This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Copyright Permission Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Liu, P. Y. Articles by Handelsman, D. J. Search for Related Content PubMed PubMed Citation Articles by Liu, P. Y. Articles by Handelsman, D. J.

Androgens and Cardiovascular Disease

ANZAC Research Institute (P.Y.L., D.J.H.), Department of Andrology (P.Y.L., D.J.H.), Concord Hospital; and Department of Medicine (A.K.D.), University of Sydney, Sydney, New South Wales 2139, Australia

Correspondence: Address all correspondence and requests for reprints to: Prof. D. J. Handelsman, ANZAC Research Institute, Concord R.G. Hospital, Building 51, Hospital Road, Sydney, New South Wales 2139, Australia. E-mail: djh{at}anzac.edu.au

	Abstract

Top Abstract I. Introduction II. Gender, Life Span,... III. Newer Aspects of... IV. Vascular Biology V. Coronary Artery Disease VI. Hypertension VII. Cardiac Hypertrophy and... VIII. Cerebrovascular Disease IX. Peripheral Arterial Disease X. Erectile Dysfunction XI. Conclusion: the Present... References

 
Globally, cardiovascular disease will continue causing most human deaths for the foreseeable future. The consistent gender gap in life span of approximately 5.6 yr in all advanced economies must derive from gender differences in age-specific cardiovascular death rates, which rise steeply in parallel for both genders but 5–10 yr earlier in men. The lack of inflection point at modal age of menopause, contrasting with unequivocally estrogen-dependent biological markers like breast cancer or bone density, makes estrogen protection of premenopausal women an unlikely explanation. Limited human data suggest that testosterone exposure does not shorten life span in either gender, and oral estrogen treatment increases risk of cardiovascular death in men as it does in women. Alternatively, androgen exposure in early life (perinatal androgen imprinting) may predispose males to earlier onset of atherosclerosis. Following the recent reevaluation of the estrogen-protection orthodoxy, empirical research has flourished into the role of androgens in the progression of cardiovascular disease, highlighting the need to better understand androgen receptor (AR) coregulators, nongenomic androgen effects, tissue-specific metabolic activation of androgens, and androgen sensitivity. Novel therapeutic targets may arise from understanding how androgens enhance early plaque formation and cause vasodilatation via nongenomic androgen effects on vascular smooth muscle, and how tissue-specific variations in androgen effects are modulated by AR coregulators as well as metabolic activation of testosterone to amplify (via 5-reductase to form dihydrotestosterone acting on AR) or diversify (via aromatization to estradiol acting upon estrogen receptor /ß) the biological effects of testosterone on the vasculature. Observational studies show that blood testosterone concentrations are consistently lower among men with cardiovascular disease, suggesting a possible preventive role for testosterone therapy, which requires critical evaluation by further prospective studies. Short-term interventional studies show that testosterone produces a modest but consistent improvement in cardiac ischemia over placebo, comparable to the effects of existing antianginal drugs. By contrast, testosterone therapy has no beneficial effects in peripheral arterial disease but has not been evaluated in cerebrovascular disease. Erectile dysfunction is most frequently caused by pelvic arterial insufficiency due to atherosclerosis, and its sentinel relationship to generalized atherosclerosis is insufficiently appreciated. The commonality of risk factor patterns and mechanisms (including endothelial dysfunction) suggests that the efficacy of antiatherogenic therapy is an important challenge with the potential to enhance men’s motivation for prevention and treatment of cardiovascular diseases.

I. Introduction

II. Gender, Life Span, and Cardiovascular Disease

III. Newer Aspects of Androgen Action

A. Genomic regulation of androgen sensitivity

B. Nongenomic effects

C. Metabolic activation of testosterone

D. Derived measures of testosterone and SHBG

E. Fetal programming and perinatal androgen imprinting

IV. Vascular Biology

A. Gender

B. Androgen treatment

C. Genomic effects

D. Nongenomic effects

E. Vascular reactivity

V. Coronary Artery Disease

VI. Hypertension

VII. Cardiac Hypertrophy and Failure

VIII. Cerebrovascular Disease

IX. Peripheral Arterial Disease

X. Erectile Dysfunction

XI. Conclusion: the Present and Future

	I. Introduction

Top Abstract I. Introduction II. Gender, Life Span,... III. Newer Aspects of... IV. Vascular Biology V. Coronary Artery Disease VI. Hypertension VII. Cardiac Hypertrophy and... VIII. Cerebrovascular Disease IX. Peripheral Arterial Disease X. Erectile Dysfunction XI. Conclusion: the Present... References

 
GLOBALLY, CARDIOVASCULAR DISEASE will remain the major cause of human deaths well into the 21st century (1). Medical progress in this field has followed identification of pathogenic factors as clues leading to novel mechanistic insights, prevention strategies, and therapeutic targets. This has created specific targets for prevention and/or treatment, including hyperlipidemia, diabetes, smoking, hypertension, clotting mechanisms, and vessel occlusion. Yet, the higher male risk of myocardial ischemia may be the oldest clue to the pathogenesis of atherosclerotic cardiovascular disease. The first known recognizable description of myocardial ischemia, written 2000 yr ago by Seneca (4 BC–65 AD), a Roman stoic philosopher (but not a physician) describing his own disease, recorded that "The attack is very short, and like a storm. It usually ends within an hour... . To have any other malady is to be sick, to have this is to be dying" (2). Since then, numerous reports eventually led to systematic clinical observations (2) that have recognized the higher male risk for at least two centuries (3). Cardiovascular disease shows a consistent male to female ratio of 2.2 (range, 1.2–4.5) among different populations despite wide variations in absolute rates (3).

Over recent decades, the gender disparity in cardiovascular disease has been interpreted primarily as reflecting estrogen-mediated protection against atherogenesis. Despite remaining unproven by prospective clinical trials, this dominant belief then shaped the direction of much mechanistic research leaving the plausible alternative, i.e., that androgens promote atherosclerosis, little studied. After the first prospective, placebo-controlled, randomized clinical trials (RCT) that showed no cardiovascular benefits of combined estrogen/progestin therapy in menopausal women (4, 5), fundamental rethinking has taken place on whether and how reproductive hormones influence cardiovascular disease, invigorating and reorientating basic and applied biomedical research. This interest is also sharpened by the widening use, misuse, and abuse of androgens in the community. The safety of wider androgen use needs careful scrutiny because even minor deleterious effects on cardiovascular disease, as the most frequent cause of death, are likely to outweigh even substantial perceived benefits from androgens in any medical, lay, or abusive context (6). This review aims to highlight newer findings as well as gaps and opportunities for research relevant to understanding the role of androgens in the male predisposition to earlier onset atherosclerosis as well as the safety of androgens as increasingly used in the community. The role of androgens in lipid metabolism, hemostasis, obesity, and insulin resistance is well reviewed elsewhere (7).

