Sarcopenia, the loss of muscle mass and function, is an important consequence of aging (59-74). The prevalence of sarcopenia, depending on the definition used, varies from 10-30% in men over the age of 70 (55-56). The principal component of the decrease in fat-free mass is the loss of muscle mass; there is little change in non-muscle lean mass (60-66). Between 20 and 80 years of age, the skeletal muscle mass decreases by 35-40% in men (64), in part due to decreased muscle protein synthesis (69). Although there is a loss of both type I and type II fibers, there is a disproportionate decrease in the number of type II muscle fibers that are important for generation of power (70-71). In spite of the significant depletion of muscle mass, body weight does not decrease, and may even increase because of the corresponding accumulation of body fat (60-66) (Figure 3).

The loss of muscle mass that occurs with aging is associated with a reduction in muscle strength (72-75). There is a substantial decrease in strength and power between 50 and 70 years of age, primarily due to muscle fiber loss and selective atrophy of type II fibers (70-75). Loss of muscle strength is even greater after the age of 70; 28% of men over the age of 74 could not lift objects weighing more than 4.5 kg (74). With increasing age, there is a progressive reduction in muscle power (127-128), the speed of strength generation, and fatigability, and the ability to persist in a task.

Loss of muscle mass and strength leads to impairment of physical function, as indicated by the impaired ability to arise from a chair, climb stairs, generate gait speed, and maintain balance (127-130). The impairment of physical function contributes to loss of independence, depression, and dependency, and increased risk of falls and fractures in older men. Therefore, anabolic interventions that can reverse or prevent aging-associated sarcopenia are desirable.

The anabolic effects of testosterone on the muscle have been a source of intense controversy for over sixty years. The athletes and recreational body builders abuse large doses of androgenic steroids with the belief that these compounds increase muscle mass and strength. Until recently, the academic community was skeptical about such claims because of the problems of study design. However, a number of studies in healthy hypogonadal men, men with chronic illness, and in healthy older men have established that testosterone administration increases skeletal muscle mass, maximal voluntary strength, and leg power.

In a systematic review of testosterone trials in healthy, hypogonadal men, testosterone therapy increased fat-free mass by an average 2.8 kg and body weight by 1.1 kg (131-139)). Some studies of testosterone replacement therapy have reported significant improvements in maximal voluntary strength (134, 137), and decreases in whole body fat mass (133, 136-138). The administration of supraphysiologic doses of testosterone in eugonadal men also increases fat-free mass, muscle size, and maximal voluntary strength (139-142). Resistance exercise training augments the anabolic response to androgens; thus men receiving testosterone and resistance exercise together demonstrate greater gains in fat-free mass and muscle strength than those receiving either intervention alone (142).

The anabolic effects of testosterone on fat-free mass, muscle size, and maximal voluntary strength are related to the administered dose and the circulating testosterone concentrations (139-141) (Figure 4). Testosterone effects on muscle performance are domain-specific: testosterone administration increases maximal voluntary strength and leg power, but does not affect fatigability or specific tension (140). The gains in maximal voluntary strength during testosterone administration are proportional to the increase in muscle mass; unlike resistance exercise training, testosterone does not improve the contractile properties of human skeletal muscle (140).

Testosterone replacement of young, hypogondal men increases muscle protein synthesis (135, 143-144); the effects of testosterone replacement on muscle protein degradation need further investigation.

Systematic reviews (132, 145-146)) of randomized, placebo-controlled trials in HIV-infected men with weight loss (147-152) have revealed that testosterone therapy for 3 to 6 months was associated with greater gains in lean body mass than placebo administration (difference in lean body mass change between placebo and testosterone therapy 1.22 kg, 95% CI 0.23-2.22 for the random effect model). In two (147, 151) out of three trials that measured muscle strength (147, 151-152), testosterone administration was associated with significantly greater improvements in maximal voluntary strength than placebo. Testosterone therapy had a moderate effect on depression indices (-0.6, 95% CI -1.0, -0.2), but did not improve any domain of quality of life (153). Changes in CD4+ T lymphocyte counts, HIV copy number, PSA, plasma HDL cholesterol, and adverse event rates were not significantly different between the placebo and testosterone-treatment groups (147-152). Overall, short-term (3-6 months) testosterone use in HIV-infected men with low testosterone levels and weight loss can induce modest gains in body weight and lean body mass with minimal change in quality of life and mood. This inference is weakened by inconsistency of results across trials, heterogeneity in inclusion and exclusion criteria, disease status, testosterone formulations and doses, treatment duration, and methods of body composition analysis (132). There are no data on testosterone effects on physical function, risk of disability, or long term safety.

