What is the normal range for N-terminal pro-brain natriuretic peptide? How well does this normal range screen for cardiovascular disease? (2024)

,

Article history

Abstract

See page 2220 for the editorial comment on this article (doi:10.1093/eurheartj/ehi470)

Introduction

Cardiovascular disease is placing an increasing burden on society¹ but may be asymptomatic or misdiagnosed once symptomatic. Accordingly, to reduce this, authors have proposed establishing cardiovascular screening programmes.^2,3 Natriuretic peptides, a family of peptide hormones released into the circulation in response to increased myocardial stretch, have been mooted as potential biomarkers of cardiovascular disease, potentially underlying such screening programmes once validated.^2,3 Plasma levels of brain-natriuretic peptide (BNP) and its co-released peptide N-terminal pro-brain natriuretic peptide (NTpBNP) both increase in a variety of cardiovascular conditions,⁴ although their full screening characteristics are yet to be firmly established. Furthermore, unlike for BNP, NTpBNP cut-off values have not been established in clinical studies, an essential pre-requisite for screening programmes.

Accordingly, the study aimed to determine upper reference values for NTpBNP in ‘normal’ subjects free from cardiovascular or renal disease, to assess which factors cause raised NTpBNP levels, and to assess how well the developed NTpBNP upper reference values screen for cardiovascular disease in both general population and higher-risk subjects, those most likely to be invited to undergo screening. A partial analysis of part of this data involving 290 normal subjects has been published previously.⁵ This is the complete analysis of all attending subjects.

Methods

Subjects

Subjects (a random sample of the community as a whole) ≥45 years old were randomly selected from the practice lists of seven representative local community practices: 1392 general population subjects were invited to attend whether or not they had any cardiovascular diagnoses, and 928 high-risk subjects were invited to attend with a general practice diagnosis of ischaemic heart disease (IHD), cerebrovascular disease (CVA), peripheral vascular disease (PVD), diabetes mellitus (DM), or a history of heavy alcohol intake (≥40 units/week). Screening took place between January 2000 and December 2001.

Subject assessment

Details of the subjects collected from questionnaires and subjects' heart rate and blood pressure measurement (the average of two readings), spirometry, electrocardiography (ECG), echocardiography and venesection for fasting serum glucose, creatinine, and NTpBNP levels were used for assessment. Values assessed on spirometry included forced expiratory volume in 1 s (FEV₁), forced vital capacity (FVC), and their ratio (FEV₁%). Spirometry was defined as abnormal if FEV₁% <60% (obstructive defect), if FEV₁% <70% with FEV₁ <80% predicted (obstructive defect), and if FEV₁% ≥70% with both FEV₁ and FVC <80% predicted (restrictive defect).

Echocardiography

Two-dimensional echocardiography was performed with a SONOS 4500 machine (Philips, Eindhoven, The Netherlands) using second harmonic imaging. Left ventricular ejection fraction (LVEF) was calculated quantitatively using Simpson's apical biplane method,⁶ the average of three readings. Borderline or worse left ventricular systolic dysfunction (LVSD) was defined as LVEF <50%, significant LVSD as LVEF <40%, as per previous epidemiological surveys.^7,8 LV mass was calculated using the Devereux-modified American Society of Echocardiography equation,⁹ with left ventricular hypertrophy (LVH) defined as LV mass index >134 g/m² for men and >110 g/m² for women. Valvular regurgitation was assessed qualitatively on a five-point scale (nil or trivial, mild, mild-to-moderate, moderate, and severe). Valvular stenosis was assessed by peak pressure gradient and estimated valve area, and again ascribed the same five-point scale. Significant valve disease was taken as mild-to-moderate or worse. Diastolic parameters assessed included isovolumic relaxation time (IVRT), mitral inflow peak E wave velocity (E), peak A wave velocity (A), E/A ratio, and E wave deceleration time (E decel). Diastolic heart failure (DHF) was defined according to the European Study Group on Diastolic Heart Failure guidelines.¹⁰

Natriuretic peptides

Serum NTpBNP levels were measured on the Elecsys™ 2010 system (Roche Diagnostics, Lewes, UK).⁵ This assay is an electrochemiluminescent sandwich immunoassay using two polyclonal antibodies directed at the NTpBNP molecule. Intra-assay and inter-assay variabilities are 1.2–1.5% and 4.4–5.0%, respectively.⁵

Subject groups assessed

General population subjects were defined as all attending general population invitees. Normal subjects were defined as all attending general population subjects with no history of IHD, DM, PVD, CVA, hypertension, heart failure or loop diuretic usage, or heavy alcohol intake; whose blood pressure was <160/90 mmHg as the mean of two readings; whose estimated creatinine clearance (Cockcrof–Gault equation) was ≥60 mL/min/1.73 m²; and who had no significant valvular heart disease (VHD), LVH, DHF, LVSD (LVEF <50%) or regional wall motion abnormalities on echocardiography. High-risk subjects were defined as all attending subjects with any cardiovascular risk factors (IHD, CVA, PVD, DM, heavy alcohol intake, or hypertension). Figure1 depicts study design.

