P. Crabbe,* V. Bogaert,* D. De Bacquer, S. Goemaere, H. Zmierczak, and J. M. Kaufman

Department of Endocrinology (P.C., V.B., J.M.K.), Unit for Osteoporosis and Metabolic Bone Diseases (S.G., H.Z., J.M.K.) and Department of Public Health (D.D.B.), Ghent University Hospital, 9000 Ghent, Belgium.

Jean-Marc Kaufman, Department of Endocrinology, 9K12 I.E., Ghent University Hospital, De Pintelaan 185, 9000 Ghent, Belgium; E-mail: jean.kaufman@ugent.be.

Abstract

Context: There is a large interindividual variation in serum (free) testosterone (FT) levels in men, underlain in part by genetic components.

Objective: The objective of the study was to explore the hypothesis that this variability results in part from differences in androgen sensitivity and feedback loop set point and assess the role of the androgen receptor (AR) polyglutamine tract polymorphism encoded by a CAG repeat of variable length in exon 1 of the AR gene.

Design/Setting/Participants: We performed a cross-sectional analysis in two independent populations of healthy men, consisting of 2322 men aged 35-59 yr (Belstress study) and 358 men aged 25-45 yr (Siblos study), respectively.

Main Outcome Measures: Serum hormonal levels and the AR gene CAG repeat length were determined.

Results: In the Belstress population, serum testosterone and calculated FT showed a positive linear association with LH (P < 0.001). In the 200 men with lowest FT, CAG repeat number was lower than in the 200 men with highest FT (P = 0.004). As studied in a larger subset of the population consisting of 857 men covering the whole FT range, FT increased progressively with CAG repeat length (P = 0.003). These findings of a positive relation of FT with both LH and CAG repeat length were confirmed in the Siblos study population (both P ≤ 0.001). Difference in FT between extreme quartiles of CAG repeat was 10 and 14% in the Belstress and Siblos study, respectively. In both study populations, CAG repeat length was also positively associated with serum total testosterone (P ≤ 0.004).

Conclusions: The data support the view that between-subject variability in serum FT in healthy men is underlain in part by differences in androgen sensitivity and feedback set point, with a contributory role of AR polymorphism. These findings have potential implications for the interpretation of epidemiological studies, diagnosis of hypogonadism, and pharmacogenetics of androgen treatment in men.

Abbreviations: ANCOVA, Analysis of covariance; AR, androgen receptor; BMI, body mass index; E₂, estradiol; FT, free T; HDL, high- density lipoprotein; T, testosterone.

Introduction

THERE IS A LARGE interindividual variation of serum total and free testosterone (T and FT) levels at all ages in healthy adult men. The physiological basis and clinical significance of this phenomenon is still poorly understood. Whereas lifestyle-related factors such as adiposity and smoking (1, 2), besides other modifiable factors, are known to contribute to this between-subject variation (3), studies in twins indicate a substantial genetic component in the determination of T blood levels (4, 5). Consistent with the latter findings are observations that, notwithstanding the moment- to-moment variations entrained by episodic pituitary LH secretion, mean serum (F)T levels in individual men tend to remain rather stable over the mid to long term (6-8). Serum T is regulated through the hypothalamic-pituitarygonadal feedback loop. In this study we explored the hy-pothesis that differences in feedback loop set point contribute to interindividual variation in serum T, with part of the between-subject differences in serum T reflecting genetically determined, subtle differences in androgen sensitivity. To this end, we assessed in healthy adult men the relationship between serum LH and (F)T and the influence of the CAG repeat polymorphism in the androgen receptor (AR) gene, one of the more obvious candidate genes that might contribute to the genetic component of between-subject variability of serum T.

As to the relationship between serum LH and (F)T, the premise was that the finding of a positive association is compatible with a genetic effect modulating androgen sen-sitivity. Indeed, a same serum (F)T level will result in a lesser feedback suppression of serum LH in a men with relatively lower androgen sensitivity, which in turn will result in in-creased LH levels and a consequent increase of serum (F)T; both LH and (F)T are thus expected to increase with decreasing androgen sensitivity. On the other hand, in the absence of differences in androgen sensitivity, genetic factors affecting testicular responsiveness to LH and Leydig cell

secretory capacity or the rate of inactivating T metabolism would rather be expected to result in no or a negative as-sociation between LH and (F)T. Indeed, men with a relatively lower Leydig cell responsiveness to LH or with increased T catabolism would require higher circulating LH to maintain a same level of serum (F)T in accordance with the feedback set point; this would result in no association between (F)T and LH or in a negative association if Leydig cell secretory capacity is a limiting factor.

