Background
Normal values of left ventricular ejection fraction (LVEF) and absolute values of global longitudinal strain (GLS) are lower in men than in women. Data concerning the association of sex hormone levels on these left ventricular systolic function surrogates are scarce. The aim of this study was to determine the association of sex hormones with systolic left ventricular function in healthy subjects and patients with congenital adrenal hyperplasia (CAH) as a model of testosterone dysregulation.
Methods
Eighty-four adult patients with CAH (58 women; median age, 27 years; interquartile range, 23–36 years) and 84 healthy subjects matched for sex and age were prospectively included. Circulating concentrations of sex hormones were measured within 48 hours of echocardiography with assessment of LVEF and left ventricular longitudinal, radial, and circumferential strain.
Results
LVEF and GLS were higher in healthy women than in healthy men (63.9 ± 4.2% vs 60.9 ± 5.1% [ P < .05] and 20.0 ± 1.9% vs 17.9 ± 2.4% [ P < .001], respectively), while there was no difference in LVEF or GLS between women and men with CAH (63.9 ± 4.5% vs 63.0 ± 4.6% [ P = NS] and 19.4 ± 2.2% vs 18.3 ± 1.8% [ P = NS], respectively). Bioavailable testosterone levels were higher in women with CAH than in female control subjects (0.08 ng/mL [interquartile range, 0.04–0.14 ng/mL] vs 0.16 ng/mL [interquartile range, 0.04–0.3 ng/mL], P < .001) and lower in men with CAH than in male control subjects (2.3 ng/mL [interquartile range, 1.3–3 ng/mL] vs 2.9 ng/mL [interquartile range, 2.5–3.4 ng/mL], P < .05). In men, LVEF and GLS were negatively correlated with bioavailable testosterone levels ( r = −0.3, P ≤ .05, and r = −0.45, P < .01, respectively), while midventricular radial strain was positively correlated with bioavailable testosterone level ( r = 0.38, P < .05). The absolute value of circumferential strain was positively correlated with follicle-stimulating hormone ( r = 0.65, P < .0001).
Conclusions
These data support that the existence of sex dimorphism concerning left ventricular systolic cardiac function is significantly associated with testosterone levels.
Highlights
- •
There is no major alteration in cardiac function and structure in treated CAH patients.
- •
Sex influences most cardiac morphological and functional parameters.
- •
In men, testosterone levels are associated with cardiac systolic contractility.
In the healthy population, women have different reference values for most cardiac morphologic and functional parameters. Men have larger and thicker left ventricles than women. This is associated with a differential pattern of the radial, circumferential, and longitudinal components of left ventricular (LV) systolic function. Women and men tend to experience different types of heart failure (HF). Men are more likely to experience systolic dysfunction, leading to HF withreduced ejection fraction, whereas women have higher rates of diastolic dysfunction, leading to HF with preserved ejection fraction. Extensive data exist concerning the influence of sex hormones on heart repolarization, but dataconcerning their influence on cardiac morphology and function are scarce and disparate. Large epidemiologic studies have shown that low serum total testosterone is a biomarker of cardiovascular mortality, and several small trials have shown promising results with testosterone for the treatment of HF. However, administration of higher supraphysiologic levels of testosterone has been associated with impaired diastolic function and postinfarction remodeling in clinical and preclinical models. Preclinical studies also have suggested that estrogens can prevent the deterioration of cardiac function and remodeling after experimental myocardial infarction. The potential influence of other sex steroid hormones showing important roles in heart repolarization, such as progesterone, or gonadotropins regulating their production, has not been studied.
To further investigate these issues, we studied the association of main sex steroid hormones and gonadotropins on LV morphology and function in healthy subjects and in a unique cohort of patients with congenital adrenal hyperplasia (CAH) due to 21α-hydroxylase deficiency, a rare disease, used in our study as a model of testosterone dysregulation. CAH corresponds to a group of inherited autosomal-recessive disorders that arise from defective steroidogenesis and result from a deficiency in one or several of the enzymes of cortisol biosynthesis. Deficiency of the 21α-hydroxylase enzyme is the most common form of CAH and is characterized by cortisol and in some cases aldosterone deficiency associated with adrenal androgen excess and renin increase in some cases ( Figure 1 ). Cortisol deficiency results in the adrenocorticotropic hormone–induced accumulation of substrate precursors such as 17-OH progesterone and progesterone and to increased secretion of adrenal androgens. Goals of treatment are to replace deficient hormones by glucocorticoids and in some cases mineralocorticoids and to control adrenal androgen excess. The optimal dose of glucocorticoid is that which fails to fully suppress 17-OH progesterone and maintains androgens in the midnormal range. There is variability in adrenocorticotropic hormone suppression by this supplementation, leading to interindividual variability in residual secretion of adrenal androgen and ultimately testosterone levels. In some patients, this leads to insufficient control of their disease, resulting in high levels of progesterone and adrenal androgens, disturbing the gonadal axis with low luteinizing hormone and follicle-stimulating hormone (FSH) release by the pituitary gland. Because of this variable response, we were expecting mean testosterone levels to be higher in women with CAH and lower in men with CAH than in sex-matched control subjects (see Figure 1 for further details).
