Dr Orwoll receives funding from Zelos Therapeutics, Novartis, Pfizer Inc., Merck & Co., Inc., Eli Lilly and Company, and Aventis. He also serves as a consultant for Procter & Gamble, TAP Pharmaceutical Products Inc., and Eli Lilly and Company. All other authors state that they have no conflicts of interest.
Free Testosterone is an Independent Predictor of BMD and Prevalent Fractures in Elderly Men: MrOS Sweden
Article first published online: 17 JAN 2006
Copyright © 2006 ASBMR
Journal of Bone and Mineral Research
Volume 21, Issue 4, pages 529–535, April 2006
How to Cite
Mellström, D., Johnell, O., Ljunggren, Ö., Eriksson, A.-L., Lorentzon, M., Mallmin, H., Holmberg, A., Redlund-Johnell, I., Orwoll, E. and Ohlsson, C. (2006), Free Testosterone is an Independent Predictor of BMD and Prevalent Fractures in Elderly Men: MrOS Sweden. J Bone Miner Res, 21: 529–535. doi: 10.1359/jbmr.060110
- Issue published online: 4 DEC 2009
- Article first published online: 17 JAN 2006
- Manuscript Accepted: 11 JAN 2006
- Manuscript Revised: 20 DEC 2005
- Manuscript Received: 15 SEP 2005
The role of androgens for bone health in elderly men is unclear. We show that free testosterone within the normal range is a predictor of BMD at predominantly cortical bone sites and of previous osteoporosis-related fractures in elderly Swedish men.
Introduction: Osteoporosis-related fractures constitute a major health concern not only in women but also in men. Previous studies have clearly shown that serum levels of estradiol are associated with BMD, whereas more conflicting data have been presented regarding the predictive value of testosterone (T) for bone health in elderly men. The aim of this study was to investigate if serum levels of T are associated with BMD and/or prevalent fractures in a large cohort of elderly men.
Materials and Methods: In the Swedish part of the MrOS study (n = 2908; average age, 75.4 years), bone parameters were measured using DXA, and prevalent fractures were recorded using standardized questionnaires and by vertebral X-ray analyses. Serum levels of total T, total estradiol (E2), and sex hormone–binding globulin (SHBG) were measured by radioimmunoassay, and free T (FT) and free E2 (FE2) were derived from the mass action equations. Height, weight, age, physical activity, smoking habits, and calcium intake were included together with FT and FE2 in regression models for BMD.
Results: FT was an independent positive predictor of BMD in total body, total hip, femur trochanter, and arm but not in the lumbar spine. The highest independent predictive value of FT was found in the arm and the hip (with a relatively high content of cortical bone). FE2 was an independent predictor of BMD at all bone sites studied, and the highest predictive value was seen for lumbar spine (with relatively high content of trabecular bone) BMD. FT but not FE2 was a positive predictor of total body bone area and BMC. FT levels below the median were independent predictors of prevalent osteoporosis-related fractures (OR, 1.56; 95% CI, 1.14–2.14; p < 0.01) and X-ray–verified vertebral fractures (OR, 2.00; 95% CI, 1.34–2.86; p < 0.001). The predictive value of FT for prevalent fractures was not affected by adjustment for BMD.
Conclusions: These findings show that variation of FT within the normal range is an independent but modest predictor of BMD at predominantly cortical bone sites and of previous osteoporosis-related fractures in elderly men. Our data indicate that not only estrogens but also androgens are of importance for bone health in elderly men. Longitudinal studies investigating the predictive value of T for fracture risk in elderly men are required.
Osteoporosis-related fractures constitute a major public health concern in women as well as in men, in whom the lifetime fracture risk at age 45 has been estimated to 24%.(1) Apart from major social and economic costs associated with fracture, mortality increases after both hip and nonhip major fractures. This mortality is even higher in men than in women.(2) Thus, it is important to clarify the pathogenesis of osteoporosis and fracture in men, who have been less widely studied than women, to aid in prevention and treatment. The fracture risk is dependent on several factors, including neuromuscular function, BMD, and bone geometry.(3,4)
Sex steroids are important for the skeletal growth and the maintenance of both the female and the male skeleton.(5–7) However, the relative contribution of androgens versus estrogens in the regulation of the male skeleton is unclear. The effects of testosterone can be exerted either directly through the androgen receptor or indirectly through aromatization to estrogens and further through estrogen receptor-α and/or -β.(6) All three of these sex steroid receptors are expressed in bone,(8–11) and experimental animal studies, using sex steroid receptor–inactivated transgenic mouse models, have indicated that each of these three receptors mediates site specific skeletal effects of sex steroids.(12–15)
Previous studies have clearly shown that serum levels of estradiol (E2) are associated with BMD, whereas possibly because of lack of power in previous studies, more conflicting data have been presented regarding the predictive value of testosterone (T) for BMD in elderly men.(16–22) However, a positive association between T and BMD has been reported in some studies.(17,20,23–25) We have recently reported that free E2 (FE2) is a negative, whereas free T (FT) is a positive, predictor of cortical bone size in young men at the age of peak bone mass.(26) This finding supports the notion that estrogens reduce, whereas androgens increase, cortical bone size, resulting in the well-known sexual dimorphism of the cortical bone geometry.
