Hypogonadotrophic Hypogonadism in Type 2 Diabetes, Obesity and ...

816

Current Molecular Medicine 2008, 8, 816-828

Hypogonadotrophic Hypogonadism in Type 2 Diabetes, Obesity and the Metabolic Syndrome Paresh Dandona*, Sandeep Dhindsa, Ajay Chaudhuri, Vishal Bhatia, Shehzad Topiwala and Priya Mohanty Division of Endocrinology, Diabetes and Metabolism, State University of New York at Buffalo and Kaleida Health, 3 Gates Circle, Buffalo, NY 14209, USA Abstract: Recent work shows a high prevalence of low testosterone and inappropriately low LH and FSH concentrations in type 2 diabetes. This syndrome of hypogonadotrophic hypogonadism (HH) is associated with obesity, and other features of the metabolic syndrome (obesity and overweight, hypertension and hyperlipidemia) in patients with type 2 diabetes. However, the duration of diabetes or HbA1c were not related to HH. Furthermore, recent data show that HH is also observed frequently in patients with the metabolic syndrome without diabetes but is not associated with type 1 diabetes. Thus, HH appears be related to the two major conditions associated with insulin resistance: type 2 diabetes and the metabolic syndrome. CRP concentrations have been shown to be elevated in patients with HH and are inversely related to plasma testosterone concentrations. This inverse relationship between plasma free testosterone and CRP concentrations in patients with type 2 diabetes suggests that inflammation may play an important role in the pathogenesis of this syndrome. This is of interest since inflammatory mechanisms may have a cardinal role in the pathogenesis of insulin resistance. It is relevant that in the mouse, deletion of the insulin receptor in neurons leads to HH in addition to a state of systemic insulin resistance. It has also been shown that insulin facilitates the secretion of gonadotrophin releasing hormone (GnRH) from neuronal cell cultures. Thus, HH may be the result of insulin resistance at the level of the GnRH secreting neuron. Low testosterone concentrations in type 2 diabetic men have also been related to a significantly lower hematocrit and thus to an increased frequency of mild anemia. Low testosterone concentrations are also related to an increase in total and regional adiposity, and to lower bone density. This review discusses these issues and attempts to make the syndrome relevant as a clinical entity. Clinical trials are required to determine whether testosterone replacement alleviates symptoms related to sexual dysfunction, and features of the metabolic syndrome, insulin resistance and inflammation.

Keywords: Hypogonadism, diabetes, obesity, testosterone, insulin resistance, hypogonadotrophic, metabolic syndrome.

INTRODUCTION Large epidemiological studies have demonstrated the association of low testosterone (T) with type 2 diabetes (T2D) [1, 2]. In these studies T was treated as a biological marker which was found to be frequently low in males with T2D, often in association with dyslipidemia and abdominal adiposity [1]. It was also shown that insulin resistance and upper abdominal adiposity were features associated with low T [3,4]. These studies did not report sex hormone binding globulin (SHBG), free T (FT) or gonadotrophin concentrations and thus could not make a comment on the mechanism underlying low T concentrations nor did they place low T concentrations in a clinically relevant context. Some recent studies have also shown low T concentrations in obese patients and in those with metabolic syndrome (MetS) [5-9]. The above evidence and our own recent work demonstrating the frequent occurrence of hypogonadotrophic hypogonadism (HH) in men with T2D, the inverse relationship between BMI and low T [10], the mar*Address correspondence to this author at the DiabetesEndocrinology Center of WNY, 3 Gates Circle, Buffalo, NY 14209, USA; Tel: (716) 887-4523; Fax: (716) 887-4773; E-mail: [email protected] 1566-5240/08 $55.00+.00

ked increase of C-reactive protein (CRP) in HH and the associated fall in hematocrit [11], emphasize a potentially important role for a low T in T2D and possibly MetS. In addition, type 1 diabetes (T1D) is not commonly associated with HH [12]. Thus, clearly HH is a feature related to obesity, insulin resistance and T2D rather than hyperglycemia per se. This growing field in endocrinology needs an organized systematic investigation including prospective controlled studies on the treatment of HH in T2D and MetS to determine whether clinical features associated with hypogonadism can be successfully reversed with T replacement. A review of the currently available literature in this area is thus clearly required.