	II. Gender, Life Span, and Cardiovascular Disease

Top Abstract I. Introduction II. Gender, Life Span,... III. Newer Aspects of... IV. Vascular Biology V. Coronary Artery Disease VI. Hypertension VII. Cardiac Hypertrophy and... VIII. Cerebrovascular Disease IX. Peripheral Arterial Disease X. Erectile Dysfunction XI. Conclusion: the Present... References

 
A strikingly consistent feature of human populations is the gender gap in life span (8, 9). At birth, life expectancy is consistently shorter for men than women across virtually all populations. In 186 of 191 (97%) United Nations member countries, men have shorter life expectancy by an average of 5.6 yr in those countries above a minimal level [Gross Domestic Product (GDP) in U.S. dollars, $3000 per capita] of economic development (Fig. 1). The gender gap in life span has not narrowed over the last century, and it seems unlikely to diminish without effective targeting. Because cardiovascular disease is the leading cause of human deaths, gender differences in its onset and severity must make an important contribution to the gender gap in life span. This contrasts with causes of death that may be more frequent among men (e.g., trauma, infectious disease) but are quantitatively minor at a population level (10).

View larger version (13K): [in this window] [in a new window]   Figure 1. Gender gap in life expectancy. Plot of gender gap in life expectancy, expressed as life expectancy at birth for women minus that for men (y-axis) against GDP per capita (in thousands of 1985 U.S. dollars) in 88 United Nations member countries for which both variables were available. Data obtained from WHO World Health Reports for 2002 (for life expectancies at birth) and 1999 (latest estimates of GDP). Note that above a threshold of approximately $3000 GDP per capita, gender difference is consistent at a mean of 5.6 yr across a wide range of economic development.

Hormonal effects are the most tractable for practical therapeutics, given the plethora of reproductive steroids available. In recent decades, clinical practice and basic mechanistic studies focused almost exclusively on the estrogen protection hypothesis—that cardiovascular disease progression was slowed by estrogenic protection in premenopausal women, a protection lost after menopause. Despite lacking proof by well-controlled prospective studies, the estrogen protection hypothesis became familiar, almost axiomatic. This status was supported by observational case-control studies finding that menopausal women using estrogens had less cardiovascular disease. As established in clinical practice, this dominant hypothesis determined that basic vascular biology studies would focus largely on mechanisms of estrogenic protection from atherosclerosis. In retrospect, selective recruiting of healthier and wealthier women, who are more likely to afford and prefer hormone replacement therapy, into case-control studies gave ultimately false reassurance. This limitation of observational research in eradicating or accounting fully for bias and confounding (17) reemphasizes a division of labor in clinical research, with observational studies being economic and efficient for hypothesis generation but well-controlled interventional studies remaining pivotal, the gold standard, for hypothesis testing. It is salient that, despite decades of clinical practice and supportive observational research, the Heart and Estrogen/Progestin Replacement Study (4) and Women’s Health Initiative (5) studies were the first placebo-controlled RCTs to test the estrogen protection hypothesis.

In fact, the dominance of the estrogen protection hypothesis overlooked contemporaneous evidence (18, 19, 20), consistently failing to confirm any break-point in female cardiovascular risk at the expected age of menopause, a key prediction of this hypothesis (21, 22). This contrasts with other definitely estrogen-dependent biological end-points, such as breast cancer and bone density (Fig. 2), that show inflection points in population data at the modal age of menopause. Moreover, cardiovascular death rate curves appear parallel for men and women, apart from an offset of approximately 5 yr in women. This is most consistent with the same disease process in both genders, but with an early head start in men rather than a continuous exposure to a greater risk in men that would tend to produce diverging lines. Such a head start might reflect biological processes that occur early in life (e.g., perinatal androgen surge in boys) or, perhaps, early in the pathogenesis of atherosclerosis. Furthermore, the estrogen protection hypothesis when applied to men, predicting that estrogen treatment reduces cardiovascular disease deaths, was conclusively refuted by the mid-1970s with two major well-controlled studies showing estrogen treatment of men caused excess cardiovascular deaths. Between 1966 and 1969, the Coronary Drug Project recruited 8341 men aged 30–64 yr from 53 centers who were randomized into various treatments or placebo for secondary prevention of myocardial infarction. Both high (5.0 mg) and low (2.5 mg) dose conjugated equine estrogen treatment arms were discontinued prematurely due to excess thromboembolism (23). In the early 1960s, the Veterans Administration Cooperative Urological Research Group recruited 2052 men with prostate cancer from 14 centers (24). Those randomized to treatment with 5.0 mg/d diethylstilbestrol had excess cardiovascular deaths compared with placebo regardless of stage. The excess mortality was a direct estrogen effect, rather than indirectly due to estrogen-induced castration, because the orchidectomy-alone treatment arm did not experience excess mortality. A caveat is that both studies used oral estrogens, which inevitably causes hepatic estrogen overdosage due to first-pass effects, further exaggerated by the high doses used. Because the oral route of administration is a major factor contributing to thromboembolic risk among users of estrogen for contraception or menopause, it is tempting to speculate that nonoral delivery of synthetic estrogen partial agonists ("selective estrogen receptor modulators") might have better outcomes.

View larger version (26K): [in this window] [in a new window]   Figure 2. Cardiovascular death rates in women do not exhibit an inflection point at expected modal age of menopause, unlike other unquestionably estrogen-dependent processes such as breast cancer and bone density that show a clear break-point at the expected age of menopause. Cardiovascular death rates in women and men in the United Kingdom, redrawn from the work of Tunstall-Pedoe (21 ) and plotted on a linear scale (top left panel) or log scale (top right panel), show no inflection point at modal age of menopause and a highly congruent curve displaced (delayed) by approximately 5 yr in women. By contrast, death rates from breast cancer (bottom left panel) and bone density (bottom right panel) show a clear inflection point around the end of the fifth decade of life. Breast cancer and cardiovascular death rates from 1962 U.S. women, redrawn from the work of Tracy (18 ). Bone density data from the National Health and Nutrition Examination Survey III (1988–94) is based on 6181 white U.S. men and women. [Top panels reproduced with permission from The Lancet (21 ).]