Furthermore, we assessed the influence of the CAG repeat polymorphism in exon 1 of the AR. This trinucleotide repeat encodes a highly polymorphic polyglutamine tract, and its length has been reported to be inversely associated with AR transactivation (9). This CAG repeat polymorphism might affect androgen action (10) and thus contribute to the vari-ability of serum T levels through modulation of the negative feedback control of the hypothalamic-pituitary-testicular axis. The CAG repeat length has been associated with a series of androgen-related clinical effects (11, 12), whereas information about its effect on circulating T levels is surprisingly scarce (13, 14).

The issue whether interindividual variation in serum (F)T in healthy men reflects differences in androgen sensitivity has potentially far-reaching implications, e.g. for a correct interpretation of epidemiological data on clinical correlates of serum T levels in men, for the diagnosis of hypoandro- genism in both young and elderly men, or eventually for individualization and optimization of T replacement therapy.

In the present study, the foregoing hypothesis that part of the interindividual variation of serum (F)T in healthy men reflects subtle differences in feedback loop set point with a contributory role of the CAG repeat polymorphism of the AR, was explored in two independent populations of healthy adult men.

Subjects and methods

Study populations

Belstress population. This study was performed in a subsample of the Belstress project, a work-floor-based study focused on job stress, cardiovascular and other health issues, and sickness absence. Objectives, design, and methodology of the Belstress project have been extensively described elsewhere (15). The subsample used for this study included 2322 male subjects aged 35-59 yr, selected according to two main criteria, i.e. serum sampled before 1000 h and from companies located in Ghent or its neighborhood. All participants signed an informed consent and completed questionnaires. Exclusion criteria comprised potentially interfering medication or extreme low T levels (≤45 ng/dl). The relationship between serum (F)T and LH levels was studied in the whole study population of 2322 men. The impact of the AR gene CAG polymorphism was assessed in the 200 men with lowest FT, compared with the 200 men with highest FT as well as in a subset of 857 men representative of the whole FT range in the study population.

Siblos population. In the Siblos study, young men were recruited from population lists of semirural communities around Ghent (Belgium) to participate in a sibling-pair study on genetic contribution to the determination of bone mass, sex steroids, and their interaction. The primary condition to be recruited was to have a brother in the same age range also willing to participate. Exclusion criteria were defined as illnesses or medication use that may affect sex steroid levels, body composition, and bone metabolisms. After applying exclusion criteria, 276 pairs, 17 triplets, and two quartets of dizygotic brothers were included in the study; furthermore, 63 men were included as single participants when their brother could not participate in the study due to the presence of an exclusion criterion.

For the present cross-sectional analysis, we used a group of 358 unrelated men consisting of one randomly chosen brother (n = 295) of each pair, triplet, or quartet and the 63 single participants. All men were in good health and completed questionnaires about previous illness and smoking. The study protocol was approved by the ethical committee of the Ghent University Hospital, and a written informed consent was obtained from all participants.

Hormonal assays

Serum was obtained between 0800 and 1000 h and stored at —80 C until analysis. Commercial direct immunoassays were used to determine serum levels of LH (ECLIA; Roche Diagnostics, Mannheim, Germany), estradiol (E₂) (clinical assay; DiaSorin s.r.l., Saluggia, Italy, according to a modified protocol with doubling the amount of serum), total T, and SHBG (Orion Diagnostica, Espoo, Finland). Serum FT was calculated from serum total T, SHBG, and albumin concentrations using a previously validated equation (16,17). For all measurements, intra- and interassay coefficients of variation were less than 10 and 15%, respectively; in particular interassay coefficient of variation was 7.5% for T (at a mean concentration of 320 ng/dl), 3.8% for SHBG (at a mean concentration of 39.7 nmol/liter), and was between 8.7 and 14.2% for FT (as calculated for two series of 10 assay runs each at a mean concentration of 13 ng/dl) (to convert T and FT to nanomoles per liter, multiply by 0.0347).

