Abstract
Background. Current data on late-onset hypogonadism, derived from healthy males in epidemiological studies, may not reflect the profile of men seen in actual clinical practice.
Objective. To examine androgen levels in relation to metabolic status and quality of life (QOL) measures in self-referred men at a hospital-based Men's Health clinic.
Methods. Cross-sectional study of 238 consecutive Asian males. Fasting total testosterone (TT), sex-hormone binding globulin (SHBG), luteinising (LH) and follicle stimulating (FSH) hormones, glucose (FPG) and lipid profile were measured. Bioavailable (cBT) and free testosterone (cFT) were calculated. Waist circumference (WC) and body mass index (BMI) were collected. Subjects also answered the modified International Index of Erectile Dysfunction (IIEF-5) and Ageing Male Symptom (AMS) questionnaires.
Results. Among non-diabetic males (N = 201), no change was noted for TT, although SHBG and gonadotrophins rose, while cBT and cFT declined, significantly with age. Sex hormones were negatively related with WC, BMI and FPG. SHBG displayed a stronger association with metabolic components than testosterone. Testosterone was not related to lipids, IIEF-5 or AMS scores. WC, not BMI, was a key determinant of TT, cBT and cFT in younger subjects, while FSH seemed a more sensitive indicator of primary hypogonadism than LH in older males.
Conclusion. The preferred measures of serum testosterone in older men are cBT and cFT. Visceral adiposity and SHBG, rather than testosterone, appeared to be the link between androgen deficiency and poorer metabolic status. QOL scores correlate poorly with androgen concentrations.
Introduction
It is well established that serum testosterone concentrations decline steadily by 1–2% per year in men with increasing age, and that this phenomenon continues even in advanced age [Citation1,Citation2]. This age-related relative androgen deficiency, also known as late-onset hypogonadism (LOH), has been implicated as a contributory cause of various sexual complaints and psychosomatic symptoms experienced by older men viz. reduced libido, erectile dysfunction, low mood and energy levels, lack of motivation and poorer work performance, among others. Apart from these quality of life (QOL) issues, LOH has also been reported to correlate with obesity, poorer metabolic status, and may even predispose to incident type 2 diabetes mellitus [Citation3,Citation4].
Although research activity in the area of LOH has intensified in the last decade, LOH is still shrouded in significant controversy. Much of this is related to the absence of any consensus on assessing androgen status in older men. Established age-specific reference ranges for total testosterone (TT) and free testosterone are lacking, screening questionnaires possess limited sensitivity and specificity [Citation5], androgen levels may not reflect symptomatology and the presence of co-morbidities as well as usage of multiple prescription drugs are known to adversely affect QOL and testosterone levels [Citation6].
Most of the current literature on LOH is derived from healthy male subjects in epidemiological studies, which, by their nature, may not reflect actual clinical practice where, with increasing public awareness, more men are seeking medical attention for perceived symptoms of LOH. These men are more likely to suffer from at least one chronic condition and to be using prescription drugs. The objective of this article is to examine androgen levels in relation to metabolic status in self-referred men, as opposed to a population study, and the usefulness of existing QOL measures in assessing androgen status in these males. We feel it would be interesting to study how such observational data from actual practice would differ from knowledge gained from population studies.
Methods
This is a cross-sectional study where data was obtained from 262 consecutive self-referred males who were seen for the first time at a hospital-based Men's Health clinic from August 2006 to April 2009 in Singapore. Most had sought consultation, on their own accord, for QOL issues such as decreased libido, erectile dysfunction (ED), perceived lack of energy and drive, mood changes and sleep disorders, among others. Others had turned up requesting for health screening, without any specific complaints. Fifteen were excluded because of incomplete data, as were eight non-Asians and one subject who had significant chronic renal failure, leaving a final sample size of 238 Asian (predominantly Chinese) men.
