Chronic heart failure (HF) is a major cause of death and disability. Predicting patients at risk of death or further cardiovascular events that require more intense monitoring and therapy is clinically challenging. Several factors are associated with increased mortality and morbidity in patients with HF, including age, a history of diabetes mellitus or renal dysfunction, advanced functional disability, reduced left ventricular ejection fraction, lower sodium concentrations, lower body mass index, lower blood pressure, the presence of ankle oedema and lower quality of life scores.1 Accordingly, multivariable risk scores that differently combine these clinical variables have been derived from well-characterised populations of patients with HF, but they may be of limited practical use.
There is currently a great interest in developing simple and economical circulating biomarkers that would be of incremental value to conventional clinical risk factors and establish rapid and accurate risk stratification of patients with HF or help in guiding therapy.2,3 To date, markers of the natriuretic peptides family are the most promising tools. Atrial natriuretic peptide (ANP) and brain natriuretic peptides (BNP) are of cardiac origin and are therefore more relevant to the pathophysiology of HF. These two hormones are secreted mainly by stretched atrial and ventricular myocytes (increased myocardial tension and intravascular volume), but also during ischaemia, and act on specific receptors and exhibit a wide range of physiological effects including natriuresis, diuresis, vasodilatation and opposition to fibrosis, cellular hypertrophy and neurohormonal activation.
ANP and BNP are secreted as inactive circulating prohormones (proANP and proBNP) that are successively processed in the respective mature C-terminal hormones and N-terminal peptides of presently unknown function.4 There are convincing data, briefly reviewed here, that the determination of the physiologically inactive N-terminal proBNP (NT-proBNP 1–76) provides powerful and independent prognostic information in patients with chronic HF in different clinical settings and at all stages of the disease.
Prognostic Value in Multicentre Controlled Clinical Trials
The strongest evidence of the prognostic value of NT-proBNP in HF arose from large-scale multicentre controlled clinical trials and robust statistical interpretation. In the Valsartan Heart Failure Trial (Val-HeFT), 5,010 patients with mild to moderate chronic HF were randomised to an angiotensin II type 1 receptor blocker or placebo.5 Blood was collected at study entry for the measurement of different biomarkers.
Mortality and the combined end-point of mortality and morbidity (mainly hospitalisation for HF), the two centrally adjudicated end-points, increased progressively with increasing plasma concentrations of NT-proBNP, starting from levels well below the diagnostic value for chronic HF (see Figure 1). An increment of 500pg/ml of NT-proBNP was associated with an increased adjusted risk of 3.8% for mortality and 3.0% for hospitalisation for HF.6 Moreover, NT-proBNP ranked as the first independent prognostic factor in these patients, ahead of traditional clinical risk factors such as functional disability by New York Heart Association (NYHA) class, age, left ventricular dilation or renal dysfunction.
The prognostic performance of NT-proBNP has been evaluated in several other clinical trials. NT-proBNP predicted one-year mortality in a sub-study of the Prospective Randomised Amlodipine Survival Evaluation (PRAISE) 2 trial that enrolled 181 patients with severe congestive HF (NYHA III-IV and left ventricular ejection fraction less than 30%) of a presumed non-ischaemic aetiology.7
NT-proBNP also predicted death and hospitalisation for HF in a subgroup of 1,011 patients with a left ventricular ejection fraction less than 25% and severe HF in the Carvedilol Prospective Randomized Cumulative Survival (COPERNICUS) trial.8 In this trial, NT-proBNP above the median concentration was a powerful predictor of one-year all-cause mortality (relative risk (RR), 2.7; 95% confidence interval (CI), 1.7–4.3; p=0.0001) and of all-cause mortality or hospitalisation for heart failure (RR, 2.4; 95% CI, 1.8–3.4; p=0.0001). There was no significant interaction between NT-proBNP and study treatment.8 The prognostic value of NT-proBNP in patients with severe HF has also been observed in smaller single-centre studies9,10 or in patients referred for consideration of, or after cardiac transplantation.11–14
Incremental Prognostic Value of Repeated Measurements of NT-proBNP
A single determination of NT-proBNP, at any time during the progression of chronic HF, provides a clinically useful tool for risk stratification. Repeated measurements of the same biomarker over time convey incremental prognostic information beyond that of a single one, as shown in different settings, and may help in monitoring the progression of disease and the clinical effect of medical therapy.2 For instance, relative changes in NT-proBNP concentrations during hospitalisation were good predictors of subsequent hospital readmission and death within six months of discharge in 182 patients admitted to hospital due to decompensated HF.15
Thus, changes in NT-proBNP may be potentially useful in assisting clinicians in making the decision to discharge HF patients. In another study16 patients with a reduction of NT-proBNP greater than 50% after cardiac resynchronisation therapy had a more favourable outcome than the rest of the population and this was associated with positive changes in left ventricular remodelling and exercise capacity. Absolute changes in NT-proBNP concentration across a given value (for instance median concentration) can also predict mortality in patients with advanced HF referred for consideration of cardiac transplantation,17 as already shown in Val-HeFT for BNP in patients with mild to moderate chronic HF.18 Likewise, absolute changes in NT-proBNP concentration across the median over four months were related to outcome in 3,488 patients with chronic HF enrolled in Val-HeFT (see Figure 2).
