| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Studies |
Bone and Mineral Research Program, Garvan Institute of Medical Research (J.R.C., T.V.N., J.A.E.), Sydney, New South Wales 2010; and University of Sydney, Royal North Shore Hospital (P.N.S.), St. Leonards, Sydney 2065, Australia
Address all correspondence and requests for reprints to: Dr. Jacqueline Center, Bone and Mineral Research Program, Garvan Institute of Medical Research, St. Vincent’s Hospital, 384 Victoria Street, Sydney, New South Wales 2010, Australia. E-mail: j.center{at}garvan.unsw.edu.au
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Several hormonal and biochemical factors known to affect bone metabolism have been shown to change with age in men in parallel with their age-related decline in bone mineral density (BMD). These parameters have therefore been implicated in the pathogenesis of aging bone loss, although whether they are causal or simply epiphenomena is still not known.
In contrast to women, in whom a sharp decline in estrogen levels at the menopause is a major and remediable cause of bone loss, men do not have a comparable menopause. However, both total and, to a greater extent, free testosterone (T) decline gradually with age in men (5, 6). In younger men, as in women, hypogonadism has been shown to be important both in attaining peak bone mass and in the maintenance of bone mass postpuberty (7, 8), and thus, it seems likely that low levels of T would contribute to osteoporotic fracture risk in the older male. Indeed, up to 50% of men with hip fractures and 20% of men with vertebral fractures (9, 10, 11) have been reported to be hypogonadal. However, it is not clear whether the normal age-related decline in T contributes to the loss of bone mineral or if there is a level of T below which there is an increased risk of osteoporosis. Although several studies have found T to correlate with BMD in elderly men (12, 13), this finding has not been universal (14, 15).
As a result of the report of a man with estrogen insensitivity (16), the importance of amortization of androgens to estrogens in normal bone mineral homeostasis in men has become more clearly established. Indeed, there is recent evidence to suggest that estrogen, rather than T, levels are more closely correlated with BMD in elderly men (17, 18, 19). Elderly men have similar or higher serum levels of estrogen compared with postmenopausal women (18, 19). Even these low levels have been shown to be associated with BMD, bone loss, and fractures in elderly women (17, 20, 21, 22, 23).
GH and insulin-like growth factor (IGF-I) have also been implicated in the pathogenesis of osteoporosis (24, 25, 26), as levels decrease with advancing age in men and women (27). However, this issue remains controversial, as not all studies have demonstrated a relationship between these growth factors and bone mass (28), and in one study there were divergent gender effects (29). The relationship between growth factors and BMD is further complicated because both estrogen and T have been shown to interact with the GH pathway (30, 31). Indeed, it has been proposed that some of the sex hormone effect on bone metabolism may be mediated via IGF-I (32).
Calcitropic hormones may also be involved in the pathogenesis of bone loss. PTH has been shown to increase with age, whereas 25-hydroxyvitamin D (25OHD) decreases (33, 34). Sustained PTH excess is generally reported to result in increased loss of cortical bone in both cross-sectional and longitudinal studies (35, 36). The effect on trabecular bone remains controversial. Thus, in studies of elderly women, secondary hyperparathyroidism, related at least in part to low levels of vitamin D, has been thought to result in increased bone turnover and bone loss (37, 38, 39, 40). Vitamin D insufficiency has been seen as an important factor for bone health in elderly people who do not have significant sunlight exposure, such as those living in institutions or in northerly latitudes, especially during the winter. However, the world-wide prevalence of vitamin D deficiency may be greater than previously recognized, and levels traditionally thought to represent deficiency may be too low (41, 42).
