BMD measurements. Thus, the cross-trait correlation between BUA and VOS was 0.24 (p< 0.05) for MZ twins and 0.04 for DZ twins, and between LS BMD versus FN BMD and TB BMD it was 0.50 (p < 0.001) in MZ pairs compared with 0.29–0.36 (p < 0.01) in DZ pairs. By contrast, between BUA and LS BMD, the correlation was 0.17 in MZ and 0.14 in DZ and between BUA and FN BMD it was 0.37 (p < 0.01) in MZ and 0.20 (p < 0.05) in DZ twin pairs.

Because the twin model with additive genetic (A) and random environmental (E) factors fitted the data adequately and was the most parsimonious, a Cholesky model of decomposition including these parameters was fitted to the variance-covariance matrices to estimate genetic and environmental correlations (Table 4). Within QUS measurements, the genetic correlations ranged between 0.36 and 0.44 (p < 0.001), which were greater than the environmental correlations, which were not significant, between 0.14 and 0.16 (p > 0.08). Within BMD measurements, both genetic and environmental correlations were significant, with genetic correlations of 0.51–0.74 (p< 0.001) and environmental correlations of 0.50–0.65 (p< 0.001). The genetic correlations between QUS and BMD measurements were moderate (0.32–0.59, p < 0.01) and consistently greater than the nonsignificant environmental correlations (0.03 [p > 0.46] to 0.28 [p < 0.05]). Variance-covariance matrices indicated that 35, 32, and 21% of genetic variance of BUA, VOS, and SOS was attributable to genetic factors that are specific to FN BMD (data not shown).

Table TABLE 4. GENETIC AND ENVIRONMENTAL CORRELATIONS
Thumbnail image of

To gain a better understanding of the underlying patterns of correlations, factor analysis was performed on phenotypic, genetic, and environmental correlations. Out of six possible factors, the first four explain ∼87, 91, and 86% of the observed phenotypic, genetic, and environmental correlations (Table 5). For the phenotypic and genetic correlation analysis, the first factor contributes to all BMD measurements, while factors II, III, and IV influence VOS, SOS, and BUA, respectively. The first factor of the environmental correlation analysis also predominantly affects BMD measurements. However, the other three factors contribute to combinations of QUS measurements rather than individual measurements, i.e., the second factor loads on VOS and SOS measurements, the third factor loads on VOS and SOS, and the fourth factor loads on BUA and SOS.

Table TABLE 5. FACTOR LOADINGS ON PHENOTYPIC, GENETIC, AND ENVIRONMENTAL CORRELATIONS
Thumbnail image of

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

It has generally been assumed that many genetic and environmental factors act in combination to determine an individual's risk of osteoporosis, including bone density. Recent attention has focused on the role of bone structure and quality in the assessment of osteoporosis. QUS measurements, postulated to measure a component of bone quality, may provide additional information about osteoporotic fracture risk beyond that obtained from bone density.32–34 However, the determinants of QUS and its relationship to BMD remain largely unclear. In the present study, a large part of variance of QUS (as much as 80%) was found to be attributable to genetic factors, making it an equally heritable trait as BMD. Furthermore, some genetic factors appear to be common to both QUS and BMD.

The role of genetic variation in determining interindividual differences in QUS has not been well documented. Indices of heritability of 53% for BUA and 61% for VOS based on least square analyses13 are comparable and within the sampling variation range of those in the present study estimated by the maximum likelihood method of model-fitting analyses. Adjusting for age and weight did not influence the estimate of the indices of heritability significantly, suggesting that the genetic affects are likely to be direct on bone mass or structure rather than secondary to effects on body size.

These data suggest that the three QUS measurements (BUA, VOS, and SOS) are under the influence of common as well as distinct sets of genetic factors because the genetic correlations between these three measurements were significant, albeit modest. Multivariate genetic and factor analyses indicate that there are likely to be at least three independent sets of genetic factors rather than environmental factors. Given the fact that these parameters measure similar properties related to the mechanical structure of bone, it is not surprising that they share some common genes.

