Heart Failure Reviews

, Volume 15, Issue 4, pp 275–291

Utility of natriuretic peptide testing in the evaluation and management of acute decompensated heart failure

Authors

    • Cardiomyopathy ProgramLoma Linda University Medical Center
  • Geoffrey T. Jao
    • Section of General Internal MedicineWake Forest University Baptist Medical Center
  • Kirkwood F. AdamsJr.
    • Departments of Medicine and RadiologyUniversity of North Carolina Chapel Hill
Article

DOI: 10.1007/s10741-009-9141-2

Cite this article as:
Chiong, J.R., Jao, G.T. & Adams, K.F. Heart Fail Rev (2010) 15: 275. doi:10.1007/s10741-009-9141-2

Abstract

The B-type natriuretic peptide (BNP) and the amino-terminal fragment of proBNP (NT-proBNP) are increased in heart failure in proportion to severity of symptoms, degree of left ventricular dysfunction, and elevation of cardiac filling pressures. These natriuretic peptides (NPs) are increasingly used for diagnostic and prognostic purposes in acute heart failure. While NP levels on admission provide independent prognostic information, serial determinations during hospitalization and at discharge better reflect adequacy of treatment and prognosis. The addition of BNP and NT-proBNP to usual clinical decision making enhances detection of high-risk patients who need aggressive follow-up and adjustment of treatment.

Keywords

Natriuretic peptidesAcute heart failurePrognosis

Introduction

Heart failure is a growing public health epidemic with more than 5 million patients afflicted in this country, even as therapies proliferate and management becomes increasingly complex [1, 2]. As our geriatric population increases, estimates suggest that the number of Medicare beneficiaries with heart failure will exceed 50 million by 2020 [3, 4]. This implies that the demand for healthcare resources in managing heart failure will increase tremendously in a very short period of time. This enormous challenge mandates continued search for new and innovative strategies that will optimize resource utilization through improved patient management.

Acute heart failure syndromes (AHFS) represent the major contributor to the health care burden from heart failure with over 1 million annual hospitalizations in the United States alone [5]. The economic cost of heart failure is estimated to be $56 billion a year and 70% of this expense is from hospitalizations [6]. Patients experiencing AHFS face substantial risk post-discharge with event rates of approximately 30–40% for readmission or death at 6 months [7]. Research efforts are intensifying to identify more effective therapies and management strategies for this lethal and debilitating syndrome [8]. An emerging approach for guiding the evaluation and management of patients with AHFS involves the use of cardiac biomarkers.

B-type natriuretic peptide (BNP) and the amino-terminal pro-natriuretic peptide (NT-proBNP) are the two most widely studied and commercially available cardiac biomarkers for the assessment of AHFS (Table 1). Originally used as diagnostic tools, recent studies demonstrate their utility as prognostic markers and their contribution to cost-effective management of acute heart failure. Their use has been incorporated into the latest heart failure guidelines for risk stratification [1]. This review focuses on the role of natriuretic peptide (NP) testing in managing hospitalized patients with AHFS and examines the latest evidence supporting the current recommendations.
Table 1

Comparison of key features of the natriuretic peptides

 

BNP

NT-proBNP

Size

32 amino acids

76 amino acids

Biologically active

Yes

No

Half Life

18 min

60–120 min

Elimination

Neutral endopeptidase

Renal clearance

Clearance receptors

Renal Clearance

Stability

Brief—4 h (molecule starts degrading in vitro after extraction)

Up to 72 h

Manufacturer recommended diagnostic cutoffs for diagnosis of heart failure

<100 pg/ml—No HF

For heart failure

100–400 pg/ml—Gray zone

<50 yo >450 pg/ml

>400 pg/ml—HF

50–75 yo >900 pg/ml

 

>75 yo >1,800 pg/ml

HF heart failure, yo years old

Diagnosis of acute heart failure

Background

Despite advances in our understanding of the pathophysiology of heart failure, the diagnosis of heart failure, especially in the acute setting, remains challenging. The Framingham criteria have been used in many studies as a standard for the diagnosis heart failure [9]. Epidemiological studies suggest that patients identified as having chronic heart failure by the Framingham criteria have reduced survival consistent with patients diagnosed by clinical consensus [10]. Whether these criteria can be treated as a score and used to identify those patients who are at high or low risk has not been well studied. The work of Gackowski et al. in acute heart failure suggests, in fact, that while a score based on the criteria showed a strong univariate association with prognosis, this measure was not significant in multivariable analysis. This lack of prognostic value in the setting of acute heart failure seemed to stem from the additional value of two clinical characteristics, the chronicity of the disease, and the severity of clinical presentation (inotrope use or not), which are not part of the criteria [11].

The evaluation of the dyspneic patient in the acute setting is arguably more difficult as disease severity and clinical immediacy may preclude obtaining a thorough history or conducting an adequate physical examination. Fewer than half of the patients with acute heart failure present with classic signs and symptoms, and approximately 50% of new heart failure patients are misdiagnosed on presentation. Routine diagnostic workup, in addition to the history and physical, often fails to show whether the dyspnea is cardiac or pulmonary in origin [12]. This is particularly true among elderly patients who have a higher prevalence of co-morbid conditions that may have similar signs and symptoms [1, 4]. A disconnect also exists between the perceived severity of heart failure by physicians and severity by other criteria [13].