THE CROSS-SECTIONAL ASSOCIATION OF HYPOGONADISM WITH TYPE 2 DIABETES, OBESITY AND METABOLIC SYNDROME Phillips in 1977 was among the first to study the relationship of T with insulin and glucose metabolism. He found that glucose intolerance, elevated insulin and high triglycerides were related inversely with estradiol/testosterone ratio in men with myocardial infarction [13]. T concentrations in men have been found to be inversely related to insulin, independently of age and obesity [14]. Several investigators have since then stu© 2008 Bentham Science Publishers Ltd.

Hypogonadotrophic Hypogonadism in Type 2 Diabetes

died the association of low T with diabetes [1,2,10,1519]. Association with Type 2 Diabetes In a population based cohort study of 985 men (4079 yrs), 110 with T2D had lower mean T and SHBG concentrations [2]. These differences were reduced but not eliminated after adjustment for age and BMI. 21% of diabetic men versus 13% of controls had a low T concentration. T was inversely related to the degree of glycemia in the whole cohort; throughout the whole range of plasma glucose, there was a step wise decrease in mean T per categorical increase in FPG, apparent in both diabetics and non diabetics. Anderson et al. also found low T and SHBG concentrations in male T2D [15]. T concentrations were inversely related to fasting insulin concentrations. In a community study, comprising 1408 older subjects [775 men] with mean age of 73 years, Goodman et al. found that in sexspecific age and BMI adjusted analyses, men with impaired glucose tolerance had significantly lower total T than those with normal glucose tolerance and that T was inversely related to fasting plasma glucose [16]. A recent study in 1200 men with erectile dysfunction found low total T in 24.5% of T2D as compared to 12.6% in non-diabetic patients; differences in the prevalence of hypogonadism retained significance after adjustment for age and BMI [17]. All these studies measured total T and SHBG but not FT because reliable assays for FT have become available only recently. SHBG concentrations are decreased in obesity. T2D men have even lower SHBG levels as compared to age and BMI matched nondiabetics [2]. It was therefore not clear whether the lower SHBG levels in T2D can account for the difference observed in total testosterone levels between diabetics and nondiabetics. Total T concentrations are determined to a large extent by circulating SHBG concentrations. In the blood of normal men, 44% of total T is bound to SHBG, 2% is unbound (FT) and 54% circulates bound to albumin and other proteins [18]. Since albumin bound testosterone has 1000 times lower affinity than SHBG, it can freely disassociate in capillaries. Virtually all the nonSHBG bound T (also called bioavailable T) is therefore available for tissue uptake [19]. Circulating SHBG concentrations are also dependent upon a number of factors, the most important association being with obesity. SHBG concentrations decrease in obesity and increase with aging. A complete assessment of hypogonadism should therefore include measurement of FT. FT measured by radioimmunoassay is considered unreliable because it represents a variable fraction [20% to 60%] of the FT measured by equilibrium dialysis (ED) [2022]. ED is considered to be the gold standard for measuring FT. FT measured by this technique represents 1.5% to 4% of total T and is not dependent upon SHBG concentrations [23]. ED is a delicate, tedious and time consuming technique and therefore may not be suitable for population based or large sized studies. FT can

Current Molecular Medicine, 2008, Vol. 8, No. 8

817

also calculated from SHBG and T using the method of Vermeulen et al. [21]. This calculated FT has been shown to correlate very well with FT measured by ED [20] and is well suited for epidemiological studies. The first study to report the prevalence of hypogonadism in T2D based on appropriately measured FT by ED was published in 2004 [10]. This study included 103 males with T2D, aged between 31 and 75 years, referred to a tertiary diabetes center in Buffalo, NY. The prevalence of hypogonadism in T2D men was 33%. The plasma concentrations of LH and FSH were inappropriately low in men with subnormal FT. Indeed, LH and FSH were significantly lower than those in patients with normal FT. In addition, GnRH induced LH and FSH release was normal in all those who were tested. This pointed to a hypothalamo-hypophyseal lesion rather than a testicular defect. Magnetic resonance imaging (MRI) of brain and the pituitary was done in 10 randomly selected hypogonadal patients. No abnormality was detected on MRI of brain and the pituitary. There was an inverse relationship of TT and FT with BMI. Thus, the more obese the patient, the lower were TT and FT. HH was also reported in male T2D with erectile dysfunction [24]. Another study has also found high prevalence of hypogonadism in male T2D (50% on that basis of calculated FT concentrations) [25]. 42% of T2D patients had symptoms of hypogonadism along with low FT concentrations. BMI and waist circumference were negatively related to total and bioavailable th T. 3/4 of the men with hypogonadism had low or inappropriately low gonadotropin concentrations. Pituitary MRI did not reveal any abnormalities in men with low gonadotropins [25]. A recently concluded multicenter study (Hypogonadism in Males [HIM]), designed to determine the prevalence of hypogonadism in males over the age of 45 years (mean age: 62 years) in the United States, found the prevalence of hypogonadism to be 38.4% [26]. The study defined hypogonadism as total T40kg/m [27, 28]. Several investigators have studied body fat distribution and insulin resistance in relation to sex hormones. FT levels have been found to correlate inversely with

