Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

REVIEWS

Testosterone and insulin resistance in the


metabolic syndrome and T2DM in men
Preethi M. Rao, Daniel M. Kelly and T. Hugh Jones
Abstract | Obesity, type 2 diabetes mellitus and the metabolic syndrome are major risk factors for
cardiovascular disease. Studies have demonstrated an association between low levels of testosterone
and the above insulin-resistant states, with a prevalence of hypogonadism of up to 50% in men with type 2
diabetes mellitus. Low levels of testosterone are also associated with an increased risk of all-cause and
cardiovascular mortality. Hypogonadism and obesity share a bidirectional relationship as a result of the
complex interplay between adipocytokines, proinflammatory cytokines and hypothalamic hormones that
control the pituitary–testicular axis. Interventional studies have shown beneficial effects of testosterone
on components of the metabolic syndrome, type 2 diabetes mellitus and other cardiovascular risk factors,
including insulin resistance and high levels of cholesterol. Biochemical evidence indicates that testosterone is
involved in promoting glucose utilization by stimulating glucose uptake, glycolysis and mitochondrial oxidative
phosphorylation. Testosterone is also involved in lipid homeostasis in major insulin-responsive target tissues,
such as liver, adipose tissue and skeletal muscle.
Rao, P. M. et al. Nat. Rev. Endocrinol. 9, 479–493 (2013); published online 25 June 2013; doi:10.1038/nrendo.2013.122

Introduction
The WHO’s statistical fact sheet from September 2011 the metabolic syndrome, is an intermediary cardio­
states that cardiovascular disease is the leading cause of vascular risk factor that contributes to the development
death and disability in the world.1 By 2030, an estimated of impaired glucose metabolism, dyslipidaemia, hyper­
23.6 million people will die from cardiovascular disease tension, endothelial dysfunction and inflammation,
every year, mainly from heart disease and stroke.1 The which all independently promote atherogenesis.
incidence of cardiovascular mortality, including prema­ Testosterone has a key role in insulin sensitivity, body
ture death, is higher in men than in women.2 According composition, metabolism of lipids and cholesterol,
to the 2012 American Heart Association fact sheet, more inflammation and vascular reactivity in men.10 Testos­
than one in three adult men has some form of cardio­ terone deficiency is associated with adverse effects on
vascular disease.3 Obesity, the metabolic syndrome and several classic modifiable risk factors for cardio­vascular
type 2 diabetes mellitus (T2DM) are major risk factors disease,11,12 contrary to the common belief that testos­
for cardiovascular disease. Changes in our dietary habits terone is ‘bad for the heart’. Furthermore, interventional
and an increasingly sedentary lifestyle have contributed studies have shown an improvement in these risk factors
to the phenomenal rise in rates of obesity throughout upon treatment with exogenous testosterone. 11,12 In
the world. A large number of studies in men over the this Review, we focus on the effects of testosterone on
past 30 years have shown that low levels of testos­terone insulin sensitivity and glycaemic control, cholesterol and
correlate significantly with the incidence of obesity, other risk factors for cardiovascular disease. The patho­
the metabolic syndrome, T2DM and cardiovascular physiological mechanisms that link hypogonadism with
disease. Furthermore, a man with hypogonadism is at body composition, insulin sensitivity and inflammation
an increased risk of developing these metabolic dis­ and the role of testosterone replacement in the meta­
orders, which supports the existence of the bidirec­ bolic syndrome and T2DM will be explored in detail.
tional relation­ship between low levels of testosterone The metabolic syndrome (Box 1), T2DM (Box 2) and
and insulin resistant states.4–9 Insulin resistance can be hypogonadism (Box 3) have been defined. Academic Unit of
Diabetes, Endocrinology
defined as a reduced capacity of the tissues to metabo­ and Metabolism,
lize glucose and free fatty acids. This condition, which Prevalence School of Medicine and
Biomedical Sciences,
is the central biochemical abnormality in T2DM and Epidemiological studies have consistently reported University of Sheffield,
a high prevalence of low levels of testosterone in men Beech Hill Road,
with the metabolic syndrome and T2DM compared with Sheffield S10 2RX, UK
(P. M. Rao, D. M. Kelly,
Competing interests the general population.13–16 Furthermore, longitudinal T. H. Jones).
T. H. Jones declares associations with the following companies:
population studies have shown that low levels of testos­
Bayer Healthcare, Clarus, Ferring, Lilly, Merck, ProStrakan. See Correspondence to:
the article online for full details of the relationships. The other terone and sex hormone-binding globulin (SHBG) are T. H. Jones
authors declare no competing interests. independent predictors of the future occurrence of these hugh.jones@nhs.net

NATURE REVIEWS | ENDOCRINOLOGY VOLUME 9  |  AUGUST 2013  |  479


© 2013 Macmillan Publishers Limited. All rights reserved
REVIEWS

Key points the Tromso study (n = 1,548), which reported negative


independent correlations for testosterone and SHBG
■■ Testosterone deficiency is highly prevalent in men with the metabolic syndrome
with waist circumference.19 In addition, the Quebec
and type 2 diabetes mellitus
■■ Low levels of testosterone are an independent risk factor that predicts
family study reported that the amount of sub­cutaneous
subsequent development of the metabolic syndrome and type 2 diabetes mellitus and visceral fat both had a negative correlation with
■■ Population studies in community-dwelling men have shown that testosterone levels of testos­terone.20 In a cross-sectional study of
deficiency is associated with increased all-cause mortality and cardiovascular 160 men who applied for medical or surgical obesity
mortality therapy, participants were assessed for isolated hypo­
■■ The hypogonadal–obesity–adipocytokine hypothesis summarises the complex gonadotrophic hypogonadism; 57.5% had subnormal
interaction of the above components and their contribution to the vicious cycle levels of total testos­terone and 35.6% had low levels of
of obesity causing hypogonadism and vice versa
free testosterone.15
■■ Interventional studies of testosterone replacement therapy have shown
improvements in insulin resistance, body composition, glycaemic control, lipid A study published in 2011 has shown that low testos­
metabolism and other cardiovascular risk factors terone is more prevalent in men with the metabolic syn­
■■ The benefit of testosterone on insulin sensitivity might be attributable to a drome than in age-matched healthy control indivi­­d­uals.21
complex regulatory influence on insulin signalling and glucose homeostasis Total testosterone, calculated free testosterone, and SHBG
in the major insulin-responsive target tissues were also lower in men with the metabolic syndrome than
in control individuals (P <0.001). Low levels of testos­
terone were observed in 19 (30%) men with the meta­
Box 1 | Definition of the metabolic syndrome bolic syndrome but only one (3.1%) of the healthy control
This definition is from a joint statement from the individuals. Men with the metabolic syndrome and
International Diabetes Federation (IDF), the National levels of calculated free testosterone <0.225 pmol/l had
Heart, Lung, and Blood Institute (NHLBI), the World Heart significantly higher HOMA‑IR (a biochemical marker
Federation, the International Atherosclerosis Society and of insulin resistance) values than men with testos­terone
the American Heart Association (AHA) issued in 2009. levels within the normal range.
Patients need to have three of five criteria to be termed Our group assessed the prevalence of clinical hypo­
as having the metabolic syndrome.
gonadism in 355 men with T2DM.13 Overt hypo­gonadism
■■ Raised waist circumference,* with population and
country-specific definitions was defined as the presence of clinical symp­toms of
■■ Raised levels of triglycerides‡§ ≥1.7 mmol/l hypogonadism and low levels of testosterone (total
■■ Reduced levels of HDL cholesterol‡ <1.0 mmol/l in testos­terone <8 nmol/l and/or bioavailable testosterone
men and <1.3 mmol/l in women <2.5 nmol/l). Borderline hypogonadism was defined
■■ Raised blood pressure,‡ systolic ≥130 mmHg and/or as the presence of symptoms and levels of total testos­
diastolic ≥85 mmHg terone of 8–12 nmol/l or bioavailable testosterone of
■■ Raised fasting levels of glucose‡§ ≥5.55 mmol/l
2.5–4.0 nmol/l. Overt hypogonadism was seen in 17%
*It is recommended that the IDF cut points be used for
nonEuropeans and either the IDF or AHA/NHLBI cut points used for
of men with levels of total testosterone <8 nmol/l and
people of European origin until more data are available. ‡Drug 14% of men with levels of bioavailable testosterone
treatment for raised levels is an alternate indicator. §The most <2.5 nmol/l. Borderline hypogonadism was found in
commonly used drugs for raised levels of triglycerides and reduced
levels of HDL cholesterol are fibrates and nicotinic acid. A patient 25% of men with total testosterone of 8–12 nmol/l and
taking one of these drugs can be presumed to have high bioavailable testos­terone of 2.5–4.0 nmol/l; 42% of the
triglycerides and low HDL cholesterol. High-dose ω‑3 fatty acid
presumes high triglycerides. Most patients with type 2 diabetes
men had calculated free testosterone levels <0.25 nmol/l
mellitus will have the metabolic syndrome by the proposed criteria. and ­therefore had hypogonadism.
In another cross-sectional study of 580 men with
T2DM and 69 men with type 1 diabetes mellitus (T1DM),
conditions.4–7 Circulating testosterone has three major 43% of men with T2DM had reduced levels of total
components: free testosterone (2%); albumin-bound testos­terone (<10 nmol/l) and 57% had reduced levels
testosterone (20–40%); and SHBG-bound testosterone of calculated free testosterone (<0.23 nmol/l). Of the
(60–80%).17 Cross-sectional studies have established that men with T1DM, 7% had low levels of total testosterone
levels of biologically measurable active testos­terone frac­ and 20.3% had low levels of calculated free testos­terone
tions (that is, free and/or bioavailable albumin-bound (<0.23 nmol/l).22 Another cross-sectional study of 115
testosterone) that are independent of SHBG are con­ men with T2DM, 93 men with T1DM and 121 healthy
siderably reduced in states of insulin resistance. 6,8,13,16 control individuals reported that 45% and 61% of men
Importantly, free testosterone was measured directly in with T2DM had low total testosterone and calculated free
one of these studies using equilibrium dialysis, which testosterone levels, respectively.23 Total testosterone levels
is the gold standard assay for accurate measurement of were not reduced in men with T1DM, but 32% had low
this fraction.16 These findings dispel concerns that testos­ levels of calculated free testosterone. After adjustment for
terone is low solely as a result of reduced levels of SHBG age and waist circumference, only calculated free testos­
in men with T2DM and the metabolic syndrome. terone in men with T2DM remained reduced compared
The HERITAGE family study found that low levels with control individuals. A study from Nigeria has also
of testosterone and SHBG predicted increased accu­ shown that the prevalence of low levels of testosterone
mulation of visceral fat on CT scanning as well as in men with T2DM is 36%.24 A meta-analysis by Corona
overall obesity. 18 These findings were supported by et al.25 confirmed that patients with T2DM have lower

