Author + information
- Received April 23, 2002
- Revision received May 28, 2002
- Accepted July 15, 2002
- Published online November 6, 2002.
- Michel F Rousseau, MD, PhD, FACC*,* (, )
- Olivier Gurné, MD, PhD*,
- Daniel Duprez, MD, PhD, FACC†,
- Walter Van Mieghem, MD, PhD‡,
- Annie Robert, PhD§,
- Sylvie Ahn*,
- Laurence Galanti, MD, PhD*,
- Jean-Marie Ketelslegers, MD, PhD∥,
- Belgian RALES Investigators
- ↵*Reprint requests and correspondence:
Dr. Michel F. Rousseau, Division of Cardiology, University of Louvain, Avenue Hippocrate 10/2800, B-1200 Brussels, Belgium.
Objectives We sought to evaluate the effects of spironolactone on neurohormonal factors in patients with severe congestive heart failure (CHF).
Background In the Randomized ALdactone Evaluation Study (RALES), spironolactone, an aldosterone receptor antagonist, significantly reduced mortality in patients with severe CHF. However, the mechanism of action and neurohormonal impact of this therapy remain to be clarified.
Methods The effects of spironolactone (25 mg/day; n = 54) or placebo (n = 53) on plasma concentrations of the N-terminal portion of atrial natriuretic factor (N-proANF), brain natriuretic peptide (BNP), endothelin-1 (ET-1), norepinephrine (NE), angiotensin II (AII), and aldosterone were assessed in a subgroup of 107 patients (New York Heart Association functional class III to IV; mean ejection fraction 25%) at study entry and at three and six months.
Results Compared with the placebo group, plasma levels of BNP (−23% at 3 and 6 months; p = 0.004 and p = 0.05, respectively) and N-proANF (−19% at 3 months, p = 0.03; −16% at 6 months, p = 0.11) were decreased after spironolactone treatment. Over time, spironolactone did not modify the plasma levels of NE and ET-1. Angiotensin II increased significantly in the spironolactone group at three and six months (p = 0.003 and p = 0.001, respectively). As expected, a significant increase in aldosterone levels was observed over time in the spironolactone group (p = 0.001).
Conclusions Spironolactone administration in patients with CHF has opposite effects on circulating levels of natriuretic peptides (which decrease) and aldosterone and AII (which increase). The reduction in natriuretic peptides might be related to changes in left ventricular diastolic filling pressure and/or compliance, whereas the increase in AII and aldosterone probably reflects activated feedback mechanisms. Further studies are needed to link these changes to the beneficial effects on survival and to determine whether the addition of an AII antagonist could be useful in this setting.
Congestive heart failure (CHF) is characterized by the progressive activation of several endocrine systems (1). Increased levels of norepinephrine (NE) and natriuretic peptides, such as atrial natriuretic peptide and brain natriuretic peptide (BNP), as well as activation of the renin-angiotensin-aldosterone system (RAAS), have been described and associated with a poor prognosis (2,3). Aldosterone promotes the retention of sodium and the loss of potassium, activates the sympathetic nervous system, stimulates the development of myocardial and vascular fibrosis, and causes endothelial and baroreceptor dysfunction (4,5). Angiotensin II (AII), which increases sympathetic drive, aldosterone release, and cardiac remodeling, is known to exacerbate the progression of CHF (6). Despite progress achieved with angiotensin-converting enzyme (ACE) inhibitors and, more recently, with beta-blockers, mortality and morbidity remain high in patients with severe CHF (7). Pitt et al. (8)demonstrated that a low dose of spironolactone, an aldosterone receptor antagonist, significantly reduced the risk of morbidity and mortality in patients with severe CHF included in the Randomized ALdactone Evaluation Study (RALES). However, the mechanism(s) by which spironolactone affects survival remain(s) unclear. Recently, the beneficial effects of spironolactone on markers of tissue collagen turnover and fibrosis were reported in a substudy of RALES (9). The aim of the present substudy was to assess, in the Belgian cohort enrolled in the RALES trial, the effects of spironolactone on several neurohormonal markers of the severity of CHF—namely, natriuretic peptides, NE, endothelin-1 (ET-1), and AII.
