Vascular endothelial function predicts mortality risk in patients with advanced ischaemic chronic heart failure†

Study design and population

Eighty-two consecutive patients with ischaemic cardiomyopathy were recruited prospectively from the Heart Failure Day Care Service at the Heart Institute, Sheba Medical Center, Tel Hashomer, Israel. Study inclusion criteria were: patients aged >20 years, with advanced NYHA class IV ischaemic CHF (defined as patients with chronic severe coronary artery disease confirmed by history of prior myocardial infarction(s), coronary angiography with at least single-vessel stenosis of >70%, and symptoms of heart failure caused by diffuse fibrosis or multiple infarctions leading to left ventricular dysfunction); left ventricular ejection fraction (LVEF) of ≤30% measured within the past 6 months; hospitalization for CHF with intravenous administration of a diuretic (minimum 40 mg furosemide or the equivalent) within the past 12 months; patients on standard CHF therapy which had to include an angiotensin converting-enzyme inhibitor (ACE-I) or an angiotensin receptor blocker (unless contraindicated or not tolerated by the patient); patients with no changes in their cardiac medications (indicated for CHF) throughout the 2 weeks prior to treatment, and signed informed consent. In addition, all patients were required to have been taking beta-blockers for at least 3 months.

Exclusion criteria were: severe valvular heart disease; alcoholic cardiomyopathy; hypertrophic cardiomyopathy; restrictive and infiltrative cardiomyopathy; arrhythmogenic right ventricular cardiomyopathy; idiopathic dilated cardiomyopathy; acute myocardial infarction within 3 months, coronary artery bypass surgery or percutaneous coronary intervention within 3 months; need for chronic intermittent inotropic therapy; malignancy; active myocarditis; early postpartum cardiomyopathy; HIV infection or immunodeficiency state; chronic viral infection; acute systemic infection requiring antibiotics; unstable angina; systemic hypertension >180/110 mmHg; permanent pacemaker; atrial fibrillation; peripheral vascular disease; deep vein thrombosis; history of drug or alcohol abuse; chronic liver disease; or refusal to sign the informed consent form. The hospital review board approved the study, and all participants gave written informed consent.

The study was only initiated when patients were optimally treated for CHF and had been in a steady state for at least 3 months. Patients were instructed to continue taking their regular medications and maintain their usual diet throughout the study. Patients underwent monthly physical examinations by the same cardiologist for a mean of 14 ± 2 months, and changes in concomitant medications during the study were recorded.

On the first day of the study after an overnight fast and discontinuation of all medications for ≥12 h, patients underwent a physical examination, assessment of brachial artery endothelial function, NYHA functional class, a 6 min walk test, and had blood drawn for the measurement of fasting blood glucose, lipids, blood cell count, electrolytes, and serum high-sensitivity C-reactive protein. Serum total cholesterol, high-density lipoprotein cholesterol (HDL-C), triglycerides, and plasma glucose concentrations were measured using standard enzymatic methods. Low-density lipoprotein cholesterol (LDL-C) was calculated by the Friedewald formula. Serum hs-C-reactive protein was analysed by the sandwich immunoassay method.

Vascular function protocol

Endothelial function in the form of endothelium-dependent flow-mediated dilation (FMD) in the brachial artery was measured as previously described.10–12 Briefly, FMD was assessed by a single ultrasonographer blinded to the clinical status of the patient. The test was performed in the subject's right arm while in a recumbent position in a temperature-controlled room (22°C) after a 10 min equilibration period. Using a 15-6 MHz linear array (15-6L HP) ultrasound (HP SONOS 5500 cv system, Agilent Technologies, Inc., Andover, MA, USA), the brachial artery was longitudinally imaged approximately 5 cm proximal to the antecubital crease. An electrocardiogram (ECG) was monitored continuously and blood pressure was taken in the left arm each minute throughout the study.

