Endothelial dysfunction in patients with chronic heart failure is independently associated with increased incidence of hospitalization , cardiac transplantation , or death

Background Endothelial dysfunction of coronary and peripheral arteries has been demonstrated in patients with chronic heart failure (CHF) and appears to be associated with functional implications. However, it is unknown whether endothelial dysfunction in CHF is independently associated with impaired outcome or progression of the disease. Methods and results We assessed the follow-up of 67 consecutive patients with CHF [New York Heart Association (NYHA) functional class II–III] in which flow-dependent, endothelium-mediated vasodilation (FDD) of the radial artery was assessed by high resolution ultrasound. The primary endpoint was defined by cardiac death, hospitalization due to worsening of heart failure (NYHA class IV, pulmonary oedema), or heart transplantation. Cox regression analysis was used to determine whether FDD was associated with these heart failure-related events. During a median follow-up of 45.7 months 24 patients had an event: 18 patients were hospitalized due to worsening of heart failure or heart transplantation, six patients died for cardiac reasons. Cox regression analysis demonstrated that FDD (P , 0.01), diabetes mellitus (P , 0.01), and ejection fraction (P , 0.01) were independent predictive factors for the occurrence of the primary endpoint. The Kaplan–Meier survival curve revealed a significantly better clinical outcome in patients with FDD above the median (6.2%) compared with those with FDD below the median (P , 0.013). Conclusion These observations suggest that endothelium-mediated vasodilation represents an independent predictor of cardiac death and hospitalization in patients with CHF, consistent with the notion that endothelium-derived nitric oxide may play a protective role in heart failure.

Background Endothelial dysfunction of coronary and peripheral arteries has been demonstrated in patients with chronic heart failure (CHF) and appears to be associated with functional implications.However, it is unknown whether endothelial dysfunction in CHF is independently associated with impaired outcome or progression of the disease.

Methods and results
We assessed the follow-up of 67 consecutive patients with CHF [New York Heart Association (NYHA) functional class II-III] in which flow-dependent, endothelium-mediated vasodilation (FDD) of the radial artery was assessed by high resolution ultrasound.The primary endpoint was defined by cardiac death, hospitalization due to worsening of heart failure (NYHA class IV, pulmonary oedema), or heart transplantation.Cox regression analysis was used to determine whether FDD was associated with these heart failure-related events.During a median follow-up of 45.7 months 24 patients had an event: 18 patients were hospitalized due to worsening of heart failure or heart transplantation, six patients died for cardiac reasons.Cox regression analysis demonstrated that FDD (P , 0.01), diabetes mellitus (P , 0.01), and ejection fraction (P , 0.01) were independent predictive factors for the occurrence of the primary endpoint.The Kaplan-Meier survival curve revealed a significantly better clinical outcome in patients with FDD above the median (6.2%) compared with those with

Introduction
2][3][4] An important consequence of endothelial dysfunction is the inability of a vessel to dilate in response to physiological stimuli, such as increases in blood flow, reflecting impaired flowdependent, endothelium-mediated vasodilation (FDD).6][7] Short-term and long-term administration of the anti-oxidant vitamin C improves FDD in patients with CHF as a result of increased nitric oxide (NO) bioavailability, suggesting that endothelial dysfunction is, at least in part, attributable to accelerated degradation of NO by oxygen radicals.
Accumulating evidence suggests that endothelial dysfunction contributes to exercise intolerance, impaired myocardial perfusion, and left ventricular remodelling in CHF.Since NO is thought to represent an antiarteriosclerotic molecule, the prognostic impact of endothelial dysfunction on vascular events has been previously studied in conditions known to represent risk factors for arteriosclerotic disease, i.e. coronary artery disease.[10] However, beyond protection from arteriosclerosis NO keeps peripheral resistance vessels in a dilated state providing afterload reduction in conditions such as CHF.Indeed, there is evidence to suggest that impaired availability of endothelial-derived NO contributes to this abnormal vasodilator response in chronic heart failure, both in the coronary and peripheral circulation. 11,12NO is required for vascular endothelial growth factor (VEGF)-induced angiogenesis 13 and its lack may be associated with reduced capillary density in cardiac muscle, a condition which has been shown to result in cardiomyopathy in animal models. 14These effects of NO raise the possibility that impaired endothelial bioavailability of NO in CHF may be associated with progression of the disease and increased mortality.If so, patients with CHF and more severe endothelial dysfunction would have a higher incidence of hospitalization for decompensation of CHF, cardiac transplantation, or cardiac death than those with relatively preserved endotheliumdependent relaxation.Accordingly, the aim of our study was to elucidate the prognostic impact of endothelial dysfunction on hospitalization for worsening of CHF, cardiac transplantation, or death.

