Testosterone (T) is the primary male sex hormone, or

ORIGINAL Endocrine ARTICLE Care Reference Ranges and Determinants of Testosterone, Dihydrotestosterone, and Estradiol Levels Measured using Liquid Chromatography-Tandem Mass Spectrometry in a Population-Based Cohort of Older Men Bu B. Yeap, Helman Alfonso, S. A. Paul Chubb, David J. Handelsman, Graeme J. Hankey, Paul E. Norman, and Leon Flicker School of Medicine and Pharmacology (B.B.Y., S.A.P.C., G.J.H., L.F.) and Western Australian Centre for Health and Ageing (H.A., L.F.), Centre for Medical Research, and School of Surgery (P.E.N.), University of Western Australia, Perth, Western Australia 6009, Australia; Department of Endocrinology and Diabetes (B.B.Y.), Fremantle Hospital, Fremantle, Western Australia 6160, Australia; PathWest Laboratory Medicine (S.A.P.C.), Fremantle and Royal Perth Hospitals, Perth, Western Australia 6000, Australia; and ANZAC Research Institute (D.J.H.), University of Sydney, Sydney, New South Wales 2139, Australia Context: Testosterone (T) levels decline with increasing age. Controversy exists over the threshold for classifying T as low vs. normal in older men. The relevance of assessing dihydrotestosterone (DHT) and estradiol (E2) remains unclear. Objective: We assessed the associations of T, DHT, and E2 in men aged 70 yr or older and established reference ranges for these in healthy older men. Participants: Community-dwelling men aged 70 89 yr residing in Perth, Western Australia, Australia, participated in the study. Main Outcome Measures: Plasma T, DHT, and E2 were assayed using liquid chromatographytandem mass spectrometry in early morning samples from 3690 men. Results: Increasing age, higher body mass index and waist to hip ratio, dyslipidemia, diabetes, and higher LH were independently associated with lower levels of T and DHT. Increasing age, diabetes, and higher LH were associated with lower E2. In a reference group of 394 men aged 76.1 3.2 yr reporting excellent or very good health with no history of smoking, diabetes, cardiovascular disease, cancer, depression, or dementia, the 2.5th percentile for T was 6.4 nmol/liter (184 ng/dl); DHT, 0.49 nmol/liter; and E2, 28 pmol/liter. Applying these cutoffs to all 3690 men, those with low T or DHT had an increased odds ratio for frailty, diabetes, and cardiovascular disease. Men with both low T and DHT had a higher odds ratio for these outcomes. Conclusions: The 2.5th percentile in a reference group of healthy older men provides age-appropriate thresholds for defining low T, DHT, and E2. Additional studies are needed to test their potential applicability and clinical utility in older men. (J Clin Endocrinol Metab 97: 4030 4039, 2012) ISSN Print 0021-972X ISSN Online 1945-7197 Printed in U.S.A. Copyright 2012 by The Endocrine Society doi: 10.1210/jc.2012-2265 Received May 21, 2012. Accepted August 20, 2012. First Published Online September 13, 2012 Testosterone (T) is the primary male sex hormone, or androgen, produced in men (1). Pulsatile secretion of GnRH regulates pituitary secretion of LH, which stimulates testicular production of T. T is converted by the intracellular enzyme 5 -reductase into dihydrotestosterone (DHT), a more potent ligand for the androgen receptor (1, Abbreviations: BMI, Body mass index; CV, coefficient of variation; CVD, cardiovascular disease; DHT, dihydrotestosterone; E2, estradiol; EMAS, European Male Ageing Study; GC-MS, gas chromatography-mass spectrometry; HIMS, Health in Men Study; LC-MS, liquid chromatography-tandem mass spectrometry; MrOS, Osteoporotic Fractures in Men Study; OR, odds ratio; T, testosterone; W1, wave 1; W2, wave 2. 4030 jcem.endojournals.org J Clin Endocrinol Metab, November 2012, 97(11):4030 4039

J Clin Endocrinol Metab, November 2012, 97(11):4030 4039 jcem.endojournals.org 4031 2). T is also converted into estradiol (E2) by the enzyme aromatase in fat, skeletal muscle, and other tissues (3). Therefore, actions of T can be mediated by T or DHT acting via the androgen receptor or by E2 acting via the estrogen receptor- or estrogen receptor-, for example, in bone (4). In men, T maintains virilization and muscle mass (1, 5). Hypogonadal men report systemic symptoms including decreased energy and libido, exhibit reduced lean mass and increased fat, and are vulnerable to osteoporosis (5 7). Current consensus guidelines recommend that men with symptomatic androgen deficiency and confirmed low T levels could be candidates for T supplementation (5). The assessment of symptoms related to androgen deficiency is problematic because they are nonspecific (8). Men with pathologically based hypogonadism exhibit consistent thresholds of T at which symptoms appear (9). However, in middle-aged and older men without pathologically based hypogonadism, T thresholds for symptoms such as reduced libido and other physical or psychological symptoms are less well defined (10 12). Therefore the criterion for defining low T levels in ageing men is controversial (5, 8, 13). Beyond the age of 40 yr, total T levels decline with increasing age (2, 14, 15). The difference in T levels has been estimated as up to 1% reduction per year of increasing age, at least partly due to a secular or temporal decline (2, 14 16). Thus, for any given threshold of T, the proportion of men with levels below this threshold will increase the older the sample of men studied (17, 18). Early-morning blood sampling and accurate assays for T are recommended (5, 8). Immunoassays for T exhibit reduced specificity and assay-specific bias (19), reinforcing the status of mass spectrometry as the reference method (20). Bhasin et al. (21) recently proposed the 2.5th percentile of T in healthy men aged 40 yr or younger, corresponding to a concentration of 12.1 nmol/liter measured by liquid chromatography-tandem mass spectrometry (LC-MS), for categorizing T levels as normal or low. By contrast, using a reference panel of men aged 21 35 yr with verified normal sexual and reproductive function, a lower limit for a mass spectrometry-based reference range of 10.4 nmol/ liter was found in healthy eugonadal men (19). We sought to establish reference ranges for T, DHT, and E2 measured by LC-MS (22), based on their distributions in a large group of healthy older men aged 70 yr or older. Threshold levels for low T, DHT, and E2 in older men were tested against the outcomes of frailty and presence of diabetes or cardiovascular disease (CVD). Materials and Methods Study population The Health in Men Study (HIMS) is a population-based cohort study of community-dwelling older men from Perth (Western Australia, Australia) (23). A total of 12,203 men completed a questionnaire and attended for a physical examination in wave 1 (W1; 1996 1999). During wave 2 (W2; 2001 2004), 4248 men of these men then aged 70 89 yr completed a second questionnaire and attended for a physical examination and venesection. The men were predominantly of Caucasian ethnic origin. The University of Western Australia Human Research Ethics Committee approved the study, and all men gave their written informed consent. Definitions of hypertension, dyslipidemia, and diabetes Men were considered to have hypertension if they reported this diagnosis at W1 or W2 or reported the use of antihypertensive medication or had a blood pressure of 140/90 mm Hg or greater at W2. Dyslipidemia was defined as having a fasting high-density lipoprotein less than 0.9 mmol/liter, a low-density lipoprotein of 3.4 mmol/liter or greater, triglycerides of 1.8 mmol/liter or greater, or total cholesterol of 5.5 mmol/liter or greater or receiving lipid-lowering therapy at W2. Men diagnosed with diabetes, reporting the use of glucose-lowering medication, or with a fasting or nonfasting glucose at W2 of 7 mmol/ liter or greater or 11.1 mmol/liter or