| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Department of Endocrinology, Ghent University Hospital, Ghent B-9000, Belgium
Correspondence: Address all correspondence and requests for reprints to: Prof. Dr. Jean M Kaufman, Department of Endocrinology 9K12, De Pintelaan 185, Ghent, B-9000, Belgium. E-mail: jean.kaufman{at}ugent.be
 |
Abstract |
|---|
 Top Abstract I. IntroductionII. Sex Steroids in...III. Clinical Significance of...IV. Diagnosis of Androgen...V. Androgen Treatment in...VI. Summary and ConclusionsReferences |
|---|
|
I. Introduction |
|---|
|
TopAbstractI. IntroductionII. Sex Steroids in...III. Clinical Significance of...IV. Diagnosis of Androgen...V. Androgen Treatment in...VI. Summary and ConclusionsReferences |
|---|
Androgens are substances that determine the differentiation of male internal and external genitalia as well as the development and maintenance of male secondary sex characteristics and male reproductive function. Besides, they have important metabolic effects on protein, carbohydrate, and fat metabolism, and as such they contribute to the determination of muscle mass and strength and to that of bone and fat mass, while they indirectly also influence insulin sensitivity. Furthermore, androgens affect behavior and cognition.
It is thus not surprising that age-associated phenomena such as a decline in virility and sexual activity, a decrease in muscle mass and strength, or an increased tendency to develop atherosclerosis and impairment of glucose metabolism have been related to an observed decline in testicular function in aging men.
|
II. Sex Steroids in Elderly Men |
|---|
|
TopAbstractI. IntroductionII. Sex Steroids in...III. Clinical Significance of...IV. Diagnosis of Androgen...V. Androgen Treatment in...VI. Summary and ConclusionsReferences |
|---|
-reduction of testosterone in peripheral tissues (5). Fifteen percent of plasma androstenedione originates from peripheral conversion of DHEA and testosterone, whereas 85% is secreted directly in approximately equal parts by the testes and the adrenals (6, 7); DHEA and DHEAS originate almost exclusively from the adrenals. Biologically, the most important plasma androgen is testosterone. It is largely bound to plasma proteins, only 1–2% being free, 40–50% being loosely bound to albumin, and 50–60% being specifically and strongly bound to the SHBG (8, 9). Unbound testosterone diffuses passively through the cell membranes into the target cell, where it binds to the specific androgen receptor (AR) (10). The serum free testosterone (FT) and the albumin-bound testosterone represent the fractions readily available for biological action. Indeed, albumin-bound testosterone dissociates during tissue transit, whereas the strong binding of testosterone to SHBG will usually not allow for substantial dissociation during the tissue transit time (11). The non-SHBG-bound testosterone, i.e., the combined free and albumin-bound testosterone, is often referred to as the "bioavailable testosterone" (bioT). However, the fraction actually available for biological action may vary according to the considered tissue and pathophysiological condition, and at present a reliable clinical or biochemical marker of androgen activity at the level of the tissues is still lacking.
The clinical significance of plasma DHT is very limited because most DHT formed in peripheral tissue acts locally (12), only a limited fraction escaping to the circulation where DHT is strongly bound to SHBG, only 0.8% being free. Androstenedione and DHEA are loosely bound to albumin, the binding to SHBG being negligible; DHEAS on the other hand is relatively strongly bound to albumin (13).
Androgenic actions of testosterone are mediated via binding to the AR, either directly or after 5-reduction to DHT, whereas part of the physiological actions of testosterone results from its aromatization to estradiol, which binds to estrogen receptors (ERs). The AR does not bind substantially androstenedione, DHEA, or DHEAS, and it is assumed that the androgenic effects of these steroids are attributable to their transformation to testosterone in the tissues. Recently, an endothelial plasma membrane DHEA binding site has been described, which still requires, however, functional proof of receptor activity (14). There is also evidence for DHEA interaction with the 
-aminobutyric acid receptor (15).
Testosterone can also exert rapid, nongenomic effects, in part via binding to a G protein-coupled membrane receptor for the SHBG-testosterone complex that initiates a cAMP-mediated, transcription-independent signaling pathway affecting calcium channels (16, 17, 18). Recently, Braun and Thomas (19) reported the presence of a high-affinity membrane AR in the Atlantic croaker.
B. Influence of aging on blood concentrations
1. Testosterone.
In healthy adult males, morning levels of serum testosterone vary between around 315 and 1000 ng/dl (11 and 35 nmol/liter) (20), the blood production rate [mean concentration multiplied by metabolic clearance rate (MCR)] ranging from 4 to 10 mg/d (14 to 35 µmol/d) (21). Plasma levels show circadian variations with amplitude of approximately 35%, highest levels in the morning and lowest levels in the late afternoon (22). Although there are also ultradian variations in testicular secretion of testosterone as a consequence of episodic stimulation by pulsatile pituitary secretion of LH, discrete testosterone secretory episodes are usually not clearly identified in peripheral blood (23).
As early as 1958, Hollander and Hollander (24) reported a decrease of spermatic vein testosterone concentration in elderly men and, soon afterward, Kent and Acone (25) reported an age-associated decline in blood production rate, which was subsequently confirmed by several other authors (26, 27, 28). However, this does not necessarily translate into lower plasma levels because the MCR also decreases with aging in men (27). Nevertheless, in the early seventies several authors reported an age-associated decline of serum testosterone levels from the fourth or fifth decade of life on. Although this has long been controversial, this decline has now been confirmed both by a large series of cross-sectional studies (for review, see Ref.29) and by several longitudinal studies (30, 31, 32, 33) (Fig. 1
). In fact, the age-associated decrease appears more important in the longitudinal than in cross-sectional studies, which might be explained by a bias toward healthier subjects in the former, whereas community-dwelling elderly are more likely to show a deterioration than an improvement of their health status during follow-up (33).
|
and 3) (33, 34, 35, 36, 37). Moreover, in the elderly the amplitude of the circadian variation of serum testosterone, FT, and bioT is reduced (34, 38, 39, 40).
|
|
and 3). Nevertheless, there is at all ages an important between-subject variability of serum FT or bioT levels (Fig. 4). In the age group above 60 yr, about 20% have serum testosterone levels in the upper normal range for young men, whereas over 20% have testosterone levels below the range for young males and an even larger proportion have levels below this limit when considering FT or bioT (41) (Fig. 5). In the Baltimore Longitudinal Study of Aging (31), 19, 28, and 49% of men over ages 60, 70, and 81 yr, respectively, had total testosterone levels below the young reference range, whereas 34, 68, and 91% had subnormal levels when the FT index (FTI: T/SHBG) was used. However, the latter index is not a valid measure of FT in males, and reference values for young males may not be applicable to the elderly (42, 43).
|
|
-reductase type 2 in the peripheral target tissues, whereas DHT that is formed in the liver under influence of 5-reductase type 1 is not released into the general circulation but is immediately glucuronoconjugated (44, 45); 5-reductase activity can be differentially regulated in different target tissues (46). Cross-sectional studies do not show substantial changes in serum levels of DHT in aging men, although due to the increase in SHBG there may be a modest decrease of the free fraction (35, 47). Nevertheless, there have also been reports of decreases (48, 49) as well as increases of serum DHT with age (33).
