Testosterone deficiency in male heart failure patients and its effect on endothelial progenitor cells

Abstract

Background: Endothelial progenitor cells (EPCs) are thought to contribute to reendothelialization and neoangiogenesis. Since it is known that EPCs express a testosterone receptor, we wanted to assess the prevalence of testosterone deficiency in patients with CHF and its impact on circulating EPCs. Methods: 137 male patients with chronic heart failure (CHF) were included (age 61 ± 13 years; BMI 29 ± 5 kg/m²; New York Heart Association classification (NYHA) I: n = 47, NYHA II: n = 51, NYHA III: n = 39). Numbers of different populations of circulating EPCs were quantified using flow cytometry. Levels of free testosterone and EPC-regulating cytokines were determined using ELISA. Results: The prevalence of testosterone deficiency in our University CHF clinic was 39%. However, there was no difference between patients with and without testosterone deficiency regarding their levels of EPCs. Testosterone levels were inversely correlated with age (R² = −0.32, p = 0.001) and NYHA status (R² = 0.28, p = 0.001) and correlated with cardiorespiratory capacity (R² = 0.26, p = 0.03). Conclusion: Testosterone deficiency is frequent in male patients with CHF but does not appear to impact the regenerative EPCs.

Keywords::

• Endothelial progenitor cells
• heart failure
• testosterone
• NYHA

Introduction

Endothelial progenitor cells (EPCs) were originally described as bone marrow derived cells circulating in the peripheral blood, expressing specific surface membrane markers such as the hematopoietic progenitor cell markers CD34, CD133 and the vascular endothelial growth factor receptor KDR [1]. EPCs contribute to normal endothelial function and renewal of blood flow in regions affected by vascular injury, and play an important role in the process of neovascularization and reendothelialization.

Several studies have shown that EPCs have the potential to differentiate into cardiomyocytes and smooth-muscle cells, therefore contributing to myocardial neovascularization and improvement of myocardial function [2].

A complex network of endogenous cytokines, such as stromal-derived factor 1 (SDF-1) and vascular endothelial growth factor (VEGF), are involved in these processes, as well as exogenous agents like certain medications, such as statins [3] or angiotensin-converting enzyme inhibitors [4]. Several other factors affect the number and function of EPCs, including cigarette smoking or physical activity [5]. Furthermore, an inverse relationship between EPC number and hypertension, diabetes mellitus, hypercholesterolemia, metabolic syndrome and family history of coronary artery disease (CAD) has been demonstrated [6].

Furthermore, evidence points towards an association between the number and function of circulating EPCs and the pathophysiology of heart failure (HF) [7]. Congestive HF is associated with selective impairment of progenitor cell function in the bone marrow and in the peripheral blood. This may contribute to an unfavorable left ventricular remodeling process [8]. Levels of circulating EPCs, determined as CD34 positive cells, have also been shown to be reduced in patients with an acutely decompensated chronic heart failure (CHF). However, in the course of hospitalization and amelioration of HF, rising numbers of circulating CD34-positive cells can be observed again [9]. Nevertheless, there are studies reporting conflicting results regarding the levels of EPCs in patients with congestive HF. Some studies report an inverse correlation between the number of circulating EPCs and various degrees of congestive HF [7] and others claim an increased count of CD34-positive cells and CD133-positive cells in patients with impaired left ventricular function and low ejection fraction [10].

With regard to the role of hormones in the regulation of EPCs, a positive correlation between estrogen levels and the number and function of circulating EPCs has been shown [11]. Due to the fact that women in their reproductive age have a markedly lower incidence of cardiovascular (CV) events compared to age-matched men, there is evidence that CV risk gradients are gender driven [12]. Thus, hormonal alterations in aging men have attracted increasing interest during the last years and new studies have been designed focusing on testosterone deficiency in aging men to verify a link between decreasing testosterone levels and the development of CV diseases (CVD). In this context, results from observational studies have already shown a higher prevalence of testosterone deficiency in male patients with congestive HF [13]. A number of studies report on age-related lowering of testosterone levels in men as a risk factor for coronary atherosclerosis [14] and an independent predictor of mortality in older men [15,16].There is evidence for low testosterone levels as a contribution to endothelial dysfunction and vascular disease, mediated through increased synthesis of inflammatory cytokines, insulin resistance, unfavorable changes in lipid profiles and among others increased visceral obesity, which also occurs in the metabolic syndrome and diabetes mellitus [17]. Testosterone deficiency in hypogonadal men is also associated with reduced number and function of EPCs, which can be restored with testosterone treatment. Immunohistochemical evidence reveals that EPCs express androgen receptor proteins, suggesting that testosterone has a direct influence on EPCs [14].