In women, it has been shown that low E2 levels are associated with increased risk of incident fractures.(27,28) In contrast, most studies have failed to show any significant predictive value of sex steroids for fractures in men. Previous studies of the predictive value of sex steroids for prevalent or incident fractures in men have in general included a low number of fractures (<55 fractures), resulting in a rather low power in these studies.(28–31) However, in one study investigating 28 men with and 324 men without prevalent vertebral fractures, it was found that E2 but not T was negatively associated with fractures in elderly men.(30)
Thus, because conflicting data from previous studies, including a relatively low number of elderly men (previous studies have included <600 subjects with available BMD data and <55 men with fractures) have been presented, the aim of this study was to investigate if T is a predictor of BMD and/or prevalent fractures in a rather large population-based cohort (2908 subjects with both BMD and sex steroids measured, including 513 subjects with at least one fracture) of elderly men. Our hypothesis is that FT is an independent predictor of BMD and prevalent fractures in elderly men.
MATERIALS AND METHODS
The MrOS study is a multicenter study including elderly men in Sweden (≅3000), Hong Kong (≅2000), and the United States (≅6000). In this study, we investigated the associations between sex steroid levels and BMD/fractures of the Swedish part of the MrOS study (Sex steroid levels, BMD, and fracture frequency were available for 2908 subjects) in subjects with an average age of 75.4 years (Table 1). Study subjects (men 69–80 years of age) were randomly identified using national population registers, contacted, and asked to participate. To be eligible for the study, the subjects had to be able to walk without aids, and they were not allowed to have bilateral hip prosthesis. There were no other exclusion criteria. The MrOS Sweden study was approved by the ethics committees at Göteborg, Lund, and Uppsala Universities. Informed consent was obtained from all study participants.
Assessment of covariates
Height and weight were measured using standard equipment. Two consecutive measurements of height were performed in the same session, and the average of these measurements was calculated. If there was a discrepancy of ≥5 mm between the first two measurements, a third measurement was performed, and the average of the two values with the least mutual discrepancy was calculated. A standardized questionnaire was used to collect information about amount of physical activity, nutritional intake, and smoking. Physical activity was the subject's average total daily walking distance, including both walking as a means of exercise and leisure, and as a means of outdoor transportation in activities of daily life. Calcium intake was calculated using information from the questionnaires about amount of calcium-containing foods (dairy products, vegetables, nuts, grains, fish etc.; Table 1).
Assessment of BMD (DXA)
Areal BMD (aBMD; g/cm2) of the total hip, femoral trochanter, and lumbar spine (L1–L4) was assessed using the Lunar Prodigy DXA (n ≅ 2000; GE Lunar Corp., Madison, WI, USA, subjects investigated in Malmö [n ≅ 1000] and Uppsala [n ≅ 1000]) or the Hologic DXA Hologic QDR 4500/A-Delphi (n ≅ 1000; Hologic, Whaltman, MA, USA, subjects investigated in Göteborg). DXA measurements of BMD, bone area, and BMC of the whole body and BMD of the left arm (arm BMD was taken from the whole body scan including the entire arm) were assessed using the Lunar Prodigy DXA (n ≅ 2000). The CVs for the aBMD measurements ranged from 0.5% to 3%, depending on application. To be able to use total hip, femoral trochanter, and lumbar spine (L1–L4) DXA measurements performed with equipment from two different manufactures (Lunar Prodigy and Hologic QDR 4500/A-Delphi), standardized BMD (sBMD) was calculated for these bone sites as previously described.(32–34) Furthermore, to minimize the possible confounding effect of using two different DXA machines, DXA type (Hologic or Lunar) was included as a covariate in all regression analyses with total hip, femoral trochanter, or lumbar spine (L1–L4) BMD as dependent factors.
