Clinical Endocrinology (2016) 84, 55–62
doi: 10.1111/cen.12842
ORIGINAL ARTICLE
Effect of testosterone treatment on cardiac biomarkers in a randomized controlled trial of men with type 2 diabetes Emily J. Gianatti*,†, Rudolf Hoermann*, Que Lam‡, Philippe Dupuis*,†, Jeffrey D. Zajac*,† and Mathis Grossmann*,† *Department of Medicine Austin Health, Austin Health, University of Melbourne, †Endocrine Unit, Austin Health, University of Melbourne, and ‡Department of Biochemistry, Austin Health, University of Melbourne, Heidelberg, Vic., Australia
Summary Objective To assess the effect of testosterone treatment on cardiac biomarkers in men with type 2 diabetes (T2D). Design Randomized double-blind, parallel, placebo-controlled trial. Patients Men aged 35–70 years with T2D and a total testosterone level ≤120 nmol/l (346 ng/dl) at high risk of cardiovascular events, median 10-year United Kingdom Prospective Diabetes Study (UKPDS) coronary heart disease (CHD) risk 21% (IQR 16%, 27%). Eighty-eight participants were randomly assigned to 40 weeks of intramuscular testosterone undecanoate (n = 45) or matching placebo (n = 43). Main Outcome Measures N-terminal pro B-type natriuretic peptide (NT-proBNP) and high-sensitivity cardiac troponin T (hs-cTnT). Result Testosterone treatment reduced NT-proBNP (mean adjusted difference (MAD) in change over 40 weeks across the testosterone and placebo groups, 179 ng/l [95% CI 324, 35], P = 0047), but did not change hs-cTnT (MAD, 041 ng/l (95% CI 056, 139), P = 062). Six men, three in each group experienced an adverse cardiac event, displaying already higher baseline NT-proBNP (P < 001) and hs-cTnT levels (P = 001). At baseline, 10-year UKPDS CHD risk was associated positively with NT-proBNP (s = 021, P = 0004) and hs-cTnT (s = 023, P = 0003) and inversely with testosterone (total testosterone s = 018, P = 002, calculated free testosterone s = 019, P = 001), but there was no significant association between testosterone and cardiac biomarkers (P > 005). Conclusions In this trial of men with T2D and high cardiovascular risk, testosterone treatment reduced NT-proBNP and did
Correspondence: Mathis Grossmann, Department of Medicine Austin Health, The University of Melbourne, 145 Studley Road, Heidelberg, Vic., 3084, Australia. Tel.: +613 9496 5000; Fax: +613 9496 3365; E-mail [email protected] Trial Registration Number: ClinicalTrials.gov NCT00613782. © 2015 John Wiley & Sons Ltd
not change hs-cTnT. Further studies should determine whether men with increased cardiac biomarkers prior to testosterone therapy are at higher risk of testosterone treatment-associated adverse cardiac events. (Received 27 January 2015; returned for revision 3 May 2015; finally revised 10 May 2015; accepted 23 June 2015)
Introduction Despite a wealth of preclinical and observational data, the effects of testosterone on the male cardiovascular system remain uncertain.1,2 This is because there have been no randomized controlled clinical trials (RCTs) of testosterone treatment adequately designed or sufficiently powered to evaluate cardiovascular outcomes. Meta-analyses of existing RCTs of testosterone therapy have been inconclusive and are limited by low numbers of adverse cardiac events.3–5 Testosterone prescriptions for older men have risen markedly in recent years.6 Cardiovascular disease is a leading cause of death in such men, especially in those with coexisting type 2 diabetes (T2D).7 Therefore, a better understanding of the effects of testosterone treatment on cardiovascular health in men with T2D is important. Several cardiac biomarkers such as troponin and brain natriuretic peptides correlate with subclinical cardiac injury in asymptomatic patients and provide cardiac risk stratification beyond conventional risk factors.8 Recently, it has been shown that both N-terminal pro B-type natriuretic peptide (NTproBNP) and high-sensitivity cardiac troponin T (hs-cTnT) improve the accuracy with which the risk of cardiovascular events can be predicted in patients with T2D.9 In this exploratory analysis, we assessed the effects of testosterone treatment on circulating NT-proBNP and hs-cTnT levels in an RCT of men with T2D and lowered testosterone levels at high risk of cardiovascular events. We have previously reported the effects of testosterone treatment on glucose metabolism,10 the primary outcomes of this trial and on constitutional and sexual symptoms11 in this population. 55
