Plasma YKL-40 levels are elevated in patients with chronic heart failure

Abstract

Objectives. Congestive heart failure (CHF) has been associated with elevated biomarker levels reflecting chronic low-grade inflammation. YKL-40 is a biomarker with increasing levels in patients with cardiovascular disease (CVD) of increasing severity. Furthermore, YKL-40 is associated with all-cause and cardiovascular mortality. We investigated plasma YKL-40 levels in patients with CHF and evaluated the possible predictive value with respect to overall mortality and recurrent cardiovascular outcomes. Design. Plasma YKL-40 was measured in 194 CHF patients and in 117 age-matched individuals without CVD. Results. Median YKL-40 levels were approximately 77% higher in patients with CHF (106 (IQR, 66-184) ng/ml vs. 60 (IQR, 42-97) ng/ml, p < 0.0001). We found a trend towards an association of YKL-40 levels with urinary albumin/creatinine ratio (UACR) (β = 0.12, p = 0.08). YKL-40 levels were not predictive of overall mortality (p = 0.59), major cardiovascular events (p = 0.23) or events of incompensation (p = 0.56). Conclusions. Plasma YKL-40 levels are elevated in patients with CHF but show no association with other clinical or paraclinical variables. YKL-40 levels were not predictive of overall mortality or incident cardiovascular events. Most likely, elevated YKL-40 levels in CHF patients are explained by the presence of concomitant diseases but a role of YKL-40 in low-grade inflammation is not excluded.

Key words::

• YKL-40
• low-grade inflammation
• chronic heart failure

Chronic heart failure (CHF) is a condition characterized by clinical symptoms due to cardiac impairment. Importantly, CHF is not merely a consequence of cardiac overload or injury but also results from a complex interaction of genetic, neurohormonal, inflammatory and biochemical changes acting on cardiac myocytes and the cardiac interstitium (1). Accordingly, CHF is associated with endothelial dysfunction and low-grade inflammation (2) as reflected by elevated levels of circulating endothelial cells (3) and elevated levels of cellular adhesion molecules such as E-selectin (E-sel, a marker of inflammation and endothelial activation) (4,5), C-reactive protein (CRP, an unspecific marker of inflammation) (6) and von Willebrand factor (vWf, a marker of endothelial damage/dysfunction). A prognostic impact of these biomarkers has been suggested among patients with CHF (3,4,7).

Elevated levels of the glycoprotein YKL-40 have been observed in inflammatory conditions as well as in conditions characterized by extracellular matrix remodelling (8). Both in vivo and in vitro, YKL-40 is secreted by a variety of cells of the innate immune system and the vasculature involved in endothelial activation and dysfunction (9–11). YKL-40 protein expression is found in macrophages and vascular smooth muscle cells in the atherosclerotic plaque (11). Clinical studies have shown an association between plasma YKL-40 levels and the presence (12–14) and extent of coronary artery disease (CAD) (12) and recently elevated YKL-40 levels have been reported in patients with myocardial infarction (MI) (15). Therefore, YKL-40 seems to be involved in both endothelial dysfunction, the earliest event in the atherosclerotic process, and in clinical conditions characterized by ischemia. Since, CAD is a major cause of CHF in the Western World it could be hypothesized that YKL-40 could also be involved in the pathogenesis of CHF. However, no previous studies have investigated YKL-40 in patients with CHF.

Since chronic low-grade inflammation and endothelial dysfunction is known to be of importance in CHF, we hypothesized that YKL-40 levels are affected by the presence of CHF and that YKL-40 has prognostic impact in these patients, particularly when ischemic heart disease (IHD) is the underlying etiology. Therefore, the objectives of the present study were as follows: (i) to examine YKL-40 levels in patients with CHF and to assess the impact of IHD; (ii) to examine possible determinants of plasma YKL-40 including the heart failure marker NT-proBNP, the cardiovascular markers high sensitive C-reactive protein (hsCRP) and urinary albumin/ creatinine ratio (UACR) and the endothelial markers E-sel and vWf; and (iii) to evaluate the predictive value of YKL-40 in terms of mortality and major cardiovascular events in patients with CHF during follow-up.

