Circulating testosterone and estradiol, autonomic balance and baroreflex sensitivity in middle-aged and elderly men with heart failure

Abstract

Background: Heart failure (HF) is considered as a cardiogeriatric syndrome. Its fundamental pathophysiological feature is autonomic imbalance (and associated abnormalities within cardiovascular reflex control), but recent evidence suggests the involvement of deranged hormone metabolism. Both these neural and endocrine pathologies have serious clinical and prognostic consequences in patients with HF. We investigated the relations between autonomic status, baroreflex sensitivity (BRS) and hormone status in men with mild systolic HF.

Methods: We examined 46 men with stable systolic HF (age: 62 ± 10 years, NYHA class I/II: 10/36 [22%/78%], ischemic aetiology: 72%, left ventricular ejection fraction: 32 ± 8%). Serum hormone levels (i.e. total testosterone [TT], dehydroepiandrosterone sulphate [DHEAS], oestradiol [E2], insulin-like growth factor type 1 [IGF-1] and cortisol) were assessed using immunoassays. Estimated free testosterone (eFT) was estimated using the Vermeulen’s equation. Heart rate variability (HRV) was assessed in time and frequency domains, based on 10-min resting recordings. BRS was estimated using the sequence method (BRS-Seq) and the phenylephrine test (BRS-Phe).

Results: Deficiencies in circulating TT, eFT, DHEAS and IGF-1 (defined as a serum hormone ≤the 10th percentile calculated for the adequate age category in the cohort of healthy men) were found in respectively 13%, 30%, 55% and 93% of men with systolic HF. Serum SHBG ≥50 nmol/L and cortisol ≥700 nmol/L characterised, respectively 44% and 29% of men with HF. In multivariable models after the adjustment for clinical variables, the following relationships were found in examined men: DHEAS and SDNN (time domain of HRV defined as a standard deviation of average R–R intervals) (β = 0.29, p = 0.03); E2 and: HRV-LF (ms²) (β = 0.37, p = 0.01), HRV-HF (ms²) (β = 0.44, p = 0.02) and BRS-Phe (β = 0.51, p = 0.008); TT and: HRV-HF (%) (β = 0.35, p = 0.02), HRV-LF/HF ratio (β = −0.35, p = 0.02) and BRS-Seq (β = 0.33, p = 0.04).

Conclusions: The observed associations between reduced circulating androgens, oestrogens and lower HRV and depleted BRS, irrespectively of HF severity suggest the pathophysiological links between these two mechanisms. These results constitute the premises to investigate whether the pharmacological supplementation of depleted hormones would enable to restore the autonomic balance and improve the efficacy of reflex control within the cardiovascular system in men with systolic HF.

• Autonomic imbalance
• hormonal status
• heart failure

Introduction

Endocrine and autonomic nervous systems are the two major systems which are responsible for the synchronisation of functioning of all other systems and organs in the body [1,2]. Hence, it is not surprising that these two important systems within the human organism are closely, structurally and functionally related to each other [3]. Neurons within autonomic centres and their afferents express receptors of numerous hormones, including androgens, oestrogens, glucosterocorticoids and mineralocorticoids. On the other hand, some endocrine glands (e.g. hypophysis and hypothalamus) are considered also as the elements of central nervous system [4].

The efficacy of endocrine and nervous systems deteriorates with age [5,6], and not surprisingly these abnormalities are involved in the pathophysiology of diseases related to aging, e.g. heart failure (HF). Crucial elements of the natural history of HF are autonomic imbalance [7] and endocrine disorders [8,9]. Importantly, these phenomena occur at the early stage of HF, and have numerous clinical and prognostic consequences [10,11].

So far, the autonomic status and hormone derangements occurring in patients with HF have been analysed in isolation. In the present study, we investigated the relations between autonomic balance, BRS and hormone status in men with mild systolic HF.

Methods

Study population

The examined men were recruited from middle-aged patients with systolic HF attending the outpatient HF clinic in Centre for Heart Diseases, Military Hospital (Wroclaw, Poland).