480  |  AUGUST 2013  |  VOLUME 9  www.nature.com/nrendo


© 2013 Macmillan Publishers Limited. All rights reserved
REVIEWS

plasma levels of testosterone than individuals who do not Box 2 | WHO definition of T1DM and T2DM
have T2DM.
1. Diabetes symptoms (that is, polyuria, polydipsia and
unexplained weight loss) plus
Testosterone and insulin resistance ■■ a random venous plasma glucose concentration
Evidence for a bidirectional mechanism >11.1 mmol/l or
Epidemiological studies as described above consistently ■■ a fasting plasma glucose concentration >7.0 mmol/l
demonstrate an association between low levels of testos­ (whole blood >6.1 mmol/l) or
terone in men and obesity, the metabolic syndrome and ■■ 2 h plasma glucose concentration >11.1 mmol/l
T2DM.13,14,16,26 In an 8‑year follow-up study—the Rancho 2 h after 75 g anhydrous glucose in an oral glucose
tolerance test
Bernardo study—low levels of testosterone predicted
2. With no symptoms, diagnosis should not be based on
insulin resistance and incident T2DM in older adults a single glucose determination but requires confirmatory
(55–89 years). 6 Both the Massachusetts Male Aging plasma venous determination. At least one additional
Study (15-year follow-up) and the NHANES III study glucose test result on another day with a value in the
have found an increased risk of the metabolic syndrome diabetic range is essential, either fasting, from a random
and T2DM in men with low levels of testos­terone, even in sample or from the 2 h postglucose load. If the fasting or
men who were not obese at the beginning of the study.8,9 random values are not diagnostic, the 2 h value should
The Quebec family study evaluated 130 nonsmoking men be used.
Abbreviations: T1DM, type 1 diabetes mellitus; T2DM, type 2
and showed that men in the upper tertile of testos­terone diabetes mellitus.
levels had a lower risk of being characterized by three
or more features of the metabolic syndrome (OR 0.24,
P <0.04), independent of age, than did men from the
Box 3 | Definition of late-onset hypogonadism
other tertiles.20 Laaksonen et al.7 looked at the associa­
tion of testosterone and SHBG with development of the This definition is based on International Society of
metabolic syndrome and T2DM in men. After 11 years Andrology (ISA), International Society for the Study of
Aging Male (ISSAM) and European Association of Urology
of follow-up, men with total testosterone, free testos­
(EAU) recommendations.163
terone and SHBG in the lower quartile had an increased
Clinical features of hypogonadism:
risk of developing the metabolic syndrome (OR 2.3,
■■ Diminished sexual desire (libido) and erectile quality
1.7, 2.8, respectively) and T2DM (OR 2.3, 1.7 and 4.3, and frequency, particularly nocturnal erections
respectively) after adjustment for age. 27 Low levels of ■■ Changes in mood with concomitant decreases
testosterone are also an independent risk factor and a in intellectual activity, cognitive functions, spatial
biomarker for the future onset of both the metabolic syn­ orientation ability, fatigue, depressed mood and irritability
drome and T2DM.4–7 Furthermore, a well-known con­ ■■ Sleep disturbances
sequence of classic hypogonadism is increased fat mass ■■ Decrease in lean body mass with associated
and decreased lean mass. In addition, the metabolic syn­ diminution in muscle volume and strength
■■ Increase in visceral fat
drome and T2DM are highly prevalent in patients with
■■ Decrease in body hair and skin alterations
Klinefelter syndrome.28 These established facts support ■■ Decreased BMD resulting in osteopaenia, osteoporosis
a bidirectional mechanism between levels of testosterone and increased risk of bone fractures
and these conditions. Plus biochemical evidence of low testosterone:
Low levels of total testosterone were previously ■■ Total testosterone levels >12 nmol/l (346 ng/dl) or
believed to reflect reduced levels of SHBG. A landmark free testosterone levels >250 pmol/l (72 pg/ml) do not
study by Dhindsa and colleagues16 confirmed this associ­ require testosterone substitution
ation by demonstrating that free testosterone measured ■■ Total testosterone levels <8 nmol/l (231 ng/dl) or
by equilibrium dialysis was lower in men with T2DM free testosterone <180 pmol/l (52 pg/ml) require
testosterone substitution
than in men without T2DM. This finding is supported
■■ Total testosterone between 8 nmol/l and 12 nmol/l,
by studies that have reported low levels of measured bio­ trials of treatment can be considered in those in
available testosterone in patients with T1DM.13,29 As these whom alternative causes of these symptoms have
assays are usually only available in research laboratories, been excluded
mathema­tical equations have been developed to measure
both free and bioavailable testosterone on the basis of
total testosterone, albumin and SHBG levels.17,29–31 has demonstrated an inverse correlation between insulin
The role of SHBG in men with T2DM has been a major resistance and serum levels of SHBG in 23 men with
confounding factor in the diagnosis of hypogonadism. T2DM that was independent of serum levels of insulin
However, low levels of SHBG are also an independent or C peptide, as well as being independent of obesity
risk factor for subsequent onset of the metabolic syn­ and accumulation of abdominal fat.34 Although levels
drome and T2DM.4,32 Production of SHBG is reduced in of SHBG are reduced overall in patients with insulin
states of insulin resistance, including obesity, the meta­ resistance, these patients do not all have low levels of
bolic syndrome and T2DM. This reduction might be this glyco­protein. For example, Kapoor and col­leagues13
because insulin is a potent inhibitor of the production found a wide range of levels of SHBG in patients with
of SHBG in the liver 33 and the above conditions have T2DM, from well below the normal range to levels above
high circulating levels of insulin. However, one study the normal range. The other major factors affecting

NATURE REVIEWS | ENDOCRINOLOGY VOLUME 9  |  AUGUST 2013  |  481


© 2013 Macmillan Publishers Limited. All rights reserved
REVIEWS

Box 4 | Factors that affect SHBG levels confirms that the reduction in the levels of testosterone
associated with T2DM cannot be entirely attributed
Increased SHBG to decreased levels of SHBG. Furthermore, a similar
■■ Ageing
associa­tion with free testosterone was demonstrated by
■■ Thyrotoxicosis
the NHANES III survey, with an odds ratio of preva­
■■ Anticonvulsants
■■ Smoking lent T2DM as high as 4.12 (95% CI 1.25–13.5; P = 0.04)
■■ Acromegaly in men with levels of calculated free testosterone in the
■■ HIV lowest tertile, even after adjusting for other confounding
■■ Cirrhosis of liver factors such as age, adiposity and ethnicity.8
■■ Thiazolidinediones
Decreased SHBG Insulin resistance and glycaemic control
■■ Insulin resistance Liver, muscle and fat are the major tissues that con­
■■ Obesity tribute to the total body insulin sensitivity. Excess intake
■■ The metabolic syndrome
of calories combined with increasing insulin resistance in
■■ Type 2 diabetes mellitus
■■ Hypothyoidism
adipose tissue impairs the complete removal of free fatty
■■ Glucocorticoids acids from the circulation by adipocytes, which leads
■■ Anabolic steroids to an overspill of free fatty acids into the liver, muscle
■■ Low albumin states and other nonadipose tissues. Hepatic steatosis and
■■ Atorvastatin non­alcoholic steatotic hepatitis, intramyocellular fat in
Abbreviation: SHBG, sex hormone-binding globulin. muscle and atherosclerosis are therefore consequences
of insulin resistance.39 Thus, insulin resistance forms a
key target of therapy for many medications, such as met­
levels of SHBG in this group of patients are age and formin and thiazolidinediones, that are commonly used
use of statins or thiazolidine­diones,35,36 as well as other in the treatment of T2DM.
well-recognized risk factors (Box 4). A state of low testosterone is associated with an
Levels of SHBG rise with age,37 which might account increased risk of developing insulin resistance and
for the disparity in levels of SHBG in patients with T2DM in healthy men.4–6 The Health in Men study
T2DM. Atorvastatin, but not simvastatin, is associated showed that low levels of total testosterone are associ­
with reduced levels of SHBG and consequently reduced ated with insulin resistance independently of measures
levels of total testosterone in men with T2DM, but of central obesity in older men (>70 years).40
does not alter levels of bioavailable testosterone.35 This Testosterone is negatively correlated with insulin
finding implies that a homeostatic regulation system resistance and levels of HbA1c.41–43 Several interventional
maintains the appropriate physiological level of biologi­ studies have shown an improvement in insulin resistance
cally active testosterone. The reduction in levels of total and glycaemic control with testosterone replacement
testosterone by atorvastatin is dose dependent, that is, therapy (TRT).43–47 Our group was the first to demon­
higher doses result in a greater reduction than low doses. strate that TRT improved insulin resistance and glycae­
Simvastatin is a weaker statin than atorvastatin, which mic control in men with hypogonadism who also had
probably explains why simvastatin does not reduce levels T2DM.43 This finding led to a multicentre European trial,
of total testosterone. the Testosterone In Metabolic Syndrome and type 2 dia­
Rosiglitazone, which reduces insulin resistance by betes study (TIMES 2), which is a large (n = 220) prospec­
~30%, increases levels of total testosterone and SHBG.36 tive randomized double-blind placebo controlled study
However, rosiglitazone also increased the levels of bio­ (RCT) in men with hypogonadism and T2DM and/or the
available testosterone, which is a further indication that metabolic syndrome. TRT reduced HOMA-IR by 15.2%
reduced levels of total testosterone in men with T2DM at 6 months and 16.4% at 12 months in the total study
cannot be fully accounted for by the changes in levels of population; similar results were found in the subgroup
SHBG. In one small study of just 10 healthy men, rosigli­ of men with hypogonadism and T2DM. Thus, TRT has
tazone did reduce total testosterone production or levels a similar efficacy to metformin with regards to insulin
of dihydrotestosterone.38 However, the study did not resistance. Metformin is used as a first-line therapy in
measure the levels of SHBG or bioavailable testosterone. T2DM, and the effects of TRT are additive to those of
A meta-analysis of 28 cross-sectional studies that metformin as 70% of the group with T2DM were already
was published in 2011 showed that levels of total testos­ on metformin. In addition, thiazolidinediones improve
terone were lower in men with T2DM (n = 1,822) than insulin resistance by 30%.48
in healthy control individuals (n = 10,009), with a mean Testosterone is known to influence body composition,
difference of –2.99 nmol/l (95% CI –3.59 to –2.4), and promoting increased lean mass and suppressing fat mass.
the association was independent of age and BMI.17 The TRT takes several weeks to affect body composition in
study showed that the level of SHBG was slightly reduced men with hypogonadism. However, changes in insulin
in patients with T2DM, which is consistent with a state sensitivity occur as early as 2 weeks after withdrawal
of insulin resistance; however, the meta-analysis also of sex steroids in young patients (aged 40.8 ± 2.8 years)
showed that measured levels of free testosterone were with idiopathic hypogonadotrophic hypogonadism.49
appreciably reduced in patients with T2DM. This finding BMI remained unchanged. This finding suggests that

482  |  AUGUST 2013  |  VOLUME 9  www.nature.com/nrendo


© 2013 Macmillan Publishers Limited. All rights reserved
REVIEWS

the mechanisms controlling insulin sensitivity are more


Hypothalamus
complex than previously thought, involving short-term
and long-term changes. –
Dysregulation of the metabolism of fatty acids that Kisspeptin

leads to inappropriate lipid accumulation in myocytes,

hepatocytes and β cells is strongly associated with insulin
resistance in patients with obesity, T2DM and impaired
glucose tolerance. 50–55 TRT decreased the accumula­
tion of fat in the liver, as assessed by CT scanning, and Pituitary Proinflammatory
gland cytokines
increased insulin sensitivity compared with placebo (TNF, IL-1β and IL-6)
after 18 weeks of treatment in 67 obese men with severe
obstructive sleep apnoea. 56 This finding suggests that
testosterone has a protective effect on the development
of hepatic steatosis.
Testosterone inhibits lipoprotein lipase, which reduces Testes
Leptin
the uptake of triglycerides by adipocytes in men.57 In –
addition, men with low levels of testosterone, including
those with reduced glucose tolerance and T2DM, have
decreased oxidative phosphorylation in muscle mito­
chondria.41 This mitochondrial dysfunction contributes Size and
to the degree of insulin resistance. Testosterone might number Adipose Insulin
Estradiol Testosterone
tissue resistance
also increase insulin sensitivity by suppressing pro­ Aromatase
inflammatory cytokines such as tumour necrosis factor Lipoprotein
Promotes adipocyte maturity
(TNF) and IL‑6, which are thought to promote insulin lipase Uptake of
resistance.58,59 The actions of testosterone on insulin sen­ triglycerides
si­tivity are discussed in more detail towards the end of
this Review. Figure 1 | The hypogonadal–obesity–adipocytokine hypothesis. Increased amounts
of adipose tissue increase the activity of aromatase (which converts testosterone to
estradiol) in adipocytes. Estradiol directly inhibits the hypothalamic–pituitary–testes
Obesity, inflammation and low testosterone axis via Kisspeptin, which leads to decreased testosterone production. Adipose
Obesity, adipocytokines and the chronic hypogonadal tissue also produces leptin and proinflammatory cytokines that have a negative
state have a complex pathophysiological interaction that feedback effect on the hypothalamic–pituitary–gonadal axis. Leptin also inhibits the
contributes to a vicious cycle of fat accumulation. This stimulatory action of gonadotropins on the Leydig cells of the testes, which results
accumulation of fat causes further reductions in testos­ in decreased androgen production from the testes. Reduced levels of testosterone
terone production, which in turn promotes additional fat in the tissues facilitates triglyceride storage in adipocytes by increasing the activity
of lipoprotein lipase. Abbreviation: TNF, tumour necrosis factor.
deposition (Figure 1).
Testosterone production is regulated by the
­hypothalamic–pituitary–testicular axis. Pulsatile release Kisspeptin (encoded by Kiss1) is a neuropeptide
of gonadotropin-releasing hormone stimulates the hormone in the hypothalamus that acts on a G‑protein
release of luteinizing hormone (and follicle-­stimulating coupled receptor to release gonadotropin-releasing
hormone), which then stimulate the testes to synthe­ hormone, which causes the release of luteinizing hor­
size and secrete testosterone. The axis is regulated by mone, follicle-stimulating hormone and testos­terone.
the direct negative feedback of testosterone on the Kisspeptin has a key role in the inhibitory effect of
hypo­t halamus. Testosterone is metabolized to estra­ estrogen on the hypothalamus.65,66 Estrogens and leptin
diol, primarily in adipose tissue by aromatase, which resistance suppress the neuronal kiss­peptin response.67
has increased activity in visceral fat. Estradiol directly A study using ovariectomised rats has shown that pro­
feeds back and inhibits the hypothalamic–pituitary–­ inflammatory agents such as lipo­polysaccharide reduce
testicular axis. Aromatase inhibitors and antiestrogens the expression of Kiss1 mRNA in the hypo­thalamus, as
such as clomiphene increase testosterone levels in men well as reducing levels of luteiniz­ing hormone.68 These
with secondary hypogonadism, 60,61 which supports effects are blocked by anti-­inflammatory agents such
these findings. Furthermore, adipocytokines, includ­ as indomethacin.68
ing the pro­i nflammatory cytokines TNF, IL‑6 and The above argument is that obesity causes hypo­
IL‑1β, inhibit the secretion of testosterone, both at the gonadism; however, evidence suggests that the low
hypothalamic–­pituitary and the testicular level.62 Leptin testos­terone state actually promotes an increase in fat
usually stimulates the release of gonadotropin-releasing mass through different mechanisms. Testosterone pro­
hormone; however, in obesity, where excess leptin is motes the maturation of pluripotent stem cells toward
produced from adipocytes, the hypothalamic–pituitary myocytes, whereas testosterone deficiency permits
axis becomes resistant to leptin.63,64 In addition, leptin the production of adipocytes. 69 This action is medi­
in­h ibits the stimulatory action of gonadotropin on ated through activation of T‑cell-specific transcription
the Leydig cells of testes, thereby further decreasing factor 4 by translocation of an androgen receptor–β-
­testosterone production.64 catenin complex into the nucleus, which downregulates