818


BMI in both obese and diabetic subjects [10, 29]. In one study of 87 Finnish males from Haffner et al., higher waist-to-hip ratio and lower T were strongly associated with a decrease in total and non oxidative whole body glucose disposal [4]. Low T was associated with high insulin concentrations in men [30]. In another study of obese men, the relationship of T with insulin/glucose was independent of BMI in obese men [31]. These findings are in striking contrast to women in whom an increased androgen concentration is associated with insulin resistance and hyperinsulinemia [3]. Women with T2D also exhibit an increase in T and low SHBG concentrations [15]. Similar to the negative impact of obesity on T concentrations in men, recent studies have highlighted the association of metabolic syndrome (MetS) with hypogonadism [29, 30]. Kaplan et al. studied 864 men (mean age 52 yrs) and demonstrated that aging men with obesity and the MetS have a significantly decreased total T (150 and 300 ng/dl less in obese and severely obese respectively) compared to aging, metabolically healthy men [8]. In a population based study of 1896 middle aged Finnish males (345 of whom had MetS), Laaksonen et al. showed that FT and SHBG were respectively 11% and 18% lower in MetS than in normal subjects [7]. Men with FT in the lowest third tertile were 1.7 times (after adjustment for age and BMI) more likely to have MetS. Low T and low SHBG concentrations were associated with MetS and its components, independently of BMI. Total T, FT and SHBG were inversely associated with concentrations of insulin, glucose, triglycerides and CRP, and positively associated with HDL. While there are no studies directly comparing the prevalence of hypogonadism in T2D and MetS, the published rates of hypogonadism in MetS [6-17%] [7] are generally less than those found in diabetic men [3352%].

Dandona et al.

were a risk factor for diabetes in men. In multivariable models adjusted for age, race/ethnicity, and adiposity, men in the lowest tertile of free T concentrations were four times more likely to have diabetes compared with men in the third tertile (odds ratio 4.12 [95% CI 1.25– 13.55]). These associations persisted even after excluding men with low total or free T concentrations (suggesting that the risk was not completely driven by hypogonadal men). Similarly, hypogonadism has also been linked with future development of MetS. Laakonsen et al. showed that T and SHBG predict MetS and T2D in middle aged men [6]. A protective role of T was thus postulated in the development of MetS, T2D and CVD in the aging male. Consistent with a possible bidirectional relationship of T with insulin sensitivity, an 11 year prospective study of 651 middle aged men showed that MetS predicted future development of hypogonadism [5]. The high prevalence of hypogonadism in T2D raises several questions: First and foremost, does this defect also occur in T1D? We compared the prevalence of hypogonadism in age matched (mean age 43 years) T1D and T2D men. 6% of T1D mean were hypogonadal as compared to 26% of T2D men [12]. SHBG in T1D was higher than that in T2D men. Thus, HH does not occur commonly in T1D, and so is not a function of diabetes or hyperglycemia per se. Interestingly, however, TT and cFT are inversely related to BMI even in T1D [12]. The next question that arises is whether HH is related to insulin resistance, especially since there is an inverse relationship of TT and FT with BMI in both T1D and T2D (Fig. 1: Correlation of calculated free testosterone with BMI in type 1 and type 2 diabetic subjects). In addition, previous studies have shown that hypogonadism is associated with upper abdominal adiposity, hyperinsulinemia and insulin resistance and features of the MetS one associates with hypogonadism in diabetes [1, 3-9, 15, 29, 30, 32, 35, 36]. Thus, there is prima facie evidence that HH may be related to