Each person arrived at 09:00 h after an overnight fast. Blood samples were obtained for TT, SHBG, luteinising hormone (LH), follicle stimulating hormone (FSH), prostate specific antigen (PSA), fasting plasma glucose (FPG), total cholesterol (TC), high-density lipoprotein (HDL) cholesterol, triglycerides (TG), electrolytes, creatinine (Cr) and albumin. Systolic (SBP) and diastolic (DBP) blood pressures (mmHg), waist circumference (WC; cm), height (m) and weight (kg) were measured, and body mass index (BMI; kg/m2) calculated. Past medical history and usage of long-term prescription drugs, if any, were obtained. Data on smoking, alcohol consumption and regular exercise were categorically scored as either ‘Yes’ or ‘No.’ Subjects who had quit smoking for less than 12 months were still classified as smokers. A subject was considered to have exercised regularly if he performed any aerobic activity of at least 30 min duration thrice weekly. Regular alcohol consumption, even in moderation not exceeding 21 units per week, was considered positive.
In addition, each individual answered the modified International Index of ED (IIEF-5) Ageing Male Symptom (AMS) questionnaires. Scoring for the IIEF-5 questionnaire ranged from a minimum of 5 to a maximum of 25. Responses were graded as no ED if the score was 22–25, mild ED (17–21), mild-to-moderate ED (12–16), moderate ED (8–11) or severe ED (5–7) [Citation7]. The AMS questionnaire [Citation8] grades QOL in relation to androgen deficiency through three subscales: sexual, psychological and somato-vegetative, with a minimum total score of 17 and maximum of 85. Responses were graded as normal if the total score was below 27, mild (27–36), moderate (37–49) or severe (≥50).
Electrochemiluminescence (by Roche) technique was used for measurement of the following: TT (coefficients of variation (CV) 2% at 3.17 nM and 3% at 17.64 nM); LH (CV 2% at both 4.77 and 20.29 mIU/ml); FSH (CV 2% at both 7.66 mIU/ml and 17.99 mIU/m); SHBG (CV 2% at 20.46 nM and 2.5% at 38.86 nM) and PSA (CV 2% at 3.64 ng/ml). Electrolytes were measured by indirect ion selective electrode analysers (by Beckman Coulter). Serum albumin was measured using bromcresol purple dye-binding (Beckman Coulter). All cholesterol components (TC, TG and HDL) were measured by enzymatic colorimetry (Beckman Coulter). Low-density lipoprotein (LDL) cholesterol was calculated using the Friedewald equation. Bioavailable (cBT) and free testosterone (cFT) were calculated using formula validated by Vermeulen [Citation9].
All statistical analyses were performed using SPSS version 16.0 software. Subjects were labeled as having type 2 diabetes mellitus (T2DM) either through self-reporting or detected through an elevated FPG of 7.0 mM or more. Non-diabetic subjects were arbitrarily divided into age groups (aged below 50 years, 50–59, or 60 and above). Values of TT, SHBG, cBT, cFT, LH, FSH and PSA were log-transformed before analysis. Simple comparison of means of variables between two groups was performed using the t-test, while comparison across the three age groups was performed using one-way ANOVA, with 95% confidence intervals shown. Data are reported as mean (±SD); values of TT, SHBG, cBT, cFT, LH, FSH and PSA are reported in both their logarithmic and inverse-logarithmic values. Correlation between variables was analysed using Spearman non-parametric correlation coefficients. Stepwise multiple linear regression was used to determine the independent determinants of TT, cFT and SHBG in non-diabetic subjects, arbitrarily separated into ‘younger’ males (aged <50 years) and ‘older’ males (aged 60 years and above). Variables entered were age, WC, BMI, SBP, DBP, LH, FSH, Cr and FPG. A two-tailed P-value of <0.05 was considered statistically significant. This study was approved by the institutional review board of Singapore Health Services (SingHealth).
Results
shows the characteristics of the measured variables in the entire cohort of 238 subjects, as well as between diabetic (N = 37) and non-diabetic (N = 201) subjects. WC and BMI were significantly higher in the diabetic group, while values of TT and SHBG were significantly lower. There was a trend towards lower cBT and cFT in diabetic men which just missed statistical significance. As it is clear that the inclusion of diabetic subjects may significantly confound the associations between variables, especially of androgens, they have been excluded in subsequent data analyses.
Table I. Characteristics of the measured variables in the whole cohort, as well as between subjects with and without diabetes.
compares the characteristics of the measured variables across different age groups amongst the non-diabetic subjects. No significant change in TT was noted between groups, although levels of SHBG, LH and FSH rose, while levels of cBT and cFT declined, significantly with increasing age. Levels of PSA also increased significantly with age, but did not exceed the upper limit of normal. Mean scores of IIEF-5 decreased significantly with age. Total AMS scores, as well as those of the psychological and somato-vegetative AMS subscales, were significantly lower in the oldest age group. Significant fluctuations in lipid components were noted between age groups, which is likely the confounding effect of a substantial number of subjects being on prescription medication.