56 E. J. Gianatti et al.
Material and methods Design overview The design of this trial is reported in detail elsewhere.10 Briefly, this 40-week, randomized, double-blind, placebo-controlled trial (ClinicalTrials.gov NCT00613782) was conducted at a tertiary referral centre (Austin Health, Melbourne, Australia) between November 2009 and February 2013 and was approved by the Human Research Ethics Committee, Austin Health. Each participant provided written informed consent prior to entering the study. Men aged 35–70 years of age were eligible to participate in this trial if they had a history of T2D and the total testosterone level (averaged from two fasting morning specimens) was ≤120 nmol/l (346 ng/dl), as measured by an electrochemiluminescence immunoassay (ECLIA).10 Although total testosterone was measured by both ECLIA and liquid chromatography–tandem mass spectroscopy (LCMS/MS), recruitment was based on ECLIA because the LCMS/MS assay was not available for routine clinical use. Therefore, samples were batched and measured by LCMS/MS at study end.10 Exclusion criteria were as described previously10,11 and included uncontrolled hypertension (>160/90 mmHg despite treatment) and cardiac insufficiency (New York Heart Association score >2). Eligible participants were randomly assigned in a concealed 1:1 allocation to either intramuscular testosterone undecanoate 1000 mg or a visually identical placebo injection at 0, 6, 18 and 30 weeks.10 Trial investigators and participants were blinded to intervention allocation. All subjects received written, uniform recommendations regarding diet and physical activity. Outcomes and follow-up The main outcome measures of the RCT were changes in glucose metabolism.10 The main outcomes of this exploratory analysis were changes in the cardiac biomarkers NT-proBNP and hs-cTnT. Assessments and measurements At 0, 18 and 40 weeks, the following relevant variables were assessed: drug treatment, adverse events, body mass index (BMI), waist circumference, blood pressure, calculated 10-year United Kingdom Prospective Diabetes Study (UKPDS) coronary heart disease (CHD) risk,12 HbA1c, lipid profile, trough total testosterone (TT), calculated free testosterone (cFT) and NTproBNP and hs-cTnT levels. All blood samples were collected in early morning, fasted state prior to trial medication administration. Changes in antihypertensive medications were allowed but recorded at each visit. TT was measured by both ECLIA and liquid chromatography–tandem mass spectroscopy (LCMS/MS), and cFT was calculated as described.10,13 Unless specified otherwise, all analyses presented are based on TT measured by a validated LCMS/MS.13 NT-proBNP and hs-cTnT were measured at the conclusion of the trial from serum stored at 70°C collected at 0, 18 and
40 weeks by ECLIA on a E602 automated platform (Roche Diagnostics, Burgess Hill, UK). Assays were performed using the manufacturer’s calibrators and quality controls. The NT-proBNP assay has an effective measuring range of 5–30 000 ng/l, and NT-proBNP levels 4mcg/l, or haematocrit >050 (n = 61). Two men who were randomized to the testosterone group had markedly elevated baseline NTproBNP levels suggestive of occult congestive cardiac failure. Both developed overt clinical congestive cardiac failure early in the trial. Data from these men were excluded, leaving 86 men in the analysis. Seventy-five men completed the trial. The most common reason for noncompletion was a predefined protocol violation of intensification of oral hypoglycaemic agents or commencement of insulin therapy, and there was no difference between the treatment groups in this regard.10 Baseline characteristics of the 86 participants are shown in Table 1. At baseline, cardiovascular risk as defined by 10-year UKPDS CHD risk was high, median 21% (IQR 16%, 27%), and 965% of men had prevalent metabolic syndrome. 116% of men had elevated NT-proBNP (≥125 ng/l). All men in the trial had detectable hs-cTnT with 140% having elevated levels ≥ = 14 ng/ l (Table 1). There was no significant difference in baseline NTproBNP (P = 023) or hs-cTnT (P = 006) between testosteroneand placebo-treated participants. At baseline, all men were assessed to be New York Heart Association (NYHA) 1 (no evidence of cardiac failure) and free of clinical symptoms suggestive of cardiovascular ischaemia. Testosterone levels Baseline TT measured by LCMS/MS was 106 nmol/l (306 ng/ dl) and 110 nmol/l (317 ng/dl) in the testosterone and placebo groups, respectively (P = 086), and TT measured by ECLIA was 87 nmol/l (251 ng/dl) and 85 nmol/l (245 ng/dl) (P = 063) (Table 1). At 40 weeks, testosterone levels increased significantly © 2015 John Wiley & Sons Ltd Clinical Endocrinology (2016), 84, 55–62
Table 1. Baseline characteristics of the participants Testosterone N = 43 Age, y Duration of diabetes, y Insulin therapy, n (%) Metabolic syndrome (ATPIII criteria), n (%) Prior cardiovascular disease, n (%) Smoking status Current smoker, n (%) Exsmoker, n (%) Nonsmoker, n (%) 10-year UKPDS CHD risk, % BMI, kg/m2 Waist circumference, cm Systolic blood pressure, mmHg Diastolic blood pressure, mmHg Total testosterone (ECLIA), nmol/l Total testosterone (LCMS/MS), nmol/l Calculated free testosterone (LCMS/MS), pmol/l eGFR, ml /min/ 173 m2 HbAlc, % Cholesterol, mmol/l LDL, mmol/l HDL, mmol/l TG, mmol/l NT-proBNP, ng/l hs-cTnT, ng/l
62 [58, 68] 7 [4, 11]
Placebo N = 43
P-value
62 [58, 67] 9 [5, 12]
082 049
6 (140)
10 (233)
027
42 (978)
41 (953)
056
12 (279)
5 (116)
010
4 (93)
3 (70)
090
23 (535) 16 (372) 20 [15, 28]
25 (581) 15 (349) 22 [17, 27]
308 [283, 348] 1100 [1040, 1200]
334 [314, 354] 1150 [1100, 1210]
004 006
140 [130, 150]
140 [129, 150]
095
72 [70, 80]
80 [70, 82]
004
87 [71, 111]
85 [73, 106]
063
106 [90, 128]
110 [82, 133]
086
225 [186, 298]
247 [183, 314]
068
840 [740, 900] 68 [64, 74] 42 [38, 48] 23 11 16 600
[17, 27] [09, 13] [11, 24] [400, 815]
70 [60, 110]
034
86 [745, 900]
089
71 [67, 75] 45 [37, 48]
007 098
22 10 18 480
[18, 28] [08, 12] [13, 24] [395, 725]
90 [65, 130]
080 018 032 023 006
Values are given as median and [interquartile range] or as n and (%). P values were calculated for the difference between groups using Wilcoxon’s test or chi-square test. P < 005 was considered significant.
in the testosterone group, while there was no significant change in the placebo group. The mean adjusted differences (MAD) in change over 40 weeks across the two groups were TT 602 nmol/l (173 ng/dl) [95% CI 38, 83 (110, 239)], P < 0001 and cFT 181 pmol/l (53 pg/ml) [95% CI 108, 255 (31, 74)], P < 0008.