Plasma androstenedione levels vary between 60 and 230 ng/liter (2 and 8 nmol/liter) in adult males aged 20–30 yr and significantly decline with age (50) (Fig. 2); the levels show a circadian rhythm, with maximal concentrations in the morning (51). The androgenic activity of androstenedione is dependent upon biotransformation to testosterone (blood conversion rate around 15%) (6); its MCR is about 2000 liters/d (52), and its blood production rate lies around 1.5 to 2 mg/d.
Plasma DHEA and DHEAS are secreted almost exclusively by the adrenals. Only about 10% of DHEA is derived from the gonads, whereas about 50 to 70% derives from desulfatation of DHEAS in peripheral tissues (53).
DHEA metabolism is very rapid, with a MCR around 2000 liters/d (7). Serum levels are highly age-dependent, with mean levels of about 430 ng/dl (15 nmol/liter) at age 20 yr, decreasing to 140 ng/dl (5 nmol/liter) at age 75 yr (35, 54, 55, 56) (Fig. 2). The age-associated decline of serum DHEA contrasts with maintained or even increased serum cortisol concentrations during aging. The serum levels are subject to a circadian rhythm with highest values in the morning; the daily production rate amounts to 2 to 7 mg. The blood conversion rate to testosterone is about 0.6%, and hence, its contribution to plasma testosterone levels is negligible in adult men. However, because this conversion occurs in peripheral tissues where testosterone may act locally, the conversion rate does not reflect the contribution of DHEA to androgenic activity in the tissues.
DHEAS is by far the most abundant androgen in plasma. Its mean concentration in young males is about 220 µg/dl (6 µmol/liter), i.e., 10 to 20 times the concentration of cortisol, decreasing however rapidly with age (35, 54, 55, 56) (Fig. 2). Its metabolism is slow (MCR around 15 liters/d), and the blood production rate in young males lies as high as 25 to 30 mg/d (57). Due to its slow metabolism, plasma DHEAS levels do not show circadian variations. Its hormonal and metabolic effects are probably essentially attributable to its transformation to testosterone and estrogens in the tissues (58). The hormonal effects resulting from local biotransformation to these active sex steroids can presently not be quantified, but the contribution to the global androgenic and anabolic effects in men is probably modest. Indeed, it can be noted that glucocorticoid-substituted adrenal insufficiency in adult men does not result in clinically manifest hypoandrogenism, whereas conversely anorchid men have no substantial residual virilization.
3. Estrogens.
There is a rapidly growing body of evidence that a number of physiological actions of testosterone in men are mediated by the ERs after its biotransformation by the aromatase cytochrome P450 enzyme in the tissues (59). Documented estrogen-mediated actions of testosterone in men include a role in the feedback regulation of LH (60, 61), a role in the regulation of skeletal homeostasis (62, 63), as well as a role in lipid metabolism and cardiovascular physiology (64, 65); among other possible estrogen actions in men, there are indications for a role in the brain (66) and in spermatogenesis (67). These estrogenic actions in men can be exerted by blood-borne estrogens as well as through local aromatization of testosterone in, or in close vicinity of, the target cell. The expression of the CYP19 gene encoding the aromatase enzyme can be differentially regulated according to the tissue (68, 69).
The conversion rate of testosterone to estradiol is around 0.2%. Up to 80% of plasma estradiol originates from aromatization of testosterone and androstenedione, mainly in (sc) fat and striated muscle, although aromatase activity is present in many other tissues, including bone and the brain; no more than 20% of estradiol in the circulation is secreted by the testes. Estradiol serum concentration in the adult male is around 20 to 30 pg/ml (70 to 110 pmol/liter), with a production rate of around 45 µg/d. Plasma estradiol is also bound to SHBG but with only half the affinity of testosterone. Total plasma estradiol levels in adult men do not vary significantly with age; indeed the decrease in precursor levels (i.e., testosterone and androstenedione) is compensated by an increase of fat mass and tissue aromatase activity with age (36, 70, 71). As a consequence of the age-associated increase in SHBG binding capacity, the serum concentrations of free estradiol and non-SHBG-bound or "bioavailable" estradiol do show a moderate age-associated decrease (36, 71, 72) (Fig. 3). It can be pointed out that estrogen serum levels in elderly males are higher than those in postmenopausal women (63).
C. Mechanisms of the age-associated decline in blood androgen levels
There are three different aspects to the changes in serum testosterone levels in aging men: first, there are primary testicular changes with a diminished testicular secretory capacity; second, there is an altered neuroendocrine regulation of the Leydig cells with apparent failure of the feedback mechanisms to fully compensate; and third, there is an independent increase of SHBG binding capacity (for review, see Refs.41 and 73).
1. Primary testicular changes.
Stimulation with human chorionic gonadotropin (hCG) (74, 75, 76, 77, 78), with pulsatile administration of GnRH (79), or with biosynthetic LH after down-regulation of endogenous LH secretion with leuprolide (80) has consistently revealed a diminished secretory capacity of the Leydig cells in the elderly compared with young men. This decrease in testicular secretory reserve appears to involve a reduction of the number of Leydig cells (81, 82, 83, 84).
In old rats at least, all enzymes involved in the synthesis of testosterone are decreased with aging, as is the steroidogenic acute regulatory protein, which is involved in the transport of cholesterol into mitochondria (85, 86, 87). There is also evidence for a shift in testicular androgen biosynthesis favoring the 
4 over the 5 steroids, analogous to the situation for the adrenals (88). In healthy, community-dwelling men over age 75 yr, mean testicular volume is reduced by about 30% relative to that in young men (89).
2. Altered neuroendocrine regulation of Leydig cell function.
Consistent with a primary testicular cause of decreased testosterone production, mean serum LH levels in the male population tend to increase with age (for review, see Refs.29 and 90), but this increase is of modest amplitude and is inconsistent (30). Many elderly men with a serum testosterone concentration below the range for young men do not have elevated LH levels. Moreover, the modest increases in basal serum LH in elderly men appear to result in part from a slower plasma clearance rather than from increased pituitary secretion (91, 92).
