3.1 Serum testosterone and type 2 diabetes risk
The low serum testosterone associated with obesity is an independent risk factor for incident T2D [
60]. The risk increases with severity of obesity and number of components of the metabolic syndrome [
61]. The degree to which this is dependent on obesity remains inconclusive; in one longitudinal cohort study of 195 older (mean age 76 years) men, non-obese by BMI, followed for 8 years a low serum testosterone concentration at baseline was not a predictor of incident T2D [
62]. Conversely in an earlier study, low baseline serum testosterone was predictive of developing incident metabolic syndrome over time, particularly in non-overweight, middle-aged men, with a BMI of < 25 kg/m
2 [
63]. Moreover, the predictive role of SHBG varies among observational studies with some [
63,
64] but not all [
60] studies reporting that low SHBG is predictive of the development of the metabolic syndrome and/ or incident T2D.
There is now accumulating evidence that explains the relationship between obesity, metabolic syndrome, testosterone, and risk of T2D in terms of high serum testosterone concentrations being protective. For example, in a metanalysis of 7 prospective studies (1966 to June 2005), the relationship between the serum testosterone concentration and T2D risk was linear up to a testosterone concentration of approximately 15.6 nmol/L (449.6ng/dL). Testosterone concentrations above this, up to 21 nmol/L (605.2 ng/dL), reduced the risk of T2D 42% [
65]. Subsequently published prospective cohort data have shown that risk of T2D increased below a total testosterone of 16 nmol/L (461.4 ng/dL) and decreased at concentrations above that [
66]. A more recent metanalysis of 10 prospective case control or nested cohort studies published between 1996 and 2016 showed a 38% reduction in T2D risk in men with higher serum testosterone [
67]. Direct evidence is provided by the recently published T4DM (testosterone for the prevention of type 2 diabetes) study (discussed in detail below) which shows that the effect of treatment with testosterone to prevent T2D is independent of the baseline serum testosterone concentration [
68].
3.2 Effect of weight loss on obesity associated functional hypogonadism
In men with low serum testosterone attributable to obesity there is a linearly inverse association between decrease in weight and increase in serum testosterone (and SHBG) concentration [
69].
Diet induced weight loss.
Modest diet induced weight loss (< 10%) prevents progression of prediabetes to T2D [
70] and increases serum testosterone concentrations[
71]. Greater magnitudes of weight loss induced by marked caloric restriction, for example using very low energy diets, may induce a remission of type 2 diabetes [
72]. In contrast to an earlier study [
73], a recently published meta-analysis has shown that the effects of weight loss induced by a hypocaloric diet, on testosterone and SHBG are not augmented by increased physical activity [
74]. The nutrient composition of the diet may affect serum testosterone. A recently published metanalysis showed a significant decrease in serum testosterone concentration in response to high protein low carbohydrate diets [
75]. However, there is marked heterogeneity among high protein diets with variation in the types of protein, carbohydrate, and fat, as well as the presence of fibre. In men who are overweight or obese, weight loss achieved with high protein, low carbohydrate diets and higher carbohydrate standard protein diets that have equivalent fibre and nutritional quality, improved testosterone, and SHBG to a similar extent [
76]. Overall, the increase in serum testosterone achievable with diet alone is relatively modest, but not insubstantial; in a meta-analysis, low-calorie diet leading to a 9.8% reduction in body weight was associated with a 2.87 mmol/L (82.7 ng/dL) increase in serum testosterone [
77]. However given that in most older men with obesity and/or T2D associated hypogonadism, serum testosterone concentrations are typically modestly reduced and fluctuate around the lower limit of the reference range [
78,
79], even modest weight loss, although this may be difficult to sustain, can be associated with reactivation of the HPT axis [
5] and in some men, biochemical normalisation of their androgen status.
Bariatric surgery.
Bariatric surgery produces substantial and durable weight loss with commensurate increases in serum testosterone SHBG, LH and FSH concentrations [
80], and potentially sustained remission of T2D [
81]. The effects of bariatric surgery on serum testosterone are substantial. In aforementioned metanalysis [
77] bariatric surgery leading to a 32% reduction in body weight was associated with a 8.73 nmol/L (251.6 ng/dl) increase in serum testosterone. Of note however whether increases in serum testosterone associated with weight loss are causal mediators of the improvement in androgen deficiency-like features observed in some obese men following weight loss remains unproven; in one study that longitudinally followed obese men after bariatric surgery, the degree of weight loss, but the not the increase in serum testosterone was associated with improvements in sexual function [
82].