Material and methods

Study population

One hundred and ninety four consecutive CHF patients from the Frederiksberg University Hospital Heart Failure Clinic were studied. The original design of the clinic, which has been operating since 1999, has been described in details previously (16). Patients with suspected or known systolic heart failure were referred to the clinic for evaluation and management of the condition by the departments of internal medicine or cardiology of the hospital or by general practitioners serving the area. The patients were offered treatment at the clinic if systolic heart failure was confirmed as defined by symptoms of heart failure combined with a left ventricular ejection fraction (LVEF) ≤ 45% as assessed by echocardiography.

At baseline, all patients were examined by a specialized nurse and a senior physician with training in cardiology at a specialist level. A medical history was obtained and medications were recorded. The physical examination was accompanied by an echocardio-graphic examination, which was supplemented by a radionuclide angiography if the echocardiography was technically insufficient.

A blood sample was obtained after an overnight fasting and 20 min resting period for the measurements of blood glucose, glycosylated hemoglobin A1c (HbA1c), lipids and N-terminal fragment of pro-brain natriuretic peptide (NT-proBNP). A first voided morning urine sample for the measurement of urinary albumin/creatinine-ratio (UACR) was also obtained.

All patients were registered in a heart failure database based on Microsoft Access software, “HJERTER+”, which served as a combined medical record and database (16). Patients were followed up for a median period (range) of 2.6 (0.5-3.9) years and none were lost to follow up. Patients were followed with respect to major cardiovascular events and overall mortality. Information on hospital admission was extracted from central discharge registers in the Danish National Board of Health specified by codes of diagnoses from the International Classification of Diseases (ICD), 10^th revision. Major cardiovascular events were defined according to the ICD-10 codes I20.0-22 and I24 (ischemic heart disease), I42.0 (cardiomyopathy), I46 (cardiac arrest), I50 (heart failure), I63 (cerebral infarction), I65-I66 (transient cerebral ischemia due to occlusion or stenosis of precerebral or cerebral arteries). All deaths were ascertained from central registers in the Danish National Board of Health and confirmed by the Danish Personal Register, which records all deaths in Denmark within two weeks of occurrence.

Control subjects

One hundred and seventeen age-matched individuals were included as a control group. They were recruited for a population study individuals of 50 years and older. The rationale, design and methodology of this study have been described previously (17). None of the participants had any prior history of cardiovascular disease (CVD), hypertension or diabetes mellitus (DM) and were not taking any medication. They were without symptoms or clinical signs of heart failure, and had normal left ventricular ejection fraction (LVEF > 60%).

All participants provided informed written consent, and the study was approved by the local ethics committee of Copenhagen. The investigations conformed to the principles of The Helsinki Declaration.

Definitions

The diagnosis of DM was based on history or ongoing treatment with oral antidiabetic agents or insulin. The diagnosis was confirmed by medical records in each case. Patients without a prior diagnosis of DM were classified as having DM based on baseline fasting blood glucose measured twice being > 6.1 mmol/l as described in detail previously (18). Left ventricular systolic dysfunction was defined as a verified LVEF ≤ 45% by echocardiography. Hypertension was defined as either systolic blood pressure ≥ 140 mmHG or diastolic blood pressure ≥ 90 mmHG. Creatinine clearance and glomerular filtration rate (eGFR) were estimated using the Cockcroft-Gault formula.

