Age-Related (Type II) Femoral Neck Osteoporosis in Men: Biochemical Evidence for Both Hypovitaminosis D– and Androgen Deficiency–Induced Bone Resorption



The problem of osteoporosis in men has recently been recognized as an important public health issue. To test the hypothesis that endocrine deficiency–mediated alterations in bone metabolism might contribute to osteoporotic fracture risk in elderly men, serum levels of 25-hydroxycholecalciferol (25(OH)D), 1,25-dihydroxycholecalciferol (1,25(OH)2D), intact parathyroid hormone (PTH), testosterone, and estradiol were measured in 40 males (mean age 73 years) who were consecutively recruited within 18 h following a fracture of the proximal femur, and in an equal number of community-living older men (mean age 72 years) who served as controls. In addition, circulating osteocalcin and urinary excretion of (deoxy)pyridinoline were determined as markers of bone formation and resorption, respectively. No differences were observed between the mean serum concentrations of osteocalcin and estradiol. Serum levels of 25(OH)D, 1,25(OH)2D, and testosterone, however, were decreased in hip fracture patients. When correcting for differences in vitamin D binding protein, differences in 1,25(OH)2D did not persist, whereas serum 25(OH)D was still significantly lower in patients than in controls (6.1 ± 4.3 vs. 7.6 ± 2.8, p = 0.01). Similarly, a highly significant deficit was observed in the free testosterone index, calculated from total testosterone and the level of sex hormone binding globulin (2.6 ± 1.3 vs. 8.2 ± 2.9, p < 0.001). Serum PTH and urinary pyridinium cross-links, however, were markedly increased in the fracture group. Moreover, in fracture patients, free 25(OH)D and free testosterone were both significant and mutually independent negative predictors of (deoxy)pyridinoline excretion. Although limited by its cross-sectional design, the present study suggests that both hypovitaminosis D and androgen deficiency may predispose to bone resorption in elderly men and in turn to remodeling imbalance and fracture risk.


MOST STUDIES OVER THE PAST decades on metabolic bone diseases have focused on the pathogenesis and treatment of osteoporosis in women. However, the age-related increase in osteoporotic fractures seen in women is evident in men as well, and reflects an increasing prevalence of skeletal fragility. In particular, the problem of fractures of the proximal femur in men has recently been recognized as an important public health issue. The incidence of hip fractures rises exponentially with age, as it does in women. In 1990, about 30% of 1.66 million hip fractures worldwide occurred in men.1 This number is projected to increase even more as the elderly population increases. In addition, an age-specific increase in fracture incidence has been consistently documented in all populations of men examined,2,3 whereas in women the rate of hip fracture appears to have stabilized. Finally, the mortality associated with a hip fracture in elderly men (aged 75 years or more) is considerably higher than in women, presumably as a result of a higher prevalence of concomitant disease.4

Both cross-sectional5–8 and longitudinal9 studies have provided evidence that the aging process in men is accompanied by changes in the hypothalamic-pituitary-gonadal axis which result in notable declines in serum levels of androgens (testosterone and, to a lesser degree, dihydrotestosterone). In view of the expanding evidence for the involvement of androgens in bone remodeling,10,11 these changes might be involved in the determination of the rate of bone loss, and fractures, in older men.

In addition to androgen deficiency, aging in men has been associated with increased parathyroid hormone (PTH) levels,12,13 reduced 25-hydroxycholecalciferol (25(OH)D) levels,14 and (in some studies) subnormal 1,25-dihydroxycholecalciferol (1,25(OH)2D) levels.15,16 In women, similar changes have been implied in the pathogenesis of senile (type II) osteoporosis.17 However, whether parallel processes are involved in femoral neck osteoporotic fracture occurrence in elderly men remains to be established.