Patients and study design
Of the 130 patients from Belgian centers who were randomized into the RALES trial, we prospectively included 127 patients in the neurohormonal substudy. Patients with the following inclusion criteria were eligible for enrollment: a history of New York Heart Association (NYHA) functional class IV within the previous six months and class III or IV at the time of randomization; full treatment with ACE inhibitors and loop diuretics; and left ventricular ejection fraction <35%. Patients were randomized to receive placebo or spironolactone (25 mg/day). The mean survival follow-up period was 24 months. Neurohormonal plasma samples were obtained at baseline and after three and six months of treatment. Of the 127 patients with baseline neurohormonal data, 20 were excluded from the follow-up because of premature death (n = 8) or dropout (n = 12). Thus, data on 107 patients (Table 1) were available for analysis at baseline and at a minimum of one follow-up time point. The number of samples available in each treatment group at any time point is listed in Table 2. Each patient gave written, informed consent, and the protocol was approved by the local institutional Review Board.
Measurements of neurohormonal markers
Venous blood samples were obtained after 30 min of rest in the supine position. Blood samples were collected in chilled tubes containing 3 mmol/l EDTA and 9 mmol/l benzamidine and stored on wet ice until centrifugation. Plasma was carefully separated and frozen at −80°C. Briefly, 3 ml of unfrozen plasma mixed with 1 g of guanidine hydrochloride was extracted on SEP-PAK C18cartridges (Waters, Milford, Massachusetts) and eluted with 0.1% trifluoroacetic acid/80% acetonitrile. Natriuretic peptides (N-terminal portion of atrial natriuretic factor [N-proANF] and BNP), ET-1, and AII were measured on the same extracts by radioimmunoassay, using specific antibodies and synthetic peptides from Peninsula (Belmont, California), as well as home-iodinated tracers purified by high-performance liquid chromatography (HPLC) (10). There was no significant cross reactivity between the assays; samples displaced tracer parallel to their respective standard curve. The percent cross reactivities of the antiserum for AII were 0.5% for AI and 100% for AIII. Recoveries, intra-assay and inter-assay coefficients of variation, and the median effective dose (ED50) were 70%, 7% and 10%, and 135 pg/ml for N-proANF; 84%, 7% and 6%, and 350 pg/ml for BNP; 68%, 12% and 16%, and 140 pg/ml for ET-1; and 95%, 2% and 4.5%, and 110 pg/ml for AII, respectively. To provide control values for natriuretic peptides and AII, blood samples were obtained in 24 healthy age-matched control subjects; normal values expressed as the geometric mean value (95% confidence interval [CI]) were 219 (105 to 453) pg/ml for N-proANF, 11 (5 to 25) pg/ml for BNP, and 9.3 (8 to 10.6) pg/ml for AII. Normal values for ET-1 were <3 pg/ml. Serum aldosterone was measured by a commercial radioimmunoassay (Abbott Diagnostics, Chicago, Illinois). Intra-assay and inter-assay coefficients of variation were 3.7% and 6.3%, and the ED50was 0.78 nmol/ml. Normal values for aldosterone were <0.4 nmol/ml. Norepinephrine was measured by HPLC with intra-assay and inter-assay coefficients of variation of 7% and 12%, respectively. Normal values for NE in our laboratory ranged from 164 to 262 pg/ml, with a normal geometric mean value of 207 pg/ml.
Data are expressed as numbers for discrete data, as the mean value ± SD for normal continuous data, and as the geometric mean value (95% CI) for neurohormonal data due to a right-skewed distribution. The placebo and spironolactone groups were compared using the Fisher exact test for discrete data and the Student ttest for continuous data. Neurohormonal data were log-transformed before statistical comparisons. Analysis of variance for repeated measures was used to analyze neurohormonal changes over time (11). Analysis of variance was computed using a grouping factor with two levels (spironolactone or placebo) and a repeated measurement factor with three levels (0, 3, and 6 months). Because only 89 patients were measured at baseline and three and six months, the degrees of freedom (df) were 2 and 174 for the interaction F test and for the time change F test. For the grouping effect F test, the df’s were 1 and 87. Because of significant interaction tests, changes over time were tested within each treatment group. Bonferroni-corrected p values were used, adjusting for four comparisons, because we considered only changes from baseline within each of the two groups. Time changes were expressed as ratios by taking the anti-log of differences at three (T3/T0) or six (T6/T0) months from baseline. All tests were two-tailed, and p < 0.05 was considered as statistically significant.
