Extremely High Brain Natriuretic Peptide Does Not Reflect the Severity of Heart Failure
Maya Guglin, MD, PhD, University of South Florida, 2 Tampa General Circle, Suite 5074, Tampa, FL 33606 E-mail: email@example.com
Brain natriuretic peptide (BNP) is important in the diagnosis and management of heart failure (HF). Sometimes, very high BNP levels encountered in clinical settings seem to be out of proportion to the severity of HF. The authors retrospectively identified 113 patients with 129 admissions with a BNP value >3000 pg/mL regardless of diagnosis. The data set was analyzed using the Student t test and bivariate analysis. Fewer than half of patients were admitted for HF. In 14 patients (10.9%), no signs of HF were found. The BNP level of those with and without HF was similar. There was no difference in BNP level in patients with and without systolic dysfunction or renal dysfunction and between different age groups. Extreme values of BNP do not necessarily correlate with the presence of HF, cardiomyopathy, or kidney dysfunction. When the magnitude of BNP elevation is very high, its clinical significance is limited. Congest Heart Fail. 2010;16:221–225. © 2010 Wiley Periodicals, Inc.
In 2008, it was estimated that heart failure (HF) accounted for nearly 1 million hospital admissions per year in the United States and accounted for a cost of $38.4 billion per year.1 In an aging population with increasingly prevalent morbid risk factors including diabetes and hypertension, the incidence of HF will continue to rise. The advent of serum assays for brain natriuretic peptide (BNP), which is a 32–amino acid neurohormone produced and released by the cardiac ventricles in response to pressure overload and ventricular dilation, has aided in the diagnosis of HF in patients presenting with dyspnea.2–6 Because BNP reflects ventricular wall stress and dilation, elevations in BNP have significant implications with regard to adverse clinical events.7–10 Numerous studies have been published indicating a significant correlation between BNP and clinical severity of HF in terms of New York Heart Association (NYHA) functional class.10 As such, BNP-guided HF therapy has further added to its clinical popularity.11
Surprisingly, some patients exhibit extreme degrees of BNP elevation but no clinical signs of HF are present, although such patients are rare. In 2007, we published our observations demonstrating that BNP levels correlate well with clinically determined severity of HF or left ventricular ejection fraction (LVEF) when they were mildly or moderately elevated, but there was no correlation when BNP reached very high values, >4000 pg/mL.12
In the present study, we collected a unique sample of patients with extremely elevated BNP values ≥3000 pg/mL and subsequently analyzed the data obtained.
Materials and Methods
Study Population. The study was approved by the institutional review boards of Tampa General Hospital and H. Lee Moffitt Cancer Center. We retrospectively screened laboratory data from both hospitals for BNP values ≥3000 pg/mL regardless of initial presentation and diagnosis. The cutoff BNP, which is 30 times the upper limit of normal, was chosen arbitrarily. We identified 139 values representing 129 admissions for 101 patients. Some patients were admitted more than once for different reasons, and some had BNP measured more than once during the same admission. From the values >3000 pg/mL measured during the same hospital admission, we calculated the average number, so that a single BNP value was recorded for each hospital admission. Each admission was analyzed separately.
The sample size comprised 129 hospital admissions for 113 patients hospitalized at 2 sites from 2007 to 2008. Their electronic medical records were analyzed for history and physical examination findings, progress notes, discharge summaries, and the data of instrumental and laboratory diagnostic tests.
Methods. The following laboratory parameters were determined at baseline: complete blood count, comprehensive metabolic profile, blood pressure, heart rate, and chest radiography. Values of right ventricular systolic pressure and LVEF obtained via echocardiography were also recorded. When LVEF was reported as a range, the average value was calculated. Documented physical examinations were reviewed for each patient, and clinical evidence of HF was determined by utilizing the Framingham criteria, which includes signs of jugular venous distention, pulmonary rales, auscultation of S3, and lower extremity edema. History of initial presentation of symptoms and clinical characteristics on physical examination in conjunction with evidence of pulmonary congestion on chest radiography were used to make the diagnosis of HF. Because categorizing HF into NYHA class was problematic, as the majority of the patients did not have any history of HF and the rest had symptoms or objective evidence of congestion, such as pulmonary edema at rest corresponding to NYHA class IV, we abandoned the use of NYHA classification. In both institutions, BNP was measured using Beckman UniCel DxI assay (Beckman Coulter, Inc, Brea, CA) with the upper normal range as 100 pg/mL.