A number of placebo-controlled, randomized trials are in agreement that testosterone replacement increases fat-free mass and decreases fat mass in older men with low testosterone levels. In a meta-analysis (132) of double-blind, randomized trials (138, 143, 144, 154-160) that included middle-aged and older men>45 years of age with low or low normal testosterone levels and without an acute illness, and that used testosterone or its esters in replacement doses for >90 days, testosterone replacement was associated with a significantly greater increase in fat-free mass (contrast 2.5 kg, 95% confidence interval 1.5-3.4), right hand grip strength (contrast 3.3 kg, 95% confidence interval 0.7-5.8 kg), and a greater reduction in whole body fat mass (contrast -2.1 k, 95% CI -3.1,-1.1) than placebo (Figure 5). The change in body weight did not differ significantly between the testosterone and placebo groups (contrast -0.7 kg, 95% CI -2.0 to +0.6). Testosterone therapy was associated with significant improvements in self-reported physical function, as assessed by the SF-36 questionnaire. Changes in lower extremity muscle strength and performance-based measures of physical function were inconsistent across trials. One study reported no changes in physical function (157), while another reported improvement in a composite measure of physical function (160). These studies were performed in asymptomatic, healthy men; the effects of testosterone supplementation in older men with symptomatic functional limitations are unknown.

One reason for the failure to demonstrate improvements in physical function is that the measures of physical function used in previous studies were relatively insensitive and had low ceiling. The widely used measures such as 0.625 m stair climb, standing up from a chair, and 20-meter walk are tasks that require only a small fraction of an individual’s maximal voluntary strength. In most healthy, older men, the baseline maximal voluntary strength is far higher than the threshold below which these measures would detect impairment. Given the low intensity of the tasks used, the relatively healthy older individuals neither show impairment in these threshold-dependent, measures of physical function at baseline, nor an improvement in performance on these tasks during testosterone administration. Another confounder of the effects of anabolic interventions on muscle function is the learning effect. For instance, subjects who are unfamiliar with weight lifting exercises often demonstrate improvements in measures of muscle performance because of increased familiarity with the exercise equipment and technique. Because of the considerable test-to-test variability in tests of physical function, it is possible that previous studies did not have adequate power to detect meaningful differences in measures of physical function between the placebo and testosterone-treated groups. It is also possible that neuromuscular adaptations needed to translate strength gains into functional improvements require a lot longer than the 3 to 6 month duration of most of the previous trials. Therefore, while there is agreement that testosterone supplementation would be expected to produce a dose-dependent increase in muscle mass and strength, further studies are needed to determine whether testosterone-induced increases in muscle mass and maximal voluntary strength translate into clinically meaningful gains in physical function and health-related outcomes.

Testosterone-induced increases in muscle mass are associated with hypertrophy of both type I and II muscle fibers (161). The absolute number and the relative proportion of type I and type II fibers do not change during testosterone administration. Testosterone-induced muscle fiber hypertrophy is associated with dose-dependent increases in myonuclear number and satellite cell number (162).

The prevalent dogma assumes that testosterone supplementation increases muscle mass by stimulating muscle protein synthesis. Indeed, testosterone administration has been shown to increase fractional muscle protein synthesis and improve the reutilization of amino acids. The effects of testosterone on muscle protein degradation have not been well studied. However, the muscle protein synthesis hypothesis does not explain the reciprocal decrease in fat mass or the increases in myonuclear and satellite cell number that occur during testosterone administration (163). Singh et al (164-165) demonstrated that testosterone promotes the differentiation of mesenchymal multipotent cells into the myogenic lineage and inhibits the differentiation of these precursor cells into the adipogenic lineage. Thus, testosterone promotes the formation of myosin heavy chain II positive myotubes in multipotent cells and up-regulates markers of myogenic differentiation, such as MyoD and myosin heavy chain. Testosterone and DHT inhibit adipogenic differentiation and downregulate markers of adipogenic differentiation, such as PPAR- and C/EBP. The hypothesis that testosterone determines the lineage of pluripotent stem cells provides a unifying explanation for the reciprocal changes in muscle and fat mass observed during testosterone administration (164-165).

