Myocardial region (right or left ventricle) and aetiology of heart failure can influence the inotropic effect of ouabain in failing human myocardium

Abstract

Aims

To investigate whether the inotropic effect of ouabain in failing human myocardium varies according to the heart chamber tested (right or left ventricle) or the aetiology of the heart disease, i.e. ischaemic or idiopathic.

Methods

The inotropic effect of ouabain was measured, as the percentage change in baseline tension, in myocardial strips isolated from right (RV; n = 21) and left ventricles (LV; n = 21) of hearts explanted from patients with idiopathic (IDC; n = 11) and ischaemic cardiomyopathy (CAD; n = 10). Concentration-effect curves obtained with ouabain (0.05–1.6 μmol l⁻¹) were analysed using the E_max sigmoidal model, and the following parameters were calculated: E_max, EC₅₀, n and EC₁₀ (threshold concentration). The influence of ventricular chamber and heart failure aetiology on these parameters was evaluated by means of a two-way anova.

Results

Age and baseline haemodynamic parameters did not differ between IDC and CAD patients. Baseline strip contractility was highly variable (range: 0.48–10.0 mN), but neither ventricular chamber nor aetiology could explain such variability. A two-way anova showed that EC₁₀ was greater in CAD than in IDC preparations (0.097±0.013 μmol l⁻¹vs 0.059±0.009 μmol l⁻¹; 95% C.I. for difference 0.043, 0.071) and E_max was lower in RV than in LV (121±21%vs 250±38%; 95% C.I. −221, −36), while EC₅₀ and n were not significantly different between groups.

Conclusions

The inotropic effect of ouabain in human myocardium may vary according to aetiology of heart failure and the ventricle being tested. Although our results do not support the hypothesis of increased sensitivity to cardiac glycosides in CAD patients, they may explain the diminished effect observed in patients with RV failure.

Introduction

The role of cardiac glycosides in the management of chronic heart failure has long been questioned because of uncertainty about their clinical benefits and safety [1–4]. There is now evidence that digoxin is still effective during long-term treatment [5, 6], in patients with advanced disease [7] but does not influence mortality [8]. Despite favourable results due to digoxin treatment, some concern still remains regarding its tolerability and/or efficacy in specific forms of heart failure. On the basis of animal studies [9–11] and clinical reports [12–14], a detrimental effect of long-term digitalis treatment has been suspected in patients suffering from myocardial ischaemia. It has also been suggested that patients with predominant right ventricular dysfunction (e.g. cor pulmonale) are resistant to the inotropic effect of digoxin [15–17] and may be predisposed to toxicity [18]. In both clinical situations, extracardiac factors, such as hypoxia, acidosis, and ionic alterations, may explain the variability in drug effect. However, it cannot be excluded that intrinsic myocardial responsiveness to digitalis may differ according to the specific aetiology of heart failure or the myocardial chamber considered (right or left ventricle).

In order to test the hypothesis that there may be disease-or chamber-related differences in intrinsic myocardial sensitivity to cardiac glycosides, we evaluated the in vitro inotropic effect of ouabain on cardiac strips of right and left ventricles obtained from hearts explanted from patients affected by postischaemic or idiopathic cardiomyopathy. Ouabain was preferred to digoxin because it has a greater water-solubility and is the reference compound for in vitro studies.

Methods

Cardiac muscle samples were obtained from the right and left ventricles of 21 patients (17 males and 4 females) undergoing orthotopic heart transplantation, who gave their oral informed consent before operation. Eleven patients suffered from idiopathic dilated cardiomyopathy (IDC) and 10 from coronary artery disease (CAD). The diagnosis of IDC was based on the exclusion of all the known causes of cardiomyopathy by means of clinical history, angiographic and laboratory criteria. Patients’ ages ranged from 15 to 67 years (mean±s.e. mean 51.9±1.8 years). The following haemodynamic variables were available from routine cardiac catheterization: mean right atrial pressure (MRAP), mean pulmonary artery pressure (MAP), mean pulmonary ‘capillary’ wedge pressure (MPCWP), cardiac index (CI) and ejection fraction (EF). All patients were on chronic treatment with a combination of digoxin, diuretics and ACE inhibitors. Twelve were also taking nitrates (10 in the CAD group) and 5 were on amiodarone for dysrhythmias (3 in the IDC group).

