Introduction

Type 2 diabetes is associated with testosterone deficiency (TD), as cross-sectional studies have shown approximately 25% to 40% of diabetic men have low testosterone [1-3] and longitudinal studies have demonstrated that men with low testosterone are at a greater risk of future development of type 2 diabetes [4-7]. While the mechanisms are not fully understood, current evidence suggests that the causative relationship between TD and diabetes might be bidirectional, or even multidirectional and interrelated with obesity, metabolic syndrome (MetS), sex hormone-binding globulins (SHBG), and other factors [8,9].

Prediabetes is a condition in which blood glucose level is higher than normal but does not reach the level of diabetes. Most diabetes organizations define prediabetes by impaired fasting glucose (IFG) and impaired glucose tolerance (IGT), while the American Diabetes Association (ADA) proposes that a glycated hemoglobin (HbA1c) of 5.7% to 6.4% should be also considered one of the criteria [10]. Prediabetes is characterized by increased insulin resistance and β-cell dysfunction, and is considered to precede the development of type 2 diabetes [11]. Data from non-diabetic men have revealed an inverse association between insulin resistance and testosterone concentrations [12-14]. This raised the question whether prediabetes, a state of increased insulin resistance, is also associated with low testosterone. However, few studies have investigated sex hormone levels in men with prediabetes, and the risk of TD in men with prediabetes has not been reported.

The present study aimed to investigate whether men with prediabetes were at an increased the risk of TD. Furthermore, as prediabetes is closely linked to obesity and MetS, both of which are associated with TD [8], we also explored whether the relationship between prediabetes and TD, if present, was independent of these factors.

Materials and Methods

Results

Among the 1,306 male participants, 577 (44.1%) were normoglycemic, 543 (41.5%) were prediabetic, and 186 (14.4%) were diabetic. The characteristics of the three groups are listed in Table 1. The mean ages of the three groups were 52.6 ± 8.8, 55.8 ± 8.5, and 59.7 ± 7.9 years (P<0.001); the mean BMIs were 24.4 ± 3.0, 25.2 ± 3.1, and 25.6 ± 3.3 kg/m² (P<0.001); the mean waist circumferences were 86.6 ± 7.4, 89.2 ± 7.5, and 90.6 ± 8.7 cm (P<0.001). Metabolic syndrome was present in 12.9%, 35.4%, and 60.5% of the patients (P<0.001), and the mean number of MetS components was 1.1 ± 1.1, 2.0 ± 1.3, and 2.8 ± 1.3 (P<0.001). The mean testosterone concentration and the prevalence of subnormal testosterone are listed in Table 2.

The age-adjusted odds ratio for subnormal total testosterone was 1.87 (95% confidence interval [CI]: 1.38-2.54) in prediabetic men and 2.38 (95% CI: 1.57-3.60) in diabetic patients (Model 1 in Table 3). The odds ratios in prediabetic men remained significant in multivariate analyses that further adjusted for BMI, waist circumference, the number of MetS components, and MetS (Models 2-5 in Table 3). After adjustment for MetS, the odds ratio for subnormal total testosterone in prediabetic men was 1.49 (95% CI: 1.08-2.06), almost equal to the odds ratio of 1.50 in diabetic patients (Model 5 in Table 3).

The presence of IFG, IPG, and HbA1c 5.7%-6.4% were all significantly associated with an increased risk of TD after adjustment for age (Model 1 in Table 3). In multivariate analyses, HbA1c 5.7%-6.4% remained significant in models that further adjusted for BMI, waist circumference, the number of MetS components, or MetS (Models 2-5 in Table 3). IFG was significant in most multivariate analyses (Models 2 to 4), but was no longer significant after adjustment for MetS (Model 5 in Table 3). However, IPG was not significant after any further adjustments (Models 2-5 in Table 3).

Figure 1 demonstrates the age-adjusted odds ratios for subnormal total testosterone in various conditions of hyperglycemia as classified by fasting glucose, postprandial glucose, and HbA1c. Compared to those with normoglycemia, men with isolated IFG, HbA1c 5.7%-6.4% (with or without elevations in fasting or postprandial glucose), newly detected diabetes, or previously diagnosed diabetes were all at an increased risk for TD.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Age-adjusted Odds Ratios for Total Testosterone <300 ng/dL in Various Conditions of Hyperglycemia.

https://doi.org/10.1371/journal.pone.0074173.g001

Table 4 shows the adjusted odds ratios for subnormal free testosterone in various glycemic groups. Neither prediabetes nor diabetes was associated with an increased risk of subnormal free testosterone after adjustment for age, BMI, waist circumference, the number of MetS, or MetS. IFG, IPG, or HbA1c 5.7%-6.4% was not associated with subnormal free testosterone in all multivariate analyses.

