JBMR
ORIGINAL ARTICLE
Reproductive Hormones and Longitudinal Change in Bone Mineral Density and Incident Fracture Risk in Older Men: The Concord Health and Aging in Men Project Benjumin Hsu,1,2 Robert G Cumming,1,2,3 Markus J Seibel,3 Vasi Naganathan,2 Fiona M Blyth,2 Kerrin Bleicher,2 Aneesh Dave,2 David G Le Couteur,2 Louise M Waite,2 and David J Handelsman3 1
School of Public Health, University of Sydney, Sydney, New South Wales, Australia Centre of Education and Research on Ageing, University of Sydney and Concord Hospital, Sydney, New South Wales, Australia 3 ANZAC Research Institute, University of Sydney and Concord Hospital, Sydney, New South Wales, Australia 2
ABSTRACT The objectives of this study were to examine relationships between baseline levels of reproductive hormones in older men and (1) change in bone mineral density (BMD) over 5 years and (2) incident fractures over an average of 6 years’ follow-up. A total of 1705 men aged 70 years and older from the Concord Health and Ageing in Men Project (CHAMP) study were assessed at baseline (2005– 2007), 2 years follow-up (2007–2009), and 5 years follow-up (2010–2013). At baseline, testosterone (T), dihydrotestosterone (DHT), estradiol (E2), and estrone (E1) were measured by liquid chromatography–tandem mass spectrometry (LC-MS/MS), and sex hormone–binding globulin (SHBG), luteinizing hormone (LH), and follicle-stimulating hormone (FSH) by immunoassay. Hip BMD was measured by dual X-ray absorptiometry (DXA) at all three time-points. Fracture data were collected at 4-monthly phone calls and verified radiographically. Statistical modeling was by general estimating equations and Cox model regression. Univariate analyses revealed inverse associations for serum SHBG, FSH, and LH and positive association for E1 but not DHT or E2 with BMD loss at the hip across the three time points. Serum levels of SHBG (b ¼ –0.071), FSH (b ¼ –0.085), LH (b ¼ –0.070), and E1 (b ¼ 0.107) remained significantly associated with BMD loss in multivariate-adjusted models; however, we were unable to identify any thresholds for accelerated BMD loss according to reproductive steroids. Incident fractures (all, n ¼ 171; hip, n ¼ 44; and nonvertebral, n ¼ 139) were all significantly associated with serum SHBG, FSH, and LH levels in univariate models but none remained significantly associated in multivariate-adjusted model. Serum T, DHT, E2, and E1 levels were not associated with incident fractures in univariate or multivariate-adjusted analyses. In older men, lower serum SHBG, FSH, and LH and higher E1 levels protected against loss of BMD without increasing fracture rate. This means these reproductive variables may be considered as novel biomarkers of bone health during male aging. © 2015 American Society for Bone and Mineral Research. KEY WORDS: REPRODUCTIVE HORMONE; BONE MINERAL DENSITY; FRACTURE; AGING; EPIDEMIOLOGICAL STUDY
Introduction
M
ale aging is associated with a decline in serum sex hormones and other major reproductive hormones, and this decline may have adverse effects on bone.(1,2) Although there are firmly established relationships between reproductive hormones and bone health in women,(3) the relationship in men remain less clear. In women, the decline in estradiol (E2) levels after menopause is a major driver of postmenopausal bone loss and hence fracture risk.(4) The significance of skeletal metabolism of E2 in men and women may be similar,(5) although circulating E2 levels are much higher for decades in premenopausal women and then lower in postmenopausal women compared with men.(6) Therefore, the relationship between hormones and bone health between older men and women may differ.