The inclusion criteria for the study were as follows: (a) men aged ≥50 years; (b) a >6-month documented history of HF (in I–II classes according to New York Heart Association, NYHA); (c) clinical stability with unchanged medications for ≥3 months preceding the study; (d) left ventricular ejection fraction (LVEF) <45% as assessed by echocardiography.

The exclusion criteria comprised: (a) HF decompensation within three months preceding the study; (b) acute coronary syndrome and/or coronary revascularization during six months preceding the study; (c) atrial fibrillation, pacemaker rhythm and/or frequent ectopics; (d) any hormonal abnormalities and associated therapy either at the time of examination or in the past.

The study was approved by the local ethics committee. All subjects gave written informed consent. The study was conducted in accordance with the Helsinki Declaration.

Study protocol

Each patient underwent a physical examination with routine laboratory tests and standard transthoracic echocardiography.

In all patients, venous blood samples were taken in the morning (8.00–10.00 am) after an overnight fasting and after a supine rest of ≥15 min. After centrifugation, the serum was aliquoted and frozen at −70 °C until being analyzed.

Assessments of serum hormone levels

In the present study, we have analysed serum levels of three major anabolic hormones (testosterone [total and free fractions], dehydroepiandrosterone sulphate [DHEAS] and insulin-like growth factor [IGF-1]) and one catabolic hormone (cortisol), all of which are involved in the progression of HF and predict prognosis [10]. Additionally, we have assessed serum levels of oestradiol in the view of a well-established important role of oestrogens in men with HF [12,13].

The assessment of serum levels of total testosterone (TT, ng/mL), DHEAS (ng/mL), E₂ (pg/mL), sex hormone binding globulin (SHBG, nmol/L) and cortisol (nmol/L) was based on electrochemiluminescence (Elecsys 2010, Roche Diagnostics GmbH, Mannheim, Germany). Serum IGF-1 (ng/mL) was assessed using immunochemiluminescence (Immulite 2000/2500, Diagnostic Products Corporation, San Francisco, CA). The inter- and intraassay variability coefficients for TT, DHEAS, E₂, SHBG, cortisol and IGF-1 were as follows: 3% and 4%, 2% and 4%, 5% and 4%, 2% and 4%, 1% and 2%, 3% and 6%, respectively. The estimated free testosterone (eFT) was calculated using the Vermeulen’s equation [14].

TT, eFT, DHEAS and IGF-1 deficiencies were defined prospectively as a serum hormone ≤the 10th percentile calculated for the adequate age category in the cohort of healthy men [15]. The values of the 10th percentiles of serum TT, eFT, DHEAS and IGF-1 were as follows: 2.90, 3.00 and 2.60 ng/mL; 53, 59 and 59 pg/mL; 863, 614 and 334 ng/mL; 218, 203 and 164 ng/mL, respectively [15].

According to DeGroot and Jameson [16], E₂ excess and E₂ deficiency in men were defined as serum E₂ ≥50 pg/mL and serum E₂ ≤10 pg/mL, respectively. The following values of serum SHBG and cortisol were used as cut-offs interpreted as the excess of these hormones: ≥50 nmol/L and ≥700 nmol/L, respectively [16].

Plasma N-terminal pro-B type natriuretic peptide (NT-proBNP, pg/mL) was measured using electrochemiluminescence on the Elecsys 1010/2010 System (Roche Diagnostics GmbH, Mannheim, Germany). Renal function was assessed based on the estimated glomerular filtration rate (GFR; mL/min/1.73m²), calculated from the Modification of Diet in Renal Disease equation [17]. Serum high-sensitive C-reactive protein (hsCRP, mg/L) was assessed using immunonephelometry (Dade Behring, Marburg GmbH, Germany).

Assessments of autonomic status and arterial baroreflex sensitivity

The assessments of autonomic imbalance were conducted in concordance with the ESC guidelines [18] and preceded by a ≥20-min resting in horizontal position. Afterwards, a digital non-invasive recording of finger arterial pressure synchronized with ECG (R–R intervals, [ms]) was acquired continuously using NEXFIN HD device (BmEye, Amsterdam, the Netherlands) during resting (30 min) and during the whole phenylephrine test (approx. 30 min). Systolic, diastolic and mean blood pressure (SBP, DBP and MAP, respectively, [mmHg]) as well as haemodynamic parameters were derived from NEXFIN using a waveform analysis, based on pressure pulse contour model [19,20].

