Chronic heart failure (CHF) has high incidence and prevalence, which are rising with ageing populations. Accurate classification of its severity has become increasingly important when deciding on drug therapies and implantable devices that have been shown to reduce morbidity and mortality, including renin–angiotensin–aldosterone1,2 and adrenergic3 system blockade and use of resynchronization and implantable defibrillation systems.4 Despite these advances, palliation and treatment of symptoms are often difficult.

The most widely used classification of CHF symptom severity is the New York Heart Association (NYHA) system, consisting of four functional classes (I–IV), which has been shown to be valuable in predicting clinical events in CHF.5 However, mild symptoms are not synonymous with mild cardiac dysfunction, and there may be little correlation between severity of symptoms and degree of ventricular dysfunction.6

Clinical evaluation of patients with signs and symptoms suggestive of CHF can be difficult and has poor reproducibility; studies assessing patients referred for echocardiography show a high rate of false positives.7 Population-based studies have also demonstrated that up to 60% of patients with left ventricular (LV) dysfunction are asymptomatic, making diagnosis more difficult.8 It is likely that these patients would also benefit from various therapies if identified and treated earlier.9

Ejection fraction (EF) is widely used as a prognostic variable in CHF patients, but the disease's course in those with systolic dysfunction, particularly with EF between 20% and 30%, is highly variable and some studies have stressed the role of hemodynamic changes such as raised LV filling pressures in determining prognosis.10,11

The aim of this study was to assess the correlation between clinical, laboratory and echocardiographic variables and symptom severity in patients with CHF, particularly the ability of the biomarkers NT-proBNP and copeptin to diagnose high functional class. We also tested the hypothesis that clinical presentation correlates more strongly with other laboratory and echocardiographic variables than with EF in patients with known LV dysfunction. Mortality was assessed at 18 months.

This was a comparative cohort study with prospective assessment of the prognosis of 37 patients with stable symptomatic CHF (NYHA class II–IV) with LV systolic dysfunction. Patients with primary valve disease, a history of chronic renal failure or serum creatinine >3mg/dl were excluded. Outpatients (n=14) were included, as well as those admitted for decompensated CHF as long as they returned to their usual functional class once discharged after clinical stabilization (n=23).

Patients were selected in our center's echocardiography laboratory following an echocardiogram showing LVEF ≤40% together with signs and symptoms of CHF. At this time, the images were recorded in digital format in accordance with the echocardiographic protocol, a clinical history was taken by interview and peripheral venous blood samples were collected. Blood pressure and heart rate were measured after five minutes in decubitus.

Clinical follow-up was the responsibility of the attending physician, the investigators being blinded to the laboratory results and interventions. The principal end-point was all-cause mortality at 18 months (548 days). Survival was assessed by telephone interview and consultation of hospital records at the end of the predefined follow-up period.

The study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of our institution. All participants gave their informed consent.

The results of our study show that mortality is higher in patients with CHF and LV dysfunction in the more advanced symptomatic stages, when they present higher levels of leukocytes, creatinine, TSH, copeptin and NT-proBNP, and E/E′ ratio, E-septum distance, PASP and IVC diameters are significantly greater. The biomarkers copeptin and NT-proBNP and the echocardiographic parameters PASP and IVC inspiratory diameter showed the best performance for diagnosis of NYHA functional class >II, but in multivariate analysis only PASP and serum creatinine were independent predictors of NYHA class >II.

Our results are in agreement with previous studies showing a correlation between NYHA functional class and occurrence of clinical events including need for hospitalization and mortality.5,16 A consistent association has been found between NYHA class and prognosis in patients with systolic heart failure17–19 and also in those with diastolic CHF.20 However, there is considerable interobserver variability in categorizing CHF patients by NYHA class.21 There are also discrepancies between the functional class attributed by physicians and that reported by patients.22 The situation is further complicated by the fact that patients’ self-assessed functional status is a prognostic predictor of hospitalizations, quality of life and death.23 Thus, despite its prognostic value, the NYHA classification system may not be applied in CHF in the same way by all health providers; the clinical assessment of a tertiary heart failure unit like ours probably cannot be applied at all levels of medical care. In an attempt to standardize the criteria, laboratory markers have been suggested as a possible way to achieve greater objectivity in clinical evaluation.24

Our study demonstrated a strong correlation between serum levels of the biomarkers NT-proBNP and copeptin and functional class, showing a good ability to distinguish between class II and >II.

