Effects of testosterone supplementation on markers of the metabolic syndrome and inflammation in hypogonadal men with the metabolic syndrome: the double-blinded placebo-controlled Moscow study

Authors

Errata

This article is corrected by:

  1. Errata: Effects of testosterone supplementation on markers of the metabolic syndrome and inflammation in hypogonadal men with the metabolic syndrome: the double-blinded placebo-controlled Moscow study Volume 75, Issue 2, 275, Article first published online: 6 July 2011

  • ClinicalTrials.gov identifier: NCT00696748

Dr Yuliya A. Tishova, Associate Professor, Chair of Endocrinology, The Faculty of Medical Staff Refresher Training, People’s Friendship University of Russia, Pokrovka, 22/1, Building 1, Moscow 125080, Russia. Tel.: +7 903 221 32 76. E-mail: yulya_tishova@mail.ru

Summary

Objective  Men with the metabolic syndrome (MetS) have low plasma testosterone (T) levels. The aim of this study was to establish whether the normalization of plasma T improves the features of the MetS.

Design  A randomized, placebo-controlled, double-blinded, phase III trial of 184 men suffering from both the MetS and hypogonadism.

Patients  One hundred and eighty-four men, aged 35–70, with the MetS and hypogonadism (baseline total T level <12·0 nm or calculated free T level <225 pm.), recruited in the outpatient andrology and urology clinic, Research Center for Endocrinology in Moscow, Russia.

Intervention  Treatment for 30 weeks with either parenteral T undecanoate (n = 113; TU; 1000 mg IM) or placebo (n = 71), administered at baseline, and after 6 and 18 weeks. One hundred and five (92·9%) men receiving TU and 65 (91·5%) receiving placebo completed the trial.

Measurements  Body weight, body mass index (BMI), waist circumference (WC), hip circumference, waist-to-hip ratio, insulin, leptin, glucose, cholesterol, triglycerides, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, C-reactive protein (CRP), interleukin-1-beta (IL-1β), interleukin-6 (IL-6), interleukin-10 (IL-10) and tumour necrosis factor-alpha (TNF-α).

Results  There were significant decreases in weight, BMI and WC in the TU vs placebo group. Levels of leptin and insulin also decreased, but there were no changes in serum glucose or lipid profile. Of the inflammatory markers, IL-1β, TNF-α and CRP decreased, while IL-6 and IL-10 did not change significantly.

Conclusions  Thirty weeks of T administration normalizing plasma T in hypogonadal men with the MetS improved some components of the MetS and a number of inflammatory markers.

Introduction

Over the last decade, a large number of review papers have highlighted the significance of depressed levels of T in the metabolic syndrome (MetS) and cardiovascular disease.1–3 Recent epidemiological studies have found that low T levels are a predictor of mortality in elderly men.4–8

Numerous studies have found inverse associations between the severity of features of the MetS and plasma T.9–13 There is an inverse relationship between waist circumference (WC), a reliable indicator of visceral obesity, and T levels over all age groups.14

Adiposity with its associated hyperinsulinism suppresses sex hormone-binding globulin (SHBG) synthesis and therewith the levels of circulating total T (tT).15,16 It may also decrease the strength of luteinizing hormone (LH) signalling to the testis.17 Further, insulin18 and leptin19 have a suppressive effect on testicular steroidogenesis. Visceral fat cells secrete a large number of cytokines which impair testicular steroidogenesis.16,20,21 There are therefore reasons to believe that adiposity is a significant factor in lowering circulating levels of T.

While it is clear that disease and, in the context of this study, in particular the MetS suppress circulating T levels, it has also been documented that low T induces the MetS.22,23 Low T and SHBG levels appeared strongly associated not only with components of the MetS but also with the MetS itself, independently of body mass index (BMI).23 The role of T is dramatically demonstrated by findings in men with prostate cancer who undergo androgen ablation therapy,24,25 particularly in the longer-term,26 which adversely affects all components of the MetS. Another study showed convincingly that acute androgen deprivation reduces insulin sensitivity in young men27 and strongly impairs glycaemic control of men with diabetes mellitus.28

Obesity, and particularly visceral fat excess, is associated with insulin resistance, hyperglycaemia, atherogenic dyslipidaemia and hypertension as well as prothrombotic and pro-inflammatory states. Low sex hormone levels are associated with inflammation.29–31. The adipocyte is now considered to be an endocrine organ secreting various peptides termed adipokines.32,33The adipokines act in a paracrine, autocrine and endocrine ways influencing lipid metabolism, glucose homeostasis and some cardiovascular risk factors, such as hypertension, as well as thrombotic and inflammatory processes. Furthermore, adipokines released from visceral fat have direct access to the liver, via the portal vein, and, therefore, a huge impact on liver metabolic functions. Some of these adipokines, such as tumour necrosis factor-alpha (TNF-α) and IL-6, are inflammatory markers and contribute to the development of insulin resistance.34–36. There is an inverse relationship between plasma T and inflammatory markers in elderly men.29,30,37 The acute withdrawal of T in young otherwise healthy hypogonadal patients caused significant increases in IL-6 and TNF-α 2 weeks after the suspension of treatment.27 Several studies have shown that administration of T to men with lower-than-normal serum T leads to a reduction in the levels of adipokines.31,38 However, these effects were not observed in a 3-month double-blinded placebo-controlled crossover study of T replacement on various adipocytokines and C-reactive protein (CRP) in type 2 diabetic men.39 T may increase lipolysis and inhibit lipid uptake, but there is also an inhibitory effect on differentiation and proliferation of preadipocytes40 and the latter may take more time to become apparent and therefore the decline of CRP might also be slower.

