Total and high molecular weight adiponectin, haemodynamics, and mortality in patients with chronic heart failure

Abstract

Aims

To evaluate whether plasma high molecular weight (HMW) adiponectin provides prognostic information in addition to that obtained from clinical, haemodynamic, and biochemical variables previously known to be associated with a high mortality in chronic heart failure (CHF) patients.

Methods and results

We measured the plasma levels of total and HMW adiponectin, atrial natriuretic peptide, brain natriuretic peptide (BNP), and N-terminal-proBNP (NT-proBNP), and haemodynamic parameters in 449 consecutive CHF patients. Based on body mass index (BMI), patients were classified into three groups: low (<21 kg/m², n = 133), normal (21–25 kg/m², n = 205), and high (>25 kg/m², n = 111). After adjustment for clinical variables associated with CHF including haemodynamics, plasma total adiponectin level was an independent prognostic predictor but HMW adiponectin was not in the overall patient group. On subgroup analyses, in patients with abnormal BMI, plasma total adiponectin level was not an independent prognostic predictor, but in patients with normal BMI, plasma levels of log NT-proBNP (P = 0.017) and log total adiponectin (P = 0.003) were independent prognostic predictors.

Conclusion

These findings indicate that total adiponectin is more useful for assessing mortality risk than HMW adiponectin and a high plasma total adiponectin is an independent prognostic predictor especially in CHF patients with normal BMI.

Adiponectin, Body mass index, N-terminal pro-brain natriuretic peptide, Prognosis, Heart failure

Introduction

Chronic heart failure (CHF) is characterized by a complex syndrome of haemodynamic, neurohormonal, and metabolic abnormalities with high mortality. Although population studies have found obesity to be a risk factor for cardiovascular disease and for the development of CHF,¹ recent data suggest that a high body mass index (BMI) is associated with a more favourable prognosis in patients with established CHF.^2–4 Moreover, CHF patients with cachexia have shown a poor prognosis.^5,6

Adiponectin, which is an adipocyte-specific cytokine, is abundant in plasma and has important metabolic effects⁷ and may predict cardiovascular events. In healthy patients and in patients with atherosclerosis, low plasma adiponectin levels have been associated with metabolic disorders and an increased risk of cardiovascular events,^8–10 recently a high plasma adiponectin level has been demonstrated to be a predictor of mortality in established CHF patients.^11,12 The mechanisms underlying these findings remained unexplained, but cachexia or obesity, which is an independent risk factor for mortality,^2–5 may influence the plasma adiponectin in patients with the CHF. Adiponectin circulates in plasma as a trimer, a hexamer, and in a high molecular weight (HMW) form.^13,14 Recent studies suggest that HMW adiponectin is an active form of this protein and that HMW adiponectin is superior to total adiponectin as a predictor of metabolic abnormalities.^15,16 However, it has not been elucidated whether plasma HMW adiponectin is a more useful prognostic predictor than total adiponectin in CHF the patients.

Plasma levels of cardiac natriuretic peptides such as atrial natriuretic peptide (ANP), brain natriuretic peptide (BNP), and N-terminal-proBNP (NT-proBNP) are well-established powerful risk markers in CHF^17–20 and are positively correlated with a left ventricular filling pressure^21,22 and inversely correlated with BMI.^11,23 Recent studies suggest a positive correlation between the plasma NT-proBNP and adiponectin in patients with coronary artery disease²⁴ and in patients with CHF,¹¹ and that cardiac natriuretic peptides have a novel lipolytic and potential lipid-mobilization effect, which is mediated by a specific adipocyte plasma membrane receptor.^25–27 However, there are no data that directly compare plasma adiponectin with physiologically active natriuretic peptides such as ANP and BNP, and haemodynamic parameters.

The present study (i) evaluated whether haemodynamic abnormalities directly contribute to the elevated total adiponectin in CHF patients, (ii) whether measuring plasma total or HMW adiponectin provides prognostic information that is additional to that obtained from clinical, haemodynamic, and biochemical variables previously known to be associated with a high mortality in CHF patients, and (iii) which CHF patients would benefit from evaluation of the plasma adiponectin to predict mortality risk.