In the present study, systemic hormones affecting bone metabolism and markers of bone remodeling were measured in 40 consecutively admitted male hip fracture patients and compared with the values in a representative sample of 40 older men living in the community. Among the bone and mineral regulating endocrine factors, 1,25- (OH)2D, intact PTH, testosterone, and estradiol were measured. In view of the high prevalence of protein depletion in elderly hip fracture patients,18 serum concentrations of 1,25(OH)2D and its precursor 25(OH)D were adjusted for vitamin D binding protein (DBP).19 Similarly, circulating testosterone and estradiol were corrected for sex hormone binding globulin (SHBG).20 Formation and resorption of bone were assessed by serum osteocalcin and urinary (deoxy)pyridinoline, respectively. Osteocalcin represents a noncollagenous protein predominantly synthesized by the osteoblast and incorporated into the extracellular matrix of bone, and circulating levels have been documented to adequately reflect bone formation.21 Pyridinium cross-links (pyridinoline and deoxypyridinoline), however, are interchain bonds present in the mature form of collagen to stabilize the molecule and appear to be sensitive and specific indices of bone resorption.22 To minimize the potentially confounding effects of the trauma and the subsequent treatment, samples from osteoporotic patients were obtained within 18 h of fracture, prior to surgery.


Study population

The study was conducted in 40 male patients with hip fracture (mean age 73.1 years, range 61–81 years) and 40 elderly controls (mean age 72.4 years, range 63–83 years). Both fracture cases and control subjects were sampled throughout the same period. Informed consent was obtained from all patients and controls, and all procedures were approved by the institutional ethical committee.

Elderly men admitted to the Department of Traumatology following a fracture of the proximal femur were recruited consecutively. To be eligible for participation, men had to be over 60 years of age, to be previously ambulatory, and to have suffered a fall resulting in a radiologically confirmed first hip fracture. Cervical and trochanteric fractures were defined from the surgical report. All osteoporotic patients were studied before surgery and within 18 h of fracture. Patients were excluded if they met any of the following criteria: (1) admitted with a pathologic fracture or fracture resulting from trauma other than a fall; (2) sustained a previous hip fracture; (3) nonosteoporotic metabolic bone disease; (4) thyroid disease, whether controlled or uncontrolled; (5) alcohol abuse; (6) calcium, fluoride, or vitamin D supplements; or (7) ever used thiazides, glucocorticoids (≥5 mg of prednisone or equivalent/day), bisphosphonates, anabolic steroids, or calcitonins for more than 3 months.

An age- and gender-matched sample of 40 community-living control subjects was selected from the Belgian participants in the European Vertebral Osteoporosis Study (EVOS), a multicenter multinational population survey of vertebral osteoporosis, for which men and women aged 50 years and older were recruited from a population-based sampling frame. Details of the recruitment of the subjects have been described previously.23

Anthropometric measurements

Anthropometric measurements were made of height and body weight. Body mass index (BMI) was calculated as body weight divided by height squared (kg/m2).