Therefore, the aim of the current study was to assess the prevalence of testosterone deficiency in a HF outpatient clinic and its association with number and function of EPCs and their regulating hormones in patients with CHF.

Materials and methods

Study subjects

The study was approved by the local Ethics Committee. One hundred and thirty-seven male patients were enrolled in the study after providing written informed consent. The study complies with the Declaration of Helsinki. All patients were recruited consecutively in our University referral outpatient clinics (Jena, Germany) for patients with CHF. Patients were included irrespectively of their HF genesis. Medical history including CV risk factors, previous and present CV events, current drug treatment and vital signs were obtained following a personal interview. Patients with acute coronary syndromes within the last three months were excluded from the study. Furthermore, only patients with stable conditions were included into the study. This was defined by no history of decompensated HF within the last three months and no change of more than one NYHA class in clinical presentation. In addition, no patients in NYHA class IV were included. Further exclusion criteria were clinical or biochemical evidence of the presence of a systemic inflammatory disease, terminal renal insufficiency (serum creatinine >250 mmol/l), malignant diseases, thrombocytopenia (<100,000/µl) or anemia (hemoglobin <10.0 g/dl). Diabetes was assessed by determination of fasting blood glucose (>6.0 mmol/l) and/or by personal medical history and an elevation of HbA1c above 5.3%. Testosterone deficiency was defined on the basis of determined free serum testosterone levels for each patient [18]. Free testosterone was measured using Active^® Free Testosterone Radioimmunassay (Beckman Coulter, Krefeld, Germany) according to the manufacturer’s instructions. According to known variations in used reagents and normal ranges between different laboratories and the used determination method age- and sex-specific cut-off values were defined by the laboratory of the University Hospital of Jena using the data provided by the manufacturer and a reference experiment [19]. For example, the reference values for a 45-year-old man were 49.0–122.0 pmol/l, whereas they were for a 55-year-old man 40.4–74.9 pmol/l.

Echocardiography

The echocardiograms (Philips iE33, Philips, Germany) were performed by cardiologists unrelated to this study, and unaware of the blood testing results. Left ventricular ejection fraction (LVEF %) was derived using Simpson’s modified biplane method; left ventricular end-diastolic diameter (LVEDD) was measured in parasternal long axis.

Electrocardiogram and laboratory investigations

Recording of the electrocardiogram and determination of blood lipids, creatinine and red blood cell count, as well as brain natriuretic peptide (BNP) were done according to internal and international standards.

Flow cytometry

Flow cytometry was used to determine the percentage of cells expressing cell surface antigens recognizing progenitor cells. Briefly, a venous blood sample (10 ml) was obtained in EDTA tubes for the enumeration of EPCs. Different progenitor subpopulations were identified by using the hematopoietic progenitor cell marker CD34, the immature hematopoietic progenitor cell marker CD133, and the endothelial cell receptor VEGFR2 (vascular endothelial growth factor receptor-2, also known as kinase domain receptor, KDR), as previously described [20,21]. Peripheral blood mononuclear cells (PBMC) were incubated with allophycocyanin (APC)-conjugated anti-human CD34 monoclonal antibody (mAb) (BD Biosciences), phycoerythrin (PE)-conjugated anti-human CD133 mAb (Milteny Biotec) and fluorescein-isothiocyanate-conjugated monoclonal anti-VEGFR2 (R&D Systems) for 60 min at 4°C. Control isotype immunoglobulin antibodies were obtained from Becton Dickinson. Following incubation, cells were washed with PBS and analyzed on a FACS Calibur flow cytometer (BD Biosciences, Franklin Lakes, NJ, USA). Each analysis acquired at least 5 × 10⁶ events. The frequency of peripheral blood cells positive for these reagents was determined by a two-dimensional side-scatter fluorescence dot-plot analysis of the samples. Threshold was adjusted in forward and side scatter dot plot to exclude cellular debris. Data were processed using the Summit software program (BD CellQuest Pro, Version 5.2.1).

