Dynamics in Insulin Resistance and Plasma Levels of Adipokines in Patients with Acute Decompensated and Chronic Stable Heart Failure

Abstract

Background

Patients with heart failure (HF) develop metabolic derangements including increased adipokine levels, insulin resistance, inflammation and progressive catabolism. It is not known whether metabolic dysfunction and adipocyte activation worsen in the setting of acute clinical decompensation, or conversely, improve with clinical recovery.

Methods and Results

We assessed insulin resistance using HOMA-IR, and measured plasma levels of NT-proBNP, adiponectin, visfatin, resistin, leptin, and TNFα in 44 patients with acute decompensated HF (ADHF) due to left ventricular (LV) systolic dysfunction and again early (<1 week) and late (> 6 months) after clinical recovery; 26 patients with chronic stable HF; and 21 patients without HF. NT-proBNP was not increased in controls, mildly elevated in patients with stable HF, markedly elevated in patients with ADHF, and decreased progressively early and late after treatment. Compared to controls, plasma adiponectin, visfatin, leptin, resistin and TNFα were elevated in patients with chronic stable HF and increased further in patients with ADHF. Likewise, HOMA-IR was increased in chronic stable HF and increased further during ADHF. Adiponectin, visfatin and HOMA-IR remained elevated at the time of discharge from the hospital, but returned to chronic stable HF levels. Adipokine levels were not related to BMI in HF patients. HOMA-IR correlated positively with adipokines and TNFα in HF patients.

Conclusions

ADHF is associated with worsening of insulin resistance and elevations of adipokines and TNFα indicative of adipocyte activation. These metabolic abnormalities are reversible, but temporally lag behind the clinical resolution of decompensated HF.

INTRODUCTION

Patients with advanced heart failure (HF) develop metabolic abnormalities and progressive multi-organ functional impairment which has been characterized as the syndrome of chronic HF. These abnormalities include derangements in metabolism leading to progressive catabolism, weight loss and cachexia (1) linked to worse prognosis in patients with chronic stable HF(2). Not limited to advanced HF, a progressive catabolic state has been described in various chronic disease states including cancer, uremia, arthritis and chronic infections affecting structure and function of various body compartments (3). These changes are accompanied by elevated circulating markers of chronic catabolism such as catabolic steroids, catecholamines, proinflammatory cytokines (1,4), and growth hormone and insulin resistance (5-7).

Although chronic disease, primarily due to infections or autoimmunity, is associated with changes in adipocyte secretory function (8), the effects of HF and especially ADHF on adipokine production are less well known. Functional changes in adipose tissue have been linked in several studies to systemic metabolic markers with both anabolic and catabolic effects (8-10). Of note, inflammatory cytokines and neurohormones such as TNFα, catecholamines and angiotensin II that are known to be elevated in HF mediate increased lipolysis and insulin resistance (4,11-13). Whether secreted factors originating from adipose tissue, so called “adipokines”, that mediate important effects in systemic metabolism including insulin sensitivity could lead to metabolic changes in ADHF is unclear (8).

Notably, changes in circulating levels of adipokines have previously been described in chronic stable HF (4,7). Intrinsic cardiac adipokine systems with active regulation of myocardial adiponectin and leptin signaling in HF have been demonstrated by different groups (14-16). Further, the adipokine adiponectin mediates important anti-inflammatory, anti-proliferative and anti-hypertrophic effects (17-19). Prior studies on patients with HF have been limited to a small and selected number of adipokines which limits the applicability of these findings.

To date, no study has systemically analyzed a cluster of adipokines with known cardiovascular and metabolic effects in the setting of acute decompensated HF and during the course of hemodynamic recovery in order to define whether a specific pattern of adipokine activation correlates with clinical status of patients with HF. To do this, we measured circulating adipokine levels implicated in insulin resistance and sensitivity in patients with HF during acute hemodynamic decompensation and after clinical recovery, in chronic stable HF and in control subjects without evidence of HF.

METHODS

Study Cohort

Patients admitted with acute decompensated heart failure (ADHF), patients with stable chronic HF and controls were included into this study. A subset of the patients with ADHF was followed during early and late hemodynamic recovery for the assessment of dynamics in serum markers. Patients with chronic stable HF were included when in stable hemodynamic condition on chronic medication for more than three months. Patients had to be euvolemic by clinical examination and without hospital admission within the prior two months.