While the prevalence of the MetS increases with ageing, ageing per se might be a factor in the decline of circulating T. However, low T levels seem to be a consistent feature of young adult men with the MetS.41

There is increasingly evidence of a beneficial effect of T treatment on body composition, visceral fat and other elements of the MetS.12,42–47 A very recent study in newly diagnosed diabetic men showed that addition of T to treatment with diet and exercise was superior to diet and exercise alone.48

This is a randomized, placebo-controlled, double-blinded, parallel group study with a duration of 30 weeks to investigate the effects of intramuscular T undecanoate (TU) on the biochemical and anthropometric components of MetS in 184 hypogonadal patients suffering from the MetS. The predefined main primary outcome variables were the measures of body composition [i.e. body weight, WC and waist-to-hip ratio (WHR)] and the lipid spectrum [i.e. total cholesterol, low-density lipoprotein (LDL) cholesterol, high-density lipoprotein (HDL) cholesterol and triglycerides (TG)].

Subjects and methods

This study was a double-blinded, placebo-controlled phase III trial (ClinicalTrials.gov identifier: NCT00696748). A randomization procedure was followed with allocation probabilities of 0·7 and 0·3 – for the active and placebo treatment groups, respectively, aiming at inclusion of 250 subjects. As the recruitment of 250 men was slower than anticipated and in view of the risk of expiration of the trial medication, eventually 184 men aged 35–70 presenting with disorders specific to the MetS at the outpatient clinic were selected for the study. It was decided to allocate treatments in unequal proportions because only small changes were expected to occur in the placebo-treated group. This allowed us to describe with more precision the changes observed in the T-treated group. The power to detect an effect of TU vs placebo on WC, WHR and total cholesterol was calculated based on a closely related crossover trial in 24 hypogonadal men with type 2 diabetes treated with IM T39 Given the numbers of 113 experimental subjects and 71 control subjects, and a two-sided level of significance of <0·05, the power was >0·90 for all three outcomes The criteria used to identify men suffering from the MetS were those defined by International Diabetes Federation.49 A further inclusion criterion was hypogonadism defined as a baseline tT level <12·0 nm or calculated free T level <225 pm. This cut-off value was specified in recent guidelines for the treatment of late onset hypogonadism.50

All patients received written recommendations on improving their diet habits and to increase physical activity to walking at least 30–40 min/day.

Eligible patients were assigned to receive treatment with either parenteral TU 1000 mg or matching placebo. The study medication was prepared by the manufacturing company of TU (Nebido) (Bayer Schering Pharma, Berlin, Germany), and the packages were numbered. Only when the code was broken was it apparent which packages had contained TU and which placebo. After baseline measurements at visit 1, patients received parenteral TU 1000 mg or placebo, and the second injection was given 6 weeks later (visit 2). Visit 3 was 18 weeks after baseline, before the third injection. Visit 4 was at week 30. From week 30 on, all patients were administered TU, and the results of this second study from this time point on will be reported later.

The following inclusion criteria were used: men aged between 35 and 70, with tT levels below 12·0 nm (i.e. 350 ng/dl) or a calculated free T level below 225 pm (i.e. 65 pg/ml) and a diagnosis of the MetS. The cut-off values for T were specified in guidelines for the treatment of late onset hypogonadism The MetS was defined by the presence of three or more of the following International Diabetes Federation (IDF) criteria:49 (i) central obesity (WC: ≥94 cm); (ii) hypertriglyceridaemia (elevated TG level: ≥1·70 mm or ≥150 mg/dl) or specific treatment for this lipid abnormality; (iii) low HDL cholesterol (<1·03 mm or <40 mg/dl) or specific treatment for this lipid abnormality; (iv) hypertension (elevated blood pressure: systolic ≥130 and/or diastolic ≥85 mmHg) or treatment of previously diagnosed hypertension; and (v) hyperglycaemia (elevated fasting glucose level: ≥5·6 mm or ≥100 mg/dl) or previously diagnosed type 2 diabetes mellitus. The IDF criteria49 specify central obesity with a WC ≥94 cm for men rather than ≥102 cm as was defined by the NCEP definition.51