greater, respectively, were considered to have diabetes (24). Assessment of medical comorbidities We assessed morbidity via the Western Australian Data Linkage System, which provides electronic linkage to records from death, hospital, and cancer registries and captures admissions to all public and private hospitals in Western Australia (25). Prevalent CVD was defined as a self-reported history of angina, acute myocardial infarction, stroke, or abdominal aortic aneurysm by questionnaire responses in W1 and W2 or hospital diagnoses of these conditions before W2. Cancer diagnoses were identified from the cancer registry between 1990 and W2. Dementia was assessed by medical history and hospital admissions data up to W2. Depression was classified according to mental health and hospital morbidity data up to W2 or by a score of 7 or more on the Geriatric Depression Scale administered at W2 (26). A generalized measure of recent medical comorbidity was constructed using the methodology of Charlson et al. (27) using questionnaire data and hospital records from 1990 to W2. Charlson index scores were categorized as no or one, two or three, or four or more medical comorbidities. Assessment of social support and socioeconomic status Participants completed the Duke Social Support Index, a measure of network support. We assessed three aspects of network support: the number of times in the past week spent with somebody not living in the same house, talked to somebody on the phone, and went out for meetings or social gatherings. These items were scored in a scale from 1 to 9 and were categorized as adequate support being a score greater than 5 (28). We used the Index of Relative Socio-Economic Disadvantage, a component

4032 Yeap et al. Testosterone in Older Men J Clin Endocrinol Metab, November 2012, 97(11):4030 4039 of the Socio-Economic Indexes for Areas of Australia (29). Lower Index of Relative Socio-Economic Disadvantage rankings indicate greater socioeconomic disadvantage. Assessment of frailty We assessed frailty at W2 with the FRAIL scale, as previously described (30). Briefly, this assesses five domains: 1) fatigue, 2) resistance (ability to climb a single flight of stairs), 3) ambulation (ability to walk one block), 4) illnesses (more than five), and 5) loss of weight (more than 5%). Participants scored positive for loss of weight if their weight decreased by more than 5% between W1 and W2. We considered participants to be frail if they presented problems in three or more of these five domains. The FRAIL scale has been validated against the outcome of mortality in older men (30). Classification of healthy older men Global perception of health was assessed at W2 with the question, In general, would you say your health is? The five responses available were excellent, very good, good, fair, and poor. This measure has been validated as a predictor of mortality in older men (31). We classified healthy older men as those reporting excellent or very good health and who also had no history of smoking, diabetes, CVD, cancer, depression, or dementia. Laboratory assays Blood samples were collected between 0800 and 1030 h at W2. Plasma was prepared immediately after phlebotomy and stored at 80 C until assayed. T, DHT, and E2 were quantified within a single LC-MS run without derivatization using atmospheric pressure photoionization in positive mode for androgens and negative mode with electrospray ionization for estrogens, from 200- l samples as previously described (22). Calibration standards were obtained from the National Measurement Institute (North Ryde, New South Wales, Australia) and prepared by diluting the working profile solutions (T and DHT, 1600 ng/ml; E2, 80 ng/ml) into a 4% albumin in PBS solution. The calibration standard for T, certified reference material M914b, is the same National Measurement Institute-certified reference standard used by the Centers for Disease Control and Prevention testosterone standardization (32). Quality control samples at three levels (high, medium, low) were prepared by spiking the 4% albumin solution with appropriate amounts of steroid stock solutions. The assay limits of detection were T, 0.04 nmol/liter; DHT, 0.01 nmol/liter; and E2, 3.4 pmol/liter. Interassay coefficients of variation (CVs; percentage) for high, medium, and low reference concentrations were as follows: T, 3.9% at 29.8 nmol/ liter, 6.8% at 5.9 nmol/liter, and 6.5% at 2.0 nmol/liter; DHT, 6.7% at 29.5 nmol/liter, 9.1% at 5.7 nmol/liter,and 13.4% at 1.9 nmol/liter; and E2, 4.8% at 1568 pmol/liter, 8.1% at 308 pmol/liter, and 8.6% at 103 pmol/liter. Precision profiles (CV percentage plotted against mean) based on 49 consecutive runs displayed a CV less than 6% for serum T levels ( 0.4 nmol/liter), less than 13% for serum DHT levels ( 0.7 nmol/liter), and less than 8% for serum E2 levels ( 25 pmol/liter). SHBG and LH had been determined by chemiluminescent immunoassay on an Immulite 2000 analyzer (Diagnostic Products Corp., Biomediq, Doncaster, Australia) (33). The sensitivity of the LH assay was 0.1 IU/liter, with an interassay CV for LH being 6.4% at 2.3 IU/liter and 5.8% at 19 IU/liter. Free T was calculated using an empirical algorithm (FTZ) as previously described (34). Statistical analyses The statistical package Stata version 11.1 (StataCorp, College Station, TX) was used to analyze the data. Baseline descriptive data are shown as mean and SDs or percentages. For continuous variables exhibiting a skewed (nonparametric) distribution, data are shown as median and interquartile range (25th to 75th percentiles). Means comparisons were performed using two-sample t tests with equal variances, which are robust for parametric and modestly skewed distributions with sufficiently large sample sizes (35). Univariate logistic regression was used to examine the associations of demographic, physical, medical, and biochemical factors with low hormone levels defined as values in the lowest quartile. Multivariate logistic regression models were constructed by retaining those factors associated with the hormone of interest at a level of P 0.1. The 2.5th percentiles of T, DHT, and E2 levels in a reference group of healthy older men were used as thresholds to define low hormone levels. These thresholds were tested in the cohort as a whole against the outcomes of frailty, diabetes, and CVD by examining the age-adjusted odds ratios for each outcome for men with hormone levels at or below the threshold compared with the remaining men. A of 0.05 was considered significant. Results Study population There were 4248 men who attended the HIMS W2 assessment and provided a blood sample. Of these men, 4230 had plasma levels of total T, DHT, and E2 assayed by LC-MS. A total of 540 men were receiving androgens (26) or antiandrogen therapy (77) or had a history of orchidectomy (56) or prostate cancer (381). Exclusion of these men left 3690 men for the analysis. Characteristics of the study population Demographic, physical, medical, and biochemical characteristics of this study population are shown in Table 1. The majority of men were in the age strata of 70 74 and 75 79 yr (n 2981, or 81% of the cohort). Consistently across age strata, most of the men were married, half had completed high school, two thirds reported ever smoking, and two thirds had adequate social support. Half reported excellent or very good health and socioeconomic status was comparable across age groups. The majority of men had hypertension or dyslipidemia. Older men had lower body mass index (BMI). There was no increase in diabetes prevalence with increasing age, but trends were apparent for increasing prevalence of CVD, cancer, depression, and dementia across older age groups. Creatinine and LH levels were higher in the oldest men, whereas T, DHT, and E2 levels were lower.