Previous studies in men have found that the agerelated decline in serum testosterone (T) and E2 are associated with loss of bone mineral density (BMD).(3) However, the majority of these studies measured T and E2 by immunoassay and reported associations mostly for derived “free” or “bioavailable” fractions of T or E2(7–24) using calculations of uncertain validity.(25–27) Studies examining the associations between hormones and fracture risk in men have, however, been inconsistent—reporting either no association at all or increased fracture risk associated with low T, low E2, or high sex hormone–binding globulin (SHBG).(23,24,28–35) Most of the previous studies have mainly restricted analyses to T, E2, and SHBG, whereas the present study will further examine other major reproductive hormones: dihydrotestosterone (DHT), estrone (E1), follicle-stimulating hormone (FSH), and luteinizing
Received in original form December 14, 2014; revised form February 17, 2015; accepted February 26, 2015. Accepted manuscript online March 3, 2015. Address correspondence to: David J Handelsman, MBBS, PhD, FRACP, ANZAC Research Institute, Sydney, New South Wales, Australia 2139. E-mail: [email protected] Journal of Bone and Mineral Research, Vol. 30, No. 9, September 2015, pp 1701–1708 DOI: 10.1002/jbmr.2493 © 2015 American Society for Bone and Mineral Research
1701
hormone (LH). It is important to examine DHT because of its higher potency and affinity as an androgen due to its high binding and affinity to, and slower dissociation from, the androgen receptor;(36–38) whereas with E1, circulating at greater concentrations but biologically less potent than E2,(39) its role in bone health remains unknown in men, but recent studies have highlighted its unsuspected role in male reproductive health.(40–42) To our knowledge, this is the first study to examine the longitudinal associations between LH and FSH with incident fracture risk and bone loss in community-dwelling older men. The objectives of this study were to (1) examine associations between baseline levels of reproductive hormones and change in hip BMD over 5 years; (2) examine the association between baseline levels of reproductive hormones and incident fractures over an average 6-year follow-up; and (3) review the recent literature from cohort and nested case-control studies on reproductive hormones and fracture risk in older men. This study will provide a comprehensive review of most of the major male reproductive hormones (T, DHT, E2, E1, SHBG, FSH, and LH).
Subjects and Methods Study participants The Concord Health and Ageing in Men Project (CHAMP) is a longitudinal, observational study of the epidemiology of male aging conducted among men living within three local government areas (Burwood, Canada Bay, and Strathfield) surrounding Concord Hospital in Sydney, New South Wales, Australia.(43) Men were selected from the New South Wales electoral roll; enrollment is compulsory in Australia. Potential subjects were community-dwelling men aged at least 70 years, with no other inclusion or exclusion criteria. A total of 1705 subjects were enrolled in the CHAMP study. The study design has been reported in detail elsewhere.(43) Baseline measurements were conducted between January 2005 and June 2007. Data were collected using self-reported questionnaires, interviewer-administered questionnaires, and a wide range of clinical assessments. Follow-up assessments were conducted between January 2007 and October 2009 for 2-year follow-up, and August 2010 and July 2013 for the 5-year follow-up, with identical measurements as at baseline.
Reproductive hormone measurement Subjects had an early morning fasting blood sample taken, with serum stored at –80°C until assay. Measurements of serum T, DHT, E2, and E1 were by liquid chromatography–tandem mass spectrometry (LC-MS/MS) as validated elsewhere.(44) Serum LH, FSH, and SHBG were measured by automated immunoassays (Roche Diagnostics Australia, Dee Why, Australia) with coefficients of variation between of 1.0% and 2.0%. The calculated free testosterone (cFT) levels were computed using an empirical formula recently validated in two large data sets consisting of more than 6000 blood samples.(25,45)
BMD measurement Whole-body and hip scans were made with dual xray absorptiometry (DXA) using Hologic Discovery-W scanner (Hologic Inc., Bedford, MA, USA). Men removed jewelry and wore light cotton gowns free from metal. The same DXA scanner was used for all scans. The coefficient of variation (CV) for scans duplicated on 30 men from the study cohort were 1.6% for the
1702
HSU ET AL.
total hip. The quality control scans were conducted daily using the Hologic whole-body phantom and indicated no shifts or drifts. Neither spinal nor whole-body BMD were included in the analyses because age-related osteoarthritis creates difficulty in interpreting spine BMD in older men.
