OUP user menu

Increase in serum adiponectin concentration in patients with heart failure and cachexia: relationship with leptin, other cytokines, and B-type natriuretic peptide

Margaret B. McEntegart, Bonaventure Awede, Mark C. Petrie, Naveed Sattar, Francis G. Dunn, Niall G. MacFarlane, John J.V. McMurray
DOI: http://dx.doi.org/10.1093/eurheartj/ehm033 829-835 First published online: 2 April 2007


Aims Adiponectin is a fat-derived hormone involved in the regulation of metabolism. Adiponectin concentration is inversely related to body weight and, in animals, causes weight loss. We, therefore, measured adiponectin concentration in patients with heart failure (HF) and cachexia.

Methods and results Serum adiponectin concentrations were measured in three groups of patients with coronary artery disease (CAD): (i) HF, reduced left ventricular systolic function, and cachexia (n = 10); (ii) HF, reduced systolic function but no cachexia (n = 20); (iii) HF-controls–patients with CAD, no HF, and preserved systolic function (n = 10); and in a healthy control group (n = 7). Patients with HF and cachexia had higher concentrations of adiponectin [23.8 (10.2–37.2) µg/mL] than all other groups: HF–no cachexia 8.1 (0.5–16.6) µg/mL; CAD-controls 7.1 (0.4–13.5) µg/mL; and healthy controls 8.7 (2.5–16.8) µg/mL) (P < 0.05 for each comparison). Adiponectin correlated negatively with body mass index, percentage of body fat, waist circumference and insulin resistance, and positively with B-type natriuretic peptide (BNP) and tumour necrosis factor-α.

Conclusion Cachexia in HF is associated with an increase in adiponectin concentration. This may represent preservation of the physiological response to change in body fat but might also suggest that adiponectin plays a role in the pathogenesis of cachexia. The correlation between BNP and adiponectin also raises the possibility that the former might increase the secretion of the latter.

  • Heart failure
  • Cachexia
  • Coronary artery disease
  • Adipose tissue
  • Metabolism
  • Adiponectin


Cachexia is a relatively uncommon but sinister development in patients with heart failure (HF).13 Involuntary weight loss and low body mass index (BMI) are associated with a poor prognosis in HF and cachexia is now known to be an independent predictor of reduced survival.13 The exact cause (or causes) of cardiac cachexia is unknown. Pro-inflammatory cytokines, increased neurohumoural activity, under-nutrition, anabolic–catabolic hormone imbalance, and immune activation have all been implicated.46 More recently, there has been interest in the newly discovered hormones that regulate appetite and metabolism, particularly leptin and ghrelin.710 Leptin production may be inappropriately low in cachectic patients, whereas there may be an appropriate, compensatory, increase in ghrelin in this syndrome.710

Adiponectin, a 244 amino-acid fat-derived peptide, is the latest hormone thought to play an important role in the regulation of energy metabolism.1113 In humans, plasma adiponectin concentrations are inversely correlated with BMI and body fat.1113 Obesity is associated with reduced adiponectin levels, whereas these are increased in anorexia nervosa.1416 In experimental animals, including genetically modified ones, administration of adiponectin reduces weight gain or leads to weight loss (depending on the model), possibly by increasing energy expenditure.1719 Plasma adiponectin concentrations have recently been shown to be increased in patients with HF.20,21 We have, therefore, examined plasma adiponectin concentrations in non-cachectic and cachectic patients with HF.


Subject recruitment

Patients were recruited between December 2002 and July 2004 from the Cardiology Clinics of the North Glasgow University Hospitals and the primary care-based HF nurse liaison service. Patient records were screened to identify subjects with HF and coronary artery disease (CAD) (with or without a history of weight loss), and a control group of patients with CAD but no HF. Following baseline assessment, six patients with HF were found to have preserved left ventricular (LV) systolic function and were excluded and two patients with HF withdrew consent. Healthy controls were recruited from patient companions at the cardiology clinics and a university staff advertisement. In addition to age-matching a lifestyle history was taken to activity-match healthy controls to the patients. The study population was initially recruited to look at the levels of gene expression in skeletal muscle and therefore the sample size was based on a power calculation using information from a previous study by our group looking at the gene expression in skeletal muscle of patients with chronic respiratory disease. The study was approved by the local Ethics Committee and written informed consent was obtained from all participants.

