OUP user menu

Long-term outcome in relation to renal sympathetic activity in patients with chronic heart failure

Magnus Petersson , Peter Friberg , Graeme Eisenhofer , Gavin Lambert , Bengt Rundqvist
DOI: http://dx.doi.org/10.1093/eurheartj/ehi184 906-913 First published online: 11 March 2005

Abstract

Aims Although cardiac sympathetic activation is associated with adverse outcome in patients with chronic heart failure (CHF), the influence of renal sympathetic activity on outcome is unknown. We assessed the hypothesis that renal noradrenaline (NA) spillover is a predictor of the combined endpoint of all-cause mortality and heart transplantation in CHF.

Methods and results Sixty-one patients with CHF, New York Heart Association (NYHA) I–IV (66% NYHA III–IV), and left ventricular ejection fraction (LVEF) 26±9% (mean±SD) were studied with cardiac and renal catheterizations at baseline and followed for 5.5±3.7 years (median 5.5 years, range 12 days to 11.6 years). Nineteen deaths and 13 cases of heart transplantation were registered. Only renal NA spillover above median, 1.19 (interquartile range 0.77–1.43) nmol/min, was independently associated with an increased relative risk (RR) of the combined endpoint (RR 3.1, 95% CI 1.2–7.6, P=0.01) in a model also including total body NA spillover, LVEF, glomerular filtration rate (GFR), renal blood flow, cardiac index, aetiology, and age.

Conclusion Renal noradrenergic activation has a strong negative predictive value on outcome independent of overall sympathetic activity, GFR, and LVEF. These findings suggest that treatment regimens that further reduce renal noradrenergic stimulation could be advantageous by improving survival in patients with CHF.

  • Heart failure
  • Congestive
  • Survival analysis
  • Sympathetic nervous system
  • Noradrenaline
  • Renal circulation

See page 861 for the editorial comment on this article (doi:10.1093/eurheartj/ehi220)

Introduction

Increased levels of natriuretic peptides, noradrenaline (NA), and renin–angiotensin–aldosterone have all been linked to adverse outcome in chronic heart failure (CHF).1,2 The renin–angiotensin–aldosterone system (RAAS) and renal sympathetic activity influence renal mechanisms resulting in sodium and water retention, hallmarks of the CHF syndrome. Renal dysfunction is common in CHF and has recently been shown to be an independent predictor of mortality.2,3 In addition, it is possible that renal dysfunction increases sympathetic activity by a sympathoexcitatory reflex initiated by afferent nerve signalling from renal chemo- or baro-receptors.4,5 A possible interaction between renal sympathetic activation and renal function thus raises the hypothesis that increased renal sympathetic activity could be associated with adverse outcome in CHF.

We have previously conducted several studies in patients with well-defined CHF in which regional and systemic NA spillover rates were assessed simultaneously with central haemodynamics and renal blood flow.610 Hence, we reviewed this CHF cohort to test the main hypothesis that increased renal NA spillover has an adverse effect on the combined outcome of all-cause mortality and heart transplantation. In an additional subgroup analysis including patients with available measurements of cardiac NA spillover rates, the prognostic value of cardiac sympathetic activity was compared with that of renal and overall sympathetic activities.

Methods

The local ethics committee and isotope committee at Sahlgrenska University Hospital approved all studies, and informed consent was obtained from each subject after the purpose and procedures of the studies had been fully explained.

The study population (Tables 1 and 2) comprised 61 patients with CHF, of whom 42 were referred for evaluation by the cardiac transplant team at Sahlgrenska University Hospital and 19 were recruited from the heart failure outpatient clinic at the same institution. Most patients had moderate to severe symptoms. The patients were studied consecutively between May 1992 and February 2001 in cardiac catheterization studies, with measurements of regional and systemic catecholamine kinetics with the radiotracer method as described by Esler et al.11 Patients entered the present study protocol at the day of catheterization and were followed until they reached either an endpoint (death or heart transplantation) or until 31 December 2003. All patients were in stable sinus rhythm, had a reduced left ventricular ejection fraction (LVEF≤35%), and a clinical history of pulmonary congestion and/or peripheral oedema without evidence of primary renal disease. Patients with diabetes mellitus, neuropathy, or known primary autonomic dysfunction were excluded. A coronary angiography was performed in all patients in conjunction with the study. The aetiology of heart failure was classified as ischaemic heart disease if the coronary angiogram revealed significant stenosis (>50% main left coronary artery or >75% for other locations). The remaining patients were considered to have non-ischaemic heart failure. LVEF was measured by echocardiography.