58 E. J. Gianatti et al. Changes in main outcome measures NT-proBNP decreased with mean adjusted difference (MAD) between groups across 40 weeks of 179 ng/l [95% CI 324, 35], P = 0047 (Table 2). NT-proBNP returned on average more frequently to below the cut off of 125 ng/l during the course of the trial in testosterone-treated men compared to placebo-treated men: MAD 5 [95% 1413, 129], P = 0027. Hs-cTnT was stable in both groups over the duration of the trial with MAD between groups across 40 weeks of 041 ng/l [95% CI 056, 139], P = 062. With respect to medications that may influence NT-proBNP levels, no participant commenced insulin during the study, and there was no between group difference in changes in total daily dose of insulin.10 At baseline, 21 men were receiving diuretic treatment, nine in the testosterone group and 12 men in the placebo group (P = 045). The most commonly used type of diuretic was hydrochlorothiazide (n = 16), and other types indapamide (n = 4), amiloride (n = 1), spironolactone (n = 1) and frusemide (n = 1), with one participant in each group taking two different diuretics. Importantly, of all men receiving diuretic treatment, only two men in the testosterone group and two men in the placebo group had minor adjustment to their diuretic dose, and no participant commenced a diuretic during the study.
Baseline associations between cardiac biomarkers, cardiovascular risk and testosterone levels At baseline in the whole cohort, 10-year UKPDS CHD risk was positively associated with both baseline NT-proBNP (s = 021 P = 00043) and hs-cTnT (s = 023 P = 00031) and negatively with baseline testosterone level: TT LCMS s = 018 P = 0017, cFT LCMS s = 019 P = 001. At baseline, neither NT-proBNP nor hs-cTnT was associated with testosterone levels: NTproBNP: TT s = 0034 (P = 065), cFT s = 0053 (P = 047), hs-cTnT: TT s = 0051 (P = 050), cFT s = 0073 (P = 034). The change in serum testosterone observed in the whole cohort over the trial duration was associated with an inverse change in NT-proBNP: delta TT LCMS/MS s = 017 (P = 0032), cFT LCMS/MS s = 017 (P = 0033), while no significant association was observed with change in hs-cTnT (Fig. 1). Adverse cardiovascular events Six cardiovascular events occurred during the trial10 among the 88 randomized participants, associated with higher baseline NTproBNP (P < 001) and higher baseline hs-cTnT (P = 001). Details are given in Table 3.
Discussion Changes in cardiovascular risk factors There was no significant between group change in BMI, blood pressure or HbA1c,10 or in 10-year UKPDS CHD risk (P = 023). As reported previously,10 testosterone treatment decreased total cholesterol (MAD 045 mmol/l [95% CI 07, 02], P < 0001), LDL cholesterol (MAD 026 mmol/l [95% CI 046, 006], P = 001) and HDL (MAD 011 mmol/l [95% CI 019, 003], P = 0002), but there was no significant between group change in CRP. There was no significant between-group difference in the proportion of participants who had changes in their antihypertensive medications during the study, 256% in testosterone-treated men and 233% in placebotreated men (P = 100).
In this RCT of men with T2D, lowered testosterone levels and high cardiovascular risk, we did not find evidence that testosterone treatment leads to an increase of NT-proBNP or hscTnT, biomarkers of subclinical cardiac injury and increased cardiovascular risk. In fact, a significant reduction in NT-proBNP was seen with treatment with no change in hs-cTnT. Although numbers were small, men who experienced cardiovascular events during testosterone treatment had increased baseline levels of NT-proBNP and hs-cTnT compared to men who were free of such events. As expected at baseline, 10-year UKPDS CHD risk was positively associated with the cardiac biomarkers and inversely associated with testosterone levels, but there was no association between testosterone and cardiac biomarkers.