Assessment of incident fracture Following the baseline assessment, men were contacted by telephone every 4 months to ascertain any incident fractures. Phone calls were made up to January 2014. If a fracture was reported, radiology reports were obtained either from the participant, or from hospital medical records and radiology practices. Additional manual searching for fractures were conducted for the men’s medical records within our health district. Only fractures confirmed by radiographic reports were included in the present analysis. Pathological fractures and fractures of hands, fingers, feet, toes, and the skull were excluded. Only the first incident fractures that met the inclusion criteria were included, regardless of trauma level or any additional subsequent fractures reported.(46,47) Time to censorship was either date of death, date of official withdrawal from the study or date of the last telephone contact. The date of first fracture was the date on the radiology report.
Statistical analysis Of the 1705 men, a total of 1657 men were included in the final analyses. Twenty men were excluded from analyses in this work because they were either on androgens or antiandrogen treatments and 28 men were excluded due to missing data. Descriptive baseline characteristics were generated for the analytic sample. The associations between baseline reproductive hormones and longitudinal change in BMD across baseline, 2-year follow-up, and 5-year follow-up were assessed by generalized estimating equations (GEEs) with exchangeable working correlation and robust variance estimator. The GEE method is known to be robust when treating missing data in longitudinal data.(48) The Cox proportional hazards model was used to assess associations between reproductive hormones at baseline, and incident hip fractures, nonvertebral fractures, and all fractures over an average 6-year follow-up. Reproductive hormones were fitted as continuous variables with results expressed in terms of a 1-SD decrease in hormone levels. In order to identify potential threshold effects of reproductive hormones on BMD loss, we investigated scatter plots, linear and nonlinear regression, and receiver operating characteristic (ROC) analysis of BMD by all reproductive hormones as suggested by previous reports of a threshold effect of serum E2 for BMD.(18) For the ROC analysis, we defined an accelerated loss of BMD as at least 1 SD below the annualized mean BMD loss, and for each hormone a series of potential hormonal cut points was created. In addition, we sought to further investigate a potential androgen threshold by accounting for net androgen effects, which required considering both T together with its more potent metabolite, DHT. Hence we defined a new variable, androgen (A), as the sum of serum testosterone plus 5 times serum DHT. This reflects the fact that DHT has twofold to 10fold higher intrinsic androgenic potency than T(36) because of its higher affinity and slower dissociation from the androgen receptor.(38) The potential cut points for serum E2 were based on previous literature(18) and for T, DHT, and A, they were based on covering a range of centiles from 1% to 75% of reference ranges derived from pooled population-based studies in three
Journal of Bone and Mineral Research
Australian cities comprising over 10,000 serum samples (Handelsman DJ, unpublished work). For ROC analysis of each hormone, we created five equally spaced cut points ranging from 20 to 40 pg/mL for E2, 1.0 to 4.2 ng/mL for T, 0.05 to 0.35 ng/mL for DHT, and 1.4 to 6.0 ng/ml for A. The model building for all analyses included known relevant covariates based on previous literatures: age, BMI, smoking status, physical activity, and number of comorbidities. Models were fitted using SPSS software version 20 (IBM Corp., Armonk, NY, USA) and SAS software 9.3 (SAS Institute Inc., Cary, NC, USA).
Results The baseline characteristics of the analytic sample are shown in Table 1. Study participants had a mean age of 76.9 5.5 years (range, 70 to 97 years) with a mean BMI of 27.8 4.0 kg/m2. Hip BMD at baseline was 0.94 0.1 g/cm2, with a mean annualized percentage loss of hip BMD of 0.4%/year (0.0038 0.008 g/cm2/ year, n ¼ 901) from baseline to 5-year follow-up (data not shown). One hundred and seventy-one (10%) men had at least one fracture over a mean follow-up time of 6 years, with 32 (2%) men having clinical vertebral fractures and 139 (8%) having nonvertebral fractures, of which 44 men (3%) had a hip fracture. When comparing men at baseline to men lost to follow-up at 2 years, there were no difference in the mean levels of all the studied reproductive hormones (data not shown). Men lost to follow-up at 2 years were significantly older (79 versus 78 years), but there were no differences in the number of comorbidities
and baseline hip BMD levels. However, when comparing men at baseline to men lost to follow-up at either 2 years or 5 years, the men lost to follow-up were significantly older (79 versus 75 years), had lower BMI (27.5 versus 28.0 kg/m2), more comorbidities (3 versus 2), and had lower hip BMD (0.92 versus 0.96 g/ cm2). The reasons for loss to follow-up over the 5-year followup were death (n ¼ 537) and other reasons (n ¼ 210).