Thus, there is evidence that declining levels of T, estradiol (E2), IGF-I, and 25OHD with a compensatory rise in PTH may contribute to deficits in BMD in the elderly. However, it is not yet clear whether these are simply aging-related changes or physiological changes and what impact, direct or indirect, they may have on BMD, particularly in men for whom there are much fewer data than for women. The aim of this study was to examine the biological interactions among the sex hormones, IGF-I, and calcitropic hormones in their determination of BMD as well as their roles in identification of osteoporosis over and above that of BMD itself in elderly men.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The data used in this analysis were obtained from participants in the Dubbo Osteoporosis Epidemiology Study. Study design and methods have been described in detail previously (1, 43, 44). Briefly, the study was carried out in Dubbo, a semiurban city 400 km northwest of Sydney, Australia, with a population of approximately 32,000, of whom 98.6% are Caucasian. The community is relatively stable with its own centralized health services, making it suitable for epidemiological study. Although open to the whole population, all participants were of Caucasian background. The study commenced in 1989 and recruited men and women over the age of 60 yr. By 1993, approximately 50% of the target population were enrolled into the study. From 1993 on, blood was collected and stored from subjects either at initial or follow-up visits. This study reports on the first 441 men for whom blood was available for analysis. Of these, 2 men were excluded because of chemotherapy that rendered them hypogonadal, and another 2 were excluded because of extremely high PTH levels. The study was approved by the ethics committee of St Vincent’s Hospital (Sydney, Australia).
Clinical data collection
Subjects were interviewed by a nurse coordinator at initial and subsequent visits, which took place at approximately 2-yr intervals. A structured questionnaire was used to collect data, including age and lifestyle factors such as cigarette smoking (pack-years of use) and alcohol use. Anthropometric variables, including weight and current height, were measured at each visit. Quadriceps strength (maximum isometric contraction) was measured in the dominant leg of the seated subject with a horizontal spring gauge calibrated to a maximum 50-kg force. Measurements used in the present study were those taken at the same or closest visit to that of the blood collection.
BMD (grams per cm2) was measured at the femoral neck (FN) and lumbar spine (LS) by dual energy x-ray absorptiometry with a DPX-L densitometer (Lunar Corp., Madison, WI). The measurement was performed at the same visit as or within 1 month of the blood sample collection in 99% of cases. In those cases where the blood was taken at a time other than at the scheduled visit, the BMD used was that which was closest to the time of blood collection. The coefficient of variation for the BMD measurement in normal subjects at our institution is 1.3% at the LS and 3.5% at the FN (45).
Measurement of biochemical parameters
Nonfasting venous samples were separated and stored at -80 C. Assays were performed during 1996 and 1997 for the following parameters: free T, E2, sex hormone-binding globulin (SHBG), PTH, 25-hydroxyvitamin D (25OHD), and insulin-like growth factor I (IGF-I). Free T was measured as a solid phase unextracted analog RIA using [125I]T analog as the labeled analyte with an interassay coefficient of variation (CV) of 9% (DPC Coat-A-Count Free T, Diagnostics Products, Los Angeles, CA). PTH was measured by a solid phase two-site immunoenzymometric assay also from Diagnostics Products, with an interassay CV of 6%. E2 was measured by commercial RIA using a polyethylene glycol- enhanced separation step with an interassay CV of 10% (Sorin Biomedica Diagnostics S.P.A., Saluggia, Italy). The interassay limit of detection for E2 was 15 pmol/L. SHBG was measured by two-site immunoradiometric assay from Orion Diagnostica with interassay CV of 6%. 25OHD was measured using an in-house preextraction competitive binding protein assay with charcoal separation, with an interassay CV of 12%. IGF-I was measured using a commercial preextraction RIA with a polyethylene glycol-enhanced second antibody separation step (Bioclone Australia Pty. Ltd., Australia) with interassay CV of 10%.