Methods

Patients

The study population was drawn from 712 consecutive symptomatic CHF patients admitted to our institute between 2000 and 2005 for management of CHF. Among these patients, the study population consisted of 449 consecutive patients with systolic CHF, which was defined as a left ventricular ejection fraction (LVEF) <45% on right-sided cardiac catheterization. All patients could be followed. LVEF was measured by two-dimensional echocardiography or ventriculography using contrast medium or radioisotope before the study. Aetiology of systolic CHF was ischaemic cardiomyopathy, dilated cardiomyopathy, or hypertensive heart disease. Patients with acute myocardial infarction, congenital heart disease, valvular heart disease, malignancy, or renal failure (serum creatinine >2.0 mg/dL) were excluded. NYHA functional class was evaluated on the day of cardiac catheterization. Informed consent was obtained from all patients for participation in the study, according to a protocol approved by the Committee on Human Investigation at our institution.

Study protocol

All patients were pre-medicated with an oral dose of diazepam (5 mg) and rested in bed in a supine position for at least 20 min. Right-sided cardiac catheterization was performed by a 6 F Swan–Ganz catheter. Blood samples for measuring plasma levels of total and HMW adiponectin, ANP, BNP, and NT-proBNP were collected from the pulmonary artery. Samples for the assay of plasma adiponectin, ANP, BNP, and NT-proBNP concentrations were transferred to chilled disposable tubes containing aprotinin (500 kallikrein inactivator units/mL). The blood samples were immediately placed on ice and centrifuged at 4°C, and the plasma was frozen in aliquots and stored at −30°C until assay.

Plasma total adiponectin level was measured by a sandwich ELISA system (adiponectin ELISA kit, Otsuka Pharmaceutical Corp. Ltd).²⁸ Plasma HMW adiponectin concentration was measured using a sandwich ELISA based on a monoclonal antibody to human HMW adiponectin (Fujirebio ELISA kit, Tokyo, Japan).¹⁶ Plasma concentrations of ANP and BNP were measured with a specific immunoradiometric assay for human ANP and BNP using a commercial kit (Shionogi, Osaka, Japan), as previously reported.¹⁷ Plasma levels of NT-proBNP concentrations were measured using Elecsys proBNP sandwich immunoassay (Roche Diagnostics, Mannheim, Germany). Renal function was represented by the estimated glomerular filtration rate according to the Cockcroft–Gault equation.

The subjects were 365 men and 84 women ranging in age from 17 to 85 years (mean 62.2 ± 12.3); 291 patients had ischaemic heart disease; 117 had dilated cardiomyopathy; 41 had hypertensive heart disease. As the recommendation criteria for obesity in our country,²⁹ patients were divided into three groups: low BMI (<21 kg/m², n = 133), normal BMI (21–25 kg/m², n = 205, and high BMI (>25 kg/m², n = 111). At entry to the study, 381 patients (85%) were treated with angiotensin-converting enzyme inhibitors or angiotensin II type 1 receptor blockers, 266 (60%) with β-blockers, 124 (28%) with spironolactone. All drugs were administered for over 3 months prior to entry in most patients.

Statistical analysis

All results are expressed as the mean ± SD. Univariate analyses were performed between two groups using Student's t-test. Patients were divided into three groups according to BMI categories [low BMI (< 21 kg/m²), normal BMI (21–25 kg/m²), and high BMI (> 25 kg/m²)]. Comparisons between the groups were performed by 1-way ANOVA or Kruskal–Wallis test for continuous variables. A χ² test was used to determine differences between groups. Because plasma levels of total and HMW adiponectin, ANP, BNP, and NT-proBNP were not normally distributed, log total adiponectin, log HMW adiponectin, log ANP, log BNP, and log NT-proBNP were used to determine correlations and regression models. Multivariable linear regression analyses, examining the correlates of log-transformed adiponectin levels, include baseline variables that were associated with adiponectin at the P < 0.10 level on univariate analyses.

In multivariable Cox proportional hazard analyses, the main models were adjusted for variables considered to reflect the severity of CHF at baseline and those were associated with mortality on univariate analyses at the P < 0.10 level. Multivariable Cox proportional hazard analyses were performed as stepwise regressions with backward elimination in the overall patient groups and in three groups divided according to BMI categories.

The sensitivity and specificity of adiponectin and NT-proBNP for predicting mortality were determined, and receiver-operating characteristic curves were constructed. Kaplan–Meier analysis was performed on the cumulative rates of survival in patients with HF stratified into subgroups based on cut-off levels of adiponectin and NT-proBNP in patients with normal BMI, and the differences between survival curves were analysed by log-rank test. A two-tailed probability value of <0.05 was considered as significant.