Assessment of fractures
Self-reported previous fractures after 50 years of age were recorded using standardized questionnaires. Seventeen percent (n = 513) of the subjects reported at least one previous fracture after 50 years of age, and 6.4% (n = 193) of the subjects had a self-reported fracture at any of the major osteoporosis-related sites (hip, vertebrae, radius, and humerus; Table 1). Vertebral fractures are the most common type of osteoporotic fracture, but more than two-thirds remain undetected.(35) Therefore, spine X-ray analyses were performed in a subsample of the cohort. Evaluated spine radiographs were available for most of the subjects (907/1000) from Malmö but not for the subjects from Göteborg or Uppsala. The X-rays were taken of both the thoracic and lumbar spine. The lateral view was examined by an experienced radiologist (IRJ). The shape of each vertebrae was evaluated, and if the vertebral body had a reduced height of 3 mm or more compared with the vertebrae above, it was classified as a vertebral fracture (vertebras are increasing in height in caudal direction). However, if there only was a minor difference, close to a reduction of 3 mm in height, together with irregularities of the endplates, the films were compared with old spinal or pulmonary X-rays (from 1960 and onward). If the shape of the vertebras was unchanged and there was no history of trauma (from the old reports), the wedge-shaped vertebra was classified as a developmental disease of the vertebrae (such as Mb Scheuermann, defined as the presence of abnormalities [predominantly as vertebral end plate irregularities] in at least three vertebrae, each with wedging of 5° or more) and thus not as a fracture. If there were no old films available, the wedged vertebra was classified as a fractured one. The spine X-ray analyses revealed that 18% of the subjects had at least one prevalent vertebral fracture (907 subjects investigated and 161 fractures diagnosed; Table 1). Statural height loss from 25 years of age was used as an additional estimation of vertebral compressions. Fourteen percent (n = 351) of the patients displayed a >5-cm age-dependent height loss (Table 1). The presence of X-ray–verified vertebral fracture was correlated to height loss >5 cm (r = 0.185, p < 0.001).
Assessment of sex hormones
Total E2 was measured using an ultrasensitive radioimmunoassay (RIA; Orion Diagnostics, Esboo, Finland, limit of detection 5 pM [140 pg/dl], intra-assay CV 3%, interassay CV 6%). Total T was measured using RIA (Orion Diagnostics, limit of detection 0.1 nM [3 ng/dl], intra-assay CV 6%, interassay CV 6%). Sex hormone–binding globulin (SHBG) was measured using immunoradiometric assay (IRMA) (Orion Diagnostics, limit of detection 1.3 nM, intra-assay CV 3%, interassay CV 7%). Two subjects had undetectable E2 levels and one had undetectable T levels.
FT and FE2 were calculated according to the method described by Vermeulen et al.(36) and Van den Beld et al.,(23) taking the concentrations of total T (TT), total E2 (TE2), and SHBG into account and assuming a fixed albumin concentration of 43 g/liter.
All samples were analyzed in duplicate in one laboratory, and the duplicates were averaged for further analyses.
Differences between characteristics for the different groups of FT (Table 2) were compared by ANOVA for continuous variables and by χ2 for categorical variables. Values are given as mean ± SD. Univariate associations among variables were examined with Pearson correlation after log-transformations of markedly nonnormally distributed variables. The independent predictors of the various bone parameters were tested using linear regression analysis, including age, height, weight, calcium intake, physical activity, and smoking (Tables 3–5).
ORs including 95% CIs were used to determine the predictive value of either levels below the median or levels in the lowest 10 percentiles of FT and FE2 for prevalent fractures (Table 5). A p value <0.05 was considered significant.
FT is a site-specific independent predictor of BMD
Univariate analyses showed that serum levels of T were negatively associated with age (TT: r = −0.058, p < 0.01; FT: r = −0.109, p < 0.001) and body weight (TT: r = −0.206, p < 0.001; FT: r = −0.071, p < 0.001), whereas they were positively associated with calcium intake (TT: r = 0.020, not significant; FT: r = 0.054, p < 0.01), smoking (TT: r = 0.045, p < 0.05; FT: r = 0.031, not significant), and physical activity (TT: r = 0.059, p < 0.001; FT: r = 0.055, p < 0.01; Table 2). E2 levels were negatively associated with age (TE2: r = −0.049, p < 0.01; FE2: r = −0.082, p < 0.001) and positively associated with body weight (TE2: r = 0.075, p < 0.001; FE2: r = 0.143, p < 0.001). Both FT and FE2 were, in univariate analyses, positively associated with BMD in the total hip, femur trochanter, arm, and spine (FT: r = 0.062–0.128, p < 0.01; FE2: r = 0.152–0.203, p < 0.01). Age, height, weight, physical activity, smoking status, and calcium intake were included as covariates in further calculations of the predictive values of E2 and T for BMD and fractures.