Statistical analysis
Statistical analyses were directed at addressing three issues: 1) the association between the hormonal and biochemical parameters and BMD, 2) the interrelationships between these biochemical parameters themselves and together with BMD, and 3) predictive values of these markers in identifying osteoporotic subjects. To address the first issue, BMD was first modeled as a linear function of each biochemical marker using a linear regression analysis. Preliminary analyses suggested that some of the relationships between the biochemical parameters and BMD were nonlinear. These relationships were explored by dividing each biochemical parameter into quartiles. Depending on the best fit between the individual parameter and BMD, they were subsequently modeled as either linear or dichotomous variables (based on quartile or normal reference cut-off points). As correlation between these markers was expected, simultaneous effects of independent markers on BMD were determined using backward and stepwise regression algorithms to search for the most empirical multiple regression model. The P value for inclusion in this final model was initially set at 0.15, but significance was only considered at P < 0.05 level. Model parameters were estimated using the least squares method.
The complex relationships between multiple biochemical and hormonal parameters and bone density can not adequately be explained by a simple linear equation or correlation coefficient. A set of equations is required. Traditionally, studies of association between hormonal parameters and bone density have been limited to multiple regression analysis, in which the covariation (e.g. correlation) between these parameters has been largely ignored. In the present study, based on biological and clinical considerations, a number of theoretical models were proposed to explain the covariation among hormonal parameters and BMD using path analysis (46). In this analysis, BMD was still considered to be the outcome or endogenous variable, whereas hormonal and biochemical factors were the factors or exogenous variables measured with random measurement error. The path model allows a representation, in diagrammatical form, of apparent direction (causality) of relationships between the exogenous and endogenous variables. A potential advantage of path analysis is that it goes beyond measuring the degree of association by the correlation coefficient or regression analysis. Instead, it looks to make explicit hypotheses about the direction and strength (quantified by path coefficients) of relationships between variables. The model (or representation) was then compared with observed data to determine its goodness of fit. Although regression analysis attempts to account for the variation of BMD, path analysis goes one step further by attempting to account for the observed intercorrelational structure between the exogenous variables and BMD. The simplest model only incorporated the variables obtained from the multiple regression model. Next, a more complex model was fitted with the correlations between the exogenous variables explicitly specified.
Evaluation of alternative models was based on four basic statistics: a
2 goodness of fit statistic, the degrees of
freedom (a function of the number of parameters in a model), a
P value, and the Aikaike Information Criterion (AIC), which
index the fit of the model and its parsimony. Large
2 values indicate a poor fit (low P
value), whereas small 2 values indicate that
the model is consistent with the observed data (high P
value). However, the P value is also dependent on the number
of degrees of freedom in each model. The AIC is a compromise measure of
the goodness of fit in relation to the degrees of freedom and is
calculated as 2 minus twice the number of
degrees of freedom. A small AIC is preferable because it indicates the
most parsimonious model, i.e. the one with fewest parameters
but which still fits the data adequately. In cases where the
2 difference test cannot unambiguously
segregate the relative fits of two models, the preferred model is the
one that yields the smallest AIC value. All estimates of regression
parameters were based on least squares method, whereas parameters of
path analyses were based on the maximum likelihood method.
To address the utility of biochemical markers in the classification of
osteoporosis, each subject was first classified as osteoporotic or
nonosteoporotic using FN BMD. Osteoporosis was defined in accordance
with the WHO criteria, i.e. 2.5 SD
below the young normal mean. In our data, mean and
SD of FN BMD for young normal subjects were 1.04
and 0.12 g/cm2, respectively. Next, the
biochemical markers were each dichotomized based on the cut-off values
determined by the regression analysis. For linear variables the
appropriate 25th or 75th quartile was used, i.e. those with
values above the 75th or below the 25th percentiles were defined as
exposed or abnormal; the remainder were defined as nonexposed or
normal. Thus, for three biochemical markers, there were eight possible
combinations of exposure/nonexposure categories (see Table 5
). The
prevalence of osteoporosis was then calculated for all possible
combinations of biochemical parameters. Those subjects who were
nonexposed or normal for all biochemical markers was defined as low
risk and taken as the basal level for all other comparisons (relative
risk = 1). Sensitivity and specificity were estimated from these
results. All analyses were performed using the SAS Statistical Analysis
System (SAS Institute, Cary, NC) (47).
|
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
As expected, age and weight were significantly associated with FN and LS BMD: FN BMD = 0.91 - 0.004 age + 0.004 weight (r2 = 0.15), and LS BMD = 0.58 + 0.003 age + 0.006 weight (r2 = 0.10).
