Towards a blood test for heart failure: the potential use of circulating natriuretic peptides

Introduction

Heart failure is one of the most common conditions of industrialized society and at its most florid encompasses a constellation of symptoms and signs associated with demonstrable left ventricular systolic dysfunction (LVSD) [1]. Many patients with clinical symptoms and signs of heart failure have no apparent abnormality of ventricular contractile function but may have diastolic heart failure; in addition many patients with left ventricular systolic or diastolic dysfunction do not have any symptoms of heart failure. For the purposes of this review, LVSD refers to demonstrable left ventricular systolic dysfunction, and heart failure to symptomatic left ventricular dysfunction.

Since heart failure is often asymptomatic and for other reasons the diagnosis of heart failure is often difficult. As a result many patients with the condition, in particular when asymptomatic, are denied appropriate pharmacological intervention. The ability to diagnose heart failure from a blood test would not have been predicted even a few years ago. However our increased understanding of heart failure as a complex clinical syndrome associated with marked neurohormonal activation has led to the search for a simple, diagnostic blood-test for the condition. That search may now be coming to fruition with the natriuretic peptides showing the most promise.

The primary aetiological factors in industrialized countries are ischaemic heart disease, hypertension and diabetes, either singly or in combination [2]. Approximately 85% of cases of heart failure in the general population are associated with either coronary disease or hypertension [3]. The clinical syndrome has an estimated prevalence of approximately 1% of the population as a whole and perhaps as high as 10% of those greater than 75 years old [4]. Many patients with significant myocardial damage pass through a period of asymptomatic left ventricular dysfunction before developing overt symptoms, i.e. clinical heart failure. Indeed at any one time there are at least as many individuals with asymptomatic (and largely undiagnosed) LVSD as there are patients with clinical heart failure; the prevalence of asymptomatic LVSD in a recent population based echocardiographic study was around 3% [5]. Clearly, the apparent incidence and prevalence of LVSD and asymptomatic heart failure depend upon definitions of the terms ‘LVSD’ and ‘symptoms’. However in the study of McDonagh et al. [5], LVSD was defined as an echocardiographic ejection fraction of ≤ 30%, a strict definition which excluded many patients who in day to day clinical practice would be considered to have LVSD (i.e. ejection fraction 30–40%).

A combination of factors explains the increasing incidence and prevalence of heart failure: improvements in survival following acute myocardial infarction, in the treatment of hypertension, and in secondary prevention following such events coupled with an increasing number of elderly individuals within the population. Increasing awareness of the problem may also be a contributory factor. It should be emphasized at this point that a diagnosis of heart failure carries with it a very poor prognosis, with mortality similar to that for the common malignant diseases. Even in the current era of vasodilator therapy for the treatment of heart failure, approximately 65% of subjects will die within 5 years of diagnosis [6]. In addition, the potential impact upon quality of life should not be understated. A number of studies of common illnesses have indicated that quality of life for patients with heart failure is worse than that in arthritis, diabetes or chronic obstructive pulmonary disease [7, 8]. In addition heart failure represents one of the major reasons for emergency hospital admission [9]. Therefore it is not surprising that heart failure is a major economic burden to the health care systems of developed countries, accounting for 1–2% of total health care expenditure, 70% of which is related to hospitalization [10]. Although there are clearly major implications incumbent on health services provision, there has been a relative neglect of the problem of heart failure by those responsible for the provision of services, a point illustrated by the absence of mention of heart failure in the Government's Health of the Nation report published in 1991 [11].

The detection of heart failure

The accurate diagnosis of advanced heart failure is straightforward in patients exhibiting overt salt and fluid retention. However, clinical features of heart failure – breathlessness, fatigue and peripheral oedema – are nonspecific and in milder cases typical features are often absent. Even when present these symptoms and signs are often difficult to interpret, particularly among elderly, obese or female patients [12]. The clinical sign with the best positive predictive value is the presence of a third heart sound [13], often a difficult sign for a general practitioner and many general physicians to detect. Isolated lung-field crepitations act as a poor predictor of LVSD having a sensitivity as low as 13% and a specificity of 35% [14].