Measurements

Plasma YKL-40 was analyzed using an ELISA (Quidel, USA). Measuring range of the assay was 20-300 ng/ml, with intra- and interassay coefficients of variation of 5.8% and 6.0%, respectively. HbA1c was measured on a Cobas Integra analyzer (Roche Diagnostics, Switzerland) using alkaline hematin D-575, reference range (95% confidence interval) 4.6-6.0% (mean 5.2%),

Plasma NT-proBNP was measured using a double antibody sandwich technique with Electro-ChemiLuminescense as signal (Elecsys NT-proBNP, Roche Diagnostics, Switzerland). Assay sensitivity was < 5.9 pmol/l with intra- and interassay coefficients of variation both < 5.0%. C-reactive protein was measured with a highly sensitive, latex-particle-enhanced immunoassay (Roche Diagnostics, Germany) with a measuring range of 0.1-300 mg/L and with a lower detection limit of 0.03 mg/L. Plasma vWf (IU/dl) was measured using ELISA (Dako, Glostrup, Denmark and a defined vWf standard from the National Institute for Biological Standards and Controls, Potters Bar, UK). E-sel (ng/ml) was measured by ELISA (R&D Systems, Abingdon, Oxfordshire, UK). Intra- and interassay coefficients of variation for these ELISAs were < 5% and < 10%, respectively.

Urinary albumin was measured by immunoturbidimetry on a Cobas Bioanalyzer (Roche Products, Switzerland). Lower detection limit was 1 mg/l, coefficients variations < 4.0%. Urinary creatinine was measured by a Jaffé reaction rate, using a kinetic principle to eliminate pseudocreatinines. Urinary albumin excretion was determined as UACR with an upper limit of normal range of 30 mg/g.

Statistical analyses

Following a test of statistical normality, data values are presented as mean ± SD or as median and interquartile range (IQR). For continuous variables, comparisons between the group of patients with CHF and the group of control subjects were performed by independent samples t-test (Student t-test), including Levene's test for equality of variance. Mann-Whitney test was used if Levene's test for equality of variance was significant, or if a variable exhibited a clear non-Gaussian distribution. For categorical variables the χ² test was used. Patterns of distribution were examined by histograms and P-Plot analyses. Baseline values of YKL-40, hsCRP, NT-proBNP and UACR were skewed and were logarithmically transformed using the natural logarithm.

Analyses of associations were performed using linear regression models with plasma YKL-40 as dependent variable and the presence of heart failure as fixed variable. Univariate analyses of correlations of YKL-40 with hsCRP, UACR, NT-proBNP, vWf and E-sel were performed prior to multivariate analyses. Univariate analyses of correlations of YKL-40 with the use of statins, beta-blockers, ACE inhibitors, angiotensin II receptor antagonists and spironolactone were also done. Beside the cardiovascular and inflammatory markers the following variables were likely to influence YKL-40 levels and were included in the multivariate analyses as independent variables: age, gender, BMI, HbA1c, presence of diabetes, IHD and hypertension, LVEF, total cholesterol, HDL, LDL, triglycerides and eGFR.

Cox proportional hazard regression analyses were used to assess adjusted hazard ratios (HR) and 95% confidence intervals (CI) for continuous YKL-40 levels and upper quartile of YKL-40 as dichotom variable with each outcome among the CHF patients: Overall mortality, major cardiovascular event (stroke, myocardial infarction or unstable angina pectoris) and admission due to worsening of CHF. Hazard ratios were calculated for each specific outcome after adjustment for age and sex and in multivariate analyses including age, sex, presence of hypertension, presence of IHD, LVEF and levels of NT-proBNP, hsCRP, UACR, E-sel and vWf. All p-values were calculated as two-sided, and a p-value < 0.05 was considered significant. Analyses were made with the statistical software package SPSS 11.5 (SPSS inc., Chicago, Il).

Results

Table I presents clinical data for the CHF and control group. The groups were matched for age, but the group of patients with CHF had a higher male to female ratio. Median LVEF was 30% (IQR, 25-35%) among the patients with CHF. Patients with CHF had higher serum creatinine values, and the mean (SD) eGFR was 60.4 (28.5) ml/min. The majority of the patients had mild or moderate CHF, 74% were classified as NYHA I/II and 26% as NYHA III-IV As expected, NT-proBNP levels were approximately five times higher in the CHF group as compared with the healthy controls.