Genotyping

Genomic DNA was extracted from EDTA-treated blood using a com-mercial kit (Purgene kit; Gentra Systems, Minneapolis, MN). The AR exon 1 region encoding the CAG repeat was amplified using PCR with forward primer 5'-GAATCTGTTCCAGAGCGTGC-3', fluorescently labeled with NED (Belstress study) or FAM (Siblos study) and reverse primer 5'-TTCCTCATCCAGGACCAGGTA-3'. Each PCR was initiated with a 5-min denaturation step at 95 C and terminated with a 20-min extension step at 72 C; in-between reaction profiles were as follows: denaturation at 95 C for 60 sec, annealing at 62 C for 60 sec, and extension at 72 C for 90 sec for 35 cycles. The PCR products were mixed with a Genescan 400HD ROX size standard and deionized formamide, heat denatured, and electrophoresed (Belstress study) on a ABI Prism 310 or (Siblos study) a 96-capillary 3730 xl genetic analyzer (ABI Prism; Perkin- Elmer Applied Biosystems, Foster City, CA).

Statistics

Comparison of the hormone levels between groups with low and high FT was performed by multiple regression analyses with additional correction for age, body mass index (BMI), and smoking. Association between LH and FT, as well as between the CAG repeat length and hormone levels, were assessed with partial Pearson correlations, using ln-transformed (for T, FT, and E₂) or sqrt-transformed (for LH and SHBG) variables when necessary. To evaluate the independent contribution of CAG repeat length in explaining the variability in hormonal levels, comparison of the distribution of hormonal levels between quartiles of AR CAG repeat length was performed using multiple regression analyses of covariance with correction for age, BMI, and smoking. To meet the necessary model assumptions, hormonal levels were ln or sqrt transformed when necessary. All statistical analyses were performed using SPSS 12.0 software. P < 0.05 was considered significant in all analyses.

Results

The working hypothesis of the existence in healthy men of a positive association between serum T and serum LH levels, on the one hand, and a positive association between serum T and the CAG repeat length in the AR receptor gene, on the other hand, was initially tested by comparing in the Belstress population men with highest and lowest (F)T levels, respec-tively. Findings for these extreme groups were subsequently assessed for consistency in subgroups of the Belstress pop-ulation representing the full (F)T concentration range. Finally, consistency with results obtained in the Belstress population was assessed in an independent population of healthy men, i.e. the Siblos population.

Belstress population

Clinical characteristics including age, BMI, smoking habits and hormones for the total study population of 2322 men are summarized in Table 1. Of this population of 2322 men, two groups of 200 men with the highest (≥15.26 ng/dl) and lowest (<9.41 ng/dl) level of calculated FT, respectively, were selected for initial further comparison. Basic clinical characteristics of both subgroups are also presented in Table 1. Highly significant differences were observed between both groups for all hormonal levels after adjustment for age, BMI, and smoking as shown in Table 2. Men with the highest FT levels also showed the highest LH, SHBG, and E₂ levels as well as a higher T to E₂ ratio. The consistency of the finding for these two extreme subgroups, that high (F)T levels were associated with higher LH levels, was confirmed across the whole study population and complete FT range (n = 2322): T and FT were positively correlated with LH after adjustment for age, BMI, and smoking (r = 0.193, P < 0.001,andr = 0.239, P > 0.001, for FT and T, respectively). This positive association of both hormonal levels is illustrated in Fig. 1, A and C), representing the means of LH adjusted for age, BMI, and smoking per quartile of T and FT [analysis of covariance (ANCOVA), P < 0.001]. For SHBG no clear linear positive effect across the FT range could be demonstrated (results not shown).

Table 1. Clinical characteristics of the Belstress study population and considered study population subsets and the Siblos study population.

Table 2. Comparison of hormonal levels and AR CAG repeat length between the men with lowest (≤ 9.41 ng/dl) and highest (≥15.26 ng/dl) FT levels in the Belstress population".