Patients were excluded if they had evidence of inflammatory or infectious conditions, severe renal or hepatic dysfunction, active malignancy, rheumatologic disorders, uncontrolled hypertension, severe hemodynamic instability, acute coronary syndrome or evidence of left ventricular outflow obstruction. Controls were recruited from patients admitted to the hospital with normal cardiac function and no history of cardiac dysfunction. All patients were recruited at Boston University Medical Center. The study protocol was approved by the internal review board and written informed consent obtained from all study participants.

Analysis of Adipokines

Fasting blood samples were collected in all individuals after inclusion into the study. In the ADHF group, samples were also collected at the time of hospital discharge and during a clinical follow-up visit. Samples from patients with stable HF were collected during one of the outpatient clinic visits. Samples from control subjects were collected during the in-hospital stay. All samples were immediately centrifuged after collection for separation of serum and plasma and stored at −80C until analysis.

NT-proBNP (Alpco Diagnostics), adiponectin (EIA Phoenix Pharmaceuticals) and visfatin (EIA Phoenix Pharmaceuticals) were analyzed in plasma samples using commercially available ELISA kits. Resistin, leptin, TNFα, insulin and IGF-1 were measured in plasma samples using multiple-ELISA (Phoenix Pharmaceuticals). All samples were run in duplicates. Routine laboratory testing was performed for all study subjects at several time points during the study period. Insulin resistance was assessed by the HOMA-IR quotient (mU/mL*mg/dL).

Statistical Analysis

Mean value ± standard deviation was calculated for all variables. Multiple group comparisons were analyzed for continuous variables utilizing the parametric Student’s t-test and the non-parametric Kruskal-Wallis test; chi-square comparisons were used for categorical variables. Correlational analysis was performed using Spearman’s rank correlation. Comparisons among multiple groups for all biomarkers were performed using ANOVA and Tukey multiple-comparison post-hoc tests, or Kruskal-Wallis test for non-normally distributed variables. Time courses of variables were analyzed using repeated measures ANOVA with appropriate post-hoc tests. A p-value of less than 0.05 was considered significant.

RESULTS

Clinical Characteristics

Baseline characteristics of all patients are described in Table 1. A total of 44 patients with ADHF, 26 patients with chronic HF and 21 controls were enrolled into this study. The mean age of patients with ADHF was 61±2 yrs in stable HF and 57±3 yrs in controls. Mean LVEF was 27±4% in patients with stable HF (p=NS between groups). Mean BMI in all patients with HF was 30.8±0.7 kg/m² versus 30.7±1.2 kg/m² in controls. A total of 32% of the study population had a history of diabetes mellitus (34% in patients with HF versus 24% in controls). 74% of all patients were previously diagnosed with hypertension (77% in patients with HF versus 62% in controls). Ischemic heart disease was the most common cause of HF (>50%).

Table 1. Baseline Characteristics.