Pharmacotherapy.
In men with T2D weight loss induced by the Glucagon-1-receptor agonists (GLP-1 RA) Exenatide was associated with an increase in serum testosterone only in men with lower serum testosterone concentrations at the outset [
83]. In a 16-week clinical trial, treatment with the long acting GLP-1 RA Liraglutide, decreased weight, improved markers of the metabolic syndrome and increased serum testosterone concentrations whereas treatment with testosterone was without benefit [
84]. Interestingly, an exploratory analysis of ‘The Researching Cardiovascular Events with a Weekly Incretin in Diabetes’ (REWIND) trial, a double-blind, placebo-controlled randomised trial of the effect of dulaglutide on cardiovascular outcomes, reported that, compared to placebo, dulaglutide may reduce the incidence of moderate or severe erectile dysfunction in men with T2D [
85]. It is unclear whether these sexual benefits were a direct effect of the GLP-1 RA. Serum testosterone concentrations were not reported in this study. Weight loss has been shown to improved erectile function [
86]. Of note, in contrast to testosterone treatment (
see Sect.
3.3below), GLP-1RA have proven cardiorenal benefits in men with T2D.
3.3 Is there a role for treatment with testosterone?
The T4DM study, a randomised double-blind placebo-controlled study, aimed to determine whether 2 years treatment with testosterone prevented T2D in men with obesity-associated functional hypogonadism. The study enrolled 1007 men aged between 50 and 74 years, with waist circumference over 94 cm, either impaired glucose tolerance or newly diagnosed T2D (20% of the total enrolled) established by an oral glucose tolerance test (OGTT) and a morning fasting serum testosterone of less than or equal to 14 nmol/L (403.5 ng/dL) as established by a platform chemiluminescent assay. All participants with serum testosterone < 8 nmol/L (230.5ng/dL) were assessed by an endocrinologist to exclude the presence of pathological hypogonadism, which if present was treated and excluded them from participation. Enrolled participants were randomised to 3 monthly intramuscular injections of 1000 mg of testosterone undecanoate or placebo on a one-to-one basis. All participants were enrolled in a WW (formerly Weight Watchers) program. At baseline and prior to each subsequent testosterone injection blood was drawn for the subsequent analysis of serum testosterone by LCMS. The primary endpoint was the proportion of participants with T2D at two years, assessed by oral glucose tolerance test (OGTT), representing either those who progressed from impaired glucose tolerance to T2D or those with newly diagnosed T2D who failed to regress [
87]. OGTT-based measures (2-h glucose ≥ 11.1 mmol/L (199.8 mg/dL) and the mean change in the 2-h glucose on OGTT at 2 years compared with baseline) were chosen as the two co-primary outcomes as OGTT has better sensitivity and specificity for diagnosing T2D compared with fasting glucose and HbA1c, and represented a gold standard test as used in the Diabetes Prevention Program (DPP) [
70]. It was also anticipated that testosterone would have an independent effect on HbA1c (
see below).
Although there was an overall correlation between testosterone measurements obtained on baseline samples by LCMS and the screening chemiluminescent assay (CLIA), there was considerable variation among individuals. The spread of values on LCMS was 4 to 30nmol/L (115–864 ng/dL), and approximately 43% of participants had a baseline testosterone over 14nmol/L (493ng/dL) as measured by LCMS (unpublished data).
In testosterone as compared to placebo treated men there was a 40% reduction in the prevalence of T2D at two years, and a significant decrease in the 2 h glucose on the OGTT, improvement in glucose tolerance, and lower fasting glucose concentration. Consistent with other studies of testosterone treatment [
88,
89], there was no effect on HbA1c. While the reasons for this finding are not fully understood, the utility of HbA1c as a marker of the glycaemic effects of testosterone treatment may be confounded by the erythropoietic actions of testosterone. While the effects of testosterone on red cell survival are not known, testosterone does increase red blood cell counts, and an increase in red cell survival could possibly explain the discrepancy between the OGTT and HbA1c findings in T4DM, although further studies are needed. In T4DM, the outcome was not dependent on either the screening or baseline testosterone concentrations. There were no differences in compliance with a lifestyle program between treatment groups and 70% of the men in each group achieved sufficient physical activity, which argues against an improvement in motivation with testosterone treatment. There was no significant difference in quality-of-life outcomes between the groups.