The intra-individual biological variation of NT-proBNP should be considered for a correct understanding of the clinical significance of repeated determinations of this biomarker, even in patients deemed clinically stable. Estimates of biological variation are not easily interpretable since they vary with the clinical criteria used for the selection of the population examined, the definition of clinical stability, the severity of disease in the population examined, the interval of time considered and the statistical methods used to calculate them. Reference change values for week-to-week variations of NT-proBNP are relatively high, in the range of 50–100%.19,20
However, at longer intervals, recent findings indicate that the biological variability could be smaller (reference range values of 23%).21,22 For instance, the year-to-year co-efficient of variation (standard deviation of repeated measurement/mean concentration) of NT-proBNP amounted to 30% in patients with chronic HF (5% when NT-proBNP was expressed on a logarithmic scale), but these data were observed in a highly selected population of patients with stable chronic HF, not necessarily representative of everyday ambulatory patients.21 Importantly, changes in NT-proBNP concentrations below the estimated biological variation have clinical relevance.
NT-proBNP and the Multimarker Approach
Head-to-head comparative studies of circulating factors generally show the superiority of NT-proBNP in terms of prognostic performance over other currently measured biomarkers (aldosterone, endothelin peptides, adrenomedullin, tumour necrosis factor-alpha, C-reactive protein, norepinephrine, erythropoietin).6,13,23
However, the various fragments of the natriuretic peptide family (NT-proANP, mid-regional proANP, BNP, NT-proBNP) seem roughly equivalent for risk stratification in HF.6,24–28 Markers of cardiac injury (cardiac troponin T,29 or troponin I30) or imaging techniques (echocardiography31,32) may be combined with NT-proBNP to further improve the prognostic performance and accuracy compared with the natriuretic peptide alone.
Prognostic Value in Patients with Chronic Heart Failure and Preserved Systolic Function
NT-proBNP increases with the severity of ventricular systolic as well as diastolic dysfunction.33–37 However, whereas the prognostic value of NT-proBNP in patients with reduced systolic function is well documented, there are few data in patients with chronic HF and preserved systolic function even though they may account for as much as 50% of the patients with symptomatic HF, particularly in the elderly. NT-proBNP was a strong predictor of one-year mortality in 161 consecutively hospitalised patients with heart failure with preserved as well as reduced systolic function.38 Similarly, NT-proBNP is useful to identify patients at risk of adverse outcome regardless of systolic function, as seen in 244 patients admitted for decompensated HF and followed for six months.39
In conclusion, NT-proBNP is a robust prognostic marker for all relevant clinical outcomes in chronic HF. Repeated determinations of this biomarker may convey additional prognostic value. More studies are needed to assess the performance of NT-proBNP in patients with chronic HF and preserved left ventricular systolic function or in clinical subgroups of patients with HF and co-morbidities (diabetes, atrial fibrillation, chronic obstructive pulmonary disease).