Subjects studied

Four groups of subjects were studied. Three were groups of patients with stable CAD: (i) those with HF, reduced LV systolic function and cachexia; (ii) those with HF, reduced systolic function, but no cachexia; and (iii) HF-controls–patients with CAD, no symptoms of HF, and preserved systolic function. The fourth group consisted of healthy controls. Only patients with CAD were studied as there may be a difference in adiponectin levels between individuals with and without CAD.

Subject characterization

A documented myocardial infarction or coronary obstruction on an angiogram was required for the diagnosis of CAD. Subjects were characterized by New York Heart Association (NYHA) classification, LV ejection fraction (LVEF), and body composition analysis. A weight history, co-morbidity, and drug therapy were documented. Cachexia was defined, as previously, as unintentional weight loss of ≥7.5% body weight over 6 months.13 Bioelectrical impedance and anthropometric measurements were used to determine percentage body fat and waist circumference. Peak VO2 was obtained from symptom-limited treadmill stress testing using the STEEP (Standardized Exponential Exercise Protocol) protocol.22

Blood sample collection and analysis

All subjects had fasting blood samples taken between 9 and 10 am. Blood samples were placed on ice, allowed to clot for 20 min, and then centrifuged at 3000 rpm for 15 min at 4°C. The serum or plasma was aliquoted and stored at −80°C until analysis. Serum adiponectin, leptin, tumour necrosis factor-α (TNF-α), and interleukin-6 (IL-6) were measured in triplicate using commercially available enzyme-linked immunosorbent assay kits (Quantikines, R&D systems). Cholesterol, triglycerides, VLDL cholesterol, LDL cholesterol, HDL cholesterol, and high-sensitivity C-reactive protein concentrations were also measured using a Hitachi 917 analyser (Roche kits). Fasting glucose and insulin were measured and the homeostasis model assessment (HOMA-IR) index was calculated to estimate insulin resistance.22 Blood for the measurement of plasma B-type natriuretic peptide (BNP) was added to a chilled tube containing ethylenediaminetetraacetic acid (EDTA). Plasma BNP was measured using the Shionoria immunoradiometric kit (Schering CIS, West Sussex, UK).

Statistical analysis

Results are expressed as mean (SD) for baseline characteristics and median (range) for biomarkers. Two-sided tests of significance were used. Normality testing and Bartlett's test for homogeneity of variance were performed and differences between groups assessed by one-way analysis of variance (ANOVA) or Kruskal–Wallis non-parametric ANOVA. Where differences were identified, a Dunnett's or Dunn's multiple comparison post hoc test was performed. Fisher's exact test was used to compare categorical characteristics. The data was logarithmically transformed for linear regression analysis and a Run's test used for departure from linearity. The relationship of adiponectin with other biomarkers was further assessed by multiple linear regression analysis adjusting for age, BMI, estimated glomerular filtration rate (eGFR), and insulin resistance. Statistical significance was taken at the 5% level.


Subject characteristics

Mean peak VO2, eGFR, and LVEF were all lower in patients with HF than in the controls (Table 1). Patients with cachectic HF had worse NYHA functional class and a lower peak VO2 than non-cachectic HF patients. Mean age was higher in the HF patients with cachexia than in the other groups, although this difference was not significant.