	ADHF (n=44)	CHF (n=26)	Controls (n=21)	P value§
Demographics
Age, years	61 ± 12	63 ± 12	57 ± 13	0.146
Body mass index, kg/m2	32.4 ± 6.3	28.1 ± 4.6	30.7 ± 5.7	0.016
Female, N (%)	11 (29.7)	5 (19.2)	8 (38.1)	0.355
Concomitant diseases
Coronary artery disease, N (%)	22 (51.2)	14 (53.9)	4 (19.1)	0.027
Ischemic/dilated cardiomyopathy, N (%)	22/22 (50/50)	14/12 (54/46)	-	0.654
Diabetes, N (%)	17 (39.5)	7 (26.9)	5 (23.8)	0.356
Hypertension, N (%)	35 (81.4)	19 (73.1)	13 (61.9)	0.240
Heart failure parameters
Left ventricular ejection fraction, %	28 ± 14	25 ± 15	63 ± 3	<0.001
NYHA*, N (%)   Class II   Class III   Class IV	0 (0.0) 0 (0.0) 43 (100)	11 (42.3) 2 (7.7) 0 (0.0)	--- --- ---
NT-proBNP, fmol/mL	1810 ± 1317	704 ± 428	331 ± 127	<0.001
Creatinine, mmol/dL	1.4 ± 0.7	1.3 ± 0.5	0.8 ± 0.2	<0.001
Systolic blood pressure, mm Hg	128 ± 20	120 ± 26	132 ± 16	0.190
Diastolic blood pressure, mm Hg	76 ± 13	71 ± 9	75 ± 8	0.156
Metabolic characteristics
Fasting insulin, μU/mL	9.1 ± 2.7	4.1 ± 1.8	3.1 ± 2.0	<0.001
Fasting glucose, mg/dL	136.4 ± 36.1	142.9 ± 41.9	92.7 ± 19.1	<0.001
HOMA-IR†, μU/mL*mg/dL	3.3 ± 1.2	2.5 ± 0.9	1.3 ± 0.9	<0.001
Adiponectin, ng/mL	42.8 ± 16.2	29.0 ± 21.3	15.0 ± 12.8	<0.001
Leptin, ng/mL	9.6 ± 5.7	7.1 ± 4.7	3.7 ± 2.7	<0.001
Resistin, ng/mL	11.5 ± 7.7	9.5 ± 5.1	5.7 ± 3.8	0.017
Visfatin, ng/mL	79.3 ± 26.2	43.4 ± 35.5	16.9 ± 12.3	<0.001
Tumor necrosis factor α, ng/mL	16.5 ± 9.0	13.8 ± 8.9	4.8 ± 2.2	<0.001
Pharmacotherapy
Beta-blockers, N (%)	34 (79.1)	24 (92.3)	9 (42.9)	<0.001
ACEI/ARBs‡, N (%)	36 (83.7)	21 (80.8)	9 (42.9)	0.001
Spironolactone, N (%)	3 (7.0)	4 (15.4)	0 (0.0)	0.413
Diuretics, N (%)	43 (100)	22 (84.6)	4 (19.1)	<0.001
Statin, N (%)	25 (58.1)	19 (73.1)	10 (47.6)	0.196
Insulin, N (%)	5 (11.6)	3 (7.0)	1 (5.0)	0.345
Oral antidiabetics, N (%)	9 (21)	7 (26.9)	3 (15.3)	0.136

Value for continuous variables are shown as mean (SD); categorical variables are represented by mean (%).

^§Difference between 3 groups by Kruskal-Wallis test (continuous variables); chi-square test utilized for categorical

^*NYHA: New York Heart Association

^†HOMA-IR: homeostasis model assessment-insulin resistance

^‡ACEI: angiotensin-converting enzyme inhibitor, ARBs: Angiotensin receptor blocker ADHF-acute decompensated heart failure; CHF-chronic stable heart failure

Mean duration of in-hospital treatment in the ADHF group was 3.5±2.1 days. Patients showed a significant weight loss under intense diuretic therapy during their in-hospital stay (-3.5 kg; p<0.05 versus admission). None of the patients died during their admission or during follow up. Mean clinical follow-up time after decompensation was 8.4±4.7 months.

Analysis of Insulin Resistance in ADHF and Chronic Stable HF

Analysis of insulin resistance using the HOMA quotient revealed baseline insulin resistance in patients with stable HF compared to controls (Figure 1E). Acute decompensation worsened insulin resistance with an approximately 50% increase in the HOMA quotient. Notably, clinical recovery of patients with ADHF corrected the HOMA quotient and, therefore, improved insulin resistance (Figure 1F). Patients with stable HF and patients who had been treated for previous ADHF had identical levels of insulin resistance (Figure 1E and F). Clinical characteristics of patients stratified for the presence of diabetes mellitus are shown in Table 2.

Figure 1. Serum Levels and Time Course of Neurohumoral Activation and Insulin Resistance in Patients with ADHF, CHF and controls.

(A) Plasma levels of NT-proBNP in controls (n=21), patients with stable CHF (n=26) and acute decompensated HF (ADHF, n=44). (B) Dynamics in circulating levels of NT-proBNP from admission to the hospital and upon discharge and at clinical recovery (n=16). (C) Plasma levels of TNFα in controls, CHF and ADHF and (D) during in-hospital treatment and after recovery. (E) Insulin Resistance assessed by HOMA quotients in controls, CHF and ADHF and (F) changes in IR during in-hospital treatment and after recovery († p<0.05 vs. controls; ‡ p<0.01 vs. controls; $ p<0.05 vs. CHF; * p<0.05 vs. admission; § p<0.01 vs. admission; # p<0.05 vs. discharge).