The initial clinical application of NPs was to aid in the diagnosis of AHFS. Subsequent to early study, NPs have become well established as markers that distinguish heart failure from pulmonary and other clinical disorders with high sensitivity and specificity [14]. These biomarkers often allow the prompt differentiation of non-heart failure causes of dyspnea and can help identify heart failure due to systolic or diastolic ventricular dysfunction. Current evidence indicates that the addition of NP testing to standard clinical assessment significantly improves diagnostic accuracy in the dyspneic patient [1421] (Table 2).
Table 2

Key studies establishing the diagnostic use of BNP and NT-proBNP in AHFS

Study

N

Year

NP

Assay

Diagnostic cutoffs (pg/ml)

LVD PSF

BNP [17]

1,586

2002

BNP

Biosite

100

Both

REDHOT [13]

464

2004

BNP

Biosite

<200

Both

PRIDE [19]

600

2005

NT-proBNP

Roche

>450 (<50 yo)

Both

>900 (>50 yo)

<300 (optimal r/o age independent)

ICON [20]

1,256

2005

NT-proBNP

Roche

>450 (<50 yo)

Both

>900 (50–75 yo)

>1,800 (>75 yo)

AHFS acute heart failure syndromes, BNP brain natriuretic peptide, LVD left ventricular systolic dysfunction, PSF preserved systolic function, NP natriuretic peptide, NT-proBNP amino-terminal pro-brain natriuretic peptide, yo years old

PRIDE study

The utility of NT-proBNP in the diagnosis of acute heart failure in patients presenting to the emergency department was investigated in the ProBNP Investigation of Dyspnea in the Emergency Department (PRIDE) study [19]. This prospective study enrolled 600 patients who presented to the emergency department with dyspnea. The “gold standard” for the diagnosis of heart failure was an independent evaluation of the study patients by physicians who were blinded to NT-proBNP results. Physicians involved in determining the diagnosis of heart failure had access to medical information from the patient’s record as well as study data obtained from enrollment through 60 days of follow-up. In 10% of the cases, there was disagreement on the diagnosis that was resolved by the application of the Framingham criteria for heart failure. Patients were classified as having new onset heart failure, acute decompensation of chronic heart failure, or dyspnea not related to heart failure.

As expected, the diagnosis of acute heart failure was highly related to NT-proBNP level [19]. The median NT-proBNP level was dramatically higher among those patients classified with acute heart failure (209 patients either new onset or established) at 4,054 pg/ml versus 131 pg/ml in those without heart failure. In the initial report, NT-proBNP cut-points of >450 pg/ml for patients <50 years of age and >900 pg/ml for patients ≥50 years of age were found to be highly sensitive and specific for heart failure. For ruling out heart failure, an NT-proBNP level of <300 pg/ml was suggested as optimal with a negative predictive value of 99%. NT-proBNP in addition to clinical judgment was best for the diagnosis of acute heart failure (Fig. 1a). NT-proBNP levels were found to correlate very well with the severity of symptoms with median values of 1,591 pg/ml in functional class II, 3,438 pg/ml in functional class III, and 5,584 pg/ml in functional class IV. Of interest, the presence of classical signs and symptoms of heart was variable and specific signs, and symptoms were generally present only in a minority of patients. Data on the diagnosis of those patients without heart failure are also of interest. In 150 cases, the diagnosis was COPD or asthma, pneumonia in 64, acute coronary syndrome in 31, acute pulmonary embolus in 19, acute bronchitis in 10, and different diagnosis was present in 116 cases.
https://static-content.springer.com/image/art%3A10.1007%2Fs10741-009-9141-2/MediaObjects/10741_2009_9141_Fig1_HTML.gif
Fig. 1

a Receiver-operating characteristic curve comparison of NT-proBNP versus clinician-estimated likelihood for the emergency department diagnosis of acute heart failure. Results of NT-proBNP testing were superior to those of clinical judgment, with significantly greater area under the curve (0.94 vs. 0.90, P < 0.006). The area under the curve from NT-proBNP testing plus clinical judgment (0.96) was superior to each diagnostic modality alone [19]. b Receiver-operating characteristic curve comparison of BNP versus clinician-estimated likelihood for the emergency department diagnosis of acute heart failure versus the combination of these assessments [16]

Breathing Not Properly study

The Breathing Not Properly study evaluated the diagnostic utility of BNP in the diagnosis of patients presenting with acute dyspnea [17]. This multi-center prospective study evaluated patients who were seen with acute dyspnea in the emergency department. The diagnosis of heart failure was established based on an independent review by two cardiologists who had access to medical records on the study patients but were blinded to BNP levels. The final diagnosis was heart failure in 47%, no finding of heart failure in 49%, and left ventricular dysfunction but no heart failure in 5%. A BNP cutoff level of 100 pg/ml had an 83.4% diagnostic accuracy and the negative predictive value of a level <50 pg/ml was 96%. The addition of BNP to clinical judgment was found to be of significant value in establishing the diagnosis of heart failure (Fig. 1b).

Diagnostic cut-points for NPs

Delineation of optimal diagnostic cut-points is critical to the application of NP testing in acute heart failure and has been an ongoing subject of investigation. Initial published studies supporting the use of BNP and NT-proBNP for the evaluation of dyspneic patients had limited numbers of patients and were restricted to single centers [19, 22, 23]. Furthermore, the patient characteristics and co-morbid conditions varied widely among these studies, making it difficult to compare results on cut-points. Results from the Breathing Not Properly study and the ICON study, discussed in detail below, have helped define optimal cut-points for these biomarkers [17, 20]. Using data from the Breathing Not Properly study, the BNP cutoff point of 100 pg/ml was judged to be optimal for making the diagnosis of heart failure [17]. This cutoff was slightly higher than that found in the pilot study (100 pg/ml vs. 80 pg/ml) and lower than that of another similar study [24, 25]. In order to provide more definitive information concerning diagnostic cut-points for NT-proBNP, several heterogeneous, multi-national patient populations were pooled for the analysis in the International Collaborative of NT-proBNP (ICON) study [20]. This combined analysis included 1,256 dyspneic subjects from four sites spanning three continents. The authors found that dividing patients into three age groups (<50, 50–75, and >75 years) yielded the best diagnostic accuracy. The following specific cut-points were identified as optimal for the respective age groups from this analysis: 450, 900, and 1,800 pg/ml. Overall, this approach yielded 90% sensitivity and 84% specificity for the diagnosis of acute heart failure. Additionally, an age-independent cut-point of 300 pg/ml was found to have a 98% negative predictive value in excluding the diagnosis of acute heart failure.

The median NT-proBNP concentration of those patients with acute heart failure exacerbation (4,639 pg/ml) was significantly higher than those with neither acute nor prior heart failure (108 pg/ml). There were 55 cases that had a history of prior heart failure but were judged not to have acute heart failure at presentation. These 55 patients had higher mean NT-proBNP concentrations (949 pg/ml) than those without acute heart failure or a history of prior heart failure. In contrast, the mean NT-proBNP levels of these 55 patients were significantly lower than that of those with acute heart failure.