NATURE REVIEWS | ENDOCRINOLOGY VOLUME 9  |  AUGUST 2013  |  483


© 2013 Macmillan Publishers Limited. All rights reserved
REVIEWS

the adipogenic transcription factors.70 Testosterone of testosterone used. In two other RCTs, which were
inhibits lipoprotein lipase and thereby decreases adipo­ published after the meta-analysis, a decrease in levels of
genesis and increases lipolysis through β‑adrenergic both total cholesterol and LDL cholesterol was noted in
receptors.71 Lipoprotein lipase catalyses the extracellular 161 men with late-onset hypogonadism and in 220 men
component of the circulating triglycerides to free fatty with the metabolic syndrome and/or T2DM.47,81 TRT
acids, which facilitates their uptake into the adipocyte consistently results in decreased levels of total cholesterol
where they are then converted back to triglycerides and in patients with the metabolic syndrome and T2DM.47,82
stored within the cell. Lack of inhibition of lipoprotein
lipase by testosterone facilitates free fatty acid uptake and HDL cholesterol
triglyceride storage, which promotes adipocyte maturity. Levels of testosterone correlate positively with levels of
The sensitivity of the androgen receptor is affected HDL cholesterol in cross-sectional studies of healthy men
by the CAG repeat polymorphism in exon 1 of the gene and those with T2DM.83,84 Interventional studies of TRT
that encodes the receptor. The lower the number of have reported conflicting effects on levels of HDL choles­
CAG repeats the greater the sensitivity of the androgen terol. The differences might be accounted for by the
receptor. Men with T2DM who have a high number of differ­ing lengths of the studies, testosterone preparations
CAG repeats within the androgen receptor gene have a used, ages of the patients and the inclusion of patients
greater waist circumference, BMI, serum levels of leptin with chemically induced hypogonadism and those with
and systo­lic blood pressure than men with a low number hypogonadism as a result of natural causes. A meta-
of CAG repeats.72 Hence, more studies are required to analysis showed that testosterone did not have a major
ascertain the relationship between CAG repeats within overall effect on levels of HDL cholesterol.85 A reduction
the androgen receptor gene and the development in levels of HDL cholesterol was seen in studies where
of obesity. levels of testosterone before treatment were increased.85
This complex interplay of adipocytokines, the immune The magnitude of the decrease in levels of HDL choles­
system, obesity and low levels of testosterone is described terol was lower in studies that used testos­terone esters
by the hypogonadal–obesity–adipocytokine hypo­ than in studies that used other preparations. Lipoprotein
thesis,73 which is an extension of Cohen’s hypothesis of lipase is suppressed by testos­terone and is involved in the
the hypogonadal–obesity cycle.74 In the early stages production of HDL choles­terol from VLDL cholesterol
of the develop­ment of obesity, homeostatic mechanisms and hence this effect con­tributes to low levels of HDL
compensate to maintain testosterone production. These cholesterol upon treatment with testos­terone.86–88 Langer
mechanisms might be particularly efficient in young et al.89 have suggested that testos­terone decreases levels
men, however, a ‘tipping point’ is reached at which these of HDL cholesterol by intensifying reverse cholesterol
mechanisms cannot correct the imbalance and a state transport, thereby enabling HDL cholesterol to remove
of hypogonadism ensues. The vicious cycle of increas­ excess cholesterol from the arterial wall and transport it
ing obesity and reduced levels of testosterone persists, to the liver for disposal, which leads to increased catabo­
which causes an increase in insulin resistance and thus lism of HDL cholesterol. If this suggestion is correct,
an increased risk of developing the metabolic syndrome, then this effect is more benefi­cial than detrimental in
prediabetes and T2DM. the process of preventing atherosclerosis.
In summary, the current data surrounding testos­terone
Testosterone and dyslipidaemia and HDL cholesterol is conflicting and therefore remains
Dyslipidaemia often occurs in patients with T2DM and controversial. Some studies have shown that testosterone
its management is critical to improving cardiovascular decreases levels of HDL cholesterol,90,91 whereas other
outcomes. Raised levels of total cholesterol and LDL studies have not.92,93 Supplementation of testosterone
cholesterol are omitted from the diagnostic criteria of in the testicular feminized mouse, which has low levels
the metabolic syndrome, whereas low levels of HDL of testosterone and an insensitive androgen receptor,
cholesterol and hypertriglyceridaemia are included in demon­strated an increase in levels of HDL cholesterol
the definition. Testosterone status affects the various compared with control mice.94 This finding suggests that
components of the lipid profile, which is discussed in the effect of testosterone on HDL cholesterol is complex
this section. and mediated through mechanisms that are dependent
on or independent of androgen receptors.
Total and LDL cholesterol
Endogenous testosterone is negatively correlated with Triglycerides
levels of both total and LDL cholesterol;29,75–79 therefore, Hypertriglyceridaemia has been associated with the
men with hypogonadism might be at increased risk of development of cardiovascular disease, particularly in
hypercholesterolaemia. A meta-analysis concluded that patients with T2DM.95–97 In castrated cholesterol-fed
TRT did not have any effect on levels of LDL cholesterol; male rabbits, serum levels of triglycerides were lower
however, overall total cholesterol was reduced, with in the testosterone-treated group than in the placebo
the effect being most pronounced in men with hypo­ group. 98 In a study of 1,619 middle-aged men with
gonadism.80 A major limitation of this meta-analysis symp­toms of hypogonadism, low levels of testosterone
was the heterogeneity of the RCTs included with respect were associated with raised levels of triglycerides. 77
to the dose, duration of treatment and the preparation A prospective observational study of 1,438 men with

484  |  AUGUST 2013  |  VOLUME 9  www.nature.com/nrendo


© 2013 Macmillan Publishers Limited. All rights reserved
REVIEWS

hypogonadism who were treated with long-acting intra­ 6–10 year follow-up period.114 An association was found
muscular testosterone undecanoate reported a notable between baseline levels of testosterone and all-cause mor­
reduction in nonfasting levels of triglyceride from tality, cardiovascular mortality and cancer-related deaths
baseline.99 Some interventional studies have shown a (P <0.001) after correcting for other comorbidities. An
decrease in levels of triglycerides after TRT,47,100 however, increase of 6 nmol/l serum testosterone (~1 SD) was
the majority of interventional studies and randomized associated with a ~14% reduction in the risk of mortality
double-blind studies have not found a difference in either over the study period. In a longitudinal follow-up study,
fasting or postprandial serum or chylomicron levels of 930 men with coronary artery disease diagnosed after an
triglycerides between testosterone-treated groups and angiography were followed up for 7 years.115 The overall
placebo groups.101 prevalence of biochemical testosterone deficiency in the
cohort with coronary artery disease using bio­available
Lipoprotein(a) testosterone <2.6 nmol/l was 20.9%, using total testos­
Lipoprotein(a) is a powerful independent risk factor for terone <8.1 nmol/l was 16.9% and using either was 24%.
the development of coronary atherosclerosis.102 In the Mortality in patients with testosterone deficiency defined
TIMES 2 study, TRT reduced levels of lipoprotein(a) in by levels of bioavailable testosterone was 21%, compared
men with hypo­gonadism with and without T2DM and/ with 12% in patients with normal levels of testosterone.
or the metabolic syndrome.47,103 Whether or not reduc­ A meta-analysis found that reduced levels of testos­
ing levels of lipoprotein(a) in men reduces their risk of terone and increased levels of estradiol correlate with
cardio­vascular disease and ultimately morbidity and increased risk of cardiovascular disease and cardio­
mortality is unknown. vascular mortality.105 Whether low testosterone is just an
association with cardiovascular risk, or an actual cause–
Cardiovascular mortality effect relationship, awaits further studies. Alternatively,
The majority of community-based longitudinal popula­ low levels of testosterone might be an indication of poor
tion studies have shown that low levels of testosterone are general health.106
associated with increased all-cause mortality and cardio­ Two retrospective studies have evaluated the effect
vascular mortality. A large population study (MrOS, of TRT on mortality in men with hypogonadism.113,116
n = 2,416) reported in 2011 that men in the highest A cohort of 1,031 male veterans aged >40 years were
quartile of testosterone levels had a reduced number of followed up for up to 4 years.113 TRT was given to 39%
cardiovascular events compared with the lower quartiles of patients as part of routine care and the remaining
over a median follow-up period of 5 years.104 However, population did not receive any treatment. Mortality was
three meta-analyses found no association between hypo­ reduced in those who received TRT (10.3%) compared
gonadism and cardiovascular events.105–107 A review of with the untreated group (20.7%). Similar findings were
cross-sectional studies found that cardiovascular disease observed during a 6 year follow-up study of 581 men
had either a positive or neutral association with low with T2DM.116 Mortality was 9% in men with a total
levels of testosterone.108 Although the evidence for a link testos­terone >10.4 nmol/l, 17.2% in untreated men (total
between low levels of testosterone and cardiovascular testosterone <10.4 nmol/l) and 8.4% in men receiving
disease is controversial and some studies have shown TRT. Although these are observational studies and not
conflicting outcomes, most of the evidence supports an RCTs, their findings do suggest that TRT might improve
association between endogenous testosterone status and survival in men with hypogonadism.
the risk of developing cardiovascular disease.
Erectile dysfunction, which is a common symptom TRT intervention studies
of hypogonadism, is an independent predictor of sub­ Four RCTs have been performed in men with hypo­
sequent atherosclerotic cardiovascular events in these gonadism and T2DM and/or the metabolic syndrome.
men.109,110 Erectile dysfunction is also associated with These studies have been supported by two non-­placebo
increased all-cause mortality, primarily through its trials and two prospective observational studies (Table 1).
association with mortality as a result of cardiovascular A double-blind randomized placebo-controlled add-on
disease.111 Low levels of testosterone were associated with crossover trial demonstrated that testosterone replace­
an increased 6‑month all-cause mortality in a geriatric ment in men with hypogonadism and T2DM improved
population (average age of 75.4 ± 6.7 years), even after insulin resistance and glycaemic control.43 Participants
correcting for other comorbidities.112 The same group (n = 24) were treated with 200 mg of testosterone intra­
evaluated mortality and low levels of testosterone in 850 muscularly or placebo every 2 weeks for 3 months
male veterans >40 years of age. Mortality over 3.5 years followed by a washout period of 1 month before under­
in men with normal testosterone levels (>12 nmol/l) going the alternate treatment phase. TRT led to a sig­
was 20.1% (95% CI, 16.2–24.1%) versus 24.6% (95% CI, nificant decrease in HOMA-IR (–1.73 ± 0.67, P = 0.02),
19.2–30.0%) in men with equivocal testos­terone levels fasting blood glucose (−1.58 ± 0.68 mmol/l, P = 0.03),
(8–12 nmol/l) and 34.9% (95% CI, 28.5–41.4%) in men HbA 1c (–0.37 ± 0.17%, P = 0.03), total cholesterol
with low levels of testosterone (<8 nmol/l).113 (−0.4 ± 0.17 mmol/l, P = 0.03) and waist circumference
The Epic–Norfolk study investigated levels of endo­ (–1.63 ± 0.71 cm, P = 0.03).43
genous testosterone and all-cause and cardiovascular The TIMES 2 47 trial examined the effect of a 2%
mortality in 11,606 healthy men aged 40–79 years over a transdermal testosterone gel on insulin resistance,