This study confirms the familial influence on bone density with estimates of heritability for the LS, FN, and TB BMD of 79, 77, and 82%, respectively, comparable with previous estimates.12–17 However, the present study also indicates that a common set of genetic and environmental variances underlies the clustering of BMD at various skeletal areas. For example, the genetic correlation between LS and TB BMD was 0.74, compared with the environmental correlation of 0.66. Similarly, the genetic correlations between LS and FN (0.64) was significantly higher than the environmental correlation (0.50). This indicates that genes that affect LS BMD are more likely to affect TB BMD than FN BMD; likewise, environmental factors that affect LS BMD are more likely to exert effects on TB BMD than on FN BMD. Indeed, estimates of heritability from the bivariate twin model suggest that over half of the genetic influence on TB BMD is mediated through genes that influence the LS and less by genes that influence FN BMD.17 Somewhat lower environmental correlations between BMD sites observed in this study may relate, in part, to the differential effects of environmental factors such as smoking and dietary calcium intakes at various skeletal sites. The pattern of genetic and environmental correlations indicate the existence of common genetic (pleiotropy) and common random environmental effects.

The parameters assessed by QUS in the fingers, heel, and BMD in the spine, hip, and TB appear to be largely controlled by separate genetic factors. However, some common genetic factors for BMD and QUS that appear to exist as the genetic correlations between QUS and BMD were significant, albeit modest. It was estimated from these data that as much as 35% of variance of BUA is attributable to genes specific to femoral BMD measurements. This implies that genes in linkage with BMD could, but not necessarily always would, be in linkage with QUS and vice versa. The finding that there are some genetic factors, which influence QUS measurements in addition to those that affect bone density, is consistent with QUS measuring additional nondensity characteristics of bone rather than being a surrogate for BMD.

If QUS and BMD measure similar characteristics of bone, and given the observation that both traits are genetically determined, it would then be expected that the traits also share common genetic factors. However, the modest genetic correlation (between 0.4 and 0.5) between the traits observed here could be due to different genetic influences but also could relate to the fact that QUS was measured in the heels (BUA and VOS) and the fingers (SOS) and would thus be expected to reflect different properties of bone from BMD measured at the hip, spine, and TB. Indeed, a recent study of reproducibility has found that correlation between BUA and BMD both measured at the heel (r = 0.76–0.81) was higher than correlation between BUA measured at the heel and BMD measured at the spine or the hip (r = 0.37–0.57) or between BMD measured at the heel and BMD measured at the hip (r = 0.51). Thus, it appears that differences in measurement sites and parameters measured may both contribute to the divergence of correlation between QUS and BMD.

Although it would be expected that bone parameters measured by BMD and QUS should have some common genetic factors, some of these “genetic” correlations may actually reflect shared family environments. Of particular interest, however, is the apparent hierarchy of common effects on this combination of measurements of bone properties. Clearly, given the observed patterning of significant genetic and environmental correlations among these measured bone parameters, the possibility of pleiotropic interactions should be particularly considered whenever QUS and BMD measures are combined as indices of skeletal fragility.

The present findings must be interpreted in the context of a number of potential limitations. The data were obtained from a Caucasian population in Australia, among whom cultural backgrounds and environmental living conditions are generally homogeneous. Also, the present results were obtained in women only, thus care should be taken when extrapolating these results to other populations and to men. Importantly, these data were obtained from twins, which are arguably not representative of the general unrelated population. However, the variances of BMD and QUS in this twin sample are comparable to those observed in unrelated populations. Furthermore, the finding that the strength of association between QUS and BMD was similar in interindividual and intrapair analyses suggests that the results can be generalized. Also, the use of intrapair differences eliminates confounding age and environmental factors inherent in cross-sectional studies.

It has been argued that MZ twins are likely to share more similar environments than DZ pairs, which could lead to an overestimation of genetic effects. However, in all traits analyzed here, the model incorporating the effects of additive genetic and specific environmental factors (AE model) fitted the data as well as the model which also contained shared environmental factors (ACE model). Thus, shared environmental factors, if contributing to the genetic effect, do so to a relatively small extent.