Table I. Baseline data on control subjects and CHF patients.

	Patients with CHF	Control subjects	p-value
N	194	117
male	139 (71.6%)	50 (42.7%)	<0.0001
Age, mean (SD), years	71.0 (7.9)	69.3 (10.3)	NS
medication:
Ace inhibitors	128 (66%)	0	<0.0001
Angiotension II receptor antagonists	23 (12%)	0	<0.0001
betablockers	60 (31%)	0	<0.0001
Spironolactone	33 (17%)	0	<0.0001
Statins	44 (23%)	0	<0.0001
Cardiovascular status:
Systolic bp, mean (SD), mmHg	139 (25)	139 (21)	NS
Diastolic bp, mean (SD), mmHg	79 (14)	83 (10)	0.004
LVEF, median (IQR), %	30 (25-35)	>60	<0.0001
NT-proBNP, median (IQR), pg/ml	1209 (469-3047)	256 (164-450)	<0.0001
Cholesterol, mean (SD), mmol/l	5.2 (1.3)	5.9 (1.0)	<0.0001
Diabetes and kidney status:
presence of DM	49 (25.3)	0	<0.0001
HbA1c, mean (SD), %	6.2 (1.2)	5.3 (0.3)	<0.0001
Creatinine, mean (SD), µmol/l	100 (32)	82 (16)	<0.0001
UACR, median (IQR), mg/g	16 (6-46)	6 (4-8)	<0.0001
biomarkers:
YKL-40, median (IQR), ng/ml	106 (66-184)	60 (42-97)	<0.0001
hsCRP, median (IQR), mg/l	4.2 (1.8-7.8)	1.8 (0.8-3.7)	<0.0001
E-selectin, median (IQR), ng/ml	37 (26-52)	36 (25-47)	NS
von Willebrand factor, median (IQR), IU/dl	96 (65-149)	81 (55-132)	NS

BP, blood pressure; DM, diabetes melletus; HbA1c, hemoglobin A1c; hsCRP high sensitive C-reactive protein; IQR, interquartile range; LVEF, left ventricular ejection fraction; N, numbers; NT-proBNP, N-terminal fragment of the prohormone brain natriuretic peptide; SD, standard deviation; UACR, urinary albumin/creatinine ratio.

YKL-40

Median levels of YKL-40 were 76% higher in patients with CHF than in control subjects (106 (66-184) vs. 60 (42-97) ng/ml, p < 0.0001) (Table I). We found no significant impact of gender neither in the total population of both CHF patients and controls subjects (91 (55-159) vs. 79 (50-146) ng/ml, p = 0.16) nor in the CHF group alone (105 (63-173) vs. 118 (84-199) ng/ml, p = 0.24, male vs. female). YKL-40 levels in the subgroup of patients with CHF and DM were not different from CHF patients without DM (104 (62-185) vs. 111 (69-186) ng/ml, p = 0.51).

Significant but weak correlations of YKL-40 was found with NT-proBNP (β = 0.096, p = 0.01) and UACR (β = 0.11, p = 0.01) in univariate analyses. Neither hsCRP (β = 0.035, p = 0.38), E-sel (β = 0.008, p = 0.93) nor vWf (β = − 0.038, p = 0.63) showed any correlation with YKL-40 levels. Similar, no correlations were found between YKL-40 and the use of betablockers (β = − 0.079, p = 0.58), ACE inhibitors (β = 0.24, p = 0.15), spironolactone (β = − 0.14, p = 0.38), angiotension II receptor antagonists (β = −0.15, p = 0.39) or statins (β = 0.06, p = 0.67). In multiple linear regression analyses of determinants of plama YKL-40, there was a trend towards an association with UACR (β = 0.12, p = 0.08) and a trend towards an inverse association with diabetes (β = − 0.48, p = 0.07). None of the other included parameters were nearly significantly associated (all p-values > 0.1).