Figure 1. Geometric means with 95% confidence interval of LH according to quintiles of FT or T in the total study Belstress population (n = 2322) (A and C) and the Siblos population (n = 358) (B and D) after adjustment for age, BMI, and smoking (ANCOVA, LH sqrt-transformed).

Figure 2. Geometric means with 95% confidence interval of FT and T according to quartiles of CAG-repeat length in the Belstress subpopulation (n = 857) (A and C) and the Siblos population (n = 358) (B and D), after adjustment for age, BMI, and smoking (ANCOVA, FT and T ln-transformed). To convert T and FT to nanomoles per liter, multiply by 0.0347.

In the 200 men with highest FT, the mean CAG repeat number in the AR was significantly higher, compared with the 200 men with lowest FT (Table 2). The consistency of this finding was confirmed in an enlarged study subset of 857 men representative of the full range of FT in the overall study population, consisting of the former 400 men and randomly selected subjects from the intermediate FT concentration strata, the numbers of included subjects being matched so that the different concentration strata covering the population range are equally represented. Their basic clinical char-acteristics are not different from the global study population (Table 1). The CAG repeat length ranged from nine to 41 (P25-P75: 20-24; median 21). Consistent with the findings for the men with extreme FT values, in those men the full range of FT, there was a significant partial correlation (adjusted for age, BMI, and smoking) between CAG repeat length and FT (r = 0.120; P < 0.001) orT(r = 0.103; P = 0.003) levels. There was no significant correlation with SHBG or LH. Quartile analysis of the CAG repeat length showed significant differences for serum levels of T (P = 0.010), FT (P = 0.003), LH (P = 0.020), and SHBG (P = 0.032). Serum LH, T, and FT were highest in the upper CAG quartiles, but for only FT, was there a progressive increase across all CAG repeat quartiles. Results for total T (ANCOVA: P = 0.01) and FT (ANCOVA: P = 0.003) are illustrated in Fig. 2, A and C); the adjusted means of T and FT in Q₄ were 6 and 10% higher than in Q₁, respectively.

Siblos population

Analysis of this population was intended as independent confirmation of the findings in the Belstress population. The characteristics of this second study population, consisting of 358 younger men, are summarized in Table 1. Consistent with the findings for the Belstress study population, in this group both total T and calculated FT were positively correlated with LH after adjustment for age, BMI, and smoking (r = 0.358 and r = 0.323, respectively, both P < 0.001), and LH showed a progressive linear increase with higher T and FT quartiles (ANCOVA, P < 0.001) (Fig. 1, B and D). As for SHBG, no significant correlation could be established with FT.

In the Siblos study population of 358 men, the AR CAG repeat length ranged from nine to 33 (P25-P75:20 -24; median 22). Significant partial correlations (adjusted for age, BMI, and smoking) were observed between the CAG repeat length and both T (r = 0.154; P = 0.004) or calculated FT (r = 0.226; P < 0.001) levels. There was no significant correlation with LH or SHBG.

Quartile analysis showed a progressive increase of FT and T with CAG repeat length (ANCOVA: P < 0.001 and P = 0.05, respectively) (Fig. 2, B and D), with the adjusted mean of T and respectively) (Fig. 2, B and D), with the adjusted mean of T and FT in Q ₄ being in average 8.5 and 14% higher than in Q₁, respectively. There was no difference across quartiles of CAG for LH (P = 0.580) and SHBG (P = 0.675).

Discussion

Whether considering either the extremes or the whole range of (F)T concentrations in the Belstress population of 2322 generally healthy men, we observed that higher (F)T levels are consistently associated with higher LH levels. These observations were confirmed in the independent younger Siblos study population. These findings are compatible with the hypothesis that between-subject variation of serum (F)T results in part from differences in androgen sensitivity and negative feedback loop set point. Furthermore, the hypothesis that the interindividual variation in serum (F)T reflects differences in androgen sensitivity is supported by the finding in both study populations of a positive association of (F)T with CAG repeat length in the AR gene, this single polymorphism explaining 6-8.5% of T and 10-14% of FT variability.