Androgen binding to androgen receptor induces a conformational change in the androgen receptor protein, promoting its association with its co-activator, beta-catenin, causing the complex to translocate into the nucleus (165). The androgen receptor – beta-catenin complex associates with TCF-4 and activates a number of Wnt target genes (165-166), thus promoting myogenic differentiation and inhibiting adipogenic differentiation.

Although the enzyme 5-alpha-reductase is expressed at low concentrations within the muscle (167-168), we do not know whether conversion of testosterone to dihydrotestosterone is required for mediating the androgen effects on the muscle. Men with benign prostatic hypertrophy who are treated with a 5-alpha reductase inhibitor do not experience muscle loss (168). Similarly, individuals with congenital 5-alpha-reductase deficiency have normal muscle development at puberty (168). These data suggest that 5-alpha reduction of testosterone is not obligatory for mediating its effects on the muscle. However, all the kindred with 5-alpha reductse deficiency that have been published to-date have had mutations of type 2 isoform of the enzyme. Similarly, finasteride is a weak inhibitor of only the type 2 isoform of the enzyme. The circulating concentrations of DHT in male patients with congenital mutation of type 2 5-alpha reductase enzyme or in men treated with finasteride are lower than eugonadal men; however, these patients still produce significant amounts of DHT and their circulating DHT concentrations are often in the lower end of the male range. It is reassuring that long term administration of dutasteride, a dual and potent inhibitor of both 5-alpha reductase isoforms, has not been associated with significant reductions in bone mineral density (169); the data on the effects of duatasteride on muscle mass are not available. This issue is important because if 5-alpha reduction of testosterone to DHT were not obligatory for mediating its anabolic effects on the muscle, then it might be beneficial to administer testosterone with an inhibitor of 5-alpha reductase or to develop selective androgen receptor modulators that do not undergo 5-alpha reduction.

Studies of aromatase knockout mice have revealed higher fat mass and lower muscle mass in mice that are null for the P450-linked CYP19aromatase gene (170). Similarly, humans with CYP19 aromatase mutations have decreased muscle mass and increased fat mass, and they exhibit insulin resistance (171). Data from these gene-targeting experiments suggest that aromatization of testosterone to estradiol might also be important in mediating androgen effects on body composition.

Sexual function in men is a complex process that includes central mechanisms for regulation of sexual desire and arousability, and local mechanisms for penile tumescence, orgasm, and ejaculation (172) (Figure 6). Primary effects of testosterone are on sexual interest and motivation (173-178). Testosterone replacement of young, androgen deficient men improves a wide range of sexual behaviors including frequency of sexual activity, sexual daydreams, sexual thoughts, feelings of sexual desire, spontaneous erections, and attentiveness to erotic stimuli (173-181). Kwan et al (177) demonstrated that androgen-deficient men have decreased frequency of sexual thoughts and lower overall sexual activity scores; however, these men can achieve erections in response to visual erotic stimuli. Hypogonadal men have lower frequency and duration of the episodes of nocturnal penile tumescence; testosterone replacement increases both the frequency and duration of sleep-entrained, penile erections (179-181). Although both orgasm and ejaculation are believed to be androgen-independent, hypogonadal men have decreased ejaculate volume and their orgasm may be delayed (Figure 6).

Although hypogonadal men can achieve erections, it is possible that achievement of optimal penile rigidity might require physiologic testosterone concentrations. Testosterone regulates nitric oxide synthase activity in the cavernosal smooth muscle (182). Testosterone administration in orchidectomized rats increases penile blood flow and has trophic effects on cavernosal smooth muscle (183-185).

In male rodents, all measures of mating behavior are normalized by relatively low testosterone levels that are insufficient to maintain prostate and seminal vesicle weight (186-187). Similarly, in men, sexual function is maintained at relatively low normal levels of serum testosterone (178, 188).