Table 2. Baseline Characteristics of Patients with and Without Diabetes Mellitus.

	Diabetes (n=29)	No Diabetes (n=62)	P value§
Demographics
Age, years	62 ± 11	60 ± 13	0.532
Body mass index, kg/m2	32.2 ± 5.7	30.1 ± 6.0	0.11
Female, N (%)	8 (27.6)	16 (25.8)	0.858
Concomitant diseases
Coronary artery disease, N (%)	16 (55.2)	24 (38.7)	0.176
Ischemic/dilated cardiomyopathy, N (%)	19/10 (65/35)	31/31 (50/50)	0.154
Hypertension, N (%)	26 (86.2)	42 (67.7)	0.077
Heart failure parameters
Left ventricular ejection fraction, %	34 ± 17	29 ± 18	0.197
NYHA*, N (%)   Class 0   Class I   Class II   Class III   Class IV	5 (17.2) 3 (10.3) 3 (10.3) 1 (3.4) 17 (58.6)	16 (25.8) 10 (16.1) 8 (12.9) 1 (1.6) 27 (43.5)	0.657
NT-proBNP, fmol/mL	1041 ± 1087	1155 ± 1154	0.663
Creatinine, mmol/dL	1.3 ± 0.7	1.2 ± 0.6	0.359
Systolic blood pressure, mm Hg	132 ± 21	124 ± 21	0.112
Diastolic blood pressure, mm Hg	76 ± 10	73 ± 11	0.194
Metabolic characteristics
Fasting insulin, μU/mL	4.3 ± 1.7	4.4 ± 2.0	0.792
Fasting glucose, mg/dL	162.3 ± 41.1	110.1 ± 37.5	<0.001
HOMA-IR†, μU/mL*mg/dL	3.0 ± 1.2	2.2 ± 1.4	0.016
Adiponectin, ng/mL	28.1 ± 15.4	34.5 ± 22.6	0.22
Leptin, ng/mL	7.36 ± 4.9	7.54 ± 5.8	0.897
Resistin, ng/mL	9.57 ± 7.6	9.5 ± 6.25	0.967
Visfatin, ng/mL	57.6 ± 37.4	58.4 ± 37.1	0.936
Tumor necrosis factor α, ng/mL	14.9 ± 11.0	11.2 ± 7.3	0.09
Pharmacotherapy
Beta-blockers, N (%)	43 (69.4)	25 (86.2)	0.121
ACEI/ARBs‡, N (%)	25 (85.9)	42 (67.8)	0.121
Spironolactone, N (%)	2 (6.9)	6 (9.7)	0.967
Diuretics, N (%)	25 (86.2)	45 (72.6)	0.188
Statin, N (%)	19 (65.5)	36 (58.1)	0.646
Insulin, N (%)	9 (30.3)	0 (0)	<0.001
Oral antidiabetics, N (%)	18 (62.1)	1 (1.6)	<0.001

Plasma Levels of Adipokines

Plasma levels of all adipokines analyzed were found to be elevated in patients with ADHF compared to controls and patients with stable HF (Figure 2). Patients with stable HF showed significant elevations of adiponectin, visfatin and resistin. Plasma levels of these adipokines were further increased in patients with ADHF. This was only partially corrected during acute clinical stabilization and recovery; notably, none of the adipokines was found to be decreased significantly during the acute recovery period between admission and discharge. Patients in the highest quartile of NT-proBNP values were found to have a two-fold increase of visfatin levels compared to lower NT-proBNP values.

Figure 2. Serum Levels and Time Course of Changes in Adipokines in Patients with ADHF, CHF and controls.

Adiponectin (A), visfatin (C), leptin (E) and resistin (G) were detected in controls (n=21), patients with stable CHF (n=26) and acute decompensated CHF at admission (n=44) to the hospital. In a subgroup, all adipokines were assessed again at discharge and after recovery (B, D, F, H) (n=16) († p<0.05 vs. controls; ‡ p<0.01 vs. controls; $ p<0.05 vs. CHF; * p<0.05 vs. admission; § p<0.01 vs. admission; # p<0.05 vs. discharge).