Aging in men is thus also accompanied by manifest alterations in the regulation of LH secretion, the regulation of FSH secretion being better maintained (89, 93). This is conceptually of significance. Indeed, albeit the testicular secretory reserve is diminished in the elderly, there is a residual secretory capacity that should allow many elderly men to substantially raise their serum testosterone levels, provided there is an adequate LH drive.
Assessment of the secretory capacity of the pituitary gonadotropes by in vivo challenge with small "near physiological" doses of synthetic GnRH has revealed a maintained (79) or, in accordance with a state of relative hypoandrogenism, even a slightly increased LH response as measured by either immunoassay or bioassay in the elderly compared with young men (91). Given that the pituitary secretory capacity is preserved in the elderly, the apparent failure of the feedback regulatory mechanisms to produce an adequate rise of serum LH must result from changes at the hypothalamic level.
The pulsatile secretion of LH, governed by episodic release of hypothalamic GnRH into the pituitary portal circulation, has in the elderly an increased irregularity compared with young men (94) with essentially unchanged (95, 96, 97) or slightly increased (98) LH pulse frequency, but with a diminished frequency of large amplitude LH pulses and a reduced mean LH pulse amplitude, which is a parameter of the stimulating effect on the Leydig cells (95, 98). By inference, it can be assumed that the diminished amplitude of LH pulses in the elderly reflects a reduced size of the GnRH bolus intermittently released into the pituitary portal circulation, which might in turn be the consequence of a reduced number of functional hypothalamic GnRH neurones, of a less efficient intermittent recruitment and/or synchronization of these neurones and/or of a down-regulation of their activity by local or systemic factors. As to the latter, an important observation is that elderly men have an increased responsiveness to the negative feedback effects of androgens (95, 96, 99, 100) (Fig. 6). It has been shown that this is not the consequence of an increased hypothalamic opioid tone (101, 102), whereas the existence of a relative leptin deficiency has also been excluded as a possible cause of the hypothalamic changes in the regulation of LH secretion in the elderly (103). The exact mechanisms underlying the decrease in hypothalamic GnRH secretion in elderly men is yet to be fully elucidated.
|
The substantial age-related increase of SHBG (about 1.2%/yr) is remarkable because it occurs in the face of an increase of fat mass and insulin levels, which are strong negative determinants of SHBG levels (35, 105, 106, 107, 108). It seems unlikely that the decreased plasma testosterone or the associated decrease of the plasma testosterone over estradiol ratio would per se be the cause, because the increase of SHBG seems to begin at a younger age (35). Presently, the mechanisms responsible for the age-associated increase of serum SHBG are yet to be uncovered. Involvement of the age-related decline in the activity of the somatotropic axis is an attractive, but yet to be validated working hypothesis presently supported only by indirect evidence, such as the existence of a negative association between serum levels of SHBG and IGF-I and the normalization of elevated serum SHBG levels in adults with GH deficiency or GH resistance during administration of substitutive doses of GH or IGF-I, respectively (35, 109, 110).
4. Mechanisms of decreased adrenal androgen secretion.
The androgen-selective decrease in plasma levels of adrenal steroids with maintained or even increased serum cortisol levels in aging men is not gender-specific, although there may be some sexual dimorphism in the observed patterns of age-related decline (111). The changes seem to be the consequence of a selective decrease in functional zona reticularis cells (112). It has been observed that after acute stimulation with ACTH, the serum DHEA response is manifestly diminished in the elderly, whereas the cortisol response is similar to that in young men (113, 114). During 3-d stimulation with 40 U of depot ACTH twice daily, increase of DHEAS in the elderly was proportional to the decreased basal level (115), which is compatible with the concept of a diminished mass of responsive cells with maintained responsiveness of the residual cells.
D. Factors affecting blood androgen levels in the healthy elderly
Healthy males show a slow and steady age-associated decline in plasma levels of testosterone and nonspecifically bound testosterone with, however, at all ages large between-subject variations. Although the mechanisms of this variability have not been completely elucidated, several physiological and lifestyle-related factors appear to play a role.
1. Intrasubject variability and random effects.
There have been reports of circannual variations in plasma testosterone with amplitude of up to 30% and maximum around October to December for studies performed in the Western hemisphere (116, 117, 118). But the reports are not consistent with some studies finding no significant variation or maximum levels rather in spring or summer (for review, see Ref.117). At present it is also not possible to differentiate between potential contributory factors such as latitude, climate, and/or diet. In any case, the large between-subject variability in serum testosterone was also seen when the study design avoided seasonal effects (73).
The circadian variation should not contribute substantially to the interindividual variability of serum testosterone levels as long as serum testosterone is being consistently evaluated in the first part of the morning (around 0700 to 1000 h). Moreover, although persisting to some extent, the circadian variation of serum testosterone is markedly blunted in the elderly (34, 38, 40, 63).
Because testosterone is being secreted episodically, one could assume that part of the between-subject variability in serum testosterone represents random effects underlined by the moment-to-moment intraindividual variation in serum testosterone. To some extent this is certainly the case, but moment-to-moment variations in circulating testosterone are rather limited and discrete pulses of testosterone are usually not discernible, at least not during daytime (119). In any case, in middle-aged men, single point measurements of testosterone have been found to reflect rather well the long-term hormonal levels (120), and we found that in 248 healthy community-dwelling men over age 70 yr, two single point measurements of total testosterone at a 1-yr interval were strongly correlated (r = 0.82) (73). Therefore, to apply single point measurements is an acceptable approach for the purpose of epidemiological studies. Nevertheless, individual elderly men have been reported to present week-to-week variations in serum testosterone and bioT, which could result in misclassification of men as having normal or low serum levels relative to the reference range if based on a single time point measurement (121).
2. Ethnicity and heredity.
Heredity plays an important role as shown by studies in twins by Meikle et al. (122, 123), which revealed that genes determine as much as 25–76% of the total variation of plasma levels of gonadotropins, testosterone, FT, estradiol, and estrone. In these studies, only 12% of the variation in serum DHT levels was explained by heredity, but there appears to be a strong genetic influence (over 40%) in the tissue formation and the production rate of DHT. Nongenetic, familial factors may also substantially contribute to the determination of plasma hormone levels, e.g. for SHBG (124). The genetic basis underlying the heredity of testosterone and FT is presently unknown. Considering the complexity of testosterone synthesis and the regulation of its secretion, there obviously is a broad range of candidate genes (125).