The natural history of complete androgen resistance due to a mutated, nonfunctional AR (36), combining male genetic sex with female phenotype, would provide a decisive test of whether life span or cardiovascular mortality followed male or female patterns. However, standard medical care of this rare disorder abrogates its natural history, making it unlikely that such data will ever be available. Alternatively, female-to-male (F2M) transgender, in which genetic females receive male testosterone doses, provides useful information. The only available follow-up study reported no excess cardiovascular disease among 293 F2M transsexuals during 2418 patient-years of exposure, compared with the general population (37). Longer and larger follow-up studies of such populations would be of considerable interest. Lower but more sustained exposure to androgen excess in women with polycystic ovary syndrome is also associated with no excess of cardiovascular disease (38) or mortality (39), despite increases in cardiovascular risk factors (40, 41), endometrial (but not breast) cancer (42), and more extensive atherosclerosis at coronary angiography (43). Because it is highly unlikely that these studies could have overlooked the greatly increased (7.4-fold) risk predicted by risk factor modeling (44), this striking discrepancy raises the possibility that androgens may have beneficial as well as detrimental effects on atherogenesis in women. If perinatal androgen imprinting is important in triggering male-pattern cardiovascular risk, both F2M transgender and women with polycystic ovary syndrome would lack this exposure, which might explain their relative freedom from cardiovascular complications.

The cardiovascular effects of androgen abuse have been extensively reviewed (7, 45, 46, 47). Adverse cardiovascular effects including myocardial infarction, hypertension, arrhythmia, cardiac failure, pulmonary embolism, stroke, and sudden death have been associated with androgen abuse, based on the temporal relationship to usage. Virtually all are single case reports of events that can occur in the absence of androgen usage and, in some, underlying medical disorders contribute to the risks (48). Without knowing the denominator of community exposure, the actual individual risks of such adverse effects compared with the general community are not clear (49). Given the highly prevalent and sustained usage among elite power athletes and bodybuilders for the last four decades, it is conceivable that the actual risks may not prove high. The East German national sports doping program involved more than 2000 elite athletes annually being treated with high-dose synthetic androgens. Regular medical monitoring recorded no cardiovascular complications, whereas liver disease and many other complications were regularly recorded (50). The longevity of former elite athletes who may have abused androgens may also shed light on the cardiovascular risks of androgen abuse. In a study of 2613 Finnish former elite athletes and age- and locale-matched conscripts, endurance and team athletes had lower cardiovascular mortality and longer lives than the general population (51), consistent with findings among Italian track and field athletes (52) and Dutch distance skaters (53), whereas those in power sports had less extended lives (51). Finnish power lifters among whom androgen abuse was suspected but not verified had 4.6-fold higher overall mortality, although life expectancy and cardiovascular mortality were the same as community controls (54). Further studies are needed to clarify the actual long-term cardiovascular and other risks associated with androgen abuse in former athletes.

In summary, the gender gap in life span is consistent in all but the poorest countries. The best available data suggest that adult male androgen exposure does not shorten men’s life span, but oral estrogen treatment has deleterious cardiovascular effects in men, as it does in women. How much of this harm is due to the oral route of administration or the use of synthetic estrogens is not clear. The risk of harmful cardiovascular effects due to high-dose androgen exposure from androgen abuse needs further clarification. If the gender differences in cardiovascular disease, which must contribute to the gender gap in life span, are due to hormonal effects, they presumably operate early in the pathogenesis of atherosclerosis.

	III. Newer Aspects of Androgen Action

Top Abstract I. Introduction II. Gender, Life Span,... III. Newer Aspects of... IV. Vascular Biology V. Coronary Artery Disease VI. Hypertension VII. Cardiac Hypertrophy and... VIII. Cerebrovascular Disease IX. Peripheral Arterial Disease X. Erectile Dysfunction XI. Conclusion: the Present... References

 
Expanding knowledge of androgen physiology has shed light on new mechanisms that may be crucial to better understanding androgen action on the vasculature for both basic and clinical research.

A. Genomic regulation of androgen sensitivity
The classical pathway of androgen action (Fig. 3) involves steroid binding to the AR, a ligand-activated transcription factor, and single copy member of the nuclear receptor superfamily, acting on the genome (36). The genomic action of AR is modulated by a large variety of coregulators, which are proteins that fine-tune target gene expression by enhancing (coactivator) or restraining (corepressor) transcription (55). Although testosterone circulates throughout the body, the factors responsible for variation in tissue androgen sensitivity remain to be further clarified. Intensity of expression of the single human AR (56) partly defines androgen sensitivity, but AR is almost ubiquitously expressed to some degree in tissues. Further biological determinants of tissue androgen sensitivity, including the functional AR polymorphisms as well as tissue distribution and regulation of AR coregulators, androgen metabolic enzymes, and nongenomic mechanisms, remain to be better defined so that their net integrated effects can be understood better.

View larger version (39K): [in this window] [in a new window]   Figure 3. Nongenomic and genomic androgen mechanisms. Left, Nongenomic effects of testosterone, which are triggered by binding to a still uncharacterized (nonclassical) membrane receptor. This activates second messengers including calcium and protein kinases, which produce typically rapid responses. Genomic effects are depicted in the middle and on the right. Testosterone (or other aromatizable synthetic androgens) crosses the cell membrane to be converted to estradiol (or the aromatic synthetic analog) by aromatase, which then binds to and activates ER or ERß. Right, DHT (or other nonaromatizable androgens) enters the cell to bind and activate the AR. Ligand-bound ER or AR dissociate their heat-shock protein (HSP), undergo conformational changes, dimerize, and translocate into the nucleus where they bind to specific sites known as estrogen response elements (ERE) or androgen response elements (ARE) located within the DNA of target nuclear genes to produce long-term genomic effects of testosterone.