Each admission episode was categorized into presence or absence of signs or symptoms of HF, presence or absence of cardiomyopathy based on an LVEF <50% or ≥50%, respectively, degree of pulmonary congestion on chest radiography on the day of BNP testing or the closest day (0, no congestion; 1, mild congestion; 2, pulmonary edema). Serum creatinine was recorded on the same day that BNP exceeded 3000 pg/mL. This value was then used as a contin-uous variable for analysis of variance test and categorized into normal (≤1.2 mg/dL) or abnormal (>1.2 mg/dL) for comparison of mean BNP in patients with and without renal dysfunction. The dataset was analyzed with Student t test comparing mean BNP in those with HF vs those without HF, those with pulmonary edema vs mild congestion vs no pulmonary congestion, renal dysfunction vs no renal dysfunction, and cardiomyopathy vs no cardiomyopathy and by bivariate analysis. Statistical significance was defined as P<.05. Statistical analysis was performed with SPSS (version 17.0; SPSS, Inc, Chicago, IL)
A total of 113 patients with 129 admissions, including 54 women and 59 men with an age ranging from 21 to 102, were included in the present study. In 56 (43.4%) of the 129 cases, the reason for admission was new-onset HF or HF exacerbation, and in 73 (56.6%) the reason was unrelated to HF.
In 14 (10.9%) of the 73 admissions admitted for reasons other than HF, no signs of HF or volume overload throughout the admission were found, while 115 (89.1%) cases showed some evidence of transient episodes of volume overload. Mean BNP in those with HF was not statistically different from those without HF (Table).
Table Table. BNP in Different Clinical Scenarios
|HF present||4113.0±1150.8||HF absent||4168.9±846.0||NS|
|Pulmonary edema||4038.7±1108.0||Minimal congestion||4374.8±1370.1||NS|
|Normal creatinine||4331.7±1640.9||Elevated creatinine||4048.21±940.9||NS|
|Ejection fraction <50%||4204.2±1215.2||Ejection fraction ≥50%||3881.43±756.0||NS|
|Age ≤50 years old||4265.0±1276.6||Age >70 years ||3844.6±950.7||NS|
|Pulmonary artery systolic pressure >35 mm Hg||4086.2±830.2||Pulmonary artery systolic pressure ≤35 mm Hg||4494.1±1521.5||NS|
|Ejection fraction <50% and elevated creatinine||4148.7±1015.3||Ejection fraction <50% and normal creatinine||4430.3±1723.7||NS|
|Ejection fraction ≥50% and elevated creatinine||3809.1±692.0||Ejection fraction ≥50% and normal creatinine||3576.1±96.6||NS|
|Ejection fraction <50% and elevated creatinine and pulmonary edema||4041.1±1136.5||Ejection fraction <50% and elevated creatinine and no congestion||4266.1±866.4||NS|
Of the 14 patients who were admitted without any signs or symptoms of HF, 11 had some form of renal dysfunction. Seven of the 14 were admitted with an underlying infection, 1 with portal hypertension, and 2 with central nervous system pathology, which included cerebrovascular accident and metabolic encephalopathy. Of interest, one of the acute renal failure admissions presented with hypovolemia and gastrointestinal bleeding. Infections included mycobacterium avis pneumonia, endocarditis (n=2), staphylococcus bacteremia, ventriculoperitoneal shunt infection, sepsis, and a line-related infection in a patient with human immunodeficiency virus.
In 81 (62.8%) cases, LVEF was <50%. There was no significant difference in the BNP value between those with normal systolic function vs those with reduced ejection fraction.