Hormones and BMD change The associations between baseline reproductive hormones and longitudinal change in hip BMD are shown in Table 2. Univariate analyses revealed significant inverse associations between longitudinal change in hip BMD for SHBG (b ¼ –0.204, p < 0.001), FSH (b ¼ –0.127, p < 0.001), and LH (b ¼ –0.129, p < 0.001), and a positive association for E1 (b ¼ 0.113, p < 0.001). Similarly SHBG (b ¼ –0.076, p ¼ 0.002), FSH (b ¼ –0.077, p ¼ 0.005), and LH (b ¼ –0.061, p ¼ 0.04) remained inversely associated, and E1 (b ¼ 0.099, p < 0.001) remained positively associated with change in hip BMD in multivariate analyses. This means higher SHBG, FSH, or LH levels, or lower E1 levels at baseline were associated with greater bone loss at the hip over 5 years. Neither univariate nor multivariate-adjusted analysis revealed any statistically significant associations between baseline T, DHT, cFT, and E2 and change in hip BMD. In subanalyses stratified by BMI status (normal: BMI < 25 kg/m2, overweight: 25 kg/m2 BMI < 30 kg/m2, and obese: BMI 30 kg/m2), the findings revealed similar relationships (data not shown).
Table 1. Characteristics of Study Participants at Baseline, 2-Year-Follow-Up, and 5-Year-Follow-Up Baseline Age (years) BMI (kg/m2) PASE Current smoker Comorbidities None One Two Three Four or more T (ng/mL) DHT (ng/mL) SHBG (nmol/L) E2 (pg/mL) E1 (pg/mL) FSH (IU/L) LH (IU/L) cFT (pg/mL) BMD (g/cm2) Total body Hip Femoral neck Fractures All sites Hip Nonvertebral
2-year follow-up
5-year follow-up
5.5 4.0 62.1 (6%)
79.5 5.3 27.3 4.0 119.7 (59.7) 52 (4%)
163 (10%) 353 (21%) 414 (25%) 308 (25%) 430 (19%) 4.3 1.9 0.4 0.2 50.1 20.7 25.3 12.4 40.4 16.3 14.7 14.9 9.6 8.8 17.2 6.5
131 (10%) 291 (22%) 305 (23%) 278 (21%) 343 (24%) 4.2 1.9 0.4 0.2 52.4 21.1 24.2 9.8 40.1 1.5 14.6 14.3 9.7 8.2 16.7 6.7
95 (10%) 209 (22%) 198 (21%) 222 (23%) 226 (24%) 3.4 1.8 0.3 0.2 57.4 24.4 36.2 15.4 29.3 12.1 16.6 15.7 11.0 9.1 13.4 6.4
1.04 0.1 0.94 0.1 0.76 0.1
1.04 0.2 0.93 0.1 0.76 0.1
1.06 0.1 0.93 0.1 0.76 0.1
76.9 27.8 124.4 101
81.4 4.6 27.0 3.8 118.4 (57.1) 40 (4%)
171 (10%) 44 (3%) 139 (8%)
Values are mean SD or n (%) as indicated. PASE ¼ physical activity scale for the elderly.