Exclusion criteria were as follows: participation in another clinical study; being incapable of giving informed consent or incapable of undergoing the necessary testing specified in the study protocol; having a serious organic or mental disease suspected based on the medical history and/or clinical examination; having prostate cancer [suspected based on the digital rectal examination of the prostate and a value of serum prostate-specific antigen (PSA) >4 μg/l], breast cancer or suspicion thereof (based on palpation of breast tissue); presence or history of hepatic tumours; having acute or chronic hepatic disease [serum alanine aminotransferase (ALT), aspartate aminotransferase (AST) four times higher than the upper limit of reference range]; presence of renal diseases with renal failure (creatinine value >150 mm); having abnormalities in biochemical or haematological laboratory values [haematocrit (Ht) >0·55], whether associated with clinical manifestations of disease or not; suspected lack of the participant’s compliance (a subjective judgment whether the patient will keep appointments); and hypersensitivity to the active substance [‘ever had an (allergic) reaction to the administration of T or other injections?’]. Some of the men were using antihypertensive (β-blockers and ACE inhibitors), lipid-lowering (statins) or antidiabetic (metformin, sulfonylurea and insulin) medication, the doses of which were held constant during the trial (Table 1).

Table 1.   Demographic and baseline clinical characteristics of the two groups
VariablenTestosteronenPlaceboP-value
  1. BMI, WC, glucose, insulin, HDL cholesterol, TG, HOMA-IR, and CRP were logarithmically transformed before analyses, and back-transformed means [95% confidence intervals (CI) in brackets] are presented.

  2. The number of MetS components was defined as: (i) WC ≥ 94 cm; (ii) elevated TG level (≥1·70 mm); (iii) low HDL cholesterol (<1·03 mm); (iv) elevated fasting glucose level (≥5·6 mm).

  3. CRP, C-reactive protein; HDL, high-density lipoprotein; HOMA-IR, homeostatic model assessment of insulin resistance; LDL, low-density lipoprotein; BMI, body mass index; MetS, metabolic syndrome; TG, triglycerides; WC, waist circumference.

Age (years)11351·6 (49·8–53·4)7152·8 (50·5–55·0)0·41
 Range (years) 35–69 35–69 
BMI (kg/m2)11335·3 (34·2–36·6)7134·2 (32·9–35·7)0·24
 Range (kg/m2) 25·4–54·2 25·1–54·8 
WC (cm)113118·0 (115·4–120·8)71116·4 (113·2–119·7)0·44
Current smoking (%)11128 (24·8%)7116 (22·5%)0·73
Glucose level (mm)1136·2 (5·9–6·4)706·1 (5·8–6·5)1·00
Insulin level (mU/l)11118·9 (16·7–21·4)6917·3 (14·8–20·3)0·45
HOMA-IR1115·4 (4·7–6·1)684·8 (4·0–5·7)0·36
Prevalent diabetes mellitus (%)11132 (28·3%)7124 (33·8%)0·43
Use of antidiabetic drugs (%)11331 (27·4%)7117 (23·9%)0·60
Total cholesterol level (mm)1135·6 (5·4–5·8)715·6 (5·3–5·9)0·97
LDL cholesterol level (mm)1133·7 (3·5–3·9)713·7 (3·4–3·9)0·63
HDL cholesterol level (mm)1131·13 (1·06–1·2)711·12 (1·03–1·21)0·93
TG level (mm)1132·04 (1·84–2·26)712·27 (2·00–2·57)0·20
Use of lipid-lowering drugs (%)11318 (15·9%)717 (9·9%)0·24
Hypertension (%)11356 (49·6%)7135 (49·3%)0·97
Use of blood pressure-lowering drugs (%)113100 (88·5%)7161 (85·9%)0·61
Number of MetS components113 71 0·17
 3 37 (32·7%) 21 (29·6%) 
 4 48 (42·5%) 23 (32·4%) 
 5 28 (24·8%) 27 (38·0%) 
Leptin (μg/l)10522·4 (19·3–26·0)6519·0 (15·8–22·9)0·24
CRP (mg/l)6729 (23–35)2729 (20–40)0·96

The study was conducted in accordance with the 2000 version of the Declaration of Helsinki and was approved by the institute’s committee on scientific investigations into human subjects by the Federal Service Of Control And Supervision In Public Health And Social Development (Roszdravnadzor), as well as by the Russian Pharmacological Committee of Ministry of Health. The participants received written information on the study objectives, the potential benefits and potential risks of participation in the study and on the requirements and burden of the study. Two copies of the informed consent were dated and signed by both the physician and the participant, one for the participant and one for the physician’s file.

Measurements

At baseline (visit 1) the following variables were assessed: age, diabetes mellitus, drug treatment, smoking, weight and height. At visits 1, 3 and 4 , body weight, BMI, WC, hip circumference (HC), WHR, haemoglobin (Hb), Ht, erythrocyte sedimentation rate, red blood cell count (RBC), white blood cell count (WBC), platelet count, tT, calculated free T (fT), estradiol (E2), SHBG, LH, insulin, leptin, glucose, cholesterol, TG, HDL cholesterol, LDL cholesterol, total PSA, bilirubin and creatinine were measured. At visit 1 and at visit 4, the following variables were measured: ALT, AST, CRP, IL-1β, IL-6, IL-10, TNF-α and prostate volume. At visits 1, 3 and 4, the International Prostate Symptom Score (IPSS) was assessed.