J Clin Endocrinol Metab, November 2012, 97(11):4030 4039 jcem.endojournals.org 4033 TABLE 1. Characteristics of the study population stratified by age All men Reference group 70 74 yr 75 79 yr 80 84 yr 85 yr n 3690 394 1395 1586 583 126 Age (yr) 77.0 (3.6) 76.1 (3.2) 73.6 (0.9) 77.3 (1.4) 82.0 (1.3) 86.2 (0.8) Completed high 1771 (48.0) 202 (51.3) 696 (50.0) 748 (47.2) 263 (45.1) 64 (50.8) school, n (%) Ever smoked, n (%) 2463 (66.7) N/A 892 (64.0) 1087 (68.5) 404 (69.3) 80 (63.5) Married, n (%) 3158 (85.6) 351 (89.1) 1231 (88.5) 1352 (85.2) 484 (83.0) 91 (72.2) Adequate social 2581 (69.9) 322 (81.7) 1028 (73.7) 1092 (68.9) 381 (65.4) 80 (63.5) support, n (%) Excellent/very good 1737 (47.1) N/A 705 (50.5) 734 (46.3) 250 (42.9) 48 (38.1) health, n (%) a Socioeconomic score 559.4 (189.6) 562.7 (184.1) 570.5 (195.0) 549.8 (182.4) 567.3 (191.5) 521.1 (201.1) BMI (kg/m 2 ) 26.5 (3.6) 26.0 (3.3) 26.8 (3.6) 26.4 (3.6) 26.0 (3.7) 25.5 (2.8) Waist to hip ratio 0.97 (0.07) 0.95 (0.06) 0.97 (0.07) 0.97 (0.07) 0.97 (0.09) 0.96 (0.07) Hypertension, n (%) 2854 (77.3) 288 (73.1) 1061 (76.1) 1226 (77.3) 472 (81.0) 95 (75.4) Dyslipidemia, n (%) 2856 (77.4) 275 (69.8) 1113 (79.8) 1225 (77.2) 440 (75.5) 78 (61.9) Diabetes, n (%) 571 (15.5) N/A 234 (16.8) 235 (14.8) 86 (14.8) 16 (12.7) CVD, n (%) 1362 (36.9) N/A 429 (31.8) 601 (39.1) 266 (48.2) 66 (54.5) Cancer, n (%) b 374 (10.1) N/A 120 (8.6) 172 (10.8) 64 (11.0) 18 (14.3) Depression, n (%) 224 (6.1) N/A 65 (4.7) 100 (6.3) 51 (8.9) 8 (6.6) Dementia, n (%) 205 (5.6) N/A 52 (3.7) 80 (5.0) 54 (9.3) 19 (15.1) Charlson s index, n (%) 0/1 2906 (78.8) 388 (98.5) 1159 (83.1) 1242 (78.3) 421 (72.2) 84 (66.7) 2/3 562 (15.2) 6 (1.5) 179 (12.8) 243 (15.3) 112 (19.2) 28 (22.2) 4 222 (6.0) 0 (0.0) 57 (4.1) 101 (6.4) 50 (8.6) 14 (11.1) Creatinine ( mol/liter) 93.5 (31.7) 88.6 (16.6) 91.2 (23.3) 94.2 (37.3) 96.1 (32.9) 100.2 (27.0) T (nmol/liter) 13.1 (4.9) 14.1 (4.8) 13.3 (4.8) 13.1 (4.9) 12.7 (5.0) 12.4 (4.5) DHT (nmol/liter) 1.4 (0.7) 1.6 (0.8) 1.5 (0.7) 1.4 (0.7) 1.3 (0.7) 1.4 (0.7) E2 (pmol/liter) 73.4 (29.1) 74.7 (28.9) 76.0 (28.9) 73.3 (29.6) 68.4 (28.2) 68.6 (23.9) LH IU/liter, median (IQR) 4.3 (3.0, 6.5) 3.9 (2.8, 6.0) 3.9 (2.7, 5.8) 4.4 (3.2, 6.8) 5.1 (3.4, 8.1) 5.2 (3.4, 8.7) Data are shown as mean (SD) unless otherwise stated as n (%) or median (IQR). To convert T from nanomoles per liter to nanograms per deciliter, divide by 0.0347; DHT from nanomoles per liter to nanograms per deciliter, divide by 0.0344; and E2 from picomoles per liter to picograms per milliliter, divide by 3.671. NA, Not applicable; IQR, interquartile range. a Men perceiving their health to be excellent or very good. b Men with prostate cancer already excluded. Factors associated with T, DHT, and E2 To examine factors associated with having T, DHT, and E2 in the lowest quartile of values, univariate logistic regression was performed (Table 2 ). Multivariate models were constructed to identify factors independently associated with T, DHT, and E2 in the lowest quartiles (Table 3). In multivariate analysis, increasing age, smoking status, higher BMI, higher waist to hip ratio, dyslipidemia, diabetes, and higher LH level were independently associated with lower T. Increasing age, socioeconomic score, BMI, waist to hip ratio, dyslipidemia, diabetes, and higher LH were independently associated with lower DHT. Increasing age, presence of diabetes, and higher LH were the only factors independently associated with lower E2. Hormone distributions and correlations of DHT and E2 with T Hormonal data approximated the normal distribution in the case of T, DHT, and E2, with LH skewed to the right (data not shown). Mean, median and percentage cutoffs for T, DHT, and E2 in the cohort as a whole are shown in Table 4. There was a strong correlation between DHT and T(r 0.69) and a moderate correlation between E2 and T(r 0.48) (Fig. 1). Reference range for T based on the distribution in healthy older men There were 394 healthy older men from this cohort, whose characteristics are shown in Table 1 and whose distributions of T, DHT, and E2 in are shown in Table 4. In this reference group of men aged 71 87 yr, the 2.5th percentile value for T was 6.40 nmol/liter (184 ng/dl), for DHT was 0.49 nmol/liter (14.2 ng/dl), and for E2 was 27.6 pmol/liter (7.5 pg/ml). Compared with the entire cohort, healthy older men had higher T levels (mean SD, 14.1 4.8 vs. 13.1 4.9 nmol/liter; P 0.001), higher DHT levels (1.6 0.8 vs. 1.4 0.7 nmol/liter; P 0.001), but comparable E2 levels (75 29 vs. 73 29 pmol/liter; P