Table II. Characteristics of the measured variables across different age groups in non-diabetic subjects.
Amongst non-diabetic men, no association was noted between TT and age (r = 0.05, P = 0.49), although both cBT (r = −0.23, P = 0.001) and cFT (r = −0.23, P = 0.001) correlated negatively with age. Highly significant correlations with age were noted for SHBG (r = 0.32, P <0.001), LH (r = 0.18, P = 0.013), FSH (r = 0.31, P <0.001), PSA (r = 0.30, P <0.001) as well as SBP (r = 0.23, P = 0.002) and FPG (r = 0.15, P = 0.033). QOL scores correlated negatively with age, particularly for IIEF-5 (r = −0.21, P = 0.004), AMS total score (r = −0.15, P = 0.033), AMS psychological (r = −0.19, P = 0.009) and AMS somato-vegetative (r = −0.15, P = 0.034) subscales. No significant associations with age were noted for WC (r = −0.06, P = 0.39) and BMI (r = −0.14, P = 0.059).
shows the correlation coefficients between sex hormones (log-transformed) and other measured variables in non-diabetic subjects, controlled for age. TT, SHBG, remained negatively related with WC and BMI. FPG correlated negatively with TT and SHBG, but not with cBT, nor cFT. Both LH and FSH correlated positively with SHBG. Only FSH, but not LH, showed a significant negative relationship with cBT and cFT. No relationship was noted between TT and either LH or FSH. Both SBP and DBP correlated negatively with SHBG, but not testosterone. There was no significant relationship between testosterone (TT, cBT and cFT) and any of the QOL scores, as well as PSA.
Table III. Coefficient correlations, controlled for age, between sex hormones and measured variables in non-diabetic subjects.
After controlling for both age and SHBG (not shown), the negative correlations of all measures of testosterone with WC, BMI and FSH remained strongly significant. In addition, a significant negative association between FSH and TT became evident (r = −0.26, P = 0.001). The observed associations between testosterone and FPG, and with LH, were lost.
In a subgroup analysis of 117 subjects (controlled for age, not shown) who were not taking any prescription medication, the only significant associations were of SHBG with HDL (r = 0.26, P = 0.007) and FPG (r = −0.27, P = 0.006). There was no correlation between testosterone, or SHBG, and any of the lipid components, nor blood pressure.
Stepwise regression analysis revealed WC to be the principal independent determinant of TT, with WC and LH being responsible in younger subjects (age <50), while WC and DBP accounted in older subjects (aged ≥60) ().
Table IV. Stepwise regression showing the determinants of total testosterone (TT) and sex hormone binding globulin (SHBG) in the non-diabetic cohort: overall, and in younger and older subjects.
Age, BMI, FPG and LH were the independent determinants of SHBG, with LH and WC being predictive for younger subjects, while age alone was predictive of SHBG in older subjects ().
The independent determinants for cBT and cFT were similar, principally being WC, FSH and age. WC, and FSH, was the main predictor in younger and older subjects, respectively ().
Table V. Stepwise regression showing the determinants of calculated bioavailable testosterone (cBT) and calculated free testosterone (cFT) in the non-diabetic cohort: overall, and in younger and older subjects.
shows the effect of smoking, exercise and medication on androgens and QOL measures in non-diabetic subjects. Levels of TT, cBT and cFT were higher among smokers, but the differences were not statistically significant. Subjects who exercised had significantly lower AMS scores, as well as a trend towards higher androgen levels, with cFT being significantly higher. Conversely, subjects on medication had significantly lower levels of TT and cFT.
Table VI. Effect of smoking, exercise and medication on androgen and quality of life parameters in non-diabetic subjects.
Discussion
We report the novel observations of WC and FSH as significant contributors to serum testosterone concentrations, independently of age. WC, but not BMI, was strongly associated with TT, as well as cBT and cFT, among younger subjects. On the other hand, FSH, but not LH, was the sole independent predictor of cBT and cFT amongst older subjects. Clinical markers of metabolic status, such as SBP, DBP, WC and BMI, were significantly related to SHBG, as was LH, which was an independent predictor of SHBG in younger subjects. There was a notable lack of association between testosterone and QOL scores.
