Given the age-related decline in T levels, it is unfortunate that age-specific population reference ranges for T are not widely available. T thresholds for diagnosis are based on the reference intervals of healthy and young men [29], or their association with specific symptoms [30]. Hence, the cut-points for partitioning hypogonadal from eugonadal older men (and indeed eligibility for reimbursement) remain controversial, relying mostly on opined recommendations [31] and consensus statements of various professional societies [32–34] that, not surprisingly, do not completely agree with each other (Table 17.1). For example, the International Society of Andrology (ISA), the International Society for the Study of Aging Male (ISSAM), the European Association of Urology (EAU), the European Academy of Andrology (EAA) and the American Society of Andrology (ASA) recommended the diagnosis of LOH when total T is below 8 nmol/L or within 8 and 12 nmol/L, if free T is below 225 pmol/L [32]. The EAU, however, suggests the threshold of 243 pmol/L for free T [34]. The Endocrine Society indicates a threshold between 9.8 and 10.4 nmol/L for total T, and for free T between 170 and 310 pmol/L while acknowledging that agreement on T thresholds is problematic and not truly evidence based [33]. There is agreement, however, on the requirement for two separate measurements of T for diagnosing LOH, based on the finding that subnormal T levels are not confirmed in almost 30% of repeat measurements [35]. Moreover, morning T levels are recommended (7.00–11.00 am) [32–34] because serum T exhibits a diurnal rhythm with peak levels in the morning [36].

A further important issue is the method of T measurement. The most commonly used for clinical practice are platform-based immunoassays, but their accuracy and precision for T immunoassays have been questioned, especially at the low levels in children, women and severely hypogonadal men [37–40]. Liquid chromatography–tandem mass spectrometry (LC-MS/MS) is now considered the gold standard for steroid measurement, but its costs and technical demands do not make it universally available for routine clinical practice. With improved quality control and methodological stringency, immunoassays are capable of providing T results in good agreement with LC-MS/MS in men [41]. Assessment of free T (non-protein-bound fraction of T) remains a controversial aspect of the biochemical evaluation of gonadal status, and its direct measurement is burdened with even more challenging analytical issues than for total T. The usefulness of evaluating free T arises from the “free hormone hypothesis” which suggests that only free unbound hormones are biologically active in target tissues. This hypothesis, though widely accepted in clinical practice, has not to date been corroborated by robust experimental evidence. In addition, there is also some evidence that T bound to SHBG can be internalized into cells by endocytosis through megalin protein [42], thus suggesting the possibility of a direct biological action also for protein-bound T. How to assess free T is also a matter of debate. The gold standard is measurement of free T by equilibrium dialysis, a technique that is expensive, time-consuming and not widely available. Alternative methods of measurement based on a labeled T analogue are unreliable and should not be used [43]. To circumvent these methodological issues, algorithms for calculating free T, based on the law of mass action and binding stoichiometry, have been introduced by Vermeulen [44], Södergård [45], Mazer [46] and Zakharov [47]. Other equations, published by Nanjee [48], Ly [49] and Sartorius [50], are empirical estimates obtained through statistical extrapolation to laboratory FT measurements by equilibrium dialysis. The equations based on the law of mass action are strongly dependent on the association constant (K _a ) for binding of T to SHBG and albumin. However, various K _a are used in different methods [44, 45, 51, 52], so that these equations carry intrinsic systematic errors based on their assumptions. Furthermore, their reliability is dependent on the accurate determination of total T and SHBG [53]. A study comparing the results of free T obtained by different equations based on the law of mass action (Vermeulen and Södergård) [44, 45] found that these equations systematically overestimate the value obtained by equilibrium dialysis [54]. The small overestimation of percentage free T may be attributed to errors in the measurement of T and SHBG with immunoassays, incorrect T-binding stoichiometry for SHBG, or the use of different K _a values [50]. Equations based on empirical assessment of free T may demonstrate better accuracy, but their applicability and transportability to different laboratories using different methods of hormone measurement is problematic. Despite the small systematic error associated with the mass action equations, a very high agreement between the calculated and the measured free hormone has been consistently found [44, 49, 55, 56]. Thus, despite the systematic, but nonetheless predictable, overestimation, the widespread application, accumulated credible clinical experience for more than 25 years, and convenience of the mass action calculations make it an acceptable method currently for evaluating free T in clinical practice until better alternatives are available. Accordingly, the U.S. Endocrine Society’s Expert Panel states that the use of equations based on the law of mass action provides the best approach for the estimation of free T concentration [57].