Journal of Bone and Mineral Research
REPRODUCTIVE HORMONES, CHANGE IN BMD, FRACTURE RISK IN OLDER MEN
1703
Table 2. Associations Between Baseline Reproductive Hormones and Longitudinal Change in Hip BMD Hip BMD Multivariatea
Univariate
T DHT SHBG E2 E1 FSH LH cFT
b
p
b
p
0.049 0.026 0.204 0.073 0.113 0.127 0.129 0.00.2
0.06 0.6
ORIGINAL ARTICLE
Reproductive Hormones and Longitudinal Change in Bone Mineral Density and Incident Fracture Risk in Older Men: The Concord Health and Aging in Men Project Benjumin Hsu,1,2 Robert G Cumming,1,2,3 Markus J Seibel,3 Vasi Naganathan,2 Fiona M Blyth,2 Kerrin Bleicher,2 Aneesh Dave,2 David G Le Couteur,2 Louise M Waite,2 and David J Handelsman3 1
School of Public Health, University of Sydney, Sydney, New South Wales, Australia Centre of Education and Research on Ageing, University of Sydney and Concord Hospital, Sydney, New South Wales, Australia 3 ANZAC Research Institute, University of Sydney and Concord Hospital, Sydney, New South Wales, Australia 2
ABSTRACT The objectives of this study were to examine relationships between baseline levels of reproductive hormones in older men and (1) change in bone mineral density (BMD) over 5 years and (2) incident fractures over an average of 6 years’ follow-up. A total of 1705 men aged 70 years and older from the Concord Health and Ageing in Men Project (CHAMP) study were assessed at baseline (2005– 2007), 2 years follow-up (2007–2009), and 5 years follow-up (2010–2013). At baseline, testosterone (T), dihydrotestosterone (DHT), estradiol (E2), and estrone (E1) were measured by liquid chromatography–tandem mass spectrometry (LC-MS/MS), and sex hormone–binding globulin (SHBG), luteinizing hormone (LH), and follicle-stimulating hormone (FSH) by immunoassay. Hip BMD was measured by dual X-ray absorptiometry (DXA) at all three time-points. Fracture data were collected at 4-monthly phone calls and verified radiographically. Statistical modeling was by general estimating equations and Cox model regression. Univariate analyses revealed inverse associations for serum SHBG, FSH, and LH and positive association for E1 but not DHT or E2 with BMD loss at the hip across the three time points. Serum levels of SHBG (b ¼ –0.071), FSH (b ¼ –0.085), LH (b ¼ –0.070), and E1 (b ¼ 0.107) remained significantly associated with BMD loss in multivariate-adjusted models; however, we were unable to identify any thresholds for accelerated BMD loss according to reproductive steroids. Incident fractures (all, n ¼ 171; hip, n ¼ 44; and nonvertebral, n ¼ 139) were all significantly associated with serum SHBG, FSH, and LH levels in univariate models but none remained significantly associated in multivariate-adjusted model. Serum T, DHT, E2, and E1 levels were not associated with incident fractures in univariate or multivariate-adjusted analyses. In older men, lower serum SHBG, FSH, and LH and higher E1 levels protected against loss of BMD without increasing fracture rate. This means these reproductive variables may be considered as novel biomarkers of bone health during male aging. © 2015 American Society for Bone and Mineral Research. KEY WORDS: REPRODUCTIVE HORMONE; BONE MINERAL DENSITY; FRACTURE; AGING; EPIDEMIOLOGICAL STUDY
Introduction
M
ale aging is associated with a decline in serum sex hormones and other major reproductive hormones, and this decline may have adverse effects on bone.(1,2) Although there are firmly established relationships between reproductive hormones and bone health in women,(3) the relationship in men remain less clear. In women, the decline in estradiol (E2) levels after menopause is a major driver of postmenopausal bone loss and hence fracture risk.(4) The significance of skeletal metabolism of E2 in men and women may be similar,(5) although circulating E2 levels are much higher for decades in premenopausal women and then lower in postmenopausal women compared with men.(6) Therefore, the relationship between hormones and bone health between older men and women may differ.