The purpose of this work was first to determine if LV morphology and systolic function are different within and between men and women depending on CAH status. Second, our aim was to investigate for the first time the association of gonadotropins and sex steroid hormones with LV function to better decipher the normal structure and function of a beating heart.
Methods
Study Design
Our work is ancillary to the Cardiovascular Risk Profile in Patients With Congenital Adrenal Hyperplasia (CARDIOHCS) study ( ClinicalTrials.gov identifier NCT01807364 ), a multicenter prospective observational case-control study comparing early cardiovascular damage in adult men and women with CAH due to 21α-hydroxylase deficiency and healthy control subjects. All patients gave written informed consent to participate, and the study was approved by each hospital ethics committee.
Study Population
Eighty-four young adults (58 women and 26 men) with CAH and 84 control subjects matched for sex, age (±5 years), and smoking status (nonsmoking, past smoking, active smoking) were prospectively included in the CARDIOHCS study between May 2011 and August 2015. Adults with CAH included were diagnosed during childhood, proved by genetic testing confirming a 21α-hydroxylase deficiency. Exclusion criteria for CAH and healthy subjects were known history of cardiovascular disease and pregnancy. Estradiol and/or progesterone contraception in the previous month was an exclusion criterion in healthy subjects, and interruption was promoted in patients with CAH. Patients with CAH were treated, as needed by their standard of care, with hydrocortisone or dexamethasone and some of them also received fludrocortisone.
Study Procedures and Laboratory Analysis
Patients with CAH were referred to the study by three endocrinology units (Pitié-Salpêtrière Hospital, Paris, France; Saint-Antoine Hospital, Paris, France; and Bicêtre Hospital, Le Kremlin-Bicêtre, France). Patients with CAH and healthy subjects were assessed at Centre d’Investigation Clinique Paris-Est (CIC-1421, Pitié-Salpétrière Hospital, Paris, France).
In the morning, before 10 am , participants underwent a clinical examination including medical history, physical examination with blood pressure measurement (using a standard sphygmomanometer), weight and height measurements (and calculation of body mass index as weight/height 2 ), and electrocardiographic recording after ≥10 min of rest in the supine position. Included subjects were asked for chronic treatment intake. Blood samples for the determination of serum concentrations of 17-OH progesterone, progesterone, testosterone, sex hormone–binding globulin, estradiol, FSH, and luteinizing hormone were collected in a dry tube and further assayed in the immunology laboratory at Pitié-Salpêtrière. Estradiol, progesterone, sex hormone–binding globulin, testosterone, FSH, and luteinizing hormone serum concentrations were assayed using chemiluminescence (Modular-E170; Roche, Mannheim, Germany) and 17-OH progesterone by radioimmunoassay (KIP1409; DIAsource ImmunoAssays, Louvain-la-Neuve, Belgium). Hexokinase/glucose-6-phosphate dehydrogenase was used for the determination of fasting plasma glucose levels (Roche). The methodology of electrocardiographic acquisition and measurements has been described elsewhere.
Echocardiography Acquisition and Analysis
Transthoracic echocardiography was performed within 48 hours of clinical and biological evaluation. It was performed using a Vivid 9 equipped with an M5S convex transducer (GE Vingmed Ultrasound, Horten, Norway) by one expert physician, blindly to clinical and biological data. Two-dimensional images, acquired with appropriate sector size and depth settings to achieve optimal visualization of all LV myocardium at the optimal frame rate of 50 to 80 frames/sec, were transferred to a workstation equipped with EchoPAC PC version 113 (GE Vingmed Ultrasound). All examinations were analyzed offline by one blinded senior cardiologist. All measurements were averaged over three to six cardiac cycles.