Study phases

Endothelium-dependent FMD: following a 2 min baseline period, a longitudinal image of 3 cm of vessel without colour flow was obtained and frozen for 5 s. The image was then unfrozen and switched to a pulse wave Doppler for 5 s at a sweep speed of 50 mm/s. A pneumatic tourniquet, placed around the upper arm proximal to the target artery (upper-arm occlusion), was inflated after the baseline phase to 50 mmHg above the subject's systolic blood pressure (or until no blood flow was observed in the brachial artery by Doppler probe), and held for 5 min. Upper arm occlusion was followed by a hyperaemic state which is mainly dependent upon local metabolic changes in favour of vasodilating substances, and only partially endothelially mediated. Increased flow was subsequently induced by sudden cuff deflation. A continuous scan was performed at deflation, and at 60 and 90 s, with frozen and Doppler measurements recorded as before.

Nitroglycerin (NTG)-induced (non-endothelium-dependent) vasodilatation: a second 2 min baseline resting scan was recorded to confirm vessel recovery 13 min after cuff deflation. Scanning was performed continuously for 5 min following administration of a sublingual NTG tablet (Nitrostat, 0.4 mg, Park-Davis).

Data analysis

Ultrasound images were recorded on an S-VHS videotape with a SLV-RS7 videocassette recorder (SONY). Brachial artery diameter was measured from the anterior to the posterior interface between the media and adventitia (‘m line’) at a fixed distance. The mean diameter was calculated from four cardiac cycles synchronized with the R-wave peaks on the ECG. All measurements were calculated at end-diastole to avoid possible errors resulting from variable arterial compliance. The internal diameter was calculated with PC Prosound software (USC, CA, USA) using a Horita Data Translation Image Processing board (DT2862-60Hz).11 The diameter percent change caused by endothelium-dependent FMD (%FMD), and endothelium-independent percent change from baseline in NTG-mediated vasodilation (%NTG) were expressed as the percent change relative to that at the initial resting scan. %FMD and %NTG were computed from the formula [(maximum diameter − baseline diameter)/baseline diameter × 100]. The intra-observer correlation coefficients for baseline and deflation diameters were 0.99. The absolute error between measurements ranged from 0 to 0.12 mm (for brachial artery diameter) and 0.02 to 2.98% (for FMD).

Echocardiographic assessment

All patients underwent two-dimensional echocardiographic assessment on the first day after endothelial function assessment by the same senior cardiologist, who was blinded to the patients' clinical status and endothelial function results. Echocardiographic data included left ventricular dimensions and volumes, left atrial dimension and volume (biplane modified Simpson's), LVEF (biplane modified Simpson's), global and regional wall motion, and valve anatomy and function.

Six-minute walk test

The 6 min walk test was conducted as described by Guyatt et al.13 A distance of 100 feet was marked in a hospital corridor, and patients were instructed to walk from end to end at their own pace, attempting to cover as great a distance as possible within 6 min. The test was supervised by a technician, who provided verbal encouragement at 30 s intervals. Patients were permitted to stop and rest, and were instructed to continue walking as soon as they felt they were able. Symptoms experienced by the patients were recorded. Distance covered was expressed in metres.

Long-term clinical follow-up

Over a mean follow-up period of 14 ± 2 months, patients underwent a monthly physical examination which was always performed by the same cardiologist. In addition, patients were contacted regularly by telephone to evaluate the incidence of combined adverse cardiovascular events, which included all-cause mortality, non-fatal myocardial infarction, and hospitalization for CHF exacerbation. The decision that a primary adverse cardiovascular event had occurred was only made after reviewing the patient's medical records, which were evaluated by cardiologists in the event of death, hospitalization, and/or angina pectoris. No patient was lost to follow-up.