In a subset of patients admitted with ADHF, dynamics of adipokine levels were studied. Clinical recovery was associated with a decrease in adipokine levels close to concentrations as seen in patients with chronic stable HF (Figure 2 B, D, F, H). This indicates acute adipocyte activation with increased circulating levels of adipokines in the setting of decompensated heart failure and correction of this dysregulation during clinical recovery.

Correlation of Insulin Resistance with Adipokine and TNFα Plasma Levels

Circulating adipokine levels correlated with insulin resistance and levels of the proinflammatory cytokine TNFα in the entire study population. Levels of adipokines and insulin resistance did not correlate in the group of patients with ADHF even after adjustment for body surface area or history of diabetes and also no correlation was found after clinical recovery. Notably, a group of adipokines including adiponectin, leptin resistin and visfatin positively correlates with insulin resistance indicating a co-regulation of these adipokines (Table 3). Of note, this correlation was only detectable in patients with no history of diabetes while diabetic patients showed metabolic derangement with lack of correlation between adipokines and insulin resistance. Fasting insulin levels correlated with leptin, resistin and visfatin and showed a trend towards a correlation with adiponectin (Supplemental Table 1). No clear co-regulation between different adipokines was found in the current study. No correlation was detectable in the study cohorts between BMI and adipokines and NT-proBNP values. TNFα levels showed a modest correlation to adiponectin (rho 0.26; p<0.05), visfatin (rho 0.39; p=0.004) and resistin levels (rho 0.26; p<0.05) (rho 0.275; p<0.05) as well as NT-proBNP (rho 0.397; p=0.001). Patients with >20% improvement in HOMA-IR levels showed higher levels of TNFα and fasting glucose levels and lower levels of fasting insulin without differences in adipokine levels at admission. Of note, patients with improvements in their HOMA-IR levels showed a non-significant trend towards a more pronounced decrease in NT-proBNP and TNFα levels compared to patients without changes or increases in HOMA-IR levels (Supplemental Table 2).

Table 3. Correlation of insulin resistance and adipokines in the entire study cohort and stratified for diabetes status.

Variables	Rho	P-Value
Adiponectin (n=63) + diabetes (n=25) − diabetes (n=38)	0.37 0.14 0.56	0.004 0.50 <0.001
Leptin (n=66) + diabetes (n=27) − diabetes (n=39)	0.27 0.21 0.32	0.03 0.28 0.04
Resistin (n=66) + diabetes (n=27) − diabetes (n=39)	0.25 0.26 0.27	0.05 0.20 0.10
Visfatin (n=57) + diabetes (n=24) − diabetes (n=33)	0.49 0.40 0.58	<0.001 0.05 <0.001
TNFα (n=67) + diabetes (n=28) − diabetes (n=39)	0.51 0.29 0.56	<0.001 0.14 <0.001

DISCUSSION

Our current study provides evidence for profound changes in plasma levels of adipocyte-derived secretory molecules during both chronic stable and acute decompensated HF, and demonstrates the reversibility of the ADHF-associated adipokine elevations after hemodynamic recovery. We here show that ADHF leads to elevations of adipokines with both hyper- and hypoglycemic action, including adiponectin, visfatin, leptin and resistin. Despite the increase in adiponectin, an adipokine associated with improved insulin actions, higher levels of adiponectin were found in the setting of the worst insulin sensitivity. The abnormalities in HF decompensation partially correct to levels seen in patients with stable chronic HF. Stable chronic HF is also associated with elevation of adipokines; however, to a lesser degree.

The syndrome of chronic HF is characterized by changes in multiple endocrine systems (1). After initial impairment of cardiac function, secondary abnormalities develop in peripheral organ systems including the lungs, liver, gastrointestinal tract, kidneys and skeletal muscle (20). This progressive multi-organ impairment is linked to an overall deterioration of functional status best evaluated by metabolic exercise testing (21). It has been demonstrated that a progressive catabolic state develops in advanced HF (1) which is associated with weight loss (2) and reduced exercise tolerance (22) that have both been associated with worse prognosis. Notably, these parameters are part of the assessment algorithm used for the evaluation of patients with advanced cardiac failure for cardiac transplantation (21).

