By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
Several lines of evidence implicate estrogen deficiency as a cause of bone loss in elderly men. Thus, in 50 elderly men (mean age ± SD, 69.1 ± 6.0 years), we performed a randomized blinded study to assess the effect of 6 months of treatment with 60 mg/day of raloxifene (a selective estrogen receptor modulator [SERM] that has an agonist effect on bone but is not feminizing) versus placebo on bone turnover markers. The mean changes in bone turnover markers, serum sex steroid, or lipid levels with treatment did not differ between groups. However, changes in urinary cross-linked N-telopeptide of type I collagen (NTX) excretion were related directly to the baseline serum estradiol level in the raloxifene (r = 0.57; p = 0.004) but not in the placebo-treated (r = 0.15; p = 0.485) men (p = 0.015 for the difference between groups). Moreover, the men in whom NTX excretion decreased after raloxifene treatment had significantly lower baseline estradiol levels (mean ± SEM, 22 ± 2 pg/ml) than the men in whom urinary NTX excretion didn't change or increased after raloxifene therapy (30± 3 pg/ml; p = 0.03), and no such difference was found in the placebo group. Thus, raloxifene has no significant effect on bone turnover markers or lipid levels in elderly men. However, the association noted between baseline estradiol levels and the change in urine NTX excretion in the raloxifene-treated men suggests that a subset of men with low estradiol levels may respond to raloxifene or other SERMs, and further studies are needed to directly test this possibility.
OSTEOPOROSIS IN men is now recognized as an important public health problem that is expected to increase in scope with expansion of the elderly population. The incidence of fractures caused by minimal-to-moderate trauma rises rapidly with aging in both men and women and reflects an increasing prevalence of skeletal fragility.(1) Elderly men have age-specific hip and vertebral fracture rates that are at least one-half of those in women.(2) Recent estimates indicate that of the $13.8 billion in total healthcare cost attributed to osteoporosis each year in the United States, $2.7 billion is accounted for by fractures in men.(3)
Because of the similarity in the pattern of rapid bone loss that follows oophorectomy(4) and orchiectomy,(5) respectively, the conventional belief is that skeletal mass is maintained mainly by estrogen in women and testosterone in men. Although men do not have a menopause equivalent and their levels of serum total testosterone decrease only marginally with age,(6) they still incur two-thirds of the bone loss sustained by women, as shown in both cross-sectional(7) and longitudinal(8) studies. Moreover, after accounting for the lack of a rapid loss phase that immediately follows menopause, aging men show continuous bone loss and an increase in serum parathyroid hormone (PTH) and bone resorption markers in a pattern that is virtually superimposable on that occurring in elderly women,(9, 10, 11) suggesting a common pathogenic mechanism.
Several recent lines of evidence indicate that estrogen may play a significant role in male bone metabolism. First, two experiments of nature indicate that normal estrogen levels are required to achieve peak bone mass. An adult man with homozygous null mutations of the estrogen receptor was found to have unfused epiphyses, marked osteopenia, and elevated bone turnover indices.(12) Subsequently, a similar skeletal phenotype was described in two men with complete aromatase deficiency.(13, 14) In both instances, bone mineral density (BMD) was significantly reduced despite normal or increased androgen levels, and in the men with aromatase deficiency, BMD improved significantly after initiation of estrogen replacement therapy.(15) Second, a number of recent cross-sectional observational studies have concluded that serum levels of estrogen predicted BMD in large male cohorts better than serum levels of testosterone.(10, 16, 17) Finally, we recently have extended these observations by showing that levels of serum estrogen, but not of serum testosterone, predict rates of bone resorption and bone loss in elderly men(18) and that estrogen is the dominant sex steroid preventing the increase in bone resorption after the induction of hypogonadism and aromatase inhibition in elderly men.(19) Moreover, serum levels of free and bioavailable estrogen decrease by approximately 50% with aging in men.(10) Thus, there is strong evidence that estrogen deficiency may be a major determinant of age-related bone loss in men.