Androgen sensitivity could be modulated by a functional polymorphism of AR that influences the strength of the genomic signal transduced from its interaction with an androgen as a bound ligand. One such functional AR polymorphism is the exon 1 triplet repeat CAG (polyglutamine) whereby the repeat length is inversely correlated with androgen sensitivity. Within the normal population, shorter repeat lengths are associated with higher risk of prostate cancer (60), whereas longer repeats are associated with reduced androgen effects on lipids and vascular reactivity (61) as well as bone (62) and sperm production (63) within the normal male population. Pathological extension (>40 repeats) causes a motor neuron disease (Kennedy’s syndrome, spinal muscular bulbar atrophy) with associated androgen insensitivity (64). Whether such fine-tuning of androgen responsiveness influences ultimate cardiovascular outcomes remains to be studied further. Further analysis of this and similar functional polymorphisms of androgen metabolizing enzymes (5-reductase, aromatase, CYP3A4; Ref. 65) may shed new light on androgen action on the cardiovascular system.

The development of the first nonsteroidal androgens (66, 67) has promise in exploiting tissue-specific differences in androgen sensitivity (68). Whether such tissue-specific partial androgen agonists (selective AR modulators), which are structurally nonaromatizable and functionally dihydrotestosterone (DHT) analogs, would have useful roles in vascular therapeutics mediated via AR remains to be clarified. Pharmacological targeting of nongenomic androgenic vasodilator mechanisms in vascular smooth muscle seems promising, in contrast to endothelial and vessel wall mechanisms that involve aromatization.

B. Nongenomic effects
There is now considerable evidence for rapid, nongenomic effects of steroids (Fig. 3), including androgens (69). Nongenomic steroid action is distinguished from genomic effects by 1) rapid onset (seconds to minutes) that is faster than genomic mechanisms, 2) insensitivity to inhibition of RNA and protein synthesis, 3) effects produced by steroids unable to access the nucleus (either covalently linked to membrane impermeable macromolecules or in cells lacking a nucleus), and 4) not usually blocked by classical antagonists due to different steroidal specificity from classical cognate nuclear receptors. As for other steroids, nongenomic androgen effects characteristically involve the rapid induction of conventional second messenger signal transduction cascades, including increases in cytosolic calcium and activation of protein kinase A, protein kinase C, and MAPK, leading to diverse cellular effects including smooth muscle relaxation, neuromuscular and junctional signal transmission and neuronal plasticity (70). In addition, nontranscriptional mechanisms have also been reported (71). Most nongenomic effects involve a membrane receptor, and putative binding sites are described for all major classes of steroids (72), including androgens (70). In mammals, only a membrane receptor for progesterone has been cloned (73, 74), but functional characterization is lacking. A membrane progestin receptor cloned from fish ovary features a heptahelical transmembrane structure typical of G protein-coupled receptors, and the recombinant protein exhibits high-affinity binding of progesterone with activation of postreceptor signal transduction pathways (75). No membrane AR has been characterized, but preliminary evidence of a low-affinity microsomal membrane binding site for alkylated androgens (76) and an endothelial cell plasma membrane dehydroepiandrostendione (DHEA) binding site (77) still require functional proof of specific receptor status. A plasma membrane SHBG receptor capable of modulating androgen action at plasma membranes and initiating intracellular cAMP signaling has been described in humans (78). The SHBG receptor remains to be fully characterized, and it is not clear whether it has any physiological role in species like rodents that lack circulating SHBG.

C. Metabolic activation of testosterone
A key issue in the biological effects of testosterone is its conversion to bioactive metabolites (Fig. 4). Although only a small fraction (<5%) of testosterone output undergoes such transformation usually in local tissues, conversion both amplifies and diversifies testosterone action. Conversion to its 5-reduced metabolite, DHT, by either type 1 or type 2 5reductase amplifies testosterone action because DHT has higher molar potency due to its more avid binding affinity and slower dissociation rate from the AR (79). In the prostate, feed-forward induction by DHT of type 2 5-reductase (80) results in virtually all testosterone entering the prostate being converted to DHT, thereby greatly enhancing local AR-mediated effects. Type 1 and type 2 5-reductase has been identified in vascular tissues based on immunoreactivity (81, 82) and enzymatic activity (83, 84, 85), but the biological consequences of androgen amplification in vessel walls remain to be clarified (84, 86, 87). Furthermore, conversion of testosterone to estradiol by the enzyme aromatase (CYP19) diversifies androgen action by activating estrogen receptors (ER). Aromatase gene expression (88, 89, 90, 91), protein (90, 91, 92), and enzymatic activity (93) have been detected in vascular tissues, including human coronary arteries (92) particularly in endothelium and smooth muscle. A more complete picture of gender differences and sex hormone regulation of aromatase and 5-reductase activity in vascular tissues is needed.

View larger version (35K): [in this window] [in a new window]   Figure 4. Pathways of testosterone action. In men, most (>95%) testosterone is produced under LH stimulation through its specific receptor, a heptahelical G protein-coupled receptor located on the surface membrane of the steroidogenic Leydig cells. The daily production of testosterone (5–7 mg) is disposed along one of four major pathways. The direct pathway of testosterone action is characteristic of skeletal muscle in which testosterone itself binds to and activates the AR. In such tissues there is little metabolism of testosterone to biologically active metabolites. The amplification pathway is characteristic of the prostate and hair follicle in which testosterone is converted by the type 2 5-reductase enzyme into the more potent androgen, DHT. This pathway produces local tissue-based enhancement of androgen action in specific tissues according to where this pathway is operative. The local amplification mechanism was the basis for development of prostate-selective inhibitors of androgen action via 5-reductase inhibition, the forerunner being finasteride. The diversification pathway of testosterone action allows testosterone to modulate its biological effects via estrogenic effects that often differ from AR-mediated effects. The diversification pathway, characteristic of bone and brain, involves the conversion of testosterone to estradiol by the enzyme aromatase, which then interacts with ER and/or ERß. Finally, the inactivation pathway occurs mainly in the liver, with oxidation and conjugation to biologically inactive metabolites that are excreted by the liver into the bile and by the kidney into the urine.