Table 2. Main outcome measures Parameter
Testosterone group median [IQR]
Placebo group median [IQR]
NT-proBNP, ng/l 0 weeks 18 weeks 40 weeks hs-cTnT, ng/l 0 weeks 18 weeks 40 weeks
n = 43 600 [400, 815] 460 [392, 640] 540 [380, 880] n = 43 70 [60, 110] 80 [60, 120] 80 [70, 120]
n = 43 480 [395, 725] 535 [418, 860] 510 [420, 790] n = 43 90 [65, 130] 95 [70, 130] 100 [75, 120]
Mean adjusted difference† [95% CI]
179 [ 324,
35]
+041 [ 056, 139]
P-Value*
0047
062
*The P-value refers to overall significance of the change between groups during follow-up. †Mean adjusted difference refers to the change over 40 weeks across groups (mixed model). © 2015 John Wiley & Sons Ltd Clinical Endocrinology (2016), 84, 55–62
Testosterone and cardiac biomarkers 59 (a)
(b) 200
Change in NT proBNP (ng/l)
Change in NT proBNP (ng/l)
200
100
0
10
0
10
100
0
200
20
Change in testosterone (nM) (c)
200
400
600
(d) 4
Change in hs cTNT (ng/l)
4
Change in hs cTNT (ng/l)
0
Change in free testosterone (pM)
2
0
2
10
0
10
20
Change in testosterone (nM)
2
0
2
200
0
200
400
600
Change in free testosterone (pM)
Fig. 1 Association of the change in serum testosterone with the change in cardiac biomarkers. The change in the cardiac biomarkers observed in the whole cohort over the trial duration is plotted vs the change in serum testosterone. (a), Change in NT-proBNP vs change in total testosterone, s = 017, P = 0032. (b), Change in NT-proBNP vs change in free testosterone, s = 017, P = 0033. (c), Change in hs-cTNT vs change in total testosterone, s = 013, P = 012. (d) Change in hs-cTNT vs change in free testosterone, s = 013, P = 013. As requirements for a linear model or Pearson’s correlation were not met, regression lines and r coefficients are not depicted in the figure, rather correlations given are based on Kendall’s tau rank correlation.
To our knowledge, this is the first randomized controlled trial of testosterone treatment in men with diabetes to assess the effect of treatment on both NT-proBNP and hs-cTnT. There is limited data regarding the effect of testosterone treatment on natriuretic peptides, and we are not aware of any data regarding hs-cTnT. NT-proBNP is a peptide secreted by the myocardium in response to volume expansion. Given clinical observations that testosterone treatment, especially if given at supraphysiological doses to older men, can promote salt and water retention, the decrease in NT-proBNP in testosterone-treated men observed in this RCT may at first appear somewhat counterintuitive. However, there is a dearth of published data regarding the effects of testosterone on salt and water homoeostasis. Small RCTs have even demonstrated improved functional outcomes in men with heart failure,17,18 although natriuretic peptides were not measured in these studies. Our findings that testosterone decreases NTproBNP are consistent with observational studies showing that © 2015 John Wiley & Sons Ltd Clinical Endocrinology (2016), 84, 55–62
circulating natriuretic peptides are 60–80% higher in women compared to men.19 In an open-label randomized 2-week study of 12 hypopituitary men, testosterone treatment increased extracellular water, although plasma atrial natriuretic peptide (ANP) levels did not change.20 ANP levels increase with myocardial stretch, but given its short half-life and lack of a robust commercial assay, it is considered a less reliable analyte for measurement than NTproBNP.21 In a secondary analysis of the testosterone in older men (TOM) trial which found an increase of cardiovascular-related events with testosterone treatment in relatively frail, mobility-limited men,22 NT-proBNP levels did not change significantly.23 In the TOM trial, NT-proBNP levels were measured at baseline and at 6 months.23 In contrast in our study, we assessed the mean adjusted difference across the testosterone and placebo groups, based on three NT-proBNP measurements per participant, over a treatment duration of 40 weeks. In accordance with our findings, in a randomized trial of men with
60 E. J. Gianatti et al. Table 3. Baseline NT-proBNP and hs-cTnT levels in men with cardiovascular events
Intervention
Time of event (Study week)
Testosterone
30
Testosterone
18
Testosterone
2
Placebo
30
Placebo
18
Placebo
18
Nature of event Sepsis-related AMI requiring CABG New onset clinical CCF responding to medical therapy New onset clinical CCF requiring CABG/AVR New onset IHD/CCF requiring CABG Palpitations, no arrhythmia found Uncontrolled hypertension, developed swelling of ankles
Baseline NT-proBNP (ng/l)
Baseline hs-cTnT (ng/l)
140
13
730
25
1460
38
165
16
38
8
111
10
AMI, acute myocardial infarction; CABG, coronary bypass graft; CCF, congestive cardiac failure; AVR, aortic valve replacement; IHD, ischaemic heart disease.
nonmetastatic prostate cancer, androgen deprivation was associated with an increase in NT-proBNP levels.24 Experimental animal studies are also consistent with the notion that androgens may exert an inhibitory effect on natriuretic peptides: orchiectomized rats had an increase in plasma ANP which was normalized by testosterone replacement.25 Overall, the current data raise the possibility that testosterone promotes salt and water retention by inhibiting the release of natriuretic peptides, although not all data concurs, and further study is required. An alternative explanation, also consistent with the current data and with RCTs of testosterone treatment reporting improvements in outcomes in men with heart failure,17,18 is the possibility that testosterone could improve myocardial function, secondarily decreasing NTproBNP. Interestingly, all three men who experienced cardiovascular events during testosterone treatment had increased baseline NTproBNP and hs-cTnT levels prior to the commencement of testosterone treatment, indicative of pre-existing cardiac disease (Table 3). In the TOM trial, changes in NT-proBNP during treatment were not associated with testosterone treatment-associated cardiovascular events, although the spectrum of events was broad.23 Clearly, the relationship of exogenous testosterone with cardiac risk will need to be better defined in adequately designed clinical trials. Whether measurement of cardiac markers may be useful to better assess cardiac risk in selected men considered for testosterone treatment requires further study.