To perform blood count analysis, blood was collected from the antecubital vein after an overnight fast using BD Vacutainer® gel tubes (Franklin Lakes, NJ, USA). The blood count analysis was performed with the automatic haematological analyser HmX (Beckman Coulter, Brea, CA, USA).

To perform biochemistry, blood collection followed the same procedure as previously described, but blood was directly centrifuged for 15 min at a temperature of +4 °C at 2000 g. Levels of glucose, total cholesterol, LDL and HDL cholesterol and TG were determined using commercial kits (Roche Diagnostics GmbH, Penzberg, Germany) on a Hitachi 912 autoanalyzer (Roche Diagnostics GmbH). Endocrine measurements were taken using a Vitros 3600 system (Ortho-Clinical Diagnostics; Johnson & Johnson company, New Brunswick, NJ, USA) with a chemiluminescence immunoassay technology. Total PSA level was determined using an ARCHITET i2000sr system (ABBOTT, Abbott Park, IL, USA) with a chemiluminescence immunoassay technology. Free T levels were estimated using the Vermeulen formula.52 Using the glucose and insulin levels measured, we derived the homeostatic model assessment of insulin resistance (HOMA-IR; calculated as fasting insulin × fasting glucose/22·5).53

To assess the concentrations of inflammatory markers, blood collection and centrifugation followed the same procedure as mentioned earlier, but the serum was frozen at −20 °C until analysis. Concentrations of interleukins and TNF-α were determined with immunoenzyme technology (ELISA) on VICTOR (Bender MedSystems, Vienna, Austria) analyzer. Levels of CRP were determined with high-sensitive (hs-CRP) immunoturbidimetric technology on COBAS (Roche) analyzer. The producers of kits used provided the following reference ranges: in healthy donors IL-1β, 10 and TNF-α are not detected, levels of IL-6 vary between 1·4 and 14·1 ng/l and CRP between 0 and 5 mg/l. All assays of hormones and cytokines and other laboratory measurements were performed in the laboratories of the Research Center for Endocrinology, Moscow, Russia.

Ultrasound examination was performed with ultrasound scanners LOGIG 7 and AU 4 IDEA with different format sensors in frequency range from 2.5–10 Megahertz (MHz). Prostate was examined with transrectal ultrasound sensor in grey-scale (B) regimen.

Statistical analysis

Because of positively skewed data, hormone levels, BMI, WC, glucose, insulin, HOMA-IR, HDL cholesterol, TG, inflammatory markers, total PSA, prostate volume, IPSS and bilirubin were logarithmically transformed before analyses, and back-transformed means [95% confidence intervals (CI) between brackets] are presented. For nontransformed continuous variables, means (95% CI) are given.

Comparisons between groups were made using multilevel regression analysis (i.e. mixed models) using a compound symmetry covariance model. Measurements in each subject were taken up to three times, and therefore a two-level structure consisted of the observations (i.e. lower level) and the subject (i.e. higher level). The overall P-value is given for the interaction term of group × time (change from baseline to follow-up). For metabolic factors, the absolute changes over 30 weeks vs baseline are also given, and these were compared between the two groups using t-tests for independent samples.

Spearman rank correlation was used to assess the relationship between continuous changes in hormonal and metabolic variables (i.e. total and free T, BMI, WC, glucose, insulin, total cholesterol, LDL cholesterol, HDL cholesterol and TG. No adjustments were made for multiple comparisons, and all tests were two-tailed with P < 0·05 denoting statistical significance. The software used was spss version 17.0 (SPSS Inc., Chicago, IL, USA).

Results

Of the total of 184 men, 14 dropped out of the study (six receiving placebo and eight receiving TU). This resulted in 105 (92·9%) men receiving TU and 65 (91·5%) receiving placebo completing the trial (Fig. 1).

Figure 1.

 Enrolment, randomization and follow-up of the study participants.

At baseline, there were no differences between the treatment group and the placebo group in any of the variables tested in this study (Table 1). In the group receiving T treatment, plasma total and free T rose significantly and SHBG did not change significantly. No changes in plasma E2 were noted. Plasma LH levels were suppressed in the treatment group (Table 2).

Table 2.   Blood hormone levels over 30 weeks of study in the two groups
 nTestosteronenPlaceboP-valueOverall P-value
  1. LH was not assessed at 18 weeks. All levels were logarithmically transformed before analyses, and back-transformed means [95% confidence intervals (CI) between brackets] are presented. The overall longitudinal P-values represent the effects of the interaction terms of group × time (change from baseline to follow-up) using multilevel linear models.