The association between testosterone and prediabetes or diabetes was also examined with multiple linear regression (Tables 5 and 6). Both prediabetes and diabetes were significant associated with total testosterone, independent of age, BMI, waist circumference, the number of MetS components, and MetS. The effect of prediabetes on total testosterone concentration was similar to that of diabetes (Model 5 in Table 5). On the contrary, the association between free testosterone and prediabetes was not significant in all multivariate analyses (Table 6).

Discussion

In the present study, we aimed to elucidate whether men with prediabetes had a greater risk for TD, while adjusting for obesity and MetS. The major findings are as follows: 1) prediabetes was associated with an increased risk of subnormal total testosterone, and the risk remained significant in multivariate analyses that adjusted for age, BMI, waist circumferences, the number of MetS components, and MetS; 2) after adjustment for MetS, the risk for subnormal total testosterone in men with prediabetes was equal to that in men with diabetes; 3) neither prediabetes nor diabetes was associated with subnormal free testosterone in multivariate analyses.

While there is a substantial body of evidence demonstrating the association between low testosterone and diabetes [1-7], few studies have reported the androgen status of men with prediabetes, which is considered the preceding stage of overt diabetes. In a study including 221 non-diabetic men, total testosterone was inversely associated with fasting glucose level and insulin resistance, and the association was independent of total body fat or abdominal fat [13]. Moreover, the Rancho Bernardo Study evaluated the sex hormone levels in 775 men and showed that men with IFG or IGT had lower total testosterone than those with normal glucose tolerance, after adjustment for age and BMI [22]. More recently, Corona et al. [23]evaluated the impact of IFG on sexual health in a consecutive 3,451 men attending the clinic for sexual dysfunction, and found that IFG increases the risk of severe erectile dysfunction, reduced penile flow, and overt hypogonadism [23]. Our findings in large part concur with the results of the previous studies, but there are some difference and several novel findings. First, the present study adopted a more updated definition of prediabetes, in which an HbA1c of 5.7%-6.4% was incorporated in the criteria of prediabetes and was proved as a significant risk factor for TD. To our knowledge, it is the first study reporting the association between HbA1c and testosterone in non-diabetic population. Second, we made a more comprehensive multivariate analysis, which adjusted for not only age but also several factors associated with obesity and MetS. The risk was reduced but remained significant after adjustment for BMI (a proxy of total body fat) or waist circumference (a proxy of visceral abdominal fat), suggesting a mechanism other than the consumption by fat is involved in the relationship between TD and prediabetes or diabetes. And we also found that men with prediabetes were at virtually the same risk of subnormal total testosterone as men with diabetes after adjustment for MetS. Third, while the Rancho Bernardo Study exclusively enrolled men aged over 55 years (mean age: 71.9 years), the present study included male participants aged 24 to 86 years (mean age: 54.9 years). The association between prediabetes and testosterone TD could therefore be applied to men over a wider age range. As the Endocrine Society recommends routine measurement of testosterone in all men with type 2 diabetes, the substantially increased risk of TD shown in the present study suggests the measurement should be performed earlier, as at the stage of prediabetes.