Troponin is a relatively specific marker of cardiac damage released into the circulation following cardiac injury. In this RCT, testosterone treatment did not change hs-cTnT, and therefore, our study provides no evidence that testosterone causes subclinical injury as reflected by hs-cTnT. Given that the 95% CI in our analysis was 056 to + 139 ng/dl, our trial makes it unlikely (
doi: 10.1111/cen.12842
ORIGINAL ARTICLE
Effect of testosterone treatment on cardiac biomarkers in a randomized controlled trial of men with type 2 diabetes Emily J. Gianatti*,†, Rudolf Hoermann*, Que Lam‡, Philippe Dupuis*,†, Jeffrey D. Zajac*,† and Mathis Grossmann*,† *Department of Medicine Austin Health, Austin Health, University of Melbourne, †Endocrine Unit, Austin Health, University of Melbourne, and ‡Department of Biochemistry, Austin Health, University of Melbourne, Heidelberg, Vic., Australia
Summary Objective To assess the effect of testosterone treatment on cardiac biomarkers in men with type 2 diabetes (T2D). Design Randomized double-blind, parallel, placebo-controlled trial. Patients Men aged 35–70 years with T2D and a total testosterone level ≤120 nmol/l (346 ng/dl) at high risk of cardiovascular events, median 10-year United Kingdom Prospective Diabetes Study (UKPDS) coronary heart disease (CHD) risk 21% (IQR 16%, 27%). Eighty-eight participants were randomly assigned to 40 weeks of intramuscular testosterone undecanoate (n = 45) or matching placebo (n = 43). Main Outcome Measures N-terminal pro B-type natriuretic peptide (NT-proBNP) and high-sensitivity cardiac troponin T (hs-cTnT). Result Testosterone treatment reduced NT-proBNP (mean adjusted difference (MAD) in change over 40 weeks across the testosterone and placebo groups, 179 ng/l [95% CI 324, 35], P = 0047), but did not change hs-cTnT (MAD, 041 ng/l (95% CI 056, 139), P = 062). Six men, three in each group experienced an adverse cardiac event, displaying already higher baseline NT-proBNP (P < 001) and hs-cTnT levels (P = 001). At baseline, 10-year UKPDS CHD risk was associated positively with NT-proBNP (s = 021, P = 0004) and hs-cTnT (s = 023, P = 0003) and inversely with testosterone (total testosterone s = 018, P = 002, calculated free testosterone s = 019, P = 001), but there was no significant association between testosterone and cardiac biomarkers (P > 005). Conclusions In this trial of men with T2D and high cardiovascular risk, testosterone treatment reduced NT-proBNP and did
Correspondence: Mathis Grossmann, Department of Medicine Austin Health, The University of Melbourne, 145 Studley Road, Heidelberg, Vic., 3084, Australia. Tel.: +613 9496 5000; Fax: +613 9496 3365; E-mail [email protected] Trial Registration Number: ClinicalTrials.gov NCT00613782. © 2015 John Wiley & Sons Ltd
not change hs-cTnT. Further studies should determine whether men with increased cardiac biomarkers prior to testosterone therapy are at higher risk of testosterone treatment-associated adverse cardiac events. (Received 27 January 2015; returned for revision 3 May 2015; finally revised 10 May 2015; accepted 23 June 2015)
Introduction Despite a wealth of preclinical and observational data, the effects of testosterone on the male cardiovascular system remain uncertain.1,2 This is because there have been no randomized controlled clinical trials (RCTs) of testosterone treatment adequately designed or sufficiently powered to evaluate cardiovascular outcomes. Meta-analyses of existing RCTs of testosterone therapy have been inconclusive and are limited by low numbers of adverse cardiac events.3–5 Testosterone prescriptions for older men have risen markedly in recent years.6 Cardiovascular disease is a leading cause of death in such men, especially in those with coexisting type 2 diabetes (T2D).7 Therefore, a better understanding of the effects of testosterone treatment on cardiovascular health in men with T2D is important. Several cardiac biomarkers such as troponin and brain natriuretic peptides correlate with subclinical cardiac injury in asymptomatic patients and provide cardiac risk stratification beyond conventional risk factors.8 Recently, it has been shown that both N-terminal pro B-type natriuretic peptide (NTproBNP) and high-sensitivity cardiac troponin T (hs-cTnT) improve the accuracy with which the risk of cardiovascular events can be predicted in patients with T2D.9 In this exploratory analysis, we assessed the effects of testosterone treatment on circulating NT-proBNP and hs-cTnT levels in an RCT of men with T2D and lowered testosterone levels at high risk of cardiovascular events. We have previously reported the effects of testosterone treatment on glucose metabolism,10 the primary outcomes of this trial and on constitutional and sexual symptoms11 in this population. 55