  2. SHBG, sex hormone-binding globulin; LH, luteinizing hormone.

Total T (nm)
 Baseline1136·7 (6·0–7·4)717·5 (6·6–8·5)0·20< 0·001
 18 weeks10811·9 (10·8–13·1)668·0 (7·1–9·1)< 0·001
 30 weeks10513·1 (11·9–14·4)657·8 (6·8–8·8)< 0·001
Calculated free T (pm)
 Baseline113120 (107–135)71130 (113–151)0·46<0·001
 18 weeks108241 (214–271)66146 (126–169)<0·001
 30 weeks105274 (243–308)65146 (125–169)<0·001
SHBG (nm)
 Baseline11331·8 (29·3–34·6)7135·9 (32·3–39·9)0·090·39
 18 weeks10731·0 (28·5–33·7)6633·2 (29·8–36·9)0·31
 30 weeks10529·5 (27·1–32·2)6531·6 (28·4–35·2)0·28
Estradiol (pm)
 Baseline10997 (89–105)6798 (88–109)0·940·13
 18 weeks105108 (99–117)6498 (88–109)0·11
 30 weeks101106 (97–115)6197 (87–108)0·17
LH (IU/l)
 Baseline1133·8 (3·2–4·4)713·3 (2·7–4·0)0·27<0·001
 30 weeks1040·9 (0·7–1·1)653·3 (2·7–4·0)<0·001

The effects of the intervention with TU on variables of the MetS are presented in the Table 3 and Fig. 2. There were significant decreases in weight, BMI, WC, HC and WHR in the treatment group compared with the placebo group. Of all the biochemical variables that were closely linked to the MetS, only plasma insulin and leptin decreased significantly. The HOMA-IR as a measure of insulin resistance also decreased in the TU vs placebo-treated group.

Table 3.   Comparisons of absolute values and changes in metabolic factors over 30 weeks time in the two groups
 nTestosteronenPlaceboP-valueOverall P-value
  1. BMI, WC, leptin, glucose, insulin, HOMA-IR, HDL cholesterol and TG levels were logarithmically transformed before analyses, and back-transformed means [95% confidence intervals (CI) between brackets] are presented. For nontransformed continuous variables, means (95% CI) are given. The overall longitudinal P-values represent the effects of the interaction terms of group × time (change from baseline to follow-up) using multilevel linear models.

  2. HDL, high-density lipoprotein; HOMA-IR, homeostatic model for assessment of insulin resistance; LDL low-density lipoprotein; BMI, Body mass index; TG, triglycerides; WC, waist circumference.