View this table:
Table 1

Baseline demographic and haemodynamic characteristics (n=61)

Age (year)53±11
Gender (F/M)8/53
Body mass index (kg/m2)27±4
NYHA class III–IV (n)40
Diagnosis (IHD/non-IHD)18/43
LVEF (%)26±9
Heart rate (b.p.m.)78±14
Mean arterial pressure (mmHg)86±14
Mean pulmonary artery pressure (mmHg)28±13
Mean pulmonary capillary wedge pressure (mmHg)17±10
Cardiac index (L/min/m2)1.5±0.7
Renal blood flow (L/min)0.85±0.31
Estimated GFR (mL/min/1.73 m2)69±16

Data are presented as mean±SD; IHD, ischaemic heart disease; NYHA, New York Heart Association.

View this table:
Table 2

Baseline indices of sympathetic activity and PRA

Total body NA spillover (nmol/min)n=614.7 (3.4–6.3)
Renal NA spillover (nmol/min)n=611.19 (0.77–1.43)
Cardiac NA spillover (nmol/min)n=400.21 (0.12–0.38)
Arterial NA concentration (nmol/L)n=612.5 (1.7–3.2)
Arterial PRA (ng and L/mL/h)n=525.0 (2.0–8.9)

Data are presented as median (interquartile range).

Study protocol

All patients underwent a combined right heart coronary sinus and right renal vein catheterization with simultaneous measurements of cardiac haemodynamics and arterial and renal venous blood sampling according to a previously described protocol,10 of whom 40 subjects also had the coronary sinus catheterized. The positions of the catheters were confirmed by fluoroscopy, blood gas analyses and, in some subjects, contrast injections.

Assays

Assays of plasma catecholamines and tritiated NA were performed as described by Eisenhofer et al.12 and Medvedev et al.13 Interassay coefficients of variation were 4.6% for endogenous NA and 3.2% for [3H]NA. Renal plasma flow was calculated according to Schnurr et al.14,15 by the paraaminohippurate (PAH) infusion clearance method. Plasma PAH concentrations were estimated by chemical analyses using a modified method of Brun,16 with a mean coefficient of variation for renal blood flow estimated by the PAH infusion clearance method of 4.5±2.4%. A commercial radioimmunoassay (RIA) was used for determination of plasma renin activity (PRA) (Renin-RIA bead; Abbot Diagnostics Division, South Pasadena, CA, USA) with an interassay coefficient of variation of 8.8%.

Calculations

Total body spillover of NA into plasma and total body plasma clearance of NA (TBCL) were calculated according to Esler et al.:17

Math

where I is the infusion rate of NA (d.p.m./min), [3H]NA the arterial concentrations of tritiated NA (d.p.m./mL), and NAA the arterial endogenous NA concentrations (nmol/L).

Organ (renal or cardiac) spillover of NA into plasma and the organ fractional extraction of NA (EXorgan) were calculated as:

Math

where Q is the organ plasma flow (mL/min), NAV the renal venous or coronary sinus NA concentration (nmol/L), and [3H]NAV the corresponding concentration of tritiated NA (d.p.m./mL). Glomerular filtration rate (GFR) was estimated with the abbreviated Modification of Diet in Renal Disease Study equation.18,19