4034 Yeap et al. Testosterone in Older Men J Clin Endocrinol Metab, November 2012, 97(11):4030 4039 TABLE 2. Factors associated with levels of T, DHT, or E2 in the lowest quartile of values in 3690 communitydwelling men aged 70 yr or older: univariate regression models T OR (95% CI) DHT OR (95% CI) E2 OR (95% CI) Age (yr) 1.03 (1.01, 1.06) <0.001 1.03 (1.01, 1.05) <0.001 1.04 (1.02, 1.06) <0.001 Completed high school 0.91 (0.78, 1.06) 0.21 0.87 (0.75, 1.01) 0.07 1.03 (0.89, 1.20) 0.67 Ever smoked 1.43 (1.21, 1.68) <0.001 1.30 (1.11, 1.53) <0.001 1.02 (0.87, 1.19) 0.82 Married 0.88 (0.72, 1.09) 0.24 0.95 (0.77, 1.17) 0.63 1.08 (0.87, 1.34) 0.47 High social support 0.94 (0.80, 1.10) 0.43 0.98 (0.84, 1.16) 0.84 0.89 (0.76, 1.05) 0.17 Excellent/very good health a 0.71 (0.61, 0.83) <0.001 0.71 (0.61, 0.83) <0.001 0.96 (0.83, 1.11) 0.58 Socioeconomic score (per SD) 1.06 (0.98, 1.14) 0.15 1.08 (1.00, 1.16) 0.05 1.02 (0.95, 1.10) 0.59 BMI (kg/m 2 ) 1.19 (1.16, 1.22) <0.001 1.15 (1.12, 1.17) <0.001 0.98 (0.96, 1.00) 0.10 Waist to hip ratio 1.08 (1.07, 1.09) <0.001 1.06 (1.05, 1.07) <0.001 1.00 (0.99, 1.02) 0.36 Hypertension 1.37 (1.13, 1.65) <0.001 1.10 (0.92, 1.32) 0.31 0.98 (0.82, 1.17) 0.84 Dyslipidemia 1.52 (1.26, 1.84) <0.001 1.49 (1.23, 1.80) <0.001 1.00 (0.84, 1.19) 0.99 Diabetes 2.11 (1.75, 2.55) <0.001 2.07 (1.71, 2.51) <0.001 1.25 (1.03, 1.53) 0.03 CVD 1.17 (1.00, 1.37) 0.04 1.22 (1.04, 1.42) 0.01 1.04 (0.89, 1.22) 0.61 Cancer 1.09 (0.85, 1.39) 0.49 0.96 (0.75, 1.24) 0.77 1.09 (0.85, 1.38) 0.51 Depression 1.74 (1.31, 2.31) <0.001 1.76 (1.31, 2.35) <0.001 1.19 (0.88, 1.61) 0.25 Dementia 1.20 (0.87, 1.64) 0.27 1.13 (0.82, 1.55) 0.45 1.27 (0.93, 1.73) 0.13 Charlson s index 2/3 1.49 (1.23, 1.82) <0.001 1.34 (1.10, 1.65) <0.001 1.25 (1.02, 1.53) 0.03 4 1.62 (1.21, 2.17) <0.001 1.54 (1.14, 2.07) <0.001 1.47 (1.10, 1.98) 0.01 Creatinine ( mol/liter), (per SD) 1.15 (1.06, 1.25) <0.001 1.04 (0.97, 1.12) 0.25 1.02 (0.95, 1.10) 0.50 LH (IU/liter) 1.04 (1.03, 1.06) <0.001 1.04 (1.02, 1.05) <0.001 1.02 (1.01, 1.04) <0.001 CI, Confidence interval. Bold text denotes OR with P 0.05. a Men perceiving their health to be excellent or very good. 0.190). The exclusion of men taking statins made minimal difference to the T levels in the reference group (data not shown). Exclusion of nonfasting men raised the 2.5th percentile for T to 6.55 nmol/liter (see Supplemental Table 1, published on The Endocrine Society s Journals Online web site at http://jcem.endojournals.org). Relationship of low T, DHT, and E2 levels to health outcomes The prevalences of frailty, diabetes, and CVD according to deciles of T are depicted graphically (Fig. 2). Using the 2.5th percentile in the reference group of healthy older men as the cutoff for defining low T, DHT, and E2 levels, we examined the association of these thresholds with the outcomes of frailty, diabetes, or CVD in the entire cohort. The results are presented in Table 5. The threshold for T of 6.4 nmol/liter defined 190 men (5.1%) as having low T levels, compared with 1673 men (45.3%) with the threshold of 12.1 nmol/liter [the 2.5th percentile of healthy men aged 40 yr or younger from the Framingham study (21)]. A T of 6.4 nmol/liter or less was associated with increased odds ratios (OR) of being classified as frail and having a TABLE 3. Factors associated with levels of T, DHT, or E2 in the lowest quartile of values in 3690 communitydwelling men aged 70 yr or older: multivariate regression models T OR (95% CI) DHT OR (95% CI) E2 OR (95% CI) Age (yr) 1.04 (1.02, 1.07) <0.001 1.04 (1.01, 1.06) <0.001 1.03 (1.01, 1.05) <0.001 Ever smoked 1.24 (1.04, 1.48) 0.02 1.14 (0.95, 1.35) 0.15 Excellent/very good health a 0.90 (0.76, 1.06) 0.22 0.86 (0.73, 1.02) 0.08 1.04 (0.89, 1.22) 0.58 Socioeconomic score (per SD) 1.07 (0.98, 1.15) 0.12 1.09 (1.01, 1.18) 0.04 BMI (kg/m 2 ) 1.16 (1.13, 1.20) <0.001 1.12 (1.09, 1.16) <0.001 0.98 (0.96, 1.00) 0.09 Waist to hip ratio 1.03 (1.01, 1.04) <0.001 1.02 (1.00, 1.03) 0.02 Dyslipidemia 1.35 (1.10, 1.67) 0.01 1.31 (1.07, 1.60) 0.01 Diabetes 1.63 (1.33, 2.01) <0.001 1.67 (1.36, 2.06) <0.001 1.23 (1.00, 1.52) 0.05 Depression 1.25 (0.90, 1.73) 0.18 1.37 (0.99, 1.89) 0.06 Charlson s index 2/3 1.20 (0.97, 1.48) 0.09 4 1.21 (0.88, 1.67) 0.24 Creatinine ( mol/liter) (per SD) 1.04 (0.97, 1.12) 0.28 LH (IU/liter) 1.04 (1.03, 1.06) <0.001 1.04 (1.02, 1.05) <0.001 1.02 (1.01, 1.03) 0.01 CI, Confidence interval. Bold text denotes OR with P 0.05. a Men perceiving their health to be excellent or very good.