Of the original group of 127 patients, 11 had died in the placebo group and five in the spironolactone group at the six-month time point. After a mean follow-up period of 24 months, there were 46 deaths (35%) in the study group: four were non-cardiovascular deaths and 42 were attributed to cardiovascular causes (19 sudden deaths, 19 due to worsening heart failure, 2 due to stroke, and 2 due to other cardiovascular causes). Consistent with the main trial results, cardiac mortality was lower in the spironolactone group than in the placebo group (21% vs. 38%, p = 0.05). Considering the mode of deaths, we observed a significant decrease in sudden deaths in the spironolactone group compared with the placebo group (8% vs. 22%, p = 0.026).
As summarized in Table 1, no significant differences between the placebo and spironolactone groups were observed for the 107 patients for whom neurohormonal data were available at baseline and at a minimum of one follow-up time point, except that patients randomized in the spironolactone group were slightly older than those in the placebo group (71 vs. 66 years, p = 0.02) and received less beta-blockers (8% vs. 16%, p = 0.07). Seventy-one percent of our study population had ischemic cardiomyopathy, 25% had idiopathic dilated cardiomyopathy, and 4% had valvular disease. Seventy-nine percent were in NYHA class III. Patients were treated with loop diuretics and ACE inhibitors in 97% and 95% of cases, respectively.
Table 2shows the neurohormonal measurements at baseline and after three and six months of follow-up. The severity of left ventricular dysfunction is demonstrated by the high baseline levels of NE, ET-1, BNP, and N-proANF, compared with the normal values of each respective assay. No significant difference was observed between the two groups.
As shown in Figure 1, the BNP plasma concentration, expressed in time change ratios, decreased by 23% in the spironolactone group compared with the placebo group (0.99 vs. 0.77, p = 0.004 and 0.96 vs. 0.77, p = 0.05, respectively at 3 and 6 months). A significant decrease of 19% in N-proANF was observed at three months in the spironolactone group (1.0 vs. 0.81, p = 0.03) (Fig. 2) and a decrease of 16% at six months (0.99 vs. 0.84, p = 0.11).
Plasma norepinephrine and endothelin-1
Compared with placebo, spironolactone did not significantly change the plasma levels of NE (0.98 vs. 1.03, p = 0.64; 0.96 vs. 1.07, p = 0.33) or ET-1 (1.03 vs. 0.94, p = 0.10; 1.11 vs. 1.03, p = 0.25) at three and six months, respectively.
Angiotensin II and aldosterone
Compared with placebo, AII increased significantly at three and six months in the spironolactone group (8.4 vs. 13.9 pg/ml, p = 0.02 and 7.9 vs. 13.2 pg/ml, p = 0.02). Furthermore, the AII ratios of 3 months/baseline and 6 months/baseline also rose markedly (0.78 vs. 1.41, p = 0.003 and 0.66 vs. 1.41, p = 0.001) (Fig. 3). As expected (Fig. 4), a significant increase in aldosterone levels was observed in the spironolactone group, both in absolute values and in the ratios of 3 months/baseline (0.92 vs. 1.75, p = 0.001) and 6 months/baseline (0.92 vs. 2.03, p = 0.001).
Our study demonstrated that the administration of spironolactone, an aldosterone receptor antagonist, significantly reduced the plasma levels of BNP and, to a lesser extent, N-proANF in patients with severe CHF, whereas AII and aldosterone levels increased and NE and ET-1 levels remained essentially unchanged.