View this table:
Table 1

Patient characteristics

CAD, HF, reduced LV systolic function, cachexia (n = 10)CAD, HF, reduced LV systolic function, no cachexia (n = 20)CAD, no HF, preserved LV systolic function (n = 10)Healthy controls (n = 7)HF cachexia vs. each other groupANOVA
Mean age (years)72.767.664.463P > 0.05P = 0.0906
Gender, M/F8/218/29/14/3P > 0.05
NYHA II/III/IV0/7/312/8/0
Mean LVEFa27.1 (11.9)35.4 (9.7)73.0 (6.4)68.4 (6.1)P > 0.05 vs. HF P < 0.01 vs. CAD or HCP < 0.0001
Peak VO2 (mL/min/kg)9.8 (2.3)13.6 (3.9)21.2 (2.0)26.4 (5.5)P < 0.05 vs. HF P < 0.01 vs. CAD or HCP < 0.0001
 Previous MI, n (%)7 (70)15 (75)2 (20)P > 0.05 vs. HF P = 0.030 vs. CAD
 Diabetes mellitus, n (%)3 (30)9 (45)1 (10)P > 0.05
 Hypertension, n (%)1 (10)5 (25)3 (30)P > 0.05
 Mean serum creatinine concentration (µmol/L)127.5 (21.5)129.1 (29.7)96.4 (14.2)85.1 (8.6)P > 0.05 vs. HF P < 0.05 vs. CAD or HCP < 0.0001
 Mean estimated glomerular filtration rate (mL/min/1.73 m2) (modification of diet in renal disease 1)46.2 (6.4)53.4 (13.8)72 (10.3)75.9 (8.4)P < 0.05 vs. HF P < 0.01 vs. CAD or HCP < 0.0001
Drug therapy
 Aspirin7 (70)17 (85)10 (100)P > 0.05
 Loop diuretic9 (90)16 (80)0 (0)P > 0.05 vs. HF P < 0.0001 vs. CAD
 Beta-blocker4 (40)15 (75)7 (70)P > 0.05
 Angiotensin-converting enzyme-inhibitor9 (90)17 (85)7 (70)P > 0.05
 Angiotensin-receptor blockers1 (10)1 (5)1 (10)P > 0.05
 Spironolactone3 (30)4 (20)0 (0)P > 0.05
 Digitalis5 (50)4 (20)0 (0)P > 0.05 vs. HF P = 0.0124 vs. CAD
 Statin7 (70)16 (80)9 (90)P > 0.05
  • aDual region of interest technique—normal ≥55%.

Body composition analysis

Mean BMI was substantially lower in the HF patients with cachexia than in the other groups: P < 0.01 compared with all other groups (Table 2). This was also the case for mean percentage body fat; P < 0.01 compared with all other groups. There was no significant difference in body composition between the HF-no cachexia, CAD-control, and healthy-control groups.

View this table:
Table 2

Body composition, serum adiponectin, and leptin concentrations

CAD, HF, reduced LV systolic function, cachexia (n = 10)CAD, HF, reduced LV systolic function, no cachexia (n = 20)CAD, no HF, preserved LV systolic function (n = 10)Healthy controls (n = 7)HF cachexia vs. each other groupANOVA
BMI (kg/m2)20.0 (5.0)29.2 (4.7)28.8 (3.7)27.0 (4.6)P < 0.01P < 0.0001
Percentage body fat (%)15.5 (3.9)31.7 (6.1)28.5 (5.6)29.2 (9.8)P < 0.01P < 0.0001
Waist circumference (mm)836 (120)1022 (135)977 (105)923 (138)P < 0.01 vs. HFP = 0.0097
Adiponectin concentration (µg/mL)23.8 (10.2–37.2)8.1 (0.5–16.6)7.1 (0.4–13.5)8.7 (2.5–16.8)P < 0.05P < 0.0001
Adiponectin concentration corrected for body fat (µg/mL/kg)2.2 (0.7–8.4)0.3 (0.02–0.9)0.3 (0.08–0.9)0.4 (0.06–1.3)P < 0.05 vs. HF or CAD P > 0.05 vs. HCP = 0.0002
Adiponectin concentration corrected for BMI (µg/mL/kg m−2)1.1 (0.4–3.3)0.2 (0.02–0.6)0.2 (0.01–0.5)0.3 (0.07–0.8)P < 0.05P < 0.0001
Leptin concentration (ng/mL)9.5 (2.1–14.0)12.6 (3.4–55.4)20.8 (12.4–26.3)11.1 (1.5–18.4)P < 0.01 vs. CADP = 0.0245
Leptin concentration corrected for body fat (ng/mL/kg)1.0 (0.2–2.8)0.5 (0.2–1.7)0.8 (0.7–1.3)0.4 (0.1–0.8)P > 0.05P = 0.0110
Leptin concentration corrected for BMI (ng/mL/kg m−2)0.6 (0.1–0.8)0.4 (0.1–1.6)0.7 (0.5–0.9)0.4 (0.1–0.7)P > 0.05P = 0.00721