Ethnic variations in serum testosterone have been reported, and in particular observations of slightly higher total testosterone and SHBG levels in men of African origin compared with Caucasian. These small differences of a few percent tend to disappear when full adjustments are made for body composition, including adjustment for some measure of abdominal adiposity. There are generally no differences in serum FT (126, 127, 128). No consistent differences are found in serum testosterone levels between Caucasians and men of Asian origin (129, 130); lower serum total testosterone, SHBG, and FT has been reported in a group of Pakistani compared with Caucasian and African-Caribbean men residing in England, part of the observed difference being again explained by differences in abdominal adiposity (131).
Recently, there has been considerable interest in a possible role of an AR gene polymorphism. The AR gene contains in exon 1 a polymorphic trinucleotide CAG-repeat, which encodes a functionally relevant polyglutamine tract of variable length. A CAG-repeat length exceeding the normal range of 15–31 results in a diminished AR transactivation capacity (132). In X-linked spinal and bulbar muscular atrophy (Kennedy’s disease), the CAG-repeats in the AR exceed 40, and the clinical picture includes signs of mild androgen resistance (133, 134, 135). Evidence from in vitro studies indicates that androgen activity might also be affected by variations in AR CAG-repeat lengths within the normal range (136, 137). Clinical studies have associated shorter repeat lengths with higher prevalence of several androgen-sensitive diseases, including prostate cancer (138), although this is not confirmed in all studies; a polymorphic GGC-repeat encoding a polyglycine tract in the AR has also been associated with prostate cancer (139). An association of shorter AR CAG-repeat length with greater longitudinal decline of serum testosterone and bioT levels has been reported for a subgroup of middle-aged men in the Massachusetts Male Aging Study (MMAS), although there was no association found between CAG-repeat length and baseline serum levels of either testosterone or FT, nor was there a consistent association with follow-up hormone levels (140). No association was found between the polymorphic AR CAG-repeat and sex steroid serum levels in a cohort of community-dwelling healthy men over age 70 yr in Belgium (141), in accordance with the lack of association between this polymorphism and serum testosterone levels in a study in Chinese and Australian men (142), in studies in German men (143, 144, 145), and in a study in Finnish men (146). As to the LH levels, which might be expected to vary according to differences in androgen sensitivity at the hypothalamic-pituitary level, there are no consistent findings with lack of association between AR CAG-repeat length and LH levels in some studies (140, 141), weak positive associations between repeat length and LH reported by others (125, 143), or even observations of a negative association in a Finnish cohort, although no longer significant after adjustment for age (146). From the whole of these studies, it can be concluded that the AR CAG-repeat polymorphism does not appear to contribute substantially to the determination of androgen levels in elderly men. On the other hand, as mentioned in other sections of this review, there have been reports of associations of the AR CAG-repeat polymorphism with clinical parameters modulated by androgen action.
In a cohort of elderly men in The Netherlands, serum total testosterone and bioT were not different in subjects with wild-type LH and those heterozygous for a polymorphic variant caused by point mutation-based substitutions of two amino acids (Trp8Arg and Ile15Thr) in the LH ß-subunit (147).
3. Fat mass and its distribution.
Adiposity as assessed by the body mass index (BMI) [i.e., body weight (in kilograms)/body height (in meters)2] is an important negative determinant of total serum testosterone levels, mainly via its effects on SHBG levels (105). The latter are in turn positively associated with insulin sensitivity, as indicated by the consistent finding of a negative correlation of SHBG with insulin serum levels (35, 48, 106, 107, 108, 131, 148, 149, 150, 151). Similarly, a negative association of serum SHBG and total testosterone with leptin levels was observed (103, 152), this negative association being (103) or not being (152) maintained after adjustment for BMI. Overall, negative associations with serum testosterone levels tend to be most pronounced for indices of abdominal adiposity (48, 107, 108, 149, 153, 154).
Although in young and middle-aged men, moderate obesity (BMI 
35 kg/m2) affects mostly the serum SHBG and total testosterone levels, morbid obesity (BMI > 35 kg/m2), and mainly abdominal fat accumulation, also affects FT levels (106, 107, 148) as a consequence of alterations in the neuroendocrine regulation of testosterone secretion, characterized by a decreased mean amplitude of LH pulses (106). Nevertheless, also within the normal variation of adiposity in the population, negative associations with FT levels can be observed, in particular in the elderly (48, 103, 108, 152). Adiposity is also associated with lower DHEA and DHEAS (35), although a positive association between DHEAS and BMI has been observed in the oldest-old (155).
When considering the effects of body composition on the androgen status in elderly men, one should keep in mind the caveat that the interrelation between androgen status and body composition is probably bidirectional with, as discussed in later sections of this review, androgen deficiency and androgen treatment, respectively, having substantial and opposite effects on body composition and insulin sensitivity.
4. Diet.
As far as the influence of diet is concerned, reports in the literature are rather divergent. In a study in elderly monks (34), plasma testosterone and FT levels were similar in vegetarian and nonvegetarian subjects. It has been reported (156) that Chinese in Beijing had lower serum testosterone and SHBG levels, but testosterone MCR similar to Chinese living in Pennsylvania and following a Western diet. Other data strongly suggest that fiber, lignan, and diets rich in isoflavone are associated with higher serum SHBG and total testosterone levels compared with Western diets (157, 158, 159, 160, 161), but FT levels may not be significantly different (159). In 1552 men aged 40–70 yr in the MMAS, fiber intake and protein intake were significant independent positive and negative determinants, respectively, of serum SHBG, whereas neither total caloric intake, nor fat or carbohydrate consumption contributed significantly; the lack of a significant role of carbohydrate and fat intake makes it unlikely that the effects of diet would be mediated only by changes in serum insulin levels (162). Interestingly, low protein intake is known to be associated with low serum IGF-I levels (163), which can be hypothesized to play a role in the age-related increase of serum SHBG (35, 109, 110).
5. Stress.
Stress evokes adaptive neuroendocrine reactions with, on the one hand, activation of the stress-responsive corticotropic, sympatho-adrenal, and somatotropic axes and, on the other hand, suppression of the gonadal axis through restraining of hypothalamic GnRH secretion (92, 164, 165, 166, 167). One of several possible mechanisms underlying the latter inhibition of GnRH secretion may involve corticotropin-stimulated secretion of endogenous opiates (168, 169, 170, 171).