Biochemical measurements

Fasting blood and urine samples were collected in the morning from all subjects. In the osteoporotic patients, samples were obtained before surgical treatment and within 18 h of fracture. Total serum calcium, inorganic phosphate, albumin, and creatinine were determined by standard analytic methods. Creatinine clearance was estimated according to Cockroft and Gault, relying on serum creatinine, weight, and age.24 Calcidiol (25-hydroxyvitamin D, 25 (OH)D) was measured by competitive binding assay, calcitriol (1,25-dihydroxyvitamin D, 1,25(OH)2D) by radioimmunoassay (RIA), and vitamin D binding protein (DBP) by single radial immunodiffusion. Details of methodology and validation have been previously reported from our laboratory.19,25–28 Both the free 25(OH)D index (based on the molar ratio of 25(OH)D to DBP) and the free 1,25(OH)2D index (based on the molar ratio of 1,25(OH)2D to DBP) were calculated. These molar ratios have been documented to be adequate estimations of the true free concentrations.29 Serum intact parathyroid hormone (PTH) was measured by a two-step immunochemical method, involving an amino-terminal capture and a midregional detecting antibody, as described previously.29 Serum concentrations of testosterone and estradiol were determined by previously reported RIAs.30,31 The intra- and interassay coefficients of variation were 3.5 and 5.1% for testosterone and 4.1 and 7.9% for estradiol, respectively. Measurement of SHBG was performed using ammonium sulphate precipitation.30 Unbound serum sex hormones (free testosterone index and free estradiol index, respectively) were computed as estimates of biologically available testosterone and estradiol from their total serum concentrations and the level of SHBG.32 Analysis of luteinizing hormone (LH) was performed using a commercial immunoradiometric assay (Biosource S.A., Belgium). Dehydroepiandrosterone sulfate (DHEAS) levels were measured by a commercial RIA kit (Diagnostic System Laboratories Inc., Webster, TX, U.S.A.). The intra- and interassay coefficients of variation were 3.2 and 7.0% for LH and 4.2 and 5.3% for DHEAS, respectively. Serum levels of androstenedione31 and human osteocalcin33 were determined by previously reported RIAs. Pyridinium cross-links (pyridinoline and deoxypyridinoline), corrected for creatinine, were measured on hydrolized urine extract by fluorescence detection after high-pressure liquid chromatography.34

Density measurements

Both men with fractured neck of femur (within 1 week after fracture) and control subjects were scanned using dual-energy X-ray absorptiometry (DXA) of the femoral neck region and the trochanteric area. Areal bone mineral density (BMD) was measured using the Lunar DPX-L scanner (Lunar Corp., Madison, WI, U.S.A.). Standard positioning was used with anteroposterior scanning of the right proximal femur except in the event of hip replacement. The precision of femoral neck BMD measurements in elderly subjects using our DXA equipment is 3.1% at the neck and 2.6% at the trochanter.35

Statistical analysis

Differences in clinical and biochemical data between patients and controls were evaluated with Student's t-test. The relation between the biochemical variables was assessed by calculating Pearson's product moment r, based on logarithmic transformation of creatinine clearance, 25OHD, PTH, osteocalcin, and (deoxy)pyridinoline. In view of the fact that no normalizing transformation was found for the parameter time (the interval elapsed after fracture), Spearman rank correlation (ρ) was used to assess the effect of this interval on the biochemical variables. Age-adjusted multiple regression models were constructed with (deoxy)pyridinoline as response and both androgen and vitamin D status as regressors. Both pyridinoline and deoxypyridinoline had to be transformed logarithmically to obtain normality. Interaction of the effects of androgen and vitamin D status was tested by entering the product term of the free testosterone index and the free 25(OH)D index in the regression models. All statistical analyses were conducted with the use of SAS (Statistical Analysis Systems Inc., Cary, NC, U.S.A.). All reported p values are two-sided. The nominal significance level was set at 0.05.


Characteristics of the population

Clinical and biochemical data from the elderly controls and the hip fracture patients are indicated in Table 1. According to the 95% reference interval of a healthy young reference group,29 41% of the fracture patients had elevated levels of PTH (serum PTH more than 40 pg/ml) compared with none of the elderly controls. Vitamin D deficiency, as evidenced by serum 25(OH)D levels below 12 ng/ml,36 was observed in 62% of the osteoporotic patients compared with 18% of the control subjects. Approximately 86% of the fracture patients had androgen deficiency (serum total testosterone less than 300 ng/100 ml) compared with only about 21% of the elderly controls.37 Deficiency of bioavailable testosterone, as evidenced by free testosterone indices below 5, was even observed in 96% of the osteoporotic patients compared with 14% of the control population.

original image

Differences in biochemical parameters and bone density between patients and controls