Most of the current literature on relative androgen deficiency in ageing males is based on epidemiological studies. On the other hand, observational studies on men in an actual clinical setting are scarce. To date, only T'Sjoen et al. [Citation10] and Kawa et al. [Citation11] had published data, but these were primarily concerned with testosterone and QOL assessments. To the best of our knowledge, this study is the only one examining the inter-relationship between sex hormones and metabolic components and QOL measures in a self-referred clinical setting. We have deliberately excluded men with diabetes from most of the analyses in order to minimise confounding bias.
The concentrations of TT in our predominantly Asian Chinese cohort were considerably lower than those reported in a large epidemiological study on healthy Chinese males [Citation12], even though similar assay methods were employed, primarily because our subjects were self-selected males who feel they may be androgen deficient, already a bias there, as opposed to a random population sample.
It is clear from , as well as from correlation analyses, that values of TT do not appear to reflect the expected decline of testosterone with age, whereas values of cBT and cFT displayed highly significant negative correlations with age. This is explained mainly by the age-related increase in SHBG concentration. As SHBG is a major binding protein of testosterone, its age-related increase may somewhat mask the expected trend of declining TT, despite the precipitous decline of bioavailable testosterone with age. Most clinical guidelines on diagnosing androgen deficiency are based on TT values, which may be misleading. Our observation, which has also been noted in the Chinese study of Li et al. [Citation12], should prompt the use of cFT or cBT, rather than TT, as the preferred measure of serum testosterone levels in the evaluation of androgen status in ageing men. We feel that cFT is preferable, as its value expressed as pmol/L offers a broader range for defining age-specific reference ranges.
Both LH and FSH rose significantly with age, probably a reflection of increasing age-related testicular dysfunction. LH, rather than FSH, has conventionally been thought to relate directly to serum testosterone. Interestingly in our cohort, however, it was FSH, but not LH, that remained significantly related to cBT and cFT after controlling for age. FSH was also found to be an independent determinant of TT, cBT and cFT in older males, but not younger individuals. This is consistent with a recent report by Lapauw et al. [Citation13] on older community-dwelling men, in whom only FSH, not LH, was found to relate independently with serum testosterone. It is, as yet, unclear why only FSH is predictive of age-related testosterone decline. We hypothesise that because levels of cFT in older subjects, although low, were not truly deficient, there might not be adequate stimulus for increased gonadotrophin drive, hence accounting for the lack of correlation between LH and testosterone. On the other hand, the maintenance of spermatogenesis, regulated by FSH and other paracrine factors, is dependent upon adequate levels of serum testosterone [Citation14]. As such, the relative deficiency of cFT in these older males, whose mean age was 63.8 ± 3.8 years, may be sufficient to impair spermatogenesis, hence leading to the compensatory rise in FSH. The feedback response of FSH to Sertoli cell dysfunction is probably much more sensitive than the response of LH to relative androgen deficiency, thus accounting for the disparity between FSH and LH in their associations with testosterone in our cohort.
In contrast, we found WC, but not BMI nor the gonadotrophins, to be a significant independent determinant of TT, cBT and cFT in younger subjects, whose mean age was 45.3 ± 3.6 years. The lack of influence of gonadotrophins in younger males is not surprising, as their testosterone levels were well within the eugonadal range. The inverse association between testosterone and obesity is well known [Citation2–4]. That WC is an independent factor is consistent with the role of visceral adiposity in influencing serum testosterone levels, where higher levels of aromatase activity facilitate the conversion of testosterone into estradiol. However, there is controversy as to whether relative androgen deficiency is a cause or effect of visceral adiposity. Our observation that WC plays a role only in younger males, who were not truly hypogonadal, suggests that lower levels of serum testosterone, if present in these men, are predominantly an effect, rather than cause, of visceral adiposity. This is supported by the observation of Niskanen et al. [Citation15] that sustained weight loss in centrally obese men led to higher levels of serum TT. On the other hand, there may be further undefined factors linking visceral adiposity with androgens, as our subjects were actually non-obese.