Erectile dysfunction and androgen deficiency are two common but independently distributed, clinical disorders that sometimes co-exist in the same patient (172, 189-191). Hypogonadism is a clinical syndrome that results from androgen deficiency (131); in contrast, erectile dysfunction is usually a manifestation of a systemic vasculopathy. Thus androgen deficiency and erectile dysfunction have distinct pathophysiology. Eight to ten percent of men presenting with erectile dysfunction have low testosterone levels (190, 192-193). The prevalence of low testosterone levels is not significantly different between middle aged and older men with impotence and those without impotence (190). Testosterone administration does not improve sexual function in men with erectile dysfunction who have normal testosterone levels (194-197). In men with sexual dysfunction who have unequivocally low testosterone levels, testosterone therapy improves libido and overall sexual activity (194-195). The response to testosterone supplementation in this group of men is variable because of the co-existence of other disorders such as diabetes mellitus, hypertension, cardiovascular disease, and psychogenic factors. Several meta-analyses of the usefulness of androgen replacement therapy concluded that testosterone administration is associated with greater improvements in sexual function compared to placebo treatment in men with sexual dysfunction and unequivocally low testosterone levels (195-197). Testosterone might also favorably affect marital interactions and intimacy due to an overall increase in sexual desire and sense of well being, independent of the change in erectile function; this hypothesis has not been examined. It is possible that testosterone might improve erectile responsiveness of men with ED to selective phosphodiesterase inhibitors. However, initial testosterone trials in men with ED who were receiving phosphodiesterase inhibitors have been inconclusive (198-199); this hypothesis should be investigated further.

In contrast to erectile dysfunction, which is usually a consequence of generalized atherosclerosis, androgen deficiency is an important cause of low sexual desire disorder. Therefore, serum testosterone concentrations should be measured in the diagnostic evaluation of hypoactive sexual desire disorder, recognizing that low sexual desire is often multifactorial; systemic illness, relationship and differentiation (the ability of individuals in a relationship to maintain their distinct identities) issues, depression, and many medications can be important antecedents or contributors to low sexual desire and sexual dysfunction.

Testosterone deficiency is associated with a progressive loss of bone mass. In one study performed in sexual offenders (200), surgical orchiectomy was associated with a progressive decrease in bone mineral density of a magnitude similar to that seen in women after menopause. Similarly, androgen deficiency induced by the administration of a GnRH agonist, surgical orchiectomy, or androgen antagonist for the treatment of prostate cancer and benign prostatic hypertrophy leads to loss of bone mass (201-203). In male rats, surgical orchiectomy or androgen blockade by administration of an androgen receptor antagonist is associated with loss of bone mass (204).

Androgen deficiency that develops before the completion of pubertal development is associated with reduced cortical and trabecular bone mass (205-206). During the pubertal years, significant bone accretion occurs under the influence of sex steroids; therefore, individuals with sex-steroid deficiency before or during peri-pubertal years may end up with suboptimal peak bone mass. Similarly, men with acquired androgen deficiency have lower bone mineral density than age-matched controls (133, 207-208).

Testosterone replacement of healthy, young, hypogonadal men is associated with significant increases in vertebral bone mineral density (133, 207-211). However, bone mineral density is typically not normalized after 1-2 years of testosterone replacement therapy (133). The reasons for the failure of testosterone replacement therapy to normalize bone mineral density in androgen-deficient men are not entirely clear. Some of the patients included in these studies had panhypopituitarism and therefore, also suffered from concomitant growth hormone deficiency. It is possible that concomitant GH replacement might be necessary for restoration of normal bone mineral density. Excessive glucocorticoid replacement might also contribute to bone loss in these patients. In addition, some participants had experienced testosterone deficiency before the onset and completion of pubertal development. It has been argued that maximal bone mass is achieved in part through bone accretion during the peripubertal period under the influence of sex-steroid hormones. If androgen deficiency occurs during this critical developmental period of bone accretion, the individual may end up with decreased peak bone mass, and testosterone administration may not be able to restore bone mass to levels seen in eugonadal age-matched controls. Many of the studies of testosterone replacement were less than 3 years in duration, and it is possible that a longer period of testosterone administration might be necessary to achieve maximal improvements in bone mineral density. Behre et al (207) reported that bone mineral density in some hypogonadal men after many years of testosterone treatment using a scrotal transdermal patch did reach the levels expected for age-matched eugonadal controls.