Enzyme linked immunosorbent assay

Circulating levels of SHBG, SDF-1, VEGF and adiponectin were determined using commercially available Enzyme Linked Immunosorbent Assay (ELISA) according to the manufacturer’s instructions (R&D, Wiesbaden, Germany).

Statistical analysis

All data are expressed as mean ± standard deviation (SD). Continuous variables were tested for normal distribution with the Kolmogorov–Smirnov test. Not normally distributed continuous variables groups were compared by the Mann–Whitney U test, while others were analyzed by t test (two-sided, including Levene’s test for equality of variances). Bivariate correlation between levels of EPCs and non-parametric variables (e.g. risk factors or drug treatment) were assessed by Spearman’s rank correlation coefficient whereas associations between levels of EPCs and parametric variables (e.g. BMI or free testosterone) were assessed by Pearson’s correlation coefficient. To identify independent influences multiple linear regression analyses were performed. Testosterone levels, systolic blood pressure, left ventricular ejection fraction, NYHA status, BMI and age were included in the model. Many of the variables included in the model are intercorrelated, which may weaken the model. Therefore, stepwise multiple regression analyses were performed to further analyze the most significant interrelation to the variation of EPC. First, all variables were entered into the model simultaneously. In each following step, the variable having the least p value was excluded from the model. Finally, all variables with p < 0.10 remained in the model. Standardized b-estimate was used to determine which variable had the strongest effect. Furthermore, for co-linearity statistics tolerance scores were used to estimate co-linearity. Statistical significance was assumed if a null hypothesis could be rejected at p ≤ 0.05. All statistical analyses were performed with IBM SPSS statistics, version 19.0 (IBM Inc.).

Results

Baseline clinical characteristics of the 137 subjects are summarized in Table I and Table II. The cohort represents a patient population of a referral HF clinic. Common comorbidities were frequently present and patients were treated according to international guidelines [22]. Most frequent etiologies were ischemic and dilated cardiomyopathy. Creatinine (116 ± 73 mmol/l) and brain-natriuretic peptide (485 ± 627 pg/ml) were above reference values.

Table I.  Baseline characteristics of the study population.

Patients’ characteristics, mean ± SD
Age (years)	61 ± 13
Body weight (kg)	91 ± 17
BMI (kg/m^2)	29 ± 5
Systolic blood pressure (mmHg)	135 ± 21
Diastolic blood pressure (mmHg)	86 ± 13
Comorbidities, % (n)
Hypertension	78% (107)
CAD	44% (61)
Hyperlipidemia	51% (71)
Diabetes	26% (35)
Co-medication, % (n)
Statin	51% (71)
ACE-Inhibitor	75% (103)
β blocker	94% (129)
Aspirin	31% (42)

ACE, angiotensin converting enzyme; BMI, body mass index; CAD, coronary artery disease; SD, standard deviation.

Table II.  Heart failure genesis and laboratory parameters in the study population.

Heart failure characteristics, % (n)
NYHA I	34% (47)
NYHA II	37% (51)
NYHA III	29% (39)
Ischemic cardiomyopathy	33% (46)
Dilated cardiomyopathy	31% (43)
Hypertensive cardiomyopathy	23% (31)
Other genesis	13% (18)
Laboratory values, mean ± SD
Creatinine (mmol/l)	116 ± 73
Cholesterol (mmol/l)	5.0 ± 1.4
LDL-cholesterol (mmol/l)	2.9 ± 1.2
BNP (pg/ml)	485 ± 627
Hemoglobin (g/dl)	14 ± 2
Free testosterone (pmol/l)	33 ± 14

BNP, brain natriuretic peptide; LDL, low density lipoprotein; NYHA, New York heart failure classification.