Methods
The patient population represents 67 consecutive patients with chronic heart failure undergoing assessment of peripheral endothelial function at our institution between 1997 and 2000.The study protocol was approved by the local Ethics Committee; written informed consent was obtained from each patient.
All patients presented with CHF classified New York Heart Association (NYHA) functional class II-III.The following clinical parameters and risk factors associated with endothelial dysfunction and prognosis in heart failure were determined: hypertension, hypercholesterolaemia, smoking status, diabetes mellitus, C-reactive protein (CRP), haemoglobin, the presence of peripheral vascular disease, chronic obstructive pulmonary disease, atrial fibrillation, serum creatinine, and ejection fraction.
Hypertension was defined as blood pressure .140/90mmHg for more than one year that required therapy with antihypertensive medication.Diabetes mellitus was defined according to the definition of the American Diabetes Association. 15All of our patients with diabetes were treated with medical therapy (oral or insulin).Hypercholesterolaemia was defined as LDL .160mg/dL or if the subject has been treated with lipidlowering medication or dietary modification.Smoking was defined as a history of smoking for more than two pack-years.All patients were asked to refrain from smoking .4h before examination.Peripheral vascular disease was defined as a history of intermittent claudication or absent peripheral pulses.The LV function was analysed by echocardiography and/or LV angiography.To distinguish between patients with reduced versus preserved systolic function an ejection fraction (LVEF) of 45% (n ¼ 30; ,45%) was used, similar to cut-off points used in clinical trials.The CRP cut-off point was 5 mg/L.
6][7] In brief, recordings of arterial diameters were obtained with a 10-MHz transducer positioned perpendicular to the vessel.Stereo Doppler guidance was used to ensure a correct vertical position of the probe over the artery.Each diameter measurement represents data digitized over a 4 s period.Wrist arterial occlusion was performed by inflating an occlusion cuff to 40 mmHg above systolic blood pressure for 8 min.Flow-dependent vasodilation (FDD) was recorded as percentage vasodilation comparing the radial artery diameter at baseline and the maximal diameter after reactive hyperaemia.Finally, all subjects received an intra-arterial infusion of sodium nitroprusside (SNP) (10 mg/min over 5 min) to assess endothelium-independent vasodilatory capacity.

Long-term follow-up
Clinical long-term follow-up was done by telephone or using a questionnaire that was sent to the patient or to the primary physician.Follow-up was performed once a year.The following clinical events were reported as endpoints: cardiac death, hospitalization due to worsening of heart failure (NYHA class IV, pulmonary oedema), or necessity for heart transplantation.The first event that occurred was used for the analysis.If a primary endpoint was reached, information regarding potential cardiovascular events was validated by review of source data, for example a hospital discharge report.The decision that a patient had reached a primary endpoint was made only after reviewing the medical records.All data are shown as mean + SEM or percentage.