J Clin Endocrinol Metab, November 2012, 97(11):4030 4039 jcem.endojournals.org 4035 TABLE 4. Hormonal distributions LC-MS diagnosis of diabetes or CVD. In each case, the OR for the association was greater than the corresponding OR for a T of 12.1 nmol/liter or less. A DHT of 0.49 nmol/liter or less was associated with an increased OR for each of these outcomes, with values for the OR lower than for a T of 6.4 nmol/liter or less. Low E2 ( 28 pmol/liter) was not associated with poorer health outcomes. The small number of men with both low T and low DHT (n 74, 2.0%) exhibited higher OR for frailty and slightly higher OR for diabetes and CVD compared with men with a single low T or DHT. Very few men had low DHT and low E2 (n 27, 0.7%), but these men had the highest OR for CVD. Discussion T (nmol/liter) Whole cohort (n 3690) DHT (nmol/liter) E2 (pmol/liter) cft (pmol/liter) FIG. 1. Correlations of DHT with T (A) and E2 with T (B) in 3690 community-dwelling older men aged 70 89 yr. T (nmol/liter) Reference group (n 394) DHT (nmol/liter) E2 (pmol/liter) cft (pmol/liter) Age (yr) 77.0 (3.6) 76.1 (3.2) Mean (SD) 13.1 (4.9) 1.4 (0.7) 73.4 (29.1) 185.1 (54.5) 14.1 (4.8) 1.6 (0.8) 74.7 (28.9) 197.1 (51.0) Median 12.53 1.34 70.1 182.6 13.45 1.52 72.0 194.1 97.5th percentile 24.05 3.12 139.9 297.2 25.65 3.16 138.7 316.0 95th percentile 21.48 2.74 124.1 272.1 23.42 2.91 126.3 291.0 75th percentile 15.75 1.83 89.9 216.3 16.55 1.97 92.5 229.2 25th percentile 9.82 0.92 53.6 150.4 10.90 1.06 54.0 160.4 Fifth percentile 6.35 0.46 32.3 102.9 7.70 0.57 34.7 122.0 2.5th percentile 5.04 0.34 25.0 83.8 6.40 0.49 27.6 103.6 Hormonal distributions in whole cohort: 3690 community-dwelling older men and in reference group: a reference group comprising the subset of 394 healthy older men defined as men perceiving their health to be excellent or very good, who were never-smokers with no history of diabetes, CVD, cancer, depression, or dementia. To convert T from nanomoles per liter to nanograms per deciliter, divide by 0.0347; DHT from nanomoles per liter to nanograms per deciliter, divide by 0.0344; E2 from picomoles per liter to picograms per milliliter, divide by 3.671. cft, Calculated free testosterone. In community-dwelling older men, increasing age was associated with lower levels of all three sex steroids: T, DHT, and E2. Higher BMI and waist to hip ratio, dyslipidemia, and diabetes were independently associated with lower T and DHT, reflecting relationships of obesity, central adiposity, and insulin resistance with reduced androgens (36). By contrast, the association of lower testosterone with CVD was attenuated after adjustment for these and other covariates. Lower levels of all three steroids were associated with higher LH. In a reference group of 394 healthy older men, the 2.5th percentile of T and DHT corresponded to levels below which the risk of frailty, diabetes, and CVD was increased. The use of a reference group of healthy older men merits consideration. There have been previous attempts to demarcate a threshold of T at which men could be regarded as having low circulating levels. One strategy was to reach consensus based on systematic reviews of existing evidence and deliberations between experts in the field. The International Society of Andrology and others recommended that men with T greater than 12 nmol/liter could generally be regarded as not requiring supplementation, whereas men with levels less than 8 nmol/liter might benefit from supplementation (37). The Endocrine Society s guidelines recommend the lower limit of the normal range in healthy young men as an appropriate threshold, citing values of 9.8 10.4 nmol/liter (5). The value of 10.4 nmol/liter is in accordance with the lower reference interval as measured by gas chromatography-mass spectrometry (GC-MS) in 147 young men aged 21 35 yr with verified normal sexual and reproductive function (19). However, within the expert panel, several favored a more rigorous threshold of 6.9 nmol/liter (13). A T less than 5.2 nmol/liter was regarded as appropriate for prompting assessment of other pituitary function. This difficulty of arriving at a consensus definition of low T is acknowledged (8, 13). Another strategy for defining low T in older men was to determine the threshold level of T at which symptoms of androgen deficiency appeared. How-

4036 Yeap et al. Testosterone in Older Men J Clin Endocrinol Metab, November 2012, 97(11):4030 4039 FIG. 2. Percentages of the 3690 older men who were classified as being frail (A) or as having diabetes mellitus (B) or CVD (C) according to the deciles of circulating testosterone. ever, although symptoms such as a lack of energy, a lack of motivation, and reduced libido appear at consistent thresholds of T when T replacement is interrupted in men with pathologically based androgen deficiency, there is considerable interindividual variation including a lower symptom threshold in men with secondary vs. primary hypogonadism (9.7 vs. 11.7 nmol/liter) (9). These thresholds are also more diffuse and difficult to define in samples of men without documented androgen deficiency (10, 11, 18). In a community-based sample of men aged 40 70 yr, libido and T were significantly associated, but the difference in T levels in men with low libido and those without was small (10). Other studies have shown poorly defined thresholds for symptoms such as loss of libido or vigor in men aged 50 86 yr (11) or relatively subtle inflection points when probability of sexual symptoms are plotted against levels of T. For example, a moderate increase in the probability of decreased frequency of sexual thoughts was seen at a T lower