If this hypothesis is correct, estrogen replacement should be effective in preventing bone loss in these men. Clearly, long-term estrogen replacement therapy is not feasible in men because of side effects. Raloxifene, a selective estrogen receptor modulator (SERM), has estrogen-agonist effects on bone but is not feminizing(20) and, thus, is likely to be better tolerated by elderly men than estrogen. However, on the average, elderly men (>70 years) have circulating estradiol levels that are approximately twice as high as those in postmenopausal women,(10) and it is possible that only a subset of men with relatively low estradiol levels will respond to raloxifene. Thus, in this study, we treated elderly men with raloxifene or placebo for 6 months and assessed the effect of treatment on bone turnover markers, sex steroid and gonadotropin levels, and blood lipids. We also tested the relationship, if any, between baseline estradiol levels and the response to treatment in the two groups.
MATERIALS AND METHODS
After approval by the Mayo Institutional Review Board and obtaining written, informed consent, we recruited 50 men aged ≥60 years for the study. Subjects were recruited from lists of participants in previous studies conducted at the Mayo Clinic, newspaper advertisements, and seminars given by the investigators. The men underwent a screening questionnaire and physical examination and were eligible to participate if they had serum bioavailable (non-sex hormone-binding globulin [non-SHBG] bound) testosterone concentrations >1 SD below the mean of men aged 20-30 years (<132 ng/dl). Because we did not have a rapid estradiol assay for screening purposes, this criterion ensured that the men had at least mild sex steroid deficiency. Exclusion criteria included current or previous treatment with medications known to affect bone metabolism (glucocorticoids, anticonvulsants, heparin, thiazide diuretics, oral fluoride preparations, calcitonin, bisphosphonates, sex steroids, calcium supplements >600 mg elemental calcium daily, and vitamin D supplements >1000 IU daily), significant medical illnesses (renal failure, malabsorption, endocrine disorders other than diabetes and treated hypothyroidism, congestive heart failure, and malignancy), a prior thromboembolic event, inability to meet follow-up requirements, and abnormalities on screening laboratory tests (complete blood count, serum sedimentation rate, creatinine, calcium, phosphorus, total alkaline phosphatase, aspartate aminotransferase [AST], total bilirubin, albumin, and prostate-specific antigen [PSA]). Of the 62 men who passed the initial screening questionnaire, 6 men were excluded because of abnormalities in non-sex steroid screening laboratory values (hypocalcemia, hyperbilirubinemia, elevated alkaline phosphatase, elevated PSA, and 2 men with anemia), and 6 additional men were excluded based on their bioavailable testosterone being above eligibility cut-off values.
Treatment protocol and follow-up studies
The men were assigned to therapy with raloxifene (Evista; kindly provided by Dr. Michael Draper at Eli Lilly Co., Indianapolis, IN, USA), 60 mg orally per day, taken in the morning or a matching placebo, based on randomization performed by our research pharmacy. Study subjects and the investigators remained blinded to the study medication. All subjects continued their usual medications and calcium/vitamin D supplements. After the screening visit and randomization, subjects were admitted to the general clinical research center (GCRC) for a baseline stay of 2 days and 3 nights. During this stay, a physical examination was performed again and serum was collected on the first morning for total and bioavailable testosterone and estradiol, estrone, SHBG, luteinizing hormone (LH), and follicular-stimulating hormone (FSH) and on the morning of both days for bone-specific alkaline phosphatase (BSAP) and PTH. Two consecutive 24-h urine collections were obtained for determination of type I collagen aminoterminal cross-linked N-telopeptide (NTX). The men were instructed to maintain physical activity similar to their usual levels, and their GCRC diets were matched for their usual intakes of calcium and calories according to a standardized diet questionnaire and interview with a dietitian. On completion of the baseline GCRC stay, the subjects were given a 3-month supply of the study medication (raloxifene or placebo) to which they were randomized.
Three months after initiating the medication, the men were questioned in person about the occurrence and severity of adverse effects and were examined according to their symptoms. Serum LH and FSH were drawn and immediately analyzed to rule out the occurrence of untoward effects of the study medication on the hypothalamic-pituitary-gonadal axis. Study subjects then were given a second 3-month supply of the same study medication to which they originally had been randomized.
Six months after starting the protocol, the men were admitted for a follow-up GCRC stay identical to the one done at baseline except for the fact that all subjects were taking their assigned study medications while serum and urine collections were performed. Subjects were questioned and examined again regarding any adverse events. Any remaining study medication tablets from both 3-month supplies were counted by our study pharmacy to assess compliance.