Table II presents YKL-40 data with regard to NYHA classification and presence of IHD. No difference in YKL-40 levels was found between patients categorized as NYHA classes I/II and III/IV (104 (66-181) vs. 116 (74-196) ng/ml, p = 0.66). Furthermore, we did not observe higher YKL-40 levels in patients with CHF due til IHD when compared to patients with CHF of other etiology (115 (60-190) vs. 103 (72-177) ng/ml, p = 0.92).

Table II. YKL-40, UACR, hsCRP, NT-proBNP, E-selectin and von Willebrand factor according to NYHA classification and etiology of CHF.

	NYHA Classification		p	Heart failure etiology		p
	I/II	III/IV		IHD	Other etiology
N	142	50		107	82
YKL-40, ng/ml	104 (66-181)	116 (74-196)	0.66	115 (60-190)	103 (72-177)	0.92
UACR, mg/g	14.5 (6-38)	17 (7-65.5)	0.14	14 (6-61)	17 (6-33)	0.46
hsCRP, mg/dl	3.3 (1.7-5.7)	7.4 (4.3-18.6)	0.0001	4.2 (2.0-7.3)	4.1 (1.8-9.3)	0.31
NT-proBNP, pg/ml	1242 (456-2760)	1055 (563-4453)	0.26	1358 (471-3349)	1141 (429-2321)	0.53
E-selectin, ng/ml	34 (22-50)	40 (32-56)	0.01	40 (26.5-52)	36 (25.5-53.5)	0.19
von Willebrand factor, IU/dl	101 (65-152.5)	87.5 (62.5-146)	0.71	66 (94.5-152)	60.5 (93-145)	0.43

All values presented as median value (IQR, interquartile range). P value for comparison between NYHA I-II and III-IV and between ischemic etiology or other etiology of CHF.

hsCRP, high sensitive C-reactive protein; IHD, ischemic heart disease; N, numbers; NT-proBNP, N-terminal fragment of the prohormone brain natriuretic peptide; UACR, urinary albumin/creatinine ratio.

Other cardiovascular or endothelial markers.

Levels of NT-proBNP, UACR and hsCRP were higher among patients with CHF as compared with the healthy controls. No significant difference between patients and controls was found for levels of E-sel and vWf (Table I). As for YKL-40, NT-proBNP, UACR and vWf levels were not higher among patients with NYHA class III/IV than patients with NYHA class I/II (Table II). In contrast, hsCRP and E-sel levels were higher among patients with CHF NYHA class III/IV than patients with CHF NYHA class I/II (Table II). As for YKL-40, none of the other markers had higher plasma levels in individuals with CHF of ischemic etiology compared to patients with other CHF etiology (Table II).

Outcomes

Age- and sex-adjusted as well as multivariate adjusted hazard ratios (HR) for YKL-40 as continuous variable and for YKL-40 levels in the upper quartile for each outcome are shown in Table III. Due to a limited number of events, Cox regression analyses for the single endpoints: cardiovascular mortality (n = 29), stroke (n = 8) and MI/unstable angina pectoris (n = 20) were not performed.

Table III. Hazard ratios for YKL-40 levels for overall mortality and cardiovascular outcomes.

	YKL-40, continuous	p	YKL-40, upper quartile	p
Overall mortality, N=44
Age and sex	0.87 (0.62-1.21)	0.40	0.86 (0.43-1.71)	0.66
multivariate**	0.88 (0.58-1.37)	0.59	1.05 (0.49-2.27)	0.90
Major cardiovascular event, N=135
Age and sex	0.95 (0.77-1.16)	0.59	0.99 (0.67-1.46)	0.96
Multivariate**	0.86 (0.67-1.10)	0.23	0.86 (0.54-1.38)	0.54
HF event, N=94
Age and sex	0.98 (0.76-1-26)	0.89	1.25 (0.80-1.95)	0.34
Multivariate**	0.91 (0.67-1.24)	0.56	1.49 (0.90-2.48)	0.13

Presented as HR (95% confidence interval), 1 standard deviation increase in ln variable.