Results

Comparison of plasma adiponectin and haemodynamic parameters and neurohumoral factors in chronic heart failure

Plasma total and HMW adiponectin level increased with the severity of NYHA functional class (Figure 1). There was a close correlation between plasma log total adiponectin and plasma log HMW adiponectin (r = 0.920, P < 0.0001). There was no significant correlation between plasma log total adiponectin and pulmonary capillary wedge pressure (r = 0.051, P = 0.278), and there were correlations between plasma log total adiponectin and other haemodynamic parameters (Table 1). There were significant correlations between plasma log total adiponectin and log ANP (r = 0.312, P < 0.0001), log BNP (r = 0.358, P < 0.0001), log NT-proBNP (r = 0.373, P < 0.0001), and BMI (r = −0.391, P < 0.0001) (Figure 2). On stepwise multivariable analyses, four parameters, low BMI (P < 0.0001), high plasma BNP level (P < 0.0001), female gender (P < 0.0001), and non-diabetes mellitus (P = 0.0126), were significant independent predictors of high plasma log total adiponectin (Table 1). Just like total adiponectin, four parameters, low BMI (P < 0.0001), high plasma BNP level (P < 0.0001), female gender (P < 0.0001), and non-diabetes mellitus (P = 0.0193), were significant independent predictors of high plasma log HMW adiponectin. There were no haemodynamic parameters that were independently associated with either elevated total or HMW adiponectin.

Figure 1

Plasma total and high molecular weight adiponectin concentrations in patients with chronic heart failure. The white box indicates the value of plasma total adiponectin and the black box indicates plasma high molecular weight adiponectin. The box defines the inter-quartile range with the median indicated by the crossbar, and differences among groups were analysed using Kruskal–Wallis testing. NYHA, New York Heart Association functional class.

Figure 2

Correlations between log total adiponection and body mass index and plasma levels of atrial natriuretic peptide, brain natriuretic peptide, and N-terminal-proBNP.

Table 1

Univariate and multivariable linear model of plasma log total adiponectin

Variables	Univariate correlation coefficient	P-value	Multivariate beta-coefficient (SE)	P-value
Gender (male = 1, female = 0)	−0.245	<0.0001	−0.117 (0.030)	<0.0001
NYHA class (III/IV = 1, I/II = 0)	0.180	0.0001
BMI (Kg/m2)	−0.391	<0.0001	−0.022 (0.004)	<0.0001
Hyperlipidaemia (yes = 1)	−0.198	<0.0001
Diabetes mellitus (yes = 1)	−0.140	0.0029	−0.062 (0.025)	0.0126
eGFR (mL/min)	−0.307	<0.0001
Log ANP (pg/mL)	0.312	0.0069
Log BNP (pg/mL)	0.358	<0.0001	0.098 (0.022)	<0.0001
Log NT-proBNP (pg/mL)	0.373	<0.0001
CI (mL/min/m2)	−0.102	0.0302
Mean PA (mmHg)	0.170	0.003
RA (mmHg)	0.111	0.018
LVEF (%)	−0.168	0.0003

NYHA, New York Heart Association; BMI, body mass index; eGFR, estimated glomerular filtration rate; ANP, atrial natriuretic peptide; BNP, brain natriuretic peptide; NT-proBNP, N-terminal-proBNP; CI, cardiac index; mean PA, mean pulmonary arterial pressure; RA, right atrial pressure; LVEF, left ventricular ejection fraction.

Total and high molecular weight adiponectin as a prognostic predictor in chronic heart failure

Table 2 summarizes the patient characteristics according to BMI categories. Forty-seven patients died during a median follow-up period of 2.7 years (inter-quartile range: 1.27–3.91 years). Mortality rate and plasma levels of ANP, BNP, NT-proBNP, total and HMW adiponectin, and HMW-to-total adiponectin ratio were higher in patients with lower BMI. HMW-to-total adiponectin ratio was significantly lower in patients with diabetes mellitus, hyperlipidaemia, and ischaemic cardiomyopathy (Figure 3). Thirteen clinical, neurohumoral, and haemodynamic variables were analysed using univariate and stepwise multivariable Cox proportional hazards regression analyses (Table 3). On stepwise multivariable analyses, high levels of plasma log NT-proBNP (P < 0.0001) and log total adiponectin (P = 0.008), cardiac index (P = 0.023), and LVEF (P = 0.01) were significant independent predictors of mortality in the overall patient group (Table 3).