than 8 nmol/liter compared with no inflection points for physical or psychological symptoms (12). Bhasin et al. (21) measured T using LC-MS in a reference group of 456 men aged 40 yr or younger from the Framingham study with no history of smoking, obesity, hypercholesterolemia, hypertension, diabetes, CVD, or cancer. The LC-MS assay for T allowed for comparability with the European Male Ageing Study (EMAS) and Osteoporotic Fractures in Men Study (MrOS) cohorts, which measured T using GC-MS (12, 38). However, the 2.5th percentile threshold of 12.1 nmol/ liter would have classified 23.5% of 40- to 79-yr-old men (EMAS) and 40.3% of men aged 65 yr or older (MrOS) as having T below the young-normal reference range (21). By contrast, we defined healthy older men from the HIMS cohort of men aged 70 89 yr as those reporting excellent or very good self-perceived health, with no history of TABLE 5. older men Associations of low T, DHT, and E2 levels with frailty, diabetes, and CVD in 3690 community-dwelling Variable T 12.1 nmol/liter (348 ng/dl) (n 1673) T 6.40 nmol/liter (184 ng/dl) (n 190) DHT 0.49 nmol/liter (14.2 ng/dl) (n 204) E2 27.55 pmol/liter (7.5 pg/ml) (n 126) Low T ( 6.40 nmol/liter) and DHT ( 0.49 nmol/liter) (n 74) Low T ( 6.40 nmol/liter) and E2 ( 27.55 pmol/liter) (n 51) Low DHT ( 0.49 nmol/liter) and E2 ( 27.55 pmol/liter) (n 27) OR, Age-adjusted odds ratio; CI, confidence interval. Frailty (n 563) OR (95% CI) Diabetes (n 571) OR (95% CI) CVD (n 1, 362) OR (95% CI) 1.82 (1.51, 2.19) <0.001 2.22 (1.85, 2.67) <0.001 1.23 (1.08, 1.41) <0.001 2.70 (1.94, 3.74) <0.001 2.76 (2.00, 3.81) <0.001 1.61 (1.20, 2.17) <0.001 1.67 (1.18, 2.37) <0.001 2.50 (1.82, 3.43) <0.001 1.47 (1.10, 1.97) 0.01 1.16 (0.73, 1.84) 0.54 0.88 (0.52, 1.47) 0.62 0.97 (0.67, 1.41) 0.89 3.28 (2.01, 5.36) <0.001 2.77 (1.69, 4.56) <0.001 1.78 (1.11, 2.85) 0.02 1.44 (0.75, 2.77) 0.27 1.09 (0.51, 2.34) 0.83 1.69 (0.96, 2.97) 0.07 2.30 (1.00, 5.26) 0.05 0.72 (0.21, 2.39) 0.59 3.48 (1.51, 8.03) <0.001

J Clin Endocrinol Metab, November 2012, 97(11):4030 4039 jcem.endojournals.org 4037 smoking, diabetes, CVD, cancer, depression, or dementia, providing a reference group of 394 men. These men constitute 11% of the cohort of community-dwelling older men and likely represent very healthy older men. The 2.5th percentile for T in this reference group identified 5.1% of the whole HIMS cohort as having low T levels. By comparison, the prevalence of late-onset hypogonadism was assessed as 5.1% in men aged 70 79 yr from EMAS (12) and that of symptomatic androgen deficiency at 6.0% of men aged 40 69 yr from the Massachusetts Male Aging Study (17) and 5.6% of men aged 30 79 yr from the Boston Area Community Health Survey (18). Androgen deficiency symptoms were not assessed specifically in HIMS (23). Therefore, the T threshold of 6.4 nmol/liter could be regarded as representing the lower end of the biochemical reference range for healthy older men and is in accordance with the more conservative views of the consensus panel (13). Of interest, the OR for frailty, diabetes, and CVD for a T of 12.1 nmol/liter or less in HIMS were comparable with the OR for adverse health outcomes reported for Framingham, EMAS, and MrOS men (21). However, ORs were consistently higher for a T of 6.4 nmol/liter or less vs. 12.1 nmol/liter or less in HIMS men, and men with both low T and low DHT exhibited the highest OR for poorer health outcomes. Men with E2 at or below the 2.5th percentile of the reference group did not show any increase in the odds for frailty, diabetes, or CVD, nor did men with both T and E2 below this threshold. In the MrOS Sweden cohort, both T and E2 in the lowest quartiles predicted mortality, with the highest hazard ratio seen in men with both low T and low E2 as measured by GC-MS (39). Additional research is needed to clarify whether E2 is a robust independent predictor of non-bone-related longitudinal health outcomes in men. The higher OR for CVD found in men with both low DHT and low E2 is intriguing and may reflect sex steroidspecific actions of DHT and E2 at a tissue level (40, 41). However, the number of men involved was small, and confirmatory studies would be appropriate before considering the inclusion of either DHT or E2 into guidelines for the assessment of gonadal status in ageing men. Strengths of our study include the large sample of community-dwelling older men; careful phenotyping using the combination of clinic assessments and data linkage for assessing medical morbidities; and the assessment of T, DHT, and E2 using an established LC-MS methodology. We defined a reference group of healthy older men to derive reference ranges for T, DHT, and E2 and tested the lower limits of the reference ranges against health outcomes relevant to aging men. We acknowledge several limitations of the study. We did not assess specific symptoms of androgen deficiency; instead we relied on men in the reference group reporting their health to be excellent or very good. This allowed men to make a positive statement of their health, excluding men who rated their health as simply good. Given the complexities inherent in assessing andrological function at this age (42), we described a reference group of healthy older men. Men returned for assessment in 2001 2004 after a previous assessment in 1996 1999. Therefore, a healthy survivor effect may be present. However, this would strengthen the selection of a reference group of healthy older men. The results are based on blood sampling at a single time point, albeit collected early in the morning to minimize any confounding from circadian variation. Until there is an international