What Comes First? Hypogonadism or Obesity/MetS/T2D? Review of literature clearly shows that hypogonadism is associated with T2D, obesity and MetS in a substantial number of men. Obesity might be the most important contributor to the presence of hypogonadism in T2D and MetS, and the linear inverse relationship of T with BMI is preserved in the presence of T2D and MetS [8, 10]. The relationship of T2D and obesity with low T is probably bidirectional. Low T can predispose to obesity while obesity perpetuates hypogonadism. But what comes first? Several epidemiological studies have looked at the predictive value of hypogonadism on the development of T2D over a number of years in men. Studies from Barrett-Conner et al. have shown that low T in men and high T in women predict IR and T2D in older adults [32]. A case control study of 176 non diabetic men followed up for 5 years showed that low levels of SHBG and T at baseline were associated with the development of T2D in men [33]. A recent analysis from the NHANES III [34] showed that low androgens

Fig. (1). Correlation of calculated free testosterone (nmol/l) 2 with BMI (kg/m ) in age matched type 1 (; r = –0.36, P < 0.05) and type 2 (; r = –0.42, P < 0.05) diabetic subjects.



819

Fig. (2). Interrelationship between insulin resistance and hypogonadism. Adapted from Cohen [43], Kapoor et al. [44] and Pitteloud et al. [40, 41].

insulin resistance and perhaps hyperglycemia in addition to insulin resistance makes the defect more frequent and possibly more severe. With a prevalence of approximately one-third in T2D, HH is a potential candidate for being the commonest complication in males with of T2D (through association rather than causality at this time) and requires further assessment in terms of the etiology of the defect and the possible consequences, complications and treatment.

HYPOGONADISM AND INSULIN SENSITIVITY Relationship Between Hypogonadism and Insulin Sensitivity Insulin sensitivity is inversely related to intramuscular, intramyocellular and intraabdominal adipose tissue. Hypogonadism is also associated with increased subcutaneous (both truncal and appendicular), intraabdominal and intramuscular adipose tissue [37, 38]. Epidemiological studies clearly show an inverse relationship between T and insulin resistance, probably mediated by total or abdominal adiposity [39]. Insulin resistance, visceral obesity and MetS have all been associated with low SHBG and low total T levels in men [1, 3-9, 15, 29, 30, 32, 35, 36, 40, 41]. A study on patients with Klinefelter’s syndrome found increased prevalence of MetS and decreased insulin sensitivity; both correla-

ted with truncal obesity [42]. Serum LDL-C, triglycerides and CRP concentrations were also markedly increased. These abnormalities were reversed partially after replacement with T, although the doses of T given for replacement were small. The relationship between insulin resistance and hypogonadism may be bidirectional: insulin resistance may promote hypogonadotropism and hypogonadism may promote obesity and insulin resistance [43, 44] (Fig. 2: Interrelationship between insulin resistance and hypogonadism. Adapted from Cohen [43], Kapoor et al. [44] and Pitteloud et al. [40, 41]). Interestingly, the effect of testosterone on insulin sensitivity may also be mediated by mechanisms other than change in adiposity. Pitteloud et al. looked at the relationship between testosterone levels, insulin sensitivity, and mitochondrial function in 60 men and found that T correlated with VO2 max and oxidative phosphorylation gene expression independent of BMI [41]. A recent trial studied the effect of withdrawal of T therapy on insulin resistance in 12 men with idiopathic HH who were on chronic replacement with T [45]. Stopping T therapy for only 2 weeks worsened insulin sensitivity by 40% compared to baseline. Insulin sensitivity was measured by HOMA-IR in the study. This was also associated with an increase in serum IL-6 concentrations but a decrease in TNF-.

820


Studies in androgen receptor knock out (ARKO) mice show that male ARKO mice develop obesity while female ARKO mice do not [46]. The obesity in male ARKO mice is not due to hyperphagia but due to decreased spontaneous activity and decreased overall oxygen consumption [47]. Interestingly, these mice had preserved insulin sensitivity, possibly due to elevated adiponectin secretion. Serum estrogen concentrations were normal in ARKO mice. A mouse model of estrogen deficiency, aromatase deficient (ArKO) mouse, also demonstrates obesity and hepatic steatosis in males but not in females [48]. Effect of Testosterone Replacement on Insulin Sensitivity T replacement leads to a dose-dependent decrease in adipose tissue and an increase in muscle mass and strength [37, 49, 50]. This effect appears to be more pronounced if the population studied has higher amounts of adipose tissue. The mechanism by which T replacement produces these effects is not well understood. Inhibition of lipoprotein lipase activity in intraabdominal adipose tissue and the differentiation of pluripotent mesenchymal precursor cells preferentially into myogenic lineage instead of adipocytic lineage may play a role [51, 52]. Marin et al. demonstrated a decrease in visceral adipose tissue and an improvement in insulin sensitivity with oral and transdermal T treatment in middle aged obese non-diabetic men [53, 54]. The subjects were not hypogonadal but insulin sensitivity improved more in subjects whose T concentrations were in the low normal range at the start of the study. The improvement in insulin sensitivity and basal plasma T concentrations were inversely related. In one of these studies, oral T undecanoate 80mg twice a day (which increased serum T to 500-600 ng/dL for a few hours after the dosing) or placebo given for 8 months reduced visceral body fat (measured by CT scan) and total body fat by approximately 6%. Subcutaneous adipose tissue did not change. Glucose disposal rate (measured by hyperinsulinemic-euglycemic clamp) increased by 20%. Change in glucose disposal rate was related to baseline T levels (r= -0.65, P40kg/m2) while these parameters were normal in