The study population was divided into two groups: (i) patients with testosterone deficiency and (ii) patients without testosterone deficiency, based on method-depending and age-adjusted reference values. Following this analysis, we observed that 39% of the population was testosterone-deficient. As expected, patients with testosterone deficiency had lower free testosterone levels (29.8 pmol/l vs. 35.3 pmol/l, p < 0.05). There were no significant differences in the genesis of HF, NYHA status or comorbidities between the two groups. Levels of cytokines such as adiponectin, SDF-1 and VEGF were also not significantly different between the two groups. Sex hormone binding globulin (SHBG) levels were significantly higher in patients without testosterone deficiency (485.6 ± 243.1 nmol/ml vs. 332.8 ± 171.2 nmol/ml, p = 0.003). Furthermore, there were no significant differences in echocardiographic findings between the two groups (Table III). Moreover, there was no significant difference in exercise capacity, as measured in the maximally achieved power stage (123 ± 56 Watt vs. 138 ± 35 Watt), nor in maximum oxygen uptake (20 ± 6 ml/kg/min vs. 22 ± 6 ml/kg/min).

Table III.  Levels of cytokines and echocardiographic parameters in the two study groups.

	Patients without testosterone deficiency n = 84	Patients with testosterone deficiency n = 53	p value
Cytokines
Adiponectin (ng/ml)	6261.67 (3868.4)	6185.54 (4160.2)	n.s.
VEGF (pg/ml)	37.8 (37.7)	33.87 (25.4)	n.s.
SDF-1 (pg/ml)	3277.5 (593.6)	3193.12 (1091.0)	n.s.
SHBG
SHBG (nmol/ml)	485.61 (243.3)	332.81 (171.2)	0.003
Echocardiographic parameters
IVSd (mm)	12 ± 3	11 ± 3	n.s.
Ejection fraction (%)	39 ± 13	38 ± 11	n.s.
PAP (mmHg + CVP)	27 ± 10	25 ± 10	n.s.
Spiroergometry/ Ergometry (reached wattage)	123 ± 56	138 ± 35	n.s.
VO2 max.(ml/kg/min)	20 ± 6	22 ± 6	n.s.

CVP, central venous pressure; IVSd, intraventricular septum, diastolic; LVEDd, left ventricular diameter, diastolic; PAP, pulmonary arterial pressure; SDF-1, stromal cell derived factor 1; SHBG, sex-hormone-binding globuline; VEGF, vascular endothelial growth factor; VO₂ max, maximal oxygen uptake.

Data are presented as mean ± SD.

Correlations between selected parameters are presented as a correlation matrix in Table IV. A negative correlation was observed between free testosterone level and age (R² = −0.32, p < 0.001), in addition to NYHA-Stage (R² = −0.28, p = 0.001). A positive correlation was observed between free testosterone levels and achieved power stage in the ergometry (R² = 0.26, p = 0.03). Furthermore, several markers for disease severity correlated with each other, such as ejection fraction, NYHA status, BNP levels and maximal oxygen uptake (Table IV).

Table IV.  Correlation matrix between different clinical parameters in the entire study population.

	Free testosterone	Age	NYHA	Systolic blood pressure	Ejection fraction	Maximum power stage	Maximal oxygen uptake	BNP
Free Testosterone	R^2	−0.317^**	−0.282^**	−0.001	−0.047	0.262^*	0.014	−0.027
	p	0.000	0.001	0.989	0.610	0.030	0.931	0.772
Age	R^2		0.481^**	0.055	−0.071	−0.517^**	−0.286	0.245^**
	p		0.000	0.530	0.437	0.000	0.070	0.007
NYHA	R^2			−0.105	−0.315^**	−0.639^**	−0.702^**	0.333^**
	p			0.238	0.000	0.000	0.000	0.000
Systolic blood pressure	R^2				0.007	−0.030	−0.006	−0.096
	p				0.940	0.806	0.971	0.308
Ejection fraction	R^2					0.306^*	0.595^**	−0.412^**
	p					0.015	0.000	0.000
Maximum power stage	R^2						0.690^**	−0.237
	p						0.000	0.071
Maximal oxygen uptake	R^2							−0.455^**
	p							0.005

R² represents regression coefficients whereas p represents p values.

