Key words
diabetes mellitus - testosterone - oral glucose tolerance test - hypogonadism
Introduction
A low testosterone level is a common finding among men with Type 2 Diabetes Mellitus
(T2DM). Several cross-sectional studies have shown that its prevalence ranges from
25 to 40% [1]
[2]
[3], reaching 50% if obesity is associated [4]. The very high prevalence of this association lead the Endocrine Society to recommend
the measurement of morning serum total testosterone in patients affected by T2DM who
have symptoms of sexual dysfunction, unexplained weight loss, weakness or mobility
limitation [5]. It is not known whether low testosterone levels play a causative role or whether
they are a consequence of T2DM and/or of its associated clinical features [6]. However, some evidences suggest that testosterone is an important modulator of
insulin resistance [7]. In fact, as already reported earlier [8]
[9], the Third National Health and Nutrition Examination Survey (NHANES III) disclosed
that, in the general population, plasma testosterone concentration is inversely related
to HOMA-IR and plasma insulin concentration, 2 proxy measures of insulin resistance
[10]. This association has also been confirmed in men with established T2DM [11], and with prediabetic conditions [12]
[13], such as impaired fasting glucose (IFG) and impaired glucose tolerance (IGT), all
characterized by increased insulin resistance [14]
[15].
Furthermore, the role of testosterone in glucose metabolism is supported by evidences
from prospective studies, which have convincingly shown that hypogonadism is a risk
factor for future development of insulin resistance and T2DM. [16]
[17]
[18]; according to a meta-analysis, men with higher testosterone levels had a 42% lower
risk of T2DM (RR, 0.58; 95% CI, 0.39–0.87) [19]. Finally, men affected by T2DM and hypogonadotropic hypogonadism, who underwent
testosterone replacement treatment, also show an increase in insulin sensitivity [20]. Taken together, these findings suggest that hypogonadism directly affects insulin
sensitivity, rather than being merely an associated condition [4]
[7].
In the past few years, low testosterone has also been related to other elements of
the Metabolic Syndrome, such as arterial hypertension [21]. In this study, we aimed to investigate whether an independent association between
the sex hormone profile and insulin resistance exists in men affected by arterial
hypertension, across all the different classes of glucose metabolism.
Methods
We collected data from male patients aged 35–70 years referred to a hypertension clinic
for essential arterial hypertension. The following exclusion criteria were applied:
a) a defined diagnosis of T2DM; b) concomitant endocrine, liver and/or renal diseases;
c) treatment with ß-blockers and/or α1-antagonists and/or thiazides within 6 months before recruitment; d) previous urologic
surgical interventions. On this basis, data from 87 patients were collected. All patients
gave informed consent to their participation to the study, which was conducted in
strict accordance to the Principles of the Declaration of Helsinki. Being a study
of clinical practice, no ethical committee approval was requested.
Anthropometric data were collected with patients wearing only light underwear, and
included the body mass index (BMI), i. e., body weight divided by the square of the
height (kg/m2) and the waist circumference (WC), measured midway between the lowest rib and the
iliac crest when standing. Data on cigarette-smoking were also obtained.
In agreement with the American Diabetes Association 2003 criteria [22], each patient underwent a standard Oral Glucose Tolerance Test (OGTT). This entails
the ingestion of 75 g of glucose and measurement of 2 glucose concentrations: Fasting
Plasma Glucose (FPG) and 2-h plasma glucose (2hPG) after the glucose challenge, measured
along with the corresponding fasting and 2-h plasma insulin concentrations (FPI and
2hPI). Glucose plasma concentration was measured by hexokinase method (ADVIA, Siemens;
detection limit 4 mg/dl), while insulin was measured by chemiluminescence (Centaur,
Siemens; detection limit 0.5 μUI/ml)
According to the ADA 2003 criteria and the results of the OGTT, a patient was classified
as having normal glucose tolerance (NGT) when FPG<100 mg/dl and 2hPG<140 mg/dl, T2DM
when FPG>125 and/or 2hPG≥200 mg/dl, IFG when FPG≥100<126 mg/dl and 2hPG<140 mg/dl,
IGT when FPG<100 mg/dl and 2hPG≥140<200 mg/dl, and a combination of IFG and IGT (IFG/IGT)
when FPG≥100<125 mg/dl and 2hPG≥140<200 mg/dl. For the purposes of the study, hypertensive
subjects with IFG, IGT, IFG/IGT were lumped into a single group called prediabetes
(preDM).