NATURE REVIEWS | ENDOCRINOLOGY VOLUME 9  |  AUGUST 2013  |  485


© 2013 Macmillan Publishers Limited. All rights reserved
REVIEWS

Table 1 | Effects of testosterone on glycaemic control and other cardiovascular risk factors
Study Age n Diagnosis Mean Mean Testosterone Treatment Effect on Other outcomes
(years) baseline TT baseline treatment duration HbA1c (%)
(nmol/l) HbA1c (%) (weeks)
Boyanov 57 48 T2DM 9.6 7.3 TU oral 120 mg/d 12 –1.8 BW, WHR,
(2003)118 (P <0.05) % body fat
Kapoor 64 24 T2DM 8.6 7.3 TE im 200 mg 12 –0.37 IR, BW, WHR,
(2006)43 2 weekly (P <0.05) % body fat, TC
and leptin
Gopal 44 22 T2DM 10.1 7.0 TE im 200 mg 12 NS None
(2010)119 2 weekly
Hackett 62 190 T2DM 9.1 7.6 TU im 1,000 mg 30* –0.42%‡ WC, TC
(2011)117 12 weekly (P <0.05)
Jones 60 220 T2DM and/or 9.4 7.3§ Transdermal 26 –0.4§ IR, WC, % body fat
(2011)47 MetS testosterone (P <0.05) TC, LDL‑C, Lp(a)
60 mg/d HDL‑C at
6 months, no
difference at
12 months
Heufelder 57 32 Newly diagnosed 10.5 7.5 Transdermal 52 –0.8 IR, WC, TG
(2009)100 T2DM and MetS testosterone (P <0.05) HDL‑C, CRP, PAI‑1
50 mg/d
Kalinchenko 71 184 MetS 6.7 ND TU im 1,000 mg 30 Not checked IR, WC, BMI, leptin,
(2010)82 (TRT group) 12 weekly CRP, TNF, IL‑1β
7.5
(placebo group)
Zitzmann 49 1,493 Men with 9.6 7.9 TU im 36–52 –1.1 WC, TG, TC,
(2013)99 hypogonadism (P <0.0001) LDL‑C, BP
HDL‑C
Bhattacharya 52 581 Men on Testim 9.0 (men Not Transdermal 52 Not checked FBS, WC, BP
(2011)120 registry with MetS) checked testosterone
(213 had MetS 10.9 (men 5–10 g/d
at baseline) without MetS)
Aversa 58 50 MetS 8.3 5.7 TU im 1,000 mg 52 –1.1 IR, WC, CRP
(2010)167 12 weekly (P <0.05)
*A further 52 weeks open labelled. ‡Patients with poorly controlled diabetes (baseline HbA1c >7.5%). §Patients with type 2 diabetes mellitus only. Abbreviations: BP, blood pressure; BW, body
weight; CRP, C-reactive protein; FBS, fasting blood glucose; HDL‑C, HDL cholesterol; im, intramuscular; IR, insulin resistance; LDL-C, LDL cholesterol; Lp(a), lipoprotein(a); MetS, the metabolic
syndrome; ND, not done; NS, not significant; PAI-1, plasminogen activator inhibitor-1; T2DM, type 2 diabetes mellitus; TC, total cholesterol; TE, testosterone esters; TG, triglycerides; TNF,
tumour necrosis factor; TT, total testosterone; TRT, testosterone replacement therapy; TU, testosterone undecanoate; WC, waist circumference; WHR, waist:hip ratio.

glycaemic control and other cardio­vascular risk factors men with hypogonadism and the metabolic syndrome
over a 12-month treatment period; no changes in hypo­ (n = 184).82 This trial found a reduction in HOMA‑IR,
glycaemic or lipid-lowering medication were allowed waist circumference, BMI, leptin and inflammatory
in the first 6 months. At the end of phase 1 (6 months), cytokines (such as IL‑1β, TNF and C‑reactive protein),
serum levels of total testos­t erone in the patients but no changes in the lipid profile.
who had received the treatment had increased by The BLAST study 117 was an RCT conducted in
19.0 ± 22.1 nmol/l versus 0.1 ± 2.9 nmol/l in the placebo ­primary-care settings. Patients were randomly assigned
group (P <0.001). TRT reduced HOMA‑IR in the overall to either receive testosterone undecanoate or placebo for
population by 15.2% at 6 months (P = 0.018) and 16.4% 24 weeks (n = 190). Levels of HbA1c were reduced in the
at 12 months (P = 0.006) in those with T2DM. In patients group treated with testosterone compared with those of
with T2DM, glycaemic control was con­siderably better in the placebo arm at 18 weeks and 30 weeks. The reduc­
the testosterone-treated group than in the placebo group tion in levels of HbA1c was greater in the group with
at 9 months (–0.446%, P = 0.035), although alterations in poorly controlled T2DM in whom the baseline level of
medications were allowed after 6 months. No change HbA1c was >7.5% (0.42% at 30 weeks and 0.72% after
in HOMA‑B (a measure of insulin secretion) was a further 52 weeks of open label medication). In all of
observed. Levels of total and LDL cholesterol (des­pite the above TRT intervention studies, HbA1c was not the
many patients being treated with statins), and lipo­ primary outcome and hence more research is needed
protein(a) decreased. A small fall was observed in in this area to understand the effects of testosterone on
levels of HDL cholesterol at 6 months that became non­ glycaemic control.
significant after 12 months. A placebo controlled RCT The first study to investigate the hypothesis that
was performed to investigate the effects of intra­muscular testos­t erone improves glycaemic control was con­
injection of long-acting testosterone undecanoate in ducted by Boyanov and colleagues.118 They looked at

486  |  AUGUST 2013  |  VOLUME 9  www.nature.com/nrendo


© 2013 Macmillan Publishers Limited. All rights reserved
REVIEWS

48 middle-aged men with T2DM, symptoms of andro­ loss in the treated group was 21 kg at 10 weeks after start­
gen deficiency and low levels of testosterone. This was ing the very low energy diet. In the treatment group,
an open label study with 24 participants receiving tes­ increases in the levels of SHBG, testosterone and HDL
tosterone un­decanoate orally and the other 24 receiving cholesterol, as well as decreases in levels of insulin and
no treatment for 3 months. Levels of HbA1c improved leptin, were maintained until the end of the 8‑month
significantly from 10.4% to 8.6%. Other parameters also follow-up period.123 Similarly, another study has shown
improved, such as body weight, waist:hip ratio and per­ that weight loss, whether by diet or surgery, leads to sub­
centage body fat. However, the study was not blinded and stantial increases in levels of total testosterone, especially
so did not include a placebo group. in men with morbid obesity, proportional to the amount
Heufelder and co-workers looked at the effect of of weight lost.122
transdermal testosterone with supervised diet and Bariatric surgery that leads to considerable weight loss
exercise in patients with hypogonadism, the metabolic also increases levels of endogenous testosterone. A study
syndrome and newly diagnosed T2DM.100 No glucose- in Salt Lake City, UT, USA, looked at 64 men with severe
lowering agents were administered prior to or during the obesity, of whom 22 had gastric bypass surgery and 42
study. Levels of HbA1c improved significantly (–0.8%, patients were controls. After 2 years, patients who had
P <0.001) in the group that received testosterone along undergone a gastric bypass had a significantly decreased
with the diet and exercise programme compared with BMI (−16.6 ± 1.2 versus −0.46 ± 0.51 kg/m 2 ) and
the group that just received the diet and exercise pro­ decreased levels of estradiol (−29.7 ± 8.8 pmol/l versus
gramme. All the patients who received testosterone 5.9 ± 5.1 pmol/l), compared with the control group.
attained levels of HbA1c <7.0%, and 87.5% achieved levels Patients who had undergone gastric bypass also had an
of HbA1c <6.5%. HOMA-IR and levels of adiponectin increase in levels of total testosterone (10.7 ± 1.65 nmol/l
and high-sensitivity C‑reactive protein also improved. versus 0.49 ± 0.53 nmol/l) and free testosterone (156.8 ±
However, a study from India did not show any notable 17.7 pmol/l versus −1.38 ± 10.4 pmol/l), compared with
improvement in levels of HbA1c or HOMA-IR following the control group. 124 In another prospective study,
intramuscular administration of testosterone. This study 20 men with morbid obesity were followed up for
looked at 22 men with T2DM and participants were 24 months. Group A included 10 patients who under­
very lean with a mean baseline BMI of 24 kg/m2, which went lifestyle modifications (exercise and diet) for
might have influenced the results.119 The TriUS (Testim 4 months and then had gastric bypass surgery, whereas
Registry In United States) study was a 12 month, multi­ Group B included 10 patients who received no inter­
centre, prospective observational study in 849 men with ventions. BMI reductions were 24.7 kg/m2 (P <0.0001)
hypogonadism who were receiving a 1% testosterone and 0.7 kg/m2 (P >0.05) for groups A and B, respec­
gel.120 After 12 months of testosterone therapy, patients tively, after 24 months. Comparing groups A and B,
with the metabolic syndrome (37% of patients had the International Index of Erectile Function 5 sexual health
metabolic syndrome at baseline) showed a decrease in score and levels of total testosterone and free testos­
fasting plasma levels of glucose, waist circumference terone increased significantly in group A (P = 0.0224,
and blood pressure. IPASS (International Multicentre 0.0043 and 0.0149, respectively).125
Post-authorization Surveillance Study) was a large study These studies confirm a beneficial effect of weight loss
of 1,438 men with hypogonadism, of whom 14% had and exercise on levels of testosterone. However, whether
T2DM, who were followed up for up to 12 months. The the modest rise in total testosterone as described above
participants were treated with injectable long-acting will ameliorate symptoms of hypogonadism and meta­
testos­terone undecanoate. A sub-analysis of 60 men with bolic imbalance is not clear; whether or not weight loss
uncontrolled T2DM, as defined by levels of HbA1c >6.1% is maintained is also unclear from these studies. Bariatric
(mean HbA1c 7.9%), demonstrated a mean fall of HbA1c surgery is more effective than lifestyle modifications at
of 1.1% after 12 months.99 A meta-analysis has shown increasing levels of testosterone but is not readily avail­
that TRT was associated with a reduction of fasting able, as access to the surgery is regulated by strict criteria
plasma levels of glucose, HOMA-IR, triglycerides and and the procedure is costly and not without surgical and
waist circumference.121 In addition, an increase of HDL postsurgical risk. In our experience of routine clinical
cholesterol was also observed. practice, only a small proportion of patients will lose a
considerable amount of weight and maintain this weight
Weight loss and bariatric surgery loss in the long term with lifestyle modifications.
Lifestyle modifications that cause weight loss improve
levels of both free and total testosterone.122 Losing Insulin-related metabolic pathways
10% of the starting body weight with diet and exercise The beneficial effects of testosterone on insulin sensi­
programmes can achieve a rise in total testosterone of tivity observed clinically might be attributable to a
2–4 nmol/l, whereas bariatric surgery can increase levels com­plex regulatory influence on insulin signalling and
of total testosterone by up to 10 nmol/l.122 An RCT was glucose homeostasis in the major insulin-responsive
performed in which one group received a 4‑month target tissues, such as skeletal muscle, liver and adipose
weight-loss programme with 10 weeks of a very low (Figure 2). Impaired insulin sensitivity in these three
energy diet and 17 behaviour modification visits and tissues is charac­terized by defects in insulin-stimulated
the other group received no interventions. The weight glucose transport activity (particularly into skeletal

NATURE REVIEWS | ENDOCRINOLOGY VOLUME 9  |  AUGUST 2013  |  487


© 2013 Macmillan Publishers Limited. All rights reserved
REVIEWS

T
Insulin +ve Glucose T
+ve
Insulin
receptor GLUT4

P
IRS1/2 Translocation T T
P +ve +ve
P P T HK2
PKC AKT +ve
Glycogen synthase
T G6P dehydrogenase
+ve G6P G6P Glycogen
T
+ve Glycogen
GLUT4 phosphorylase
vesicle
6-P-gluconolactone Fructose-6-P
T
T –ve
PFK +ve