Figure 3

Comparisons of plasma high molecular weight-to-total adiponectin ratios. IHD, ischaemic heart disease, DM, diabetes mellitus. * = P < 0.05, ** = P < 0.01.

Table 2

Patient characteristics according to body mass index

	Total group	BMI < 21 (kg/m2)	21 ≤ BMI ≤ 25 (kg/m2)	25 < BMI (kg/m2)	P-value
Characteristics	(n = 449)	(n = 133)	(n = 205)	(n = 111)
Age (year)	62.2 ± 12.3	63.7 ± 12.5	63.1 ± 11.3	58.7 ± 13.3	0.0025
Male, n (%)	365 (81)	96 (72)	173 (84)	96 (86)	0.0021
BMI (kg/m2)	23.1 ± 3.6	19.2 ± 1.4	23.0 ± 1.1	27.9 ± 2.6	<0.0001
NYHA class III/IV, n (%)	108 (24)	42 (32)	38 (19)	28 (25)	0.022
Death, n (%)	47 (10)	21 (16)	20 (10)	6 (5)	0.0264
Aetiology of heart failure
Ischemic cardiomyopathy	291 (65)	80 (60)	131 (64)	79 (71)	0.230
Dilated cardiomyopathy	117 (26)	38 (29)	56 (27)	23 (21)	0.325
Hypertensive heart disease	41 (9)	15 (11)	18 (9)	9 (8)	0.785
Hypertension, n (%)	205 (46)	44 (33)	95 (46)	66 (59)	0.0002
Diabetes Mellitus, n (%)	141 (31)	32 (24)	68 (33)	41 (40)	0.074
Hyperlipidaemia, n (%)	205 (46)	50 (38)	89 (43)	66 (59)	0.002
Heart rate (beats/min)	71.2 ± 15.4	70.5 ± 15	70.8 ± 15	72.8 ± 15	0.449
Mean blood pressure (mmHg)	86.7 ± 15.5	83.8 ± 15.6	85.7 ± 14.7	91.5 ± 15.9	0.003
LVEF (%)	31.2 ± 8.4	29.4 ± 8.4	31.7 ± 8.3	32.3 ± 8.4	0.01
PCWP (mmHg)	12.1 ± 6.5	12.3 ± 6.8	11.4 ± 6.4	12.8 ± 6.2	0.154
RA (mmHg)	3.7 ± 3.1	3.4 ± 3.0	3.4 ± 3.0	4.6 ± 3.1	0.0015
Mean PA (mmHg)	16.1 ± 7.0	16.5 ± 7.7	15.5 ± 6.9	16.7 ± 6.3	0.258
CI (L/min/m2)	2.17 ± 0.49	2.13 ± 0.5	2.18 ± 0.48	2.2 ± 0.5	0.464
ANP (pg/mL)	89.2 ± 90	107.9 ± 97	80 ± 89	82.5 ± 79	0.005
Median value	59	78.9	53	54
BNP (pg/mL)	183.2 ± 265	262 ± 335	167 ± 251	119 ± 148	<0.0001
Median value	95	170	80	69.9
NT-pro-BNP (pg/mL)	1125.1 ± 3099	1908 ± 5230	952 ± 1953	506 ± 636	<0.0001
Median value	429.5	736.8	350	293
Total adiponectin (μg/mL)	9.46 ± 6.7	13.3 ± 7.9	8.5 ± 5.8	6.7 ± 4.3	<0.0001
Median value	7.6	12	7	6.2
HMW adiponectin (μg/mL)	5.69 ± 5.5	8.42 ± 6.7	5.1 ± 4.9	3.49 ± 3.1	<0.0001
Median value	4.0	7	3.6	2.6
HMW to total adiponectin ratio	0.547 ± 0.21	0.60 ± 0.21	0.55 ± 0.2	0.49 ± 0.2	0.0001
Haemoglobin (g/dL)	12.8 ± 1.8	12.0 ± 1.8	12.9 ± 1.7	13.6 ± 1.8	<0.0001
Creatinine (mg/dL)	0.94 ± 0.28	0.91 ± 0.30	0.97 ± 0.28	0.92 ± 0.25	0.121
eGFR (mL/min)	75.9 ± 33	61.9 ± 21	71.9 ± 26	100 ± 43	<0.0001
Treatments
Spironolactone, n (%)	124 (28)	45 (34)	52 (25)	27 (24)	0.157
ACE-I or ARB, n (%)	381 (85)	111 (83)	176 (86)	94 (85)	0.883
β blockers, n (%)	266 (60)	82 (62)	118 (58)	66 (59)	0.754

BMI, body mass index; LVEF, left ventricular ejection fraction; PCWP, pulmonary capillary wedge pressure; RA, right atrial pressure; PA, pulmonary arterial pressure; CI, cardiac index; ANP, atrial natriuretic peptide; BNP, brain natriuretic peptide; NT-proBNP, N-terminal-proBNP; HMW, high molecular weight; eGFR, estimated glomerular filtration rate; ACE-I, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker.

