No impact of sex and age on beta-adrenoceptor-mediated inotropy in human right atrial trabeculae

2.1 Clinical characteristics

We analysed data from 535 trabeculae obtained from 165 patients. Samples were obtained from patients undergoing isolated coronary bypass grafting (CABG; n = 89, 53.9%), valvular surgery (n = 54, 31.7%) or combined CABG and valve procedures (n = 22, 13.4%). The mean patient age was 67 ± 11 years, and 26% of patients were women. Mean body mass index (BMI) was 27.6 kg/m². One third of patients had diabetes mellitus. The main indication for cardiac surgery was coronary artery disease (67%), followed by heart valve disease. Left ventricular end-diastolic diameter (LVEDD; 51.4 mm) and left ventricular ejection fraction (LVEF; 53.4%) were preserved, and left ventricular end-diastolic pressure (16.8 mmHg) and left atrial size (43.2 mm) only slightly increased. The majority of patients were treated with ACE-inhibitors or AT1-blockers. One third of patients were on beta-blockers. Details regarding clinical characteristics are given in Table 1.

2.2 Univariable analysis

First, a univariate analysis including all 535 trabeculae from 165 patients was performed. There was no effect of age or sex on beta-AR sensitivity, basal force and maximum force. The same holds true for BMI and diabetes (Table 2). Hyperlipidaemia was associated with higher sensitivity of beta-ARs. Impairment of left ventricular function was associated with a decreased beta-AR sensitivity (lesser sensitivity [=higher EC₅₀ value of beta-AR agonists] with larger LVEDD and with smaller LVEF). The presence of valve disease was associated with a lower basal force, but without effects on maximum force. Pretreatment with diuretics or carvedilol was associated with reduced beta-AR sensitivity, without effect on force values. All other variables did not have a significant influence on beta-AR sensitivity or force measurements in the univariate analysis (Figure 1).

2.3 Multivariable analysis

For multivariable analysis, all patients with missing clinical variables had to be excluded. About 221 muscles from 65 patients could be used for multivariable analysis. There was no effect of age and sex neither on beta-AR sensitivity nor on basal or stimulated force (Table 3, Figures 2-4). In contrast to the univariate analysis, now a clear association between beta-blocker treatment and beta-AR sensitivity was observed (Table 3, Figure 3). Beta-blocker treatment (exception carvedilol) was associated with a higher beta-AR sensitivity. The beta-blocker treatment-associated increase in beta-AR sensitivity was higher for beta₂-AR (activated with adrenaline) than for beta₁-AR (activated with noradrenaline, Figure 2), but was independent of age and sex (Figure 3). In line with the findings from the univariate analysis, carvedilol was associated with a marked decrease of the apparent beta-AR sensitivity with larger effects on beta₂-AR than on beta₁-AR (Figure S1). Effects of hyperlipidaemia, LVEDD, LVEF and diuretics on beta-AR-mediated sensitivity that were seen in the univariate analysis could not be confirmed in multivariable modelling (Figure S2).

3.1 Principal findings

In our in vitro study of human atrial heart muscle samples, we did not find evidence for age- or sex-related effects on beta-AR sensitivity. Patients in our study had common cardiovascular pathologies such as coronary artery disease, valvular heart disease, hypertension, obesity and diabetes. Furthermore, patients with heart failure and reduced left ventricular ejection fraction were included. Therefore, our cohort should be a representative of contemporary patients with CVDs.

3.2 Methodological context: application of multivariable models

Using a multivariable analysis, we measured the impact of clinical variables, including age and sex, on functional properties of human heart tissue under in vitro conditions. Multivariable analysis using mixed models represents state of the art when the impact of clinical characteristics is measured in epidemiological studies. Mixed models have remained uncommon in the analysis of in vitro studies. In this context, it should be noted that experimental evidence for an age-dependent decline in cardiac beta-AR sensitivity comes from smaller size studies,^{1, 7} not allowing application of multivariable analysis. Sex-dependent differences in beta-AR inotropy are expected from animal studies with even less individuals.

3.3 Methodological context: relevance of findings in atrial tissue

Our results obtained in right atrial tissue cannot be translated one-to-one to the situation in left or right ventricles. In atrial tissue, sensitivity of β-AR to produce positive inotropic effects is around 0.7 log units higher than in ventricles in the hearts of mice and rat.^{6, 7} Reason for that difference is unclear. The same hold true for human tissue.^{8, 9} However, results are hard to interpret since β-AR remodelling in ventricular vs atrial tissue may differ between individual patients. It is almost impossible to get right atrial trabeculae and ventricular tissue from the same patient. To best of our knowledge, there is only one study where such a head-to-head comparison was done. Authors found heart failure-associated decrease in β-AR density in ventricles nicely paralleled in left and right atrial tissues.¹⁰ This finding makes us confident that atrial tissue can be useful to investigate basic mechanisms of β-AR physiology. Further work is needed to verify if age does not affect β-AR sensitivity in human ventricular tissue.