Data analysis and statistical considerations

Baseline characteristics were expressed as mean±SD, or median values with quartile range. Baseline differences between the patients, divided on the basis of the median renal NA spillover rate, were assessed with an unpaired t-test for normally distributed variables, Mann–Whitney U test for skewed variables, and χ2 test for categorical data. Correlation between different NA spillover measurements and haemodynamic and clinical data were performed with the non-parametric Spearman method. To account for the inflation of Type I error, correction according to Bonferroni was used. Univariate and multivariable Cox proportional hazard analyses were performed to investigate the relationship between NA spillover rates and the combined primary outcome variable of heart transplantation and mortality. In addition, the Kaplan–Meier method with log-rank tests was used to determine the differences in time-dependent cumulative transplantation-free survival rates in relation to median renal and total body NA spillovers. Day 1 was defined as the day of the renal and cardiac catheterizations. All Swedish citizens have a unique personal identification number (PIN). For the purpose of the present study, the PIN was used to link the present patients with the Cause of Death Register during 1992–2003.

We first assessed the effect on the combined endpoint of each separate NA spillover rate (cardiac, renal, and total body NA spillovers), plasma NA concentration, arterial PRA, GFR, renal blood flow, and the clinical and haemodynamic data with separate univariate Cox proportional hazard models. Power calculations were developed using the PS power and sample size program.20 Baseline catheterizations were performed during an 8 year period with a median follow-up time of 5.5 years. We assumed a median survival time of between 4 and 5 years.2123 Given this assumption, the present sample size would allow detection of a difference in RR of 2.6–2.8 with a power of 0.8 with alpha established at 5%. The proportional hazards assumption was assessed by adding the covariates by log–time interactions to the models and testing their significance by the Wald test. In addition, Schoenfeld residuals were plotted and visually checked.

Secondly, we performed three multiple regression Cox analyses: (i) renal and total body NA spillover measurements (n=61) as independent variables; (ii) with addition of the clinical and echocardiographic variables with a univariate relationship with outcome (LVEF, cardiac index, GFR, and renal blood flow, P<0.05) and, in addition, diagnosis (ischaemic vs. non-ischaemic aetiology) and age (n=61) and subgroup analyses; and (iii) with all three NA spillover measurements (n=40). All multiple regression models used case-wise deletion of missing data. The regression analyses were primarily performed with the continuous variables dichotomized according to their respective median values. Next, all models were repeated with the variables in continuous form. Arterial NA levels correlate with renal and total body NA spillover rates and were subsequently not used in the multivariable models, as our main objective was to compare renal NA spillover with total body NA spillover. The mean pulmonary artery pressure, capillary wedge pressure, arterial pressure, and heart rate can all carry prognostic information, but were not used in the multivariable analysis due to the correlation with the NA spillover rates and to the baroreflex-mediated interdependence between these measurements. All P-values were two-sided, and a P-value of <0.05 was considered to indicate statistical significance.

Results

The follow-up time was 5.5±3.7 years (median 5.5 years, range 12 days to 11.6 years). There were 19 deaths and 13 cases of heart transplantation in the cohort, giving a total of 32 endpoints (52% uncensored observations).

Univariate results

The baseline renal NA spillover rate for the 61 patients was 1.13±0.47 nmol/min (mean±SD, median 1.19, interquartile range 0.8–1.4; Figure 1). In patients with renal NA spillover above the median value, 22 endpoints were registered (10 deaths and 12 cases of heart transplantation), whereas there were 10 endpoints in patients with renal NA spillover below the median value (nine deaths and one case of heart transplantation). Baseline characteristics are presented in Table 1 and neurohormonal data in Table 2. Correlations between indices of sympathetic activity and haemodynamic variables are given in Table 3.

Figure 1 Histogram illustrating renal NA spillover rates in the present study population (the black line indicates the median value at 1.2 nmol/min).