Association among biochemical parameters, age, and anthropometric characteristics
With increasing age there were lower free T (r = -0.20;
P = 0.0001), 25OHD (r = -0.18; P
= 0.0001), and IGF-I (r = -0.19; P = 0.0001)
levels and higher SHBG (r = 0.21; P = 0.0001) and
PTH (r = 0.25; P = 0.0001) levels. This could also
be seen in the increasing number of individuals in either the lowest
(for free T, 25OHD and IGF-I) or highest (for SHBG and PTH) quartiles
for each 10-yr age group increase (Table 1). These changes paralleled the decline
in FN BMD (Table 1). However, there was no clear relationship between
E2 and age.
|
Association between biochemical parameters and BMD
Univariate analysis. The biochemical parameters were all
normally distributed, except for PTH, which had a left skew
(skewness = 2.24). Descriptive analysis by quartile for FN BMD
(Fig. 1) suggested a nonlinear
relationship between FN BMD and both the sex hormones and PTH. For the
sex hormones, the lowest quartile (compared with upper three quartiles)
was associated with a lower FN BMD. For PTH, the highest quartile
(compared with the lower three) was associated with a lower FN BMD.
These parameters were therefore analyzed as dichotomous variables in FN
BMD regression models.
|
) the
relationship with serum levels of E2, free T, and
SHBG appeared to be linear. Thus, these parameters were analyzed as
linear variables in LS BMD regression analysis.
|
31 nmol/L) and E2 (56
nmol/L). This coincided with the lower limit of the normal reference
range for T. For PTH, the cut-off used was the clinical upper limit of
the normal reference range (
6 pmol/L), which was slightly higher than
the upper quartile cut-off level (5 pmol/L).
In univariate linear regression (Table 2), low FN BMD was associated with low
estradiol (56 pmol/L; P = 0.0003), but there was only
borderline significance between low FN BMD and low T (P
= 0.06). There was a negative association between FN BMD and SHBG
(P = 0.0001), and low FN BMD was associated with high
PTH (6 pmol/L; P = 0.0006). There was no association
between FN BMD and 25OHD or IGF-I. Results for trochanteric BMD were
consistent with those at the FN.
|
Multivariate analysis. In multiple regression analysis (Table 3), after adjusting for age and weight,
both E2 (P = 0.01 for FN and LS
BMD) and SHBG (P = 0.0002 for FN and P
= 0.0001 for LS BMD) remained significant independent predictors of
both FN and LS BMD. In addition, PTH (P = 0.03) was an
additional independent predictor of FN BMD, whereas free T
(P = 0.02) was an independent predictor of LS BMD.
Although the relationship between E2 and BMD was
that of a threshold effect at the FN and was linear at the LS, an
E2 level of 56 pmol/L or less was associated with
a 6.5% lower FN BMD and a 6.3% lower LS BMD in multivariate analysis.
The relationships between E2 and BMD at the spine
and hip were not significantly different.
|
Path analysis of interrelationships between hormonal factors.
Although the sex hormones, SHBG and PTH, were independent predictors of
FN and/or LS BMD in multiple regression analysis, they were also
interrelated (Table 4). The strongest
correlation was between free T and E2 (r =
0.50; P = 0.0001). Free T and E2
were also positively correlated with 25OHD (r = 0.18;
P = 0.0001 and r = 0.17; P =
0.0006), but only free T was correlated with IGF-I (r = 0.14;
P = 0.005). SHBG was positively correlated with
E2 (r = 0.15; P = 0.002),
but was not related to free T (P = 0.58). SHBG was
inversely related to IGF-I (r = -0.16; P =
0.0006). PTH and 25OHD were inversely related (r = -0.18;
P = 0.0002).
|
) and LS (Fig. 4) BMD. Multiple regression analysis
cannot account for the intercorrelation between the independent
variables, whereas path analysis allows for the concurrent interaction
between the serum parameters, age, and weight as well as with BMD.