56 E. J. Gianatti et al.
Material and methods Design overview The design of this trial is reported in detail elsewhere.10 Briefly, this 40-week, randomized, double-blind, placebo-controlled trial (ClinicalTrials.gov NCT00613782) was conducted at a tertiary referral centre (Austin Health, Melbourne, Australia) between November 2009 and February 2013 and was approved by the Human Research Ethics Committee, Austin Health. Each participant provided written informed consent prior to entering the study. Men aged 35–70 years of age were eligible to participate in this trial if they had a history of T2D and the total testosterone level (averaged from two fasting morning specimens) was ≤120 nmol/l (346 ng/dl), as measured by an electrochemiluminescence immunoassay (ECLIA).10 Although total testosterone was measured by both ECLIA and liquid chromatography–tandem mass spectroscopy (LCMS/MS), recruitment was based on ECLIA because the LCMS/MS assay was not available for routine clinical use. Therefore, samples were batched and measured by LCMS/MS at study end.10 Exclusion criteria were as described previously10,11 and included uncontrolled hypertension (>160/90 mmHg despite treatment) and cardiac insufficiency (New York Heart Association score >2). Eligible participants were randomly assigned in a concealed 1:1 allocation to either intramuscular testosterone undecanoate 1000 mg or a visually identical placebo injection at 0, 6, 18 and 30 weeks.10 Trial investigators and participants were blinded to intervention allocation. All subjects received written, uniform recommendations regarding diet and physical activity. Outcomes and follow-up The main outcome measures of the RCT were changes in glucose metabolism.10 The main outcomes of this exploratory analysis were changes in the cardiac biomarkers NT-proBNP and hs-cTnT. Assessments and measurements At 0, 18 and 40 weeks, the following relevant variables were assessed: drug treatment, adverse events, body mass index (BMI), waist circumference, blood pressure, calculated 10-year United Kingdom Prospective Diabetes Study (UKPDS) coronary heart disease (CHD) risk,12 HbA1c, lipid profile, trough total testosterone (TT), calculated free testosterone (cFT) and NTproBNP and hs-cTnT levels. All blood samples were collected in early morning, fasted state prior to trial medication administration. Changes in antihypertensive medications were allowed but recorded at each visit. TT was measured by both ECLIA and liquid chromatography–tandem mass spectroscopy (LCMS/MS), and cFT was calculated as described.10,13 Unless specified otherwise, all analyses presented are based on TT measured by a validated LCMS/MS.13 NT-proBNP and hs-cTnT were measured at the conclusion of the trial from serum stored at 70°C collected at 0, 18 and
40 weeks by ECLIA on a E602 automated platform (Roche Diagnostics, Burgess Hill, UK). Assays were performed using the manufacturer’s calibrators and quality controls. The NT-proBNP assay has an effective measuring range of 5–30 000 ng/l, and NT-proBNP levels 4mcg/l, or haematocrit >050 (n = 61). Two men who were randomized to the testosterone group had markedly elevated baseline NTproBNP levels suggestive of occult congestive cardiac failure. Both developed overt clinical congestive cardiac failure early in the trial. Data from these men were excluded, leaving 86 men in the analysis. Seventy-five men completed the trial. The most common reason for noncompletion was a predefined protocol violation of intensification of oral hypoglycaemic agents or commencement of insulin therapy, and there was no difference between the treatment groups in this regard.10 Baseline characteristics of the 86 participants are shown in Table 1. At baseline, cardiovascular risk as defined by 10-year UKPDS CHD risk was high, median 21% (IQR 16%, 27%), and 965% of men had prevalent metabolic syndrome. 116% of men had elevated NT-proBNP (≥125 ng/l). All men in the trial had detectable hs-cTnT with 140% having elevated levels ≥ = 14 ng/ l (Table 1). There was no significant difference in baseline NTproBNP (P = 023) or hs-cTnT (P = 006) between testosteroneand placebo-treated participants. At baseline, all men were assessed to be New York Heart Association (NYHA) 1 (no evidence of cardiac failure) and free of clinical symptoms suggestive of cardiovascular ischaemia. Testosterone levels Baseline TT measured by LCMS/MS was 106 nmol/l (306 ng/ dl) and 110 nmol/l (317 ng/dl) in the testosterone and placebo groups, respectively (P = 086), and TT measured by ECLIA was 87 nmol/l (251 ng/dl) and 85 nmol/l (245 ng/dl) (P = 063) (Table 1). At 40 weeks, testosterone levels increased significantly © 2015 John Wiley & Sons Ltd Clinical Endocrinology (2016), 84, 55–62
Table 1. Baseline characteristics of the participants Testosterone N = 43 Age, y Duration of diabetes, y Insulin therapy, n (%) Metabolic syndrome (ATPIII criteria), n (%) Prior cardiovascular disease, n (%) Smoking status Current smoker, n (%) Exsmoker, n (%) Nonsmoker, n (%) 10-year UKPDS CHD risk, % BMI, kg/m2 Waist circumference, cm Systolic blood pressure, mmHg Diastolic blood pressure, mmHg Total testosterone (ECLIA), nmol/l Total testosterone (LCMS/MS), nmol/l Calculated free testosterone (LCMS/MS), pmol/l eGFR, ml /min/ 173 m2 HbAlc, % Cholesterol, mmol/l LDL, mmol/l HDL, mmol/l TG, mmol/l NT-proBNP, ng/l hs-cTnT, ng/l
62 [58, 68] 7 [4, 11]
Placebo N = 43
P-value
62 [58, 67] 9 [5, 12]
082 049
6 (140)
10 (233)
027
42 (978)
41 (953)
056
12 (279)
5 (116)
010
4 (93)
3 (70)
090
23 (535) 16 (372) 20 [15, 28]
25 (581) 15 (349) 22 [17, 27]
308 [283, 348] 1100 [1040, 1200]
334 [314, 354] 1150 [1100, 1210]
004 006
140 [130, 150]
140 [129, 150]
095
72 [70, 80]
80 [70, 82]
004
87 [71, 111]
85 [73, 106]
063
106 [90, 128]
110 [82, 133]
086
225 [186, 298]
247 [183, 314]
068
840 [740, 900] 68 [64, 74] 42 [38, 48] 23 11 16 600
[17, 27] [09, 13] [11, 24] [400, 815]
70 [60, 110]
034
86 [745, 900]
089
71 [67, 75] 45 [37, 48]
007 098
22 10 18 480
[18, 28] [08, 12] [13, 24] [395, 725]
90 [65, 130]
080 018 032 023 006
Values are given as median and [interquartile range] or as n and (%). P values were calculated for the difference between groups using Wilcoxon’s test or chi-square test. P < 005 was considered significant.
in the testosterone group, while there was no significant change in the placebo group. The mean adjusted differences (MAD) in change over 40 weeks across the two groups were TT 602 nmol/l (173 ng/dl) [95% CI 38, 83 (110, 239)], P < 0001 and cFT 181 pmol/l (53 pg/ml) [95% CI 108, 255 (31, 74)], P < 0008.