There have been contradictory reports with regard to the usefulness of the electrocardiogram (ECG) in identifying patients likely to have heart failure. An abnormal ECG has been shown to have a sensitivity of 94% and a specificity of 61% in identifying patients with LVSD and it has been suggested that LVSD is unlikely to be present if an ECG is normal or shows only minor abnormalities [15]. However less promising findings have been demonstrated in other studies: Houghton suggested that 10% of patients with significant LVSD would be missed if the ECG was to be used as the screening tool for suspected heart failure [16]. In a recent hospital based study we found that in patients with severe LVSD (i.e. a wall motion index score ≤ 1.2, equivalent to left ventricular ejection fraction ≤ 35%) 1 in 3 had a normal ECG [17]. A poor relationship exists between the heart size and LV systolic function [18] and the chest-X-ray is an insensitive diagnostic tool in detecting the presence or severity of LVSD.

In view of the evidence that current routine clinical and investigative parameters are inadequate in the diagnosis of systolic dysfunction and heart failure, it is hardly surprising that the accurate diagnosis of heart failure is difficult. In primary care the level of false positive diagnosis, made on clinical grounds, ranges from 30 to 50% [19]. False positive diagnosis is often a result of obesity, chronic pulmonary disease or unrecognized myocardial ischaemia. Furthermore, asymptomatic LVSD is clinically undetectable without assessing myocardial function via echocardiography or ventriculography. Of critical value in reducing the high morbidity and mortality associated with heart failure is accurate assessment of left ventricular function in at-risk populations in order to initiate appropriate therapy at an early stage. This has led to the suggestion that all patients in whom a diagnosis of heart failure is considered should be referred for formal assessment of ventricular function. However, existing methods for assessing this are not without their limitations. Radionuclide ventriculography is a valid method but is limited as a screening procedure in terms of cost, access and the requirement for administration of radioisotope. The echocardiogram is the single most effective tool available to the physician and is the current ‘gold standard’ for the diagnosis of heart failure. Echocardiography requires the availability of trained personnel and adequate equipment and remains restricted in relation to the number of patients in whom the diagnosis of ventricular dysfunction requires confirmation or exclusion – current waiting lists for echocardiography are long. Considerable expansion in the provision of echocardiographic services would be essential for it to be considered as a routine method of community based screening for LVSD even in high risk groups. Such considerations have led to the search for an alternative screening test for heart failure. To date the natriuretic peptides have shown the greatest promise for development of a ‘blood test’ for heart failure.

‘Brain’ natriuretic peptide

The family of natriuretic peptides have the potential to form the basis of a noninvasive test which could enable screening of appropriate populations for LVSD. Brain natriuretic peptide (BNP) is a 32 amino acid peptide [20] originally isolated as a putative neurotransmitter in porcine brain [21]. The primary site of secretion of BNP is now known to be the ventricle of the heart [22]. BNP-32 is the physiologically active product of a high molecular mass pro-hormone proBNP, which has a vasodilatory and natriuretic action as well as antagonizing the physiological effects of the renin-angiotensin-aldosterone system. N-terminal-proBNP (NT-proBNP) is the more recently identified circulating aminoterminal of proBNP [23]. Increased myocardial wall stress appears to be the main stimulus for cardiac natriuretic peptide secretion [22, 24].

The association of BNP with LVSD

Much of the previous work concerning secretion of natriuretic peptides has involved their role in the detection of LVSD following acute myocardial infarction (AMI). Indeed, there are a now several studies documenting a strong association of plasma BNP level with indices of ventricular dysfunction and with prognosis following AMI [25–28].