We note, in our cohort, that the association between sex hormones and clinical components of the metabolic syndrome [Citation16] were not particularly striking. Negative correlations between sex hormones (TT, cBT, cFT and SHBG) and WC, as well as FPG, were well established. SHBG also mirrored TT in its correlation with TG and HDL. In addition, only SHBG, but not testosterone, related to both systolic and diastolic blood pressure. Both cBT and cFT did not correlate with blood pressure, TG or HDL. While various reports have identified hypoandrogenism and low SHBG as factors predicting the metabolic syndrome [Citation17–19], our observations are more consistent with that of Chubb et al. [Citation17], that SHBG displays a stronger association with components of the metabolic syndrome than TT, cBT or cFT.
The mechanisms underlying the complex relationships between sex hormones and the metabolic syndrome are still unclear. There is persuasive evidence implying low SHBG to be the main driver between testosterone and the metabolic syndrome, and that interventions aimed at preventing the metabolic syndrome should be directed at raising levels of SHBG, rather than increasing TT or cFT per se [Citation17]. The relative lack of influence of cBT and cFT in our study lends further weight to the primary role of SHBG. There are also reports that cFT in older non-diabetic men, whose SHBG would have risen with age, is not associated with the metabolic syndrome [Citation20,Citation21]. Visceral adiposity, leading to insulin resistance and lower SHBG, had been thought to be the physical link, as Phillips et al. [Citation22], using magnetic resonance to quantify visceral adipose tissue, had reported that the relationship between sex hormones and metabolic risk disappeared upon controlling for visceral adiposity. However, subsequent studies [Citation18,Citation19] had found significant reduction in the risk of developing metabolic syndrome with each increment of standard deviation of TT and SHBG, even after adjusting for insulin levels and body composition.
IIEF-5 scores correlated negatively with increasing age, which was consistent with expectations of increasing ED related to ageing. The IIEF-5 correlated well with total AMS (r = −0.21, P = 0.003) and the sexual subscale (r = −0.42, P <0.001), indicating that both instruments would tend to identify similar subjects.
We note, however, a lack of association between sex hormones (TT, cBT, cFT and SHBG) and both the IIEF-5 and the AMS, even though the mean IIEF-5 (13.9 ± 5.8, 95% CI: 13.0–14.7) and total AMS (41.4 ± 11.4, 95% CI: 39.8–43.0) scores implied symptoms of moderate severity, against a background of age-related androgen decline. Both T'Sjoen et al. [Citation10] and Kawa et al. [Citation11], in studies based on self-referred men, had also reported a lack of correlation between testosterone (TT and cFT) and AMS scores in subjects with moderately severe symptoms and ED. In our study, only WC (r = 0.16, P = 0.028) and BMI (r = 0.15, P = 0.038) correlated with the AMS somato-vegetative subscale.
The lack of significant association between AMS scores and sex hormones is not surprising, for it was originally developed to quantitatively measure QOL in ageing men, without any particular reference to androgen status at all [Citation23]. In a couple of screening tests to detect men with androgen deficiency, arbitrarily defined as TT <13.8 nM, it returned poor degrees of sensitivity and specificity, ranging only between 50 and 75% [Citation24].
Rather than using quantitative measures, Zitzmann et al. [Citation25] found that the progressive decline in TT concentrations correlated closely with specific symptoms, with the presence of ED being 90% specific for TT ≤7.8 nM. In our study, men with ED of at least moderate severity (N = 84, defined as IIEF-5 score of 5–11) were found to have significantly lower levels of cFT (287 ± 78 pM vs. 314 ± 82 pM, P = 0.038) than those with mild or no ED at all (N = 73, defined as IIEF-5 score of 17–25). However, this apparent difference in cFT is probably a function of age, as men with ED were significantly older (57.6 ± 8.0 years vs. 53.1 ± 7.3 years, P <0.001). Moreover, we found no correlation between IIEF-5 scores and sex hormones after controlling for age. Our findings are consistent with other reports [Citation10,Citation11] indicating that the presence of ED, or poorer QOL scores as measured using the AMS, does not necessarily indicate androgen deficiency. At best, the IIEF-5 and AMS serve only as screening tools for possible androgen deficiency. There is, as yet, no consensus on diagnosing androgen deficiency in ageing males. Although symptomatic improvement has been reported with empirical testosterone therapy in males with ‘moderately severe’ AMS scores, regardless of pre-treatment testosterone levels [Citation26], we are of the opinion that androgen deficiency should be diagnosed based on both the presence of relevant symptoms as well as biochemical evidence of hormonal insufficiency on a case-by-case basis.