Serum adiponectin concentration

The HF and cachexia group had a significantly higher median serum concentration of adiponectin than those in each of the other groups: P < 0.05 for all comparisons (Table 2). This remained true after adiponectin concentration was adjusted for fat mass or BMI. Because adiponectin shows sexual dimorphism (with higher serum concentrations in females), we also repeated this analysis in male subjects only. Adiponectin levels remained significantly higher in HF patients with cachexia: 22.5 (10.2–37.2) µg/mL compared with 7.1 (0.5–16.6) µg/mL in the HF no-cachexia group and 7.0 (0.4–10.7) µg/mL in the HF controls and 9.8 (2.5–16.8) µg/mL in the healthy controls.

Serum leptin concentration

Patients in the HF-cachexia group had the lowest median serum leptin concentration, although there was no statistically significant difference from the other groups with the exception of the CAD-control group (Table 2). This difference in leptin was no longer significant after adjusting for fat mass or BMI.

Concentrations of other serum cytokines

Serum TNF-α concentration was significantly higher in the HF patients with cachexia than in the HF patients without cachexia and the healthy controls: P < 0.01 for both comparisons. Although TNF-α concentration was higher in HF patients with cachexia than in the CAD controls, the difference was not statistically significant (Table 3). Although IL-6 concentration was higher in HF patients with cachexia than in all other groups this was only significant on comparison with the CAD controls (P < 0.05).

View this table:
Table 3

Serum cytokine concentrations

CAD, HF, reduced LV systolic function, cachexia (n = 10)CAD, HF, reduced LV systolic function (n = 20)CAD, no HF, preserved LV systolic function (n = 10)Healthy controls (n = 7)HF cachexia vs. each other groupANOVA
Tumour necrosis factor-α (pg/mL)27.2 (21.8–132.8)11.7 (2.6–27.3)25.6 (5.3–51.3)10.8 (3.6–16.3)P < 0.01 vs. HF or HC P > 0.05 vs. CADP < 0.0001
Interleukin-6 (pg/mL)12.2 (3.1–25.7)8.2 (3.4–22.8)5.5 (2.6–10.4)5.4 (3.1–13.6)P > 0.05 vs. HF or HC P < 0.05 vs. CADP = 0.0280
B-type natriuretic peptide (pg/mL)484 (79–1609.5)151 (22.5–557)58.5 (3.0–173.7)22.5 (5.5–72.0)P < 0.01 vs. CAD or HC P > 0.05 vs. HFP < 0.0001
C-reactive protein (mg/L)3.8 (0.2–12.5)3.7 (0.2–28.6)2.3 (0.3–18.4)1.6 (0.5–3.7)P > 0.05P = 0.2732

C-reactive protein

There was no significant difference in C-reactive protein concentration between the groups (Table 3).

B-type natriuretic peptide

Patients in the HF-cachexia group had higher median plasma BNP concentration than HF patients without cachexia, although this difference was not statistically significant. BNP concentrations were significantly higher in the HF-cachexia group than in the CAD-control group and the healthy controls: P < 0.01 for both comparisons (Table 3).