Among patients diagnosed with acute heart failure, there was a significant increase in median NT-proBNP level as symptom severity increased, although significant overlap existed between these clinical groups. Patients with acute heart failure, but preserved left ventricular systolic function, had lower mean NT-proBNP concentrations (3,070 pg/ml) compared to those with impaired systolic function (6,536 pg/ml) but levels of this biomarker were useful in both types of patients. Similar results have been reported with BNP determinations [26].

The gray zone values

As expected NP values may fall in an indeterminate range between a lower level that rules out heart failure and a higher level that effectively rules in this diagnosis. As discussed in detail below, there is evidence that certain clinical characteristics enhance the likelihood that patients with or without acute heart failure will have indeterminate levels. As inter-individual variability in NP levels among patients with AHFS is also common, it is important not to interpret these biomarker results in a strictly dichotomous fashion.

Results from a substudy of PRIDE led to the development of a simple, rapid, and accurate bedside tool (PRIDE Acute Heart Failure Score, Table 3) to assist in finalizing a diagnosis in patients with gray zone values [27]. This score emphasizes the strengths and minimizes the weaknesses of both clinical and biomarker diagnostic approaches. The score includes eight readily available variables found to be statistically significant independent predictors of acute heart failure at presentation: elevated NT-proBNP results (considered elevated if ≥450 pg/ml in patients aged <50 years or ≥900 pg/ml in patients ≥50 years), interstitial edema on chest X-ray, orthopnea, absence of fever, current loop diuretic use, age >75 years, rales on lung examination, and absence of cough. This score can be considered as a continuous estimate of the likelihood of acute heart failure, but can be considered in a categorical way to optimize clinical utility. Patients with a score of 0–5 points have a low probability, patients with a score of 6–8 points have an intermediate probability, while those with a score of 9–14 points have a high likelihood for AHFS (Fig. 2). This categorical approach maintains the benefit of a dichotomous score of <6 points for ruling out the diagnosis of heart failure (with a negative predictive value of 98%), while emphasizing the continuously increasing likelihood of heart failure with rising scores. The overall median score was five points. A score threshold of ≥6 points for the diagnosis of acute heart failure yielded a sensitivity and specificity of 96% and 84%, respectively. A suggested integrated approach using the PRIDE Acute Heart Failure Score, NT-proBNP measurements, and other medical data are shown in Fig. 3 [28].
Table 3

Definition of the PRIDE Acute Heart Failure Score

Predictor

Points

Elevated NT-proBNP

4

Interstitial edema on chest X-ray

2

Orthopnea

2

Absence of fever

2

Current loop diuretic use

1

Age > 75 years

1

Rales on lung examination

1

Absence of cough

1

A score of ≥6 has a high predictive accuracy for the diagnosis of acute heart failure [27]

https://static-content.springer.com/image/art%3A10.1007%2Fs10741-009-9141-2/MediaObjects/10741_2009_9141_Fig2_HTML.gif
Fig. 2

Distribution of PRIDE Acute Heart Scores among a all the patients in the entire PRIDE study and b expressed as a function of the presence (black bars) or absence (gray bars) of acute heart failure [27]

https://static-content.springer.com/image/art%3A10.1007%2Fs10741-009-9141-2/MediaObjects/10741_2009_9141_Fig3_HTML.jpg
Fig. 3

Diagnostic algorithm based on NT-proBNP, clinical assessment, and PRIDE heart failure score [28]

Performance of the PRIDE Acute Heart Failure Score was specifically evaluated among the subgroup of patients in whom the diagnosis of heart failure was uncertain during initial clinical evaluation [27]. A score ≥6 points for the diagnosis of acute heart failure yielded a sensitivity of 90% and a specificity of 87% in this subpopulation, consistent with the overall score performance in the PRIDE study (Fig. 4).

https://static-content.springer.com/image/art%3A10.1007%2Fs10741-009-9141-2/MediaObjects/10741_2009_9141_Fig4_HTML.gif
Fig. 4

Performance of the PRIDE Acute HF Score among patients for whom the diagnosis was uncertain at the time of clinical evaluation. A score ≥6 points for the diagnosis of acute HF yielded a sensitivity of 90% and a specificity of 87% in this subpopulation, consistent with the overall score performance in the PRIDE study [27]

This scoring system emphasizes that the signs and symptoms of AHFS may also be found in other disease states. This score also encourages clinicians to consider ruling out non-cardiac causes of NP elevation during the diagnostic evaluation, and potentially reduce improper medication administration for those without heart failure.

Impact of clinical characteristics on diagnostic utility of NPs

Since their emergence, limitations are now evident in the diagnostic utility of NPs. Elevation of BNP and NT-ProBNP is not fully specific for heart failure. Several factors, particularly older age, depressed renal function, sepsis, and pulmonary embolus may be associated with increased NPs even when AHFS is not present [24, 26, 2934]. In addition, a number of demographic and clinical characteristics commonly co-exist in patients presenting with acute heart failure and may influence the degree of elevation of NPs present in this syndrome. These characteristics must be taken into account when interpreting the results of biomarker testing. Using a single reference range for the NPs to diagnose heart failure is clearly inadequate in patients with these characteristics. A number of studies reviewed below suggest appropriate cutoff values in clinical subsets of this diverse patient population.

Age and gender

There is a direct relationship between age and NP levels, which is independent of the presence or absence of heart failure [35, 36]. This relationship between older age and increased levels of NPs is likely a consequence, in part, of decreased left ventricular compliance and worsened renal function [3739]. While being modest, when considering criteria for the assessment of left ventricular function in community populations, age-related increases in BNP and NT-proBNP are important to consider.

The magnitude of age-related increase in NPs does not suggest that age would be an important consideration in the diagnostic use of these biomarkers. However, data from the Breathing Not Properly study and the ICON study do suggest that among acutely dyspneic patients, higher NP levels have decreased specificity for heart failure in older patients. In the case of BNP, this effect has been stated to be modest [40]. The argument is made that upward adjustment of cut-points in older subjects may be undesirable since failing to treat cases of heart failure may be worse than treating negative cases [40].