Despite lacking theoretical validity, such derived testosterone measures might still be useful empirically if they provided prediction of, or correlation with, independent biological androgen effects that were superior to those of standard testosterone measurements. Such empirical validation, however, is conspicuously lacking. Some derived testosterone measures clearly lack face validity. For example, among derived measures purporting to estimate free testosterone, both the free androgen (or testosterone) index calculation (111) and the free testosterone analog assays (112, 113, 114) are well-known to be invalid in men. Other measures such as equilibrium dialysis to estimate the free fraction or the bioavailable (corresponding to free plus loosely albumin-bound fractions) may be technically reproducible (95, 96), but there is no evidence that they provide significantly superior or additional biological information to the measurement of total testosterone as judged by independent biological effects. In summary, derived testosterone measures lack theoretical or empirical validity, add little or no explanatory power to clinical or physiological research beyond measurement of total testosterone, and should be used, if at all, only with concomitant independent empirical validation of those measures. Further studies of the biological role of SHBG in modulating androgen action are needed.

E. Fetal programming and perinatal androgen imprinting
The role of early life environmental exposures on late life cardiovascular disease has recently been recognized. Barker (115, 116) has compiled evidence from a wide variety of sources implicating prenatal environmental programming as a major determinant of susceptibility to diseases of later life, notably cardiovascular disease (115) and type 2 diabetes (116), the latter termed the thrifty phenotype hypothesis. He proposes that fetal adaptation to intrauterine malnutrition, to prioritize protection of vital organs, conditions preferred metabolic pathways in developing organ systems (developmental plasticity). The perpetuation of such adaptations into postnatal life may foster mechanisms that are deleterious in adult life. The precise mechanisms leading to cardiovascular disease remain to be elucidated, but accelerated early growth following low birth weight is characteristic (117), especially in boys (118). Neonatal androgen imprinting determines the sexually dimorphic mature blood pressure patterns of SHR rats (119) and susceptibility to diet-induced hypertension (119). This may be an important clue, when considered in conjunction with the natural history of gender differences in cardiovascular mortality, that points to atherogenesis having similar progression in men and women but with men having a head start at some undefined early stage in pathogenesis. A key event in early male life is the perinatal androgen surge when blood testosterone concentrations reach adult levels for months. This epoch is critical for hormonal imprinting of brain (120), prostate (121, 122, 123), and probably other androgen-sensitive tissues, perhaps including vascular tissues. An informative clinical model to test the effects of perinatal androgen exposure is congenital adrenal hyperplasia. Women with 21-hydroxylase deficiency would usually experience marked androgen excess until effective treatment is instituted. This early life androgen excess is known to influence gender identity and role (124), but effects on life span and cardiovascular disease are not known. Interestingly, castration of inbred rats at birth prolongs life span, whereas castration at weaning or maturity has no effect (35). More specific examination of gender in relation to fetal environmental programming and related perinatal events, such as the androgen surge and hormonal imprinting, may be informative.

	IV. Vascular Biology

Top Abstract I. Introduction II. Gender, Life Span,... III. Newer Aspects of... IV. Vascular Biology V. Coronary Artery Disease VI. Hypertension VII. Cardiac Hypertrophy and... VIII. Cerebrovascular Disease IX. Peripheral Arterial Disease X. Erectile Dysfunction XI. Conclusion: the Present... References

 
A. Gender
Gender differences are characteristic of animal models of atherosclerosis. Males develop earlier and more extensive atherosclerotic plaques independent of lipid levels in diet-induced models in the nonhuman primate (125) and in rabbits with (126) and without (127) intimal injury. Gender differences are also observed in the mouse atherosclerosis models involving apolipoprotein E (ApoE; Ref. 128) and low-density lipoprotein (LDL) receptor (129) deficiency as well as in the modulating influences on these models of hypertension (130), P-selectin deficiency (131), and peroxisome proliferator activator receptor agonist (132). Gender differences are reversed, with females developing diet-induced atherosclerotic plaques faster than males in certain mouse models such as ApoE-deficient mice (128, 133) and inbred mouse strains (134), although exogenous testosterone accelerates atherosclerosis in these models (133, 135). Maternal hypercholesterolemia in pregnant LDL receptor-deficient mice enhances atherosclerosis susceptibility of their male progeny, providing a mechanism for prenatal determination of susceptibility to cardiovascular disease (136). Mouse genetic susceptibility to atherosclerosis (137) or postinjury restenosis (138) varies between strains, with distinct susceptibility genes for each model (139). Although gender influences the relatively rapid vascular remodeling of postinjury restenosis in humans (140), the impact of gender on experimental restenosis models remains to be clarified.

B. Androgen treatment
Testosterone treatment consistently inhibits atherosclerosis in castrate, cholesterol-fed male rabbits (141, 142, 143). The most detailed study showed that castration of male rabbits increased, whereas both testosterone and DHEA treatment inhibited, aortic atherosclerosis. The higher testosterone concentrations produced by an injectable testosterone ester provided significantly better protection than an oral testosterone ester or DHEA. Testosterone effects were not explained by lipid mechanisms, whereas aortic ER (but not AR) content was down-regulated in parallel with atheroprotective effects. This suggestion of the importance of aromatization within the vessel wall for atheroprotection has been supported by observations that treatment with DHEA (readily aromatized) inhibits atherosclerosis in intact, cholesterol-fed rabbits (144, 145, 146), whereas a synthetic nonaromatizable androgen (stanozolol) had no protective effect (147). Similarly, in LDL receptor-deficient male mice, an aromatase inhibitor (anastrozole) blocked the atheroprotective effect of both endogenous and exogenous testosterone (91).

In female animals, however, proatherogenic effects of androgens are reported. Treatment of ovariectomized nonhuman primates for 1–2 yr with nandrolone decanoate (148) or 8 months with testosterone (149) increased coronary artery atherosclerotic plaque size compared with treatment controls and lower level androgen exposure (androstenedione plus estrone). The enhancement of atherogenesis was not explained by lipid changes or by the limited aromatizability of nandrolone. However, androgens also enlarged coronary diameter (148) and enhanced endothelium-dependent acetylcholine vasodilator responses (149), consistent with vasodilator effects of androgens.