Lipid profile

There was no significant difference in the lipid profile of patients with HF and cachexia than the other patient groups (Table 4).

View this table:
Table 4

Lipid profile

CAD, HF, reduced LV systolic function, cachexia (n = 10)CAD, HF, reduced LV systolic function (n = 20)CAD, no HF, preserved LV systolic function (n = 10)Healthy controls (n = 7)HF cachexia vs. each other groupANOVA
Cholesterol (mmol/L)3.4 (2.2–5.5)3.8 (2.6–5.4)3.7 (3.1–5.1)4.5 (3.2–6.4)P > 0.05P = 0.1501
Triglycerides (mmol/L)1.2 (0.8–4.9)1.8 (0.9–5.0)1.2 (0.7–2.0)0.7 (0.6–1.4)P > 0.05P = 0.0053
VLDL cholesterol (mmol/L)0.5 (0.4–2.3)0.8 (0.4–2.3)0.5 (0.3–0.9)0.3 (0.3–0.6)P > 0.05P = 0.0054
LDL cholesterol (mmol/L)1.4 (0.9–3.3)2.1 (0.9–3.6)2.0 (1.2–3.1)2.6 (1.6–4.2)P > 0.05P = 0.1545
HDL cholesterol (mmol/L)0.8 (0.5–1.6)0.8 (0.6–1.5)1.1 (0.9–1.6)1.3 (1.3–2.2)P > 0.05P = 0.0022

Insulin resistance

The median HOMA-IR index was 1.3 (0.5–33.1) in the HF-cachexia group, 5.9 (0.7–24.8) in the HF-no cachexia group, 2.2 (0.7–7.3) in the CAD-control group, and 1.5 (0.7–3.7) in the healthy control group. The groups were not statistically significantly different.


In all patients studied (excluding healthy controls), BMI, percentage body fat, and waist circumference negatively correlated with adiponectin (r = −0.47, r = −0.47, and r = −0.44, respectively; P < 0.01 for all relationships) and positively correlated with leptin concentration (r = 0.56, r = 0.52, and r = 0.52, respectively; P ≤ 0.003 for all relationships).

BNP positively correlated with adiponectin (r = 0.55, P = 0.0004) and negatively correlated with leptin (r = −0.36, P = 0.0363). There was an inverse correlation of BNP with BMI (r = −0.37, P = 0.0225) and percentage body fat (r = −0.45, P = 0.006).

There was an inverse relationship between adiponectin and insulin resistance (r = −0.43, P = 0.0077).

Adiponectin positively correlated with IL-6 (r = 0.34, P = 0.0325) while no relationship was observed with TNF-α (r = 0.27, P = 0.0928; Run's test P < 0.05 with significant departure from linearity).

Multiple linear regression analysis, using a model adjusting for age, BMI, eGFR, and insulin resistance, demonstrated a positive relationship between adiponectin and BNP (r = 0.57, P = 0.0017) and TNF-α (r = 0.41, P = 0.0313).


We found that patients with HF and cachexia had striking increases in plasma adiponectin concentrations compared to those with HF without cachexia.

At the very least, this suggests preservation of the supposed ‘physiological’, inverse, relationship between adiponectin and fat mass.12,13 More interestingly, our finding also raises the possibility that adiponectin contributes to cachexia in HF, as suggested by recent studies showing that adiponectin causes weight loss in experimental animals.1719 Alternatively, an increase in adiponectin might represent a physiological response designed to either improve fatty acid utilization in such patients or dampen inflammation.