Population based screening with BNP

Several studies have assessed the predictive accuracy of the natriuretic peptides in screening for LVSD in primary care. Cowie measured cardiac natriuretic peptide levels in 122 patients referred to a rapid access heart failure clinic [29]. Plasma BNP had a higher sensitivity and specificity for the identification of LVSD than ANP or NT-ANP and proved a useful indicator of which patients were likely to have LVSD and should therefore be referred for additional clinical assessment. A plasma BNP level higher than the reference cut-off value positively identified 70% of patients with LVSD. McDonagh et al. undertook a larger and nonselective population based survey involving 1252 patients screened in primary care [30]. The sensitivity and specificity of BNP for identifying LVSD in this community based population was 77% and 87%, respectively, which improved to 92% and 72% in patients over 55 years of age. The positive predictive value for the detection of LVSD was disappointingly low (16%) but is of a level comparable to those of currently used screening tests such as the use of prostate specific antigen for the identification of prostate malignancy [31]. This study identified raised levels of BNP in patients with asymptomatic LVSD. While the absolute numbers of patients with LVSD was low, the authors concluded that BNP could be used to screen the general population for LVSD, particularly targetting high-risk groups [32]. Indeed positive predictive value of an elevated BNP level was approximately doubled in patients with a history of ischaemic heart disease.

NT-proBNP represents a marker of cardiac impairment potentially more discerning than BNP-32 itself. NT-proBNP is elevated in asymptomatic LVSD and the proportional and absolute increases of levels in plasma exceed those of BNP-32 in both asymptomatic LVSD and in heart failure [32]. Furthermore, by analogy with the atrial natriuretic peptide (ANP) hormonal system, where the N-terminal precursor provides a more sensitive index of cardiac dysfunction than ANP itself, it seems likely that a similar association exists for the BNP family of peptides [33]. Plasma NT-proBNP has recently been shown to correlate with left ventricular wall motion index, a regional measurement of ventricular function. In a population of patients referred for echocardiography; positive and negative predictive values were 58% and 93%, respectively, for the detection of left ventricular wall motion index of ≤ 1.2 (i.e. severe LVSD) [17]. In a separate study circulating NT-proBNP ≥ 145 pmol l⁻¹, measured 2–4 days following AMI, had positive and negative predictive values of 56% and 80%, respectively, for the detection of LVEF ≤ 40% [34]. The negative predictive value of the assay was again > 90%. Both these studies have addressed the value of NT-proBNP in relatively high risk populations. Routine assay of NT-proBNP may assist in the risk stratification of the post infarct population, a group clearly at high risk of progressing to symptomatic heart failure. Studies are ongoing with regard to the value of NT-proBNP in the identification of patients with LVSD in primary care. Again interest will centre on the relative value of the test in unselected cohorts as compared to patients at high risk of LVSD.

Recently a novel, sensitive and specific immunoluminometric assay for NT-proBNP has been developed [35]. This nonradioactive assay is inexpensive and easy to perform and would be appropriate for use in screening for LVSD since large numbers of specimens can be processed rapidly. In addition, NTproBNP remains stable in whole blood at room temperature for up to 48 h, representing a critical attribute for any putative biochemical test [36]. These factors together would facilitate measurement of NT-proBNP as an index of ventricular function and provide an aid to clinical assessment alone. Although such a biochemical marker should not be used to provide absolute confirmation or exclusion of a diagnosis it may serve as a means of directing individuals suspected of having heart failure for further assessment when access to investigations such as echocardiography are limited by expense, inconvenience or the availability of equipment or expertise. This would facilitate rapid and accurate assessment of left ventricular function and effect the early introduction of appropriate therapies of known benefit, namely angiotensin converting enzyme inhibitors [37, 38] and β-adrenoceptor-blockers [39–41], at a time when the potential for clinical benefit is at a maximum. Furthermore the presence of high concentrations of NT-proBNP in LVSD and heart failure is an ideal characteristic for the potential development of a near patient assay for this peptide.

Table 1 summarizes the most important recent studies in identifying LVSD or heart failure using the natriuretic peptides [17, 29, 30, 34]. It is clear that these peptides in general have high negative predictive values for detection of LVSD, so that a normal test virtually excludes significant impairment of cardiac function. It should be emphasized that while measurement of natriuretic peptide levels is by no means perfect in terms of diagnosing or excluding LVSD or heart failure, this test is superior to clinical assessment alone. Due to the relatively low positive predicitive values and specificities, an elevated plasma concentration should indicate the need for further detailed cardiological investigations.