However, the variables entered into the initial path analysis model
were those suggested by the multiple regression models as described
above.
|
|
For LS BMD, the most parsimonious model (P = 0.71; AIC = 5.9) was the one suggesting the following relationships: 1) age, weight, E2, SHBG, and free T were directly related to BMD; 2) age contributed to the variations in E2, free T, SHBG, and IGF-I; 3) weight contributed to the variations in E2 and SHBG; and 4) the same complex interrelationship was noted between the sex steroids, IGF-I and SHBG, as seen for FN BMD.
Thus, in both models, there was a direct association between BMD and age, weight, SHBG, and E2 for both FN and LS BMD with the addition of PTH (for FN BMD) and free T (for LS BMD). These relationships persisted even after allowing for an age and weight adjustment for the serum parameters as well as for the covariation between the sex steroids, SHBG and IGF-I.
Biochemical Parameters and Risk of Osteoporosis
Sensitivities and specificities for the prediction of osteoporosis
(defined as 2.5 SD below young normal values,
i.e. FN BMD <0.74 g/cm2) were
calculated for combinations of abnormal E2, SHBG,
and PTH levels. These were based on the lowest and highest quartiles
for E2 and SHBG, respectively, and a PTH of 6
pmol/L or more (Table 5
). Odds ratios for
osteoporosis were compared with those for subjects who did not have any
abnormal serum parameters. Thus, a subject with any one abnormal
biochemical parameter had a 3.8-fold risk of osteoporosis compared with
a subject who had all normal levels. Although the sensitivity for
osteoporosis detection was quite high (81.3%), the specificity was low
(50.8%). By contrast, a subject with all three abnormal serum
parameters had an 11-fold risk of osteoporosis with a high specificity
(99.5%) but, as expected, a low sensitivity (4.2%).
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
56 pmol/L), higher levels of SHBG
and high PTH (6 pmol/L) were independent predictors of a lower age-
and weight-adjusted FN BMD. Moreover, lower levels of both
E2 and free T as well as higher levels of SHBG
were independent predictors of a lower age- and weight-adjusted LS BMD.
However, of the total variance (20% for FN BMD and 17% for LS BMD)
explained by the combination of these variables, age and weight
accounted for the major proportion, with the biochemical parameters
explaining 5% of the variance in FN BMD and 7% of the variance in LS
BMD. Other parameters, such as 25OHD and IGF-I, were not directly
related to BMD, but a complex interaction was demonstrated among IGF-I,
the sex steroids, and SHBG. The precise mechanism by which E2, T, and SHBG interact to together or individually influence bone density is not entirely clear, but recent data have highlighted some possibilities. The role of estrogen deficiency in the rapid postmenopausal bone loss that occurs in women has been well demonstrated, and estrogen receptors have been shown in human osteoblastic and osteoclastic cells from men as well as women. Several recent studies have demonstrated the importance of estrogen on bone density and skeletal maturation in men. In young male subjects with homozygous mutations of either the estrogen receptor (blocking estrogen action) or the aromatase gene (blocking T conversion to estrogen), bone density was very low, and levels of bone turnover markers were high. Subsequently, a number of epidemiological studies have demonstrated a stronger relationship between estrogen and BMD than between T and BMD in elderly men (19). This is consistent with the findings of the current study, where estrogen was independently associated with both FN and LS BMD, but free T was only associated with LS BMD.