The following measurements were performed according to current guidelines : LV internal diameter and interventricular septal and posterior wall thicknesses at end-diastole from M-mode images. LV mass was derived according to the Devereux formula. LV relative wall thickness was defined as (2 × LV wall thickness)/LV diastolic diameter, measured at end-diastole. LV wall thickness was calculated from the mean of posterior and interventricular septal wall thicknesses. Left atrial maximal volume was measured using the biplane method from apical four- and two-chamber views. LV volumes and LV ejection fraction (LVEF) were measured using the biplane Simpson method. Measurements were indexed to body surface area as appropriate. Transmitral E-wave velocity (E) was determined using Doppler echocardiography. Using pulsed tissue Doppler mode, we acquired e′, the mitral annular early diastolic velocity, in the four-chamber apical view at the septal level. Septal E/e′ was then calculated and considered as a surrogate of LV filling pressure. LV longitudinal strain, radial strain (RS), and circumferential strain (CS) were measured offline and blinded by one expert cardiologist, with the use of standard commercially available software (Q-Analysis, EchoPAC PC). RS and the absolute value of CS were assessed on LV short-axis images at the papillary muscle level (midventricular) with speckle-tracking analysis. For all subjects, global longitudinal strain (GLS) values were derived from the average of LV apical four-chamber and two-chamber views, using speckle-tracking analysis ( Figure 2 ). Three apical views were missing or of insufficient quality in 43 of 168 subjects (25.6%), precluding the use of 18-segment derived GLS in this study. Intraobserver variability of LVEF, LV strain, and E/e′ measurements in our laboratory is excellent (intraclass correlation coefficients ≥ 0.83) and has been reported elsewhere.
Statistical Analyses and Power of the Study
Data are reported as mean ± SD or as median and interquartile range, as appropriate. Comparisons of quantitative variables were made by using Wilcoxon paired t tests, Mann-Whitney U tests, and analysis of variance (one-way or two-way) with Tukey or Sidak posttests, as appropriate. Qualitative variables were compared by using χ 2 tests. Correlations between linear variables were assessed by calculating the Pearson or Spearman coefficient ( r ), as appropriate, using Prism 6 (GraphPad Software, La Jolla, CA). Multivariate analysis was performed using analysis of covariance (XLSTAT; Addinsoft, Paris, France). Explanatory variables were selected to be tested in multivariate analysis only if they were associated with the dependent variable in univariate analyses, except for CAH status, which was always tested. The best model was selected driven by Schwarz’s Bayesian criterion. Statistical significance was accepted at P ≤ .05.
This study was sufficiently powered to detect the expected sex difference in absolute value of LVEF and GLS (∼2%–3%) within the CAH and healthy subgroups. However, our study lacked power (<70%) to study sex differences for absolute values of RS and CS. In detail, the study had 82% power to detect a difference in LVEF of ≥3% between healthy men and any other of the three other groups by analysis of variance (α risk = 0.05, LVEF SD = 4%, expected mean LVEF = 65%, considering n = 25 in each male subgroup and 50 in each female subgroup). The study had 87% power to detect a difference in GLS of ≥|2%| between healthy men and any other of the three other groups by analysis of variance (α risk = 0.05, GLS SD = |2.5%|, expected mean GLS = |20%|, considering n = 25 in each male subgroup and 50 in each female subgroup).
In men with no exogenous sex hormone intake and delay < 48 hours between hormonal sampling and echocardiography ( n = 41), the study had power of 80% to detect a significant correlation (with r > |0.41|, α risk = 0.05) between hormone levels and LVEF or strain value.
We could not analyze the relations between sex hormone levels and echocardiographic parameters in the female subset, because women are characterized by very steep variations in hormone levels during the menstrual cycle, as opposed to men. In fact, most echocardiographic examinations were performed ≥24 to 48 hours after blood sampling.
Results
Clinical and Hormonal Evaluations
The clinical characteristics of patients included in this study are shown in Table 1 . In total, 168 adult subjects (median age, 27 years; interquartile range [IQR], 23 to 36 years), of whom 116 were women (56 with CAH and 56 healthy) and 52 men (26 with CAH and 26 healthy), were included and available for analysis. Subjects had no known histories of cardiovascular disease or treated hypertension. Only two women with CAH were treated for dyslipidemia or diabetes.