Acute fasting for 48 or 84 h has been reported to result in a substantial reduction of serum testosterone in healthy men through reduced LH pulse frequency and amplitude (92, 164, 172), which can be reversed by pulsatile administration of GnRH (167) and may involve a specific metabolic signal rather than a nonspecific reaction to stress (173). Interestingly, elderly men appear to be relatively resistant to the metabolic stress of fasting compared with young men (92).
For several types of acute physical stress (e.g., temperature, pain, injury, strenuous exercise) or psychological stress, it has been reported that they can inhibit gonadal function (174, 175, 176, 177, 178, 179). For these various forms of stress, there is little information on whether the elderly may be less or more susceptible. In a study of young and older athletes completing a triathlon (lasting 9–12 h in young men and 11–16 h in older men), there were no differences in observed absolute decrease in serum testosterone levels (180). As assessed in a small group of subjects, insulin-induced hypoglycemia resulted in a significant decline of serum testosterone in healthy young men but not in elderly men, although the cortisol response was robust and even slightly greater than in the young (34). It has also been reported that serum testosterone levels appear more affected in the acute phase after myocardial infarction in middle-aged men with a mean age of 49 yr compared with elderly men with mean age of 70 yr (34). Overall, it does appear that in elderly men plasma testosterone levels, albeit lower than in young men, may be less susceptible to decrease in response to acute stress.
6. Other lifestyle-related factors.
Serum total testosterone and SHBG are reported to increase transiently during acute physical exercise of moderate intensity (181, 182), an effect that appears to result from hemoconcentration and decreased testosterone MRC (181, 183).
There is little information on the effects of dosed exercise programs on circulating androgen levels in elderly men, although such programs might have an effect on testosterone levels, in particular through changes in body composition and activity of the somatotropic axis. Deslypere and Vermeulen (34) found no consistent effect of a revalidation physical training program after myocardial infarction; Houmard et al. (184) observed no effects on serum androgen levels in middle-aged sedentary men of an exercise program that did result in significantly improved insulin sensitivity. Kraemer et al. (182) found a small increase of resting FT levels at the end of a 10-wk heavy-resistance training program in a group of eight men of 30 yr but not in men of 62 yr of age.
At all ages in adult men, serum testosterone and FT levels are 5–15% higher in (actual) smokers compared with nonsmokers (34, 35, 185, 186, 187). Smoking is also associated with higher DHEA and DHEAS serum levels (35, 187, 188).
Moderate alcohol consumption has no marked effect on serum testosterone (49, 162). Alcohol abuse, also in the absence of cirrhosis of the liver, accelerates the age-associated decline of serum testosterone and FT levels and is accompanied by increased estradiol levels (189, 190, 191, 192).
A controversial issue is whether and how sexual activity influences mean serum testosterone levels; the data available is inconsistent and does not allow for a conclusion (193, 194, 195, 196).
E. Contributory role of diseases and their treatment
As has been discussed in previous sections, aging per se induces a slow progressive decrease in plasma total and nonspecifically bound testosterone, with a substantial proportion of men over 65 yr who are in apparently good health presenting with serum levels below the reference range for young men. However, at all ages serum testosterone levels may be transiently or more permanently also affected by diseases or their treatment (41, 197, 198, 199, 200). Considering that the incidence and prevalence of diseases and consumption of medication with possible adverse effect on androgen status increase with age, in daily practice age-related decline in serum testosterone is not uncommonly accentuated by concomitant disease. An exhaustive inventory of the effects of disease on the androgen status falls beyond the scope of this review, and we limit the discussion to a few examples that are potentially relevant to the situation in elderly men.
1. Acute critical illness.
Acute critical illness, such as in patients with surgical trauma or myocardial infarction, is commonly associated with temporary, but often profound and prolonged decrease of serum testosterone levels (197, 201, 202, 203). The hypogonadism in critically ill men involves alteration in all compartments of the hypothalamo-pituitary-testicular axis, with observations of transient elevations of serum LH in the initial phase and of a state of hypogonadotropic hypogonadism during more prolonged illness (201, 202, 204, 205, 206).
2. Chronic systemic diseases.
Both testosterone and SHBG tend to decrease in elderly men with diabetes mellitus (207). Testosterone levels were found to be lower in men over 60 yr with type II diabetes than in nondiabetic controls (154, 208), which is in line with a negative association of serum testosterone with adiposity and insulin resistance in nondiabetic men.
Most cross-sectional studies report lower or similar testosterone levels in patients with coronary artery disease (CAD) compared with controls (209, 210, 211). Similarly, inverse associations between serum testosterone levels and other measures of atherosclerosis have been reported (212, 213). In a longitudinal study, Zmuda et al. (32) found a more rapid age-associated decline of testosterone levels in patients at increased risk for CAD. However, in neither case-control studies nor prospective cohort studies could an independent association between endogenous testosterone levels and subsequent development of fatal or nonfatal CAD be observed (for review, see Ref.214). An inverse correlation has been reported between (F)T levels and blood pressure (215, 216), possibly mediated by the abdominal obesity that is associated with atherosclerosis and increased blood pressure.
Chronic obstructive pulmonary disease (COPD) is often accompanied by decreased serum testosterone concentrations, also in the absence of systemic glucocorticoid treatment, with normal or decreased gonadotropin levels pointing toward hypothalamo-pituitary dysfunction; this is possibly the consequence of hypercapnea or of hypoxia (217, 218, 219, 220). It has been reported that men with obstructive sleep apnea not uncommonly present with low morning testosterone levels, which can be reversed by treatment with nasal continuous positive airways pressure (221, 222, 223).
Patients with chronic liver disease, with or without hepatic cirrhosis, tend to present with hypogonadism characterized by decreased FT levels and increased serum concentrations of SHBG, androstenedione, and estrogens (224, 225); alcohol has an additive effect on the hypogonadism of cirrhosis (226). Hypogonadism is also a classical feature of hemochromatosis, which is predominantly the consequence of a pituitary lesion caused by the iron overload (227).
Hypogonadism is a common sequel of chronic renal failure (228), which is underlied by a complex pathophysiology. On the one hand, mean gonadotropin levels are usually elevated, probably mainly a consequence of an increase of plasma half-life (229). On the other hand, there are abnormalities in LH secretion resulting from altered GnRH release (228, 229), and there is also evidence for the existence of impaired Leydig cell function (229).