All laboratory analyses were performed on-site at our institution, and the same assay was used for comparison of a subject's results over the course of the study. Testosterone was measured using a competitive chemiluminescent immunoassay (interassay CV < 10%; Bayer Diagnostics Corp., Tarrytown, NY, USA). Estradiol was measured by a high-sensitivity, double antibody radioimmunoassay (interassay CV < 16%; Diagnostic Products Corp., Los Angeles, CA, USA). Estrone was measured using a competitive radioimmunoassay (interassay CV < 13%; Diagnostic Systems Laboratories, Webster, TX, USA). Bioavailable testosterone and estradiol were measured by differential precipitation of SHBG by ammonium sulfate after equilibration of the serum sample with tracer amounts of tritium-labeled testosterone or estradiol, respectively, as previously described.(10) Other screening tests were done at the Mayo Central Clinical Laboratory. Urinary NTX was measured using a competitive immunoassay technique (interassay CV < 8%; Ortho-Clinical Diagnostics, Rochester, NY, USA). Serum lipids, including total cholesterol, triglycerides, high-density lipoprotein (HDL), and calculated low-density lipoprotein (LDL) were measured in samples obtained after an overnight fast.
Baseline characteristics for raloxifene and placebo groups were compared using two-sided two-sample t-tests for quantitative data and χ2 tests for qualitative data. The mean changes in endpoints for the raloxifene and placebo groups were compared using one-sided two-sample t-tests. Linear regression models were used to test whether the relationship between study endpoints and baseline estradiol levels differed between the treatment groups. The 95% CI for the x intercepts of the linear regression models was calculated using bootstrapping. The occurrence of side effects was compared using the one-sided Fisher's exact test. Correlations between baseline characteristics and treatment effects were calculated using Pearson correlation coefficients. The level of statistical significance was set at p < 0.05. Data processing and analysis were performed with SAS software (SAS Institute, Cary, NC, USA). Unless otherwise indicated, the data are reported as the mean ± SEM.
Clinical characteristics of the study subjects
Table 1 shows the baseline clinical characteristics of the study subjects. As is evident, the subjects were well matched for age and anthropometric indices. All other baseline characteristics, including history of osteoporosis or fracture, calcium and vitamin D supplementation, ethanol consumption, smoking, and prostatism symptoms were similar between placebo and treatment groups (data not shown).
Table Table 1.. Baseline Clinical Characteristics of the Study Subjects
Baseline biochemical data
Table 2 shows the baseline values for serum PTH and bone turnover markers, sex steroids and gonadotropin levels, and serum lipids in the placebo and raloxifene groups. Thus, the subjects were well matched for all of these variables except for serum BSAP, which was slightly higher in the placebo group.
Table Table 2.. Baseline Values for Serum PTH and Bone Turnover Markers, Sex Steroid and Gonadotropin Levels, and Serum Lipids in the Study Patients
Changes in PTH and bone turnover markers
In assessing the effect of treatment with placebo or raloxifene, we analyzed both the changes in the various endpoints in the two groups and the impact of baseline sex steroid levels in the response to treatment in the two groups. Thus, as shown in Table 3, the changes in serum PTH, BSAP, and urine NTX were not different between the groups. However, as shown in Fig. 1, there was a significant relationship between baseline estradiol levels and the change in urinary NTX excretion in the raloxifene group (Fig. 1A), but there was no significant relationship between these parameters in the placebo group (Fig. 1B; p = 0.015 for the difference in estradiol effect between the groups). The regression analysis in Fig. 1A also served to define an estradiol level of approximately 26 pg/ml (95% CI, 20, 32) in which raloxifene had no effect on NTX excretion (ΔNTX = 0). There was no relationship between the change in NTX excretion and age in either group, and total estradiol levels were not related to age in either group (data not shown).