In contrast, NT-proBNP levels definitely increase with age and influence diagnostic accuracy [19, 20, 38, 41, 42]. In a younger population (<50 years old), a NT-proBNP cutoff level of 900 pg/ml has 73% sensitivity and 96% specificity for the diagnosis of heart failure. However, this same cut-point was only 91% sensitive and 80% specific in older patients (≥50 years old). However, when age-adjusted cut-points from the ICON study are applied to NT-proBNP levels for diagnosis of heart failure (450 pg/ml for age <50; 900 pg/ml for age 50–75; and 1,800 pg/ml for age >75 years), then no additional adjustment is needed for age. The study of Chenevier-Gobeaux et al. [43] provides some interesting, clinically relevant data on NT-proBNP levels in very elderly patients (age ≥85) who presented with acute dyspnea. They found similar threshold values for BNP among this very elderly patient group compared to elderly patients aged 65–84 years (290 pg/ml vs. 270 pg/ml), but NT-proBNP threshold values were higher in patients ≥85 years (1,700 pg/ml vs. 2,800 pg/ml). Additional studies seem warranted to better define cutoff values for patients 85 years or older. In contrast, age stratification is not necessary to exclude the diagnosis of acute heart failure. Interestingly, results from the ICON study indicated that a single cut-point of 300 pg/ml (95% confidence interval 241–369 pg/ml) demonstrated a sensitivity of 99%, a specificity of 60%, and a negative predictive value of 98%. Using the approach of confirmation and exclusion cutoff points for NT-proBNP will help clinicians more confidently utilize this marker in the evaluation of the dyspneic patient, preserving sensitivity for younger patients with suspected heart failure, while optimizing specificity for elderly patients [19, 41].

In contrast to age, in the ICON study, there was no difference in NT-proBNP levels between men and women with acute heart failure. Female gender was associated with a trend toward higher median NT-proBNP in patients without acute heart failure (190 pg/ml vs. 160 pg/ml, P = 0.12), but the median NT-proBNP concentration in female patients with acute heart failure (5,801 pg/ml) did not differ significantly from male patients with acute heart failure (5,645 pg/ml, P = 0.66). There is a suggestion that NPs might convey more prognostic information in women than men admitted with AHFS [44].

Obesity and NPs

The differential diagnosis of acute dyspnea in obese patients can be particularly difficult. In these patients, classic signs and symptoms are not always present. Body habitus may mask signs of edema and may muffle the heart and lung sounds during auscultation. History also can be less reliable, because obese patients frequently have co-morbid conditions that can mimic heart failure, including dyspnea caused by deconditioning and orthopnea due to increased abdominal size.

In addition to these clinical pitfalls, NP levels are significantly affected by body mass. Levels are lower when body weight is elevated for the same degree of clinical heart failure [39, 45]. The study of Daniels et al. [45] revealed an inverse correlation between log BNP and body mass index (BMI), which was even stronger in patients with AHFS. They utilized patients enrolled in the Breathing Not Properly study to investigate appropriate cutoff points to maintain 90% sensitivity for the diagnosis of heart failure in patients presenting with acute dyspnea. Their results showed that a BNP cutoff of 170 pg/ml was appropriate for lean subjects (BMI <25), 110 pg/ml for overweight or obese subjects (BMI ≤25 to <35), and 54 pg/ml in severely or morbidly obese patients (BMI ≥ 35). These cutoff values were reflected in the mean BNP levels of these three body weight groups, which were 643, 462, and 247 pg/ml, respectively. Obesity is common in patients with AHFS, and totally in 19% of the patients in the study reported by Daniels et al. were severely or morbidly obese (BMI ≥ 35).

Frankenstein et al. [39] found that NT-proBNP concentrations were lower in obese patients with chronic stable heart failure. This study found that BMI was a significant univariate predictor of NT-proBNP level, with a 4% decrease in NT-proBNP concentration per unit increase of BMI. Taylor et al. found that NP concentrations were lower in obese patients despite the presence of higher left ventricular end diastolic pressures. This raises at least the possibility that decreased concentrations of NPs may play a role in the development of heart failure in obese patients [46]. The PRIDE study also confirms that overweight and obese dyspneic patients with AHFS (BMI ≥ 25) have significantly lower NT-proBNP concentrations compared with lean patients (BMI < 25) with this diagnosis. Linear regression analysis showed that BMI was inversely associated with NT-proBNP concentrations, even after adjustment for all other significant covariates [47]. Although clinical studies suggest that obesity has less effect on the diagnostic accuracy of NT-proBNP, some decrease in sensitivity is evident with this biomarker as well. Using results from the ICON study, Bayes-Genis et al. [48] reported that there was some decline in sensitivity of NT-proBNP from 84% to 76% in lean to obese (BMI ≥ 30) patients. Interestingly, Hermann-Arnhof et al. [49] found that NT-proBNP was increased (median 356 pg/ml) in severely obese patients (BMI > 40) without known cardiac disease compared to a median of 289 pg/ml in stable NYHA Class I patients, suggesting this biomarker may be useful as an indicator of potential cardiovascular disease in very overweight patients.

The link between obesity and low NP levels is not yet fully understood. It is postulated that the inverse relationship observed between BMI and the BNP concentration may be due to increased expression of NP clearance c-receptor by adipose tissue, resulting in increased removal of BNP in obese subjects [33]. However, the association between higher BMI and lower NT-proBNP levels suggests that other mechanisms must be considered as NT-proBNP is an inactive peptide unlikely to be actively cleared by specific receptors [48].

Diabetes mellitus

Diabetes is a very common co-morbidity in acute heart failure. The Acute Decompensated Heart Failure National Registry (ADHERE) showed that 44% of all the patients hospitalized because of worsening heart failure have diabetes [5]. NP levels are increased in diabetics even in the absence of structural heart disease [50]. The PRIDE substudy suggests that NT-pro-BNP levels may also be slightly higher in dyspneic patients with diabetes mellitus in the absence of acute heart failure [51]. Diabetes was common among those enrolled (26.2%) and this characteristic was an independent predictor that acute heart failure was present. When heart failure was present, the median concentrations of NT-proBNP were similar in patients with and without diabetes (4,784 pg/ml vs. 3,382 pg/ml, respectively, P = 0.93). Among patients who had shortness of breath due to another cause, NT-proBNP was elevated in diabetics compared to non-diabetics in dyspneic subjects (242 pg/ml vs. 115 pg/ml, P = 0.01). However, this difference was no longer evident after considering factors known to affect NT-proBNP concentration, such as impaired renal function. Importantly, NT-proBNP remained highly sensitive and specific for the evaluation of dyspnea in diabetics when using the same age-adjusted cut-points that have been established for the population at large. Data from the Breathing Not Properly Multinational Trial also found that the BNP concentrations were identical in diabetic versus non diabetic subjects after the parameters known to affect BNP concentrations (i.e., age, sex, BMI, and renal function) were matched suggesting that diabetes is not a confounding variable to be considered when interpreting BNP concentrations in patients who present with acute dyspnea [52].