*The correlation is significant at the p<0.05 level.

**The correlation is significant at the p<0.01 level.

No significant differences were found between levels of EPCs in the two study groups (Figure 1). Levels of CD34-positive cells were similar in both groups (0.93 cells/µl vs. 0.97 cells/µl) (Figure 1A). Furthermore, no difference was detected between levels of the various EPC subtypes examined (Figure 1B–1D). However, a significant positive correlation was observed between the number of CD34/CD133/KDR cells and free testosterone levels (R² = 0.18, p = 0.031). In addition, a significant positive correlation was also found between levels of CD34/CD133-positive cells and BMI (R² = 0.26, p = 0.003) and SHBG levels (R² = −0.236, p = 0.006).

Figure 1.  Endothelial progenitor cell subpopulations detected in patients with chronic heart failure divided with and without testosterone deficiency. Shown subpopulations include A: CD34⁺, B: CD34⁺CD133⁺, C: CD34⁺KDR⁺ and D: CD34⁺CD133⁺KDR⁺. TD, testosterone deficiency. Data presented as mean ± SD.

Multiple linear regression analysis revealed that for CD34-positive cells only NYHA stage (p = 0.034, β: −0.26) and BMI (p = 0.08, β: 0.21) remained in the model. For more endothelial specific progenitor cell populations such as CD34/CD133/KDR cells only age remained in the model (p = 0.05, β: −0.23).

Discussion

The aim of this study was to evaluate the prevalence of testosterone deficiency in male HF patients in a real-life scenario and its association with numbers of circulating EPCs. The presence of testosterone deficiency according to age-specific reference values is 39%. Although there was no difference in absolute levels of circulating EPCs between with normal or testosterone deficient individuals, we did observe a positive correlation between levels of CD34/CD133/KDR-positive EPCs and free testosterone.

In recent years, the existence of a complex interaction between hormonal, metabolic and nutritional changes has been observed in patients with a progression of HF. In this context, anabolic deficiency in terms of free testosterone, dihydroepiandrosterone sulfate (DHEAS) and insulin-like-growth-factor-1 (IGF-1) has been frequently detected in men with CHF and negatively influences their prognosis [23,24]. In elderly men (>60 years) and in chronic disease states, a decrease in male sex hormones below the lower limit of normal for young men is not frequently observed [25].

As for the role of EPCs, their prognostic impact has been confirmed in a large study including patients with congestive HF. Levels of circulating EPCs were shown to be the most important determinant of prognosis with advanced age and diabetes mellitus following Cox proportional regression analysis [26]. Changes in EPC biology are recognized as surrogate markers of CV risk and likewise as a therapeutic target for drugs exerting effects on the CV system. It has also been observed that hypogonadism and HF are accompanied by reduced numbers of EPCs [7,27], as well as immunohistochemical analysis demonstrated the presence of androgen-receptor protein in EPCs. In these terms, it has been hypothesized that testosterone could exert a direct effect on EPC function [14].

Several studies have pointed out that commonly observed symptoms of aging men such as fatigue, an impaired libido, mood changes with depressive tendencies, sleep problems, anemia, loss of muscle strength or bone mass with resulting osteopenia or osteoporosis result from a decrease in the production of male sex hormones similar to the changes during the female menopause [19]. Besides that, testosterone deficiency has been related to obesity, dyslipidemia, insulin resistance, hypertension, high fibrinogen levels as well as the development of diseases accompanied by a high CV risk profile like diabetes mellitus or metabolic syndrome [17]. It has also been shown that testosterone deficiency favors the development of atherosclerosis. Concerning the relationship between testosterone deficiency and CAD, there are studies investigating androgens and CAD which show a weak association between hypogonadism and coronary events [13]. However, many studies have failed to observe a statistically significant relationship between endogenous testosterone production and the incidence of CVD in men [28,29].