BMI (kg/m2)
 Baseline11335·3 (34·2–36·6)7134·2 (32·9–35·7)0·24<0·001
 18 weeks10934·7 (33·6–35·9)6733·9 (32·5–35·4)0·58
 30 weeks10534·1 (33·0–35·2)6534·1 (32·7–35·6)0·75
 Δ30 weeks vs baseline105−1·32 (SEM 0·23)65−0·11 (SEM 0·21)<0·001
Weight (kg)
 Baseline113115·1 (110·6–119·6)71111·0 (105·4–116·7)0·28<0·001
 18 weeks109113·0 (108·6–117·5)67110·0 (104·3–115·6)0·69
 30 weeks105110·8 (106·3–115·3)65110·6 (105·0–116·3)0·78
 Δ30 weeks vs baseline105−4·31 (SEM 0·75)65−0·40 (SEM 0·71)<0·001
WC (cm)
 Baseline113118·0 (115·4–120·8)71116·4 (113·2–119·7)0·44<0·001
 18 weeks109114·2 (111·7–116·9)67114·2 (111·1–117·5)0·83
 30 weeks105112·2 (109·7–114·7)65114·9 (111·7–118·2)0·15
 Δ30 weeks vs baseline105−6·02 (SEM 0·77)65−1·46 (SEM 0·54)<0·001
Hip circumference (cm)
 Baseline113119·1 (116·7–121·5)71116·6 (113·6–119·6)0·21<0·001
 18 weeks109117·6 (115·2–120·0)67115·7 (112·7–118·7)0·49
 30 weeks105116·1 (113·8–118·5)65116·7 (113·6–119·7)0·72
 Δ30 weeks vs baseline105−2·94 (SEM 0·67)650·02 (SEM 0·47)<0·001
Waist-to-hip ratio
 Baseline1131·00 (0·99–1·01)711·01 (0·99–1·02)0·590·04
 18 weeks1080·98 (0·97–0·99)670·99 (0·98–1·01)0·10
 30 weeks1050·97 (0·96–0·99)650·99 (0·98–1·01)0·03
 Δ30 weeks vs baseline105−0·027 (SEM 0·005)65−0·013 (SEM 0·005)0·07
Leptin (μg/l)
 Baseline10522·4 (19·3–26·0)6519·0 (15·8–22·9)0·240·001
 18 weeks10416·6 (14·3–19·3)6218·3 (15·1–22·1)0·33
 30 weeks9613·8 (11·8–16·1)5916·0 (13·2–19·3)0·17
 Δ30 weeks vs baseline96−11·6 (SEM 1·7)57−5·8 (SEM 2·3)0·04
Glucose (mm)
 Baseline1136·2 (5·9–6·4)706·1 (5·8–6·5)1·000·23
 18 weeks1076·0 (5·8–6·3)666·0 (5·7–6·4)0·94
 30 weeks1055·9 (5·6–6·2)646·1 (5·8–6·5)0·28
 Δ30 weeks vs baseline105−0·37 (SEM 0·17)63−0·10 (SEM 0·12)0·26
Insulin (mIU/l)
 Baseline11118·9 (16·7–21·4)6917·3 (14·8–20·3)0·450·07
 18 weeks10814·1 (12·4–16·0)6618·0 (15·3–21·1)0·01
 30 weeks10415·6 (13·7–17·7)6517·4 (14·8–20·4)0·24
 Δ30 weeks vs baseline104−5·48 (SEM 2·93)631·08 (SEM 2·79)0·13
HOMA-IR
 Baseline1115·4 (4·7–6·1)684·8 (4·0–5·7)0·360·04
 18 weeks1074·0 (3·4–4·6)654·9 (4·1–5·8)0·02
 30 weeks1044·3 (3·7–4·9)644·9 (4·1–5·8)0·18
 Δ30 weeks vs baseline104−1·49 (SEM 0·85)610·20 (SEM 0·84)0·19
Total cholesterol (mm)
 Baseline1135·6 (5·4–5·8)715·6 (5·4–5·9)0·970·32
 18 weeks1075·4 (5·2–5·6)675·4 (5·1–5·7)0·92
 30 weeks1055·4 (5·1–5·6)655·5 (5·2–5·8)0·38
 Δ30 weeks vs baseline105−0·24 (SEM 0·10)65−0·10 (SEM 0·11)0·32
LDL cholesterol (mm)
 Baseline1133·7 (3·5–3·9)713·7 (3·4–3·9)0·630·07
 18 weeks1073·5 (3·3–3·7)673·5 (3·3–3·8)0·95
 30 weeks1053·4 (3·2–3·5)643·5 (3·2–3·7)0·66
 Δ30 weeks vs baseline105−0·39 (SEM 0·08)64−0·16 (SEM 0·10)0·08
HDL cholesterol (mm)
 Baseline1131·10 (1·05–1·16)711·09 (1·01–1·16)0·690·17
 18 weeks1071·13 (1·07–1·19)671·03 (0·96–1·10)0·02
 30 weeks1051·18 (1·12–1·25)641·10 (1·02–1·17)0·11
 Δ30 weeks vs baseline1050·076 (SEM 0·036)640·003 (SEM 0·068)0·30
TG (mm)
 Baseline1132·04 (1·84–2·26)712·27 (2·00–2·57)0·200·46
 18 weeks1071·91 (1·71–2·12)672·03 (1·77–2·31)0·28
 30 weeks1051·77 (1·58–1·97)652·08 (1·82–2·37)0·02
 Δ30 weeks vs baseline105−0·32 (SEM 0·12)65−0·15 (SEM 0·11)0·32
Figure 2.

 Geometric mean waist circumference, glucose levels, high-density lipoprotein cholesterol levels and triglyceride levels at baseline and after 18 and 30 weeks (on logarithmic scales). P-values by multilevel regression analysis (i.e. mixed models) for the time × group effect.

In the TU-treated group, there were significant associations between absolute changes in endocrine and metabolic characteristics over 30 weeks. Decreases in BMI, WC, total cholesterol and LDL cholesterol correlated with the increases in tT. Decreases in BMI and WC also correlated with the increases in calculated free T. Decreases in BMI correlated with the decreases in WC, total cholesterol, LDL cholesterol and TG, and decreases in WC correlated with the decreases in total cholesterol, LDL cholesterol and TG. Changes in serum glucose levels correlated with the changes in TG, and changes in serum cholesterol with changes in LDL cholesterol and TG. In the placebo group, changes in BMI and WC were more weakly linked to lipid changes than in the TU-treated group.

The effects of T administration on markers of inflammation are presented in Table 4. Levels of TNF-α and CRP declined profoundly, while the effect on levels of IL-1β was less pronounced but significant. Levels of IL-6 and IL-10 were not affected by T administration. Changes in CRP were correlated with changes in total and free T and changes in serum LDL cholesterol levels. Otherwise, no correlations could be established between the changes in markers of inflammation on the one hand and changes in T or markers of the MetS on the other.

Table 4.   Inflammatory markers over 30 weeks in the two groups
 nTestosteronenPlaceboP-valueOverall P-value
  1. All inflammatory markers were logarithmically transformed before analyses, and back-transformed means [95% confidence intervals (CI) between brackets] are presented. The overall longitudinal P-values represent the effects of the interaction terms of group × time (change from baseline to follow-up) using multilevel linear models.