Although male hypogonadism is associated with osteoporosis and can be ameliorated by T replacement, the actual contribution of testosterone vs. estrogen in elderly men with hormone levels in the normal or low normal range remains unknown. Elderly men have levels of estrogen similar to or even greater than those in postmenopausal women (18) consistent with levels in the current study. Even these low levels of estrogen in women have been reported to be correlated with bone density and fracture rates and, when rendered even lower by blockade of peripheral conversion of androgens, have been associated with increased bone resorption (20). Thus, it is plausible that the mechanism of action of the sex steroids on bone in elderly men relates more to available estrogen action obtained from conversion of androgen, although these studies have not yet been carried out.
In this study, SHBG increased with advancing age, as reported by Khosla and others (18) and, similar to E2 was found to be an independent determinant of FN and LS BMD. SHBG binds both E2 and T, leading to reduced bioavailability of the sex steroids and possibly decreased clearance. Regardless of the mechanism, SHBG was an independent predictor of BMD even after adjusting for levels of the sex steroids. SHBG has been previously demonstrated to be associated with bone density in elderly men, although in one study, it was only significantly related at the greater trochanter and not as good a predictor as E2 (19). However, in women it has been shown to be independently associated with bone mass (48), bone loss even after adjusting for E2 levels (21, 48), and hip and vertebral fracture (23).
The interrelationships between all of the biochemical parameters makes it difficult to determine which of the parameters directly or indirectly affect BMD. The use of path analysis, which can account for interrelated data, may help in generating useful hypotheses. For example, age, weight, SHBG, and E2 were directly related to both FN and LS BMD suggesting direct causative effects. PTH for FN and free T for LS BMD were also directly related, similarly suggesting a direct causal relationship. Age contributed further to the variations in E2, SHBG, free T, PTH, and IGF-I in the expected directions, whereas there was a positive association between weight and E2 and a negative association between weight and SHBG as reported by others. The well recognized positive correlation between E2 and free T was demonstrated, consistent with peripheral conversion of androgens to estrogens. There was a positive association between SHBG and E2, but not between SHBG and free T. This may relate to the fact that the major part of total E2 is bound to SHBG, whereas free circulating T (as measured) is not bound to SHBG.
A complex relationship was also demonstrated between the sex steroids, SHBG and IGF-I, independent of age and weight. IGF-I and free T were positively, albeit weakly, correlated with each other, whereas there was a negative association between IGF-I and SHBG. These associations, at least theoretically, would act on bone turnover in the same direction, as lower IGF-I, lower free T (and estradiol), and higher SHBG are all associated with lower BMD. Of these interrelated variables, only SHBG and E2 were directly associated with FN BMD, with the addition of free T for LS BMD. Although mechanisms of action cannot be deduced from this study, this observation suggests that any influence of IGF-I on BMD may be indirect, via the sex steroids or by changes in SHBG.
There is considerable evidence in the literature to support an interrelationship between the activity of the sex steroids and the GH/IGF-I axis. Both undergo parallel large increases during the pubertal growth spurt (49). Increases in IGF-I have been demonstrated after exogenous T administration in both prepubertal boys (50) and adult men (31). In light of the recent focus on E2 rather than T as the primary modulator of bone density, antiestrogen blockade shows that the effect of T on the somatotropic axis in males is at least partly via amortization of T to E2 (51, 52). On the other hand, some researchers have postulated that IGF-I may itself be a regulator of sex steroids at both the pituitary and hypothalamic LH levels (53, 54). Considering the relationship between IGF-I and SHBG, several studies also support an inverse association (29, 55, 56, 57). Indeed, some researchers have postulated that IGF-I may mediate changes in SHBG levels, possibly at the level of liver synthesis, which, in turn, could affect levels of the bioavailable sex steroids (55, 57). Thus, the current analysis has highlighted the complex relationships between these hormone systems, which need to be further explored to enhance our understanding of bone biology.