Forty-five minutes before surgery, patients were given a 2 mg oral dose of flunitrazepam. Anaesthesia was induced by i.v. administration of fentanyl (1–10 mg kg⁻¹), droperidol (5 mg) and vencuronium (0.1 mg kg⁻¹) and maintained with 10 mg kg⁻¹ fentanyl boluses (up to a total dose of 40–50 mg kg⁻¹) and 2 mg h⁻¹ pancuronium. Hypnosis was obtained with propofol (2 mg kg⁻¹ h⁻¹) or 0.4–0.6% isoflurane.

Immediately after heart removal, the posterior papillary muscles of both ventricles were excised and put into an ice-cold oxygenated solution (see below for composition) until the experiments could be performed (usually within 15–30 min). With the aid of a dissection microscope, tissue strips, parallel to fibre direction, were prepared (length×diameter = about 7–9 mm×1 mm), macroscopically fibrotic areas being avoided. The endocardium was removed in all cases. Preparations were suspended vertically in a 15 ml Blinks bath at 35 °C and connected to an isometric force transducer (F30, Hugo Sachs Electroniks). The composition of the physiological salt solution was (mmol l⁻¹): NaCl 138; KCl 4; CaCl₂ 2.5; MgCl₂.6H₂O 1; HEPES 5; Na pyruvate 2; glucose 10. The pH was adjusted to 7.35 with NaOH and the solution was bubbled with 100% oxygen. The strips were field-stimulated by square-wave impulses (duration = 5 ms; voltage 20% above threshold) at 0.5 Hz. During the equilibration period (1.5–2 h), the muscles were stretched to the length at which the force of contraction was maximal.

The inotropic effect of ouabain was measured at concentrations of 0.05, 0.1, 0.2, 0.4, 0.8 and 1.6 μmol l⁻¹. The drug was cumulatively added to the bath after apparent equilibration of the effect of the previous dose (i.e. every 30 min), until the maximum inotropic effect was reached or toxicity occurred (i.e. reduction in peak tension, increase in diastolic tone or occurrence of premature contractions). The drug effect was evaluated as a percentual change in baseline peak tension. The concentration-effect curves could be described by the E_max sigmoidal model:

1

where E_max is the maximal peak tension increase, EC₅₀ the concentration producing 50% of E_max and n is a shape factor.

Equation 1 was fitted to the individual concentration-effect data using the nonlinear best fit programme of GraphPad Prism software, and the optimal value of each parameter was obtained.

The threshold for the appearance of the drug effect was defined as the concentration producing 10% of E_max (EC₁₀), and was calculated by rearranging. (1) as follows:

2

Data are presented as means±s.e. mean. Student’s unpaired t-test was used to compare the age and haemodynamic parameters of CAD and IDC patients. Differences in baseline contractility of CAD right ventricles (CAD-RV), IDC right ventricles (IDC-RV), CAD left ventricles (CAD-LV) and IDC left ventricles (IDC-LV) were checked by means of a one-way anova, followed by the Newman-Keuls posthoc test. In order to verify whether sampling site (right or left ventricle) and/or aetiology of the disease (CAD or IDC) could affect baseline contractility and/or the response to ouabain, a two-way anova was applied to data from the four experimental groups (CAD-RV, IDC-RV, CAD-LV and IDC-LV). Significance level was P <0.05. When appropriate, 95% confidence intervals (95% C.I.) are indicated.

This study was approved by the ethics committee of our Medical School.

Results

The age and haemodynamic characteristics of patients with CAD and IDC did not differ significantly, although mean age tended to be greater and ejection fraction lower in the CAD group (Table 1). Mean baseline tension developed by all ventricular preparations tested (n = 42) was 4.42 mN (±0.46 mN). The one-way anova indicated that some difference between subgroups was present (P = 0.049) (Table 2), but post hoc test failed to detect any significant difference between specific group pairs. Two-way anova did not reveal an influence of the disease (IDC or CAD) or the sampling site (LV or RV) in explaining intergroup variability.

Table 1.

Haemodynamic parameters (means±s.e.mean).

Table 2.

Mean baseline tension (mN).

Ouabain increased peak tension dose-dependently in most, but not all, preparations. Indeed, in three out of 11 experiments with IDC-RV, ouabain did not elicit any evident change in contractility, up to the highest concentration tested (1.6 μmol l⁻¹). These preparations were therefore excluded from the next analysis.