A variety of endocrine diseases are known to induce hypogonadism, which includes primary testicular lesions, Cushing’s syndrome, prolactinoma as well as other secreting or nonsecreting pituitary tumors. There is a broad overlap between the symptomatology of normal aging, of hypogonadism, and of hypothyroidism, the latter being itself a possible cause of hypogonadism (230). Thyrotoxicosis takes a special position in that it is characterized, even in the absence of evident clinical signs of thyrotoxicosis, by greatly increased serum total testosterone levels, secondary to a marked increase of SHBG levels, with generally normal levels for FT (104, 230, 231).
As to diseases of more particular concern in some countries or ethnic groups, it can be mentioned that men with sickle-cell disease can present with low serum testosterone levels (232) and that in men previously treated for leprosy, hypogonadism is not uncommon as a sequel of Mycobacterium leprae-caused orchitis (233, 234, 235).
3. Drugs.
In the elderly, there is a rather high prevalence of use of medication and in particular frequent concomitant use of multiple drugs. Age-related decline of Leydig cell function, which may already be worsened by intercurrent disease may thus also be accentuated by use of drugs. A typical example of the latter situation is systemic administration of glucocorticoids in older men with COPD (236). Adverse drug actions on adult Leydig cell function and their underlying mechanisms have not been extensively studied. Here, we discuss only a few examples of potential adverse drug effects.
If only as a reminder, one has to mention here that the hormonal treatment of prostate cancer, the most common nondermatological cancer in men in Western countries, consists in suppression of endogenous testosterone production with use of a GnRH-analog and/or blockade of androgen effects by administration of an antiandrogen. As to the use of inhibitors of the 5-reductase in men with benign prostate hypertrophy, treatment with finasteride results in slightly elevated serum LH and testosterone levels (237), but the treatment mitigates androgenic effects in those tissues that are largely dependent on local production of DHT, which can result in mild symptoms or signs of hypogonadism (238). Chemotherapy with alkylating agents, which has major effects on spermatogenesis, can also result in mild Leydig cell insufficiency (239, 240).
Systemic glucocorticoids have dose-dependent adverse effects on testosterone production, which may result from both direct testicular effects and inhibition of gonadotropin secretion (220, 236, 241, 242).
Opiates and cannabinoids suppress testosterone production through inhibition of gonadotropin secretion (243, 244). Similarly, LH secretion and Leydig cell function may be adversely affected by hyperprolactinemia during chronic use of neuroleptic drugs and related compounds (245). Ketokonazol inhibits the c17/c20 lyase, decreasing testosterone synthesis (246), whereas spironolactone reduces the 17 hydroxylase-lyase activity leading also to reduced testosterone biosynthesis (247). Several classes of antihypertensive drugs, including ß-blockers, can interfere with normal erectile function, and hypertensive patients under treatment may often show a moderate reduction in serum testosterone levels (198, 248), but in view of the many confounding factors in these subjects, a drug effect on the testosterone levels cannot be considered as established.
F. Tissue levels of androgens and androgenic action
Although testosterone itself exerts androgenic and anabolic actions through binding to the nuclear AR in target cells, it is essentially a prohormone, being reduced to the more active androgen DHT in tissues expressing 5-reductase (type II in the urogenital tract and hair follicles; type I in skin and liver), whereas a fraction of testosterone can be aromatized to estradiol in tissues expressing the cytochrome P450 aromatase enzyme. Hence, action of testosterone is the complex resultant of tissue availability and locally achieved testosterone concentrations, of local testosterone metabolism, of expression of AR and/or ERs, as well as the expression of a number of coactivators and repressors of these receptors. Each of these determining factors of testosterone action can be differentially regulated in the tissues, thus offering a close to unlimited potential for differential regulation of sex steroid action in the tissues. This complexity of testosterone action should be kept in mind when discussing clinical implications of declining androgen levels in aging men and possible merits and risks of androgen treatment to the elderly.
Both the AR and ER (ER and/or ERß) are expressed in a wide range of tissues. The AR is found in the highest concentrations in male accessory sex organs (249) and in some areas of the brain, whereas lower concentrations of AR are measured in skeletal muscle (250), in the heart and vascular smooth muscle (251), and in the bone (62). Testosterone and DHT bind to the same receptor, but the affinity of DHT for the AR is greater than that of testosterone and in many tissues DHT mediates most androgenic effects of testosterone. In muscle, however, testosterone itself is the active androgen, 5-reductase activity being extremely low and no DHT formed (252).
Androgen concentration in target tissues depends on plasma concentration of bioavailable androgen, local androgen metabolism, and the presence of AR. As expected, total androgen (testosterone + DHT) concentration is highest in the prostate and scrotal skin, and higher in pubic skin than in striated muscle. In prostate and scrotal skin, DHT is quantitatively by far the major androgen, the DHT concentration being 5–10 times higher than that of testosterone; in muscle DHT levels are, as expected, extremely low (253) (Table 1). In the prostate, induction by DHT of type 2 5-reductase has as a consequence that almost all androgenic effects are exerted by DHT, enhancing the AR-mediated androgen effects (254).
|
The concentration of AR is influenced by androgens (increase), estrogens (increase of AR in prostate), and aging, which has been reported to be accompanied by a decrease of AR concentration in different tissues (257, 258, 259, 260, 261). Androgen sensitivity could be modulated by functional AR receptor polymorphisms, such as the aforementioned variation in functionally important polyglutamine and polyglycline tracts, encoded by a polymorphic trinucleotide CAG-repeat and GGC-repeat, respectively, contained in exon 1 of the AR gene (136, 137, 138, 139).
G. Androgen metabolism
As discussed in Section II.F, part of the metabolism of testosterone is activating, consisting in its conversion to the bioactive metabolites DHT and estradiol. Most testosterone entering prostate tissue is biotransformed to DHT, and in most tissues, with the important exception of muscle tissue, DHT is the principal active androgen, which acts mainly locally, only a small fraction escaping into the general circulation. Blood production rates of DHT and estradiol are lower than the total quantity of these steroids actually formed in the organism, a large fraction of locally produced hormone being further metabolized in situ.
Testosterone catabolism involves 5/5ß reduction of the double bond between carbons 4 and 5, 3/3ß reduction in ring A, and 17ß hydroxyl oxidation, this enzymatic degradation taking place to some extend in peripheral tissues and for a large part in the liver.