BNP and its precursor NT-proBNP are widely used in cardiovascular risk assessment, as reflected in current guidelines.25 High levels of these peptides are associated with advanced age, renal failure, arrhythmias, systolic and diastolic dysfunction, and mortality.26 They reflect individual patient risk and are thus essential tools in stratifying risk.27–29

To our knowledge, this is the first study to assess copeptin in a Portuguese population with CHF. Copeptin is a fragment of pre-provasopressin that is secreted in equimolar quantities to vasopressin, a hormone that, unlike natriuretic peptides, has antidiuretic and vasoconstrictor properties.30,31 There is strong evidence of a correlation between vasopressin and CHF severity, which probably extends to prognosis.32–34 However, it is difficult to measure since it is an unstable peptide, binds strongly to platelets and is rapidly eliminated.35 The main advantages of assessing copeptin are its stability and ease and accuracy of measurement in plasma or serum. Although copeptin correlates with serum levels of BNP and NT-proBNP, its prognostic value appears to be superior in population-based studies.24 The value of copeptin is enhanced by the fact that vasopressin blockade is being investigated as a therapeutic target in CHF. Clinical trials of V2 receptor antagonists have shown improvement in congestive symptoms due to the drugs’ diuretic properties but benefits in survival have yet to be demonstrated.36,37

TSH was significantly higher in more symptomatic patients, and there is a known relation between elevated TSH and prevalence of CHF.38 Hypothyroidism has also been implicated as a cause of anemia, which in turn causes fatigue and decompensation of CHF, as well as being a predictor of mortality.39 This suggests that thyroid function should be assessed in CHF, particularly in more symptomatic patients, since the symptoms of hypothyroidism and heart failure overlap and the presence of one increases the probability of the other also being present.

The origin of symptoms in CHF is poorly understood and continues to be the subject of research.40–42 We used echocardiography to analyze the hemodynamic and structural changes associated with symptom severity in systolic CHF. A weak but statistically significant correlation was found between E-septum distance and NYHA functional class, but most parameters of LV systolic function (EF, fractional shortening and dP/dT) did not correlate with symptoms. Although the small sample size may have contributed to this result, it is not unexpected since there is a known lack of correlation between ventricular dysfunction and severity of symptoms,25 as shown by the results of clinical trials. For example, patients with mild symptoms in the EMPHASIS-HF study43 had a mean EF of 26%, virtually the same as the more symptomatic patients in the RALES trial.44 This highlights the discrepancy between functional class and LV function in CHF patients.

Elevated LV filling pressures are the cause of dyspnea and pulmonary edema in acute heart failure. However, studies performed during exercise in CHF have shown only a weak correlation between pulmonary capillary wedge pressure and functional capacity.40 Nevertheless, we found a correlation between LV filling pressure, as estimated by the E/E′ ratio, and NYHA functional class at rest in patients with stable CHF. This is consistent with the relation found between congestive symptoms and diastolic dysfunction as assessed by pulsed Doppler study of transmitral flow in patients with dilated cardiomyopathy.45

LV dysfunction results in increased end-diastolic ventricular pressure, passive increases in left atrial and pulmonary venous pressure and development of postcapillary pulmonary hypertension.46 This in turn leads to RV dysfunction and systemic venous hypertension, and their clinical and hemodynamic consequences. These changes are highlighted by the association of increased jugular venous pressure and increased LV filling pressures on cardiac catheterization.47,48 This explains our study's finding of increased LV filling pressures, PASP and IVC inspiratory and expiratory diameters, of which only PASP was an independent predictor of more severe symptoms.