Of the 129 admissions, creatinine was found to be >1.2 mg/dL in 98 (76.0%) cases. The mean BNP of those with a normal creatinine was 4331.7±1640.0 pg/mL and the BNP of those with an elevated creatinine was 4048.2±940.0 pg/mL. This difference was not statistically different.
A combination of decreased ejection fraction and elevated creatinine was found in 58 (45%) of cases. Only 3 cases (2.3%) were found in which both systolic function and renal function were normal.
Plain chest radiographic findings suggestive of pulmonary edema were observed in 44 patients (34.1%). BNP of those patients with and without pulmonary edema was 4038.7±1108.0 and 4374.8±1370.0 pg/mL, respectively; this difference was not statistically significant. Elevated pulmonary artery systolic pressure, estimated from echocardiography by maximal velocity of tricuspid regurgitation, also was not associated with higher BNP (Table).
From our findings, as evidenced in Table, the BNP level was similar regardless of the presence of HF, renal failure, and severity of pulmonary congestion.
Using bivariate analysis, age, the presence of HF, decreased systolic function, elevated creatinine, and evidence of pulmonary congestion by chest radiography were not associated with higher values of BNP. BNP values in different age groups were also found to be similar (Table).
Curiously, one patient who was admitted twice, once for HF exacerbation and on another occasion for dehydration, was found on both admissions to have a BNP >3000 pg/mL.
It is well-established that natriuretic peptides such as BNP and N-terminal prohormone brain natriuretic peptide have become available as useful tools for both the diagnosis of HF and prognostic stratification of patients with HF.13,14 According to the Heart Failure Society of America (2006) and the American Heart Association (2005), BNP levels >100 pg/mL in conjunction with symptoms of dyspnea are predictive of an acute exacerbation of HF.5 A number of studies have published a significant correlation between BNP and clinical severity of HF in terms of NYHA class.15–19 However, in most cases, the BNP remained within 1000 to 2000 pg/mL.
In a recent paper published by Guglin and colleagues,12 patients were stratified into 3 groups based on whether their admission BNP was between 500 and 1000 pg/mL (mildly elevated), 2000 and 3000 pg/mL (moderate elevated), or 4000 and 20,000 pg/mL (highly elevated). Patients found in the mildly elevated and highly elevated BNP cohorts were found to exhibit similar LVEF as well as similar clinical presentations. However, patients with a BNP of 2000 to 3000 pg/mL showed no statistically significant differences in terms of clinical presentation and LVEF when compared with those patients with a BNP in excess of 4000 pg/mL.12
In our paper, we present a unique data set of patients with extremely high BNP elevation with the arbitrary cutoff at 30 times the upper normal range, measured upon admission to the hospital. In 11% of cases, we were unable to find any signs of HF or volume overload throughout the admission. In the majority of cases, either serum creatinine was elevated (76%) or LVEF was reduced (63%) or both (45%), and only in 2% of admissions were both renal and left ventricular systolic function normal.
Several papers have discussed a correlation between mild to moderate renal dysfunction and elevation of BNP values.20,21 Although no linear relationship has been shown, it has been suggested that in patients with renal failure, a correction factor based on glomerular filtration rate may benefit BNP evaluation. While BNP is not completely dependent upon renal excretion, eventual saturation of up-regulated peripheral clearance receptors could provide a possible calculable etiology, thus making a linear correlation between markedly elevated BNP values and renal dysfunction plausible. Again, in our sample of extremely high BNP levels there was no statistical association between patients with and without renal failure.
In addition, BNP typically increases with age.8 In our data set, there was no difference between BNP levels in younger and older patients, with a BNP of 4265.0±1276.6 pg/mL in patients younger than 50 years old, 4245.1± 1131.0 pg/mL in patients 50 to 70 years old, and 3844.6±950.7 pg/mL in patients older than 70. Furthermore, serum BNP is usually inversely proportionate to LVEF.16 However, in our study, BNP levels were not found to be statistically different when comparing those patients with an LVEF of <50% to those with a normal ejection fraction. No association was found between LVEF, creatinine, severity of pulmonary congestion by chest radiography, or presence and absence of HF and BNP values. Therefore, we deduce that there must be factors other than vascular congestion that contribute to such elevations in BNP.