6-P-gluconate Fructose-1,6-P
Pentose
phosphate Gluconeogenesis Glycolysis
shunt Krebs
Ribulose-5P Glyceraldehyde-3-P cycle
Oxidative
phosphorylation
PEP

PEP T UQCRB Citrate


Nucleotide PEPCK +ve
synthesis
Oxaloacetate Pyruvate
kinase Acetyl-coenzyme A

Pyruvate Pyruvate Pyruvate


dehydrogenase

Figure 2 | Potential mechanism of testosterone action on cellular insulin sensitivity and glucose homeostasis.
Testosterone increases insulin receptor expression and IRS expression and phosphorylation, which enhances cellular
responsiveness to insulin. GLUT4 expression and membrane translocation is also raised in response to testosterone,
which increases cellular glucose uptake and utilization potentially via an increase in the phosphorylation of AKT and PKC.
Key enzymes in the glycolysis pathway are additionally influenced by testosterone, with HK2, PFK and (G6P) dehydrogenase
expression increased; this results in glucose utilization via the pentose phosphate shunt or pyruvate citric acid cycle.
Furthermore, the activity of glycogen synthase, the rate-limiting enzyme in glycogenesis, is increased by testosterone,
whereas glycogen phosphorylase activity is decreased to reduce glycogen breakdown and diminish cellular levels of free
glucose. Mitochondrial oxidative phosphorylation is increased by testosterone via upregulation of UQRCB, which is involved
in the activation of ubiquinol cytochrome c reductase and has a critical role in the biochemical generation of ATP. Dashed
arrows indicate pathways in which multiple steps are involved but not shown. Abbreviations: GLUT4, glucose transporter 4;
G6P, glucose‑6-phosphate; HK2, hexokinase 2; IRS, insulin receptor substrate; T +ve indicates targets or activity increased
by testosterone; T -ve indicates targets decreased by testosterone; P, phosphate; PEP, phosphoenolpyruvate; PEPCK,
phosphoenolpyruvatecarboxykinase; PFK, phosphofructokinase; PKC, protein kinase C; UQCRB, ubiquinol cytochrome c
reductase-binding protein.

muscle), impaired insulin-mediated inhibition of glucose and lipids in muscle also impairs insulin signalling,
production in the liver and stimulation of glycogen syn­ which leads to decreased insulin sensitivity. Ultimately,
thesis in the liver, and a reduced ability of insulin to these effects lead to peripheral hyperinsulinaemia, sys­
inhibit lipolysis in adipose tissue.126 In addition, fat depo­ temic insulin resistance and hepatic steatosis. There­
sition in nonadipose tissue, including skeletal muscle fore, a reduction in the size and number of adipocytes
and the liver occurs as a ‘spillover’ effect of dissociated by testos­terone would reduce the release of free fatty
fatty acid release from adipose tissue surplus to energy acids and, in turn, insulin resistance. Moreover, a lack of
requirements in other organs.127 In turn, this lipid accu­ testos­terone action decreases the activity of key enzymes
mulation contributes to impaired insulin responsiveness in the regulation of hepatic fatty acid synthesis in animal
and abnormalities in glucose control. Interestingly, levels models, including sterol regulatory element binding
of fat in the liver and muscle correlate more strongly with protein‑1c, stearoyl‑CoA desaturase 1, acetyl CoA car­
insulin sensitivity than levels of abdominal fat in rats.128 boxylase, fatty acid synthase and peroxisome prolifera­or
Indeed, free fatty acids decrease hepatic insulin bind­ activated receptor‑γ.128,130
ing, increase hepatic gluconeogenesis and increase Insulin-stimulated uptake of glucose into muscle
hepatic insulin resistance.129 Build-up of triglycerides and adipose tissue is largely mediated by glucose

488  |  AUGUST 2013  |  VOLUME 9  www.nature.com/nrendo


© 2013 Macmillan Publishers Limited. All rights reserved
REVIEWS

transporter type 4 (GLUT4). GLUT4 translocates from returned the enhanced glycogen phosphorylase activ­
cyto­plasmic vesicles to the cell membrane in response ity in castrated rats to a normal level, thus reducing gly­
to insulin via signalling through the insulin receptor, cogen breakdown and the subsequent rise in levels of
subsequent binding of insulin receptor substrate (IRS) 1 free glucose. Indeed, castration is followed by decreased
and activation of intracellular signalling pathways. 131 levels of glycogen in rat perineal and levator ani muscles
Thus, di­minished expression of GLUT4 correlates with and the administration of testosterone induces a con­
decreased insulin responsiveness, and defects at the siderable increase in glyco­gen content.140,143,144 In addi­
level of expression of the insulin receptor, IRS‑1 and tion, glucose‑6-­phosphate dehydrogenase activity, the
GLUT4 have been observed in patients with T2DM.132 rate limiting enzyme in the pentose phos­phate pathway,
Testosterone increased the expression of GLUT4 in is increased in the rat levator ani muscle fo­l lowing
cultured skeletal muscle cells, hepatocytes and adipo­ adminis­tration of testosterone propionate.145
cytes,133–135 as well as augmenting GLUT4 membrane Regulation of cellular fuel metabolism is impor­
translocation and promoting glucose uptake in adipose tant for maintaining homeostatic energy balance and
and skeletal muscle tissue.134 can affect whole-body metabolic homeostasis and the
Similarly, testosterone treatment increased the expres­ emergence of metabolic disorders. The mitochondria
sion of insulin receptors in the Chang human adult liver have a central role in fuel utilization and energy pro­
and human larynx carcinoma cell line (HEp‑2), which duction, with dis­ordered mitochondrial function at the
resulted in enhanced insulin binding and responsive­ cellular level linked to the pathophysiology of T2DM.
ness through raised utilization of glucose compared Evidence suggests that an association exists between
with the basal response.136,137 Testosterone administra­ serum levels of testosterone and mitochondrial func­
tion in orchidectomized adult male rats increased insulin tion, with testosterone deficiency directly contributing
receptor and IRS‑1 mRNA and protein expression in to ineffi­cient energy utilization in cases of increased
liver tissue and was associated with a normalization of insulin resistance.41,49
the castration-induced impairment of glucose oxida­ An inverse correlation exists between levels of testos­
tion.138 Supporting the effects on insulin receptor signal­ terone and adverse mitochondrial function in men
ling, IRS‑1 was upregulated in cultured adipocytes and with varying degrees of glucose control. 41 Low levels
skeletal muscle cells following testosterone treatment,133 of testos­terone were associated with decreased expres­
as was IRS‑2 in cells isolated from male human skeletal sion of genes involved in the mitochondrial oxidative
muscle biopsy samples.139 Serine phosphorylation of phosphorylation pathway in skeletal muscle biopsy
IRS‑1, which is known to attenuate insulin signalling by samples. Of these genes, those that encode ubiqui­
inhibiting tyrosine phosphorylation, was raised in cas­ nol cytochrome c reductase binding protein and the
trated mice, with testosterone treatment increasing this transcription factor proteasome proliferator activated
effect.138 Increased phosphorylation of AKT and protein receptor‑γ co-­activator 1α (PGC‑1α) had the largest
kinase C, as key steps in the insulin receptor signalling expression difference between normal and diabetic
pathways for regulation of GLUT4 trans­location, was also muscle and was correlated with levels of testosterone
shown upon testosterone treatment of skeletal muscle and insulin resistance. 41 PGC‑1α downregulation in
cells.134 These effects were blocked by the 5α-reductase skeletal muscle is considered to be the mechanism
inhibitor, finasteride, which suggests that local conver­ that causes ­testosterone-deficiency-induced insulin
sion of testosterone to dihydro­testosterone and activation resistance, potentially via a subsequent reduction
of the androgen receptor might be important for glucose of key mito­c hondrial transcriptional factors (for
uptake, at least in isolated rat skeletal muscle cells. example, transcription factor A, mitochondrial and
In addition to insulin signalling and glucose uptake, nuclear respiratory factor‑1) and signalling molecules
experiments have shown that testosterone influences (such as, AKT) to decrease oxidative phosphorylation
key enzymes involved in glycolysis. Indeed, testos­terone and mitochondrial biogenesis. 146,147 Similarly, genes
increased the activity of phosphofructokinase and hex­o­ involved in oxidative phosphorylation and ubiquinone
kinase in cultured rat skeletal muscle cells and increased pathways are downregulated in young male mice fol­
hexokinase activity in castrated rats.134,140 Preliminary lowing orchidectomy, an effect that was reversed by
work in the testicular feminized mouse, whereby testosterone treatment.148
affected animals present low endogenous testos­terone In elderly (28 months old) male mice, testosterone
levels and a nonfunctional androgen receptor, has also treatment in combination with physical exercise training
demonstrated that expression of hexokinase 2 was increased expression of the genes that encode carnitine
reduced in skeletal muscle tissue of testicular feminized palmitoyltransferase‑1β and pyruvate dehydrogenase
mice compared with wild-type mice.141 Testosterone kinase 4 in the mitochondria to improve fatty acid oxida­
treatment did not alter the expression of hexokinase 2, tion and favour fatty acid utilization as the mito­chondrial
which suggests an action that is dependent on the fuel.149 Indeed, testosterone treatment in female rats
androgen receptor. An increase in glycogen synthase increases skeletal muscle levels of fatty acid binding
activity in skeletal muscle is apparent in castrated male protein, which has an important role in the delivery of
rats supplemented with testosterone, diminishing the fatty acids to the mitochondria for oxidation.150 By regu­
raised blood levels of glucose seen in untreated control lating mitochondrial function and efficient fuel metabo­
rats. 142 Concurrently, testosterone administration lism, testosterone might protect against a cellular energy

NATURE REVIEWS | ENDOCRINOLOGY VOLUME 9  |  AUGUST 2013  |  489


© 2013 Macmillan Publishers Limited. All rights reserved
REVIEWS

imbalance that in turn leads to chronically raised levels Conclusions


of glucose and accumulation of lipids that can mediate Epidemiological studies have established an associa­
insulin resistance. tion between hypogonadism in men and an increased
risk of cardiovascular events and metabolic disorders.
Use of TRT Studies have also shown increased mortality in men
Certain studies have shown improvements in angina with hypogonadism, especially mortality as a result
symptoms following TRT151–153 as a result of the effect of cardio­vascular disease. Short-term interventional
of testosterone on coronary vasodilatation 154,155 and studies have shown beneficial effects of testosterone
improvements in exercise capacity and symptoms in therapy on various cardio­vascular risk factors; these
men with heart failure.156,157 Testosterone therapy also include improvements in insulin resistance and glycae­
improves sexual function, especially erectile dysfunc­ mic control, as well as a decrease in visceral fat mass and
tion.158 Furthermore, numerous studies have also shown reduced levels of circulating inflammatory cytokines
that testosterone has profound effects on improvement and markers, an increase in vasodilatation and vaso­
in mood and cognition.159–162 reactivity, improvement in endothelial function (includ­
However, a diagnosis of late-onset hypogonadism ing the number of endo­thelial progenitor cells), and a
should be based on international and/or national guide­ decrease in total levels of cholesterol.10
lines and recommendations.163,164 A decision to initiate It is imperative that TRT should be monitored carefully
TRT should take into account the benefits versus risks with regular assessment of haematocrit, prostate specific
of potential adverse effects in individual patients. TRT antigen and testosterone levels at 3, 6 and 12 months and
is contraindicated in the presence of metastatic prostate then annually. Further research is required in this impor­
carcinoma, breast cancer, prostate nodules and raised tant field of health, including larger and longer-term
levels of prostate specific antigen that have not been RCTs to evaluate fully the beneficial effects and risks of
investigated and prostate cancer excluded, severe lower TRT over time.
urinary tract symptoms associated with benign prostatic
hypertrophy, severe uncontrolled heart failure and levels Review criteria
of haematocrit >50%. The incidence of adverse effects is
A search for original and review articles that focus on
low when testosterone is replaced to achieve levels, by
testosterone, hypogonadism, metabolic syndrome and
titration, within the normal range. The most common type 2 diabetes mellitus was performed in MEDLINE.
adverse effect attributable to TRT is secon­dary poly­ The key search terms were “testosterone”, “androgen”,
cythaemia. This symptom can be treated by dose reduc­ “hypogonadism”, “obesity”, “metabolic syndrome”,
tion, venesection or switching the mode of testosterone “diabetes”, “body composition”, “insulin resistance” and
delivery. No evidence indicates that TRT increases the “glycaemic control”. We also searched the reference lists
risk of prostate carcinoma; however, as this condition can of the identified articles for further papers. Articles were
restricted to English language and no date restrictions
occur in the general population of men >40 years, annual
were applied.
assessment is advised.165,166