DHEA is first metabolized to androstenedione, under the influence of a 3ß-hydroxysteroid dehydrogenase/4,5 isomerase, the subsequent metabolism being identical to that of testosterone. The end metabolites of endogenous androgens, i.e., androsterone, etiocholanolone, and 5/5ß androstane-3,17ß diol are either glucuronidated under the influence of uridine diphosphate glucuronyltranferase or sulfated under the influence of a sulfokinase, these hydrosoluble conjugates being excreted by the kidneys (262). Androstanediol glucuronide (ADG) is considered by many as an important parameter of androgen action in women (263, 264), but in males its major precursor being testosterone (70%), with 30% deriving from DHEAS (265), determination of ADG does not offer much interest. The urinary excretion of ADG decreases significantly with age (266); the ratio of urinary 5/5ß metabolites decreases with age, a consequence of a decrease of 5 reductase type 2 activity (267).
|
III. Clinical Significance of the Age-Associated Decrease in Androgen Levels |
|---|
|
TopAbstractI. IntroductionII. Sex Steroids in...III. Clinical Significance of...IV. Diagnosis of Androgen...V. Androgen Treatment in...VI. Summary and ConclusionsReferences |
|---|
Does the decrease of androgen levels translate clinically? Arguments indicating such clinical significance could be sought in similarities between the symptomatology of aging and that of androgen deficiency in young hypogonadal men, as well as in associations between (severity of) symptoms and androgen levels. It should, however, be realized that aging is accompanied by a decline of almost all physiological functions such as cardiac output, pulmonary ventilatory capacity (both reducing work capacity), renal clearance, or GH and melatonine secretion, which in conjunction with age-associated changes in lifestyle such as retirement or relative sedentarism, may all contribute to the symptomatology of aging. The decrease in GH and IGF-I levels is associated with changes in lean body mass, bone density, and abdominal obesity, similar to the changes observed in hypogonadal states, whereas the age-associated decrease in melatonin secretion might play a role in the age-associated sleep disorders.
B. Similarities between symptoms of aging and hypogonadism in young men
Frequent clinical manifestations of aging in males are decreased libido and sexual activity or impotence; decreased virility, with decreased sexual body hair and beard growth; decreased energy, work capacity and cognitive function with, as objective signs, decreased muscle mass and strength; decreased bone mineral density (BMD), with increased fracture risk; increased (abdominal) obesity; and slightly decreased hematocrit. The latter changes are frequently accompanied by signs of altered tone of the autonomous nervous system as manifested by nervousness, insomnia, and sometimes hot flushes. The analogy with the general symptomatology of hypogonadism in young males is striking: impaired virilization with poor development of sexual body hair and beard growth, decreased bone and muscle mass with decreased physical strength, weakness, decreased libido, and often erectile dysfunction, abdominal obesity, and difficulty with concentration.
This symptomatology in the elderly develops, however, slowly and progressively, the symptoms being subtle, variable, and not specific. Hence, the clinical symptomatology does at best only suggest the possibility of a hypoandrogenic state in the elderly.
C. Associations between clinical manifestations of aging and sex steroid status
In view of the multifactorial origin of aging symptoms, strong correlations with FT or bioT levels can hardly be expected, whereas the multiplicity of contributing factors renders meaningful multivariate regression analysis difficult. Furthermore, cross-sectional association cannot establish causality, whereas prospective observational studies are rare.
1. Senile osteoporosis.
Aging in men is associated with continuous loss of bone and an exponential increase of the incidence of fractures of the hip (268, 269) and the spine (270, 271). Moreover, in older men the consequences of fractures in terms of morbidity and mortality appear to be more severe than in their female counterparts (272, 273). Acquired profound hypogonadism in men induces high bone turnover and accelerated bone loss (274), which has also been confirmed to be the case in the elderly (275, 276). However, whether and in how far an age-related partial androgen deficiency may be instrumental in senile osteoporosis in men is not definitely established.
Although the complex role of testosterone in the regulation of bone metabolism is yet to be fully elucidated, a large body of evidence allows us to conclude that besides direct androgen actions, aromatization to estrogen plays an important role in the regulation of both the acquisition of adult bone mass and the preservation of adult skeletal integrity (62, 63). Indeed, cross-sectional studies, in which age and BMI (or body weight) are major confounders, have yielded inconsistent results as to the association of serum testosterone levels with prevalent BMD in elderly men, with some studies failing to show an independent association (277, 278, 279), whereas others did show weak but significant positive correlations, in particular with FT or bioT (72, 280, 281, 282, 283, 284). On the other hand, in a series of recent cross-sectional studies, multivariate analysis consistently indicated that free or bioavailable estradiol is a better predictor of prevalent BMD in elderly men than FT or bioT (72, 283, 284, 285, 286, 287, 288, 289, 290, 291). Biochemical indices of bone turnover increase moderately with aging in men and are inversely related to prevalent BMD in the elderly (292), with markers of bone resorption more clearly (negatively) associated with serum levels of estradiol than with those of testosterone (63, 284, 289, 291, 292), although correlations are generally weak.
Cohort studies have shown that bioavailable estradiol is negatively associated with prospectively assessed bone loss, without independent association of serum (bioavailable) testosterone (284, 290, 293). Moreover, prospectively assessed bone changes in elderly men were also found to be associated with a polymorphism of the CYP19 gene that encodes the aromatase enzyme, independently of serum estradiol levels, suggesting indirectly that local aromatization of testosterone in bone tissue might play a role (290). Although there has been a report of association of a CAG-repeat polymorphism of the AR with BMD assessed by ultrasound in men aged 20–50 yr (143), in community-dwelling men over age 70 yr, no association was found between this polymorphism and either BMD or biochemical indices of bone turnover (141).
The important role of aromatization of testosterone to estrogen in the regulation of bone metabolism in elderly men has been further elegantly demonstrated in a short-term intervention study with selective manipulation of testosterone and estradiol levels (294). There is evidence indicating that there may be a threshold level for the association of bioavailable estradiol with bone loss and indices of bone metabolism (287, 293, 295, 296, 297), although other studies did not demonstrate such a threshold (290).
As to the association of sex steroid levels with fracture risk in elderly men, little data are available. In case-control studies, there was a higher prevalence of low serum testosterone in men recruited after a hip fracture (298, 299, 300), but testosterone levels assessed after a traumatic event should be interpreted with caution. In the Rancho Bernardo study, lower estradiol levels, but not serum testosterone concentrations, were associated with a higher prevalence of vertebral fracture in older men (301). In a case-control study in men aged 67.7 ± 6.8 yr as part of the Rotterdam Study, there was no significant association between vertebral fracture and either bioavailable estradiol or bioT (302). There was a preliminary report from the Mr. OS study in 5995 community-dwelling men 65 yr or older of an association of incident fracture risk with both bioT and bioavailable estradiol, with only the latter assocation being apparently mediated by a lower BMD (303). In community-dwelling men over age 70 yr, an aromatase gene polymorphism, associated with longitudinal changes in BMD, was also significantly associated with self-reported clinical fractures at the spine, hip, and/or wrist and with the occurrence of these fractures in their first-degree relatives, but there was no association of fracture history with circulating bioavailable estradiol (290).