Previous studies in men have found that the agerelated decline in serum testosterone (T) and E2 are associated with loss of bone mineral density (BMD).(3) However, the majority of these studies measured T and E2 by immunoassay and reported associations mostly for derived “free” or “bioavailable” fractions of T or E2(7–24) using calculations of uncertain validity.(25–27) Studies examining the associations between hormones and fracture risk in men have, however, been inconsistent—reporting either no association at all or increased fracture risk associated with low T, low E2, or high sex hormone–binding globulin (SHBG).(23,24,28–35) Most of the previous studies have mainly restricted analyses to T, E2, and SHBG, whereas the present study will further examine other major reproductive hormones: dihydrotestosterone (DHT), estrone (E1), follicle-stimulating hormone (FSH), and luteinizing
Received in original form December 14, 2014; revised form February 17, 2015; accepted February 26, 2015. Accepted manuscript online March 3, 2015. Address correspondence to: David J Handelsman, MBBS, PhD, FRACP, ANZAC Research Institute, Sydney, New South Wales, Australia 2139. E-mail: [email protected] Journal of Bone and Mineral Research, Vol. 30, No. 9, September 2015, pp 1701–1708 DOI: 10.1002/jbmr.2493 © 2015 American Society for Bone and Mineral Research
1701
hormone (LH). It is important to examine DHT because of its higher potency and affinity as an androgen due to its high binding and affinity to, and slower dissociation from, the androgen receptor;(36–38) whereas with E1, circulating at greater concentrations but biologically less potent than E2,(39) its role in bone health remains unknown in men, but recent studies have highlighted its unsuspected role in male reproductive health.(40–42) To our knowledge, this is the first study to examine the longitudinal associations between LH and FSH with incident fracture risk and bone loss in community-dwelling older men. The objectives of this study were to (1) examine associations between baseline levels of reproductive hormones and change in hip BMD over 5 years; (2) examine the association between baseline levels of reproductive hormones and incident fractures over an average 6-year follow-up; and (3) review the recent literature from cohort and nested case-control studies on reproductive hormones and fracture risk in older men. This study will provide a comprehensive review of most of the major male reproductive hormones (T, DHT, E2, E1, SHBG, FSH, and LH).
Subjects and Methods Study participants The Concord Health and Ageing in Men Project (CHAMP) is a longitudinal, observational study of the epidemiology of male aging conducted among men living within three local government areas (Burwood, Canada Bay, and Strathfield) surrounding Concord Hospital in Sydney, New South Wales, Australia.(43) Men were selected from the New South Wales electoral roll; enrollment is compulsory in Australia. Potential subjects were community-dwelling men aged at least 70 years, with no other inclusion or exclusion criteria. A total of 1705 subjects were enrolled in the CHAMP study. The study design has been reported in detail elsewhere.(43) Baseline measurements were conducted between January 2005 and June 2007. Data were collected using self-reported questionnaires, interviewer-administered questionnaires, and a wide range of clinical assessments. Follow-up assessments were conducted between January 2007 and October 2009 for 2-year follow-up, and August 2010 and July 2013 for the 5-year follow-up, with identical measurements as at baseline.
Reproductive hormone measurement Subjects had an early morning fasting blood sample taken, with serum stored at –80°C until assay. Measurements of serum T, DHT, E2, and E1 were by liquid chromatography–tandem mass spectrometry (LC-MS/MS) as validated elsewhere.(44) Serum LH, FSH, and SHBG were measured by automated immunoassays (Roche Diagnostics Australia, Dee Why, Australia) with coefficients of variation between of 1.0% and 2.0%. The calculated free testosterone (cFT) levels were computed using an empirical formula recently validated in two large data sets consisting of more than 6000 blood samples.(25,45)
BMD measurement Whole-body and hip scans were made with dual xray absorptiometry (DXA) using Hologic Discovery-W scanner (Hologic Inc., Bedford, MA, USA). Men removed jewelry and wore light cotton gowns free from metal. The same DXA scanner was used for all scans. The coefficient of variation (CV) for scans duplicated on 30 men from the study cohort were 1.6% for the
1702
HSU ET AL.
total hip. The quality control scans were conducted daily using the Hologic whole-body phantom and indicated no shifts or drifts. Neither spinal nor whole-body BMD were included in the analyses because age-related osteoarthritis creates difficulty in interpreting spine BMD in older men.