A number of cross-sectional surveys are in agreement that age-related decline in sex hormones is associated with age-related changes bone mineral density and the increased risk of osteoporotic fractures (108-115). Older men with hip fractures have lower testosterone levels than age-matched controls (212). In the Rancho Bernardo Study (109), bioavailable testosterone and bioavailable estradiol, but not total testosterone, levels were positively correlated with bone mineral density of the spine, hip, and distal radius. The Mayo Clinic Study in the Olmsted County also found bioavailable estradiol to be a better predictor of bone mineral density than total testosterone (111, 114). In the Dubbo Study (112), estradiol levels were more strongly correlated with bone mineral density than free testosterone.

Two long-term studies of testosterone replacement of relatively healthy older men have demonstrated significant increases in vertebral bone mineral density (214); however, another study did not find significant differences between the change in vertebral or femoral bone mineral density between testosterone and placebo groups (215). The latter study (215) supplemented participants with calcium and vitamin D; it is therefore possible that the increase in bone mineral density in the placebo-treated men might have been due to calcium and vitamin D supplement. Many of the men in this study had normal testosterone levels at baseline; bone mineral density improved in men with unequivocally low baseline testosterone levels (215).

Testosterone increases bone mass by several mechanisms. Short-term studies of androgen replacement have shown inconsistent increases in markers of bone formation, but a more consistent reduction in markers of bone resorption (216-219). These observations suggest that testosterone increases bone mineral density in part through its aromatization to estrogen, which inhibits bone resorption. Estrogen deficiency contributes to increased bone resorption and remodeling by multiple mechanisms. Estrogens regulate the activation frequency of bone functional basic multicellular units, the duration of the resorption phase and the formation phase, and osteoclast recruitment (228). The protective effects of estrogen on bone in both male and female mice during growth and maturation are mediated largely through estrogen receptor-alpha (220-226). In men androgens and estrogens both play independent roles in regulating bone resorption (228). In addition, there is increasing evidence that testosterone might also directly stimulate osteoblastic bone formation. Androgen receptors have been demonstrated on osteoblasts and on mesenchymal stem cells (227). Testosterone stimulates cortical bone formation (229). Testosterone also stimulates the production of several growth factors within the bone, including IGF-1; these growth factors may contribute to bone formation (230). Testosterone increases muscle mass, which may indirectly increase bone mass by increased loading. Testosterone might inhibit apoptosis of osteoblasts through non-genotropic mechanisms (231-232). In addition to its effects on bone mineral density, testosterone might reduce fall propensity because of its effects on muscle strength and reaction time.

Testosterone replacement has been shown to increase vertebral bone mineral density in young and older men with unequivocally low testosterone levels (5). Testosterone increases bone mass by multiple mechanisms. However, testosterone effects on fracture risk have not been studied. Efficacy trials of the effects of testosterone replacement on fracture risk would require several thousand men randomized to placebo and testosterone arms; such studies are being planned.

Androgens effects on cognitive function are domain-specific. For instance, observations that men outperform women in a variety of visuo-spatial skills suggest that androgens enhance visuo-spatial skills (233). In !Kung San Bushmen of Southern Africa , testosterone, but not estradiol, levels correlated with better spatial ability and with worse verbal fluency (234). Circulating levels of dihydrotestosterone, a metabolite of testosterone that is not converted to estrogen, positively correlated with verbal fluency (234). Barrett-Conner et al found positive associations between total and bioavailable testosterone levels, and global cognitive functioning and mental control, but not with visuospatial skills (235). In the Baltimore Longitudinal Study of Aging (236), higher free testosterone index was associated with better scores on visual and verbal memory, visuospatial functioning, and visuomotor scanning. Men with low testosterone levels had lower scores on visual memory and visuospatial performance (236). Neither total testosterone level nor the free testosterone index was correlated with verbal knowledge, mental status, or depressive symptoms (236). Other studies have reported a complex relationship between androgen levels and spatial ability (237-240). Women with congenital adrenal hyperplasia with high androgen levels score higher on tests of spatial cognition than their age- and gender-matched siblings (241). 46, XY rats with androgen insensitivity perform worse on tests of spatial cognition than their age-matched controls (242).