The autonomic status was assessed based on indices of heart rate variability (HRV) (calculated from 10-min recordings of an acceptable quality which were selected from the whole resting recordings):

1. time domains (meanRR – mean duration of R–R intervals [ms]; SDNN [ms] – standard deviation of average R–R intervals, RMSSD [ms] – the square root of the mean of the sum of the squares of differences between adjacent R–R intervals) [18,21]

2. frequency domains (spectral components were distinguished in HRV spectrum: low [LF], 0.04 to 0.15 Hz; and high [HF], 0.15 to 0.4 Hz frequencies expressed in absolute values [ms²] and as a LF/HF ratio in order to describe the balance between sympathetic and parasympathetic drive within autonomic nervous system) [18,21]. Spectral analysis of HRV was performed using standard autoregressive methods [22].

The high-frequency component of HRV is interpreted as an index of efferent vagal activity [23], which is usually depleted in the course of HF [24]. LF component is interpreted as a marker of sympathetic modulation [23], which typically is increased in patients with HF, particularly in those not treated with β-blockers [25].

The values of HRV were compared to the reference values calculated for healthy adult males examined in our physiological laboratory [26].

BRS was assessed using:

1. the sequence method (based on the selection of three sequences of changes in systolic BP (with a difference of ≥1.0 mmHg) accompanied by changes in R–R interval (with a difference of ≥5.0 ms) which occurred most frequently within the whole analysed fragment. BRS-Seq was interpreted as an average of all regression slopes relating BP to R–R intervals (ms/mmHg) [27,28].

2. the phenylephrine method (consisted of three injections of efficacious dose of phenylephrine (resulting in an increase in systolic BP ≥15 mmHg). If an increase did not occur with the initial dose (2 μg/kg), the dose of phenylephrine was increased by 2 μg/kg in each of the next injections (with an interval of >5 min) until the required systolic BP increase appeared. For the final analysis, three recordings of the BP rise after injections with an optimal dose of phenylephrine (still separated by >5-min intervals) were selected. BRS was calculated as the linear regressions of HR and systolic BP including all of the points between the beginning and end of the increase in systolic BP. The final slope is the mean of the three linear regressions coefficients expressed in ms/mmHg [29,30].

Despite the methodological differences, spontaneous (BRS-Seq) and pharmacological (BRS-Phe) BRS values are significantly related to each other [31]. BRS-Phe is based on the analysis of HR response to instant and acute SBP increase, whereas the physiological role of baroreceptors is to be sensitive primarily to an instant decrease of SBP, in order to maintain blood perfusion through critical organs (brain, heart) during, e.g. haemorrhage [32]. In the sequence method, the reactions to both SBP increase and decrease are analysed [33]. On the other hand spontaneous BRS (BRS-Seq) is based on lower SBP oscillations as compared to SBP increase induced rapidly by a phenylephrine injection [34,35]. That is why we have decided prospectively to apply two methods of BRS assessment with a slightly different physiological interpretation [31].

The values of BRS were compared to the reference values calculated for healthy adult males examined in our physiological laboratory [36].

It is difficult to discuss the reference values for HRV and BRS parameters, as the absolute values reflecting autonomic imbalance are strongly related to the methodology applied in particular laboratories. Therefore, in our opinion and also according to the experts, each laboratory should provide its normal values measured in a group of healthy controls. In the past we measured HRV and BRS in healthy subjects and published the results in two papers [26,36].

Statistical analyses

The results for each variable were tested for normality using the Kolmogorov–Smirnov method. Normally distributed continuous variables were presented as means ± standard deviations. The inter-group differences were tested using the Student’s t-test. Variables with a skewed distribution were expressed as medians with lower and upper quartiles, and were log transformed (and proved to be normalized afterwards) before the parametric analyses were performed. The categorical variables were expressed as numbers with percentages. The inter-group differences were tested using the χ² test.

The relationships between serum hormone levels, and variables reflecting autonomic status and BRS were assessed in univariate regression models (Pearson’s correlatory coefficients), and if they were statistically significant, they were included in multivariate models with the adjustment for indices of HF severity (i.e. plasma NT-proBNP, NYHA class).