A comparison of both groups is presented in Table 1. No significant differences were observed for mean age and height between patients and controls. Osteoporotic patients, however, had a lower body weight and a lower BMI. Serum albumin was significantly decreased in fractured men as well. Total calcium and phosphate levels, serum creatinine and calculated creatinine clearance, and osteocalcin were not statistically different. Serum concentrations of the vitamin D metabolites (both 25(OH)D and 1,25(OH)2D) were lower in patients than in the control subjects. However, when correcting for differences in DBP, the difference in serum 1,25(OH)2D was no longer significant. Concentrations of PTH and LH were significantly increased in the fracture group compared with the control group. Mean levels of circulating estradiol and SHBG were not significantly different, whereas the free estradiol index was marginally decreased in patients compared with controls. Serum concentrations of total and free testosterone, however, were markedly lower in patients than in the control subjects. Similarly, a highly significant difference was observed in circulating DHEAS. In contrast, levels of androstenedione were not different between both groups. Finally, the femoral bone density in hip fracture patients was decreased compared with elderly controls, while the urinary resorption of pyridinoline and deoxypyridinoline was significantly increased.

Relationship between biochemical parameters of bone metabolism

Both in elderly controls and in fracture patients, circulating levels of total testosterone and estradiol were positively interrelated (r = 0.71 and 0.60, respectively, p < 0.001), whereas no relationship was observed between serum testosterone and androstenedione (r = 0.20, p = 0.22 and r = 0.18, p = 0.27, respectively) or DHEAS (r = 0.05, p = 0.75 and r = 0.12, p = 0.65, respectively). Serum osteocalcin was related positively to urinary pyridinolines in both groups but unrelated to any of the calciotropic or sex hormones in any group (data not shown). Negative correlations were found between 25(OH)D and PTH, both in controls and patients (r = −0.34, p = 0.05 and r = −0.39, p = 0.02, respectively). In fracture patients, vitamin D status (both 25(OH)D and the free 25(OH)D index) was found to be negatively correlated with pyridinoline (r = −0.44, p = 0.006 and r = 0.43, p = 0.007, respectively) and deoxypyridinoline (r = −0.40, p = 0.01 and r = −0.40, p = 0.01, respectively). Similarly, negative relationships were observed in patients between androgen status (both total testosterone and the free testosterone index) and pyridinoline (r = −0.45, p = 0.003 and r = −0.60, p < 0.001, respectively) or deoxypyridinoline (r = −0.50 and −0.65, respectively, p < 0.001). In contrast, pyridinium cross-links were statistically unrelated to androstenedione or DHEAS (data not shown). In the patient population, multiple regression models indicated that the free testosterone index and the free 25(OH)D index were both significant and mutually independent negative predictors of (deoxy)pyridinoline excretion (Table 2). No significant interaction of free testosterone and free 25(OH)D was observed for pyridinoline (p = 0.69) or deoxypyridinoline (p = 0.14). Including androstenedione or DHEAS into these models did not improve the model precision in predicting pyridinoline (p = 0.47 and 0.44, respectively) or deoxypyridinoline (p = 0.21 and 0.24, respectively). In line with these models, the highest levels of bone resorption were observed in patients with both hypovitaminosis D and testosterone deficiency (Fig. 1). Similar relationships between androgen or vitamin D status and markers of bone resorption, however, were not statistically significant in the control group (data not shown).

Figure FIG. 1.

Bone resorption in male hip fracture patients resulting from the combined effects of hypovitaminosis D and androgen deficiency.

original image

Effect of time elapsed since fracture

According to the cross-sectional distribution of single values from all the patients, there was no significant correlation between the time elapsed after fracture and serum albumin (r = 0.17, p = 0.23), DBP (r = 0.09, p = 0.86) or SHBG (r = 0.11, p = 0.77). Similarly, serum osteocalcin (r = 0.19, p = 0.18) and urinary pyridinium cross-links (r = 0.16, p = 0.26) did not correlate with the time elapsed between sustaining the fracture and the sampling. Finally, no relationship was observed between the time since fracture and vitamin D or androgen status (data not shown).