In men with functionally intact hypothalamo-pituitary compartment of the gonadal axis, serum LH levels are a marker of tissue exposure to sex steroids at the hypothalamo- pituitary level, with the combination of a low or normal serum T together with a high LH being the hallmark of primary deficiency of testicular Leydig cell secretory capacity. The association of an elevated serum LH together with increased serum (F)T is an expected and classical finding in patients with functional mutations of the AR and patholog-ically decreased androgen sensitivity. It has been suggested that determination of the product of serum LH and T con-centrations might be useful in identifying patients with min-imal forms of androgen insensitivity (18). However, it should be recognized that the combination of elevated LH and T levels is not pathognomonic of androgen insensitivity as other situations affecting the LH feedback regulation, in particular estrogen deficiency, can induce similar changes (19). In our study, it is unlikely that the observed positive association between LH and FT levels would have been underlain by variation in estrogen feedback action. Indeed, it has been shown that circulating estrogen levels rather than local aro- matization of T in the hypothalamo-pituitary tissues is critical for estrogen feedback regulation of LH secretion (20), and when comparing the Belstress subgroups with low and high FT, we found that the latter group had not only higher LH but also higher serum E₂ concentrations. The latter higher serum E₂ is likely explained by higher T substrate availability for aromatization and possibly also plays a role in the observed higher SHBG concentrations in the subgroup with high FT.

Our study is based on single time point sampling in sizable populations. It is unlikely that temporal association of LH and T secretion within individuals would have contributed substantially to the observed cross-sectional association of LH and FT in the study population. Indeed, time of sampling was random relative to occurrence of endogenous LH pulses, and whereas intraindividual concentrations of LH and T are associated, this is with a time shift (21). Furthermore, single time point T levels have been reported to be fairly repre-sentative of midlong term mean serum T (6, 7).

The length of the CAG repeat of the AR affects AR mRNA and protein levels (10, 22) as well as transcriptional efficacy of the AR through changes in conformation and association with coactivators (10, 23). This AR polymorphism has been reported to affect tissular androgen action in men (24-26) and prostate cancer risk (27, 28). Along the same line, the CAG repeat polymorphism of the AR might be expected to affect hypothalamic-pituitary feedback regulation with longer CAG repeats being associated with diminished androgen feedback and relative elevation of circulating (F)T, as was indeed observed in our two study populations. Previous limited literature data on circulating T in relation to CAG repeat length have not been univocal. Mifsud et al. (11) reported a significant (P = 0.046) bivariate positive correlation between CAG repeat length and FT in a small group of 91 fertile men with mean age of 35.6 yr but did not discuss the impact of confounders. Walsh et al. (14) reported a nonsignificant trend of increasing bioavailable T according to ter-tiles of CAG repeat length in 294 men aged 55-93 yr; a significant similar trend in a second cohort of 171 men aged 19-90 yr appeared to be restricted to the older subjects. In a cohort of 882 men aged 40-70 yr from the Massachusetts Male Aging Study, Krithivas et al. (29) found no association between CAG repeat and T levels at baseline evaluation (mean age 53 ± 8 yr), although they did report a significantly sharper decline of (F)T on follow-up examination (mean age 61 ± 8 yr) in those men with shorter CAG repeat length of the AR. No significant association between serum T and CAG repeat was found in three other independent studies by Van Pottelbergh et al. (30) in 273 healthy community-dwelling men over 70 yr of age, Harkonen et al. (25) in 172 men aged 41-70 yr, and Canale et al. (31) in 91 healthy men aged 23-45 yr. Apparent discrepancies between studies may be underlain by differences in population genetic background, differences in subjects selection criteria, and small sample size for several of these studies, whereas in the older men, lack of association might be explained by age-related alterations in androgen feedback regulation (30).

In any case, the consistent positive association between CAG repeat length and (F)T in the present study as observed in two independent, sizable and more homogenous study populations is predicted and biologically plausible. The 1014% of between-subject variability of FT explained by the CAG repeat polymorphism in the Belstress and Siblos study populations, respectively, is not negligible and is in the same order of magnitude as the effects of well-established modi-fiable factors affecting serum T levels, such as smoking (3). Moreover, from a genetic perspective, serum (F)T is un-doubtedly a polygenic trait and the polymorphism of the AR is obviously only one of many potentially contributing ge-netic factors underlying the substantial heritability of serum (F)T (5).