A value of p < 0.05 was considered statistically significant.

Results

The baseline clinical and laboratory data are presented in Table 1.

Table 1. Baseline characteristics of the examined men with mild systolic HF.

Variables	Men with HF (n = 46)
Age (years)	62 ± 10
BMI (kg/m^2)	28.9 ± 6.0
NYHA class (I/II) n, (%)	10/36 (22/78)
Aetiology (CAD) n, (%)	33 (72%)
Hypertension (yes)	33 (72%)
DM (yes)	15 (33%)
LVEF (%)	32 ± 8
Haemoglobin (g/dL)	14.4 ± 1.0
hsCRP (mg/L)	1.3 (0.9–3.1)
GFR (mL/min/1.73m^2)	69 ± 16
Na (mEq/L)	142 ± 3
NT-proBNP (pg/mL)	648 (257–946)
Total cholesterol (mg/dL)	192 ± 38
Uric acid, mg/dL	6.9 ± 1.7
Applied treatment
ACE inhibitor and/or ARB, n(%)	41 (89%)
Aldosterone antagonist, n(%)	7 (15%)
β-blocker, n(%)	45 (98%)
Loop diuretic, n(%)	20 (43%)

Data are presented as a mean ± standard deviation, medians with lower and upper quartiles or percentage where appropriate. HF, heart failure; BMI, body mass index; NYHA, New York Heart Association; CAD, coronary artery disease; DM, diabetes mellitus; LVEF, left ventricular ejection fraction; hsCRP, high-sensitivity C-reactive protein; GFR, glomerular filtration rate; NTproBNP, plasma N-terminal pro-B-type natriuretic peptide; ACE, angiotensin converting enzyme; ARB, angiotensin receptor blocker.

The medians (with lower and upper quartiles) of serum hormone levels along with the prevalence of hormonal abnormalities (hormone deficiencies and excesses) among examined men with HF are shown in Table 2. Deficiencies in circulating TT, eFT, DHEAS and IGF-1 were found in, respectively, 13%, 30%, 55% and 93% of men with systolic HF. Serum E2 was within the reference range (10–50 pg/mL) in all the examined men with HF. Excesses of SHBG and cortisol were found, respectively, in 44% and 29% of men with HF.

Table 2. Serum hormone levels and the prevalence of hormone deficiency and excess in examined men with mild systolic HF.

Variables	Patients with systolic HF (n = 46)	Hormone deficiency (%)	Hormone excess (%)
TT (ng/mL)	5.00 (4.30–5.80)	9	–
eFT (pg/mL)	71 (53–88)	30	–
DHEAS (pg/mL)	558 (96–1214)	47	–
IGF-1 (ng/mL)	114.5 (103.0–141.0)	91	–
SHBG (nmol/L)	46.3 (38.5–57.3)	–	38
E2 (pg/mL)	29.4 (25.1–31.8)	0	0
Cortisol (nmol/L)	613 (517–718)	–	30

Data are presented as a median with lower and upper quartiles and percentage where appropriate; HF – heart failure; TT – total testosterone; eFT – estimated free testosterone; DHEAS – dehydroepiandrosterone sulfate; IGF-1 – insulin-like growth factor type 1; SHBG – seh hormone binding globulin, E₂- oestradiol; Deficiencies of TT, eFT, DHEAS and IGF-1) were defined as ≤the 10th percentile in the corresponding age category in the reference population. SHBG ≥50 nmol/L and cortisol ≥700 nmol/L were interpreted as exceeded values of these hormones.

The mean values of the calculation of HRV indices and BRS assessments are presented in Table 3. There were no differences in HRV between the examined men with HF and the reference values [26], whereas values of BRS were lower as compared to the reference values [36] (both p < 0.05).

Table 3. HRV and BRS in examined men with mild systolic HF.