Vitamin D deficiency has been implicated in the induction of hyperparathyroidism-mediated osteoporosis,38 among other factors. In the Baltimore Longitudinal Study of Aging, lower radial bone density in men was significantly related to higher PTH levels and lower 25(OH)D concentrations.39 Similarly, an inverse relationship was observed between PTH levels and BMD of the proximal femur in a recent study involving 133 community-based elderly men, aged 65–76 years.40 After adjustment for age and BMI, PTH showed a consistent downward trend with increasing bone density at all femoral measurement sites, supporting the view that secondary hyperparathyroidism contributes to bone loss in elderly men. In line with this concept, male hip fracture patients in this study were found to have significantly lower levels of 25(OH)D, coupled with higher concentrations of serum PTH and urinary pyridinium cross-links. The differences in circulating 25(OH)D and intact PTH between patients and controls agree with recent reports in women41–43 and cannot be explained by seasonal variation44 because both groups were equally distributed over the year.

In view of the fact that 25(OH)D and 1,25(OH)D are more than 99% bound to serum proteins, in particular to DBP,19,28 part of the decrease in concentration of serum vitamin D metabolites in the osteoporotic patients might be related to the low levels of circulating proteins. However, when correcting for DBP, serum 25(OH)D levels were still significantly lower in fracture cases. In contrast, estimates of the free 1,25(OH)D concentrations showed no significant differences between patients and controls, suggesting that 1,25(OH)D levels are maintained in the normal range in male hip fracture patients by an increase in serum PTH. Similar findings have recently been reported in elderly women with hip fracture.43 These results support the concept that the sensitivity of the 25(OH)D 1α-hydroxylase to PTH is not impaired in elderly osteoporotic patients.36 The absence of a significant decrease of the active 1,25-hydroxylated metabolite may explain the low incidence of osteomalacia in cases of hip fracture.45,46

Compared with healthy elderly controls, male hip fracture patients were found to be significantly androgen deficient. The reduced testosterone levels in osteoporotic subjects were associated with a rise in serum LH, consistent with primary testicular failure. Even after adjusting for differences in SHBG,20 systemic testosterone concentrations were still markedly lower in fracture cases. Androgens may indeed be essential for the maintenance of bone mass, since the development of hypogonadism in adult men is associated with osteopenia.47 In fact, studies that have compared bone density between clinically distinct groups of hypogonadal males and age-matched controls have invariably shown significant differences in bone mass.48,49 In addition, there have been several reports of increases in BMD after testosterone replacement in hypogonadal adult men50,51 although, in general, the response of bone mass to androgen replacement has been modest. Finally, in recent retrospective case-controlled studies (involving 17 and 28 hip fracture cases, respectively), reduced serum levels of free testosterone were observed in elderly patients compared with controls,52,53 suggesting that age-associated testosterone deficiency may increase the risk of fracture. However, these retrospective studies included selected control populations. In addition, serum samples were drawn in institutionalized patients up to 24 months following the occurrence of minimal trauma hip fracture. Since chronic medical conditions may diminish testosterone secretion,8 it cannot be determined whether hypogonadism preceded the fracture or was secondary to fracture-induced debilitation.54 In the present study, patients were consecutively recruited within 18 h of trauma and compared with a representative sample of community-living older men. Moreover, testosterone deficiency in fracture patients was found to be significantly related to an increase in urinary (deoxy)pyridinoline, suggesting that—in addition to vitamin D deficiency—low bioavailable testosterone levels may contribute to bone resorption in elderly men and in turn to remodeling imbalance and fracture risk. These results are consistent with experimental evidence for an inhibitory effect of androgens on bone resorption in various orchidectomized rat models.55–57

In postmenopausal women, Ooms et al. recently reported higher serum PTH levels and, by implication, increased bone resorption in women in whom low serum 25(OH)D was combined with high serum SHBG, i.e., low sex hormone concentrations.58 In line with these findings, they subsequently observed that the effect of vitamin D supplementation on femoral BMD was greater in women with high serum SHBG than in those with lower serum levels.59 In this study, however, no significant interaction of free testosterone and free 25(OH)D was observed for pyridinoline or deoxypyridinoline. These findings suggest that, in elderly men, the effects of hypovitaminosis D and androgen deficiency on bone resorption are additive but not synergistic.