Table Table 3.. Changes in PTH and Bone Turnover Markers, Serum Sex Steroid and Gonadotropin Levels, and Serum Lipids in the Two Groups After Treatment
Consistent with the foregoing findings, the men in whom urinary NTX excretion decreased after raloxifene therapy had significantly lower baseline serum estradiol levels than the men in whom urinary NTX excretion did not change or increased after raloxifene therapy (22 ± 2 pg/ml vs. 30 ± 3 pg/ml; p = 0.03). By contrast, no such difference was present in the placebo-treated men (28 ± 3 pg/ml vs. 28 ± 3 pg/ml; p = 0.832 for the corresponding analysis in the placebo group). There was no difference in age between men in whom NTX increased after raloxifene therapy compared with the men in whom NTX decreased after raloxifene therapy (data not shown). In addition, as shown in Table 4, baseline testosterone levels were not related to the change in NTX excretion in either the raloxifene or the placebo-treated men, nor did they influence the relationship between baseline estradiol levels and the change in urine NTX in either the raloxifene or placebo groups (data not shown). Changes in serum PTH or BSAP also were not related to baseline sex steroid levels in either group (Table 4).
Table Table 4.. Pearson Correlation Coefficients Relating Baseline Estradiol and Testosterone Levels to Changes in PTH and Bone Turnover Markers, Serum Sex Steroid and Gonadotropin Levels, and Serum Lipids in the Two Groups After Treatment
Similar to total estradiol levels, the change in urinary NTX excretion after treatment also was dependent on the baseline bioavailable estradiol level in the raloxifene but not in the placebo-treated men, although the strength of this association was not as strong as for total estradiol (R = 0.42 and p = 0.052 and R = −0.06 and p = 0.768 for the relationship between baseline bioavailable estradiol and ΔNTX in the raloxifene and placebo groups, respectively; p = 0.119 for difference between groups).
Changes in serum sex steroid and gonadotropin levels
There were no significant differences between the groups in changes in serum sex steroid levels (Table 3). Serum LH and FSH increased in the men treated with raloxifene compared with the placebo-treated men, although this was statistically significant only for the change in FSH. Again, the effect of baseline estradiol levels on the changes in serum LH and FSH was clearly different between the groups (Table 4). This is depicted graphically in Fig. 2 for FSH. As evident, there was a significant relationship between baseline estradiol levels and the change in FSH in the men treated with raloxifene (Fig. 2A) but not in the placebo (Fig. 2B) group. Moreover, after raloxifene treatment, the regression analysis predicted a ΔFSH of zero at a baseline estradiol level of approximately 14 pg/ml (95% CI, −26, 23), which is considerably lower than the corresponding value of 26 pg/ml for NTX. Thus, as opposed to NTX, in which only a subset of men had an increase in NTX excretion after raloxifene therapy (Fig. 1A), the majority of the men (21/25) had an increase in FSH levels after raloxifene treatment (Fig. 2A). By contrast, the men on placebo clustered around a ΔFSH of zero (Fig. 2B). As shown in Table 4, baseline testosterone levels were not related to changes in serum FSH or LH.
Changes in serum lipids
There were no significant differences in changes in serum lipid levels between the two groups (Table 3). In addition, changes in lipid levels in either group were not related to baseline sex steroid levels (Table 4).
Compliance and side effects
During the course of the study, there were no dropouts and compliance with study medication use, as assessed by tablet counts, was 98% with no significant difference between placebo and raloxifene groups. The present dose of raloxifene was well tolerated in this group of men aged ≥60 years. No severe adverse reactions occurred during this study (including thromboembolic events). As shown in Table 5, the only significant difference in side effects between the men treated with placebo and the men treated with raloxifene was in the occurrence of hot flushes, which were reported by none of the men receiving placebo treatment but was reported by four of the men receiving raloxifene treatment.
Table Table 5.. Frequency of Side Effects in the Placebo- and Raloxifene-Treated Men
Because declining levels of serum bioavailable estrogen are associated with decreases in BMD(10, 17, 18) and with increased rates of bone resorption and bone loss(18) in elderly men, we tested the hypothesis that the SERM raloxifene would reduce bone resorption in these men. The overall results of the study were clearly negative, because raloxifene was no different from placebo in terms of its effects on changes in urine NTX. Recently, similar results have been reported by Uebelhart et al.(21) who studied 43 men with a mean age of 56 years and found no effect of 120 mg daily of raloxifene on urinary deoxypyridinoline excretion.