  2. CRP, C-reactive protein.

Erythrocyte sedimentation rate (mm/h)
 Baseline1126·5 (5·8–7·4)716·9 (5·9–8·1)0·530·78
 18 weeks1076·9 (6·1–7·8)678·1 (6·9–9·4)0·16
 30 weeks1056·5 (5·7–7·3)657·1 (6·0–8·3)0·45
CRP (mg/l)
 Baseline6729 (23–35)2729 (20–40)0·96<0·001
 30 weeks6519 (15–24)2638 (27–52)0·004
Interleukin 1 beta (ng/l)
 Baseline671·7 (1·4–2·1)272·1 (1·5–2·8)0·300·008
 30 weeks651·5 (1·2–1·8)262·8 (2·1–3·6)0·001
Interleukin 6 (ng/l)
 Baseline671·1 (0·9–1·4)271·2 (0·9–1·7)0·5460·07
 30 weeks651·1 (0·9–1·4)260·9 (0·6–1·3)0·317
Interleukin 10 (ng/l)
 Baseline664·9 (4·5–5·3)275·6 (4·9–6·3)0·0780·15
 30 weeks655·1 (4·7–5·5)265·4 (4·7–6·1)0·528
Tumour necrosis factor-alpha (ng/l)
 Baseline663·5 (2·8–4·4)274·7 (3·3–6·5)0·1970·03
 30 weeks652·4 (1·8–3·0)264·7 (3·3–6·6)0·003

There are safety concerns about the administration of T, particularly to elderly men. In the 30-week period of TU administration, levels of Hb rose significantly as did the RBC counts and the Ht, but values did not exceed the upper limit of the reference range. The mean absolute increases in Hb levels in the TU group were +6·3 (SD 10·5) g/l and +7·7 (SD 11·4) g/l after 18 and 30 weeks, respectively, and the increases in Ht were +2·0%(SD 3·4) and +2·0%(SD 3·9) after 18 and 30 weeks, respectively. In both the placebo group and the T treatment group, values of PSA did not rise significantly. Prostate volume did not increase over the treatment period. Levels of PSA and prostate volume were not significantly different after 30 weeks in the men with T treatment compared with the placebo group. Over the 30-week study, there were no changes in the IPSS (Table 5).

Table 5.   Safety markers of prostate and haematopoiesis over 30 weeks in the two groups
 nTestosteronenPlaceboP-valueOverall P-value
  1. Prostate volume was not assessed at 18 weeks. Total PSA, prostate volume, IPPS and bilirubin were logarithmically transformed before analyses, and back-transformed means [95% confidence intervals (CI) of the mean between brackets] are presented. For other variables, means (95% CI of the mean) are given. The overall longitudinal P-values represent the effects of the interaction terms of group × time (change from baseline to follow-up) using multilevel linear models.

  2. PSA, prostate specific antigen.

Total PSA level (μg/l)
 Baseline1130·8 (0·7–0·9)710·8 (0·7–1·0)0·490·13
 18 weeks1070·8 (0·7–1·0)660·9 (0·7–1·1)0·73
 30 weeks1040·8 (0·7–1·0)651·0 (0·8–1·2)0·19
Prostate volume (ml)
 Baseline5227·9 (25·2–30·9)3034·0 (29·7–39·0)0·030·45
 30 weeks4628·3 (25·5–31·4)2233·4 (29·0–38·4)0·09
International Prostate Symptomatic Score (IPPS)
 Baseline1133·5 (2·8–4·4)713·9 (3·0–5·2)0·520·92
 18 weeks1083·0 (2·4–3·8)673·6 (2·7–4·8)0·63
 30 weeks1052·9 (2·3–3·7)653·4 (2·5–4·5)0·75
Haemoglobin (g/l)
 Baseline113154 (151–156)71155 (151–158)0·61<0·001
 18 weeks107160 (157–162)67152 (149–155)0·001
 30 weeks105161 (159–164)65152 (149–155)<0·001
Haematocrit (%)
 Baseline11045·6 (44·8–46·5)6845·1 (44·0–46·1)0·43<0·001
 18 weeks10647·6 (46·8–48·5)6744·6 (43·6–45·7)<0·001
 30 weeks10547·7 (46·8–48·5)6444·6 (43·6–45·7)<0·001
Red blood cells (×1012 cells/l)
 Baseline1135·12 (5·03–5·21)715·17 (5·06–5·29)0·47<0·001
 18 weeks1075·36 (5·27–5·45)675·11 (4·99–5·22)<0·001
 30 weeks1055·39 (5·30–5·48)655·15 (5·04–5·27)0·005
White blood cells (×109 cells/l)
 Baseline1127·55 (7·2–7·91)717·42 (6·98–7·87)0·660·10
 18 weeks1077·97 (7·61–8·33)677·69 (7·24–8·14)0·33
 30 weeks1057·88 (7·52–8·24)657·33 (6·87–7·78)0·02
Bilirubin (μm)
 Baseline11111·4 (10·4–12·5)6912·8 (11·5–14·4)0·130·02
 18 weeks10510·2 (9·3–11·2)6611·9 (10·6–13·3)0·07
 30 weeks10411·9 (10·9–13·1)6411·2 (10·0–12·6)0·42