For each patient we also calculated the HOMA-IR as follows: (FPI µU/ml×FPG mmol/L)/22.5
[23]. Contextually to OGTT, patients underwent a further morning plasma sampling; follicle-stimulating
hormone (FSH), luteinizing hormone (LH), testosterone, sex hormone-binding-globulin
(SHBG) were measured by chemiluminiscent immunoassays, and free-testosterone by a
radioimmunoassay (Advia Centaur XP, Siemens Healthcare Diagnostics, Ireland/USA).
Based on the testosterone threshold for hypogonadism [24], patients with a plasma concentration below 350 ng/dl were compared to those whose
concentration was above this limit. Comparisons between percentages were performed
by χ2 test. Continuous variables were shown as medians with upper and lower quartile (InterQuartile
Range, IQR) and compared as appropriate by the Mann-Whitney and Kruskal-Wallis tests,
followed by post-hoc analysis. The correlation between continuous variables was analyzed
calculating the Spearman Rho correlation coefficient, while multiple linear regression
analysis was used to predict the variability of testosterone and free-testosterone.
Statistical significance was set at a p-value<0.05 for all statistical tests. The
statistical software package StatSoft 5.1 STATISTICA Inc. (2300 East 14th Street, TULSA, OK 74104, USA) was used for statistical analysis.
Results
We recruited 87 male subjects with a median age of 59 (50–65) years; median BMI was
28.6 (26.0–31.9) kg/m2 and BMI was≥30 kg/m2 in 35.6% of patients (n=31). Median WC was 99 (93–105) cm, while 37.95% (n=33) of
subjects had a WC≥102 cm. 57.5% (n=50) were active smokers. According to OGTT results,
40 subjects (45.9%) were classified as NGT, while preDM was diagnosed in 35 (40.3%;
25 IFG, 3 IGT and 7 IFG/IGT) and T2DM in 12 (13.8%) patients.
[Table 1] shows the medians (interquartile range, IQR) of FSH, LH, testosterone, free-testosterone
and SHBG in NGT, preDM and T2DM groups. Testosterone and SHBG decreased significantly
from NGT to T2DM group (p<0.01 and p<0.01, respectively). In post-hoc analysis, testosterone
was significantly lower in the group of patients with T2DM when compared to NGT group
(p<0.01), while SHBG was significantly lower in preDM than in the NGT group (p<0.04).
Furthermore, in T2DM patients free-testosterone was significantly lower in comparison
to preDM (p<0.05).
Table 1 Hormonal profile according to glucose tolerance stage.
|
NGT (n=40)
|
PreDM (n=35)
|
T2DM (n=12)
|
Kruskal-Wallis
|
FSH (µU/ml)
|
5.1 (4.4–8.7)
|
5.1 (3.9–7.9)
|
7.0 (5.9–10.3)
|
n.s.
|
LH (µU/ml)
|
4.5 (3.5–6.8)
|
4.3 (2.9–6.5)
|
4.6 (3.3–9.9
|
n.s.
|
TST (ng/dl)
|
431.5 (330.4–514.2)†
|
387.0 (344.0–452.0
|
311.9 (207.8–410.0)
|
p<0.01
|
FTST (pg/ml)
|
10.7 (8.7–13.7)
|
11.6 (8.7–12.9)#
|
8.5 (6.2–11.1)
|
n.s.
|
SHBG (nmol/L)
|
46.5 (40.0–51.6)°
|
38.9 (33.1–45.7)
|
30.4 (25.0–61.9)
|
P<0.01
|
In [Table 2] we compared anthropometric measures and glucose metabolism data between 30 subjects
with hypogonadism (testosterone below 350 ng/dl) and 57 subjects with normal plasma
testosterone concentration. The former group had median values of BMI (p<0.05), FPG
(p<0.05), FPI (p<0.0001), 2hPI (p<0.01) and HOMA-IR (p<0.0001) significantly higher
than the latter group. The percentage of subjects with altered OGTT, although higher
in the group of subjects with hypogonadism, was not significantly different between
the two groups (63.3% vs. 49.1, χ2=1.6, n.s.).