Adiponectin levels have been inversely correlated with body mass and fat in the disordered metabolic states of obesity and anorexia nervosa. Adiponectin concentrations are decreased in obesity and weight loss in obese subjects induced by dieting or gastric surgery is associated with an increase in adiponectin concentrations.12,13,23 However, in recent studies in non-obese healthy volunteers, weight loss has been shown to be associated with no increase or, commonly, a fall in plasma adiponectin concentrations.24,25 Thus, the relationship between change in weight and change in adiponectin may not be the same in non-obese and obese individuals and the response of adiponectin to weight loss in cardiac cachexia appears to be quite different than that in healthy volunteers. In fact, the directional change in plasma adiponectin concentrations in cardiac cachexia appears similar to that in patients with anorexia nervosa, although eating disorder involves starvation (which is not a feature of cachexia) and neuropsychiatric abnormalities, which may alter endocrine function. Furthermore, the increases in adiponectin reported in anorexia nervosa have been much less marked than we found in cardiac cachexia.1416 Obviously, a more direct way of establishing a ‘cause-and-effect’ relationship between increased adiponectin and reduced weight would be to examine change in weight following administration of the peptide. This has not been done in humans. However, to date, the studies conducted in experimental animals have shown a consistent effect of adiponectin to either prevent weight gain or induce weight loss.1719

What are the mechanisms for the increased plasma concentrations of adiponectin in patients with HF? The factors controlling adiponectin secretion are still poorly understood and controversial, though it is striking that our cachectic patients had much higher levels despite a greatly reduced ‘endocrine’ mass (i.e. fat tissue). TNF-α has been suggested to increase adiponectin secretion, although this literature is conflicting.2629 We found elevated TNF-α levels in patients with HF complicated by cachexia, as shown previously.1,30,31 This finding could also be interpreted as being consistent with the proposed anti-inflammatory role of adiponectin.

We also found an association between BNP and adiponectin, raising the possibility that natriuretic peptides might also promote adiponectin secretion. Recently, a novel lipolytic and potential lipid-mobilizing effect of natriuretic peptides has been identified.31 These actions appear to be mediated by specific adipocyte membrane receptors, which operate via a cGMP-dependent pathway and they may indirectly stimulate adiponectin production.31 Our findings are consistent with Kistorp et al.20 who found a positive relationship (albeit of lesser magnitude than seen in the present study) between plasma adiponectin levels and NT-proBNP concentration. BNP has previously been demonstrated to correlate inversely with BMI.32 In addition to observing a similar inverse relationship of BNP with BMI in our study, we also found an inverse relationship between BNP and percentage body fat.

Reduced renal clearance, rather than increased secretion, might also account for our findings in cachectic HF patients, since adiponectin is cleared from the circulation by the kidneys.12,13 While many of our patients had reduced renal function, as evidenced by a reduced glomerular filtration rate, there was no marked difference in renal function between cachectic and non-cachectic patients, whereas there was in adiponectin. Thus, renal dysfunction is unlikely to explain the higher adiponectin concentrations in HF patients with cachexia.

Adiponectin is also thought to regulate glucose and fatty acid metabolism, acting as an insulin sensitizer. It was surprising, however, to find that adiponectin was increased in our patients as HF is a state of insulin resistance, which is usually associated with a decrease in adiponectin concentrations. Patients with advanced HF often show an increase in plasma fatty acids with resultant decrease in cardiac glycolysis. Increased adiponectin secretion might, therefore, occur in an attempt to attenuate this by enhancing fatty acid metabolism.33

In summary, cachexia in HF is associated with a significant increase in serum adiponectin concentrations. This may represent preservation of the ‘physiological’ response of adiponectin to change in body weight but also raises the possibility that adiponectin may play a role in the pathogenesis of cachexia, given the experimental evidence that administration of this hormone leads to weight loss. The answer to whether increased adiponectin is a cause or consequence of cachexia requires further study.

Our study has limitations, particularly due to the cross-sectional design and small number of subjects. We would point out that patients with HF and cachexia are uncommon and are often so sick as to preclude inclusion in a study like this. It is possible that the lack of significant difference observed between some of the biochemical parameters we measured might be the result of a type II error.


Supported by British Heart Foundation Fellowship FS/03/062/15890.

Conflict of interest: none declared.


View Abstract