Renal disease

Among clinical conditions outside of cardiovascular disease, renal dysfunction has one of the most important and greatest independent effects on NP levels. Renal function definitely needs to be carefully considered during the clinical interpretation of NP levels. Blood levels of NPs are directly and closely correlated with kidney function even in the absence of heart failure with levels rising as renal performance declines [53]. In addition, at any given level of renal function, BNP and NT-proBNP levels tend to be higher in patients with abnormal ventricular function. In the case of BNP, clearance from the body is complicated [54]. There are specific clearance receptors (NP c-receptor) that are largely responsible for removal of BNP (Fig. 5). Neuropeptidases, in the plasma and the renal tubular epithelium, also degrade BNP to a modest degree. In addition, BNP is cleared by the kidney. In contrast, NT-proBNP is primarily cleared by the kidneys. This would suggest that this NP may be more affected by renal disease, but results discussed below do not support this conclusion. Nevertheless, these clearance pathways indicate that blood levels of BNP and NT-proBNP will be increased when renal impairment is present [55].
https://static-content.springer.com/image/art%3A10.1007%2Fs10741-009-9141-2/MediaObjects/10741_2009_9141_Fig5_HTML.gif
Fig. 5

Proposed renal mechanisms to account for increased levels of BNP in patients with chronic kidney disease [54]. NP natriuretic peptide receptors located on the renal tubular cells, GTP guanine triphosphate, cGMP cyclic guanine monophosphate

In the extreme case of end-stage renal disease (ESRD) where endogenous renal function is essentially absent, BNP and NT-proBNP levels are significantly increased. The picture is further complicated by the common occurrence of cardiovascular disease in patients with ESRD and patients with less renal impairment in the form of chronic kidney disease (CKD). Given that NPs are influenced by ventricular mass and cardiac function, both of which may be disturbed in ESRD and CKD, cardiac and renal factors both contribute to elevated levels of these biomarkers [56]. It is well known that left ventricular mass increases with progression of renal disease and that there is a high prevalence of left ventricular dysfunction in patients with advanced renal failure [57, 58]. These considerations help explain why NPs also reflect the prognosis of patients with CKD [59].

A number of studies have considered the optimal use of NPs in the diagnosis of acute dyspnea when renal dysfunction is present. First, in a substudy of the Breathing Not Properly Multinational study, McCullough et al. [38] have clearly shown that heart failure is more common among patients with dyspnea who also have advanced CKD. They found that BNP level was strongly and independently associated with heart failure when holding all other predictors equal. Second, they found that the optimal cutoff for BNP to aid in the diagnosis of heart failure in dyspneic patients was significantly influenced by the degree of renal dysfunction present. Estimated GFR (eGFR) was available in 1,452 of the 1,586 study patients (91.6%). Patients with an eGFR <15 ml/min/1.73 m2 or who were on dialysis therapy were excluded from the analysis. The following raw and log–log transformed correlations were found between BNP and eGFR values, r = −0.19 and r = −0.17, respectively, for those with heart failure and r = −0.20 and r = −0.31, respectively, for those without heart failure (both P < 0.0001). Patients were divided into eGFR groups and mean BNP levels were reported. For patients with a final diagnosis of heart failure, those with eGFR (ml/min/1.73 m2) ≥90 had a mean BNP of 562 pg/ml, 60–89 the mean BNP was 648 pg/ml, 30–59 the mean BNP was 746 pg/ml, and <30 the mean BNP was 851 pg/ml. For patients without a final diagnosis of heart failure, the mean BNP levels in these eGFR groups were 85, 132, 297, and 285 pg/ml, respectively. The areas under the receiver-operating characteristic curve and optimum cut-points for BNP was 0.91 (71 pg/ml), 0.90 (104 pg/ml), 0.81 (201 pg/ml), and 0.86 (225 pg/ml) for these same eGFR groups, respectively. These investigators concluded that renal function correlated sufficiently with BNP to influence the optimal cut-point for this biomarker in the diagnosis of acute dyspnea, particularly when eGFR was <60 ml/min/1.73 m2.

NT-proBNP levels are also influenced by renal function. deFilippi et al. [60] studied the influence of renal function on the interpretation of NT-proBNP values used in the differential diagnosis of patients with acute dyspnea. They found that the proportional increase in NT-proBNP was similar to that for BNP in patients with renal insufficiency. They reported data on median NT-proBNP values seen in patients with dyspnea, with and without heart failure at different levels of renal function. In patients with heart failure compared to those without heart failure (both with impaired renal function, eGFR <60), the median NT-proBNP was 5,305 pg/ml compared to 1,331 pg/ml (P = 0.001). In patients with heart failure versus those without heart failure (both without significant renal dysfunction, eGFR ≥60), NT-proBNP was 2,916 pg/ml versus 451 pg/ml (P = 0.001). Despite this relationship with renal function, no routine adjustment in the age-adjusted cutoffs for NT-proBNP is needed in patients presenting with dyspnea [32]. Age acts as an adjustment for renal dysfunction, but for younger patients <50 years old, who have severe CKD, a cut-point of 1,200 pg/ml for NT-proBNP may be more reasonable [32].