Based on these considerations, pending further studies demonstrating the clinical applicability of free T in the assessment of the androgenic status in men, total T measured by reliable methods, is still favored for the initial assessment and diagnosis of hypogonadism in men [57, 58]. However, recent data from EMAS showed that men with low total T levels do not have symptoms and signs of androgen deficiency if free T is normal, but when free T is low, even when total T is in the normal range, symptoms and signs of androgen deficiency are prevalent [59]. These results support the pivotal importance of measuring free T in conditions associated with alterations in SHBG levels, such as obesity or T2DM (low SHBG), or HepC or alcoholic liver cirrhosis, HIV infection or use of certain anticonvulsants (high SHBG) [32–34]. In these situations, the measurement of free T is essential for the accurate identification of LOH and to minimize its over-diagnosis [32–34].

Besides the uncertainty of T thresholds for hypogonadism in the elderly, the lack of precise knowledge on specific clinical features or consequences arising from pathologically low T levels in older men makes the diagnosis of LOH ever more challenging. Recommendations from professional societies require that the diagnosis of LOH requires low T levels together with androgen deficiency symptoms [32–34] (Table 17.1). However, symptoms of androgen deficiency are generally non-specific and overlap substantially with those of aging, depression and ill health. An attempt to create a working definition of LOH was based on the strength of the association between low T levels and the presence of symptoms [30]. After considering a large set of possible symptoms of hypogonadism, the presence of three sexual symptoms (i.e., erectile dysfunction, decreased sexual desire and impaired morning erections) was found to form a strong syndromic clustering with low T. Accordingly, LOH was defined as the simultaneous presence of erectile dysfunction, decreased sexual desire, and impaired morning erections in men with total and free T below 11 nmol/L and 220 pmol/L, respectively [30]. The specificity of sexual symptoms, but not physical and psychological ones, to low T, has been confirmed in the longitudinal extension of the EMAS, which showed that the fall of total T into the hypogonadal range during 4.3 years of follow-up is associated with the development of new sexual symptoms or the worsening of those already present at baseline [60]. Interestingly, testosterone replacement therapy (TRT), compared to placebo in 800 older hypogonadal men in a randomized clinical trial (RCT), has recently been shown to improve sexual symptoms, with variable to negative results on physical and psychological outcomes [61]. Another RCT, involving more than 700 adult men (>18 years of age) receiving T gel or placebo for 3 months, showed that TRT improved sexual symptoms, but not energy level or mood [62].

LOH, as defined by low T and three sexual symptoms, has an overall population prevalence of 2.1% [30]. The prevalence increases with age, BMI, and comorbidity (Fig. 17.2) [30]. This prevalence is much lower than that reported (5.6 and 12%) in other surveys of similar populations [25, 63], presumably due to the inclusion of less stringent or specific clinical features.

The relationship between T and obesity is complex and bidirectional. If a role of obesity in inducing hypogonadism is well recognized, some data also suggest that low androgen levels can lead to the development of obesity. Important information on the role of androgens in metabolism has been provided by the studies on androgen receptor (AR) knockout mice (ARKO). ARKO mice develop obesity with an increase in both subcutaneous and visceral fat, and they show a derangement in brown adipose tissue characteristics, which suggest a shift to the white adipose tissue profile [96]. Furthermore, ARKO mice display an increase in circulating triglycerides as well as in those stored within skeletal muscle and liver, and an impairment of glucose tolerance with hyperinsulinemia, hyperleptinemia, and hypoadiponectinemia [97]. In vitro studies demonstrated that androgens could inhibit the differentiation of pre-adipocytes in mice and in humans [98, 99]. Similar results have been provided by in vivo studies [100, 101]. In addition, androgen treatment of pluripotent mesenchymal cells increased myogenic but inhibited adipogenic differentiation [102]. Taken together, these data suggest a role for androgens in regulating both central and peripheral insulin sensitivity and body composition, which is best shown when androgen action is abolished rather than diminished.