In the Belstress population, the CAG repeat length was also associated with serum LH, although not consistently progressive over the four quartiles of CAG repeat length. This association was not observed in the Siblos population. In view of the much higher moment-to-moment variability of serum levels of LH, compared with (F)T, to demonstrate more consistently an association between AR CAG repeats and serum LH would probably require repeated sampling and/or larger study populations.

Results of this study, which support the view that inter-individual differences in serum (F)T in healthy men partly reflect subtle differences in androgen sensitivity, strongly suggest that healthy men with (F)T in the upper normal range do not necessarily have a higher tissular androgen activity than those with lower (F)T levels. These findings have obvious potential implications as to the interpretation of epidemiological reports on associations of T levels within the normal range with clinical and biological parameters, and they might help explain some apparently inconsistent find-ings. For example, below and above this normal range of serum T in men, a higher T tend to be associated with less favorable cardiovascular risk factors [e.g. with lower high- density lipoprotein (HDL)-cholesterol)], whereas, on the con-trary, within the normal range, higher serum T is associated with more favorable cardiovascular risk factors (e.g. higher HDL-cholesterol) without having demonstrable positive ef-fects in terms of clinical cardiovascular end point (32). If a higher serum T within the normal range does not reflect a higher androgen activity but is accompanied by higher E₂ levels, this might explain seemingly paradoxical findings such as a positive association of higher serum T with higher HDL-cholesterol seen only within the window of T levels around the normal range. When the feedback regulation of androgen levels is not fully operational, as is the case in elderly men (33), failure to compensate for AR CAG repeat polymorphism-related differences in androgen sensitivity might then translate into differences in androgen action. In this light Canale et al. (31) reported that given similar circu-lating T levels in hypoandrogenized men and controls, the final net androgenic phenotypical effect is related to AR polymorphism.

Our study results imply that assessment of androgen re-ceptor polymorphisms may provide additional information when studying associations between clinical variables and androgen levels in men. Conversely, it seems preferable to take into account serum T levels when studying associations of clinical variables such as cardiovascular risk factors and the AR CAG repeat polymorphism, as was done in some (33) but not all (34, 35) studies. Whether and how integrated interpretation of serum (F)T level and CAG repeat polymor-phism can have useful applications in the clinical management of individuals with suspected alteration of the androgen status remains to be established in studies with clinical end points.

Limitations of our study are those inherent to epidemio-logical studies with single time point assessment, in particular as far as the hormonal levels are concerned, and in the fact that in this explorative study a single candidate gene was considered. A strength of the study is that we had access to a large cohort of men, all sampled in the first part of the morning and for whom data were available to adjust our analyses for major confounders. Furthermore, we confirmed our findings on the relationship of FT with LH and AR CAG repeat length in a second, independent group of younger men.

In conclusion, this report presents evidence that between- subject variability in serum (F)T in healthy men is underlain in part by differences in androgen sensitivity, with a con-tributory role of the polymorphic glutamine tract in the AR encoded by the CAG repeat polymorphism in exon 1 of the AR gene. These findings have potential implications for in-terpretation of epidemiological studies, diagnosis of hypo-gonadism in borderline situations, and possibly individual-ization of androgen therapies in men.

Acknowledgements

The authors are grateful to K. Toye, M. Daems, R. De Muynck, H. Myny, H. Vlieghe, M. Masschelin, I. Bocquaert, and K. Mertens for their excellent technical assistance.

References

Vermeulen A, Kaufman JM, Giagulli VA 1996 Influence of some biological indexes on sex hormone-binding globulin and androgen levels in aging or obese males. J Clin Endocrinol Metab 81:1821-1826 

Ukkola O, Gagnon J, Rankinen T, Thompson PA, Hong Y, Leon AS, Rao DC, Skinner JS, Wilmore JH, Bouchard C 2001 Age, body mass index, race and other determinants of steroid hormone variability: the HERITAGE Family Study. Eur J Endocrinol 145:1-9 

Kaufman JM,Vermeulen A 2005 The decline of androgen levels in elderly men and its clinical and therapeutic implications. Endocr Rev 26:833-876 