Variables	Mean ± SD	Median(Q1–Q3)
SDNN, ms	36.3 ± 20.2	33.2 (19.2–46.2)
RMSSD, ms	25.3 ± 15.4	24.4 (11.4–33.0)
HRV-VLF, ms^2	653 ± 813	320 (91–878)
HRV-LF, ms^2	207 ± 226	109 (41–296)
HRV-HF, ms^2	181 ± 247	72 (23–210)
HRV-LF/HRV-HF	2.10 ± 1.81	1.45 (0.76–2.35)
BRS-Seq, ms/mmHg	7.8 ± 4.2	6.6 (4.8–10.2)
BRS-Phe, ms/mmHg	4.8 ± 2.5	3.9 (2.9–6.7)

Data are presented as mean with standard deviation and median with higher and lower quartile. SD, standard deviation; q1, lower quartile; q3, upper quartile; SDNN, standard deviation of average R–R intervals; RMSSD, the square root of the mean of the sum of the squares of differences between adjacent R–R intervals; HRV, heart rate variability; VLF, very low frequency; LF, low frequency; HF, high frequency; BRS-Seq, baroreflex assessed using the sequence method; BRS-Phe, baroreflex assessed in the phenylephrine test.

Relations between serum hormone levels and clinical variables reflecting heart failure severity

Among the clinical variables reflecting the severity of HF, lower LVEF was related to a higher level of SHBG (r = 0.42. p = 0.004) and a higher level of NTproBNP was accompanied by a lower level of DHEAS (r = 0.37, p = 0.01).

There were no other correlations between serum hormone levels and clinical parameters (NYHA class, HF aetiology, etc.).

Relations between serum hormone levels and indices of heart rate variability

In univariate regression models in men with systolic HF, serum IGF-1 (ng/mL) was related to SDNN (ms) (r = 0.34 p = 0.02) and RMSSD (ms) (r = 0.37, p = 0.055).

E2 (pg/mL) was related to RMSSD (ms) (r = 0.67, p = 0.03) and HRV-HF (ms²) (r = 0.54, p = 0.006; Figure 1). DHEAS was related to HRF LF/HF ratio (r = −0.45 p = 0.03).

Figure 1. The relation between serum oestradiol and frequency domain of HRV (high frequency) in men with mild systolic HF (0.54, p = 0.006).

In multivariable regression models, after the adjustment for plasma NT-proBNP and NYHA class, the following relationships remained statistically significant: serum TT and HRV–LF/HF ratio (β = −0.35, p = 0.02), DHEAS and SDNN (ms) (β = 0.29, p = 0.03). Moreover, E2 was related to both HRV-LF (ms²) (β = 0.37, p = 0.01) and HRV-HF (ms²) (β = 0.44, p = 0.02).

Relations between serum hormone levels and baroreflex sensitivity

In univariate regression models in men with systolic HF, serum TT (ng/mL) was related to BRS-Seq (ms/mmHg) (r = 0.52, p = 0.006; Figure 2), whereas both serum DHEAS (ng/mL) and serum E2 (pg/mL) were related to BRS-Phe (ms/mmHg) (r = 0.48 p = 0.03 and r = 0.73, p = 0.001, respectively; Figure 3).

Figure 2. The relation between serum testosterone and BRS calculated using the sequence method in men with mild systolic HF (r = 0.52, p = 0.006).

Figure 3. The relation between serum oestradiol and BRS calculated using the phenylephrine method in men with mild systolic HF (0.73, p = 0.001).

In multivariable regression models, after the adjustment for plasma NT-proBNP and NYHA class, the following relationships remained statistically significant: serum TT (ng/mL) and BRS-Seq (ms/mmHg) (β = 0.33, p = 0.04).

Discussion

In our study, we have found that middle-aged and elderly men with stable mild systolic HF, receiving drugs antagonizing the sympathetic and the renin-angiotensin-aldosterone systems, still demonstrate at the early stage of heart disease the derangements within reflex mechanisms controlling cardiovascular system (i.e. diminished BRS as compared to the reference values [36]). Moreover, a substantial percentage of the examined men with HF reveals abnormalities within the levels of investigated hormones related to male aging. Importantly, there are several associations between hormone milieu (testosterone, estradiol, DHEAS, but not cortisol) and autonomic status in men with systolic HF.

A lack of significant reduction of HRV revealed in men with HF compared to the reference values is probably related to the treatment based on β-blockers which are able to restore the optimal HRV profile in patients with HF [25,26].