Several small clinical trials in elderly hypogonadal men have provided conflicting results (243-249); not surprisingly, a systematic review of clinical trials revealed no significant effects of testosterone on cognition (5). Janowsky et al (243) tested verbal and visual memory, spatial cognition, motor speed and cognitive flexibility in a group of healthy older men who received 3 months of testosterone supplementation. Testosterone replacement was associated with a significant improvement in spatial cognition only. Serum testosterone levels were not significantly correlated with spatial performance, but estradiol levels showed a significant inverse relationship with spatial performance suggesting that estradiol might inhibit spatial ability. Sih et al (156) found no effect of testosterone administration on cognition, while Cherrier et al (244, 246-247) reported an effect on visuo-spatial cognition. Testosterone also enhanced verbal fluency. Hypogonadal men performed worse on tests of verbal fluency than eugonadal men, and showed improvement after testosterone replacement (249). In transsexual males (251-252), administration of anti-androgen and estrogen, prior to surgery for gender reassignment, decreased anger and aggression proneness, sexual arousability, and spatial skills, and increased verbal fluency ability. Conversely, testosterone administration to females decreased verbal fluency and increased spatial skills. Testosterone administration may also improve verbal memory in women (252).

Testosterone is aromatized to estrogen in the brain, and some effects of testosterone on cognition might be mediated through its conversion to estradiol. However, androgen receptors are expressed in the brain (253), and androgen effects on brain organization during development (254-255) are mediated through androgen receptor. Androgens increase neurite arborization, facilitating intercellular communication (254-257). Testosterone also has nongenomic effects, and affects serotonin, dopamine, acetylcholine (257), and calcium signaling (258).

The literature on testosterone and cognition is highly equivocal; some, but not all studies, demonstrate improvements in tests of spatial cognition, verbal fluency and verbal memory. The inconsistency in findings cannot yet be interpreted as evidence that there is no effect. Rather methodological problems appear to limit the generalizability of results. Limitations of previous studies include limited sample sizes with heterogeneous, poorly defined samples; the use of a variety of neuropsychological tests, including some that lack prior psychometric validation; and the use of differing protocols in clinical trials. The effects of testosterone therapy on clinically important outcomes in men with cognitive impairment have been studied.

Cross-sectional studies of middle-aged men found a direct, rather than an inverse, relationship between serum testosterone levels and plasma HDL-cholesterol concentrations (277-279). Lower testosterone levels in men are associated with higher levels of dense LDL particles (278) and prothrombotic factors (280).

The effects of androgen supplementation on plasma lipids depend on the dose, the route of administration (oral or parenteral), the type of androgen (aromatizable or not) and the subject population (whether young or old, and hypogonadal or not). Supraphysiological doses of testosterone and non-aromatizable androgens frequently employed by body-builders undoubtedly decrease plasma HDL-cholesterol levels (281-284). However, physiological testosterone replacement in older men has been associated with only a modest or no decrease in plasma HDL-cholesterol (5, 155-160).

It has been suggested that the decrease in HDL cholesterol with testosterone administration might be the result of increased cholesterol efflux from endothelial macrophages stimulating reverse cholesterol transport, and therefore, a beneficial effect, rather than the result of increased HDL catabolism (285).

Spontaneous (133) and experimentally induced (286) androgen deficiency is associated with increased fat mass, and testosterone replacement decreased fat mass in older men with low testosterone levels (5). In epidemiologic studies, low testosterone levels are associated with higher levels of abdominal adiposity (287-288). Testosterone supplementation of middle-aged men with truncal obesity is associated with a reduction in visceral fat volume, serum glucose concentration, blood pressure, and an improvement in insulin sensitivity (289-291). Testosterone administration promotes the mobilization of triglycerides from the abdominal adipose tissue in middle-aged men (289). Surgical castration in rats impairs insulin sensitivity; physiologic testosterone replacement reverses this metabolic derangement (292). However, high doses of testosterone impair insulin sensitivity in castrated rats (292). Androgens increase insulin-independent glucose uptake (293) and modulated LPL activity in a region-specific manner (294).

Testosterone levels are lower in patients with type 2 diabetes mellitus compared with controls (295-298). Low total and free testosterone levels are associated with a higher risk of insulin resistance (299) and of developing type 2 diabetes (295-298). Free testosterone levels are negatively correlated with glucose, insulin, and C-peptide levels independent of body mass index (295). Some of these studies measured free testosterone by a tracer analog method which, however, is not independent of SHBG effects.