Changes in lipid profiles of men affected by decreased serum testosterone levels have been claimed as a major factor in the pathogenesis of CAD atherosclerosis. There is evidence that levels of total cholesterol, triglycerides, LDL-C and oxidized LDL increase, while HDL-C levels decrease under low serum testosterone levels [30]. Indeed, results from our study demonstrate that low levels of free testosterone are clearly related with the manifestation of the CV risk factor hyperlipidemia. In addition, the often mentioned link between obesity and decreased testosterone levels [31,32] can be substantiated, showing that a high BMI is associated with decreased levels of free testosterone and SHBG levels. In addition, due to an altered ratio between testosterone and SHGB, this can result in a lower quantity of free, biologically active hormone. This could contribute to a greater anabolic hormone deficiency and is related with a higher risk of CV events and cardiac death [33].

Typical symptoms such as fatigue, sleep problems or muscle weakness are particularly aggravated in HF as well as the severity of the disease may be directly affected by low levels of androgens [34]. In this regard, similar to the study of Jankowska et al. [35], our results show that low circulating testosterone levels are associated with exercise intolerance.

Previous studies examining the effects of testosterone on EPC number and function show conflicting results. One study demonstrated that hypogonadal men had a significant decrease in ex vivo determined circulating EPCs and pharmacological treatment with testosterone restored the number of circulating EPCs [27]. Another study showed potential favorable effects of androgens on EPCs in vitro [14]. In contrast, it has also been stated that testosterone and its active metabolite dihydrotestosterone exert no effects on expansion and function of late EPCs isolated from the peripheral blood of healthy human adult males in vivo and in vitro [36]. Possible explanations could be the method used to quantify circulating EPCs, as it is a well known problem that the non-uniform methods to study EPCs can yield different results [37]; or the not sufficiently considered influence of 17β-oestradiol, which is known to positively influence mobilization and enhancement of EPC number and function [38,39].

In light of these various factors, this study verifies the participation of the gonadal axis in the disease process of HF as well as a direct effect of androgens on number of countable EPCs, especially given the fact that the number of CD34/CD133/KDR-positive cells positively correlated with levels of free testosterone. But in consideration of other pathologic changes ongoing in the development and progress of HF, like unfavorable changes in the adrenal-axis or increased expression of inflammatory cytokines, e.g. the myelosuppressive tumor-necrosis factor-α, interleukin-1 or interleukin-6 [40], which are known to exert a negative influence on the hypothalamic-pituitary-testicular-axis [41] and mobilization and function of EPCs [42], it is plausible that those effects dominate over the negative effects caused by a decreased testosterone secretion in patients affected by congestive HF and explain the difference to the findings published earlier in patients without heart failure by Foresta and coworkers [14,43]. In addition, the role of inflammatory cytokines as confounders in the relationship between testosterone, cardiac function, heart failure and EPCs is unclear. Furthermore, the selective functional exhaustion of EPCs in post-ischemic HF has been recently demonstrated, as well as the fact that bioavailability of nitric-oxide (NO) is restricted under this condition. This results in a decreased release of EPCs from bone marrow niches, as NO acts on MMP-9 activity which is acquired for mobilization and proliferation of EPCs [8], and which supports this theory and explains the general low count of circulating EPCs in this study.

Our study does have several limitations that need to be addressed. The sample size is limited and the study was designed as an observational and cross-sectional study without randomization. However, our patient population is well characterized and represents a real-life scenario in an academic referral center for CHF. Furthermore, testosterone deficiency was defined on the basis of determined free serum testosterone free and not totals testosterone. Furthermore, the patient cohort had a high prevalence of obesity. This should be taken into consideration while interpreting the results of the present study.

In conclusion, testosterone deficiency is frequent in male patients with CHF but does not appear to impact the regenerative EPCs.

Acknowledgements

The authors thank Annett Schmidt for excellent technical assistance.

Declaration of interest: The authors declare no conflicts of interest.