The definition of prediabetes differs among health organizations. It is generally composed of IFG and IGT, although different cutoff values to define IFG were adopted by the ADA and the World Health Organization (WHO) (100 mg/dL or 110 mg/dL, respectively) [10]. Greater controversy exists over whether HbA1c of 5.7% to 6.4% should be considered as prediabetes. Currently, an intermediate HbA1c range is considered prediabetes by the ADA [10], but not by the WHO. It has to be noted that the selection of glycemic measures and cutoff values to define prediabetes is based on the risk of future development of overt diabetes. However, elevations in fasting and postprandial glucose levels may be caused by different mechanisms and may represent distinct stages of diabetes development [11,24-27]. Prediabetes identified by IFG, IGT, or an intermediate HbA1c range may represent different population and is associated with different features of metabolic derangement [28,29]. Moreover, fasting and postprandial hyperglycemia have been shown to be associated with different levels of risk of disease, such as cardiovascular events and mortality [30,31]. Intrigued by these findings, we further examined whether the risk of TD might differ in men with IFG, IPG, or HbA1c 5.7%-6.4%. Our data show that HbA1c appeared to be a stronger predictor of subnormal total testosterone, and it remained significant across all multivariate analyses. Men with HbA1c 5.7%-6.4%, with or without fasting or postprandial hyperglycemia, were at a significantly increased risk of subnormal total testosterone. IFG was also a reliable predictor in most multivariate adjustments, and loss of significance was only observed after adjustment for MetS. An isolated IFG, even with HbA1c of <5.7%, was significantly associated with an increased risk of subnormal total testosterone. In contrast, compared with fasting glucose or HbA1c levels, postprandial glucose was only weakly associated with TD in the present study. Although postprandial glucose has been used in lieu of a standard OGTT to define prediabetes in the literature [19], the clinical application of postprandial glucose has been controversial due to low reproducibility and a lack of a standardized method of measurement [32-34]. Postprandial hyperglycemia has been considered as a surrogate of a much more complex series of metabolic events that occur in the postprandial period [33]. These facts support our finding that the relationship between postprandial hyperglycemia and testosterone is confounded by obesity or MetS, or can be explained by an overlap with fasting hyperglycemia or elevated HbA1c. Nonetheless, from a clinical perspective, men with elevated fasting glucose, postprandial glucose, or HbA1c, with or without other metabolic disorder, should be tested for TD.

Total testosterone is composed of free (2-3%), albumin-bound (20–40%), and SHBG-bound testosterone (60-80%). While the present study found that prediabetes was significantly associated with subnormal total testosterone, its association with free or bioavailable testosterone was not significant. This finding is supported by most but not all of the previous studies. In the Rancho Bernardo Study, the total but not the bioavailable testosterone concentration correlated with the insulin resistance and the risk of subsequent development of type 2 diabetes in middle-aged men after adjustment for age, BMI, or systolic blood pressure [5]. Similar results were also observed in two other cohort studies [7,35], in both of which total testosterone, but not free testosterone, was associated with the incidence of diabetes. There is also evidence that the relationship between free testosterone and diabetes is confounded by total body fat or abdominal fat [9,13]. Similarly, we also found that the relationship between prediabetes and free testosterone was confounded by general or central obesity. The finding that prediabetes or diabetes is significantly associated with total testosterone but not free testosterone may be explained by a mechanism associated with SHBG. SHBG has been traditionally considered a protein which binds and transports sex steroids and regulates circulating concentrations of free-form hormones. However, recent studies showed that low SHBG can predict the future development of type 2 diabetes, independent of the testosterone concentration [7,35]. And there is also evidence that mechanisms associated with the SHBG gene is involved in the pathogenesis of type 2 diabetes [36]. As SHBG tightly binds testosterone, which constitutes the largest portion of total testosterone, the low total testosterone in prediabetic or diabetic patients may be simply the consequence of low SHBG. On the other hand, the low SHBG also cause a shift of testosterone from SHBG-bound to free form, which relatively preserves the amount of free testosterone. While the exact mechanism may be complex and involve more factors, the SHBG level, at least in part, explains the discrepant levels of total testosterone and free testosterone in men with prediabetes or diabetes.

The identification of TD in men with prediabetes may have clinical implications. Testosterone replacement therapy (TRT) in hypogonadal men with type 2 diabetes and/or metabolic syndrome improves insulin sensitivity in the short-term [37-40]. It may be of interest to investigate whether hypogonadal men with prediabetes could also benefit from TRT, in terms of improving glucose control and, in a long-term perspective, preventing the future development of overt diabetes. Furthermore, as it has been shown that both prediabetes and low testosterone are independent risk factors for cardiovascular disease and event, and all-cause mortality [41,42], it would be interesting to evaluate whether TRT could decrease the risk of cardiovascular events and mortality in men with both TD and prediabetes. However, it should be noted that the current consensus on the guideline is that TRT be considered only in men with both TD and clinical symptoms [43]. In this way, defined by a combination of total testosterone of <300 ng/dL and Aging Males’ Symptoms (AMS) scale >=27, 14.7% of the prediabetic men in our dataset (unpublished data) could be considered candidates for TRT.

Conclusions

Men with prediabetes are at an increased risk of subnormal total testosterone, but not free testosterone. The risk is reduced, but remains significant after adjustment for BMI, waist circumference, the number of MetS components, or MetS. After adjustment for MetS, the risk for TD in men with prediabetes is almost equal to that of men with diabetes. The substantially increased risk suggests that testosterone should be measured routinely in men with prediabetes.