**Adjusted for age, sex, LVEF, presence of hypertension, IHD and diabetes and levels of UACR, hsCRP, NT-proBNP, E-selectin and von Willebrand Factor.

HF, heart failure; hsCRP, high sensitive C-reactive protein; IHD, ischemic heart disease; LVEF, left ventricular ejection fraction; N, numbers; NT-proBNP, N-terminal fragment of the prohormone brain natriuretic peptide; UACR, urinary albumin/creatinine ratio.

Among patients with CHF a total of 44 deaths, 135 major cardiovascular events and 94 events of admission due to incompensation were recorded during follow up. No association between continuous YKL-40 levels and any of the outcomes was observed, neither in the age- and sex-adjusted models nor in multivariate models. Insignificant results were also obtained in analyses using upper quartiles of YKL-40 as dichotomous variable (Table III).

Discussion

To our knowledge this is the first study of the inflammation marker YKL-40 in patients with CHF. We report approximately 77% higher YKL-40 levels in patients with CHF compared with age-matched healthy control subjects. We found no difference the genders in between, and the overweight of males in the CHF group could not explain the results for reasons previously discussed in detail (8). Approximately 25% of the patients with CHF were also diagnosed with DM, but we found no differences in YKL-40 levels between CHF patients with or without DM. Despite a known association between YKL-40 and diabetes (19–21), our data did not support, that increased YKL-40 levels in patients with CHF were due to comorbidity with DM.

YKL-40 levels showed only weak correlations with NT-proBNP and UACR in univariate analyses, but were not influenced by other clinical characteristics in multivariate analyses. Similar, medication did not have an effect on YKL-40 levels. This suggests that YKL-40 might play an independent role in the inflammatory state and that YKL-40 levels increase through other mechanisms or signalling cascades than e.g. hsCRP. This is in accordance with previous studies, where no or only weak correlations between YKL-40 and hsCRP have been found (19,22,23). In contrast, a single study has documented a correlation between YKL-40 and hsCRP but as in the studies where no correlation between the markers was found, no other biomarkers were investigated concomitantly in that study (14).

Previous studies indicate that YKL-40 plays a role in endothelial dysfunction in relation to cell migration, reorganization and tissue remodelling during atherogenesis by promoting chemotaxis, cell attachment, spreading and migration of vascular endothelial cells (9,10,24). High YKL-40 protein expression is found in vivo in both macrophages and vascular smooth muscle cells in the atherosclerotic plaque (11). In the last few years, several clinical studies have described elevated YKL-40 levels in several cardiovascular conditions as well as described an association between YKL-40 and mortality. Studies show, that elevated YKL-levels are independently associated with the presence (12–14) and the extent (12) of CAD as assessed by echocardiography. In patients suffering MI even higher YKL-levels are documented (13,15) and elevated YKL-40 levels have also been found to be associated with all-cause as well as cardiovascular mortality not only in patients with stable CAD (13) but also in the general population above 50 years of age without known DM or CHD (22). Lately, elevated YKL-40 levels have also been documented in individuals with atrial fibrillation (23). All together this suggests YKL-40 as a possible screening modality/diagnostic marker for progressing coronary atherosclerosis in its early phase before development of angina which could be of relevance in terms of prophylactic treatment or intervention.

In patients with Type 1 Diabetes, increasing YKL-40 levels are seen with increasing levels of albuminuria and the apparant role of YKL-40 in endothelial dysfunction at a very early stage suggests that YKL-40 might be associated with other markers of endothelial dysfunction such as E-sel, VCAM and ICAM-1. However, our data could not support this hypothesis, since we did not find a significant association with neither E-sel nor vWf. VWf and E-sel have previously been examined in this population, but no significant increase of these markers in patients with CHF was found except for CHF patients with co-existing DM (25). In contrast, we found elevated YKL-40 levels in CHF patients, without further increased levels in CHF patients also having diabetes. The finding of elevated YKL-40 levels but not of E-sel and vWf suggests that YKL-40 might have an independent role in the endothelial dysfunction observed in patients with CHF perhaps even preceding or modulating the role of E-selectin and vWf.