Table 3

Univariate and multivariable predictors of mortality in 449 patients with chronic heart failure

Variables	Univariate χ2	P-value	Multivariate χ2	P-value
NYHA class (III/IV = 1, I/II = 0)	13.10	0.003
Ischemic aetiology (yes = 1, no = 0)	21.427	<0.0001
BMI (Kg/m2)	10.327	0.0013
RA (mmHg)	5.594	0.018
CI (L/min/m2)	11.894	0.0006	5.169	0.023
Mean PA (mmHg)	28.039	<0.0001
PCWP (mmHg)	10.891	0.0010
LVEF (%)	28.405	<0.0001	6.477	0.0109
Log ANP (pg/mL)	21.996	<0.0001
Log BNP (pg/mL)	37.058	<0.0001	18.322	<0.0001
Log NT-proBNP (pg/mL)	45.404	<0.0001
Log adiponectin (μg/mL)	27.108	<0.0001	6.888	0.0087
Log HMW Adiponectin (μg/mL)	21.994	<0.0001

Abbreviations are listed in Table 1.

On subgroup analyses, a high plasma log total adiponectin level was not an independent predictor in patients with abnormal BMI (<21 or >25 kg/m²), a high plasma log NT-proBNP level (P = 0.008) and low cardiac index (P = 0.004) were independent predictors in low BMI patients, and only a low BMI was an independent predictor in high BMI patients (P = 0.03). In patients with normal BMI (21–25), plasma levels of log NT-proBNP (P = 0.017) and log total adiponectin (P = 0.003), and low LVEF (P = 0.002) were independent prognostic predictors. Receiver operating characteristic curves of total and HMW adiponectin and BNP and NT-proBNP demonstrating mortality risks in patients with normal BMI are shown in Figure 4. The cut-off level for total adiponectin was determined as 8.8 µg/mL (95% CI: 7.1–13.4), giving a sensitivity of 86% (95% CI: 63.7–96.9) and specificity of 69% (95% CI: 61.4–75.3). The cut-off level of NT-proBNP was determined as 633 pg/mL (95% CI: 604–1645), giving a sensitivity of 86% (95% CI: 63.7–96.9) and specificity of 70% (95% CI: 63.1–76.7).

Figure 4

(A) Receiver operating characteristic curves for the ability of brain natriuretic peptide and N-terminal-proBNP to detect mortality in chronic heart failure patients with normal body mass index. (B) Receiver operating characteristic curves for the ability of total adiponectin and high molecular weight adiponectin to detect mortality in CHF patients with normal body mass index. AUC, area under the curve.

Kaplan–Meier lifetime analysis

Patients were divided in two groups based on cut-off levels for total adiponectin and NT-proBNP, and Kaplan–Meier survival curves were constructed (Figure 5). The CHF patients with normal BMI were stratified into four groups based on the cut-off levels of plasma concentrations of total adiponectin and NT-proBNP and cumulative survival curves were constructed by Kaplan–Meier survival methods (Figure 6). The hazard ratio of patients with plasma NT-proBNP > 633 pg/mL and total adiponectin > 8.8 µg/mL was 17.76 (95% CI: 6.4–49.1) compared to those with plasma BNP < 633 pg/mL or total adiponectin < 8.8 µg/mL for mortality (P < 0.0001).

Figure 5

Kaplan–Meier survival curves according to the cut-off level of plasma concentrations of total adiponectin and N-terminal-proBNP. (A) Patients with low body mass index (BMI < 21 kg/m², n = 133). (B) Patients with normal BMI (21–25 kg/m², n = 205). (C) Patients with high BMI (BMI > 25 kg/m², n = 111).

Figure 6

Kaplan–Meier survival curves stratified into four groups based on the cut-off level of plasma concentrations of total adiponectin and NT-proBNP in chronic hear failure patients with normal body mass index.