3.4 Methodological context: heart rate response as a measure for β-AR sensitivity

There are some doubts about taking heart rate responses upon stress/exercise as a measure for β-AR sensitivity. Heart rate increase in aged people is decreased and often taken as an indication for reduced β-AR sensitivity. However, activity of major contributors of pacemaking in mice heart (L-Type and T-type Ca²⁺ channels as well as funny channels) is decreased in old mice, while response of the respective channels to β-AR stimulations is nicely preserved.¹¹ These findings indicate that intrinsic electrophysiological properties of sinus node cells rather than desensitized β-AR are involved in blunted heart rate response. Situation is less clear for diabetes. In sinus node cells of diabetic rats, not only the expression for gene coding for L-Type and T-type Ca²⁺ and funny channels but also for NCX is reduced,¹² and data on β-AR responses of the channels are not yet available. In humans, the rate-lowering effect of the HCN blocker ivabradine was correlated with the intrinsic heart rate¹³ and exercise training reduces heart rate via downregulation of HCN4.¹⁴ Collectively, these results may explain the apparent discrepancy between our results and clinical reports about decreased heart rate response in elderly people, patients with diabetes or restricted exercise capacitance.^15-17

3.5 Clinical context: age

Physical working capacity declines with age. Smaller contribution of sympathetic nervous system activity to adaptation was suspected from a smaller propranolol-sensitive component of exercise-induced increase in cardiac output in old vs young healthy individuals. Since catecholamine levels were found to increase with age,¹⁸ it seemed reasonable to suspect desensitization of beta-AR with age. In line with this assumption, heart rate increase upon stimulation by exogenous agonists was consistently smaller in older healthy people. However, increase in heart rate and cardiac output upon activation of beta-AR may be affected by several autonomous reflex mechanisms, impeding interpretation of these findings.

Measurements in isolated cardiomyocytes seem ideal to elaborate changes in intrinsic beta-AR function. Cell shortening response upon isoprenaline was smaller in cardiomyocytes isolated from unused donor hearts obtained from older individuals.¹ The reduction of isoprenaline response was restricted to cells from the left ventricle. There were no age-dependent changes in myocytes isolated from right ventricles. The age-dependent reduction in isoprenaline effects was later confirmed in intact isometrically contracting right ventricular trabeculae, again from unused donor hearts. Interestingly, the 50% reduction in maximum isoprenaline effect came along with a 10-fold higher concentration of isoprenaline to evoke half-maximum effects (EC₅₀) in trabeculae obtained from older humans.² In the same year, the 10-fold higher EC₅₀ for isoprenaline to evoke positive inotropy was confirmed by a retrospective study using atrial instead of ventricular human tissue. In this study, no data about maximum effects of isoprenaline have been reported.³ However, it should be noted that later work using cell shortening in isolated ventricular cardiomyocytes failed to reproduce the higher EC₅₀ for isoprenaline in cells from older healthy donors.¹⁹

In our study, we neither saw an increase in EC₅₀ for noradrenaline and adrenaline to evoke positive inotropy nor a decrease in maximum effects of either compounds. Discrepancies to the previously published studies are difficult to explain. Methodological differences should be of minor importance, since the two studies reporting higher EC₅₀ values in tissue from older donors^{2, 3} used an almost identical experimental approach: isometrically contracting muscle strips. However, the two above-mentioned studies used rather small sample sizes from a heterogeneous patient population. Younger donors/patients including children were used as tissue donors, in this study. The authors stated: ‘The changes we observed in this study may not have been due to ageing process but instead may have occurred during the maturation process. We cannot rule out this possibility, and it will be rather impossible to obtain access to tissues from non-failing human hearts with a more appropriate age distribution’.³ This was an almost prophetic statement, since up to now there is no follow-up study that has solved the issue.

In our present study, we used a different approach. Because of restricted access to healthy human tissue, we collected a large database from patients in sinus rhythm requiring open heart surgery and used a multivariable statistical model, commonly used in epidemiological studies, to adjust for potential bias due to heterogeneous baseline patient characteristics. As a result, we here provide clear evidence that age does not reduce sensitivity of beta-AR in human hearts, neither in men nor in women.