Despite the substantial evidence of the association between elevated YKL-40 levels and CAD, we did not find higher YKL-40 levels in patients with CHF due to IHD when compared to patients with non-ischemic cardiomyopathy. Nor were levels higher in patients with more severe CHF as assessed by NYHA classification. Hence, once CHF has developed YKL-40 is elevated irrespectively of the under-lyimg cause and severity. Whether or not YKL-40 plays a pathogenetic role in the development of CHF remains speculative. Follow up data further revealed that YKL-40 was not significantly associated with neither overall mortality nor any specific cardiovascular outcome. However, the moderate number of participants and events make conclusions difficult.

CHF is a complex disease modality often complicated by comorbidities as e.g. renal dysfunction or COPD among several others (26). Unfortunately, we do not have baseline description of concomitant diseases such as kidney malfunction, COPD or other diseases who either predispose to CHF development or interrelate with the CHF progression and prognosis (26). However, we did not find any association between eGFR and YKL-40 levels in univariate analyses and in multivariate analyses we only found a trend towards an association with UACR. It could be, that elevated YKL-40 levels in CHF patients are explained by the presence of concomitant diseases also characerized by elevated YKL-40 levels (8,27), but it is not excluded that low-grade inflammation, endothelial dysfunction or matrix remodelling could be other explanations for the elevated YKL-40 levels in CHF. YKL-40 has previously been found to counteract the mediative effects of IL-1 and TNFα by reducing the catabolic effect of IL-8 and metalloproteinases (MMPs) (28) indicating a protective role in remodelling processes similar to pathological changes in the myocardial and vascular matrices in CHF patients. To assess this possibility further, future studies should evaluate a possible relation between MMPs and YKL-40 in CHF populations.

The main limitations of the present study are the moderate numbers of patients and events and the lack of information on concomitant diseases. Furthermore, the study has only evaluated YKL-40 levels in relatively stable CHF patients with moderate symptoms, and results cannot be extrapolated to patients with acute heart failure. However, based on the findings there is little reason to expect that YKL-40 would provide more prognostic information in patients with more advanced CHF. Rather, it seems logical to study the marker more closely in patients earlier in the disease process. Therefore, future prospective studies examining the development of CHF and processes preceding the disease might be explanatory in terms of markers of the earliest pathogenetic events such as YKL-40.

In summary, this is the first study of YKL-40 in patients with CHF. Independently of co-existing DM and levels of cardiovascular markers as well as markers of low-grade inflammation and endothelial dysfunction, patients with CHF have elevated YKL-40 levels. However, YKL-40 levels were not higher in patients with CHF caused by myocardial ischemia and YKL-40 levels were not predictive of overall mortality or non-fatal cardiovascular outcomes. YKL-40 showed no association with E-selectin, vWf or other markers of endothelial dysfunction, and an independent role of YKL-40 in low-grade inflammation and matrix remodelling in the pathogenesis of CHF can not be excluded. However, missing information of the likely presence of concomitant diseases also characterized by elevated YKL-40 levels e.g. COPD could be another reason for elevated YKL-40 levels in these CHF patients. Further studies are needed.

Acknowledgements

The study was supported by grants from The Research Foundation of Herlev Hospital; The Research Foundation of Copenhagen County; The Foundation of Advancement of Medical Science of The A.P. Moeller and wife Chastine McKinney Moeller Foundation. We wish to thank Ulla Kjaerulff-Hansen, Tonni Loeve Hansen and Debbie Nadelmann, Endocrine Research lab. 54O4, Herlev Hospital, University of Copenhagen, Denmark, for skilful laboratory assistance. There is no conflict of interest to be declared.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.