It is important to emphasize that these findings are partly dependent on the validity of the assumptions in the classical twin model stated earlier. Some important assumptions, such as twins of the two zygosities sharing equal environmental influence and the lack of genotype-environmental interactions, are difficult to verify. Moreover, there was evidence suggesting that genetic influence in BMD may reduce with advancing age,16 and thus the heritability reported here may be an overestimate. Finally, the genetic and environmental correlations are, of course, subject to sampling variation.

In summary, this study demonstrates that bone parameters assessed by calcaneal and digital QUS are under strong genetic influence and share some common genetic factors with those assessed by bone density. Defining the shared and distinct genetic and environmental regulators of parameters assessed as bone density and ultrasound characteristics may provide a better understanding of the factors contributing to the development of osteoporosis.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

This study was supported, in part, by a grant from the National Health and Medical Research Council (NHMRC). We acknowledge the invaluable assistance of Sisters Libby Powell and Joan Birmingham with data collection and coordination of the study. Recruitment of twins was assisted by the NHMRC Twin Registry, to whom we are grateful. We also thank the twins themselves for their enthusiasm and support to osteoporosis research. Finally, we thank Igea Ultrasonics for the use of their digital ultrasound densitometer.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES
  • 1
    Hui SL, Slemenda CW, Johnston CJ 1988 Age and bone mass as predictors of fracture in a prospective study J Clin Invest 81:18041809.
  • 2
    Nguyen T, Sambrook P, Kelly P, Jones G, Lord S, Freund J, Eisman J 1993 Prediction of osteoporotic fractures by postural instability and bone density BMJ 307:11111115.
  • 3
    Melton LJ III, 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 8:12271233.
  • 4
    Cummings SR, Black DM, Nevitt MC, Browner W, Cauley J, Ensrud K, Genant HK, Palermo L, Scott J, Vogt TM 1993 Bone density at various sites for prediction of hip fractures Lancet 341:7275.
  • 5
    Kaufman JJ, Einhorn TA 1993 Perspectives: Ultrasound assessment of bone J Bone Miner Res 8:517525.
  • 6
    Bauer DC, Glüer CC, Cauley JA, Vogt TM, Ensrud KE, Genant HK, Black DM 1997 Bone ultrasound predicts fractures strongly and independently of densitometry in older women: A prospective study Arch Intern Med 157:629634.
  • 7
    Glüer CC, Cummings SR, Bauer DC, Stone K, Pressman A, Mathur A, Genant HK 1996 Osteoporosis: Association of recent fractures with quantitative US findings Radiology 199:725732.
  • 8
    Bauer DC, Glüer CC, Genant HK, Stone K 1995 Quantitative ultrasound and vertebral fracture in postmenopausal women J Bone Miner Res 10:353358.
  • 9
    Ross P, Huang C, Davis J, Imose K, Yates J, Vogel J, Wasnich R 1995 Predicting verterbal deformity using bone densitometry at various skeletal sites and calcaneus ultrasound Bone 16:325332.
  • 10
    Heaney RP, Avioli LV, Chesnut CH III, Lappe J, Recker RR, Brandenburger GH 1995 Ultrasound velocity through bone predicts incident vertebral deformity J Bone Miner Res 10:341345.
  • 11
    Hans D, Dargent-Molina P, Schott AM, Sebert JL, Cormier C, Kotzki PO, Delmas PD, Pouilles JM, Breart G, Meunier PJ 1996 Ultrasonographic heel measurements to predict hip fracture in elderly women: The EPIDOS prospective study Lancet 348:511514.
  • 12
    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 80:706710.