1. World Health Organization. Fact sheet No317, 6. Oh, J. Y., Barrett-Connor, E., Wedick, N. M., 13. Kapoor, D., Aldred, H., Clark, S., Channer, K. S. &
cardiovascular diseases [online], http://www. Wingard, D. L. & Rancho Bernardo, S. Jones, T. H. Clinical and biochemical
who.int/mediacentre/factsheets/fs317/en/ Endogenous sex hormones and the development assessment of hypogonadism in men with type 2
(2011). of type 2 diabetes in older men and women: diabetes: correlations with bioavailable
2. British Heart Foundation. Coronary heart disease the Rancho Bernardo study. Diabetes Care 25, testosterone and visceral adiposity. Diabetes
statistics in England 2012 [online], http://www. 55–60 (2002). Care 30, 911–917 (2007).
bhf.org.uk/plugins/PublicationsSearchResults/ 7. Laaksonen, D. E. et al. Testosterone and sex 14. Ding, E. L., Song, Y., Malik, V. S. & Liu, S. Sex
DownloadFile.aspx?docid=e3b705eb‑ceb3‑ hormone-binding globulin predict the metabolic differences of endogenous sex hormones and
42e2‑937d‑45ec48f6a797&version=‑1&title syndrome and diabetes in middle-aged men. risk of type 2 diabetes: a systematic review and
=England+CHD+Statistics+Factsheet+2012& Diabetes Care 27, 1036–1041 (2004). meta-analysis. JAMA 295, 1288–1299 (2006).
resource=FactsheetEngland (2012). 8. Selvin, E. et al. Androgens and diabetes in men: 15. Hofstra, J. et al. High prevalence of
3. American Heart Association. Men and results from the Third National Health and hypogonadotropic hypogonadism in men referred
cardiovascular diseases [online], http://www. Nutrition Examination Survey (NHANES III). for obesity treatment. Neth. J. Med. 66, 103–109
heart.org/idc/groups/heart-public/@wcm/ Diabetes Care 30, 234–238 (2007). (2008).
@sop/@smd/documents/downloadable/ 9. Kupelian, V. et al. Low sex hormone-binding 16. Dhindsa, S. et al. Frequent occurrence of
ucm_319573.pdf (2012). globulin, total testosterone, and symptomatic hypogonadotropic hypogonadism in type 2
4. Stellato, R. K., Feldman, H. A., Hamdy, O., androgen deficiency are associated with diabetes. J. Clin. Endocrinol. Metab. 89,
Horton, E. S. & McKinlay, J. B. Testosterone, sex development of the metabolic syndrome in 5462–5468 (2004).
hormone-binding globulin, and the development of nonobese men. J. Clin. Endocrinol. Metab. 91, 17. Vermeulen, A., Verdonck, L. & Kaufman, J. M.
type 2 diabetes in middle-aged men: prospective 843–850 (2006). A critical evaluation of simple methods for the
results from the Massachusetts male aging 10. Jones, T. H. Testosterone deficiency: a risk estimation of free testosterone in serum. J. Clin.
study. Diabetes Care 23, 490–494 (2000). factor for cardiovascular disease? Trends Endocrinol. Metab. 84, 3666–3672 (1999).
5. Haffner, S. M., Shaten, J., Stern, M. P., Endocrinol. Metab. 21, 496–503 (2010). 18. Couillard, C. et al. Contribution of body fatness
Smith, G. D. & Kuller, L. Low levels of sex 11. Kelly, D. M. & Jones, T. H. Testosterone: and adipose tissue distribution to the age
hormone-binding globulin and testosterone a vascular hormone in health and disease. variation in plasma steroid hormone
predict the development of non‑insulin‑ J. Endocrinol. 217, R47–R71 (2013). concentrations in men: the HERITAGE Family
dependent diabetes mellitus in men. MRFIT 12. Jones, T. H. Effects of testosterone on Study. J. Clin. Endocrinol. Metab. 85, 1026–1031
Research Group. Multiple Risk Factor type 2 diabetes and components of the (2000).
Intervention Trial. Am. J. Epidemiol. 143, metabolic syndrome. J. Diabetes 2, 146–156 19. Svartberg, J., von Mühlen, D., Sundsfjord, J. &
889–897 (1996). (2010). Jorde, R. Waist circumference and testosterone

490  |  AUGUST 2013  |  VOLUME 9  www.nature.com/nrendo


© 2013 Macmillan Publishers Limited. All rights reserved
REVIEWS

levels in community dwelling men. The Tromsø and reduces waist circumference in hypogonadal 55. Unger, R. H. & Orci, L. Lipotoxic diseases of
study. Eur. J. Epidemiol. 19, 657–663 (2004). men with type 2 diabetes. Diab. Vasc. Dis. Res. 5, nonadipose tissues in obesity. Int. J. Obes. Relat.
20. Blouin, K. et al. Contribution of age and declining 135–137 (2008). Metab. Disord. 24 (Suppl. 4), S28–S32 (2000).
androgen levels to features of the metabolic 37. Feldman, H. A. et al. Age trends in the level of 56. Hoyos, C. M. et al. Body compositional and
syndrome in men. Metabolism 54, 1034–1040 serum testosterone and other hormones in cardiometabolic effects of testosterone therapy
(2005). middle-aged men: longitudinal results from the in obese men with severe obstructive sleep
21. Singh, S. K., Goyal, R. & Pratyush, D. D. Is Massachusetts male aging study. J. Clin. apnoea: a randomised placebo-controlled trial.
hypoandrogenemia a component of metabolic Endocrinol. Metab. 87, 589–598 (2002). Eur. J. Endocrinol. 167, 531–541 (2012).
syndrome in males? Exp. Clin. Endocrinol. 38. Vierhapper, H. & Nowotny, P. Reduced production 57. Marin, P., Oden, B. & Bjorntorp, P. Assimilation
Diabetes 119, 30–35 (2011). rates of testosterone and dihydrotestosterone in and mobilization of triglycerides in subcutaneous
22. Grossmann, M. et al. Low testosterone levels healthy men treated with rosiglitazone. abdominal and femoral adipose tissue in vivo in
are common and associated with insulin Metabolism 52, 230–232 (2003). men: effects of androgens. J. Clin. Endocrinol.
resistance in men with diabetes. J. Clin. 39. Ginsberg, H. N. Insulin resistance and Metab. 80, 239–243 (1995).
Endocrinol. Metab. 93, 1834–1840 (2008). cardiovascular disease. J. Clin. Invest. 106, 58. Hotamisligil, G. S. Inflammatory pathways and
23. Biswas, M., Hampton, D., Newcombe, R. G. & 453–458 (2000). insulin action. Int. J. Obes. Relat. Metab. Disord.
Rees, D. A. Total and free testosterone 40. Yeap, B. B. et al. Lower serum testosterone 27 (Suppl. 3), S53–S55 (2003).
concentrations are strongly influenced by age is independently associated with insulin 59. Malkin, C. J. et al. The effect of testosterone
and central obesity in men with type 1 and resistance in non-diabetic older men: the Health replacement on endogenous inflammatory
type 2 diabetes but correlate weakly with In Men Study. Eur. J. Endocrinol. 161, 591–598 cytokines and lipid profiles in hypogonadal men.
symptoms of androgen deficiency and diabetes- (2009). J. Clin. Endocrinol. Metab. 89, 3313–3318 (2004).
related quality of life. Clin. Endocrinol. 76, 41. Pitteloud, N. et al. Relationship between 60. Guay, A. T., Bansal, S. & Heatley, G. J. Effect of
665–673 (2012). testosterone levels, insulin sensitivity, and raising endogenous testosterone levels in
24. Ogbera, O. A., Sonny, C., Olufemi, F. & Wale, A. mitochondrial function in men. Diabetes Care 28, impotent men with secondary hypogonadism:
Hypogonadism and subnormal total testosterone 1636–1642 (2005). double blind placebo-controlled trial with
levels in men with type 2 diabetes mellitus. J. Coll. 42. Mårin, P. et al. The effects of testosterone clomiphene citrate. J. Clin. Endocrinol. Metab. 80,
Physicians Surg. Pak. 21, 517–521 (2011). treatment on body composition and metabolism 3546–3552 (1995).
25. Corona, G. et al. Type 2 diabetes mellitus and in middle-aged obese men. Int. J. Obes. Relat. 61. Guay, A. T., Jacobson, J., Perez, J. B.,
testosterone: a meta-analysis study. Int. Metab. Disord. 16, 991 (1992). Hodge, M. B. & Velasquez, E. Clomiphene
J. Androl. 34, 528–540 (2011). 43. Kapoor, D., Goodwin, E., Channer, K. S. & increases free testosterone levels in men with
26. Wang, C. et al. Low testosterone associated with Jones, T. H. Testosterone replacement therapy both secondary hypogonadism and erectile
obesity and the metabolic syndrome contributes improves insulin resistance, glycaemic control, dysfunction: who does and does not benefit? Int.
to sexual dysfunction and cardiovascular visceral adiposity and hypercholesterolaemia in J. Impot. Res. 15, 156–165 (2003).
disease risk in men with type 2 diabetes. hypogonadal men with type 2 diabetes. Eur. 62. Jones, T. H. & Kennedy, R. L. Cytokines and
Diabetes Care 34, 1669–1675 (2011). J. Endocrinol. 154, 899–906 (2006). hypothalamic-pituitary function. Cytokine 5,
27. Laaksonen, D. E. et al. The metabolic syndrome 44. Marin, P. Testosterone and regional fat 531–538 (1993).
and smoking in relation to hypogonadism in distribution. Obes. Res. 3 (Suppl. 4), 609S–612S 63. Mantzoros, C. S. The role of leptin in human
middle-aged men: a prospective cohort study. (1995). obesity and disease: a review of current
J. Clin. Endocrinol. Metab. 90, 712–719 (2005). 45. Simon, D. et al. Androgen therapy improves insulin evidence. Ann. Int. Med. 130, 671 (1999).
28. Bojesen, A., Host, C. & Gravholt, C. H. sensitivity and decreases leptin level in healthy 64. Isidori, A. M. et al. Leptin and androgens in male
Klinefelter’s syndrome, type 2 diabetes and the adult men with low plasma total testosterone: obesity: evidence for leptin contribution to
metabolic syndrome: the impact of body a 3‑month randomized placebo-controlled trial. reduced androgen levels. J. Clin. Endocrinol.
composition. Mol. Hum. Reprod. 16, 396–401 Diabetes Care 24, 2149–2151 (2001). Metab. 84, 3673–3680 (1999).
(2010). 46. Naharci, M. I., Pinar, M., Bolu, E. & Olgun, A. Effect 65. Herbison, A. E. Multimodal influence of estrogen
29. Barrett-Connor, E. Lower endogenous androgen of testosterone on insulin sensitivity in men with upon gonadotropin-releasing hormone neurons.
levels and dyslipidemia in men with idiopathic hypogonadotropic hypogonadism. Endocr. Rev. 19, 302–330 (1998).
non‑insulin‑dependent diabetes mellitus. Ann. Endocr. Pract. 13, 629–635 (2007). 66. Smith, J. T., Clifton, D. K. & Steiner, R. A.
Intern. Med. 117, 807–811 (1992). 47. Jones, T. H. et al. Testosterone replacement in Regulation of the neuroendocrine reproductive
30. de Ronde, W. et al. Calculation of bioavailable hypogonadal men with type 2 diabetes and/or axis by kisspeptin-GPR54 signaling.
and free testosterone in men: a comparison of 5 metabolic syndrome (the TIMES2 study). Reproduction 131, 623–630 (2006).
published algorithms. Clin. Chem. 52, Diabetes Care 34, 828–837 (2011). 67. George, J. T., Millar, R. P. & Anderson, R. A.
1777–1784 (2006). 48. Lebovitz, H. E. & Banerji, M. A. Insulin resistance Hypothesis: kisspeptin mediates male
31. Morris, P. D., Malkin, C. J., Channer, K. S. & and its treatment by thiazolidinediones. Recent hypogonadism in obesity and type 2 diabetes.
Jones, T. H. A mathematical comparison of Prog. Horm. Res. 56, 265–294 (2001). Neuroendocrinology 91, 302–307 (2010).
techniques to predict biologically available 49. Yialamas, M. A. et al. Acute sex steroid 68. Iwasa, T. et al. Decreased expression of
testosterone in a cohort of 1072 men. Eur. withdrawal reduces insulin sensitivity in healthy kisspeptin mediates acute immune/
J. Endocrinol. 151, 241–249 (2004). men with idiopathic hypogonadotropic inflammatory stress-induced suppression of
32. Kupelian, V. et al. Low sex hormone-binding hypogonadism. J. Clin. Endocrinol. Metab. 92, gonadotropin secretion in female rat.
globulin, total testosterone, and symptomatic 4254–4259 (2007). J. Endocrinol. Invest. 31, 656 (2008).
androgen deficiency are associated with 50. Boden, G. & Shulman, G. I. Free fatty acids in 69. Singh, R., Artaza, J. N., Taylor, W. E., Gonzalez-
development of the metabolic syndrome in obesity and type 2 diabetes: defining their role Cadavid, N. F. & Bhasin, S. Androgens stimulate
nonobese men. J. Clin. Endocrinol. Metab. 91, in the development of insulin resistance and myogenic differentiation and inhibit
843–850 (2006). beta-cell dysfunction. Eur. J. Clin. Invest. 32 adipogenesis in C3H 10T1/2 pluripotent cells
33. Peiris, A. N. et al. Relationship of insulin (Suppl. 3), 14–23 (2002). through an androgen receptor-mediated pathway.
secretory pulses to sex hormone-binding 51. McGarry, J. D. Banting lecture 2001: dysregulation Endocrinology 144, 5081–5088 (2003).
globulin in normal men. J. Clin. Endocrinol. of fatty acid metabolism in the etiology of type 2 70. Singh, R. et al. Testosterone inhibits adipogenic
Metab. 76, 279–282 (1993). diabetes. Diabetes 51, 7–18 (2002). differentiation in 3T3‑L1 cells: nuclear
34. Birkeland, K. I., Hanssen, K. F., Torjesen, P. A. & 52. Savage, D. B., Petersen, K. F. & Shulman, G. I. translocation of androgen receptor complex with
Vaaler, S. Level of sex hormone-binding globulin Mechanisms of insulin resistance in humans beta-catenin and T‑cell factor 4 may bypass
is positively correlated with insulin sensitivity in and possible links with inflammation. canonical Wnt signaling to down-regulate
men with type 2 diabetes. J. Clin. Endocrinol. Hypertension 45, 828–833 (2005). adipogenic transcription factors. Endocrinology
Metab. 76, 275–275 (1993). 53. Jacob, S. et al. Association of increased 147, 141–154 (2006).
35. Stanworth, R. D., Kapoor, D., Channer, K. S. & intramyocellular lipid content with insulin 71. Singh, A. B. et al. The effects of varying doses
Jones, T. H. Statin therapy is associated with resistance in lean nondiabetic offspring of of T on insulin sensitivity, plasma lipids,
lower total but not bioavailable or free type 2 diabetic subjects. Diabetes 48, apolipoproteins, and C‑reactive protein in healthy
testosterone in men with type 2 diabetes. 1113–1119 (1999). young men. J. Clin. Endocrinol. Metab. 87,
Diabetes Care 32, 541–546 (2009). 54. Petersen, K. F. et al. 13C/31P NMR studies on 136–143 (2002).
36. Kapoor, D., Channer, K. S. & Jones, T. H. the mechanism of insulin resistance in obesity. 72. Stanworth, R. D., Kapoor, D., Channer, K. S. &
Rosiglitazone increases bioactive testosterone Diabetes 47, 381–386 (1998). Jones, T. H. Androgen receptor CAG repeat