However, we did find that the effect of raloxifene on bone resorption in elderly men may be dependent on their endogenous estradiol levels. The men in whom urinary NTX excretion declined after raloxifene treatment had lower estradiol levels than men in whom NTX excretion didn't change or increased after treatment. Moreover, regression analyses suggested that those men with estradiol levels below 26 pg/ml (which corresponds to serum bioavailable estradiol levels of approximately 9 pg/ml) may respond differently to raloxifene than men with estradiol levels above these values. However, our study was not specifically designed to test this directly, and additional studies with sufficient numbers of men with low and high estradiol levels are needed to test this hypothesis.
Nonetheless, our findings with raloxifene are consistent with the reported effects of another SERM, tamoxifen, on bone in premenopausal women. Thus, although tamoxifen treatment preserves or increases BMD in ovariectomized mice,(22) rats,(23) and postmenopausal women,24-27) it is associated with bone loss in premenopausal women.(25, 26) Although serum estradiol levels in pre- and postmenopausal women are either well above or below 26 pg/ml, respectively, estradiol levels in elderly men hover around this value. Thus, in population-based studies, we have found that the median serum estradiol level in postmenopausal women of the same age as the men in our study is 12 pg/ml, with approximately 90% of these women having serum estradiol levels below 26 pg/ml,(10) and several previous studies have shown that comparable postmenopausal women respond to raloxifene treatment with a reduction in bone resorption.28-32) By contrast, the median estradiol level for premenopausal women of 102 pg/ml(10) is well above 26 pg/ml. However, elderly men have a median estradiol level of 30 pg/ml,(10) and approximately 35% of these men have total estradiol levels below 26 pg/ml. Further studies are needed to determine if this subset of elderly men would respond to raloxifene therapy with a reduction in bone resorption and rates of bone loss.
Consistent with the findings of this study, we recently found in population-based studies that elderly men with total and bioavailable estradiol levels below approximately 31 pg/ml and 11 pg/ml, respectively, have higher rates of bone resorption and bone loss than men with estradiol levels above these values.(18) Our findings also are consistent with recent data showing that elderly men with estradiol levels below approximately 20 pg/ml have higher rates of hip fracture than men with estradiol levels above this value.(33) Thus, while the precise estradiol level at which the male skeleton becomes relatively estrogen deficient may differ slightly because of differences in assays, study design, or the outcome assessed (i.e., response to raloxifene, rates of bone loss, or fracture risk), the data from multiple independent studies are consistent with the notion that the skeleton becomes relatively estrogen deficient in men (and perhaps also in women) at estradiol levels below approximately 20-30 pg/ml. However, direct studies assessing the dose-response relationship between bone turnover markers and serum estradiol levels in men are needed to test this.
As for urine NTX, we found that baseline estradiol levels were associated with the change in serum FSH levels after raloxifene therapy, with a similar (but nonsignificant) trend for changes in serum LH. However, the serum total estradiol level that predicted a zero change in serum FSH after raloxifene therapy (14 pg/ml) was considerably lower than the corresponding estradiol value at which the change in NTX excretion after raloxifene therapy was zero (26 pg/ml). Thus, virtually all the men had an increase in FSH secretion after raloxifene therapy. This suggests that even at low endogenous estradiol levels, raloxifene continues to function as an estrogen antagonist at the hypothalamic/pituitary level.
In contrast to studies in postmenopausal women,(28, 29) we did not observe any consistent changes in serum lipids in these men. Overall, raloxifene was well tolerated, with the only significant side effect being an increased incidence of hot flushes in the raloxifene-treated men compared with the placebo-treated men.
In summary, we found that raloxifene had no significant effect on bone turnover markers or lipid levels in elderly men. However, our findings do suggest that a subset of men with low estradiol levels may respond to raloxifene with a reduction in bone resorption. Additional studies comparing the effects of raloxifene or other SERMs in men with low versus normal or high estradiol levels are needed to directly test this hypothesis.
We are indebted to our study volunteers; the staff at the Mayo GCRC; Kenneth Mahoney, for assistance with recruitment of study volunteers; Celia Wright, R.N., for management of the study volunteers; and Sara Achenbach, for assistance with statistical analyses. This work was supported by grants AG-04875 and RR-00585 from the National Institutes of Health (NIH).