| Variable | Men with CAH | Healthy men | Women with CAH | Healthy women | P |
|---|---|---|---|---|---|
| Number of subjects | 26 | 26 | 58 | 58 | |
| Age (y) | 28.8 ± 7.9 | 28.7 ± 8.5 | 30.7 ± 9.1 | 30.9 ± 9 | NS |
| Body surface area (m 2 ) | 1.80 ± 0.19 † , ‡ | 1.90 ± 0.15 ∗ , † , § | 1.63 ± 0.17 ∗ , ‡ , § | 1.72 ± 0.13 † , ‡ | <.0001 |
| BMI > 35 kg/m 2 | 0 | 0 | 3 (5%) | 0 | NS |
| SBP (mm Hg) | 117.4 ± 9.3 ∗ , † | 118.9 ± 8.6 ∗ , † | 110.0 ± 14 ‡ , § | 108.5 ± 10.2 ‡ , § | <.0001 |
| DBP (mm Hg) | 69.1 ± 9.8 | 69.6 ± 7.2 | 68.2 ± 12.3 | 68.3 ± 7.9 | NS |
| Treated hypertension | 0 (0%) | 0 (0%) | 0 (0%) | 0 (0%) | NS |
| Cardiovascular disease in family | 0 (0%) | 1 (4%) | 1 (2%) | 5 (9%) | NS |
| Current smoker/past or nonsmoker | 7/19 (27/73%) | 7/19 (27/73%) | 16/42 (31/69%) | 16/42 (31/69%) | NS |
| Treated dyslipidemia or diabetes | 0 (0%) | 0 (0%) | 2 (3%) | 0 (0%) | NS |
| Fasting glycemia (≥7 mmol/L) | 0 (0%) | 0 (0%) | 0 (0%) | 1 (2%) | NS |
| Fludrocortisone (μg/d) | 100 (50–100) † | NA | 50 (0–94) § | NA | <.01 |
| Equivalent of hydrocortisone (mg/d) | 30 (20–40) † | NA | 20 (10–25) § | NA | <.001 |
∗ Significant compared with healthy women.
† Significant compared with women with CAH.
‡ Significant compared with healthy men.
The results of biological evaluations in women are shown in Table 2 . Compared with healthy women, women with CAH had higher levels of progesterone (median, 2.7 ng/mL [IQR, 1–8.5 ng/mL] vs 0.9 ng/mL [IQR, 0.5–3 ng/mL]; P < .01) and bioavailable testosterone (median, 0.16 ng/mL [IQR, 0.04–0.32 ng/mL] vs 0.08 ng/mL [0.04–0.14 ng/mL]; P < .001).
| Variable | CAH women ( n = 58) | Healthy women ( n = 58) | P |
|---|---|---|---|
| Women | |||
| 17-OH progesterone (ng/mL) | 14.7 (4.4–50.8) | 1.2 (0.7–2.1) | <.0001 |
| Progesterone (ng/mL) | 2.7 (1–8.5) | 0.9 (0.5–3) | <.01 |
| SHBG (nmol/L) | 63 (31–91) | 64 (54–87) | NS |
| Bioavailable testosterone (ng/mL) | 0.16 (0.04–0.3) | 0.08 (0.04–0.14) | <.001 |
| Estradiol (pg/mL) | 65 (42–162) | 87 (40–170) | NS |
| FSH (IU/L) | 5 (3.2–6.4) | 6.4 (4.1–8.3) | <.01 |
| LH (IU/L) | 5.5 (2.8–8.3) | 6.7 (5.1–12.8) | <.01 |
| Renin (pg/mL) | 19.3 (13.4–44) | 11.3 (7.6–16.8) | <.001 |
| ACTH (pg/mL) | 30.3 (9–67.3) | 16.5 (10.8–24.6) | <.0001 |
| CAH men ( n = 26) | Healthy men ( n = 26) | ||
|---|---|---|---|
| Men | |||
| 17-OH progesterone (ng/mL) | 15.1 (4.0–44.3) | 2 (1.6–2.4) | <.0001 |
| Progesterone (ng/mL) | 1.0 (0.4–3.6) | 0.8 (0.6–1) | <.05 |
| SHBG (nmol/L) | 36.2 (28.2–55) | 40.2 (24.5–46.3) | NS |
| Bioavailable testosterone (ng/mL) | 2.3 (1.3–3) | 2.9 (2.5–3.4) | <.05 |
| Estradiol (pg/mL) | 29.5 (19.8–32.6) | 28.3 (24.7–37.3) | NS |
| FSH (IU/L) | 3.9 (2.3–7) | 3.4 (2.6–5.2) | NS |
| LH (IU/L) | 4.8 (2–5.8) | 4.5 (3.5–5.3) | NS |
| Renin (pg/mL) | 26.0 (15.9–44.6) | 13.1 (10–16.5) | <.01 |
| ACTH (pg/mL) | 44.9 (18–161.8) | 27.2 (14.8–38) | <.01 |
The results of biological evaluations in men are shown in Table 2 . Compared with healthy men, men with CAH had higher levels of progesterone (median, 1.0 ng/mL [IQR, 0.4–3.6 ng/mL] vs 0.8 ng/mL [IQR, 0.6–1 ng/mL]; P < .05). In contrast, bioavailable testosterone levels were lower in men with CAH than in healthy men (median, 2.3 ng/mL [IQR, 1.3–3 ng/mL] vs 2.9 ng/mL [IQR, 2.5–3.4 ng/mL]; P < .01).