Van Kimmenade et al., using results from the ICON study, found that NT-proBNP and eGFR were each significantly related to mortality and combining both led to proper identification of heart failure patients at the highest risk for early death after the diagnosis of acute heart failure. The combination of NT-proBNP and eGFR revealed that heart failure patients with moderate or worse renal insufficiency but lower NT-proBNP concentrations had 60-day outcomes comparable to those without significant renal insufficiency. Importantly, the mortality risk associated with worsening creatinine on presentation was only significant when NT-proBNP concentrations were elevated [61]. The combined use of objective parameters of cardiac (NT-proBNP) and renal (eGFR) functions allowed identification of heart failure patients at highest risk for early mortality and argues that elevations of NT-proBNP in the setting of renal insufficiency are still prognostic. Nevertheless, these findings point out that NP values should not be interpreted in isolation and should be integrated with other findings in the diagnostic evaluation.

Elevated NPs in other disease states

The ready availability of NP determinations has led to their assay in a wide variety of conditions, sometimes with unexpected elevations being detected. Although not usually a problem in differential diagnosis, NT-proBNP is emerging as potentially useful as a marker of prognosis in these conditions. Recently, investigators studied the prognostic role of NT-proBNP in patients with pulmonary vascular diseases. Suntharalingam et al. [62] studied the prognostic role of NT-proBNP in patients with chronic thromboembolic pulmonary hypertension, who were undergoing endarterectomy for this condition. They found that pre-operative levels of NT-proBNP predicted survival to 3 months post-op. Interestingly, perioperative declines in NT-proBNP levels in survivors correlated with improvements in pulmonary vascular resistance and 6-min walk test results at 3 months. Reesink et al. [63] reported that levels of NT-proBNP showed a high correlation with hemodynamic parameters, such as pulmonary vascular resistance. These results are encouraging, but the role of NT-proBNP in predicting long-term survival still needs to be addressed.

Natriuretic peptides have also been shown to be elevated in patients with septic shock and to correlate with prognosis [64]. NP levels were found to be markedly elevated in one study of patients in septic shock with elevation comparable to those seen in acute heart failure [65]. In this study, the 24 patients with severe sepsis had a median (range) BNP level of 572 pg/ml (13–1,300) and a median NT-proBNP level of 6,526 (198–70,000) pg/ml. These median values were similar to those of the 51 patients in the study who had acute heart failure with BNP 581 pg/ml (6–1,300) and NT-proBNP 4,300 pg/ml (126–70,000). Kandil et al. [66] also found that patients with septic shock on admission had significantly higher BNP levels compared to patients in early sepsis without a final diagnosis or a control group. Interestingly, BNP levels were not significantly elevated in patients with early sepsis. Follow-up showed NP levels were positively correlated with the extent of organ failure and predicted outcome. The work of Vila et al. [67] demonstrates that part of the elevation in NPs is related directly to the inflammatory state of septic shock not just to changes in cardiac function. They administered Escherichia coli endotoxin (2 ng/kg) to 10 healthy volunteers in a randomized, placebo-controlled, cross-over design. Plasma NT-proBNP reached peak values after 6 h (41 ± 7.9 pg/ml vs. 16 ± 3.2 pg/ml on placebo days, P = 0.023). Changes in NT-proBNP correlated with changes in body temperature (P < 0.001), heart rate (P = 0.005), C-reactive protein (P < 0.001), but not to blood pressure. Clearance of NPs is also diminished in the septic state, which contributes to the elevations seen in these patients [34].

Prognosis in dyspneic or critically ill patients

Interestingly, NT-proBNP has been found to predict outcome in patients with dyspnea whether due to acute heart failure or not [68]. The NT-proBNP cutoff for increased mortality was 1,000 pg/ml in both groups. Surprisingly, the mortality rate at 1 year in patients with NT-proBNP above this value was higher in patients without heart failure.

Natriuretic peptides are often elevated in patients cared for in intensive-care units so their diagnostic interpretation needs to undertaken carefully in this setting [69]. The prognostic role of NT-proBNP has also been tested in an unselected cohort of critically ill patients hospitalized in an academic intensive care unit with a variety of illnesses including acute heart failure [70]. Survivors had significantly lower NT-proBNP values compared with non-survivors (3,394 pg/ml vs. 6,776 pg/ml), suggesting that NT-proBNP may be an independent prognostic marker of outcome in critically ill patients. NT-proBNP levels were not significantly different in patients with a primary diagnosis of heart failure compared with the critically ill non-cardiac patients. Also suggesting that a single measurement of NT-proBNP might facilitate triage of emergency and intensive care unit patients [16].

Monitoring clinical response in AHFS

Despite current therapy, close to half of all the patients with AHFS remain symptomatic at discharge and a third is readmitted within 6 months [71]. These dismal statistics illustrate the limitations of standard therapy, but also suggest that current methods of assessing patient status and the application of available therapies in patients with AHFS are suboptimal. The high rate of re-admission indicates that the present criteria for discharge, typically based mostly on subjective impressions, correlates poorly with clinical stabilization. Traditional parameters, including changes in body weight, seem to be inaccurate predictors of adequate volume status [72]. Timing of discharge in AHFS continues to be determined primarily by subjective assessment of improvement in signs and symptoms of congestion and tolerance of desired medication changes. In the last 5 years, a number of studies have explored the most appropriate clinical indicator, laboratory parameter or a combination thereof, to determine adequacy of therapy and readiness for discharge. Evidence supporting the use of NPs to help guide the timing of discharge will be explored below.

Results from the ADHERE study indicate the prognostic importance of NPs for in-hospital mortality [73]. This analysis was based on a subset of the overall study population (48,629 or 63% of the 77,467 hospitalization episodes recorded in the protocol) who had BNP levels determined within 24 h of presentation. In-hospital mortality was related to the following quartiles (Q) of BNP (pg/ml): Q1 (<430), Q2 (430–839), Q3 (840–1,729), and Q4 (≥1,730). There was a strong direct relationship between BNP quartiles and in-hospital mortality: Q1 (1.9%), Q2 (2.8%), Q3 (3.8%), and Q4 (6.0%), P < 0.0001 which persisted after adjustment for a number of important clinical characteristics including age, gender, systolic blood pressure, blood urea nitrogen, creatinine, sodium, pulse, and dyspnea at rest. The adjusted odds ratio for Q4 versus Q1 from this analysis was 2.23 (95% confidence interval 1.91–2.62, P < 0.0001).