One must keep in mind in clinical practice the extent of influence on androgen levels and QOL that external factors, such as usage of long-term medication, smoking and exercise, may bring upon. In our cohort, exercise significantly improved QOL scores, while usage of long-term medication had an adverse effect on testosterone levels, independently of SHBG (). Almost 40% of our subjects were on medication, usually an anti-hypertensive agent or a lipid-lowering drug, or both. Not surprisingly, chi-square analysis (not shown) revealed that usage was significantly higher with increasing age, which may further depress serum testosterone concentrations [Citation6]. The trend towards higher concentrations of testosterone in smokers is consistent with current knowledge, but lacked statistical significance probably because of the relatively small proportion of smokers (20.9%) compared to non-smokers.
Scores for the AMS psychological and somato-vegetative subscales improved significantly with increasing age. As our cohort consisted mainly of men with self-perceived QOL issues, we had expected this to be reflected as higher AMS scores with age. That the subscores correlated negatively with age, instead, is probably peculiar to our cohort, as sexual dysfunction was the principal complaint of most of our subjects, rather than psycho-somatic issues. It is also unclear whether the fact that a higher proportion of middle-aged and older males exercise, while more of younger men smoked, may be contributory.
Conclusions
In this observational study on self-referred ageing males in clinical practice, we noted FSH, rather than LH, to be a more sensitive indicator of declining serum testosterone concentration, which is better reflected by values of cFT, rather than TT. Androgen concentrations in younger males appear to be principally influenced by visceral adiposity, reflected by WC. The link between androgen deficiency and poorer metabolic status seemed to be mediated more by SHBG, rather than testosterone itself. Both the IIEF-5 and AMS, in isolation, has little value in assessing androgen status, as they did not relate with androgen levels. Extraneous factors may influence the assessment of androgen status as chronic medication usage adversely affected testosterone concentrations, while exercise improved QOL scores.
Acknowledgements
We gratefully acknowledge the assistance of Mr. Shiow-Pin Tan with the laboratory assays.
References
- HarmanSM, MetterEJ, TobinJD, PearsonJ, BlackmanMR. Longitudinal effects of aging on serum total and free testosterone levels in healthy men. Baltimore Longitudinal Study of Aging. J Clin Endocrinol Metab2001;86:724–731.
- ChenRY, WittertGA, AndrewsGR. Relative androgen deficiency in relation to obesity and metabolic status in older men. Diabetes Obes Metab2006;8:429–435.
- LaaksonenDE, NiskanenL, PunnonenK, NyyssonenK, TuomainenTP, SalonenR, RauramaaR, SalonenJT. Sex hormones, inflammation and the metabolic syndrome: a population-based study. Eur J Endocrinol2003;149:601–608.
- StellatoRK, FeldmanHA, HamdyO, HortonES, McKinlayJB. Testosterone, sex hormone-binding globulin, and the development of type 2 diabetes in middle-aged men: prospective results from the Massachusetts male aging study. Diabetes Care2000;23:490–494.
- HeinemannLA, SaadF, HeinemannKT. Can results of the Aging Males' Symptoms (AMS) scale predict those of screening scales for androgen deficiency?Aging Male2004;7:211–218.
- TravisonTG, AraujoAB, KupelianV, O'DonnellAB, McKinlayJB. The relative contributions of aging, health, and lifestyle factors to serum testosterone decline in men. J Clin Endocrinol Metab2007;92:549–555.
- RosenRC, CappelleriJC, SmithMD, LipskyJ, PenaBM. Development and evaluation of an abridged, 5-item version of the International Index of Erectile Function (IIEF-5) as a diagnostic tool for erectile dysfunction. Int J Impot Res1999;11:319–326.
- HeinemannLA, SaadF, ZimmermannT, NovakA, MyonE, BadiaX, PotthoffP, T'SjoenG, PollanenP, GoncharowNP, KimS, GiroudetC. The Aging Males' Symptoms (AMS) scale: update and compilation of international versions. Health Qual Life Outcomes2003;1:15.