Total levels as well as free levels of both T and E2 were predictors of BMD at all bone sites when included separately in regression models including height, weight, age, physical activity, smoking habits, and calcium intake (Table 3). For all bone sites studied, FT and FE2 were stronger predictors of BMD than the respective total sex hormone levels (Table 3).
Because serum levels of FT and FE2 were correlated (r = 0.59, p < 0.001), we next included FT and FE2 simultaneously in regression models (Table 4). These analyses showed that FT was an independent positive predictor of BMD in total body, total hip, femur trochanter, and arm, whereas it did not predict the lumbar spine BMD (Table 4). Separate evaluation of the large group of men analyzed by the Lunar densitometers (n ≅ 2000) revealed a similar pattern that FT was an independent predictor of total hip and femur trochanter but not of lumbar spine BMD (data not shown). The highest predictive value was found in bones with a relatively high content of cortical bone, including the arm and the hip. After the exclusion of subjects in the lowest 10 percentile of FT, the independent predictive value was clearly reduced for most of the above-mentioned bone sites, suggesting that a threshold might exist for FT as a predictor of BMD (data not shown).
FE2 was an independent positive predictor of BMD at all bone sites studied (Table 4). The highest independent predictive value of FE2 was seen for the lumbar spine, which is the bone site with the relatively highest content of trabecular bone (Table 4).
We next studied the independent predictive value of FT and FE2 for bone area and BMC. FT but not FE2 was a positive independent predictor of total body bone area (FT: β = 0.062, p = 0.004; FE2: β = 0.012, p = 0.575) and BMC (FT: β = 0.089, p < 0.001; FE2: β = 0.044, p = 0.059).
FT predicts prevalent fractures independently of BMD
FT levels below the median were positive predictors of previous fractures after 50 years of age, osteoporosis-related fractures, and prevalent X-ray–verified vertebral fractures (Table 5). The predictive value of FT for prevalent fractures was not affected by adjustment for age, height, weight, smoking status, physical activity, and calcium intake (Table 5). A possible involvement of BMD for the predictive value of FT for prevalent fractures was studied by adjustment for BMD. The predictive value of FT for prevalent fractures was neither affected by adjustment for BMD (Table 5) nor by adjustment for grip strength (data not shown). FT in the lowest 10 percentile was predictive not only for the self-reported previous fractures described above but also for height loss >5 cm (Table 5). FT levels below the median predict previous fractures also after the exclusion of subjects in the lowest 10 percentile of FT (X-ray–verified vertebral fractures: OR, 1.92; 95% CI, 1.30–2.84; p < 0.001; osteoporosis-related fractures: OR, 1.45; 95% CI, 1.04–2.03; p < 0.05), supporting the notion that that FT levels within the normal range predict previous fractures.
FE2 below the median did not significantly predict any of the fracture-related parameters. However, FE2 in the lowest 10 percentile was a strong positive predictor of X-ray–verified vertebral fractures and height loss >5 cm (Table 5). The predictive value of very low FE2 for prevalent fractures remained both after adjustments for age, height, weight, smoking status, physical activity, and calcium intake and after adjustment for BMD (Table 5).
Sex steroids are important regulators of the male skeleton. Previous studies have clearly shown that E2 is of importance for bone health in men, whereas conflicting results have been presented regarding the role of T. We show that FT is an independent predictor of BMD and prevalent fractures in a large cohort of elderly men.
For all bone sites studied, FT and FE2 were stronger predictors of BMD than the respective total sex hormone levels, supporting the notion that free (or bioavailable) sex hormone levels should be considered when evaluating the predictive value of sex steroids for bone health.(22,23,36) This study confirms several previous studies, showing that E2 is a predictor of BMD in elderly men.(16–22) The finding, that the highest predictive value of FE2 for BMD was found in the lumbar spine, indicates that estrogens are predominately of importance for the trabecular and less for the cortical bone compartment in elderly men.