3.6 Clinical context: sex

Most of our knowledge about sex effects on beta-AR function in cardiomyocytes comes from small animal studies. Mouse, rat, guinea pig and kitten were used for that purpose and gave conflicting results even within the same species. Basal force was reported lower in ventricular preparations from females compared to males in some studies, while others found no difference. The density of beta-AR was reported to be lower in rat ventricular cardiomyocytes from females than males, with lower increase of cAMP and calcium currents upon high concentrations of isoprenaline.²⁰ Later studies found only a trend towards lower beta-AR density in ventricular myocytes from females (same species: rat).²¹ The sensitivity to isoprenaline to evoke positive inotropy in intact muscle preparations from rats was not different in the majority of studies.^22-25 Only one study reported markedly lower sensitivity for isoprenaline in female vs male rat ventricular tissue.²⁶ No difference in beta-AR density and in maximum inotropy with isoprenaline stimulation was seen between donor heart tissues obtained from male vs female humans (no data on beta-AR sensitivity from this study).²⁷

3.7 Clinical context: obesity and diabetes mellitus

The mean BMI of patients in our study was 27.7 kg/m². One third of patients had diabetes mellitus. Despite relevant prevalence of obesity and diabetes mellitus, we could not detect any impact of either entity on basal force or beta-AR-evoked inotropy. This finding may be surprising since many studies report a negative impact of both obesity and diabetes mellitus on cardiac performance.^{28, 29} Our results should be interpreted with caution, since clinical and basic science reports focus on ventricular function and our study was done on atrial tissues. It remains unclear whether obesity/diabetes-related pathophysiological changes affect ventricles and atria to a similar extent. On the other hand, our findings may indicate that functional impairment in obesity and diabetes may not relate to remodelling of the myocardium but rather to functional effects of secretory products released from adipocytes.^30-32

3.8 Clinical context: beta-blocker treatment

Beta-blockers are a mainstay in the treatment of heart failure. In patients with advanced heart failure, beta-blocker treatment was associated with increased myocardial mRNA expression for SERCA and α-myosin heavy chain. Chronic treatment with metoprolol has been shown to reverse beta-AR downregulation in heart failure patients.^{32, 33} However, it is known that beta-blocker treatment sensitizes beta-AR, as described almost 30 years ago.^{34, 35} Both studies reported large effects of beta-blocker treatment on beta₂, but not on beta₁-AR, coming to the possibly wrong conclusion that beta-blocker-induced sensitization is selective for beta₂-AR. From our larger study, we can conclude that both beta₁- and beta₂-AR are sensitized with a significantly larger effect size on beta₂-AR. Theoretically, sensitization of beta-AR by beta-blocker treatment could help to overcome the desensitization of beta-AR in heart failure patients. However, the amount of sensitization of beta-AR we and others found^{34, 35} (about half a log unit) is in the same order of magnitude of the expected rightward shift of the concentration-response curve by metoprolol at typical plasma concentrations,⁹ indicating no net effect at rest. This may change under exercise, but the question whether beta-AR sensitization contributes to the beneficial effect of chronic beta-blocker treatment in heart failure remains open. In any case, our results clearly indicate that neither age nor sex affect chronic effects of beta-blockers on beta-ARs. Regarding carvedilol, we can confirm our earlier findings that beta-AR blockage by carvedilol persists under in vitro condition with larger effect on beta₂-AR than on beta₁-AR³⁶ qualifying carvedilol as a beta-blocker that preferentially inhibits beta₂-AR. Carvedilol but not metoprolol reduces beta-adrenergic responsiveness after complete elimination from plasma in vivo,³⁷ suggesting that carvedilol accumulates in sarcolemma or sticks on β-AR.³⁶ The data are also well in line with earlier studies showing a persistent antagonism by carvedilol.^{37, 38}

In conclusion, our contemporary data in a large cohort suggest that no clinically relevant age or sex differences exist in the treatment with beta-blockers. Clinicians should be encouraged to provide guideline-recommended beta-blocker therapy across age groups and in both genders.

5.1 Drugs

CGP 20712A was from Novartis (Basel, Switzerland). ICI 118 551 was from Tocris (Bristol, UK) and (−)-adrenaline and (−)-noradrenaline were from Sigma Chemie. Phenoxybenzamine was from Röhm Pharma.

5.2 Statistics

−LogEC₅₀M values for the catecholamines were estimated from fitting a Hill function with variable slopes to concentration-effect curves from individual experiments using GraphPad Prism 5 software (GraphPad Software Inc.). Patient characteristics are presented as mean ± standard deviation and as absolute and relative frequencies. Mixed models were fit in order to find associations between patient and/or measurement characteristics and measurement (‘log_10_Control’, ‘log_10_EC50’ and ‘log_10_E_max’) and multivariable mixed model regressions were performed.⁴³ Outliers were identified by visual inspection of frequency histograms for ‘log_10_Control’, ‘log_10_EC50’, ‘log_10_E_max’ and removed (four muscles with logEC₅₀ <-9 and two muscles with log_10_Control<25). Experimental values (‘log_10_EC50’, ‘log_10_Control’ and ‘log_10_E_max’) were transformed before being used in the fitting algorithm in order to improve the normality of the residuals of the fit, using the following polynomial functions (found by fitting the measured values to quantile normalized values):

The beta-coefficients obtained in the mixed model regressions with the above transformed values represent the change of transformed measurement value per unit change in patient characteristic. Here we reported beta-coefficients which represent an increase of untransformed measurement value per unit increase of patient characteristic, by dividing the mixed model beta-coefficients by slope of the polynomial at the average measurement value as a first order approximation.