standardization of LC-MS assays, caution is needed before applying any reference range broadly. Men in HIMS are predominantly Caucasian in ethnicity; therefore, we cannot extrapolate these findings to other racial groups (43). However, HIMS can be compared with Framingham, EMAS, and U.S. MrOS (21). The difference between the lower thresholds of the reference ranges for healthy young men 40 yr old or younger and healthy older men 70 yr old or older [12.1 nmol/liter (21) or 10.4 nmol/liter (19) vs. 6.4 nmol/liter] represents an interval of more than 3 decades and reflects the age-associated decline in T levels previously reported (14 16). Whether this difference arises from aging itself or the accumulation of adverse health, lifestyle, behavioral, or environmental exposures remains to be determined. We propose the 2.5th percentile of T in this reference group of healthy older men may be considered as a threshold for defining low T levels in men aged 70 yr or more. This may represent a conservative threshold, and it is possible that some men with levels greater than 6.4 nmol/liter might benefit from T supplementation. Additional studies are needed to test the potential applicability and clinical utility of this threshold in older men. Acknowledgments We thank the staff of PathWest Laboratory Medicine and Fremantle and Royal Perth Hospitals (Perth, Western Australia, Australia) and the ANZAC Research Institute (Sydney, New South Wales, Australia) for their excellent technical assistance. We also thank the staff and management of Shenton Park Hospital (Shenton Park, Western Australia, Australia) for their support of the study. We especially thank all the men and staff who participated in the Western Australian Abdominal Aortic Aneurysm Program and the Health in Men Study. Address all correspondence and requests for reprints to: Bu Beng Yeap, M.B.B.S., Ph.D., School of Medicine and Pharmacology, Level 2, T Block, Fremantle Hospital, Alma Street,

4038 Yeap et al. Testosterone in Older Men J Clin Endocrinol Metab, November 2012, 97(11):4030 4039 Fremantle, Western Australia 6160, Australia. E-mail: byeap@ cyllene.uwa.edu.au. B.B.Y. is recipient of a Clinical Investigator Award from the Sylvia and Charles Viertel Charitable Foundation, New South Wales, Australia. The Health in Men Study was funded by Project Grants 279408, 379600, 403963, 513823, and 634492 from the National Health and Medical Research Council of Australia. The funding sources had no involvement in the planning, analysis, or writing of the manuscript. Disclosure Summary: The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported. References 1. Bhasin S 2008 Testicular disorders. In: Kronenberg HM, Melmed S, Polonsky KS, Larsen PR, eds. Williams textbook of endocrinology. 11th edition. Philadelphia: Saunders Elsevier; 645 699 2. Kaufman JM, Vermeulen A 2005 The decline of androgen levels in elderly men and its clinical and therapeutic implications. Endocr Rev 26:833 876 3. Jones ME, Boon WC, McInnes K, Maffei L, Carani C, Simpson ER 2007 Recognizing rare disorders: aromatase deficiency. Nat Clin Pract Endocrinol Metab 3:414 421 4. Khosla S, Melton 3rd LJ, Riggs BL 2002 Estrogen and the male skeleton. J Clin Endocrinol Metab 87:1443 1450 5. Bhasin S, Cunningham GR, Hayes FJ, Matsumoto AM, Snyder PJ, Swerdloff RS, Montori VM 2010 Testosterone therapy in men with androgen deficiency syndromes: an Endocrine Society clinical practice guideline. J Clin Endocrinol Metab 95:2536 2559 6. Fink HA, Ewing SK, Ensrud KE, Barrett-Connor E, Taylor BC, Cauley JA, Orwoll ES 2006 Association of testosterone and estradiol deficiency with osteoporosis and rapid bone loss in older men. J Clin Endocrinol Metab 91:3908 3915 7. Meier C, Nguyen TV, Handelsman DJ, Schindler C, Kushnir MM, Rockwood AL, Meikle AW, Center JR, Eisman JA, Seibel MJ 2008 Endogenous sex hormones and incident fracture risk in older men: the Dubbo Osteoporosis Epidemiology Study. Arch Intern Med 168: 47 54 8. Cunningham GR, Toma SM 2011 Why is androgen replacement in males controversial? J Clin Endocrinol Metab 96:38 52 9. Kelleher S, Conway AJ, Handelsman DJ 2004 Blood testosterone threshold for androgen deficiency symptoms. J Clin Endocrinol Metab 89:3813 3817 10. Travison TG, Morley JE, Araujo AB, O Donnell AB, McKinlay JB 2006 The relationship between libido and testosterone levels in aging men. J Clin Endocrinol Metab 91:2509 2513 11. Zitzmann M, Faber S, Nieschlag E 2006 Association of specific symptoms and metabolic risks with serum testosterone in older men. J Clin Endocrinol Metab 91:4335 4343 12. Wu FC, Tajar A, Beynon JM, Pye SR, Silman AJ, Finn JD, O Neill TW, Bartfai G, Casanueva FF, Forti G, Giwercman A, Han TS, Kula K, Lean ME, Pendleton N, Punab M, Boonen S, Vanderschueren D, Labrie F, Huhtaniemi IT 2010 Identification of late-onset hypogonadism in middle-aged and elderly men. N Engl J Med 363:123 135 13. Anawalt BD 2010 Guidelines for testosterone therapy for men: how to avoid a mad (T)ea party by getting personal. J Clin Endocrinol Metab 95:2614 2617 14. Harman SM, Metter EJ, Tobin JD, Pearson J, Blackman MR 2001 Longitudinal effects of aging on serum total and free testosterone levels in healthy men. J Clin Endocrinol Metab 86:724 731 15. Feldman HA, Longcope C, Derby CA, Johannes CB, Araujo AB, Coviello AD, Bremner WJ, McKinlay JB 2002 Age trends in the level of serum testosterone and other hormones in middle-aged men: longitudinal results from the Massachusetts Male Aging Study. J Clin Endocrinol Metab 87:589 598 16. Travison TG, Araujo AB, Hall SA, McKinlay JB 2009 Temporal trends in testosterone levels and treatment in older men. Curr Opin Endocrinol