Meikle AW, Stringham JD, Bishop DT, West DW 1988 Quantitating genetic and nongenetic factors influencing androgen production and clearance rates in men. J Clin Endocrinol Metab 67:104-109 

Ring HZ, Lessov CN, Reed T, Marcus R, Holloway L, Swan GE, Carmelli D 2005 Heritability of plasma sex hormones and hormone binding globulin in adult male twins. J Clin Endocrinol Metab 90:3653-3658 

Kaufman JM, T'Sjoen GG, Vermeulen A 2006 Androgens in male senescence. In: Nieschlag E, Behre HM, eds. Testosterone, action, deficiency, substitution. Cambridge, UK: Cambridge University Press; 497-541 

Vermeulen A, Verdonck G 1992 Representativeness of a single point plasma testosterone level for the long term hormonal milieu in men. J Clin Endocrinol Metab 74:939-942 

Andersson AM, Carlsen E, Petersen JH, Skakkebaek NE 2003 Variation in levels of serum inhibin B, testosterone, estradiol, luteinizing hormone, follicle- stimulating hormone, and sex hormone-binding globulin in monthly samples from healthy men during a 17-month period: possible effects of seasons. J Clin Endocrinol Metab 88:932-937 

Chamberlain NL, Driver ED, Miesfeld RL 1994 The length and location of CAG trinucleotide repeats in the androgen receptor N-terminal domain affect transactivation function. Nucleic Acids Res 22:3181-3186 

Albertelli MA, Scheller A, Brogley M, Robins DM 2006 Replacing the mouse androgen receptor with human alleles demonstrates glutamine tract length- dependent effects on physiology and tumorigenesis in mice. Mol Endocrinol 20:1248-1260 

Mifsud A, Sim CKS, Boettger-Tong H, Moreira S, Lamb DJ, Lipshultz LI, Yong EL 2001 Trinucleotide (CAG) repeat polymorphisms in the androgen receptor gene: molecular markers of risk for male infertility. Fertil Steril 75: 275-281 

Zitzmann M, Nieschlag E 2003 The CAG repeat polymorphism within the androgen receptor gene and maleness. Int J Androl 26:76-83 

Mifsud A, Choon AT, Fang D, Yong EL 2001 Prostate-specific antigen, testosterone, sex-hormone binding globulin and androgen receptor CAG repeat polymorphisms in subfertile and normal men. Mol Hum Reprod 7:1007-1013 

Walsh S, Zmuda JM, Cauley JA, Shea PR, Metter EJ, Hurley BF, Ferrell RE, Roth SM 2005 Androgen receptor CAG repeat polymorphism is associated with fat-free mass in men. J Appl Physiol 98:132-137 

Coetsier P, De Backer G, De Corte W 1996 Etude belge du stress au travail: apercu du modele de recherche et des outils d'investigation. Psychologie Psychometrie 17:17-35 

Vermeulen A, Verdonck L, Kaufman JM 1999 A critical evaluation of simple methods for the estimation of free testosterone in serum. J Clin Endocrinol Metab 84:3666-3672 

Van Uytfanghe K, Stockl D, Kaufman JM, Fiers T, De Leenheer A, Thienpont 

LM 2005 Validation of 5 routine assays for serum free testosterone with a candidate reference measurement procedure based on ultrafiltration and iso-tope dilution-gas chromatography-mass spectrometry. Clin Biochem 38:253-261 

Aiman J, Griffin JE, Gazak JM, Wilson JD, MacDonald PC 1979 Androgen insensitivity as a cause of infertility in otherwise normal men. N Engl J Med 300:223-227 

Hayes FJ, Seminara SB, DeCruz S, Boepple PA, Crowley Jr WF 2000 Aromatase inhibition in the human male reveals a hypothalamic site of estrogen feedback. J Clin Endocrinol Metab 85:3027-3035. 