In ApoE-deficient male mice, atherosclerotic plaque size is decreased by estradiol treatment of intact (150) or orchidectomized (151, 152) males. However, reported effects of testosterone are conflicting. A study using a GnRH antagonist (Cetrorelix) to deplete endogenous testosterone found that castration reduces, and testosterone treatment increases, atherosclerosis (133), whereas another study reported that orchidectomy had no effect and testosterone treatment reduces atherosclerosis (152). In LDL receptor-deficient male mice, orchidectomy increases, whereas testosterone or estradiol treatment reduces, atherosclerosis, and an aromatase inhibitor (anastrozole) blocks the atheroprotective effects of endogenous and exogenous testosterone (91). The speculation that these differences are due to extragonadal LH effects (133) lacks basis because no vascular effects of LH outside the reproductive tract have been established (153). Overall, animal models do not yet provide a fully coherent picture, but most evidence supports testosterone having an atheroprotective effect requiring aromatization in males, whereas in females androgens are proatherogenic. The role of 5-reduction in these testosterone effects has not been reported. Further studies using pharmacological probes (nonaromatizable androgens, blockers of androgen metabolism) and genetic mouse models to dissect components of testosterone action on the vasculature will be of continuing interest.

C. Genomic effects
AR is expressed in all cells of the vasculature, including endothelial cells, smooth muscle cells, myocardial fibers, macrophages, and platelets (Fig. 5). In earlier studies, myocardial (154) and aortic (155, 156, 157, 158, 159) AR content was similar in male and female rats (155, 156, 159), rabbits (158), and nonhuman primates (154, 157). However, males showed more nuclear localization, consistent with greater AR activation by endogenous testosterone (154, 155, 156, 157). Recent studies using more sensitive detection methods find consistent gender differences in vascular tissue AR content. Higher AR expression in males is reported for rat vascular smooth muscle (160), human macrophages from peripheral blood (161) or synovium (162), and mesenteric artery and endothelial cells (163). Hormonal regulation of AR protein levels in nonreproductive tissues including the vasculature is not well defined. Ligand binding initially stabilizes AR protein, but prolonged exposure leads to down-regulation (164). Male-specific expression of AR, such as in macrophages and smooth muscle cells, implies long-term AR protein up-regulation with prolonged exposure to endogenous male testosterone concentrations. Short-term exposure of rabbit arterial neointimal plaque to testosterone increased AR mRNA while inhibiting plaque development in culture (165).

View larger version (48K): [in this window] [in a new window]   Figure 5. Sites of androgen action on the vasculature, including sites where androgens have been shown to interact with the vessel wall and bloodstream components to initiate early stages of atherosclerosis, to propagate and enlarge the lesions, and to modify the late stage of plaque rupture.

Apoptotic damage of vascular endothelial cells is an important cellular mechanism in atherogenesis, leading to increased platelet adhesiveness and thus increased tendency for thrombus formation (173). Testosterone enhances endothelial cell apoptosis provoked by serum deprivation (as an in vitro model of arterial wall damage; Ref. 174). This testosterone effect is blocked by flutamide but is not replicated by DHEA or estradiol, indicating involvement of AR but not aromatization. Any role of 5-reduction has not been reported.

In the intact arterial wall, smooth muscle cells regulate the arterial tone and produce the extracellular matrix. Proliferation and migration of smooth muscle cells are important steps in the formation of neointima and stenoses. Testosterone and DHT treatment stimulate proliferation of rat vascular smooth muscle cells (84), whereas estradiol inhibits their proliferation and migration (175, 176). Consequently, postinjury neointimal plaque size is larger in males than females and increases after ovariectomy (but not orchidectomy) with restoration of intact female levels by estradiol replacement (177).

Macrophages play a key role in atherosclerosis by migrating into the vessel wall where they internalize large amounts of exogenous lipids by various unregulated scavenger receptors and phagocytosis. Lipid accumulation transforms macrophages into foam cells, forming the fatty streaks characteristic of early atherosclerosis. Experimentally, androgens enhance foam cell formation because lipid loading of male (but not female) macrophages is increased by DHT treatment, an effect blocked by flutamide consistent with an AR-dependent mechanism (161). Conversely, testosterone also enhances reverse cholesterol transport, which might retard development of fatty streaks. Testosterone at a physiological concentration increases expression of the scavenger receptor-B1 at mRNA and protein levels in human hepatocytes and monocyte-derived macrophages (178). The functional consequences of increased scavenger receptor-B1 were observed as testosterone-dependent enhanced transfer of cholesterol esters from monocyte-derived macrophages to high-density lipoprotein-3 during in vitro culture (178). Whether the testosterone effect required the AR or aromatization was not reported, but because estrogens reduce scavenger receptor B1 in nonsteroidogenic tissues (179), aromatization is not required. Any role of 5-reduction has not been reported. Testosterone treatment of murine macrophages inhibits nitrite release via inhibition of the inducible nitric oxide (NO) synthase (NOS) enzyme, although the underlying mechanism remains unknown (180). This inhibition of inducible NOS could increase platelet aggregation and thrombosis risk associated with androgen treatment by eliminating the antiaggregatory effects of NO.

D. Nongenomic effects
Among vascular cells, nongenomic effects of testosterone (Fig. 5) are described for macrophages (181, 182), endothelial cells (163, 183), and vascular smooth muscle. Macrophage cell lines (IC-21, RAW264.7) that do not express the nuclear AR show rapid and repeated increased cytosolic calcium responses to testosterone, DHT, and testosterone rendered membrane-impermeable by conjugation to BSA (T-BSA) but not to inactive androgens (5ß-DHT, 1-DHT) (181, 182). Testosterone effects were not blocked by classical antiandrogens (cyproterone, flutamide) or antiestrogens (raloxifene, tamoxifen, ICI182, 780) excluding the involvement of AR and aromatization, but the role of 5-reduction remains unclear. T-BSA was bound to plasma membranes and internalized by an energy- and cytoskeleton-dependent process. Based on blocker experiments, the calcium response originated from primarily internal stores coupled to phospholipase C via a G protein-coupled receptor (184). In RAW264.7, murine macrophages stably transfected with c-fos promoter (RAW-fos13), the functional effects of nongenomic testosterone signaling were illustrated by testosterone attenuation of lipopolysaccharide-activated c-fos promoter activity, p38MAPK, and NO production (182). In human endothelial cells, testosterone, T-BSA, and DHT all stimulate increased cytosolic calcium via membrane influx that was abolished by removing extracellular calcium or blocking membrane calcium channels but not by blockade of intracellular calcium stores (163). By contrast, in cultured rat, aortic endothelial cells testosterone inhibited bradykinin-induced increases in intracellular calcium but had no effect itself (183).