The work of Johnson et al. [74] suggests interventions to improve hemodynamic status may acutely reduce a variety of markers of neurohormonal activation in patients hospitalized with severe hemodynamic compromise. NPs have been shown to improve with standard therapy for AHFS, as reflected by the significant decline in biomarker levels from admission to discharge noted in numerous studies (Table 4). Results from the Evaluation study of Congestive Heart Failure and Pulmonary Artery Catheterization Effectiveness (ESCAPE) study are typical of these findings [75]. In this trial, patients with severe AHFS were randomly assigned to receive therapy guided by clinical assessment or by pulmonary artery catheterization. In a subset of study patients reported by Shah et al., serial BNP levels showed a significant, but modest correlation between BNP levels and hemodynamics suggesting that this biomarker is not an instantaneous surrogate for cardiac filling pressures. Specifically, there was a significant correlation between initial BNP and right atrial pressure (r = 0.47, P = 0.005) and pulmonary capillary wedge pressure (r = 0.54, P = 0.001) measured at baseline.
Table 4

BNP and NT-proBNP results at admission and discharge in patients with AHFS

Study

Year

Study size

Type HF

NP

Assay

Median NP (pg/ml) admit

Median NP (pg/ml) discharge

ADHERE [73]

2007

48,629

LVD, PSF

BNP

N/A

840

N/A

ESCAPE [75]

2007

141

LVD

BNP

Biosite

783

483

Valle R et al. [76]

2008

186

LVD, PSF

BNP

Biosite

716

404

Hogenhuis et al. [77]

2006

601

LVD, PSF

BNP

Biosite

N/A

492

Bettencourt et al. [78]

2004

182

LVD, PSF

NT-proBNP

Roche

6,778

4,137

Verdiani et al. [79]

2008

120

LVD, PSF

NT-proBNP

Roche

10,912a

4,701

Cournot et al. [80]

2008

157

LVD, PSF

BNP

Abbott

1,057

657

Wasywich et al. [83]

2008

24

LVD, PSF

BNP

Biosite

721

508

Verdiani et al. [84]

2005

100

LVD, PSF

BNP

Shionogi

739

414

Feola M et al. [85]

2008

250

LVD, PSF

BNP

Biosite

N/A

643a

Abbreviations as in Table 2

aMean

The median (25th percentile, 75th percentile) BNP level of substudy patients was 783 pg/ml (329, 1,565) at baseline and declined to 468 pg/ml (240, 946) at discharge. Thus, BNP did decrease significantly during the entire hospitalization, approximately 1 week for most patients, even in this severely ill patient population. These results suggest in-hospital changes in this biomarker may aid in monitoring response to therapy [75]. Initial BNP level was strongly associated with outcome. Patients with baseline BNP levels >1,500 pg/ml had greater mortality at 6 months and almost twice the length of stay as patients with BNP levels <500 pg/ml (10.1 days vs. 5.7 days, P = 0.002).

A recent study by Valle et al. [76] reviewed the predictive value of the extent and rapidity of BNP change during hospitalization and found that a rapid decline in BNP to levels <250 pg/ml suggested sufficient clinical stability for a safe discharge. This retrospective study investigated the association of changes in BNP levels during hormone-guided treatment and noninvasive assessment of fluid status with clinical course after discharge. The study enrolled a total of 186 patients admitted with AHFS. All the patients underwent serial bioelectrical impedance measurements and BNP determinations during the study. The treatment goal was to reach a BNP value of <250 pg/ml during the hospitalization as often as possible. Follow-up data demonstrated that a BNP value at discharge of <250 pg/ml was associated with an event rate of 16% at 6 months, while BNP levels >250 pg/ml were associated with a 78% risk of adverse events during the same period. Despite the common sense desirability of reducing BNP levels to a normal level (<100 pg/ml) by hospital discharge, this is very difficult to achieve and in one study only occurred in 10% of the patients [77].

Bettencourt et al. [78] performed an observational analysis of the relationship of discharge NT-proBNP and outcomes following admission for AHFS. Their study included 182 patients consecutively admitted for AHFS. The patients were classified into three groups based on the change of their discharge NT-proBNP from admission: decrease of ≥30%, increase of ≥30%, and values between these ranges. Risk was substantially lower when NT-proBNP levels fell by more than 30% of the baseline value compared to patients with no significant change or increased NT-proBNP levels. This threshold of NT-proBNP change was confirmed in the recent study by Verdiani et al. [79] who also found that patients discharged with NT-proBNP reductions of <30% from admission levels were at increased risk. In this study, these patients had more than twice the risk of cardiovascular death or rehospitalization due to heart failure exacerbation compared to patients with a reduction of ≥30%. Patients with NT-proBNP percent reductions of <15% were found to be at very high risk of death due to cardiovascular causes. A prospective study on elderly patients by Cournot et al. [80] also showed that both pre-discharge BNP and the change in BNP between admission and discharge were strong markers of outcome. These investigators found that the optimal cut-point for BNP at discharge was 360 pg/ml and the optimal cut-point for change was 50% reduction in admission BNP. These criteria were found to provide complementary information with the highest event rate observed in patients who had a high BNP level at discharge (>360 pg/ml) and a failure of BNP to decline by more than 50% during the hospitalization, while the lowest event rate during follow-up occurred in patients who met both of these cut-points for improvement.

Additional research is needed to fully define the optimal utilization of NPs during hospitalization. Tracking both discharge NP levels and the change in NP levels during hospitalization may enable the timing of the patient’s discharge to be optimized. The ability of NPs to identify heart failure patients at higher risk for adverse outcomes after discharge suggests that tailoring medical therapy to BNP or NT-proBNP levels during hospitalization may be desirable. Integrating this approach into the routine assessment of patients with AHFS could allow clinicians to more accurately identify high-risk patients who may derive the most benefit from intensive in-hospital management strategies [76].