milder obesity [27]. One study suggested that obesity can also blunt the responsiveness of pituitary gonadotropin secretion to GnRH stimulation [63]. Other studies showed a rise in LH or FSH levels after weight loss [64, 65]. It should also be noted that LH/hCG induced T secretion by Leydig cells is inversely correlated with insulin sensitivity (as measured by hyperinsulinemic euglycemic clamp) among men with varying degrees of glucose tolerance [40]. Insulin receptors are present on Leydig cells [66]. Thus, the lesion resulting in hypogonadism in obesity and T2D may occur at multiple levels of hypothalamic-pituitary-gonadal axis [48-50]. However, the absence of hypergonadotrophinism indicates that the primary defect in T2D is at the hypothalamohypophyseal level. Interestingly, the decline in T with aging appears to be associated with similar mechanisms. Mulligan et al. [67] in 1999 performed a randomized double blinded placebo controlled study in which a 2 week pulsatile GnRH infusion unmasked a dual (hypothalamic and Leydig cell) defect in the healthy aging male. This study did not include obese or diabetic patients. c) Role of Estradiol Elevated estradiol and leptin concentrations observed in obesity probably mediate a part of the hypogonadotrophic effect of obesity , but they do not account for the full effect of obesity on hypothalamohypophyseal axis [65, 68, 69]. Testosterone and androstenedione in the male can be converted to estrogen (estradiol and estrone respectively) via aromatization in extraglandular tissues, most significantly in adipose tissue. The rate of extraglandular aromatization increases with age and obesity. Even though aromatase activity increases with age, estrogen concentrations fall with age in males [70]. This is due to the fact that testosterone decreases with age and therefore is not available for conversion to estrogen. Both testosterone and estradiol can act independently on pituitary/hypothalamus to induce negative feedback. Estradiol levels are increased in mild to moderate obesity but do not rise further in massive obesity [27, 28]. However, free testosterone continues to fall as the obesity progresses. Treating obese men with an aromatase inhibitor leads to a reduction in serum estrone and estradiol, and an increase in serum testosterone and LH/FSH levels [71]. Weight loss leads to an increase in testosterone and FSH concentrations; however serum estrogen levels either decrease or do not change with weight loss [65, 68]. This discrepancy may be related to the final weight achieved. It appears that even mild to moderate obesity can raise serum estrogen levels and achievement of normal weight normalizes serum estrogen levels. On the other hand, massively obese males may not have a significant change in their serum estrogen levels after weight loss if they are still obese at the end of a weight loss study, even though they may have an improvement in their free testosterone levels.