Testosterone-induced vasodilatation, first reported by Waldman in 1945 (185), is well established with in vitro vasodepressor effects reported in precontracted arteries of the rat (13, 186, 187, 188, 189, 190), mice (191), rabbit (192), pig (193, 194), guinea pig (195), ferret (196), and dog (197). These effects involve primarily the vascular smooth muscle without requiring the presence of endothelium, although an endothelial contribution is apparent in some studies (186, 197). Testosterone acts via a nongenomic mechanism because the responses are rapid; present in tfm rats with mutated, nonfunctional AR (186, 191, 198); reproduced by testosterone conjugated to membrane-impermeable macromolecules (189, 199, 200); and not blocked by inhibitors of DNA and protein synthesis (199, 201), classical AR antagonists (190, 199, 201, 202), aromatase inhibition (190, 192, 203), or ER blockers (197, 203). The mechanisms involved include endothelium-derived NO (where endothelium is involved) and, more regularly, blockade of membrane calcium influx via voltage-operated calcium channels (191, 204) and potassium efflux involving voltage-operated (13, 189) and BK_Ca (194) potassium channels (190) in vascular smooth muscle, including in humans (205). The steroidal selectivity of the putative membrane receptor involved in testosterone-induced vasodilation is unusual, with classical antiandrogens (cyproterone, flutamide) ineffective, whereas 5ß-androstanes (androsterone, etiocholanolone, 5ß-DHT) are more potent vasodilators than 5-androstanes (5-DHT, 3-diol), which is the reverse of their potency at the AR. Neither 3 nor 17 conjugation impairs vasodilatory activity (200). Testosterone induces vasodilatation in all arterial beds studied, comprising coronary, mesenteric, iliac, renal, and femoral (206), although sensitivity varies between them (192, 197, 206, 207, 208) and is reduced by aging (209).

The consistent requirement for high (micromolar) concentrations of testosterone for vasodilatory responses raises questions about the physiological nature of these responses. Few studies are reported at lower testosterone doses, but a recent in vivo study of pigs reported widespread arterial vasodilatory responses using testosterone infusions calculated to produce concentrations of approximately 1 µg/liter (206). At similar physiological concentrations of testosterone, however, vasoconstrictor responses have been observed (199, 201), including one study in which the higher, supraphysiological testosterone concentrations produced vasodilatation (201). Vasoconstrictor responses have long been known in tissues obtained from androgen-treated animals (195, 210, 211, 212, 213), but the effective tissue androgen concentrations involved are not clear. Further study of physiological testosterone concentrations would be of considerable interest.

E. Vascular reactivity
The vascular endothelium, a single cell layer separating blood and vascular smooth muscle, regulates vessel tone through release of vasoactive factors such as NO, endothelins, and prostanoids (214). As a cellular plane, it forms the surface on which blood cells, vessel wall growth, and adhesive factors interact to form the nidus of atherogenic lesion. The pivotal significance of endothelial NO release causing vasodilatation has facilitated the development of endothelial function tests. These tests are based on comparing the vascular effects of stimuli that cause endothelial NO release (endothelium-dependent stimuli) with endothelium-independent stimuli that deliver NO directly to vascular smooth muscle, bypassing the endothelium. The difference then represents the effect of endothelial-mediated NO release and vasodilatation. During coronary angiography, for example, acetylcholine represents an endothelium-dependent stimulus to be compared with nitrates as an endothelium-independent stimulus (215). The development of a noninvasive test [flow-mediated dilatation (FMD)] for arterial endothelial function (216) has allowed wider investigation of endothelial function, whereas biochemical tests of endothelium-derived markers remain of uncertain validity (214). FMD is estimated by measuring brachial artery diameter with ultrasound before and after reactive hyperemia (where shear stress causes endothelial NO release) and administration of sublingual glyceryl trinitrate (where drug release of NO is an endothelium-independent stimulus). Impaired FMD is a feature of early atherosclerosis, corresponding closely with extent of coronary heart disease (217), but it is also influenced by other variables that must be controlled for studies of FMD (214).

FMD studies provide insight into androgen effects on the human vasculature (Table 1). FMD is similar in men and age-matched women (218, 219) except for higher values during the estrogen-dominated follicular and luteal phases of the menstrual cycle (218). FMD declines with age in both men and women, although the decline starts earlier in men (220). Both endogenous and exogenous testosterone impair vascular reactivity in men. Vascular reactivity is enhanced by castration in older men with advanced prostate cancer compared with age-matched and cancer controls (221). In a large well-controlled study of 36 newly diagnosed, nonsmoking hypogonadal men, androgen deficiency was associated with markedly increased FMD compared with age-matched eugonadal controls (222). Testosterone replacement therapy reduces FMD (214, 222), although to levels lower than age-matched eugonadal controls (222), presumably reflecting limitations of the testosterone delivery system compared with endogenous production. A study of non-androgen-deficient men with coronary artery disease reported that FMD, but not nitrate-induced dilatation, was increased after 8-wk treatment with 40 mg tamoxifen daily (223), which suggests effects of endogenous testosterone involving aromatization although a direct estrogenic effect of the drug, a partial estrogen agonist, cannot be excluded. Hence, within the physiological range, testosterone impairs FMD, although the involvement of aromatization, 5-reduction, AR, or nongenomic androgen effects remains to be clarified.

In healthy, community-dwelling older men (over the age of 60 yr) with mild age-related lowering of blood testosterone concentrations, administration of transdermal testosterone for 1 yr (228) or recombinant human chorionic gonadotropin (229) or transdermal DHT gel (230) for 3 months had no significant effect on vascular reactivity. The latter study using DHT, a nonaromatizable androgen, indicates that the unimproved vascular reactivity in the two studies using testosterone was unlikely to be due to the balancing effect of estradiol via aromatization from testosterone. These findings in older men are best explained by their not being sufficiently androgen deficient, or the treatments not being used to produce sufficiently high androgen exposure, to change vascular reactivity. Alternatively, the effects of age may not be reversible by modest net changes in androgen exposure.