Individual change in NPs

Declines in NP levels, as filling pressures are reduced, lend support to the evolving concept that these biomarkers may be useful in evaluating a patient’s hemodynamic response to therapeutic interventions [81]. However, when considering therapeutic adjustments in an individual patient, the expected spontaneous variation of these biomarkers must be taken into consideration. In the outpatient environment, there is a considerable intra-individual variability from test to test in these peptides even in stable heart failure patients. Week-to-week changes of approximately 50% and 66% for NT-proBNP and BNP, respectively, are needed to exceed expected assay variation and to indicate an altered clinical status [82]. Since all the patients are treated during hospitalization, spontaneous changes are difficult to quantify. Due to relief of congestion, NPs would be expected to decline, and measurements are generally made more closely in time than in the outpatient setting. Moreover, the higher levels often observed during hospitalization may not vary as much spontaneously, on a percentage basis, as the lower levels frequently observed in the outpatient clinic. Until evidence from randomized clinical trials is available, a reasonable approach would be to administer treatment in order to achieve a 30% reduction in initial NP concentration or to a target level, in the case of NT-proBNP <1,000 pg/ml, and BNP <250 to 350 pg/ml [7480, 8385].

Cost effectiveness of NP determination in acute heart failure

Much has been written in the last decade advocating pharmacotherapy and multidisciplinary disease management as key strategies in the care of patients with heart failure [86]. Pay for performance has been implemented to encourage multidisciplinary approaches to improving the quality of care with incentives paid to top performers [87, 88]. The drive to incorporate biomarkers in day-to-day management is partly due to the wave of quality initiatives mandated by regulatory agencies, third party payers, or health plans. Measurement of NPs on presentation, in addition to clinical judgment, appears to be a cost-effective approach that allows the practitioner to accurately diagnose or rule out heart failure and predict adverse clinical outcomes, including death. Hospitals and clinicians embrace and implement quality improvement programs with appreciation of the cost and time commitment required. The concept of differential payment for the delivery of high-quality care is increasingly common in the healthcare landscape, especially in heart failure. Encouraging outcomes in research studies on the cost effectiveness of NPs provide further support for healthcare payers to pursue this avenue.

A number of studies, including two prospective randomized trials discussed below, demonstrate that the routine use of NP testing in the diagnostic assessment and management of dyspneic patients appears to be cost effective. Costs were lower in patients managed with NP determinations due to fewer hospital days during follow-up [89, 90], shorter initial lengths of stay [21, 91, 92], and the performance of fewer echocardiograms [89]. The impact of this approach on mortality is difficult to assess at present. Current studies have shown no survival benefit, but samples sizes and event rates to date have been too modest to rule out a clinically important benefit [8992].

BASEL study

The B-type natriuretic peptide for Acute Shortness of Breath Evaluation (BASEL) study was a prospective, randomized, controlled, single-blind investigation designed to test the hypothesis that knowledge of the BNP level at the time of admission would improve the evaluation and treatment of patients presenting with acute dyspnea [17]. Knowledge of the BNP level reduced the need for hospitalization (75% vs. 85%, P = 0.0008) and admission to the intensive care unit (15% vs. 24%, P = 0.01). The median time to discharge was 8.0 days in patients who had BNP values available compared to 11 days in the control group (P = 0.001). As expected from these differences, the mean cost of treatment was lower in the study group. The rate of death in the two groups was similar at 30 days. This study provides strong evidence that the rapid determination of BNP in the emergency department results in improved evaluation and treatment of patients with acute dyspnea [17].

IMPROVE-CHF study

The addition of NP testing to the traditional clinical evaluation of patients with dyspnea reduced the duration of emergency department visits in a Canadian multicenter study [21]. This trial randomized 500 adult patients with dyspnea believed to be of cardiac origin to standard clinical evaluation with or without the addition of NT-proBNP testing. In the biomarker group, NT-proBNP measurements were obtained in the emergency department to guide initial management and after 72 h in patients who were admitted to help guide subsequent management. NT-proBNP testing was associated with reduced duration of emergency department visits (from 6.3 to 5.6 h, P = 0.031) and lower 60-day rehospitalization rates (reduced by 35% from 51 to 33 events, P = 0.046). In contrast, initial hospital length of stay, initial intensive care unit admissions, and 60-day mortality rates were not statistically different between the study groups. The frequency of additional outpatient testing, including echocardiography, radionuclide ventriculography, and chest CT, was lower in the NT-proBNP group. Total direct medical costs throughout follow-up were significantly lower when biomarker testing was employed [21].

Discussion

Natriuretic peptides have emerged as major adjuncts to clinical judgment in the differential diagnosis of patients presenting with dyspnea [93]. Major studies have substantiated the added value of NT-proBNP and BNP assessment beyond standard clinical evaluation in patients seen acutely for suspected AHFS. Additional work indicates that determination of these NPs assists in the initial triage of patients seen in the emergency department or in urgent care settings. NT-proBNP and BNP are also powerful tools for risk stratification of patients with AHFS. BNP and NT-proBNP levels on admission and during hospitalization added significant prognostic information, even after adjusting for clinical variables that are routinely obtained during admission for heart failure. NPs measured during hospitalization for acute heart failure have been found in many studies to independently predict increased risk of cardiovascular mortality and all-cause mortality in AHFS, both during the initial admission and post-discharge. NP levels at discharge are an even stronger predictor of clinical outcomes in this population compared with baseline levels.

These observational data on the ability of NPs to identify patients, with AHFS and chronic heart failure, who are at higher risk for adverse outcomes support the conduct of large-scale, randomized outcome trials of tailoring medical therapy based on these biomarkers [9496]. Prospective studies are still needed to verify the utility of specific NPs and to identify the target levels and other details of monitoring necessary for the practical application of biomarker guidance. Meanwhile, available data suggest that clinicians should strive for pre-discharge BNP levels of <250–350 pg/ml and NT-proBNP levels of <1,000 pg/ml. We advocate closer outpatient monitoring in patients discharged with higher than target NP levels as persistent elevation of this degree correlates with adverse outcomes, including death and rehospitalization.

Conclusions

NPs provide incremental value to clinical judgment for diagnosing heart failure especially in the acute setting where timely diagnosis is both important and difficult. Incorporation of NP measurements in evaluating and managing patients with dyspnea in the hospital provides powerful information on prognosis and optimal allocation of resources. Observational data suggest NP levels may help set targets for individualized treatment, but large-scale randomized trials are needed to confirm and extend promising initial results. The judicious use of BNP and NT-proBNP measurements, in the setting of thorough and accurate clinical assessment and reasoning, has the potential to improve the management of patients with AHFS [97100].

Copyright information

© Springer Science+Business Media, LLC 2009