The observed relations between lower LVEF and higher SHBG as well as between lower level DHEAS and higher NT-proBNP are consistent with the previous findings suggesting that the levels of SHBG and DHEAS correlate with the measures of HF severity [37–39].

In our previous studies, we have demonstrated several unfavourable consequences of TT deficiency in men with systolic HF, including impaired exercise capacity [10,40], more severe depressive symptoms [40,41] and high mortality [10]. Other reports indicate that TT deficiency in patients with HF is related to insulin resistance and an increased risk of the development of diabetes [42,43], greater prevalence of anaemia [44], more augmented inflammation [45] and an increased mortality risk [37,46]. Patients with lower level of TT are often characterized by impaired renal function, greater prevalence of atrial fibrillation, lower blood pressure, and more often are treated with diuretics and cardiac glycosides [37].

Also, Wehr et al. [47] demonstrated that low circulating TT, measured in 2078 men with symptomatic HF (age: medians with interquartile ranges of age in I–IV free testosterone quartiles were 68[61–73], 64[58–70], 61[55–68] and 57[49–63]) referred for coronary angiography was accompanied by the higher NYHA classes, impaired LVEF function and most importantly in this group of patients, testosterone deficiency was associated with an increased mortality for congestive HF and myocardial infarction during the median follow-up of 7.7 years [47].

In the present analysis, we have found that TT deficiency is accompanied also by a decrease in parasympathetic tone and reduced BRS in men with mild HF, which is another example of negative effects of TT depletion in men with HF. This is in accordance with the results of the studies performed in hypogonadal men, which showed the analogous associations between the depleted serum TT and reduced HRV [48]. In a population of healthy men, both circulating TT and autonomic function decrease with age, but the relation between TT and autonomic status remains independent of age and anthropometric factors [21,49,50]. Our observational data showing associations between the level of TT and BRS are in accordance with the interventional study performed by Caminiti et al. [51] where they demonstrated in a group of 70 patients with chronic HF, who were older, with more severe HF (46% in NYHA class III), lower level of TT and more impaired BRS as compared to patients included in the present study, that three-month testosterone supplementation increased the circulating TT level as well as improved BRS. These two kinds of evidence are in favour to consider the TT supplementation in men with HF as there are premises (our study and the Caminiti et al.’s [51] study) that it may significantly improve autonomic imbalance. The latter is a critical element of HF pathophysiology, which when restored could potentially reverse unfavourable history of HF. However, this issue is not so straightforward, because there are some reports showing that TT supplementation may be associated with a greater frequency of adverse outcomes, particularly cardiovascular, respiratory, and dermatologic events [52]. However, the study performed by Basaria et al. [52] had a few weaknesses (comments in N Engl J Med, e.g. [53]): the authors did not consider differences in physical activity (which can affect TT levels as well as the probability of CV events [54] among the men receiving testosterone supplementation vs. those receiving placebo; the potential changes of the level of physical activity over the course of the study period have not been reported. Moreover, the study lacks information regarding the levels of luteinizing and follicle-stimulating hormones before an onset of the treatment. The authors overlooked the questions of the relation between TT supplementation and an induced increase in plasma oestrogen and/or the influence of androgens on erythropoiesis. And finally, the doses of TT in the study performed by Basaria et al. [52] were relatively high, thus the observed harms could be related to simple TT oversupplementation. The issue of the potential harms associated with TT supplementation is extremely important and remains controversial [46], thus it needs to be studied further.

On the other hand, there are premises that stimulation of the autonomic system, performed at the peripheral level (excluding classical and well-known hypothalamus–hypophysial axis) participates in testosterone release from the testis [55].

One of the features of HF pathophysiology is an anabolic–catabolic imbalance favouring the catabolic processes [7,10], which ultimately may lead to the cardiac cachexia [56].

In patients with HF, even without cachexia, increased cortisol is related to unfavourable prognosis [57]. Although the other authors found some associations between the presence of cachexia and autonomic imbalance [58], neither us nor the others confirmed the associations between cotrisolemia and autonomic status in the clinical settings of HF.