The coefficient of correlation (r) between the E_max-sigmoidal model and data was in every case greater than 0.99. The mean values (±s.e.mean) of the parameters (E_max, EC₅₀, n, EC₁₀) obtained in the four experimental groups are shown in Figure 1. The two-way anova (Table 3) demonstrated that sampling site and disease aetiology were able to explain some variability between groups, i.e. (a) E_max was smaller in right ventricles than in left ventricles (121±21%vs 250±38%; P = 0.009; 95% C.I. −221, −36); (b) EC₁₀ was greater in CAD than in IDC preparations (0.097±0.013 μmol l⁻¹vs 0.059±0.009 μmol⁻¹; P = 0.036; 95% C.I. 0.043, 0.071). Although EC₅₀ and n tended to be greater in CAD than in IDC preparations (EC₅₀ = 0.264±0.049 μmol l⁻¹vs 0.177±0.022 μmol l⁻¹; C.I. −0.025, 0.196; n = 2.88±0.31 vs 2.15±0.20; C.I. −0.011, 1.49), the difference was not significant (P = 0.131 and 0.070, respectively).

Figure 1.

Mean values (±s.e. mean) of E_max, EC₅₀, EC₁₀and n obtained from concentration/effect curves for ouabain, in patients with idiopathic dilated cardiomyopathy (IDC) and ischaemic cardiomyopathy (CAD). Open bars: right ventricle preparations; hatched bars: left ventricle preparations.

Table 3.

Two-way anova: P values.

For better visualization of the results, the mean E_max, EC₅₀ and nvalues of each group were substituted in Equation 1. The resulting concentration-effect curves are shown in Figure 2.

Figure 2.

Concentration/effect curves for ouabain, obtained in right (RV) and left ventricles (LV) of patients with idiopathic dilated cardiomyopathy (IDC) and ischaemic cardiomyopathy (CAD), by substituting mean parameter values in E_max model equation. Continuous lines: IDC data; dashed lines: CAD data.

Discussion

Several studies have been performed in human and experimental heart failure to investigate possible alterations in digitalis action. Most previous in vitro investigations on human myocardium have reported that the inotropic effect of cardiac glycosides, like that of external calcium, is substantially similar in failing and nonfailing hearts [7, 18, 19]. One study [20] found that trabeculae with more severe morphological alterations were more sensitive to ouabain (lower EC₅₀) and exhibited a lower maximum response to both ouabain and calcium than preparations with moderate alterations. Other authors [21] observed that the time to reach the peak effect of ouabain was shorter in failing than in nonfailing myocardium, and interpreted this as an increase in drug potency. Studies at a molecular level in diseased human hearts have noted that the density of the ‘digitalis receptor’, sarcolemmal Na/K-ATPase, is either unchanged [6, 7, 22] or decreased [20, 23–26], whereas the mRNA expression of the three Na/K ATPase α subunits is found to be unchanged [20, 27]. Similar studies in experimental heart failure models have consistently reported that Na/K-ATPase density (or activity) is reduced [28–31] and α3-subunit expression is decreased [31, 32]. It has also been shown that sympathetic hyperactivity associated with heart failure can modulate digitalis receptor density and is at least partly responsible for its downregulation [31, 32], and that β-adrenoceptor blocker pretreatment can prevent these effects [30].

To our knowledge, no direct evidence has so far been provided that the inotropic effect of ouabain in failing human hearts varies according to the aetiology of the disease or the ventricular chamber tested. One of the most important results of our study is that, irrespective of the ventricle considered, the ouabain threshold concentration (EC₁₀) is about two times higher in the ischaemic than in idiopathic cardiomyopathy. This difference cannot be justified by inhomogeneity in baseline twitch tension or drug treatment. Although statistical analysis did not reveal significant differences in age or haemodynamic parameters, the two groups are not perfectly matched for age and ejection fraction (Table 1); CAD patients tended to be older (P = 0.114) and to have a lower ejection fraction (P = 0.058) than IDC patients. This factors may have contributed to the difference in EC₁₀. Since the EC₅₀ also tends to be higher (although not significantly), the concentration-effect curves of ischaemic cardiomyopathy are shifted to the right with respect to those obtained with idiopathic cardiomyopathy (Figure 2). Thus, the original hypothesis of the steady hypersensitivity of CAD myocardium to digitalis is not supported by our results. This, of course, does not exclude that an unfavourable interaction between myocardial ischaemia and digitalis may occur during acute ischaemic episodes. Although no clinical observation suggests reduced sensitivity of CAD patients to the inotropic effect of digitalis, some experimental data are in keeping with this possibility. Dixon et al. [33], who induced heart failure in rats by ligating the left ventricular coronary artery, reported that the affinity of ouabain for the high-and low-affinity ATPase binding sites decreased 8 and 16 weeks after myocardial infarction, whereas no change in maximal binding capacity occurred. Another explanation for the rightward shift of the concentration-effect curve of ouabain in CAD patients may be the greater density of sarcolemmal Na/K ATPase sites, although no direct demonstration of this has so far been provided. However, in view of the inverse relationship between adrenergic stimulation and ATPase activity in experimental heart failure, it is possible that a different sympathetic drive (and, consequently, a different degree of ATPase down-regulation) exists in CAD and DCM hearts. Indeed, Bristow et al. [34] found a decreased contractile response to the β-adrenoceptor agonist isoprenaline in RV trabeculae of CAD compared with IDC patients. The same research group also reported a smaller improvement in cardiac function by β-adrenoceptor blocking agents in CAD than in IDC patients [35]. It is tempting to suggest that the blunted response to isoprenaline observed in CAD, is associated with higher Na/K ATPase levels and, accordingly, to decreased sensitivity to ouabain. Further studies on this aspect are clearly needed.