Statistical analysis
Cox regression analysis was performed to examine the potential relationship between variables and events during the follow-up period.Multivariate analysis using Cox regression techniques was performed to examine potential interactions among the entered covariates.Variables generally accepted as being of prognostic value in heart failure were included in the Cox model including age, gender, FDD, endothelium-independent vasodilation after infusion of SNP, NYHA functional class, ejection fraction, hypertension, hypercholesterolaemia, smoking status, diabetes mellitus, CRP, haemoglobin, serum sodium, serum creatinine, the presence of peripheral vascular disease, chronic obstructive pulmonary disease, or atrial fibrillation.
The proportional hazard assumption of the model was assessed by inspection of the log time-log hazard plot for all covariates.Statistical comparisons were performed by Kaplan-Meier survival curves to analyse the cumulative event rate linked to FDD.The Kaplan-Meier plots were constructed to illustrate the results.For this illustration, the patient population was divided into two groups.One group had FDD lower than the median (6.2%), the other group had FDD greater than the median.Day 0 in the survival analysis was the day the FDD measurement was done.Comparisons of the baseline characteristic variables between the two groups (FDD lower or greater than the median) were made for continuous variables with a two-sided Student's t-test and for categorical variables with Fisher's exact test.P , 0.05 was considered statistically significant.All statistical analysis was performed using SPSS for Windows 12.0 (SPSS Inc).

Results
We included 67 patients in our study (12 female, 55 male).The mean age was 60.7 + 11.1 years.The baseline features are outlined in Table 1.In Table 2, the baseline characteristics are depicted separately for each group.Both groups are comparable at baseline in all features with the exception of the LDL cholesterol (P ¼ 0.08).The median period of follow-up was 45.7 months (range 1 to 69, interquartile range 27.4).Follow-up was completed in all patients.Patients were treated with ACE-inhibitiors (83.6%), b-blockers (44.8%), diuretics (58.2%), spironolactone (10.4%), and digoxin (37.3%).There were no significant differences in the medical therapy between patients with FDD above and below the median.Twenty-four patients had a classified event.Fifteen patients were hospitalized due to worsening of heart failure, three patients underwent heart transplantation, and six patients died from cardiac causes.Events occurred at a median 16.5 months (range 1-61, interquartile range 38.75).There were no non-cardiac deaths in this population.Cox regression analysis revealed that FDD {P , 0.01, hazard ratio [Exp(B)] 0.665, standard error (SE) 0.182} was an independent predictive factor for the occurrence of events.In contrast, endotheliumindependent vasodilation after infusion of SNP was not associated with clinical events (P ¼ 0.37).Other independent predictive factors for the occurrence of events in our patient population were, as expected, diabetes mellitus [P , 0.01, Exp(B) 0.055, SE 0.946] and ejection fraction [P , 0.01, Exp(B) 0.054, SE 0.894].Age (P ¼ 0.47), gender (P ¼ 0.14), serum creatinine (P ¼ 0.21), CRP (P ¼ 0.55), haemoglobin (P ¼ 0.88), serum sodium (P ¼ 0.10), hypertension (P ¼ 0.06), NYHA functional class (P ¼ 0.09), hypercholesterolaemia (P ¼ 0.56), chronic obstructive pulmonary disease (P ¼ 0.18), peripheral arterial disease (P ¼ 0.14) and atrial fibrillation (P ¼ 0.86) were not significantly associated with events, probably due in part to the sample size of this population.The Kaplan-Meier survival curve depicting the cumulative event rate depending on FDD shows a significantly better outcome in patients with FDD above the median (6.2%) compared with those with FDD below the median (P ¼ 0.013) (Figure 1).