58 E. J. Gianatti et al. Changes in main outcome measures NT-proBNP decreased with mean adjusted difference (MAD) between groups across 40 weeks of 179 ng/l [95% CI 324, 35], P = 0047 (Table 2). NT-proBNP returned on average more frequently to below the cut off of 125 ng/l during the course of the trial in testosterone-treated men compared to placebo-treated men: MAD 5 [95% 1413, 129], P = 0027. Hs-cTnT was stable in both groups over the duration of the trial with MAD between groups across 40 weeks of 041 ng/l [95% CI 056, 139], P = 062. With respect to medications that may influence NT-proBNP levels, no participant commenced insulin during the study, and there was no between group difference in changes in total daily dose of insulin.10 At baseline, 21 men were receiving diuretic treatment, nine in the testosterone group and 12 men in the placebo group (P = 045). The most commonly used type of diuretic was hydrochlorothiazide (n = 16), and other types indapamide (n = 4), amiloride (n = 1), spironolactone (n = 1) and frusemide (n = 1), with one participant in each group taking two different diuretics. Importantly, of all men receiving diuretic treatment, only two men in the testosterone group and two men in the placebo group had minor adjustment to their diuretic dose, and no participant commenced a diuretic during the study.
Baseline associations between cardiac biomarkers, cardiovascular risk and testosterone levels At baseline in the whole cohort, 10-year UKPDS CHD risk was positively associated with both baseline NT-proBNP (s = 021 P = 00043) and hs-cTnT (s = 023 P = 00031) and negatively with baseline testosterone level: TT LCMS s = 018 P = 0017, cFT LCMS s = 019 P = 001. At baseline, neither NT-proBNP nor hs-cTnT was associated with testosterone levels: NTproBNP: TT s = 0034 (P = 065), cFT s = 0053 (P = 047), hs-cTnT: TT s = 0051 (P = 050), cFT s = 0073 (P = 034). The change in serum testosterone observed in the whole cohort over the trial duration was associated with an inverse change in NT-proBNP: delta TT LCMS/MS s = 017 (P = 0032), cFT LCMS/MS s = 017 (P = 0033), while no significant association was observed with change in hs-cTnT (Fig. 1). Adverse cardiovascular events Six cardiovascular events occurred during the trial10 among the 88 randomized participants, associated with higher baseline NTproBNP (P < 001) and higher baseline hs-cTnT (P = 001). Details are given in Table 3.
Discussion Changes in cardiovascular risk factors There was no significant between group change in BMI, blood pressure or HbA1c,10 or in 10-year UKPDS CHD risk (P = 023). As reported previously,10 testosterone treatment decreased total cholesterol (MAD 045 mmol/l [95% CI 07, 02], P < 0001), LDL cholesterol (MAD 026 mmol/l [95% CI 046, 006], P = 001) and HDL (MAD 011 mmol/l [95% CI 019, 003], P = 0002), but there was no significant between group change in CRP. There was no significant between-group difference in the proportion of participants who had changes in their antihypertensive medications during the study, 256% in testosterone-treated men and 233% in placebotreated men (P = 100).
In this RCT of men with T2D, lowered testosterone levels and high cardiovascular risk, we did not find evidence that testosterone treatment leads to an increase of NT-proBNP or hscTnT, biomarkers of subclinical cardiac injury and increased cardiovascular risk. In fact, a significant reduction in NT-proBNP was seen with treatment with no change in hs-cTnT. Although numbers were small, men who experienced cardiovascular events during testosterone treatment had increased baseline levels of NT-proBNP and hs-cTnT compared to men who were free of such events. As expected at baseline, 10-year UKPDS CHD risk was positively associated with the cardiac biomarkers and inversely associated with testosterone levels, but there was no association between testosterone and cardiac biomarkers.