Assessment of incident fracture Following the baseline assessment, men were contacted by telephone every 4 months to ascertain any incident fractures. Phone calls were made up to January 2014. If a fracture was reported, radiology reports were obtained either from the participant, or from hospital medical records and radiology practices. Additional manual searching for fractures were conducted for the men’s medical records within our health district. Only fractures confirmed by radiographic reports were included in the present analysis. Pathological fractures and fractures of hands, fingers, feet, toes, and the skull were excluded. Only the first incident fractures that met the inclusion criteria were included, regardless of trauma level or any additional subsequent fractures reported.(46,47) Time to censorship was either date of death, date of official withdrawal from the study or date of the last telephone contact. The date of first fracture was the date on the radiology report.
Statistical analysis Of the 1705 men, a total of 1657 men were included in the final analyses. Twenty men were excluded from analyses in this work because they were either on androgens or antiandrogen treatments and 28 men were excluded due to missing data. Descriptive baseline characteristics were generated for the analytic sample. The associations between baseline reproductive hormones and longitudinal change in BMD across baseline, 2-year follow-up, and 5-year follow-up were assessed by generalized estimating equations (GEEs) with exchangeable working correlation and robust variance estimator. The GEE method is known to be robust when treating missing data in longitudinal data.(48) The Cox proportional hazards model was used to assess associations between reproductive hormones at baseline, and incident hip fractures, nonvertebral fractures, and all fractures over an average 6-year follow-up. Reproductive hormones were fitted as continuous variables with results expressed in terms of a 1-SD decrease in hormone levels. In order to identify potential threshold effects of reproductive hormones on BMD loss, we investigated scatter plots, linear and nonlinear regression, and receiver operating characteristic (ROC) analysis of BMD by all reproductive hormones as suggested by previous reports of a threshold effect of serum E2 for BMD.(18) For the ROC analysis, we defined an accelerated loss of BMD as at least 1 SD below the annualized mean BMD loss, and for each hormone a series of potential hormonal cut points was created. In addition, we sought to further investigate a potential androgen threshold by accounting for net androgen effects, which required considering both T together with its more potent metabolite, DHT. Hence we defined a new variable, androgen (A), as the sum of serum testosterone plus 5 times serum DHT. This reflects the fact that DHT has twofold to 10fold higher intrinsic androgenic potency than T(36) because of its higher affinity and slower dissociation from the androgen receptor.(38) The potential cut points for serum E2 were based on previous literature(18) and for T, DHT, and A, they were based on covering a range of centiles from 1% to 75% of reference ranges derived from pooled population-based studies in three
Journal of Bone and Mineral Research
Australian cities comprising over 10,000 serum samples (Handelsman DJ, unpublished work). For ROC analysis of each hormone, we created five equally spaced cut points ranging from 20 to 40 pg/mL for E2, 1.0 to 4.2 ng/mL for T, 0.05 to 0.35 ng/mL for DHT, and 1.4 to 6.0 ng/ml for A. The model building for all analyses included known relevant covariates based on previous literatures: age, BMI, smoking status, physical activity, and number of comorbidities. Models were fitted using SPSS software version 20 (IBM Corp., Armonk, NY, USA) and SAS software 9.3 (SAS Institute Inc., Cary, NC, USA).
Results The baseline characteristics of the analytic sample are shown in Table 1. Study participants had a mean age of 76.9 5.5 years (range, 70 to 97 years) with a mean BMI of 27.8 4.0 kg/m2. Hip BMD at baseline was 0.94 0.1 g/cm2, with a mean annualized percentage loss of hip BMD of 0.4%/year (0.0038 0.008 g/cm2/ year, n ¼ 901) from baseline to 5-year follow-up (data not shown). One hundred and seventy-one (10%) men had at least one fracture over a mean follow-up time of 6 years, with 32 (2%) men having clinical vertebral fractures and 139 (8%) having nonvertebral fractures, of which 44 men (3%) had a hip fracture. When comparing men at baseline to men lost to follow-up at 2 years, there were no difference in the mean levels of all the studied reproductive hormones (data not shown). Men lost to follow-up at 2 years were significantly older (79 versus 78 years), but there were no differences in the number of comorbidities
and baseline hip BMD levels. However, when comparing men at baseline to men lost to follow-up at either 2 years or 5 years, the men lost to follow-up were significantly older (79 versus 75 years), had lower BMI (27.5 versus 28.0 kg/m2), more comorbidities (3 versus 2), and had lower hip BMD (0.92 versus 0.96 g/ cm2). The reasons for loss to follow-up over the 5-year followup were death (n ¼ 537) and other reasons (n ¼ 210).