Table 2 Antropometric data and glucose metabolism parameters according to Testosterone levels.
|
TST<350 ng/dl (N=30) Median (IQR)
|
TST≥350 ng/dl (N=57) Median (IQR)
|
Mann-Whitney test
|
Age (years)
|
57.5 (51–65)
|
59 (50–65)
|
n.s.
|
BMI (Kg/m
2
)
|
29.4 (27.8–33.4)
|
28 (25.2–30.2)
|
p<0.05
|
WC (cm)
|
100 (94–108)
|
99 (91–103)
|
n.s.
|
FPG (mg/dl)
|
106 (94–124)
|
98 (93–107)
|
p<0.05
|
2hPG (mg/dl)
|
116 (98–161)
|
104 (94–139)
|
n.s.
|
FPI (μU/ml)
|
16.4 (12.3–23.2)
|
10.8 (7–15)
|
p<0.0001
|
2hPI (μU/ml)
|
89.1 (48–165.4)
|
48 (31.1–69.3)
|
p<0.01
|
HOMA-IR
|
4.38 (3.4–6.4)
|
2.65 (1.72–3.8)
|
p<0.0001
|
As detailed in [Table 3], in the study population, total testosterone plasma concentration, free-testosterone
and SHBG are inversely and extensively related to glucose metabolism.
Table 3 Spearman’s correlations between sex hormones and glucose metabolism.
|
Total testosterone
|
Free testosterone
|
SHPG
|
BMI (Kg/m
2
)
|
r=− 0.25; p<0.01
|
r=− 0.23; p<0.05
|
r=− 0.08; n.s.
|
WC (cm)
|
r=− 0.27; p<0.01
|
r=− 0.28; p<0.01
|
r=− 0.15; n.s.
|
FPG (mg/dl)
|
r=− 0.40; p<0.0001
|
r=− 0.26; p<0.05
|
r=− 0.35; p<0.001
|
2hPG (mg/dl)
|
r=− 0.29; p<0.01
|
r=− 0.21; n.s.
|
r=− 0.29; p<0.01
|
FPI (μU/ml)
|
r=− 0.42; p<0.0001
|
r=− 0.28; p<0.01
|
r=− 0.20; n.s.
|
2hPI (μU/ml)
|
r=− 0.42; p<0.0001
|
r=− 0.27; p<0.01
|
r=− 0.15; n.s.
|
HOMA-IR
|
r=− 0.46; p<0.0001
|
r=− 0.30; p<0.01
|
r=− 0.27, p<0.05
|
The inverse correlations of both testosterone and free-testosterone with HOMA-IR (r=− 0.47,
p<0.01 and r=− 0.34, p<0.05, respectively) was confirmed also in the subgroup (N.
40) of NGT subjects.
[Table 4] shows the 2 multivariate analysis models we constructed considering, as regressors,
age, BMI, WC and HOMA-IR and, as dependent variable, either testosterone or free-testosterone.
In both the analyses, HOMA-IR was the only variable independently and inversely related
to testosterone (p<0.001) and to free-testosterone (p<0.05).
Table 4 Multiple regression models: factors associated to testosterone and free-testosterone
plasma concentration (after log-transformation).
|
Testosterone
|
Free-testosterone
|
Age (years)
|
0.39, p=n.s.
|
−0.94, p=n.s.
|
BMI (Kg/m
2
)
|
−1.34, p=n.s.
|
−0.93, p=n.s.
|
HOMA-IR
|
−3.97, p<0.0001
|
−2.32, p<0.05
|
Discussion
Our data indicate that, among male patients with hypertension, the sex hormone profile
is related to glucose metabolism; in fact the gonadal axis is significantly impaired
in patients affected by T2DM. Furthermore, testosterone and free testosterone plasma
concentrations are inversely related to insulin resistance, independently from other
determinants of glucose metabolism, being this association confirmed in all OGTT classes.