Diabetes Obes 16:211 217 17. Araujo AB, O Donnell AB, Brambilla DJ, Simpson WB, Longcope C, Matsumoto AM, McKinlay JB 2004 Prevalence and incidence of androgen deficiency in middle-aged and older men: estimates from the Massachusetts Male Aging Study. J Clin Endocrinol Metab 89: 5920 5926 18. Araujo AB, Esche GR, Kupelian V, O Donnell AB, Travison TG, Williams RE, Clark RV, McKinlay JB 2007 Prevalence of symptomatic androgen deficiency in men. J Clin Endocrinol Metab 92: 4241 4247 19. Sikaris K, McLachlan RI, Kazlauskas R, de Kretser D, Holden CA, Handelsman DJ 2005 Reproductive hormone reference intervals for healthy fertile young men: evaluation of automated platform assays. J Clin Endocrinol Metab 90:5928 5936 20. Shackleton C 2010 Clinical steroid mass spectrometry: a 45-year history culminating in HPLC-MS/MS becoming an essential tool for patient diagnosis. J Steroid Biochem Mol Biol 121:481 490 21. Bhasin S, Pencina M, Jasuja GK, Travison TG, Coviello A, Orwoll E, Wang PY, Nielson C, Wu F, Tajar A, Labrie F, Vesper H, Zhang A, Ulloor J, Singh R, D Agostino R, Vasan RS 2011 Reference ranges for testosterone in men generated using liquid chromatography tandem mass spectrometry in a community-based sample of healthy nonobese young men in the Framingham Heart Study and applied to three geographically distinct cohorts. J Clin Endocrinol Metab 96: 2430 2439 22. Harwood DT, Handelsman DJ 2009 Development and validation of a sensitive liquid chromatography tandem mass spectrometry assay to simultaneously measure androgens and estrogens in serum without derivatization. Clin Chim Acta 409:78 84 23. Norman PE, Flicker L, Almeida OP, Hankey GJ, Hyde Z, Jamrozik K 2009 Cohort profile: the Health In Men Study (HIMS). Int J Epidemiol 38:48 52 24. The Australian Diabetes, Obesity, and Lifestyle Study 2000 Diabesity and associated disorders in Australia 2000. Melbourne, Australia: International Diabetes Institute 25. Holman CD, Bass AJ, Rouse IL, Hobbs MS 1999 Population-based linkage of health records in Western Australia: development of a health services research linked database. Aust N Z J Public Health 23: 453 459 26. Almeida OP, Flicker L, Norman P, Hankey GJ, Vasikaran S, van Bockxmeer FM, Jamrozik K 2007 Association of cardiovascular risk factors and disease with depression in later life. Am J Geriatr Psychiatry 15:506 513 27. Charlson ME, Pompei P, Ales KL, MacKenzie CR 1987 A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 40:373 383 28. Pachana NA, Smith N, Watson M, McLaughlin D, Dobson A 2008 Responsiveness of the Duke Social Support sub-scales in older women. Age Ageing 37:666 672 29. Pink B 2008 An Introduction to Socio-Economic Indexes for Areas (SEIFA) 2006. Canberra, Australia: Australian Bureau of Statistics 30. Hyde Z, Flicker L, Almeida OP, Hankey GJ, McCaul KA, Chubb SA, Yeap BB 2010 Low free testosterone predicts frailty in older men. The Health in Men Study. J Clin Endocrinol Metab 95:3165 3172 31. Alfonso H, Beer C, Yeap BB, Hankey GJ, Flicker L, Almeida OP 2012 Perception of worsening health predicts mortality in older men: the Health in Men Study. Arch Gerontol Geriatr 55:363 368 32. Vesper HW, Botelho JC 2010 Standardization of testosterone measurements in humans. J Steroid Biochem Mol Biol 121:513 519 33. Yeap BB, Almeida OP, Hyde Z, Norman PE, Chubb SA, Jamrozik K, Flicker L 2007 In men older than 70 years, total testosterone remains stable while free testosterone declines with age. The Health In Men Study. Eur J Endocrinol 156:585 594 34. Ly LP, Sartorius G, Hull L, Leung A, Swerdloff RS, Wang C, Han-

J Clin Endocrinol Metab, November 2012, 97(11):4030 4039 jcem.endojournals.org 4039 delsman DJ 2010 Accuracy of calculated free testosterone formulae in men. Clin Endocrinol (Oxf) 73:382 388 35. Armitage P, Berry G, Matthews JNS 2002 Statistical methods in medical research. 4th ed. Oxford, UK: Blackwell Science; 83 146 36. Grossmann M 2011 Low testosterone in men with type 2 diabetes: significance and treatment. J Clin Endocrinol Metab 96:2341 2353 37. Wang C, Nieschlag E, Swerdloff R, Behre HM, Hellstrom WJ, Gooren LJ, Kaufman JM, Legros JJ, Lunenfeld B, Morales A, Morley JE, Schulman C, Thompson IM, Weidner W, Wu FCW 2008 Investigation, treatment and monitoring of late-onset hypogonadism in males. Eur J Endocrinol 159:507 514 38. LeBlanc ES, Nielson CM, Marshall LM, Lapidus JA, Barrett-Connor E, Ensrud KE, Hoffman AR, Laughlin G, Ohlsson C, Orwoll ES 2009 The effects of serum testosterone, estradiol, and sex hormone binding globulin levels on fracture risk in older men. J Clin Endocrinol Metab 94:3337 3346 39. Tivesten A, Vandenput L, Labrie F, Karlsson MK, Ljunggren O, Mellström D, Ohlsson C 2009 Low serum testosterone and estradiol predict mortality in elderly men. J Clin Endocrinol Metab 94:2482 2488 40. Sieveking DP, Lim P, Chow RW, Dunn LL, Bao S, McGrath KC, Heather AK, Handelsman DJ, Celermajer DS, Ng MK 2010 A sexspecific role for androgens in angiogenesis. J Exp Med 207:345 352 41. Nofer JR 2012 Estrogens and atherosclerosis: insights from animal models and cell systems. J Mol Endocrinol 48:R13 R29 42. Hyde Z, Flicker L, Hankey GJ, Almeida OP, McCaul KA, Chubb SA, Yeap BB 2010 Prevalence of sexual activity and associated factors in men aged 75 95 years: a cohort study. Ann Intern Med 153:693 702 43. Orwoll ES, Nielson CM, Labrie F, Barrett-Connor E, Cauley JA, Cummings SR, Ensrud K, Karlsson M, Lau E, Leung PC, Lunggren O, Mellström D, Patrick AL, Stefanick ML, Nakamura K, Yoshimura N, Zmuda J, Vandenput L, Ohlsson C 2010 Evidence for geographical and racial variation in serum sex steroid levels in older men. J Clin Endocrinol Metab 95:E151 E160