Hormones and BMD change The associations between baseline reproductive hormones and longitudinal change in hip BMD are shown in Table 2. Univariate analyses revealed significant inverse associations between longitudinal change in hip BMD for SHBG (b ¼ –0.204, p < 0.001), FSH (b ¼ –0.127, p < 0.001), and LH (b ¼ –0.129, p < 0.001), and a positive association for E1 (b ¼ 0.113, p < 0.001). Similarly SHBG (b ¼ –0.076, p ¼ 0.002), FSH (b ¼ –0.077, p ¼ 0.005), and LH (b ¼ –0.061, p ¼ 0.04) remained inversely associated, and E1 (b ¼ 0.099, p < 0.001) remained positively associated with change in hip BMD in multivariate analyses. This means higher SHBG, FSH, or LH levels, or lower E1 levels at baseline were associated with greater bone loss at the hip over 5 years. Neither univariate nor multivariate-adjusted analysis revealed any statistically significant associations between baseline T, DHT, cFT, and E2 and change in hip BMD. In subanalyses stratified by BMI status (normal: BMI < 25 kg/m2, overweight: 25 kg/m2 BMI < 30 kg/m2, and obese: BMI 30 kg/m2), the findings revealed similar relationships (data not shown).
Table 1. Characteristics of Study Participants at Baseline, 2-Year-Follow-Up, and 5-Year-Follow-Up Baseline Age (years) BMI (kg/m2) PASE Current smoker Comorbidities None One Two Three Four or more T (ng/mL) DHT (ng/mL) SHBG (nmol/L) E2 (pg/mL) E1 (pg/mL) FSH (IU/L) LH (IU/L) cFT (pg/mL) BMD (g/cm2) Total body Hip Femoral neck Fractures All sites Hip Nonvertebral
2-year follow-up
5-year follow-up
5.5 4.0 62.1 (6%)
79.5 5.3 27.3 4.0 119.7 (59.7) 52 (4%)
163 (10%) 353 (21%) 414 (25%) 308 (25%) 430 (19%) 4.3 1.9 0.4 0.2 50.1 20.7 25.3 12.4 40.4 16.3 14.7 14.9 9.6 8.8 17.2 6.5
131 (10%) 291 (22%) 305 (23%) 278 (21%) 343 (24%) 4.2 1.9 0.4 0.2 52.4 21.1 24.2 9.8 40.1 1.5 14.6 14.3 9.7 8.2 16.7 6.7
95 (10%) 209 (22%) 198 (21%) 222 (23%) 226 (24%) 3.4 1.8 0.3 0.2 57.4 24.4 36.2 15.4 29.3 12.1 16.6 15.7 11.0 9.1 13.4 6.4
1.04 0.1 0.94 0.1 0.76 0.1
1.04 0.2 0.93 0.1 0.76 0.1
1.06 0.1 0.93 0.1 0.76 0.1
76.9 27.8 124.4 101
81.4 4.6 27.0 3.8 118.4 (57.1) 40 (4%)
171 (10%) 44 (3%) 139 (8%)
Values are mean SD or n (%) as indicated. PASE ¼ physical activity scale for the elderly.
Journal of Bone and Mineral Research
REPRODUCTIVE HORMONES, CHANGE IN BMD, FRACTURE RISK IN OLDER MEN
1703
Table 2. Associations Between Baseline Reproductive Hormones and Longitudinal Change in Hip BMD Hip BMD Multivariatea
Univariate
T DHT SHBG E2 E1 FSH LH cFT
b
p
b
p
0.049 0.026 0.204 0.073 0.113 0.127 0.129 0.00.2
0.06 0.6