- VermeulenA, VerdonckL, KaufmanJM. A critical evaluation of simple methods for the estimation of free testosterone in serum. J Clin Endocrinol Metab1999;84:3666–3672.
- T'SjoenG, FeyenE, De KuyperP, ComhaireF, KaufmanJM. Self-referred patients in an aging male clinic: much more than androgen deficiency alone. Aging Male2003;6:157–165.
- KawaG, TaniguchiH, KinoshitaH, MatsudaT, UrakamiM, SawaragiI. [Aging male symptoms and serum testosterone levels in healthy Japanese middle-aged men]. Nippon Hinyokika Gakkai Zasshi2008;99:645–651.
- LiJY, LiXY, LiM, ZhangGK, MaFL, LiuZM, ZhangNY, MengP. Decline of serum levels of free testosterone in aging healthy Chinese men. Aging Male2005;8:203–206.
- LapauwB, GoemaereS, ZmierczakH, Van PottelberghI, MahmoudA, TaesY, De BacquerD, VansteelandtS, KaufmanJM. The decline of serum testosterone levels in community-dwelling men over 70 years of age: descriptive data and predictors of longitudinal changes. Eur J Endocrinol2008;159:459–468.
- WalkerWH, ChengJ. FSH and testosterone signaling in Sertoli cells. Reproduction2005;130:15–28.
- NiskanenL, LaaksonenDE, PunnonenK, MustajokiP, KaukuaJ, RissanenA. Changes in sex hormone-binding globulin and testosterone during weight loss and weight maintenance in abdominally obese men with the metabolic syndrome. Diabetes Obes Metab2004;6:208–215.
- Executive Summary of The Third Report of The National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). JAMA2001;285:2486–2497.
- ChubbSA, HydeZ, AlmeidaOP, FlickerL, NormanPE, JamrozikK, HankeyGJ, YeapBB. Lower sex hormone-binding globulin is more strongly associated with metabolic syndrome than lower total testosterone in older men: the Health in Men Study. Eur J Endocrinol2008;158:785–792.
- MullerM, GrobbeeDE, den TonkelaarI, LambertsSW, van der SchouwYT. Endogenous sex hormones and metabolic syndrome in aging men. J Clin Endocrinol Metab2005;90:2618–2623.
- LaaksonenDE, NiskanenL, PunnonenK, NyyssonenK, TuomainenTP, ValkonenVP, SalonenR, SalonenJT. Testosterone and sex hormone-binding globulin predict the metabolic syndrome and diabetes in middle-aged men. Diabetes Care2004;27:1036–1041.
- KupelianV, PageST, AraujoAB, TravisonTG, BremnerWJ, McKinlayJB. Low sex hormone-binding globulin, total testosterone, and symptomatic androgen deficiency are associated with development of the metabolic syndrome in nonobese men. J Clin Endocrinol Metab2006;91:843–850.
- MaggioM, LauretaniF, CedaGP, BandinelliS, BasariaS, BleA, EganJ, PaolissoG, NajjarS, Jeffrey MetterE, ValentiG, GuralnikJM, FerrucciL. Association between hormones and metabolic syndrome in older Italian men. J Am Geriatr Soc2006;54:1832–1838.
- PhillipsGB, JingT, HeymsfieldSB. Relationships in men of sex hormones, insulin, adiposity, and risk factors for myocardial infarction. Metabolism2003;52:784–790.
- HeinemannLA, ThielC, AssmannA, ZimmermannT, HummelW, VermeulenA. Sex differences in ‘climacteric symptoms’ with increasing age? A hypothesis-generating analysis of cross-sectional population surveys. Aging Male2000;3:124–131.
- KratzikC, HeinemannLA, SaadF, ThaiDM, RucklingerE. Composite screener for androgen deficiency related to the Aging Males' Symptoms scale. Aging Male2005;8:157–161.
- ZitzmannM, FaberS, NieschlagE. Association of specific symptoms and metabolic risks with serum testosterone in older men. J Clin Endocrinol Metab2006;91:4335–4343.
- DingerJ, ZimmermannT, HeinemannLA, StoehrD. Quality of life and hormone use: new validation results of MRS scale. Health Qual Life Outcomes2006;4:32.