By contrast, vascular reactivity is increased by long-term administration of high-dose estrogens to genetic males (M2F transsexuals; Refs. 231, 232, 233), by 8-wk oral estradiol valerate treatment in older men castrated for advanced prostate cancer (234), and by parenteral administration of low-dose estradiol to healthy young men (235). Acute administration of estradiol had no effect in healthy young men (236), and FMD was absent (although nitrate- and estradiol-induced dilatation was present) in a young man with a nonfunctional, mutated ER gene (237). These findings that estradiol, mediated probably via ER in endothelial cells, augments vascular reactivity in men of all ages suggest that aromatization contributes to the vascular reactivity effects of exogenous and endogenous testosterone.

Administration of testosterone to women has less clear results. Genetic females having long-term administration of testosterone (F2M transsexuals) maintaining male blood testosterone concentrations have larger brachial artery diameter and reduced nitrate-induced response, but FMD was not significantly different from age-matched female controls (238). For estrogen-treated postmenopausal women having parenteral testosterone (producing a 5-fold elevation to supraphysiological blood testosterone concentrations for women), FMD was apparently increased, although the small effect size was difficult to interpret due to baseline mismatch for vascular reactivity (239). The effects of age, aromatization, and testosterone dose in these contrasting findings remains unclear.

	V. Coronary Artery Disease

Top Abstract I. Introduction II. Gender, Life Span,... III. Newer Aspects of... IV. Vascular Biology V. Coronary Artery Disease VI. Hypertension VII. Cardiac Hypertrophy and... VIII. Cerebrovascular Disease IX. Peripheral Arterial Disease X. Erectile Dysfunction XI. Conclusion: the Present... References

 
The relationship of androgens to coronary artery disease has been well reviewed (240, 241, 242). Observational studies show a consistent inverse relationship between endogenous testosterone and adverse cardiovascular events (240, 243, 244, 245, 246). These findings may be interpreted as suggesting the testable hypothesis that chronically lowered blood testosterone may increase risk of cardiovascular disease. However, such findings from cross-sectional studies cannot distinguish the direction of causality (either of which is plausible) or exclude a common cause. The opposite interpretation is that blood testosterone is mildly lowered due to heart disease, as occurs for many chronic disorders (247, 248, 249, 250). Longitudinal studies do not support the predictive value of lowered blood testosterone concentration for further cardiovascular events (240, 251, 252), which favors the decrease in cross-sectional studies being an effect, rather than a cause. Nevertheless, the only definitive test of this important concept is a prospective interventional study of sufficient power and duration of surveillance to estimate or exclude an important protective effect size.

An important predictive relationship for blood DHEA sulfate (DHEAS) with further cardiovascular disease was first reported by the Rancho Bernardo study, which found that low blood DHEAS concentration predicted cardiovascular disease 12 yr later among community-dwelling older men, but not women (253). Subsequent observational studies confirmed these findings, albeit with a lower risk (252), whereas other similar prospective studies with more than 5-yr follow-up fail to confirm the original observation (254, 255, 256). Further analysis of the original study cohort at 19 yr has attenuated and qualified the risk (257). The apparent protective effects of DHEA remain hard to explain. DHEA and its sulfated ester DHEAS are weak androgens of adrenal origin that circulate at high blood levels in young adults before undergoing a steep decline from the third decade of life. Yet, DHEA has no convincing hormonal effects in its own right (other than due to conversion to bioactive steroids), consistent with the absence of a specific DHEA receptor or distinctive functional effects. Preliminary reports of DHEA binding (77, 258, 259, 260) have not yet led to characterization of a specific DHEA receptor. However, given the plethora of orphan receptors in the nuclear receptor gene superfamily (261), a specific DHEA receptor cannot be discounted. Similarly, there is evidence for DHEA interacting with neurotransmitter (N-methyl-D-aspartate , -aminobutyric acid_A) receptors (262), having vasodilator properties (192, 205, 263) and many other nonspecific biological effects (264), but none that convincingly explains cardioprotective effects. Although mechanistic studies have still not identified plausible explanatory mechanisms for protective effects of DHEAS, further studies are warranted. Alternatively, methodological factors may be the explanation. For example, the very steep age-related decline in DHEAS concentrations and complex interrelationships with other age-related cardiovascular risk factors may be difficult to fully account for confounding by age despite sophisticated analyses (252). It remains to be clarified whether predictive effects of DHEAS on cardiac mortality reflect confounding by the strong age dependence of its circulating concentration (265) or other factors such as cardiac failure, which lowers DHEAS in proportion to its severity (266).

Interventional studies of androgen effects on symptomatic coronary artery disease are limited (Table 2). Three RCTs of chronic androgen therapy involving objective clinical cardiovascular end-points have been reported. Earlier studies having no controls (267) or only ad hoc controls (268, 269) reported that androgen therapy improved symptomatic angina (270). The three well-controlled studies consistently report improvement in objective measures of cardiac ischemia (exercise stress testing) with subjective improvements with 2- to 12-wk testosterone therapy. The first study randomized 50 symptomatic men with cardiographic evidence of ischemia (postexercise ST segment depression) to weekly injections of testosterone cypionate (200 mg) or oil vehicle for 8 wk (271). Testosterone caused a highly significant reduction (32% at 4 wk, 51% at 8 wk) in postexercise ST segment depression compared with placebo, which had no effect (<2%). These findings were confirmed by a placebo-controlled, crossover study of 62 older men with established ischemic heart disease randomized to treatment with either testosterone undecanoate (120 mg/d for 2 wk, then 40 mg/d for 2 wk) or placebo, and then crossed over to the other treatment after a 2-wk washout period (272). Dramatic improvement in cardiac ischemia in subjective (77% vs. 7% with angina symptoms) and objective criteria (electrocardiogram, 69% vs. 8%; Holter, 75% vs. 8%), but no change in cardiac function (echocardiography) was observed. This latter study was apparently unblinded, and the objective scales were not described; in addition, another report gives an inconsistent account of the placebo treatment (273). Most recently, a third study involved 46 men with stable angina randomized to receive daily testosterone (5 mg) transdermal patch or matching placebo for 12 wk in addition to their current antianginal therapy (274). Testosterone lengthened time to ST segment depression and improved subjective pain perception and physical role function compared with placebo. In concert, these studies suggest that testosterone is a coronary vasodilator in men with established atherosclerosis. The consistent improvement in objective measures of cardiac function indicates that, despite the fact that mood elevating and motivational effects of testosterone might enhance effort-dependent exercise tests, such subjective reactions to testosterone cannot explain all cardiac effects of androgens.