Table 2. Main outcome measures Parameter
Testosterone group median [IQR]
Placebo group median [IQR]
NT-proBNP, ng/l 0 weeks 18 weeks 40 weeks hs-cTnT, ng/l 0 weeks 18 weeks 40 weeks
n = 43 600 [400, 815] 460 [392, 640] 540 [380, 880] n = 43 70 [60, 110] 80 [60, 120] 80 [70, 120]
n = 43 480 [395, 725] 535 [418, 860] 510 [420, 790] n = 43 90 [65, 130] 95 [70, 130] 100 [75, 120]
Mean adjusted difference† [95% CI]
179 [ 324,
35]
+041 [ 056, 139]
P-Value*
0047
062
*The P-value refers to overall significance of the change between groups during follow-up. †Mean adjusted difference refers to the change over 40 weeks across groups (mixed model). © 2015 John Wiley & Sons Ltd Clinical Endocrinology (2016), 84, 55–62
Testosterone and cardiac biomarkers 59 (a)
(b) 200
Change in NT proBNP (ng/l)
Change in NT proBNP (ng/l)
200
100
0
10
0
10
100
0
200
20
Change in testosterone (nM) (c)
200
400
600
(d) 4
Change in hs cTNT (ng/l)
4
Change in hs cTNT (ng/l)
0
Change in free testosterone (pM)
2
0
2
10
0
10
20
Change in testosterone (nM)
2
0
2
200
0
200
400
600
Change in free testosterone (pM)
Fig. 1 Association of the change in serum testosterone with the change in cardiac biomarkers. The change in the cardiac biomarkers observed in the whole cohort over the trial duration is plotted vs the change in serum testosterone. (a), Change in NT-proBNP vs change in total testosterone, s = 017, P = 0032. (b), Change in NT-proBNP vs change in free testosterone, s = 017, P = 0033. (c), Change in hs-cTNT vs change in total testosterone, s = 013, P = 012. (d) Change in hs-cTNT vs change in free testosterone, s = 013, P = 013. As requirements for a linear model or Pearson’s correlation were not met, regression lines and r coefficients are not depicted in the figure, rather correlations given are based on Kendall’s tau rank correlation.
To our knowledge, this is the first randomized controlled trial of testosterone treatment in men with diabetes to assess the effect of treatment on both NT-proBNP and hs-cTnT. There is limited data regarding the effect of testosterone treatment on natriuretic peptides, and we are not aware of any data regarding hs-cTnT. NT-proBNP is a peptide secreted by the myocardium in response to volume expansion. Given clinical observations that testosterone treatment, especially if given at supraphysiological doses to older men, can promote salt and water retention, the decrease in NT-proBNP in testosterone-treated men observed in this RCT may at first appear somewhat counterintuitive. However, there is a dearth of published data regarding the effects of testosterone on salt and water homoeostasis. Small RCTs have even demonstrated improved functional outcomes in men with heart failure,17,18 although natriuretic peptides were not measured in these studies. Our findings that testosterone decreases NTproBNP are consistent with observational studies showing that © 2015 John Wiley & Sons Ltd Clinical Endocrinology (2016), 84, 55–62
circulating natriuretic peptides are 60–80% higher in women compared to men.19 In an open-label randomized 2-week study of 12 hypopituitary men, testosterone treatment increased extracellular water, although plasma atrial natriuretic peptide (ANP) levels did not change.20 ANP levels increase with myocardial stretch, but given its short half-life and lack of a robust commercial assay, it is considered a less reliable analyte for measurement than NTproBNP.21 In a secondary analysis of the testosterone in older men (TOM) trial which found an increase of cardiovascular-related events with testosterone treatment in relatively frail, mobility-limited men,22 NT-proBNP levels did not change significantly.23 In the TOM trial, NT-proBNP levels were measured at baseline and at 6 months.23 In contrast in our study, we assessed the mean adjusted difference across the testosterone and placebo groups, based on three NT-proBNP measurements per participant, over a treatment duration of 40 weeks. In accordance with our findings, in a randomized trial of men with
60 E. J. Gianatti et al. Table 3. Baseline NT-proBNP and hs-cTnT levels in men with cardiovascular events
Intervention
Time of event (Study week)
Testosterone
30
Testosterone
18
Testosterone
2
Placebo
30
Placebo
18
Placebo
18
Nature of event Sepsis-related AMI requiring CABG New onset clinical CCF responding to medical therapy New onset clinical CCF requiring CABG/AVR New onset IHD/CCF requiring CABG Palpitations, no arrhythmia found Uncontrolled hypertension, developed swelling of ankles
Baseline NT-proBNP (ng/l)
Baseline hs-cTnT (ng/l)
140
13
730
25
1460
38
165
16
38
8
111
10
AMI, acute myocardial infarction; CABG, coronary bypass graft; CCF, congestive cardiac failure; AVR, aortic valve replacement; IHD, ischaemic heart disease.
nonmetastatic prostate cancer, androgen deprivation was associated with an increase in NT-proBNP levels.24 Experimental animal studies are also consistent with the notion that androgens may exert an inhibitory effect on natriuretic peptides: orchiectomized rats had an increase in plasma ANP which was normalized by testosterone replacement.25 Overall, the current data raise the possibility that testosterone promotes salt and water retention by inhibiting the release of natriuretic peptides, although not all data concurs, and further study is required. An alternative explanation, also consistent with the current data and with RCTs of testosterone treatment reporting improvements in outcomes in men with heart failure,17,18 is the possibility that testosterone could improve myocardial function, secondarily decreasing NTproBNP. Interestingly, all three men who experienced cardiovascular events during testosterone treatment had increased baseline NTproBNP and hs-cTnT levels prior to the commencement of testosterone treatment, indicative of pre-existing cardiac disease (Table 3). In the TOM trial, changes in NT-proBNP during treatment were not associated with testosterone treatment-associated cardiovascular events, although the spectrum of events was broad.23 Clearly, the relationship of exogenous testosterone with cardiac risk will need to be better defined in adequately designed clinical trials. Whether measurement of cardiac markers may be useful to better assess cardiac risk in selected men considered for testosterone treatment requires further study.
Troponin is a relatively specific marker of cardiac damage released into the circulation following cardiac injury. In this RCT, testosterone treatment did not change hs-cTnT, and therefore, our study provides no evidence that testosterone causes subclinical injury as reflected by hs-cTnT. Given that the 95% CI in our analysis was 056 to + 139 ng/dl, our trial makes it unlikely (