There are premises that among all sex steroid hormones, DHEAS is most strongly related to the functioning of central nervous system (being considered as an important neurotransmitter), and also revealing relationships with autonomic status [59]. Circulating DHEAS decreases rapidly with age [60], and this decline is even much evident in men with systolic HF [10]. DHEAS deficiency has unfavourable consequences [61] also in men with HF, regardless of severity of heart disease and the concomitant abnormalities in other anabolic hormones, i.e. increased mortality [10] and more severe depressive symptoms [40,41]. In this paper, we have found a positive relation between the indices of HRV and DHEAS, which is consistent with the previous findings [62], suggesting another potentially favourable effect of optimal serum DHEAS on homeostasis of men with heart disease.

The results regarding the potential pathophysiological implications of the deficiency of anabolic sex steroids in patients with HF are equivocal [63]: although TT is very often decreased in elderly patients with systolic HF, being also related to the disease severity, it does not predict mortality [64]. In patients with HF and erectile dysfunction, TT is inversely related to ejection fraction, hemodynamic parameters and exercise capacity [65]. Moreover, it is suggested that the clinical status of patients with HF is related to aldosterone/dehydroepiandrosterone synthetic balance [66] and that DHEAS is an independent predictor of exercise capacity in patients with mild HF [67]. The level of eFT is known to be inversely related to hsCRP and NT-proBNP levels and to be higher in HF patients who have a history of smoking and/or hypertension [37].

It is worthy to mention that not only androgens but also oestrogens are important for the survival of both healthy men [68] and those with HF [12]. There are theories that at least some beneficial effects of estrogens on male health are driven through the interactions with central nervous system and the modulation of functioning of autonomic nervous centres [62] Experiments performed in rats showed that E2 injected directly to the autonomic nervous system as well as to peripheral blood resulted in an increase in BRS [69]. It has been also proven that E2 is positively related to sympathetic activity, and negatively to the indices of parasympathetic activation [70]. We have also found that men with lower (but not extremely low) serum estradiol had depleted HRV and reduced BRS.

In conclusion, we have demonstrated associations between hormone parameters and autonomic status in men with mild systolic HF, and suggest the existence of potential hormonal modulation of these neural reflex mechanisms. This is particularly important that these relations occur at the early stage of HF, and are found in those who are treated with drugs antagonizing neurohormonal activation. We have to acknowledge that our study was observational, and that our findings should be verified in interventional studies which could confirm the causal character of revealed associations. Nevertheless, the obtained results constitute the premises to consider the pharmacological supplementation of depleted hormones and investigate whether such an intervention would restore the autonomic imbalance and normalize the reflex mechanisms controlling the functioning of cardiovascular system.

Limitations of the study

We are aware that our study has some limitations that need to be acknowledged. Firstly, according to some experts, the fraction of free circulating TT reflects the biologically active TT which can reach the target tissues. In our study we did not measure directly free TT but we estimated free TT using the equation by Vermeulen [14,71].

Secondly, as we have emphasized this in the “Methods” and also later in the “Discussion” sections, our study was prospectively designed to measure the associations between hormones and autonomic status. Due to the complex and time-consuming methodology, most studies measuring autonomic balance have been performed with relatively small number of subjects. In the present study we have included all consecutive men who fulfilled the inclusion/exclusion criteria (i.e. mild HF, clinical stability, sinus rhythm). Thus, we have to acknowledge that our findings should be verified in interventional studies which could confirm the causal character of revealed associations.

What is more, TT is characterised by inter- as well as intra-individual variability, which results in, e.g. a broad range of normal values. Moreover, the majority of methods used to assess TT (including methodology applied in the current study) are known to be indirect. Although methods that were applied are proven to be valid and reliable, we have to underline this as a strong limitation of the present study. However, we tried to minimize the intra-individual variability in circulating TT by collecting each blood sample at exactly the same time (i.e. in the morning) in a unified condition (quite, isolated, air-conditioned room). Moreover, the presented study is focused not on the level of the particular hormone in the light of its deficiency, but on its level in relation to the autonomic status.

Declaration of interest

This research was financially supported by the State Committee for Scientific Research (Poland), grant no. NN519 580838. The authors report no conflicts of interest. The authors alone are responsible for the content and writing of this article.