While the structural and functional impairment of skeletal muscle in HF has been well studied, only limited data are available on adipose tissue function, metabolism and structure (4,7,12,23). This is surprising given the central role of adipose tissue as a regulator of metabolism but might reflect the previous assumption that it merely functions as an energy storage pool (8,24). Multiple studies have demonstrated secretory, metabolic and structural changes of adipose tissue in obesity (8), chronic inflammation (9,10) and cancer (25). Adipose tissue secretion of a number of proteins including adiponectin, visfatin, leptin and resistin (24,26) has been assessed and TNFα, PAI-1 and IL-6 have been used as markers of systemic inflammation (11,24). Limited data are available on the regulation of these adipokines in the setting of chronic HF. We and others have previously demonstrated increased levels of leptin (7,12,27,28) and the soluble leptin receptor (4) in chronic stable HF. A small study showed increased plasma levels of resistin in HF (29). Although adiponectin improves insulin sensitivity, high adiponectin levels have been linked to increased mortality in patients with HF (30-32) but reduced mortality in patients at risk for myocardial infarction (33). Multiple molecular effects of this adipokine both systemically and also in the myocardium have been described including its function as an anti-inflammatory and anti-atherogenic molecule (17-19).

The current study analyzed for the first time levels of an entire cluster of adipokines in patients over a range of clinical stages of HF including ADHF and chronic stable HF as well as in controls. Circulating adipokine levels correlated with insulin resistance in the study cohort but this finding was absent when the group of patients with ADHF was analyzed independently. The lack of direct correlation between adipokine levels and insulin resistance in the ADHF group underlines our hypothesis that metabolic derangements in this severely compromised group disrupt physiologic interactions and signaling networks. Further, we demonstrated the time-course of its correction during late but not early hemodynamic recovery. Although others have noted increased circulating levels of adiponectin and leptin in stable HF (4,12,30-32), our study not only provides a cross-sectional analysis in different clinical stages of HF but also demonstrates the rapid induction and slow correction of these markers of adipocyte function in the setting of acute cardiac decompensation and during clinical recovery. We further provide evidence of a relation of adipokine levels to markers of inflammation and the degree of insulin resistance. We hypothesize that changes in energy homeostasis during progression of HF symptoms are due, at least in part, to adipocyte activation leading to increased circulating levels of adipokines.

Our study included adipokines with both hyper- and hypo-glycemic effects. The adipokines adiponectin, leptin and visfatin have been described as insulin-synergistic with significant insulin-sensitizing function (8,26). In contrast, resistin and TNFα have been described as counteracting insulin function through inhibition of the insulin signaling cascade (7,8,11,34). Notably, the current study demonstrates a synergistic induction of all adipocyte markers implicating a general, non-discriminating activation of adipocytes most notably in the setting of acute cardiac decompensation. At this point, we hypothesize that adipocyte activation is mediated through circulating factors in the setting of acute stress.

A limitation of the current study is the analysis of circulating levels of adipokines only. While adipokines are predominantly expressed and secreted by adipocytes, our current study can only speculate about the specific origin of its expression and secretion. It is known that adipokines are expressed and secreted in both the subcutaneous and visceral adipose tissue. Accumulation of visceral adipose tissue carries a higher risk of cardiovascular disease as well as negative metabolic effects including type II diabetes than subcutaneous adipose tissue (35). Also, our finding of increased insulin resistance in patients with ADHF is based on the homeostasis model assessment for insulin resistance (HOMA-IR) instead of the gold standard of steady state plasma glucose using insulin suppression tests. Further, specific analysis of gene expression and protein levels as well as functional studies in adipose tissue biopsies might answer this question in upcoming studies.

In conclusion, chronic HF is associated with increased circulating levels of various adipokines that correlate with markers of inflammation and insulin resistance in both chronic stable and acutely decompensated patients. Several of these adipokines have important effects on insulin signaling and energy homeostasis. Therefore, they may contribute to progressive metabolic abnormalities in patients with HF.

ABBREVIATIONS LIST

ADHF

acute decompensated heart failure

ELISA

enzyme-linked immunosorbent assay

HF

heart failure

HOMA-IR

homeostasis model of assessment – insulin resistance

IGF-1

insulin-like growth factor-1

NT-proBNP

N-terminal – pro-brain natriuretic peptide

TNFα

tumor necrosis factor alpha