In cross-sectional studies, a direct correlation was found between circulating testosterone concentrations and tissue plasminogen activator activity (300), and an inverse relationship between testosterone and plasminogen activator inhibitor-1 activity, fibrinogen, and other prothrombotic factors (300), suggesting an antithrombotic effect of testosterone. In a prospective study of effect of androgens on serum inflammatory markers in men, increasing testosterone concentrations by hCG administration had no significant effect on inflammation sensitive markers (301). Similarly, in another study, even supraphysiological doses of testosterone did not affect C-reactive protein (302).

Whether variation of testosterone within the normal range is associated with risk of coronary artery disease remains controversial. Of the 30 cross-sectional studies reviewed by Alexandersen (121), 18 reported lower testosterone levels in men with coronary heart disease, 11 found similar testosterone levels in controls and men with coronary artery disease and 1 found higher levels of DHEAS. Prospective studies have failed to reveal an association of total testosterone levels and coronary artery disease (122-126, 303-305). The Rotterdam Study found that the common carotid artery intima media thickness, a marker of generalized atherosclerosis, was the highest in older men in the lowest quartile of serum testosterone levels (126).

One interventional study (306), reported that testosterone undecanoate given orally improved angina pectoris in men with coronary heart disease. Testosterone infusion acutely improves coronary blood flow in a canine model and in men with coronary artery disease (307-313). Short-term administration of testosterone induces a beneficial effect on exercise-induced myocardial ischemia in men with coronary artery disease (312). This effect may be related to a direct coronary-relaxing effect. Testosterone replacement has been shown to increase the time to 1-mm ST-segment depression (309). However, in another study, there were no differences among the placebo or testosterone groups in peak heart rate, systolic blood pressure, maximal rate pressure product, perfusion imaging scores, or the onset of ST-segment depression. Studies by Yue et al (313) demonstrated testosterone-induced endothelium independent relaxation of rabbit coronary arteries via potassium conductance.

In a mouse model of atherosclerosis that is LDL-receptor deficient (314) surgical castration accelerated, and testosterone administration retarded the progression of atherosclerosis. The magnitude of testosterone effect on atherosclerosis progession is similar to that observed with estrogen administration. Favorable effects of testosterone on atherosclerosis in this mouse model are antagonized by concomitant administration of an aromatase inhibitor, suggesting that testosterone effects are possibly mediated through its conversion to estrogen in the vessel wall (314). Testosterone effects in retarding atherosclerosis progression were independent of plasma lipids (314). Many, though not all the studies in cholesterol-fed, castrated male rabbits are in agreement that testosterone does not promote atherogenesis (315). Taken together, these data provide evidence that testosterone, through its conversion to estradiol, can retard the progression of atherosclerosis in these animal models.

These data suggest that serum testosterone levels in the range that is mid-normal for healthy young men are consistent with an optimal cardiovascular risk profile at any age, and that testosterone concentrations either above or below the physiological male range may increase the risk of atherosclerotic heart disease. The cohort and cross-sectional studies collectively suggest a neutral or favorable effect of testosterone on coronary heart disease in men, although the evidence is far from conclusive. The effects of testosterone replacement on cardiovascular risk in men have not been examined and are particularly important because even small changes in incidence rates could have significant public health impact.

Data from Cross-sectional Studies. Overall, in cross-sectional, epidemiological studies, there has not been a consistent association between circulating androgen levels and the occurrence of prostate cancer (330-349). While one meta-analysis found no association between serum testosterone levels and prostate cancer (335), another found a slightly increased risk of prostate cancer in men with the highest testosterone levels (343). However, in this latter meta-analysis, the increased risk was largely attributable to one study that demonstrated a significantly higher risk in men with higher testosterone levels than in men with lower testosterone levels (336). Two prospective, large, longitudinal, epidemiological studies that used random probability sampling, namely the MMAS and the Rancho Bernardo Study (347-348) did not report a significant correlation between circulating concentrations of testosterone or DHT and the subsequent detection of clinical prostate cancer.