821

It is not yet known if hypogonadal T2D males have higher or lower estrogen concentrations than eugonadal T2D males. d) Role of Leptin Obesity is associated with increased plasma levels of leptin. Isidori et al. studied the relationship of leptin with sex hormones(one fasting blood sample) at varying degrees of obesity, normal (BMI40, n=14) [69]. Moderate obese subjects had higher leptin than nonobese subjects but did not differ in terms of most sex hormones (FT was lower but was measured by RIA and therefore unreliable). Massive obese subjects had higher leptin and estradiol levels and lower TT, FT, and SHBG levels (as compared to normal subjects). LH/FSH concentrations were not different between the groups. In multiple regression analysis of T with estradiol, leptin, SHBG and LH, leptin was the best predictor of fasting T concentrations. When all the subjects were stimulated with a single dose of hCG, obese subjects had a lower peak T concentration than nonobese subjects. Another study also found a negative correlation between leptin concentrations and hCG stimulated T levels [40]. Leptin receptors are present on Leydig cells [72]. This implies that leptin may play a role in decreased Leydig cell responsiveness to gonadotropin stimulation in obesity. Leptin is known to play a permissive role in the regulation of reproductive axis. Absence of leptin results in HH [73]. Leptin appears to serve as a signal of energy reserves to regulate the hypothalamo-pituitarygonadal axis in relation to nutritional status [74]. It is not yet clear what role leptin resistance might play in the HH associated with insulin resistance. e) Role of Inflammation Both obesity and diabetes have been shown to be associated with oxidative stress [75-78] and inflammation [79-83]. The concentration of pro-inflammatory cytokines including TNF is elevated in T2D [81] and may contribute to HH in this condition. Obesity is also associated with an increase in pro-inflammatory cytokines like TNF [80, 84, 85]. This is consistent with the increase in hypogonadism in obesity and may also explain the effect of BMI on T concentrations in T2D and T1D. The understanding of mechanisms leading to oxidative stress and inflammation is important. One possible reason why obesity and T2D are associated with oxidative stress and inflammation is the state of insulin resistance. This is due to the fact that 1) Insulin resistance is associated with the presence of pro-inflammatory factors including cytokines [86]. These cytokines also interfere with insulin signal transduction [86] and this may also interfere with GnRH secretion since insulin is known to facilitate GnRH secretion both in vivo and in vitro [87], 2) insulin has been shown to exert an antiinflammatory and anti-oxidant effects at the cellular and molecular level both, in vitro, and in vivo [88-91]. It is

822


therefore relevant that insulin infusion induces an increase in plasma T [92]. Inflammatory mediators may impair insulin mediated mechanisms which facilitate GnRH, gonadotrophin and T secretion. This impairment is likely to be at the hypothalamic level in view of the defect described above. TNF and IL-6 are known to reduce insulin receptor tyrosine phosphorylation, and IRS tyrosine phosphorylation and the phosphorylation of AkT kinase (protein kinase B). These effects have been shown in adipocytes, human aortic endothelial cells and hepatocytes [93]. TNF has also been shown to reduce insulin receptor expression in human aortic endothelial cells [67]. It is probable that insulin signal transduction related defects would also be observed in neurons. More recently, it has been shown that SOCS-3, a protein which suppresses insulin signal transduction at IRS-1 level is induced by TNF and IL-6 [94, 95]. It is elevated in the obese and thus may have a role in the pathogenesis of a putative insulin resistant state at the level of GnRH neuron [96]. It has been suggested that hypogonadism is associated with an increase in inflammatory mediators and treatment of hypogonadism leads to a reduction in inflammation [97]. In vitro studies have demostrated that T induces an inhibition of IL-6 production by human monocytes [98]. T treatment of human aortic endothelial cells results in an inhibition of TNF- induced vascular cell adhesion molecule (VCAM)-1 mRNA expression and nuclear factor-B (a key inflammatory mediator) activation [99]. Intramuscular T replacement in hypogonadal males results in a decrease in proinflammatory cytokines (IL-1 and TNF-) and an increase in anti-inflammatory cytokine (IL-10) levels in serum [100]. However, some studies were unable to demonstrate an anti-inflammatory effect of T [55, 101]. It is possible that the effects of T replacement on inflammation depend on the population studied. A comprehensive study on inflammation and T replacement needs to be done to clarify the issue. Inflammation is known to be a cardinal causative factor in inducing atherosclerosis [102, 103]. Adherence of circulating monocytes and lymphocytes to the arterial endothelial lining is one of the earliest detectable events in human and experimental atherosclerosis. Whether adherence of monocytes is reduced by T replacement needs to be investigated in future studies.

TYPE 2 DIABETES, HYPOGONADISM AND ATHEROGENESIS Epidemiological studies have shown that hypogonadism is associated with increased incidence of atherosclerosis [104]. Patients with CAD have lower testosterone levels than healthy controls, and this is inversely related to the degree of CAD [105]. Hak et al. studied abdominal aortic calcifications (detected radiographically) in 504 non-smoking men in a population based Rotterdam study [106]. They found an inverse relationship between T and abdominal calcifications.

Dandona et al.

The age-adjusted relative risk of having abdominal calcifications was only 0.4 in men in the highest tertile of T and BT as compared to men in the lowest tertile. A population-based prospective study over 16 years [107] found that incident ischemic heart disease had a positive association of cortisol and a negative association with T. These relationships appeared to be mediated by measures of insulin resistance. Thus it may appear that there is an inverse relation of T with atherosclerosis, part of which may be explained by obesity. A recent population based prospective study in elderly men found that men in the lower quartile of testosterone (