Non-cardiac factors that were not assessed in our study may have contributed to CHF symptoms, including changes in muscle structure and function,49,50 endothelial dysfunction51 and ventilatory dysfunction.52

The independent predictors of functional class >II in our population were PASP and creatininemia.

PASP was the strongest independent predictor of functional capacity in our CHF patients. LV dysfunction is a known cause of pulmonary hypertension53 but its development is highly variable and the contributing factors are not fully understood. As mentioned above, the finding of increased E/E′ ratio, PASP and IVC diameters suggests an interrelationship between high ventricular filling pressures and development of pulmonary hypertension and elevated central venous pressure. It has also been demonstrated that there may be a precapillary component to pulmonary hypertension in systolic CHF.54 Our results are in agreement with studies implicating pulmonary hypertension in functional changes and in prognosis.55

One mechanism through which pulmonary hypertension can contribute to CHF symptoms is increased RV afterload,56 and a strong correlation has been found between oxygen consumption and RV function at rest and during exercise.57 There is also a known correlation between pulmonary hypertension and changes in ventilatory efficiency that contribute to the development of hyperpnea and dyspnea.58,59 In addition to functional repercussions, pulmonary hypertension has also been shown to be associated with prognosis,60 probably related to its impact on RV function. The present study found a correlation between functional class and pulmonary hypertension as assessed by PASP, which may explain the poor prognosis recorded in higher NYHA classes. This finding is of clinical importance as it suggests that noninvasive determination of PASP adds valuable information when assessing hemodynamic changes in patients with systolic CHF, particularly those in NYHA class III or IV. Although effective therapies for such pulmonary hypertension have yet to be identified, recent evidence suggests a role for type 5 phosphodiesterase inhibitors.61

While adding only limited value to the logistic regression model, creatininemia, as a marker of renal function, was shown to have predictive value independently of the other variables analyzed. There is a known relationship between elevated serum creatinine and risk for CHF in elderly patients.62 More recent studies using cystatin C as a marker of renal function have demonstrated that the relationship is even closer, beginning in the preclinical phase of renal dysfunction, which serum creatinine levels do not reveal.63,64 Renal failure makes it more difficult to control heart failure and is one of the strongest markers of adverse clinical events in CHF,65 the coexistence of heart and renal disease being termed cardiorenal syndrome.66 However, the presence of symptoms characteristic of heart failure in patients with chronic renal failure but without diagnosed CHF is common,67 which may be due to the congestive symptoms shared by the two conditions. Thus, the association of elevated creatinine and greater functional impairment may arise from a mutual potentiation of symptoms.

The study has certain limitations. Extrapolation of the results to the general population is limited by the use of a convenience sample. Furthermore, the number of patients included in the study is relatively small, but it has the advantage of being a prospective assessment of important clinical variables such as mortality. The differences in clinical characteristics between the two groups could be considered a limitation in terms of the selection method used. However, the fact that patients in the class >II group were more often included during hospitalization may merely reflect the greater severity of their symptoms. Increased hospitalization has been reported in patients in higher functional classes.5 In addition, although PASP was an independent predictor of functional class, it could not be estimated in all patients due to the absence of, or presence of only mild, tricuspid regurgitation.

Our results suggest that mortality is higher in the more advanced symptomatic stages of systolic CHF. The biomarkers copeptin and NT-proBNP showed a good correlation with symptom severity and high sensitivity and specificity for diagnosis of high functional classes. They may thus be useful in defining a standardized functional classification of patients with LV dysfunction.

The main structural and hemodynamic changes associated with higher functional classes were elevated LV filling pressure, PASP and central venous pressure.

PASP and creatinine levels are independent predictors of NYHA functional class >II in patients with systolic CHF, and so their noninvasive measurement provides important information on hemodynamics and should always be assessed in these patients.

The authors have no conflicts of interest to declare.