NATURE REVIEWS | ENDOCRINOLOGY VOLUME 9  |  AUGUST 2013  |  491


© 2013 Macmillan Publishers Limited. All rights reserved
REVIEWS

polymorphism is associated with serum 89. Langer, C. et al. Testosterone up-regulates lipoprotein(a) in men. Am. J. Cardiol. 77,
testosterone levels, obesity and serum leptin in scavenger receptor BI and stimulates 1244–1247 (1996).
men with type 2 diabetes. Eur. J. Endocrinol. cholesterol efflux from macrophages. Biochem. 104. Ohlsson, C. et al. High serum testosterone is
159, 739–746 (2008). Biophys. Res. Commun. 296, 1051–1057 associated with reduced risk of cardiovascular
73. Jones, T. H. Testosterone associations with (2002). events in elderly men: The MrOS (Osteoporotic
erectile dysfunction, diabetes, and the metabolic 90. Bagatell, C. J., Heiman, J. R., Matsumoto, A. M., Fractures in Men) study in Sweden. J. Am. Coll.
syndrome. Eur. Urol. Suppl. 6, 847–857 (2007). Rivier, J. E. & Bremner, W. J. Metabolic and Cardiol. 58, 1674–1681 (2011).
74. Cohen, P. G. The hypogonadal–obesity cycle: role behavioral effects of high-dose, exogenous 105. Corona, G. et al. Hypogonadism as a risk factor for
of aromatase in modulating the testosterone– testosterone in healthy men. J. Clin. Endocrinol. cardiovascular mortality in men: a meta-analytic
estradiol shunt—a major factor in the genesis of Metab. 79, 561–567 (1994). study. Eur. J. Endocrinol. 165, 687–701 (2011).
morbid obesity. Med. Hypotheses 52, 49–51 91. Rossouw, J. E. et al. Risks and benefits of 106. Ruige, J. B., Mahmoud, A. M., De Bacquer, D. &
(1999). estrogen plus progestin in healthy Kaufman, J.‑M. Endogenous testosterone and
75. Barud, W., Palusinski, R., Beltowski, J. & postmenopausal women: principal results from cardiovascular disease in healthy men: a meta-
Wojcicka, G. Inverse relationship between total the Women’s Health Initiative randomized analysis. Heart 97, 870–875 (2011).
testosterone and anti-oxidized low density controlled trial. JAMA 288, 321–333 (2002). 107. Araujo, A. B. et al. Endogenous testosterone and
lipoprotein antibody levels in ageing males. 92. Zgliczynski, S. et al. Effect of testosterone mortality in men: a systematic review and meta-
Atherosclerosis 164, 283–288 (2002). replacement therapy on lipids and lipoproteins in analysis. J. Clin. Endocrinol. Metab. 96,
76. Simon, D. et al. Association between plasma hypogonadal and elderly men. Atherosclerosis 3007–3019 (2011).
total testosterone and cardiovascular risk 121, 35–43 (1996). 108. Jones, R. D., Nettleship, J. E., Kapoor, D.,
factors in healthy adult men: The Telecom Study. 93. Uyanik, B. S., Ari, Z., Gumus, B., Yigitoglu, M. R. Jones, H. T. & Channer, K. S. Testosterone and
J. Clin. Endocrinol. Metab. 82, 682–685 (1997). & Arslan, T. Beneficial effects of testosterone atherosclerosis in aging men. Am. J. Cardiovasc.
77. Makinen, J. I. et al. Endogenous testosterone undecanoate on the lipoprotein profiles in Drugs 5, 141–154 (2005).
and serum lipids in middle-aged men. healthy elderly men. A placebo controlled study. 109. Chew, K. K. et al. Erectile dysfunction as a
Atherosclerosis 197, 688–693 (2008). Jpn Heart J. 38, 73–82 (1997). predictor for subsequent atherosclerotic
78. Barrett-Connor, E. & Khaw, K. T. Endogenous sex 94. Nettleship, J. E., Jones, T. H., Channer, K. S. & cardiovascular events: findings from a linked-
hormones and cardiovascular disease in men. A Jones, R. D. Physiological testosterone data study. J. Sex. Med. 7, 192–202 (2010).
prospective population-based study. Circulation replacement therapy attenuates fatty streak 110. Jackson, G. et al. Cardiovascular aspects of
78, 539–545 (1988). formation and improves high-density lipoprotein sexual medicine. J. Sex. Med. 7, 1608–1626
79. Haffner, S. M., Mykkanen, L., Valdez, R. A. & cholesterol in the Tfm mouse: an effect that is (2010).
Katz, M. S. Relationship of sex hormones to independent of the classic androgen receptor. 111. Araujo, A. B. et al. Erectile dysfunction and
lipids and lipoproteins in nondiabetic men. Circulation 116, 2427–2434 (2007). mortality. J. Sex. Med. 6, 2445–2454 (2009).
J. Clin. Endocrinol. Metab. 77, 1610–1615 95. Scott, R. et al. Effects of fenofibrate treatment 112. Shores, M. M. et al. Low testosterone is
(1993). on cardiovascular disease risk in 9,795 associated with decreased function and
80. Isidori, A. M. et al. Effects of testosterone on individuals with type 2 diabetes and various increased mortality risk: a preliminary study of
sexual function in men: results of a meta- components of the metabolic syndrome. The men in a geriatric rehabilitation unit. J. Am.
analysis. Clin. Endocrinol. 63, 381–394 (2005). Fenofibrate Intervention and Event Lowering in Geriatr. Soc. 52, 2077–2081 (2004).
81. Permpongkosol, S., Tantirangsee, N. Diabetes (FIELD) study. Diabetes Care 32, 113. Shores, M. M., Matsumoto, A. M., Sloan, K. L. &
& Ratana-olarn, K. Treatment of 161 men with 493–498 (2009). Kivlahan, D. R. Low serum testosterone and
symptomatic late onset hypogonadism with long- 96. Jeppesen, J., Hein, H. O., Suadicani, P. & mortality in male veterans. Arch. Intern. Med.
acting parenteral testosterone undecanoate: Gyntelberg, F. Triglyceride concentration and 166, 1660–1665 (2006).
effects on body composition, lipids, and ischemic heart disease: an eight-year follow-up 114. Khaw, K. T. et al. Endogenous testosterone and
psychosexual complaints. J. Sex. Med. 7, in the Copenhagen Male Study. Circulation 97, mortality due to all causes, cardiovascular
3765–3774 (2010). 1029–1036 (1998). disease, and cancer in men: European
82. Kalinchenko, S. Y. et al. Effects of testosterone 97. Hokanson, J. E. & Austin, M. A. Plasma prospective investigation into cancer in Norfolk
supplementation on markers of the metabolic triglyceride level is a risk factor for (EPIC-Norfolk) Prospective Population Study.
syndrome and inflammation in hypogonadal men cardiovascular disease independent of high- Circulation 116, 2694–2701 (2007).
with the metabolic syndrome: the double-blinded density lipoprotein cholesterol level: a 115. Malkin, C. J. et al. Low serum testosterone and
placebo-controlled Moscow study. Clin. metaanalysis of population-based prospective increased mortality in men with coronary heart
Endocrinol. 73, 602–612 (2010). studies. J. Cardiovasc. Risk 3, 213–219 (1996). disease. Heart 96, 1821–1825 (2010).
83. Van Pottelbergh, I., Braeckman, L., 98. Alexandersen, P., Haarbo, J., Byrjalsen, I., 116. Muraleedharan, V., Marsh, H. & Jones, H. in
De Bacquer, D., De Backer, G. & Kaufman, J. M. Lawaetz, H. & Christiansen, C. Natural Society of Endocrinology, BES P163
Differential contribution of testosterone and androgens inhibit male atherosclerosis: a study (Birmingham, UK, 2011).
estradiol in the determination of cholesterol and in castrated, cholesterol-fed rabbits. Circ. Res. 117. Hackett, G., Cole, N., Deshpande, A., Hall, A. &
lipoprotein profile in healthy middle-aged men. 84, 813–819 (1999). Wilkinson, P. The BLAST study: treating
Atherosclerosis 166, 95–102 (2003). 99. Zitzmann, M. et al. IPASS: a study on the hypogonadism in type 2 diabetes with long
84. Stanworth, R. D., Kapoor, D., Channer, K. S. & tolerability and effectiveness of injectable acting testosterone undecanoate versus
Jones, T. H. Dyslipidaemia is associated with testosterone undecanoate for the treatment of placebo significantly improves HbA1c, waist
testosterone, oestradiol and androgen receptor male hypogonadism in a worldwide sample of circumference, aging male symptom scores and
CAG repeat polymorphism in men with type 2 1,438 men. J. Sex. Med. 10, 579–588 (2013). all sexual function domains of the IIEF. Results
diabetes. Clin. Endocrinol. 74, 624–630 100. Heufelder, A. E., Saad, F., Bunck, M. C. & continue to improve for 12 to 18 months.
(2011). Gooren, L. Fifty‑two‑week treatment with diet and Presented at the European Society of Sexual
85. Isidori, A. M. et al. Effects of testosterone on exercise plus transdermal testosterone reverses Medicine meeting, Milan (2011).
body composition, bone metabolism and serum the metabolic syndrome and improves glycemic 118. Boyanov, M. A., Boneva, Z. & Christov, V. G.
lipid profile in middle-aged men: a meta-analysis. control in men with newly diagnosed type 2 Testosterone supplementation in men with
Clin. Endocrinol. 63, 280–293 (2005). diabetes and subnormal plasma testosterone. type 2 diabetes, visceral obesity and partial
86. Kirkland, R. T. et al. Decrease in plasma high- J. Androl. 30, 726–733 (2009). androgen deficiency. Aging Male 6, 1–7 (2003).
density lipoprotein cholesterol levels at puberty 101. Agledahl, I., Hansen, J. B. & Svartberg, J. Impact 119. Gopal, R. A. et al. Treatment of hypogonadism
in boys with delayed adolescence. Correlation of testosterone treatment on postprandial with testosterone in patients with type 2
with plasma testosterone levels. JAMA 257, triglyceride metabolism in elderly men with diabetes mellitus. Endocr. Prac. 16, 570–576
502–507 (1987). subnormal testosterone levels. Scand. J. Clin. (2010).
87. Zmuda, J. M. et al. The effect of testosterone Lab. Invest. 68, 641–648 (2008). 120. Bhattacharya, R. et al. Effect of 12 months of
aromatization on high-density lipoprotein 102. Rosengren, A., Wilhelmsen, L., Eriksson, E., testosterone replacement therapy on metabolic
cholesterol level and postheparin lipolytic Risberg, B. & Wedel, H. Lipoprotein (a) and syndrome components in hypogonadal men:
activity. Metabolism 42, 446–450 (1993). coronary heart disease: a prospective case- data from the Testim Registry in the US (TRiUS).
88. Kekki, M. Lipoprotein-lipase action determining control study in a general population sample of BMC Endoc. Disord. 11, 18 (2011).
plasma high density lipoprotein cholesterol level middle aged men. BMJ 301, 1248–1251 (1990). 121. Corona, G. et al. Testosterone and metabolic
in adult normolipaemics. Atherosclerosis 37, 103. Zmuda, J. M., Thompson, P. D., Dickenson, R. & syndrome: a meta-analysis study. J. Sex. Med. 8,
143–150 (1980). Bausserman, L. L. Testosterone decreases 272–283 (2011).