View this table:
Table 3

Correlations between indices of sympathetic activity and haemodynamic variables

GFR (R, P-value)Renal blood flow (R, P-value)Cardiac index (R, P-value)LVEF (R, P-value)Mean arterial pressure (R, P-value)Mean pulmonary artery pressure (R, P-value)
Renal NA spillover (n=61)−0.09, 0.5−0.16, >0.5−0.16, >0.5−0.46, 0.001−0.35, 0.050.44, 0.003
Cardiac NA spillover (n=40)−0.37, >0.5−0.60, 0.0003−0.13, >0.5−0.16, >0.5−0.06, >0.50.30, 0.4
Total body NA spillover (n=61)−0.17, >0.5−0.14, >0.5−0.06, >0.5−0.21, >0.5−0.13, >0.50.49, 0.0004
Arterial NA concentration (endogenous) (n=61)−0.36, 0.03−0.51, 0.0001−0.32, 0.07−0.34, 0.04−0.22, >0.50.58, 0.000006

R, Spearman rank order correlations. All P-values adjusted according to Bonferroni, an adjusted P<0.05 was considered significant.

Outcome: univariate regression results

The transplantation-free survival was significantly lower in patients with a renal NA spillover above the median when compared with patients with a value below the median (Figure 2), whereas no statistically significant effect on outcome was found for total body NA spillover. Renal NA spillover and arterial NA concentrations above the median and renal blood flow, LVEF, and cardiac index carried the highest RR for adverse outcome when analysed separately in univariate Cox regression models (Table 4).

Figure 2 Kaplan–Meier graphs illustrating the proportion of heart failure patients with transplantation-free survival in relation to median renal and total body NA spillover rates (n=61). Solid lines below and hatched lines above median values. Log–rank tests.

View this table:
Table 4

Effect on the combined outcome of all-cause mortality and heart transplantation: separate univariate Cox proportional hazard models

PredictornRR95% CIP-value
Above median renal NA spillover613.51.6–7.40.001
Above median total body NA spillover611.80.9–3.60.1
Above median cardiac NA spillover404.80.7–32.60.1
Above median plasma NA level613.41.6–7.20.001
Above median PRA521.00.5–2.00.9
Below median GFR612.41.2–5.10.01
Below median renal blood flow612.81.3–5.80.007
Below median LVEF612.81.4–6.00.006
Below median cardiac index612.41.2–4.80.02

Outcome: multivariable regression results

A multivariable Cox model comparing outcome in relation to median total body NA spillover and renal NA spillover revealed only renal NA spillover as an independent negative predictor (RR 2.9, 95% CI 1.3–6.1, P=0.006). The same model with continuous variables yielded a similar result (RR 3.4, 95% CI 1.6–7.4, P=0.001). When aetiology, age, and dichotomized variables of GFR, renal blood flow, cardiac index, and LVEF were also included, only renal NA spillover above median (RR 3.1, 95% CI 1.2–7.6, P=0.01) remained an independent predictor of outcome (Table 5). When using continuous data, the result remained similar with only renal NA spillover being independently associated with adverse outcome (RR 3.7, 95% CI 1.5–9.0, P=0.004). In the subgroup including patients with available simultaneous renal, cardiac, and total body NA spillover measurements (n=40), a multivariate comparison of the effect on outcome of these variables was performed. In this model, only renal NA spillover above median was independently associated with adverse outcome (RR 4.6, 95% CI 1.5–13.7, P=0.006, n=40; Table 6), a result that remained valid when handling the spillover rates in continuous form (RR 3.0, 95% CI 1.0–8.6, P=0.05).

View this table:
Table 5

Effect on the combined outcome of all-cause mortality and heart transplantation: multiple regression Cox model (n=61)

Independent variableRR95% CIP-value
Aetiology (ischaemic vs. non-ischaemic)0.50.2–1.20.1
Age1.00.95–1.040.9
Below median GFR1.70.7–4.20.2
Below median LVEF1.50.6–4.00.4
Below median cardiac index1.80.8–4.10.2
Below median renal blood flow1.40.6–3.60.5
Above median renal NA spillover3.11.2–7.60.01
Above median total body NA spillover1.50.7–3.40.3

P for global test=0.003; χ2=23.6.