In addition to 25(OH)D and testosterone deficiency, aging in men is accompanied by a progressive decline of the serum levels of the adrenal androgens dehydroepiandrosterone (DHEA) and androstenedione.60,61 DHEAS, the sulfate ester of DHEA, constitutes about 80% of the mixture of adrenal androgens in the circulation and declines at approximately 2% per year, leaving a residual value of approximately 10–20% during the eighth and ninth decades of life.62 This decrease is not the result of a change in the metabolism of DHEAS, but of a diminished adrenal secretory rate.63 Recent in vitro data have suggested a role for DHEAS in the regulation of human osteoblast function,64 but the extent to which adrenal androgens are involved in the maintenance of bone mass in vivo remains to be established.65,66 In this study, reduced serum levels of DHEAS in male hip fracture patients were statistically unrelated to the excretion of pyridinium cross-links, suggesting that adrenal androgen deficiency does not contribute to the increase in bone resorption and bone fragility in aging men.

The present analysis has several limitations. In particular, markers of bone resorption67 as well as androgen levels68 might be affected by the post-traumatic reaction. The generation of diagnostically useful values may therefore depend critically on the timing of samples. Our osteoporotic patients were sampled within 18 h, prior to surgery, and protein concentrations, circulating testosterone, and pyridinium excretion were unrelated to the time elapsed after fracture. Therefore, both the androgen deficiency and the increase in bone resorption observed in the present study are likely to reflect pre-existent alterations, rather than short-term metabolic changes induced by the trauma. Nevertheless, we acknowledge that it is impossible to control completely for the effect of the moment of trauma in a cross-sectional analysis and that confirmation would require a longitudinal study design. Moreover, cross-sectional correlations such as those presented here can only suggest cause and effect relationships between predictor variables and measures of bone resorption. Finally, our study was not designed to address the therapeutic value of hormone substitution. In fact, whether vitamin D or androgen therapy offers any clinical benefit to older men remains to be established. In elderly (institutionalized) women, supplemental calcium and cholecalciferol is associated with a suppression of PTH levels, a decreased rate of bone loss at the proximal femur, and a reduced incidence of hip fractures.17 In contrast, no prospective studies addressing this issue have been reported in elderly men. Similarly, no attempts have been made to examine the antifracture efficacy of testosterone substitution. In a recent preliminary trial, Tenover reported that parenteral testosterone supplementation in 13 elderly men aged 57–76 years reduced urinary hydroxyproline excretion,69 consistent with a suppressive effect of testosterone on bone resorption. However, BMD was not assessed, and it cannot be excluded that the urinary hydroxyproline changes were caused by the effects of testosterone on the skin. In this regard, the findings of the present study support the need for further investigations to assess the efficacy of vitamin D substitution as well as the potential of exogenous testosterone in preventing or reversing osteopenia and lowering the rate of hip fracture in selected aging men.

In summary, our data suggest that among older men both hypovitaminosis D and androgen deficiency are associated with an increase in bone resorption and bone fragility. The progressive decline in serum 25(OH)D and testosterone levels over time could be important pathophysiologic components of the age-related osteoporotic syndrome in men.


The expert data management by S. Breemans is gratefully acknowledged. We are grateful to E. Van Herck, J. Peeters and I. Jans for skillful laboratory assistance. This work was supported in part by a grant from the Hellemans Foundation (Fonds Prof. Dr. Hellemans).