Discussion

There is an abundance of studies showing that men with the MetS, cardiovascular disease and type 2 diabetes mellitus have low plasma T levels.2,54,55 Visceral obesity suppresses T production55 and low T levels induce visceral obesity and insulin resistance,2 as shown in men on androgen deprivation for prostate cancer.56 There are relatively few studies of men with the MetS and hypogonadal values of plasma T, which analyse the effects of restoring T to eugonadal levels. The present study is the first randomized double-blinded, placebo-controlled study that investigates the effect of correction of hypogonadal values of plasma T in men with MetS. It was performed in a large cohort of 184 hypogonadal MetS men between the ages of 35 and 70, of whom 170 completed the trial, receiving T or placebo, over a period of 30 weeks. There were significant improvements in weight, BMI, WC, HC and WHR in the treatment group compared with the placebo group. BMI, WC and WHR are predictors of the MetS and cardiovascular risk.57,58 Of all the biochemical variables of the MetS, only plasma insulin and leptin decreased significantly.

In the group receiving TU injections, plasma T levels were quantitatively not much elevated. This might be attributed to the fact that blood samples for the measurement of T were taken just before the next TU injection was due, and the T levels generated from the TU injections were at their lowest.59 It is very likely that the average plasma T level would have been higher over the injection interval. In this double-blinded study, dose adjustments were not possible and this might have been desirable. The benefits of T treatment are related to achieved levels of T42, and we might have observed quantitatively better effects on the study variables if the plasma T levels had been higher, which would have been the case with shorter intervals between TU injections.

The improvements in variables of the MetS were modest and mostly not apparent in the biochemical variables. It may be that changes in the latter are the result of changes in body fat, and it might be that the duration of the study was too short for these effects to become manifest.47,60

Serum SHBG did not change significantly. An increase had been expected because features of the MetS had improved,61,62 but the increase may have been offset by the increase in circulating T.

It has been reported that acute T withdrawal is associated with an increase in plasma insulin levels and reduced insulin sensitivity in hypogonadal men.27 The decrease in plasma insulin in our study may be interpreted as a sign of improved insulin sensitivity upon normalization of plasma T.

There are relatively few studies assessing the effect of T treatment on body composition, visceral fat and other elements of the MetS,12,42,44,46,47,63,64 and our results are largely in agreement with those findings by a reduction in body fat and a reduction in visceral fat (with WC, HC and WHR ratio serving as indices). The much longer study of Page et al.47 over 36 months also found improvements in lipid profiles, not encountered in our study, but found in other nonblinded studies over 24 months.

Obesity is a state of low-grade inflammation, with adipose tissue generating substantial quantities of proinflammatory molecules. This explains the observed relationships between insulin resistance and endothelial dysfunction better than a model revolving around insulin resistance.34,35 There is an inverse relationship between the levels of plasma T and inflammatory markers.29,30 In our study, levels of TNF-α and CRP declined profoundly, while the effect on levels of IL-1β was less pronounced but significant. Levels of IL-6 and IL-10 were not affected by T administration. This is in agreement with other studies of T replacement finding a reduction in IL-1β, IL-6 and TNF-α31,38 but was not confirmed in a study of T administration over 3 months.39 In a recent study, we found that serum levels of CRP decline slowly following the administration of T65, and the observation period of 3 months may have been too short for a reduction in cytokines to become manifest in the study of Kapoor et al.39 This reduction may be linked to a reduction in visceral fat.34 These changes as well as a reduction in body weight and visceral fat are associated with a reduced risk of cardiovascular disease and type 2 diabetes, risks which, importantly, are increased in hypogonadal men with the MetS.34,35

In this short period of 30 weeks of T administration, there were no adverse events with regard to an increase in the Ht or prostate pathology. Another study found no effect on prostate size over a similar duration of T administration.66 Evidently, such a short period does not allow the conclusions as to the safety of T treatment on the prostate.

There are some limitations to our study. It was of relatively short duration (30 weeks), and effects on blood pressure over time were not assessed. Studies of longer duration of T administration have shown that the beneficial effects are progressive over time.47,65

In conclusion, in this short study of administration of T to men with the MetS, a number of risk factors for diabetes mellitus and cardiovascular disease showed improvement. Our study confirms that restoring plasma T to eugonadal values in men with the metabolic syndrome decreases body weight, body mass index, waist circumference and visceral fat and reduces plasma insulin and leptin levels. Levels of glucose and lipids did not improve significantly, but their changes were correlated with changes in plasma T. Also, some markers of inflammation decreased. However, longer-term studies with clinical end-points are necessary to determine whether these improvements contribute to a reduced risk of developing diabetes and/or cardiovascular disease.

Conflicts of interest

Farid Saad reports being an employee of Bayer Schering Pharma AG, which produces the intramuscular depot formulation of T undecanoate. The other authors declare no conflict of interests.

Funding

The study was partially financially supported by Bayer Schering Pharma, Berlin, Germany.

Ancillary

Advertisement