View this table:
Table 6

RR ratios for cardiac, renal, and total body NA spillovers for the combined outcome of all-cause mortality and heart transplantation, multivariable Cox proportional hazard model (n=40)

Independent variableRR95% CIP-value
Above median cardiac NA spillover0.90.3–2.50.8
Above median renal NA spillover4.61.5–13.70.006
Above median total body spillover2.91.0–9.00.06

P for global test=0.008; χ2=11.6.

Discussion

The main and novel finding of this study was the strong association between renal sympathetic activity, as measured by renal NA spillover rate, and the combined outcome of all-cause mortality and heart transplantation in patients with moderate to severe CHF. However, total body NA spillover, which more specifically reflects overall NA release rate than plasma NA levels,24 appeared not to carry prognostic information. Furthermore, the adverse outcome linked to increased renal NA spillover was found to be independent of LVEF and renal function, as well as of total body NA spillover. In addition, when comparing the effect of increased renal and cardiac NA spillovers in this cohort of patients, mostly on treatment with β-receptor blockers, renal noradrenergic activity seems to have a stronger association with adverse outcome. This suggests a central role of renal sympathetic activation in CHF and further defines the impact of renal abnormalities in this syndrome.

The link between renal sympathetic activity and renal dysfunction

Renal sympathetic activation promotes salt and water retention, which increases cardiac load and hence contributes to the progression of CHF. Furthermore, and importantly, renal sympathetic nerve stimulation increases renin release.25 Studies in patients with CHF indicate that β-receptor blocking agents lower renin release, consistent with a clinically relevant interaction between RAAS and renal sympathetic nerve activity.26 This linkage is also supported by the finding that the ACE-inhibitor enalaprilat reduces muscle sympathetic nerve activity (MSNA) in predialytic patients.27 Ample evidence linking renal failure in humans with sympathetic activation was reported by Converse et al.4, who found increased MSNA in patients undergoing chronic haemodialysis, but only in a subset who had not been subjected to bilateral nephrectomy. Studies by Campese et al.2830 have provided support for a possible renal afferent reflex, whereby renal chemo- or baro-receptors generate excitatory afferent signalling to the posterior hypothalamus resulting in an increased sympathetic outflow. Treatment with non-hypotensive doses of moxonidine, a centrally acting sympatholytic agent, reduced morphologic progression of kidney damage and reduced albuminuria in another rat model.31 This was paralleled by lower intrarenal NA release in treated animals vs. untreated, thus further supporting an effect associated with decreased renal sympathetic activation. The possibility of a renal afferent sympathoexcitatory reflex is supported by the present finding of a correlation between renal blood flow and cardiac NA spillover, as well as arterial NA concentrations, and the univariate association between low renal blood flow and adverse outcome.

The patients in the present study with high renal NA spillover rates could hypothetically have a progressively faster decline of renal function over time, affecting outcome,2,3 although they did not differ significantly in renal function at baseline. High renal sympathetic activity may thus constitute an important pathogenic factor in terms of adverse outcome in CHF patients.

Plasma NA spillover rates as prognostic indicators in CHF

The prognostic value of different NA spillover rates in CHF was first assessed by Kaye et al.,32 who demonstrated that a cardiac NA spillover rate above median was an independent predictor of mortality in patients with severe heart failure without β-receptor blockade, whereas total body NA spillover did not predict outcome. These results were later confirmed in a paper from Brunner-La Rocca et al.,33 where cardiac NA spillover carried a higher hazard ratio for mortality than total body NA spillover. Total body NA spillover thus seems to be only a weak prognostic marker for mortality, concordant with the present findings.

To our surprise, cardiac NA spillover lacked predictive value in the present subgroup of 40 patients with cardiac catecholamine measurements, thus differing from the results in previous reports.32,33 Furthermore, when we compared renal and cardiac NA spillover rates, together with total body NA spillover in a multivariable model, only renal NA spillover remained an independent predictor of outcome. This finding must, however, be interpreted with care: the present subgroup was small and the outcome might also be influenced by the use of β-receptor blockers, the latter contrasting with the study by Kaye et al.32 Although only 40% of the present study population had β-blocker therapy at the time of cardiac catheterization, the vast majority were treated with the maximal tolerated dose during the follow-up period, whereas none received this therapy in the study by Kaye et al.32 and only 7% in the study by Brunner-La Rocca et al.33 This could possibly modify the influence of chronic cardiac noradrenergic stimulation and explain the difference between our studies in terms of increased cardiac NA spillover and outcome.