This study clearly shows that FT is an independent predictor of BMD at selected bone sites. The conflicting findings in previous studies, regarding the role of T for BMD in elderly men, might be caused by lack of power in these studies (<600 subjects in previous studies(16–20,22–25)). Interestingly, the highest predictive value of FT for BMD, in this study, was found at bone sites with relatively high content of cortical bone. In a similar manner, Khosla et al.(18) previously suggested that T may have a greater influence on BMD in men at sites with predominantly cortical bone than at sites with predominantly trabecular bone. Furthermore, we recently reported that FT is a positive predictor of cortical bone size in young men at the age of peak bone mass.(26) This finding supports the notion that androgens increase cortical bone size. A stimulatory effect of androgens but not estrogens on cortical bone size is further supported by the finding that FT but not FE2 was an independent positive predictor of height-adjusted total body bone area. Because not only BMD but also bone size is of importance for the mechanical strength of bones,(37) we propose that the size of the bones should be evaluated when judging the role of testosterone on bone strength and possibly also fracture risk in men.
The most important and novel finding in this study is that low levels of FT were clear positive predictors of prevalent fractures in our large cohort of elderly men. Previous studies investigating the role of T for prevalent or incident fractures in men have in general included a low number of subjects with fractures, and they all failed to detect any predictive role of T for fractures.(28–31) Because this cohort is rather large, resulting in a high number of prevalent fractures, it has an adequate power to study if serum levels of testosterone are associated with fracture risk. In this study, including >500 subjects with at least one fracture, we showed that FT levels below the median were positive independent predictors of self-reported previous osteoporosis-related fractures and prevalent X-ray–verified vertebral fractures in elderly men. Thus, FT levels within the normal range are of predictive value for prevalent fractures. The predictive value of FT for prevalent fractures was not affected by adjustment for BMD, suggesting that the major part of the associations with fractures is distinct from any effect on BMD. Effects independent of BMD might be markers of biological age and health status not measured by methods that are more traditional, and as such, may be useful in identifying those at a high risk of fracture. A BMD-independent mechanism is supported by the fact that BMD incompletely separates those who will fracture from those who will not fracture. Other risk factors, such as the quality of bone and the propensity to fall, are determinants of fracture risk, although they are harder to quantify.(31) We propose that T may modulate fracture risk by affecting bone structure/quality or by affecting neuromuscular function. However, one should emphasize that this cross-sectional study only gives information about associations and does not provide any information about cause and effect. Well-powered longitudinal studies, investigating the predictive value of T for incident fracture risk in elderly men, are required and are ongoing.
FE2 levels in the lowest 10 percentile but not FE2 levels below the median were positive predictors of prevalent X-ray–verified vertebral fractures, indicating, similarly as has previously been suggested,(6,27,38) that a threshold level exists for E2 to affect bone health. An inverse association between E2 levels and prevalent vertebral fractures in men has previously been described in the Rancho Bernardo study.(30) However, in some other studies, no association between E2 and prevalent or incident fractures in men was found.(28,29,31) The conflicting data regarding the relationship between E2 and fractures in men might, in a similar manner as described above for the relationship between T and fractures, be caused by the fact that very large cohorts are required to, with an adequate power, be able to identify a predictive factor for fractures.
The main limitation of this study is that it is a cross-sectional study. Although the subjects studied were part of a prospective cohort study, the fracture data were cross-sectional and thus causal inferences cannot be made. Therefore, these findings need to be confirmed in prospective studies of elderly men. Furthermore, the cohort is mainly of white origin. Thus, our results may not apply to other populations. Similarly, as in most previous studies, serum sex steroid levels were, in this cohort, measured using standardized RIA. Levels of E2, T, and sex steroid metabolites can also be measured with the more specific but also much more expensive and time-consuming gas chromatographic/mass spectroscopic or liquid chromatographic/mass spectroscopic methods. The main strength of this study is the very large sample size, including BMD, extensive fracture data, and information about several other known risk factors to be included in the regression models for the calculation of the independent predictive value of sex steroids for BMD and fractures.
In conclusion, these findings show that variation of FT within the normal range is an independent but modest predictor of BMD at predominantly cortical bone sites and of previous osteoporosis-related fractures in elderly men. Our data indicate that not only estrogens but also androgens are of importance for bone health in elderly men. Longitudinal studies investigating the predictive value of T for fracture risk in elderly men are required.
The authors thank Anette Hansevi, Maud Petterson, Ann-Charlotte Adolfsson, Marja Gustafsson, Inger Abrahamsson, Violja Mixhue, and Karin Önnby for excellent technical assistance and Professor J. M. Kaufman for valuable advice regarding calculations of free sex steroid levels. This study was supported by the Swedish Research Council, the Swedish Foundation for Strategic Research, European Commission Grant QLK4-CT-2002-02528, the ALF/LUA research grant in Gothenburg, the Lundberg Foundation, the Torsten and Ragnar Söderberǵs Foundation, Petrus and Augusta Hedlunds Foundation, and the Novo Nordisk Foundation