BNP can be elevated because of other confounders not related to fluid overload. Neurologic disorders and sepsis are the most commonly cited reasons for BNP elevations not associated with HF. Meaudre and colleagues22 found that BNP levels did not correlate with left ventricular filling pressure as estimated by echocardiography in patients admitted to the hospital with subarachnoid hemorrhage. Similar results were reported by Burjonroppa and colleagues23 in a retrospective trial in patients with cancer who were admitted to MD Anderson Cancer Center, Houston, Texas, in 2003 with an elevated BNP value >1000 pg/mL. They reported no association between the level of BNP elevation and the presence of volume overload or left ventricular dysfunction. It has been thought that in such cases, there is an increase in sympathetic stimulation causing the hypothalamus to secrete an excess of vasopressin, which causes fluid retention in the kidneys.
Furthermore, several studies have been published to support the theory that the immune system plays a significant part in falsely elevating BNP levels.24 McLean25 demonstrated that in 13 patients with septic shock, BNP was found to be elevated, with a mean BNP value of 849 pg/mL, yet no signs of HF existed. In a study preformed at MD Anderson Cancer Center, 52 of the 99 patients analyzed were found to meet criteria for sepsis, and approximately 90% of these patients had no evidence of left ventricular dysfunction or clini-cal evidence of HF.23 Because sepsis involves the up-regulation of many pro-inflammatory cytokines, there remains a possibility that these cytokines play a role in falsely elevating BNP.
Uniquely, Masters and colleagues26 looked at patients with transplanted hearts after an acute rejection episode and observed that BNP was again elevated independently of LVEF. In fact, Shaw and colleagues24 demonstrated that patients 1 month after cardiac transplant had a significant increase in BNP despite the absence of ventricular stretch or elevated ventricular pressure. The mechanism of BNP release after cardiac transplant remains an enigma, although one theory again proposes the indirect connection between cytokine release and elevation in BNP. Rudiger and colleagues27 then demonstrated, in a smaller sample size, a correlation between inflammatory markers (C-reactive protein, leukocyte count) and BNP levels, thus suggesting a possible interaction between natriuretic peptides and the systemic inflammatory response. Interleukin 1, which has been shown to be released in systemic inflammatory response, is a transcription activator of the promoter in the BNP gene. In addition, tumor necrosis factor α (TNF-α), which is secreted by CD4 and CD8 cells, macrophages, and neutrophils, independently up-regulates BNP mRNA levels. Expression of TNF-α has been found to occur in both acute allograft rejection and in the setting of sepsis.24 As such, it is very possible that elevations in BNP are likely influenced by multiple mechanisms at the molecular level, including inflammation, alloimmune activation, cardiac remodeling, and vascular injury and repair.28
Another theory challenges the contention that there are 2 forms of BNP and that there may be multiple forms of BNP. Hawkridge and colleagues29 have found that altered higher-molecular forms of BNP, which do not carry any active biologic function, such as uncleaved monomers of prohormone BNP, and noncovalently linked trimers or tetramers of prohormone BNP with high molecular mass interact with commercial BNP assays and could be another factor confounding an elevation in BNP.
Dokainish and colleagues30 discussed the lack of sensitivity of natriuretic peptides in the setting of critically ill patients, and the question of their potential prognostic utility was raised. Although there have been discrepancies within the literature regarding this role, further research is needed to evaluate this potential function.
There are several limitations to the present study. This was a retrospective and nonrandomized study in which misclassification bias was possible and difficult to quantify. In addition, the BNP value of 3000 pg/mL was an arbitrary cutoff.
When BNP reaches very high levels, exceeding 3000 pg/mL, it cannot be used as a reliable indicator of the severity of heart failure. Approximately one-quarter of patients with extreme levels of BNP elevation did not exhibit signs of decompensated HF. Although extreme BNP elevation most often results from a combination of decreased systolic function, acute fluid overload, and renal insufficiency, our study demonstrates that BNP can be equally elevated without any or all of these conditions.