None of the testosterone trials in middle-aged or older men has had sufficient power to detect meaningful differences in prostate event rates between testosterone and placebo-treated men. A systematic review of randomized testosterone trials in middle-aged and older men found higher rates of prostate events in testosterone-treated men than in placebo-treated men. Men treated with testosterone in these trials were significantly at higher risk for undergoing prostate biopsy than placebo-treated men. Because of the high prevalence of subclinical prostate cancer in older men, the higher number of prostate biopsies in testosterone-treated men is likely to yield higher detection rates of prostate cancer in comparison with placebo-treated men. Thus, testosterone therapy of middle-aged and older men is associated with a higher risk of prostate biopsy and a bias towards detection of a higher number of prostate events.

In a recent randomized controlled trial, Marks et al (350) measured intraprostatic testosterone and DHT levels in older men treated with placebo or testosterone. Surprisingly, intraprostatic DHT concentrations were not significantly higher in testosterone-treated men than in placebo-treated men (350). Similarly, the expression levels of androgen-dependent genes in the prostate were not significantly altered by testosterone administration (350). In a separate study, lowering of circulating testosterone levels by administration of a GnRH antagonist was not associated with changes in intraprostatic androgen concentrations (351).

Serum PSA levels are lower in testosterone–deficient men and are restored to normal following testosterone replacement (5, 352-371). Lowering of serum testosterone concentrations by withdrawal of androgen therapy in young, hypogonadal men is associated with a decrease in serum PSA levels. Similarly, treatment of men with benign prostatic hyperplasia with a 5-alpha reductase inhibitor, finasteride, is associated with a significant lowering of serum and prostatic PSA levels (360-361). Conversely, testosterone supplementation increases PSA levels (352-359). However, serum PSA levels do not increase progressively in healthy hypogonadal men with replacement doses of testosterone. In two placebo–controlled trials of testosterone administration in older men, the change in serum PSA levels over three–years was not significantly different between placebo–and testosterone–treated men (157-158). The increase in PSA levels during testosterone replacement might trigger evaluation and biopsy in some patients (5, 316).

More intensive PSA screening and follow-up of men receiving testosterone replacement might lead to an increased number of prostate biopsies and the detection of subclinical prostate cancers that would have otherwise remained undetected (5, 316). Serum PSA levels tend to fluctuate when measured repeatedly in the same individual over time (362-364). When serum PSA levels in androgen deficient men on testosterone replacement therapy show a change from a previously measured value, the clinician has to decide whether the change warrants detailed evaluation of the patient for prostate cancer, or whether it is simply due to test–to–test variability in PSA measurement. Therefore, it is important to set criteria for monitoring PSA changes during testosterone supplementation. Criteria that use very low thresholds for performing prostate biopsy relative to test-retest variability will likely result in an excessive number of biopsies with their associated costs, psychological trauma, and morbidity. On the other hand, criteria that use unreasonably high thresholds for performing prostate biopsies may fail to detect clinical prostate cancers at an early stage.

There is considerable test-retest variability in PSA measurements (362-364). Some of this variability is due to the inherent assay variability, and a significant portion of this variability is due to unknown factors. Fluctuations are larger in men with high mean PSA levels. Variability can be even greater if measurements are performed in different laboratories that use dissimilar assay methodology (362-364).

From a clinical perspective, an important issue is what increment in PSA level should warrant a prostate biopsy in older men receiving testosterone replacement. To address this issue, we conducted a systematic review of published studies of testosterone replacement in hypogonadal men (316). This review indicated that the weighted effect size of the change in PSA after testosterone replacement in young, hypogonadal men is 0.68 standard deviation units with a 95% confidence interval of 0.55 to 0.82, which is statistically significant. This means that the average effect of testosterone replacement therapy is to increase PSA levels by about 0.68 standard deviations over baseline. Because the average standard deviation was 0.47 in this systematic analysis, the standard deviation score of 0.68 translates into an average increase in serum PSA levels of about 0.30 ng/ml (316). There is considerable variability in the magnitude of change in PSA after testosterone supplementation among these studies, in part due to heterogeneity of study populations, inclusion of older men in some studies but not others, and differences in PSA assays. In addition, many patients who were enrolled in these studies were likely receiving testosterone replacement therapy previously; we do not know whether the washout period was sufficient to revert PSA levels to baseline. Therefore, it is possible that because of inadequate washout, the increments in the increments in serum PSA levels after testosterone administration might have been under-estimated.

To evalua