Discussion

Plasma total adiponectin level has been reported to be an independent prognostic predictor in patients with CHF.^11,12 The present study for the first time evaluated the prognostic role of HMW adiponectin, which may be a more sensitive marker of metabolic abnormality than total adiponectin.^15,16 On multivariable Cox proportional hazard analyses, high levels of plasma log total adiponectin (P = 0.008) and log NT-proBNP (P < 0.0001) were significant independent predictors, suggesting that total adiponectin is more useful for assessing mortality risk than HMW adiponectin in our cohort patients. The reason HMW adiponectin, an active form, was a less useful prognostic predictor than total adiponectin is not be easily explained; a high HMW adiponectin might have a cardioprotective effect and an increased total adiponectin might reflect the severity of CHF.

Which CHF patients would benefit from evaluation of the plasma adiponectin to predict mortality is a clinically important issue because the increase in plasma adiponectin level is smaller than those of cardiac natriuretic peptides, established markers of prognosis.^17–20 Therefore, we performed sub analysis in three groups divided by BMI. Our findings may indicate that a high plasma total adiponectin level is an independent and useful prognostic predictor in CHF patients especially in patients with normal BMI and that the prognostic power of plasma adiponectin is attenuated in CHF patients with abnormal BMI. The prevalence of obesity, defined by the World Health Organization as BMI > 30 is not more than 2–3% in the Japanese population, in contrast to the 10–20% in Europe and the USA.³⁰ In the present study, patients were divided into three groups: low BMI (<21 kg/m²), normal BMI (21–25 kg/m²), and high BMI (>25 kg/m²) as the recommendation of criteria for obesity disease in Japan.²⁹ However, further studies in Europe and the USA are needed to confirm our findings.

We also evaluated whether haemodynamic abnormalities directly contribute to elevated adiponectin in CHF patients. Plasma total and HMW adiponectin level did not correlate with pulmonary capillary wedge pressure, but positively correlated with plasma cardiac natriuretic peptides. Among the clinical parameters including haemodynamic and neurohumoral factors, metabolic parameters such as high BMI and diabetes mellitus were independently associated with decreased adiponectin, which is consistent with previous studies.^11,24 The present study suggested that the increase in plasma adiponectin is associated with the physiologically active BNP rather than inactive NT-proBNP and that any haemodynamic parameter is not independently associated with increased adiponectin in CHF patients. A recent study³¹ showing that BNP increases adiponectin mRNA in cultured adipocytes via a functional GC-A receptor may support our findings.

Plasma cardiac natriuretic peptides such as ANP, BNP, and NT-proBNP have been established as a risk factor not only in patients with CHF^17–20 and coronary artery disease but also in the general population. There was a significant correlation between plasma NT-proBNP and adiponectin in CHF patients¹¹ and coronary heart disease.²⁴ Plasma levels of BNP, NT-proBNP, and adiponectin were inversely correlated with BMI.^11,22,24 The prognostic value of a high NT-proBNP has been documented in CHF regardless of the low BMI or high BMI;³² however, just like the ‘obesity paradox’,⁴ subjects with a high adiponectin level had lower risk for myocardial infarction,⁹ whereas established CHF patients with a high adiponectin level had poor prognosis,^11,12 suggesting an association between the ‘obesity paradox’ and ‘adiponectin paradox’. The mechanisms underlying these findings remained unexplained, these findings suggest that the prognostic significance of adiponectin is affected by BMI in CHF. Taking ethnic differences into account, the increased energy expenditure may be more important than the actual BMI level in relation to prognosis. As weight loss increases the plasma adiponectin level, high adiponectin levels in CHF patients could be a marker of the wasting process, which may be one explanation of the association between high adiponectin levels and increased mortality risk in the present CHF population. We did not measure changes in weight in the present study, future studies are needed to address this issue. The low number of patients and events for subgroup analyses were the limitations of the study.

In conclusion, the increase in plasma total and HMW adiponectin in CHF patients is not independently associated with haemodynamic abnormalities but is mainly associated with BMI and cardiac natriuretic peptides. Furthermore, total adiponectin is more useful to assess mortality risk than HMW adiponectin and a high plasma total adiponectin is an independent prognostic predictor especially in CHF patients with normal BMI.

Acknowledgements

We wish to thank Aoi Murata for excellent technical assistance. We also express thanks to Mr Daniel Mrozek for assistance in preparing the manuscript. This study was supported by a Grant-in-Aid for Scientific Research in Japan.

Conflict of interest: none declared.