Discussion
The present study shows that endothelial dysfunction represents an independent predictor of clinical outcome in patients with heart failure, i.e. it is associated with higher incidences of hospitalization for decompensation of heart failure, cardiac transplantation, or cardiac death.As these clinical endpoints are thought to reflect progression of the disease and impaired prognosis, our observations are consistent with the notion that endothelial dysfunction is associated with accelerated progression of heart failure.In contrast, endothelium-independent vasodilator responses with SNP were not predictive of cardiovascular events, arguing against a non-specific observation.
It should be noted that impaired FDD in heart failure is a generalized abnormality occurring in both the peripheral and coronary circulation and appears to be due to reduced bioavailability of NO. 6 In this respect it is interesting to note that a common polymorphism of endothelial NO-synthase (eNOS) shows increased vasoconstrictive response to phenylephrine and may be associated with decreased NO-synthase (NOS) activity.A recent report showed that this polymorphism, the Asp298 variant of endothelial NOS, is associated with poorer event-free survival in patients with heart failure. 16The impaired survival in patients with this common eNOS polymorphism and potentially reduced eNOS bioavailability is consistent with our observation that reduced FDD (reflecting impaired eNOS bioavailability) is associated with clinical endpoints reflecting progression of heart failure.In contrast, Jones et al. 17 have provided compelling evidence for a beneficial effect of eNOS-derived NO in a mouse model of CHF following over-expression of eNOS, supporting the idea that physiological levels of NO exert beneficial effects in the setting of heart failure.
There are several potential mechanisms by which endothelial dysfunction may contribute to disease progression in patients with CHF.Since endothelial vasodilator function is involved in control of tissue perfusion, impaired exercise-induced release of NO may contribute to reduced exercise capacity in chronic heart failure and more severe symptoms.Recent experimental studies have demonstrated that impaired endothelium-dependent vasodilation after NOS inhibition was associated with reduced exercise capacity. 18Conversely, the improvement of endothelium-dependent vasodilation in patients with CHF after exercise training was closely correlated with the achieved increase in peak oxygen uptake, suggesting that improved endothelial function contributed to increased exercise capacity after physical training in patients with CHF. 19educed endothelium-dependent, NO-mediated vasodilation may contribute to increased peripheral vasoconstriction and enhanced afterload.Moreover, endothelial dysfunction may contribute to myocardial perfusion abnormalities in patients with CHF and further augment myocardial damage.In this respect there is evidence that inhibition of NOS results in impaired myocardial perfusion during adenosine-induced hyperaemia, suggesting that endothelium-derived NO plays a significant role in the regulation of myocardial perfusion. 20In patients with dilated cardiomyopathy, impaired myocardial blood flow was observed in response to pacing tachycardia or dipyridamol infusion. 21Furthermore, the degree of myocardial blood flow reduction (in response to dipyridamol) is an independent prognostic predictor in patients with idiopathic left ventricular dysfunction. 22In addition, recent experimental studies suggest that endothelial NOS-derived NO production lowers myocardial oxygen consumption 23 and limits left ventricular remodelling. 24Ventricular dysfunction, remodelling, and mortality are greater in eNOS-deficient mice compared with wild-type mice after myocardial infarction, even after correction for differences in blood pressure. 24Our observations in patients with heart failure are consistent with these experimental findings.
The findings of our present study support the notion that endothelial dysfunction may represent a novel therapeutic target in patients with CHF.It is conceivable that novel therapeutic agents, i.e. statins, may exert beneficial effects in patients with CHF by improving endothelial function.In fact, we have recently shown that statins improve survival post-infarction by an eNOS-dependent mechanism. 25Moreover, there is experimental and clinical evidence that ACE-inhibitors improve endothelial dysfunction in heart failure 7 and it is conceivable that this mechanism of action contributes to the long-term effect of ACE-inhibitors in this condition. 11hile the sample size of the present study is limited, a large number of clinical events occurred, which made possible an analysis of endothelial dysfunction as an independent predictor.Nevertheless, other established predictors of mortality in CHF did not reach significance in the present series, possibly due to the sample size.Notably, our analysis did not include previously identified prognostic markers in heart failure such as B-type natriuretic peptide (BNP), or endothelin.
In summary, our study suggests that endothelial dysfunction in patients with CHF is independently associated with an increase in clinical events which are thought to reflect progression of heart failure.Improvement of eNO availability may represent an attractive objective for novel therapeutic interventions.

Figure 1
Figure1Relationship between endothelium-dependent vasodilation and cardiovascular events in the follow-up of patients with chronic heart failure.Kaplan-Meier analyses demonstrating proportion of patients surviving free from heart failure events during long-term follow-up.Study cohort is divided into those with flow-dependent vasodilation (FDD) .median(6.2%) and , median.(A ) shows patients with FDD above median, (B ) shows patients with FDD below median.

Table 1
Baseline characteristics of study cohort (n ¼ 67) Figures in parentheses are percentages.

Table 2
Baseline characteristics separated for groups