Circulating plasma levels of cardiac natriuretic peptides are inversely related to the severity of left ventricular dysfunction and have been found to be prognostic predictors (10,12). More specifically, the plasma BNP level is directly correlated to changes in ventricular wall stress (13). Furthermore, natriuretic peptides have several biologic functions, including vasodilation, increased compliance in large vessels, enhanced baroreceptor sensitivity, and renal effects, particularly sodium excretion (14). Spironolactone can influence the progression of left ventricular remodeling by a reduction in interstitial fibrosis and reorganization of the collagen matrix (15–17). Thus, the reduction of BNP and N-proANF levels could be related to an improvement of left ventricular diastolic properties and/or filling pressures. Spironolactone may prevent myocardial fibrosis by blocking the aldosterone effects on collagen formation, as suggested by decreased levels of collagen markers such as the procollagen type III aminoterminal peptide (9,13). Our data are also consistent with the results of Tsutamato et al. (13), who showed, after four months of spironolactone therapy (25 mg/day), a significant decrease in BNP and ANF levels in a small group of patients with mild to moderate non-ischemic cardiomyopathy.
Levels of NE and ET-1 also identify a CHF population with a poor prognosis (18,19). The high-risk profile of our population is confirmed by the marked NE and ET-1 levels at baseline. We did not observe a decrease in NE and ET-1, and these well-established neuroendocrine prognostic markers failed to predict the beneficial survival effect, suggesting that the effects of spironolactone on mortality are related to mechanisms independent of adrenergic and endothelin systems. Similarly, the plasma levels of NE and ET-1 were not changed after spironolactone therapy in the study performed by Tsutamoto et al. (13). The role of aldosterone in sympathetic modulation is controversial, as indicated in the study of Yee et al. (20), where whole-body NE clearance and spillover did not appear to be significantly affected by spironolactone therapy compared with placebo. In that study, spironolactone reduced the heart rate and improved heart rate variability and QT dispersion in patients with CHF, and these beneficial effects seem to be related to the modulation of parasympathetic tone. Moreover, an antagonizing aldosterone effect could improve baroreflex function, which is an important determinant of sudden cardiac death. Therefore, spironolactone could possess properties able to reduce life-threatening arrhythmias (21,22).
In our study, we observed direct evidence that spironolactone therapy activated the RAAS. Spironolactone significantly increased the plasma levels of AII (and/or its metabolite AIII, as cross-reactivity existed in our assay) and aldosterone during the follow-up period. This observation raises two additional questions: first, regarding the physiologic impact of this rise in AII, and second, the mechanisms by which AII is further increased in the presence of an ACE inhibitor. One can only speculate about the physiologic consequences of the increase in angiotensin in the presence of a decrease in BNP, as well as the likely blockade of the effects of aldosterone at the cellular level. There were no obvious changes in blood pressure, plasma creatinine, or potassium levels. Furthermore, two other markers of vasoconstrictor activity—NE and ET-1—were also unchanged. Admittedly, this rise is modest, and the statistical significance is also driven by the fact that the values decreased slightly in the placebo group. However, AII can stimulate left ventricular hypertrophy and perhaps cardiac myocyte apoptosis. Accordingly, and despite the fact that the changes remained essentially within a normal range, it would be tempting to assess the changes in cardiac mass during follow-up in these patients or to determine the effects of an AII antagonist. Some benefit of an AII antagonist has been recently reported in patients already treated with an ACE inhibitor in the Valsartan Heart Failure Trial (Val-HeFT) (23). However, because only 5% of the Val-HeFT patients were receiving spironolactone, no conclusion can be drawn yet regarding the safety and efficacy of a combination of an ACE inhibitor/AII receptor blocker and spironolactone (23). With respect to the mechanism underlying the AII and aldosterone escape, this probably reflects activated feedback mechanisms on the RAAS (24). The possibility also exists that renin or ACE expression in the failing heart could contribute to this enhanced production of AII. Sun et al. (25)demonstrated that after myocardial infarction in the rat, cardiac renin production was induced and contributed to local AII generation. Mizuno et al. (26)also showed in failing ventricles that the levels of aldosterone had a highly significant positive correlation with levels of ACE activity, suggesting that increased activity of local ACE, causing conversion of AI to AII, may stimulate production of aldosterone in heart failure. Silvestre et al. (27)recently showed that cardiac aldosterone is activated in the rat heart with myocardial infarction, and that this is mediated primarily by cardiac AII. Thus, cardiac aldosterone may play a major role in the progression of heart failure, and spironolactone, an aldosterone receptor antagonist, may improve heart failure by blocking the action of locally produced aldosterone in the failing heart.