In summary, the evidence suggests that declining sex steroid levels in the elderly may adversely affect the preservation of skeletal integrity and indicates that aromatization of testosterone to estradiol is a major component of the regulation of bone metabolism in the elderly, the skeletal effects of a relative testosterone deficiency in the elderly being modulated by aromatase activity, which in turn is affected by factors such as adiposity and heredity.
2. Body composition.
Aging is associated with important changes in body composition (108, 304, 305). In healthy men, fat mass was found to increase from around 22% of body weight in young men to around 30% in the elderly, lean body mass being 30% lower in the elderly group (108). Similar changes with a decrease of lean body mass and increase in fat mass are observed in hypogonadal men compared with age- and BMI-matched controls (306). It thus seems logical to suspect a role of the decreased androgen levels in the changes in body composition seen in elderly men.
In healthy men, there is a negative correlation between serum testosterone and FT with visceral fat (307), and in a study involving 61 middle-aged men and 271 elderly community-dwelling men aged 70–85 yr (108), BMI and fat mass were found to be negatively correlated to FT and IGF-I levels, the correlation of fat mass with FT persisting after correction for IGF-I and age. Multivariate analysis revealed that the negative correlation of FT with fat mass was determined primarily by abdominal fat mass. These findings are in agreement with findings by others (72, 192, 305). Khaw and Barrett-Connor (149) in a cohort study of 571 men aged 30–79 yr observed that low testosterone levels predict central obesity in men as estimated 12 yr later. The negative correlation of serum testosterone with abdominal fat might be related to the inhibition by testosterone of triglyceride uptake and lipoprotein-lipase activity in abdominal, but not in femoral, sc fat (308). Moreover, testosterone stimulates lipolysis and thus reduces fat storage in the fat cells (304).
The highly significant negative correlation between FT and (abdominal) body fat may be both a cause and a consequence of abdominal obesity in the elderly. Indeed, as discussed in Section II.D.3, increased adiposity is itself partially responsible for a decrease of testosterone levels (35). Moreover, decreased GH levels, as observed in elderly males (309, 310, 311, 312, 313) may also play a role in the age-associated changes in body composition, with GH substitution being possibly more effective than testosterone administration to reduce abdominal fat in elderly men (312).
The reduction of muscle mass by about 30% between ages 30 and 80 yr (313, 314, 315) is accompanied by a proportional decrease of muscle strength (316). This decreased muscle strength contributes to frailty and is a risk factor for falls, hip fractures, and loss of independent living conditions.
Again, the similarity with the occurrence of low muscle mass in hypogonadal young men raises the hypothesis that the age-associated decrease in androgen levels may be in part instrumental in the sarcopenia of elderly men. Few data concerning the correlation between endogenous androgen levels and muscle mass in elderly men are available, however. In an already mentioned study (108) involving middle-aged and elderly community-dwelling men, no correlation was found between testosterone or FT levels and lean mass. Similar negative findings were reported by van den Beld et al. (72) and by Roy et al. (317) in men aged 20–90 yr old, who observed nevertheless a positive association of FT levels with muscle strength. Szulc et al. (291) in a cross-sectional analysis in men 51–85 yr found that low FT is associated with lower muscle mass, functional impairment in the legs, and occurrence of falls in the past year. In institutionalized healthy elderly men, Abbasi et al. (318) observed a correlation between testosterone levels and severity of sarcopenia. Also, Baumgartner et al. (319) observed a positive association of FT with muscle mass, but not with muscle strength.
In summary, the literature indicates that the age-related decline of androgen levels does affect body composition, although the data on muscle mass and strength is limited and not consistent. Indeed, in aging males the interrelations between FT and muscle mass or strength are obscured by confounding effects of relative physical inactivity as well as by the effects of the age-associated decrease of somatotropic stimulation.
3. Atherosclerosis and CAD.
Atherosclerosis and CAD increase almost exponentially with age and are much more frequent in men than in premenopausal women (320), the gender gap narrowing after the age of 50 yr (321, 322). This gender gap with male preponderance is observed in all countries, notwithstanding widely divergent rates in cardiovascular mortality (322, 323). A gender difference is also observed in the plasma lipid profile: before menopause, women have higher high-density lipoprotein-associated cholesterol (HDL-C) and lower low-density lipoprotein-associated cholesterol (LDL-C) and lipoprotein (a) levels than men, whereas women with hyperandrogenism show a higher risk lipid profile (321, 324). In men, suppression of plasma testosterone with GnRH analogs increases HDL-C (325, 326), an effect that is abolished by coadministration of testosterone (326, 327). All this has led to the common belief that androgens are harmful and estrogens beneficial.
Review of the literature, however, does not provide convincing evidence that the gender difference can be explained by distinctive patterns of endogenous sex hormones (for review, see Refs.322 and 323). Indeed, of 32 cross-sectional studies on the relationship between endogenous testosterone levels and CAD (209), 16 showed lower testosterone levels and 16 showed similar levels in patients with CAD compared with controls. In more recent studies, Barrett-Connor (328), English et al. (329), as well as Hak et al. (212) observed a consistent inverse relationship between endogenous plasma testosterone levels and CAD. In a study involving 403 men aged 73–94 yr, van den Beld et al. (213) observed that serum testosterone levels, adjusted for age, were inversely correlated with intima media thickness (IMT) of the carotid artery (213). Although many of the subjects in the latter study suffered from angina pectoris, recent myocardial infarction, or diabetes mellitus, which might be responsible for both the decreased testosterone levels and CAD, the authors showed nevertheless that the associations of IMT with testosterone levels were as powerful in subjects free of CAD as in subjects with prevalent CAD. In a subgroup of 195 of these men, lower FT was also associated with greater progression of IMT, independently of cardiovascular risk factors (330).
Prospective cohort studies (32, 212) and case-control studies (214, 331, 332) involving thousands of patients did not show correlations of testosterone with CAD or a predictive value of testosterone levels for CAD (for review, see Refs.322 and 323).
As to the correlation of testosterone levels with risk factors for CAD in men, data in the literature suggest an inverse correlation between endogenous plasma FT or bioT and total cholesterol as well as LDL-C, fibrinogen, and plasminogen activator type 1 (for review, see Refs.322 and 323). Moreover, most studies (333, 334, 335) suggest a highly significant positive correlation between total testosterone and FT on the one hand and serum HDL-