The prognostic information of arterial NA levels in patients with CHF

The arterial NA level at baseline in the present study was a statistically significant predictor of adverse outcome, thus confirming previous results regarding plasma NA concentrations and adverse outcome in CHF.1,34,35 It may seem contradictory that total body NA spillover did not predict outcome given the fact that plasma NA concentrations are included in the calculation. However, as cardiac output and tissue perfusion are reduced in CHF, and consequently also plasma NA clearance—the other factor which is needed for calculation of total body NA spillover—the prognostic value found in plasma NA concentrations is lost in total body NA spillover related to the reduced NA clearance that prevails in CHF. This stresses the importance of regional NA spillover measurements. Both the elevated NA concentrations and the increased renal NA spillover showed prognostic information. Given that the contribution of renal NA spillover to total body NA spillover is substantial (∼20%11,36), it may be speculated that the increased plasma NA concentrations observed in patients in the present study and in several other previous studies1,34,35,37 in fact emanate from the kidneys. The present study thus importantly focuses on the kidneys as one possible source of the elevated plasma NA and, consequently, an organ to be further targeted with various treatment regimens.

Limitations of the study

With respect to the generalizability of the findings, the patients in the present cohort were younger than in a common heart failure population.23 Ischaemic heart disease was not the dominating aetiology and patients with concomitant conditions such as diabetes and atrial fibrillation were excluded. Thus, caution must be exercised when extrapolating the present findings to the general CHF population. Although patients with symptoms varying from mild to severe heart failure were studied, the most severely ill patients generally came to our attention when they were referred for consideration of heart transplantation owing to end-stage CHF. These patients thus had multiple clinical risk factors indicating a high risk of mortality, thus contrasting with the patients with mild to moderate symptoms, who mostly were recruited from our out-patient clinic. Testing the effect of functional class on outcome was accordingly not possible in the present cohort. Selection bias could further introduce a risk that the main effect of increased renal NA spillover on outcome was related to cardiac transplantation. There were, however, only two cases of heart transplantation after 1 January 2000 and considering that all transplanted patients had end-stage CHF, the effect could most likely be ascribed as related to mortality. Although only all-cause mortality was registered, the results from large clinical studies in patients with CHF indicate that cardiac deaths account for >90% of the total mortality.38,39 Data regarding medications were only recorded at baseline, thus making analyses of treatment changes impossible. Treatment with β-blockers was, however, mandatory when tolerated in the clinical heart failure protocol at our institution during the follow-up period.

Conclusions

This study demonstrates a strong negative predictive value of renal NA spillover on outcome in patients with moderate to severe CHF, an effect that appears to be independent of total body NA spillover, as well as LVEF and GFR. We hypothesize that renal noradrenergic activation could aggravate CHF by increasing both cardiac load and the progression rate of renal dysfunction. Although cardiac sympathetic activity and renal function were not independent prognostic variables in the present cohort, results from previous studies have established their importance for prognosis, warranting a cautious interpretation of this part of our findings. On the basis of the present data, one may surmise that renal sympathetic activation has a pathophysiological role in parallel with renal function, cardiac sympathetic activation, and other neurohormonal abnormalities operative in heart failure. Taken together, these findings further strengthen the concept that treatments which reduce renal neurohormonal activity and improve renal dysfunction in CHF may have beneficial effects on outcome in this disorder.

Acknowledgements

We are grateful to Gun Bodehed Berg for invaluable technical assistance and to the staff of the Cardiology Laboratory at Sahlgrenska University Hospital. This study was supported by grants from the Swedish Heart Lung Foundation, The Sahlgrenska Academy at Göteborg University and Göteborg Medical Society.

References

View Abstract