Table 1.

Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of the natriuretic peptides identifying HF or LVSD.

	Sensitivity (%)	Specificity (%)	PPV(%)	NPV (%)
ANP, Heart failure [29]	97	72	55	98
ANP, Post MI LVF [34]	79	71	50	90
N-ANP, Heart failure [29]	97	66	54	98
N-ANP, Post MI LVF [27]	76	76	54	89
BNP, Heart failure [29]	97	84	70	98
BNP, LVSD [30]	77	87	16	98
BNP, Post MI LVF [34]	85	73	54	93
N-BNP, LVSD [17]	94	55	58	93
N-BNP, Post MI [34]	82	69	50	91

BNP and ‘diastolic dysfunction’

A significant proportion of patients with signs and symptoms of heart failure have preserved systolic function but may have echocardiographic evidence of abnormalities of left ventricular relaxation, so-called ‘diastolic dysfunction’. Plasma BNP levels correlate with prognosis in this group of patients [42]. While identification of these patients using either echocardiography or plasma BNP level is possible, the therapeutic implications of so doing are unclear as the appropriate treatment for diastolic dysfunction has not been established.

Additional uses of BNP in heart failure

In addition to its ancillary role in the diagnosis of LVSD, the cardiac natriuretic system may have the potential to aid the physician's treatment of patients with heart failure. A recent study sought to determine whether plasma BNP was of value in the individual optimization of heart failure therapy [43]. Titration of the dose of ACE inhibitor and other drug therapies until reduction of plasma BNP to below 200 pg ml⁻¹ resulted in a more favourable outcome compared with those patients where standard therapy was administered without knowledge of the BNP response. Similarly, plasma natriuretic peptide levels may represent a sensitive measure of disease exacerbation and progression. Whether such findings can be translated into useful clinical application of the test remains to be established.

The cardiac natriuretic peptide system itself also represents an attractive site for therapeutic intervention in heart failure. The diuretic, natriuretic and vasodilatory actions of ANP and BNP may in theory be harnessed by blocking the natriuretic peptide clearance receptor or by inhibition of neutral endopeptidase (the enzyme responsible for the hydrolysis of ANP and BNP), an area of active research at the current time. While short-term infusion of BNP has beneficial haemodynamic effects in heart failure [44], the oral agents investigated to date have been disappointing in terms of their natriuretic and diuretic efficacy. Newer agents such as the combined ACE/neutral endopeptidase inhibitor omapatrilat show great promise as therapeutic agents in heart failure. In contrast it has been suggested that the natriuretic peptides may in some way contribute to the progression of the syndrome of heart failure. There is little evidence to support such an assertion.

Conclusions

Heart failure represents a serious personal and public health issue. Interest is currently focused upon the early identification of LVSD and prevention of its progression. Many patients with LVSD will be missed if clinical signs alone are used to select patients for appropriate therapy. One of the most promising areas of current cardiological research is the study of secretion and pathophysiological effects of the natriuretic peptides. A strong association exists between circulating concentrations of BNP with LV function and patient prognosis. NT-proBNP represents a potentially more discerning marker of these parameters. Further work is essential in determining the value of measuring NT-proBNP as a means of population based detection of LVSD. Such screening programmes are likely to be of the greatest value in identifying patients at high risk of the condition, e.g. those with a history of ischaemic heart disease, hypertension or diabetes. The lack of complex handling and storage requirements and its ease of measurement in plasma represent ideal attributes for such a putative screening test. The use of the natriuretic peptides is as yet unlikely to replace established methods of assessing ventricular dysfunction such as echocardiography and radionuclide ventriculography. However measurement of plasma markers (particularly when aimed at high risk individuals) may better identify those individuals in whom further investigation is required than clinical criteria alone. This would limit inappropriate and unnecessary investigations and moreover would allow for the earlier identification of patients with both symptomatic and asymptomatic left ventricular dysfunction. The potential impact of population based screening for HF is widespread and represents an exciting and challenging prospect. The precise role of cardiac natriuretic peptides in population based screening programs requires further and more detailed investigation.