A human model for the metabolic derangements arising from chemical castration is represented by men with metastatic prostatic cancer who are treated with androgen deprivation therapy (ADT). Several prospective studies demonstrated that ADT for prostate cancer is associated with significant weight gain and changes in body composition from the first months of therapy [103]. A meta-analysis of longitudinal studies shows that ADT is associated with a mean increase of 7.7% in fat mass and a decrease of 2.8% in lean mass from baseline, befitting the clinical feature of “sarcopenic obesity” [104]. These changes in body composition were accompanied by impairment in lipid profile and insulin sensitivity [103], and patients are at higher risk for developing metabolic syndrome and diabetes [105]. The role of T and androgen deficiency in modifying body composition was assessed in a study conducted in more than 400 healthy young men aged 20–50 years in whom severe hypogonadism was experimentally induced by a GnRH analogue [106]. Volunteers were then randomly assigned to receive for 16 weeks placebo or different doses of T gel with or without an aromatase inhibitor. Results showed that experimental androgen deprivation is associated with a reduction in lean mass and leg-press strength as well as an increase in total and subcutaneous fat mass, but no change in visceral fat [106]. Testosterone replacement with aromatase inhibition (pure T effect) maintained lean mass and muscle strength, but not fat mass [106]. These data suggest that the adverse effects of severe hypogonadism on lean mass are mediated by testosterone, and in adipose tissue, by estrogen deficiency.

It should be highlighted that hypogonadism induced by GnRH analogues, as in ADT in men with prostate cancer or experimental hypogonadism in eugonadal men [106], is associated with castrate T levels, which are much lower than those commonly found in LOH. A few longitudinal observational studies in the general population have evaluated the relationship between T and obesity with conflicting results. In the Rancho Bernardo study [107], a negative linear relationship between T or SHBG at baseline and waist-to-hip ratio at follow-up was found even for T levels within the eugonadal range (mean baseline total T of 18 nmol/L for those in the highest quartile of waist-to-hip ratio at follow-up). This observation is supported by a longitudinal study of 110 second-generation Japanese–American healthy male volunteers followed for 7.5 years [108]. In this study, the increase in intra-abdominal fat, as assessed by computed tomography, showed a linear negative relationship with total T at baseline, with a mean baseline total T of 14.8 nmol/L in the group with the highest increase in intra-abdominal fat [108]. More recently, longitudinal data from the EMAS showed that although low T at baseline is associated with incident metabolic syndrome, it was not associated with development of visceral obesity (determined by waist circumference) after 4.3 years [109]. Taken together, unlike the short-term changes associated with castration, chronic exposure over many years to moderately low T may predispose to abnormalities in central fat metabolism and body composition in men, but a causal relationship cannot be assumed from these observational data.

A number of uncontrolled registry studies appear to support the effect of TRT in improving body composition [110–115]. Several small RCTs of the metabolic outcomes of T therapy have been conducted. Meta-analysis of these results showed that T treatment was not associated with any change in BMI, body weight, or waist circumference [116]. However, a significant decrease in total fat mass and an increase in lean mass were consistently found in men treated with T, irrespectively of their gonadal status [116] (Table 17.3). The improvement in body composition by treating hypogonadal men with TRT is not necessarily accompanied by a substantial improvement in insulin sensitivity. In another meta-analysis of double-blind RCTs conducted in hypogonadal men with metabolic syndrome and/or diabetes mellitus [117], TRT showed only a modest, although significant, improvement in HOMA-IR as compared to placebo, whereas the effect of TRT on glycated hemoglobin was not different from placebo (Table 17.3). A recent RCT of 44 men with type 2 diabetes mellitus and low free testosterone with normal or low LH levels showed that i.m. T treatment for 24 weeks improved insulin sensitivity, as assessed by hyperinsulinemic-euglycemic clamp and fasting glycemia, without any change in glycated hemoglobin [118].