492  |  AUGUST 2013  |  VOLUME 9  www.nature.com/nrendo


© 2013 Macmillan Publishers Limited. All rights reserved
REVIEWS

122. Grossmann, M. Low testosterone in men with phosphorylation in target tissues of adult male 153. Malkin, C. J. et al. Testosterone replacement in
type 2 diabetes: significance and treatment. rat. Mol. Cell. Biochem. 352, 35–45 (2011). hypogonadal men with angina improves
J. Clin. Endocrinol. Metab. 96, 2341–2353 (2011). 139. Salehzadeh, F., Rune, A., Osler, M. & Al-Khalili, L. ischaemic threshold and quality of life. Heart 90,
123. Kaukua, J., Pekkarinen, T., Sane, T. & Testosterone or 17β-estradiol exposure reveals 871–876 (2004).
Mustajoki, P. Sex hormones and sexual function sex-specific effects on glucose and lipid 154. English, K. M., Jones, R. D., Jones, T. H.,
in obese men losing weight. Obes. Res. 11, metabolism in human myotubes. J. Endocrinol. Morice, A. H. & Channer, K. S. Testosterone acts
689–694 (2003). 210, 219–229 (2011). as a coronary vasodilator by a calcium
124. Hammoud, A. et al. Effect of Roux‑en‑Y gastric 140. Bergamini, E., Bombara, G. & Pellegrino, C. The antagonistic action. J. Endocrinol. Invest. 25,
bypass surgery on the sex steroids and quality effect of testosterone on glycogen metabolism 455–458 (2002).
of life in obese men. J. Clin. Endocrinol. Metab. in rat levator ani muscle. Biochim. Biophys. Acta 155. Webb, C. M., McNeill, J. G., Hayward, C. S.,
94, 1329–1332 (2009). 177, 220–234 (1969). de Zeigler, D. & Collins, P. Effects of testosterone
125. Reis, L. O. et al. Erectile dysfunction and 141. McLaren, D., Kelly, D., Akhtar, S., Channer, K. & on coronary vasomotor regulation in men with
hormonal imbalance in morbidly obese male is Jones, T. Low testosterone is associated with coronary heart disease. Circulation 100,
reversed after gastric bypass surgery: decreased expression of glut‑4 and hexokinase 2 1690–1696 (1999).
a prospective randomized controlled trial. Int. in muscle of the testicular feminised mouse. 156. Malkin, C. J. et al. Testosterone therapy in men
J. Androl. 33, 736–744 (2010). Endocrine Abstracts 29, P559 (2012). with moderate severity heart failure: a double-
126. Kelly, D. M. & Jones, T. H. Testosterone: 142. Ramamani, A., Aruldhas, M. M. & blind randomized placebo controlled trial. Eur.
a metabolic hormone in health and disease. Govindarajulu, P. Differential response of rat Heart J. 27, 57–64 (2006).
J. Endocrinol. 217, R25–R45 (2013). skeletal muscle glycogen metabolism to 157. Pugh, P. J., Jones, R. D., West, J. N., Jones, T. H.
127. Yu, Y. H. & Ginsberg, H. N. Adipocyte signaling testosterone and estradiol. Can. J. Physiol. & Channer, K. S. Testosterone treatment for men
and lipid homeostasis sequelae of insulin- Pharmacol. 77, 300–304 (1999). with chronic heart failure. Heart 90, 446–447
resistant adipose tissue. Circ. Res. 96, 143. Leonard, S. L. The effect of castration and (2004).
1042–1052 (2005). testosterone propionate injection on glycogen 158. Jain, P., Rademaker, A. W. & McVary, K. T.
128. Lim, S. et al. Fat in liver/muscle correlates more storage in skeletal muscle. Endocrinol. 51, Testosterone supplementation for erectile
strongly with insulin sensitivity in rats than 293–297 (1952). dysfunction: results of a meta-analysis. J. Urol.
abdominal fat. Obesity 17, 188–195 (2012). 144. Apostolakis, M., Matzelt, D. & Voigt, K. D. The 164, 371–375 (2000).
129. Lam, T. K. T., Van de Werve, G. & Giacca, A. Free effect of testosterone propionate on glycolytic 159. Cherrier, M. M. et al. Testosterone
fatty acids increase basal hepatic glucose and transamination enzyme activities in the supplementation improves spatial and verbal
production and induce hepatic insulin resistance liver, biceps muscle and levator ani muscle in memory in healthy older men. Neurology 57,
at different sites. Am. J. Physiol. Endocrinol. rats [German]. Biochem. Z. 337, 414 (1963). 80–88 (2001).
Metab. 284, E281–E290 (2003). 145. Max, S. R. Androgen-estrogen synergy in rat 160. Janowsky, J. S., Oviatt, S. K. & Orwoll, E. S.
130. Kelly, D. M. et al. Testosterone increases hepatic levator ani muscle: glucose‑6‑phosphate Testosterone influences spatial cognition in older
liver X receptor and ApoE expression and dehydrogenase. Mol. Cell. Endocrinol. 38, men. Behav. Neurosci. 108, 325–332 (1994).
improves lipid metabolism in the testicular 103–107 (1984). 161. Cherrier, M. M. et al. The role of aromatization in
feminized mouse: a potential protective 146. Mauvais-Jarvis, F. Estrogen and androgen testosterone supplementation: effects on
mechanism against atherosclerosis and fatty liver receptors: regulators of fuel homeostasis and cognition in older men. Neurology 64, 290–296
disease. Endoc. Rev. 33, OR22–5 (2012). emerging targets for diabetes and obesity. (2005).
131. Bryant, N. J., Govers, R. & James, D. E. Regulated Trends Endocrinol. Metabol. 22, 24–33 (2011). 162. Wang, C. et al. Testosterone replacement
transport of the glucose transporter GLUT4. Nat. 147. Traish, A. M., Abdallah, B. & Yu, G. Androgen therapy improves mood in hypogonadal men
Rev. Mol. Cell Biol. 3, 267–277 (2002). deficiency and mitochondrial dysfunction: —a clinical research center study. J. Clin.
132. Pessin, J. E. & Saltiel, A. R. Signaling pathways implications for fatigue, muscle dysfunction, Endocrinol. Metab. 81, 3578–3583 (1996).
in insulin action: molecular targets of insulin insulin resistance, diabetes, and cardiovascular 163. Wang, C. et al. ISA, ISSAM, EAU, EAA and ASA
resistance. J. Clin. Invest. 106, 165–170 (2000). disease. Horm. Mol. Biol. Clin. Investig. 8, recommendations: investigation, treatment and
133. Chen, X., Li, X., Huang, H., Li, X. & Lin, J. F. 431–444 (2011). monitoring of late-onset hypogonadism in males.
Effects of testosterone on insulin receptor 148. Ibebunjo, C., Eash, J. K., Li, C., Ma, Q. & Int. J. Impot. Res. 21, 1–8 (2008).
substrate‑1 and glucose transporter 4 Glass, D. J. Voluntary running, skeletal muscle 164. Bhasin, S. et al. Testosterone therapy in men with
expression in cells sensitive to insulin [Chinese]. gene expression, and signaling inversely androgen deficiency syndromes: an Endocrine
Zhonghua Yi Xue Za Zhi 86, 1474–1477 (2006). regulated by orchidectomy and testosterone Society clinical practice guideline. J. Clin.
134. Sato, K., Iemitsu, M., Aizawa, K. & Ajisaka, R. replacement. Am. J. Physiol. Endocrinol. Metab. Endocrinol. Metab. 95, 2536–2559 (2010).
Testosterone and DHEA activate the glucose 300, E327–E340 (2011). 165. Fernandez-Balsells, M. M. et al. Clinical review 1:
metabolism-related signaling pathway in skeletal 149. Guo, W. et al. testosterone plus low-intensity Adverse effects of testosterone therapy in adult
muscle. Am. J. Physiol. Endocrinol. Metab. 294, physical training in late life improves functional men: a systematic review and meta-analysis.
E961–E968 (2008). performance, skeletal muscle mitochondrial J. Clin. Endocrinol. Metab. 95, 2560–2575
135. Muthusamy, T., Murugesan, P. & biogenesis, and mitochondrial quality control in (2010).
Balasubramanian, K. Sex steroids deficiency male mice. PloS ONE 7, e51180 (2012). 166. Fenely, M. R. & Carruthers, M. Is testosterone
impairs glucose transporter 4 expression and its 150. van Breda, E. et al. Modulation of fatty‑acid‑ treatment good for the prostate? Study of safety
translocation through defective Akt binding protein content of rat heart and skeletal during long-term treatment. J. Sex. Med. 9,
phosphorylation in target tissues of adult male muscle by endurance training and testosterone 2138–2149 (2012).
rat. Metabolism 58, 1581–1592 (2009). treatment. Pflugers Arch. 421, 274–279 (1992). 167. Aversa, A. et al. Effects of testosterone
136. Sesti, G. et al. Androgens increase insulin 151. Wu, S. Z. & Weng, X. Z. Therapeutic effects of an undecanoate on cardiovascular risk factors and
receptor mRNA levels, insulin binding, and insulin androgenic preparation on myocardial ischemia atherosclerosis in middle-aged men with late-
responsiveness in HEp‑2 larynx carcinoma cells. and cardiac function in 62 elderly male coronary onset hypogonadism and metabolic syndrome:
Mol. Cell. Endocrinol. 86, 111–118 (1992). heart disease patients. Chin. Med. J. (Engl.) 106, results from a 24-month, randomized, double-
137. Parthasarathy, C., Renuka, V. N. & 415–418 (1993). blind, placebo-controlled study. J. Sex. Med. 7,
Balasubramanian, K. Sex steroids enhance 152. English, K. M., Steeds, R. P., Jones, T. H., 3495–3503 (2010).
insulin receptors and glucose oxidation in Chang Diver, M. J. & Channer, K. S. Low-dose
liver cells. Clin. Chim. Acta 399, 49–53 (2009). transdermal testosterone therapy improves Author contributions
138. Muthusamy, T., Murugesan, P., Srinivasan, C. & angina threshold in men with chronic stable All authors researched data for the article,
Balasubramanian, K. Sex steroids influence angina: A randomized, double-blind, placebo- contributed to discussion of the content and wrote
glucose oxidation through modulation of insulin controlled study. Circulation 102, 1906–1911 the article. P. M. Rao and T. H. Jones reviewed and
receptor expression and IRS‑1 serine (2000). edited the manuscript before submission.

NATURE REVIEWS | ENDOCRINOLOGY VOLUME 9  |  AUGUST 2013  |  493


© 2013 Macmillan Publishers Limited. All rights reserved

You might also like