The lack of direct association between IGF-I and bone density in men is contradictory to some (58), but not all (29), studies in the literature. Many of the positive studies have examined IGFBP-3, the major binding protein of IGF-I, and found that it is a better predictor of BMD than IGF-I (24, 25). Circulating levels of IGF-I and IGFBP-3 are principally determined by the same factors, but levels of the binding protein have been reported to be more stable. Thus, any association between IGFBP-3 and BMD may be easier to detect than between IGF-I and BMD. This study did not measure IGFBP-3 and so cannot address this relationship. One cross-sectional study by Johansson et al. found that 77% of the variation in BMD was accounted for by IGFBP-3, IGF-I, GH, PTH, age, and BMI (24). An important difference between this and the present study is that the study by Johansson et al. examined a younger population of men, aged 25–59 yr. Changes in the IGF-I/GH axis in relation to BMD may be more obvious in this younger sample. According to the path model in the current study, any influence of IGF-I on BMD may be mediated in part via changes in sex steroids or SHBG.
An increase in PTH with age is consistent with data in the literature (34, 59), as is an association between PTH and BMD (24, 59, 60). Indeed, in one study (24), as in the present study, an association between PTH and BMD was observed for the FN, but not for the LS. This may relate to the greater percentage of cortical to trabecular bone in the FN compared with the LS, as increased levels of PTH have been shown to increase bone turnover predominately in areas of cortical bone.
The lack of association between 25OHD and bone density is perhaps not surprising considering the relatively high levels of 25OHD in this population compared with those in European populations, where lack of sunlight is a major concern (33, 61). However, consistent with other studies, we did observe a decrease in 25OHD with age (34). There are several studies demonstrating low vitamin D levels in hip fracture subjects (predominately women) (59, 62) and some benefit of vitamin D therapy on hip fracture prevention, but this clearly involves different populations from that of the current study of ambulatory healthy men.
Despite significant associations demonstrated between these biochemical
parameters and BMD, they only accounted for a small percentage of the
variance in either FN or LS BMD. Thus, the utility of these parameters
in the prediction of osteoporosis is limited. For one or more abnormal
serum parameters (i.e. estradiol 56 pmol/L, PTH 6
pmol/L, and SHBG >50 nmol/L), the sensitivity for osteoporosis
prediction (<2.5 SD below young normal BMD) was
81%, with a specificity of 51%, despite a 4-fold increase in
osteoporosis risk. By including all three abnormal parameters (11-fold
risk of osteoporosis), the specificity increased (99.5%), but
the sensitivity decreased (4%). Thus, these biochemical
parameters may be of use in identifying particularly high risk
individuals, but not in screening elderly men for
osteoporosis.
This study should be interpreted in the light of several considerations. The cross-sectional design of this study does not allow for cause and effect inferences of the biochemical parameters considered. The results were based on a single serum sample measurement, which may underestimate the true relationships between the parameters measured and BMD. A measurement at one point in time does not account for changes throughout a subject’s lifetime. In addition, methodological limitations of the measurements may have weakened the associations found. For example total E2 may not be as precise a measure, as biologically available serum E2 and, similarly, total IGF-I may be less precise than free IGF-I. The analysis of sensitivity and specificity of these hormones was based on small sample sizes; therefore, estimates may be statistically unstable. In addition, no conclusions can be drawn about the role of any of the examined biochemical parameters on peak bone mass, arguably one of the most important predictors of current bone density. However, this study has considerable advantages. This is one of the few epidemiological studies in this area involving a large sample of elderly community-dwelling men aged 60 yr and over. All of the men were community dwelling, thus representing a healthy subsection of the elderly population. This may act to weaken any observed relationship between the biochemical parameters and BMD, as a less healthy population may be assumed to have more abnormal marker values. In addition, as serum samples were not taken at a prescribed time of day but, rather, at the time that the participant was seen, diurnal variation would have acted to weaken any relationship between these serum parameters and BMD. Thus, subtle associations may not have been observed, but the positive results are unlikely to have been exaggerated.