Raven G, de Jong FH, Kaufman JM, de Ronde W 2006 In men, peripheral estradiol levels directly reflect the action of estrogens at the hypothalamo- pituitary level to inhibit gonadotropin secretion. J Clin Endocrinol Metab 91:3324-3328 

Veldhuis JD, King JC, Urban RJ, Rogol AD, Evans WS, Kolp LA, Johnson ML 1987 Operating characteristics of the male hypothalamo-pituitary-gonadal axis: pulsatile release of testosterone and follicle-stimulating hormone and their temporal coupling with luteinizing hormone. J Clin Endocrinol Metab 65:929-941 

Choong CS, Kemppainen JA, Zhou ZX, Wilson EM 1996 Reduced androgen receptor gene expression with first exon CAG repeat expansion. Mol Endocrinol 10:1527-1535 

Buchanan G, Yang M, Cheong A, Harris JM, Irvine RA, Lambert PF, Moore NL, Raynor M, Neufing PJ, Coetzee GA, Tilley WD 2004 Structural and functional consequences of glutamine tract variation in the androgen receptor. Hum Mol Genet 13:1677-1692 

Hiort O, Holterhus PM, Horter T, Schulze W, Kremke B, Bals-Pratsch M, Sinnecker GHG, Kruse K 2000 Significance of mutations in the androgen receptor gene in males with idiopathic infertility. J Clin Endocrinol Metab 85:2810-2815 

Harkonen K, Huhtaniemi I, Makinen J, Hubler D, Irjala K, Koskenvuo M, Oettel M, Raitakari O, Saad F, Pollanen P 2003 The polymorphic androgen receptor gene CAG repeat, pituitary-testicular function and andropausal symptoms in ageing men. Int J Androl 26:187-194 

Zitzmann M, Depenbusch M, Gromoll J, Nieschlag E 2003 Prostate volume and growth in testosterone-substituted hypogonadal men are dependent on the CAG repeat polymorphism of the androgen receptor gene: a longitudinal pharmacogenetic study. J Clin Endocrinol Metab 88:2049-2054 

Hardy DO, Scher HI, Bogenreider T, Sabbatini P, Zhang ZF, Nanus DM, Catterall JF 1996 Androgen receptor CAG repeat lengths in prostate cancer: correlation with age of onset. J Clin Endocrinol Metab 81:4400-4405 

Hsing AW, Gao YT, Wu G, Wang X, DengJ, Chen YL, Sesterhenn IA, Mostofi FK, Benichou J, Chang C 2000 Polymorphic CAG and GGN repeat lengths in the androgen receptor gene and prostate cancer risk: a population-based case- control study in China. Cancer Res 60:5111-5116 

Krithivas K, Yurgalevitch SM, Mohr BA, Wilcox CJ, Batter SJ, Brown M, Longcope C, McKinlay JB, Kantoff PW 1999 Evidence that the CAG repeat in the androgen receptor gene is associated with the age-related decline in serum androgen levels in men. J Endocrinol 162:137-142 

Van Pottelbergh I, Lumbroso S, Goemaere S, Sultan C, Kaufman JM 2001 Lack of influence of the androgen receptor gene CAG-repeat polymorphism on sex steroid status and bone metabolism in elderly men. Clin Endocrinol (Oxf) 55:659-666 

Canale D, Caglieresi C, Moschini C, Liberati CD, Macchia E, Pinchera A, Martino E 2005 Androgen receptor polymorphism (CAG repeats) and andro- genicity. Clin Endocrinol (Oxf) 63:356-361 

Liu PY, Death AK, Handelsman DJ 2003 Androgens and cardiovascular disease. Endocr Rev 24:313-340 

Zitzmann M, Brune M, Kornmann B, Gromoll J, von Eckardstein S, von Eckardstein A, Nieschlag E 2001 The CAG repeat polymorphism in the AR gene affects high density lipoprotein cholesterol and arterial vasoreactivity. J Clin Endocrinol Metab 86:4867-4873 

Hersberger M, Muntwyler J, Funke H, Marti-Jaun J, Schulte H, Assmann G, Luscher TF, von Eckardstein A 2005 The CAG repeat polymorphism in the androgen receptor gene is associated with HDL-cholesterol but not with coronary atherosclerosis or myocardial infarction. Clin Chem 51:1110-1115 

Page ST, Kupelian V, Bremner WJ, McKinlay JB 2006 The androgen receptor gene CAG repeat polymorphism does not predict increased risk of heart disease: longitudinal results from the Massachusetts Male Ageing Study. Clin Endocrinol (Oxf) 65:333-339