A notable outcome was that glucose tolerance at two years normalised in 43% of the men in the placebo group highlighting the efficacy of even a modest lifestyle intervention.
Testosterone treatment was also associated with beneficial changes in body composition, (decreased total and visceral fat mass, and increased skeletal muscle mass), muscle strength, and sexual function. In the placebo group fat mass, lean body mass and muscle strength all decreased [
68]. Testosterone also improved bone density as assessed by DEXA. In a sub-group of participants (n = 177) assessed by high resolution-peripheral quantitative computed tomography (HR-pQCT) the predominant effect was an increase in cortical area and thickness with only minimal effects on trabecular bone accompanied by a reduction in bone remodelling markers [
90].
There were no overall differences in serious adverse events between the groups. Cardiac and cerebrovascular safety were particularly reassuring. A similar number of placebo and testosterone treated men were diagnosed with prostate cancer over the 2 years of the study; there were fewer high-grade cancers in the testosterone treated men. There were 106 (21%) testosterone treated men who developed a haematocrit greater than 0.54%. However, only 25 of these men had treatment withdrawn prior to the final testosterone dose [
68]. It should be noted that obstructive sleep apnoea (OSA) did not preclude enrolment in T4DM [
87]. Approximately 25% of men over the age of 40 years have moderate to severe OSA and the proportion is higher with increasing age and obesity [
91]. Although speculative, it seems likely that the relatively high frequency of increased haematocrit in this population of testosterone treated men is attributable to underlying untreated OSA.
In contrast to the benefits of testosterone treatment observed in the T4DM study, a recent 12-week trial showed that over weeks, exercise but not testosterone had a benefit on vascular function and health [
92]. As discussed elsewhere in this issue, the long-term effects of testosterone treatment on cardiovascular health remain uncertain, due to the current lack of availability of dedicated cardiovascular outcome trials powered for clinical cardiovascular events.
3.4 Mechanism of effect of testosterone on glucose metabolism
Glucose tolerance depends on the interplay between cellular mechanisms of glucose uptake and utilisation and insulin production by pancreatic beta cells.
There is some evidence to show that testosterone increases glucose stimulated insulin secretion from human cadaveric islets [
93], a process dependent on the presence of 5 alpha reductase and possibly aromatase in pancreatic beta cells [
94]. However, there are no data to show that acute or short-term administration of testosterone to hypogonadal men increases glucose dependent insulin production.
The potential effects of testosterone on insulin signalling and glucose utilisation have recently been reviewed [
95]. In adipose tissue testosterone increases insulin receptor β subunit, insulin receptor substrate-1, protein kinase B and glucose transporter type 4 expression. In skeletal muscle testosterone increases adenosine 50-monophosphate-activated protein kinase expression and activity. Testosterone treatment also decreases markers of inflammation linked and free fatty acids linked to inhibition of insulin signalling [
95].
In men with functional hypogonadism and type 2 diabetes, topical testosterone treatment for 24 weeks, but not 3 weeks, improved insulin sensitivity an effect associated with a decrease in adipose tissue mass. Free fatty acids decreased from 15 weeks, but any effect on insulin sensitivity could not be disentangled from the change in body composition [
96]. Treatment of men with prediabetes [
97] or type 2 diabetes [
98] and functional hypogonadism with 3 monthly intramuscular testosterone undecanoate injections shows a cumulative time dependent effect to decrease waist circumference and other anthropometric obesity indices paralleled by progressive improvements in glucose tolerance.
Taken together these data suggest that the mechanism by which testosterone treatment improves glucose metabolism in men is dependent on favourable changes in body composition. In accordance with this, a preliminary mediation analysis of data from the T4DM study suggests that the beneficial effects of testosterone on glucose metabolism in was predominantly (but not entirely) mediated by decreased fat mass [
68]. Ongoing analyses from this study may provide further clarification as to the mechanisms by which testosterone produces beneficial effects on glucose metabolism.