In summary, this study of elderly community-dwelling men has demonstrated that E2 and SHBG were common independent predictors of age- and weight-adjusted FN and LS BMD, with PTH and free T additional independent determinants of FN and LS BMD, respectively. These findings support recent data showing the importance of estrogen in bone biology in men. In addition, a complex relationship between the sex steroids, IGF-I and SHBG, was observed, suggesting that IGF-I may act on bone via changes in the sex steroids, perhaps mediated by SHBG. These biochemical parameters would be weak predictors of bone density in an individual. However, overall they appear to play an important role in the determination of BMD in the elderly and may be particularly relevant in high risk individuals. To fully exploit opportunities for hormone-related management of osteoporosis in elderly men, a better understanding of the precise nature of their action on BMD is needed.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
2 Recipient of a medical postgraduate scholarship from the National
Health and Medical Research Council of Australia.
3 Current address: Wright State University School of
Medicine, Dayton, Ohio 45387.
Received September 14, 1998.
Revised June 28, 1999.
Accepted July 1, 1999.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  S. Khosla, S. Amin, and E. Orwoll Osteoporosis in Men Endocr. Rev., June 1, 2008; 29(4): 441 - 464. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. B. Araujo, T. G. Travison, B. Z. Leder, and J. B. McKinlay Correlations between Serum Testosterone, Estradiol, and Sex Hormone-Binding Globulin and Bone Mineral Density in a Diverse Sample of Men J. Clin. Endocrinol. Metab., June 1, 2008; 93(6): 2135 - 2141. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Bjornerem, L. A. Ahmed, R. M. Joakimsen, G. K R. Berntsen, V. Fonnebo, L. Jorgensen, P. Oian, E. Seeman, and B. Straume A prospective study of sex steroids, sex hormone-binding globulin, and non-vertebral fractures in women and men: the Tromso Study Eur. J. Endocrinol., July 1, 2007; 157(1): 119 - 125. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. A. Fink, S. K. Ewing, K. E. Ensrud, E. Barrett-Connor, B. C. Taylor, J. A. Cauley, E. S. Orwoll, and for the Osteoporotic Fractures in Men Study Group Association of Testosterone and Estradiol Deficiency with Osteoporosis and Rapid Bone Loss in Older Men J. Clin. Endocrinol. Metab., October 1, 2006; 91(10): 3908 - 3915. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. L. Greenspan, P. Coates, S. M. Sereika, J. B. Nelson, D. L. Trump, and N. M. Resnick Bone Loss after Initiation of Androgen Deprivation Therapy in Patients with Prostate Cancer J. Clin. Endocrinol. Metab., December 1, 2005; 90(12): 6410 - 6417. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. M. Kaufman and A. Vermeulen The Decline of Androgen Levels in Elderly Men and Its Clinical and Therapeutic Implications Endocr. Rev., October 1, 2005; 26(6): 833 - 876. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Mueller, R. Dittrich, H. Binder, W. Kuehnel, T. Maltaris, I. Hoffmann, and M. W Beckmann High dose estrogen treatment increases bone mineral density in male-to-female transsexuals receiving gonadotropin-releasing hormone agonist in the absence of testosterone Eur. J. Endocrinol., July 1, 2005; 153(1): 107 - 113. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Gennari, R. Nuti, and J. P. Bilezikian Aromatase Activity and Bone Homeostasis in Men J. Clin. Endocrinol. Metab., December 1, 2004; 89(12): 5898 - 5907. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. Blain, A. Vuillemin, A. Blain, F. Guillemin, N. D. Talance, B. Doucet, and C. Jeandel Age-Related Femoral Bone Loss in Men: Evidence for Hyperparathyroidism and Insulin-Like Growth Factor-1 Deficiency J. Gerontol. A Biol. Sci. Med. Sci., December 1, 2004; 59(12): 1285 - 1289. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
V.-V. Valimaki, H. Alfthan, K. K. Ivaska, E. Loyttyniemi, K. Pettersson, U.-H. Stenman, and M. J. Valimaki Serum Estradiol, Testosterone, and Sex Hormo |