Echocardiographic Evaluations
Heart morphologic parameters such as initial thoracic aortic diameter, indexed LV telediastolic volume, and indexed LV mass were not different between healthy men and men with CAH or between healthy women and women with CAH. However, these parameters were significantly higher in men than in women ( P < .0001; Table 3 ).
| Variable | Healthy women ( n = 58) | Women with CAH ( n = 58) | Men with CAH ( n = 26) | Healthy men ( n = 26) | P for interaction (sex-disease) | P for sex influence | P for disease influence |
|---|---|---|---|---|---|---|---|
| Ascending aorta (mm) | 29.2 ± 3.9 ‡ , § | 28.8 ± 2.7 ‡ , § | 32.6 ± 2.7 ∗ , † | 33.0 ± 3.1 ∗ , † | NS | <.0001 | NS |
| LV morphology | |||||||
| LV end-diastolic volume index (mL/m 2 ) | 48.8 ± 8.5 ‡ , § | 47.3 ± 10.7 ‡ , § | 55.9 ± 12.3 [1] ∗ , † | 59.4 ± 12.4 ∗ , † | NS | <.0001 | NS |
| LV end-systolic volume index (mL/m 2 ) | 17.9 ± 4.3 ‡ | 17.1 ± 4.6 ‡ , § | 20.9 ± 6.1 [1] † | 23.5 ± 6.8 ∗ , † | NS | <.0001 | NS |
| LV wall thickness (mm) | 7.09 ± 0.90 ‡ , § | 7.12 ± 0.82 ‡ , § | 7.85 ± 0.88 ∗ , † | 8.25 ± 1.08 ∗ , † | NS | <.0001 | NS |
| LV mass index (g/m 2 ) | 63.4 ± 11.9 ‡ , § | 67.0 ± 12.6 ‡ , § | 76.5 ± 15.4 ∗ , † | 83.2 ± 16.1 ∗ , † | <.05 | <.0001 | NS |
| LV relative wall thickness | 0.30 ± 0.04 | 0.30 ± 0.05 | 0.31 ± 0.04 | 0.32 ± 0.05 | NS | <.05 | NS |
| LV diastolic function | |||||||
| Left atrial volume index (mL/m 2 ) | 31.7 ± 7.3 | 29.5 ± 6.0 ‡ | 31.8 ± 7.1 [1] | 34.9 ± 7.7 † | NS | <.05 | <.05 |
| E/e′ ratio | 6.6 ± 1.9 [4] | 6.5 ± 1.4 | 6.4 ± 1.5 [3] | 6.0 ± 1.4 | NS | NS | NS |
| LV systolic function | |||||||
| LVEF (%) | 63.9 ± 4.2 ‡ | 63.9 ± 4.5 ‡ | 63.0 ± 4.6 [1] | 60.9 ± 5.1 ∗ , † | NS | <.05 | NS |
| A4C longitudinal strain (%) | −19.9 ± 2.2 [2] ‡ , § | −18.9 ± 2.4 [4] ‡ | −17.9 ± 2.0 [1] ∗ | −17.2 ± 2.6 [1] ∗ , † | NS | <.0001 | NS |
| A2C longitudinal strain (%) | −20.6 ± 2.2 [3] ‡ , § | −19.8 ± 2.4 [4] | −18.7 ± 2.1 [1] ∗ | −18.8 ± 2.7 [3] ∗ | NS | <.001 | NS |
| GLS (%) | −20.0 ± 1.9 [3] ‡ , § | −19.4 ± 2.2 [4] ‡ | −18.3 ± 1.8 [1] ∗ | −17.9 ± 2.4 [3] ∗ , † | NS | <.0001 | NS |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