Our findings strengthen the hypothesis that testosterone does not act exclusively
on sexually-related chemical and attitudinal events, but that it also plays a key
role in carbohydrate metabolism [6].
Many observational studies have already documented a high prevalence of hypogonadism
in men with T2DM [1]
[2]
[3]
[4] or prediabetes [13]. In our study, we decided to investigate selectively a subgroup at high risk for
both hypogonadism [25] and insulin resistance [26], hypertensive men. We confirm a strong association between sex hormones and glucose
metabolism, also in this subset of subjects: in fact, we observed a progressive reduction
of testosterone, free-testosterone and SHBG plasma concentration moving from NGT to
T2DM. Furthermore all these 3 hormones were inversely related to HOMA-IR. To date,
the mechanism underlying the association between low testosterone levels and insulin
resistance has not been completely elucidated [6]; it has been suggested that some confounding factors can mediate this relationship,
such as elements of the metabolic syndrome, particularly, obesity [27]
[28]. Anyway, our data are in line with previous reports, which agreed in detecting an
independent association between insulin resistance and testosterone/free-testosterone
plasma concentration [13]
[18]
[29], which is confirmed after correcting for BMI, age and WC.
As expected according to previous findings, low testosterone levels were not associated
to increased gonadotropins plasma concentrations, entailing the condition of hypogonadotropic
hypogonadism [5]
[17]. It has been reported that insulin promotes gonadotropin release from pituitary
cell cultures [30], and that insulin signaling in the brain plays an important role in regulating the
reproductive function [31]. Moreover, insulin receptors are expressed by Leydig cells and insulin enhances
testosterone secretion both in vivo and in vitro [32]
[33]. Therefore, a low testosterone level in insulin resistant states could indicate
a functional defect at one or more levels of the hypothalamic-pituitary-gonadal axis.
In addition to the impact of insulin resistance on testosterone secretion, there is
evidence to support an effect of testosterone itself on insulin sensitivity [6]
[33]. For example, in male rats, acute castration induces significant insulin resistance
by reducing lipolysis, which is reversed by testosterone replacement [34]. In men, low testosterone levels predispose to visceral obesity [35], leading to dysregulation of fatty acid metabolism, which, in turn, promotes insulin
resistance [6]
[7]. Furthermore, consistent data from prospective studies support the hypothesis that
hypogonadism is an independent risk factor for the development of diabetes [16]
[17]
[18]
[19]. Finally, although still controversial [5]
[6]
[36], early interventional studies suggest that testosterone treatment in hypogonadal
men with T2DM exerts beneficial effects on insulin resistance and glucose control
[37]
[38]
[39]. Taken together, these findings suggest that glucose metabolism and the hypothalamus-pituitary-gonadal
axis influence one each other in a complex relationship.
Importantly, we have documented that the inverse association between testosterone
and insulin resistance is evident even in case of normal glucose tolerance; in other
words, it extends across the full continuum of glucose tolerance. This finding suggests
to consider testing the gonadal axis, not only in T2DM, but also in prediabetes and
in insulin resistant NGT subjects. Further ad hoc studies are required to investigate
whether hypogonadism correction in NGT and prediabetic subjects could prevent progression
towards T2DM.
Our study has some limitations: first of all, we decided to focus our attention on
patients referred to our clinic because of a diagnosis of arterial hypertension. Therefore,
our findings can not be extended to the general population. Furthermore the sample
size is limited and larger cohorts should be assessed to better identify and characterize
the association between sexual hormonal profile and glucose tolerance, as well as
to implement useful management algorithms.
In conclusion, a bidirectional association appears to exist between low testosterone
and insulin resistance. In males with hypertension, the link between insulin sensitivity
and hypothalamic-pituitary-gonadal axis is maintained along the entire spectrum of
glucose tolerance.