Another noteworthy finding of our study is that right ventricles of both CAD and DCM patients show a smaller maximal response to ouabain than left ventricles. This finding is reinforced by the lack of any detectable effect in three out of 11 ICM-RV preparations. From a general point of view, this difference may be ascribed either to a native characteristic of the two ventricles or to different adaptation of ventricles to heart failure. Our data do not allow us to discriminate between these possibilities. Nevertheless, experimental evidence indicates that the two ventricles do have peculiar biochemical and physiological properties [36], which may in turn lead to differing sensibility to drugs. The changing inotropic responses of right and left ventricles to drugs have been reported in vitro for ryanodine [37] (less negative inotropic effect in LV than in RV of DCM patients) and in vivo for noradrenaline [38] (inotropic effect more pronounced in RV than in LV, in rats).

One explanation for the differential ouabain effect is that the left ventricle has relatively lower density of ‘digitalis receptor’ than the right. Indeed, a study [39] on autoptic specimens of normal-weight and hypertrophied hearts found ouabain binding reduced by 25–35% in the left but not in the right ventricle of hypertrophied hearts. However, using fresh tissue from explanted hearts, Shamraj et al. [20] were not able to note any difference in ouabain binding between the ventricles, and Sylven et al. [27] found that mRNA expression of the three Na⁺/K⁺ ATPase α subunits was similar in right and left ventricles of both normal and DCM human hearts. Thus, the ‘ATPase hypothesis’ has weak, if any, experimental support. Alterations of many other cellular processes may influence the inotropic effect of digitalis, such as: (1) Na⁺/Ca²⁺ exchange activity; (2) Ca²⁺ uptake from the sarcoplasmic reticulum (Ca²⁺ ATPase); (3) Ca²⁺ release from the sarcoplasmic reticulum (ryanodine receptor); (4) Ca²⁺ interaction with the contractile machinery. Although Na⁺/Ca²⁺ exchange expression and activity have been found to be increased in failing hearts [40, 41], no information is available about possible ventricle-related differences. Various authors confirm that both Ca²⁺ ATPase [42–44] and Ca²⁺ release channel expression [44, 45] is lower in failing than in nonfailing hearts, but no significant differences are detectable between right and left ventricles in either group [44]. Also, the inotropic response to Ca²⁺ was found to be similar in nonfailing and failing human hearts [7, 18, 19, 46] and in right and left ventricles [47]. In summary, available evidence in human disease does not explain our observations at a molecular level. In a clinical trial, Mathur et al. [16] showed that, in patients with pulmonary heart disease and depressed right ventricular contractility, 8-week digoxin treatment did not increase the right ventricular ejection fraction, while left ventricle function was significantly improved. This mean improvement was largely due to a subgroup of patients with concomitant left ventricle dysfunction, who also showed a slight but significant increase in the right ventricle ejection fraction. Since right ventricle performance is considered highly dependent on afterload changes, the above authors suggested that the drug-induced improvement in left ventricle function reduced right ventricle afterload. Although this view seems reasonable, our findings of a differential digitalis effect in the two ventricles provide a more straightforward explanation for these clinical data.

In conclusion, the present study demonstrates that the in vitro response to ouabain of human diseased myocardium may vary widely between preparations, owing to heart failure aetiology and ventricle-related factors. Thus, our results, together with others previously published, strengthen the idea that the heart is physiologically and pharmacologically inhomogeneous, so that observations made in a specific myocardial region or disease cannot be automatically extended to other regions or diseases. From a clinical point of view, our data provide experimental support for the asserted resistance of right-sided heart failure to digitalis, but do not explain the greater sensitivity of CAD patients, at least in the absence of ischaemic episodes. Further in vitro experiments on human CAD myocardium are warranted to ascertain whether, as in animal models, an unfavourable interaction exists between digitalis and ischaemia.