  • 13
    Arden NK, Baker J, Hogg C, Baan K, Spector TD 1996 The heritability of bone mineral density, ultrasound of the calcaneus and hip axis length: A study of postmenopausal twins J Bone Miner Res 11:530534.
  • 14
    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 Bone 8:207209.
  • 15
    Flicker L, Hopper JL, Rodgers L, Kaymakci B, Green RM, Wark J 1995 Bone density determinants in elderly women: A twin study J Bone Miner Res 10:16071613.
  • 16
    Slemenda CW, Christian JC, Williams CJ, Norton JA, Johnston CJ 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.
  • 17
    Nguyen TV, Howard GM, Kelly PJ, Eisman JA 1998 Bone mass, lean mass and fat mass: Same genes or same environmemts Am J Epidemiol 147:316.
  • 18
    Roux C, Fournier B, Laugier P, Chappard C, Kolta S, Dougados M, Berger G 1996 Broadband ultrasound attenuation imaging: A new imaging method in osteoporosis J Bone Miner Res 11:11121118.
  • 19
    McCloskey EV, Murray SA, Miller C, Charlesworth D, Tindale W, O'Doherty DP, Bickerstaff DR, Hamdy NA, Kanis JA 1990 Broadband ultrasound attenuation in the os calcis: Relationship to bone mineral at other skeletal sites Clin Sci 78:227233.
  • 20
    Glüer CC, Vahlensieck M, Faulkner KG, Engelke K, Black D, Genant HK 1992 Site-matched calcaneal measurements of broad-band ultrasound attenuation and single X-ray absorptiometry: Do they measure different skeletal properties? J Bone Miner Res 7:10711079.
  • 21
    Hrubec Z, Robinette C 1984 The study of human twins in medicine and research N Engl J Med 310:320339.
  • 22
    Nguyen TV, Sambrook PN, Eisman JA 1997 Source of variability in bone density: Implication for study design and analysis J Bone Miner Res 12:124135.
  • 23
    Howard GM, Nguyen TV, Pocock NA, Kelly PJ, Eisman JA 1997 Influence of handedness on calcaneal ultrasound: Implication for assessment of osteoporosis and study design Osteoporos Int 7:190194.
  • 24
    Myers RH 1986 Classical and Modern Regression with Applications. Duxbury Press, Boston, MA, U.S.A., pp. 204207.
  • 25
    Donner A, Koval JJ 1980 The estimation of interclass correlation in the analysis of family data Biometrics 36:1925.
  • 26
    Heath A, Neale M, Hewitt J, Eaves L, Fulker D 1989 Testing structural equation models for twin data using LISREL Behav Genet 19:936.
  • 27
    Neale MC, Miller MB 1997 The use of likelihood-based confidence intervals in genetic models Behav Gen 27:113120.
  • 28
    Neale MC, Cardon L 1992 Methodology for Genetic Studies of Twin and Families, Kluwar Academic, Dordrecht, The Netherlands.
  • 29
    Harris RJ 1975 A Primer of Multivariate Statistics. Academic Press, New York, NY, U.S.A., pp. 205220.
  • 30
    Neale MC 1997 Mx: Statistical Modelling, 4th ed, Richmond, VA, U.S.A.
  • 31
    SAS Institute Inc 1990 SAS/SAT User's Guide, release 6.03. SAS Institute Inc., Cary, NC, U.S.A.
  • 32
    Kleerekoper M, Villanueva AR, Stanciu J, Rao DS, Parfitt AM 1985 The role of three-dimensional trabecular micro-structure in the pathogenesis of vertebral compression fractures Calcif Tissue Int 37:594597.
  • 33
    Hayes WC, Piazza SJ, Zysset PK 1991 Biomechanics of fracture risk prediction of the hip and spine by quantitative computed tomography Radiol Clin N Am 29:118.
  • 34
    Recker RR 1989 Low bone mass may not be the only cause of skeletal fragility in osteoporosis Proc Soc Biol Med 191:272274.
  • 35
    Graafmans WC, Lingen A, Ooms ME, Bezemer PD, Lips P 1996 Ultrasound measurements in the calcaneous: Precision and its relation with bone mineral density of the heel, hip and lumbar spine Bone 19:97100.