One potential limitation of this study could be a bias introduced by the unbalanced attrition of the placebo and spironolactone-treated groups and by the small imbalances noted for age and use of beta blockers at baseline. However, the numbers of deaths, dropouts, and missing samples at six months were relatively well balanced, and the baseline characteristics of the two subsets of patients, in whom six-month data were available, were not only similar but also comparable to those of the whole study group. Furthermore, differences in age and use of beta-blockers would have tended to underestimate the benefit of spironolactone and were therefore unlikely to affect the conclusions. Another potential limitation of the study relates to the specificity of the AII assay. The anti-serum used for the AII assay cross reacted with AIII, another peptide known to produce vasoconstriction and stimulate aldosterone production. Thus, because AIII is also a biologically active peptide, with effects qualitatively similar to those of AII, the high concentration detected by the anti-serum used indicates an abnormal activation of the RAAS (24). Because of the disparity of baseline BNP levels in our study compared with other studies, the specificity of the BNP assay should also be considered. The higher baseline BNP values observed in other laboratories likely resulted from utilization of commercially available assays with technical differences in the extraction procedure, standardization, and antibody affinity for various circulating forms of BNP (28). In our RALES cohort, we observed a sevenfold increase in baseline BNP levels, compared with control values, and these levels appeared consistent with severe CHF. Therefore, variability in normal values related to the type of assay used is unlikely to qualitatively affect our conclusions.
The present study indicates that spironolactone has the opposite effects on BNP and N-proANF, which are lowered during treatment, and AII and aldosterone, which are increased. The lack of a decrease in the neuroendocrine prognostic markers, NE and ET-1, suggests that the beneficial effects of spironolactone are mainly related to mechanisms independent of the adrenergic and endothelin systems. The escape of AII and aldosterone, probably reflecting activated feedback mechanisms, confirms the specific activity of spironolactone on the RAAS system and supports the hypothesis that the beneficial effects of spironolactone on the progression of heart failure are mediated by the blockade of aldosterone receptors.
☆ This study was supported in part by a grant from Pharmacia, Brussels, Belgium. With respect to a potential conflict of interest, Dr. Ketelslegers received a grant from Pharmacia Belgium for the biochemical assays.
- angiotensin II
- angiotensin-converting enzyme
- brain natriuretic peptide
- congestive heart failure
- N-terminal portion of pro-atrial natriuretic factor
- New York Heart Association
- renin-angiotensin-aldosterone system
- Randomized ALdactone Evaluation Study
- Received April 23, 2002.
- Revision received May 28, 2002.
- Accepted July 15, 2002.
- American College of Cardiology Foundation
- Francis G.S.,
- Benedict C.,
- Johnstone D.E.,
- et al.
- Rouleau J.L.,
- Packer M.,
- Moye L.,
- et al.
- Omland T.,
- Aakvaag A.,
- Bonarjee V.S.S.,
- et al.
- Farquharson C.A.,
- Struthers A.D.
- Swedberg K.,
- Eneroth P.,
- Kjekshus J.,
- et al.
- Zannad F.,
- Alla F.,
- Dousset B.,
- et al.
- Glantz S.,
- Slinker B.
- Hall C.,
- Rouleau J.L.,
- Moye L.,
- et al.
- Tsutamato T.,
- Wada A.,
- Maeda K.,
- et al.
- Weber K.T.,
- Brilla C.G.
- Benedict C.R.,
- Shelton B.,
- Johnstone D.E.,
- et al.
- Yee K.-M.,
- Pringle S.D.,
- Struthers A.D.
- Macfadyen R.J.,
- Barr C.S.,
- Struthers A.D.
- Rousseau M.F.,
- Konstam M.A.,
- Benedict C.R.,
- et al.
- Mizuno Y.,
- Yoshimura M.,
- Yasue H.,
- et al.
- Silvestre J.S.,
- Robert V.,
- Heymes C.,
- et al.
- Fischer Y.,
- Filzmaier K.,
- Stieger H.,
- et al.