Statistical analysis

Group data are expressed as mean ± SD. Survival data were analysed by the Kaplan–Meier method, log-rank tests for univariate analysis, and Cox proportional-hazards models for multivariate analysis. Cox regression analysis was performed to examine the potential relationship between variables and events during follow-up. Multivariate analysis using Cox regression techniques was performed to examine potential interactions among the entered covariates. Variables, generally accepted as prognostic values in heart failure, were incorporated into the Cox model and included age, gender, FMD, NTG, LVEF, hypertension, hypercholesterolaemia, smoking status, diabetes mellitus, hs-C-reactive protein, haemoglobin, serum sodium, serum creatinine, the presence of peripheral vascular disease, chronic obstructive pulmonary disease, or atrial fibrillation. The multivariate model also included baseline brachial artery diameter. The proportional hazard assumption of the model was assessed by inspection of the log time–log hazard plot for all covariates. In the survival analysis, FMD measurements were performed on day 0. Baseline characteristics were compared between the two groups using a two-sided Student's t-test for continuous variables, and Fisher's exact test for categorical variables. P < 0.05 was considered statistically significant. All statistical analyses were performed using SPSS for Windows 12.0 (SPSS, Inc.).

Endothelial function in chronic heart failure subjects

The mean FMD of all subjects was 5.1 ± 0.4% (median 4.6%) and the mean NTG was 8.3 ± 4.7% (median 8.2%). The CHF patients were divided into two groups: those with an FMD equal to or below the median FMD of 4.6% (n = 41) and those with an FMD above the median (n = 41). Both groups were comparable regarding CAD risk factors, LVEF, lipid panel, fasting glucose, homocysteine, hs-C-reactive protein, haemoglobin, serum creatinine and renal function tests, resting blood pressure and heart rate, 6 min walk distance, and concomitant medications (Table 1). However, baseline brachial artery diameter was significantly larger in patients with an FMD lesser than or equal to the median of 4.6%, compared with those with an FMD above the median (P < 0.01). This phenomenon is indicative of advanced atherosclerosis and vessel remodelling, as well as vasodilatation (possibly drug-induced), a factor which further reduces the possibility of an increase in diameter, since the vessel has already reached its maximal dilatory capacity. In addition, %NTG was significantly higher in CHF patients with an FMD above the median, compared with those with an FMD equal to or below the median (Table 1).

Long-term adverse cardiovascular endpoints

Thirty patients (36.5%) from the total study cohort had composite adverse cardiovascular events (including all-cause mortality, non-fatal myocardial infarction, and hospitalization for CHF exacerbation) throughout the follow-up (Table 2). Hospitalizations for CHF exacerbation in patients with an FMD lesser than or equal to the median were twice those of CHF patients with an FMD above the median (39.0 vs. 19.5%, P < 0.05). Furthermore and significantly, more patients with advanced NYHA class IV CHF and an FMD equal to or below the median died (n = 5) during follow-up (Table 2), compared with no deaths in patients with an FMD above the median (P < 0.03 by log-rank test). In Cox proportional hazard models adjusted for other known prognostic factors, decreased FMD remained significantly associated with increased mortality risk when considered as a continuous variable [adjusted hazard ratio (HR) estimate for a 1% decrease in FMD = 1.20; 95% confidence interval (CI) 1.01–1.69; P < 0.03] or dichotomized at the median value (adjusted HR estimate for below median FMD = 2.04; 95% CI 1.09–5.1, P < 0.03). Despite a significant correlation between FMD and baseline brachial artery diameter, the latter was not a prognostic predictor in our study cohort.

Study limitations

The observational nature of this study prohibits drawing conclusions about a causal link between endothelial dysfunction and mortality, since unmeasured confounders may have contributed to the observed associations. We studied a small number of advanced CHF patients on a relatively intensive regimen of ACE-I, statins, spironolactone, diuretics, and cardiac glycosides. These drugs could only be discontinued for 12–24 h prior to the study measurements, and therefore they may have had an effect on endothelial function, thereby possibly influencing our findings. Nevertheless, based on our positive findings, further studies with larger numbers of advanced NYHA class IV CHF patients are indicated. The sample size and relatively small number of events limited the number of variables in our regression models, providing sufficient statistical power to detect only extensive risk differences. Furthermore, our analysis did not include previously identified prognostic markers in heart failure, such as B-type natriuretic peptide, endothelin, or various cellular adhesion molecules, such as E-selectin and the von Willebrand factor.30 E-selectin is exclusively expressed on endothelial cells and is a specific biomarker of endothelial activation. Finally, it should be noted that changes in therapy over time could be a possible confounding factor in patients with advanced CHF.