Statistical analysis

A multivariable linear regression model using forward-conditional entry into the model (P<0.05 for model entry and P>0.1 for removal from the model) was developed to predict NTpBNP levels in normal subjects and collinearity statistics were assessed. Factors assessed included age, gender, ethnicity, heart rate, blood pressure, serum creatinine, serum glucose, creatinine clearance, lung function, and weekly alcohol intake. The upper reference value for NTpBNP was defined as the 97.5th percentile of NTpBNP levels stratified by any multivariable predictors in normal subjects, with 95% confidence intervals calculated by the bootstrap method.¹¹ Raised NTpBNP levels were defined as those exceeding upper reference values. Univariate predictors of raised NTpBNP levels were calculated: categorical data by the Yates corrected χ² test or Fisher's exact test as appropriate, normally distributed continuous data by the Student's t test, and non-normally distributed continuous data by the Mann–Whitney U test. A multivariable logistic regression model using forward-conditional, backward-conditional, and forced entry into the model was then developed to predict raised NTpBNP levels in all attending subjects, initially entering all univariate predictors. Model validity was tested by Hosmer and Lemeshow goodness-of-fit, Nagelkerke R² value, and area under the receiver-operating-characteristic curve (AUC). Cardiovascular screening characteristics for the developed NTpBNP upper reference values were calculated in both general population and high-risk subjects. Receiver operating characteristic (ROC) curves, their AUCs, and best cut-off values were calculated for NTpBNP to detect cardiovascular disease. Data were analysed using Analyse-it for Microsoft Excel version 1.48 (Leeds, UK) and SPSS version 10.0.5 (Chicago, IL, USA).

Results

Subject demography

In total, 1205 subjects (52%) attended, 734 (53%) were general population invitees and 471 (51%) high-risk invitees. Of these, 1176 (98%) subjects had NTpBNP levels analysed, 1165 LVEF calculated (97%), and 1136 (94%) both tests performed. Among those, 290 general population attendees had one or more cardiovascular risk factors. Thus, 761 high-risk subjects attended in total, out of which 397 attendees fulfilled the criteria for normal subjects, 389 of whom (98%) had NTpBNP levels analysed. Figure1 depicts attendees. Subject demography is shown in Table1. There were no significant differences between attendees and non-attendees except for age, with attendees 3 years younger on average (P<0.0001).

NTpBNP normal range and upper reference values

Female gender (P<0.0001) and increasing age (P<0.0001) were the only independent predictors of increasing NTpBNP levels in normal subjects. Table2 depicts NTpBNP serum levels in normal subjects, with upper reference values in bold.

Univariate and multivariable predictors of raised NTpBNP levels

Tables3 and 4, respectively, show the univariate and multivariable predictors of raised NTpBNP serum levels in all attending subjects. For the multivariable model, identical models were seen whether forward-conditional, backward-conditional, or forced-entry methods were used, with goodness-of-fit P=0.49 suggesting a valid model and R²=0.49 and AUC=0.87 (0.84–0.89) suggesting excellent discrimination.

NTpBNP cardiovascular screening characteristics

The cardiovascular screening characteristics of NTpBNP upper reference values in predicting cardiovascular disease are shown in Table5. The AUCs and best single cut-offs for NTpBNP to detect cardiovascular disease in all attendees are shown in Table6.

Table7 shows the cumulative prevalence of underlying multivariable and then univariate risk factors in those found to have raised NTpBNP levels. Figure2 shows the prevalence of significant cardiovascular disease [LVSD, LVH, DHF, VHD, atrial fibrillation (AF), or cor pulmonale] stratified by NTpBNP levels. Of the 64 subjects with NTpBNP levels more than four times the upper reference value, 61 (95%) had significant cardiovascular disease and the other three having significant renal impairment.

Discussion

Natriuretic peptides are peptide hormones released into the circulation in response to increased myocardial stretch and wall tension, producing vasodilatation, natriuresis, and inhibition of the rennin–angiotensin and sympathetic nervous systems.⁴ BNP is co-released with NTpBNP predominantly from the left ventricle in response to such stimuli. NTpBNP (i) is extremely stable in plasma, serum, and whole blood, allowing community-based venesection,⁵ (ii) is a discerning marker of early cardiac dysfunction,¹² and (iii) can be measured by routinely available fully automated high-throughput laboratory equipment,⁵ all are pre-requisites for community-based screening. This is one of the first clinical studies to calculate upper reference values for NTpBNP and the first to assess how well these upper reference values screen for cardiovascular disease.

NTpBNP upper reference values

This study found that NTpBNP levels increase with female gender and increasing age in normal individuals, requiring age- and gender-specific cut-offs when defining upper reference values. Similar results have been seen elsewhere for NTpBNP^13–16 and BNP,^17,18 confirming these findings further.

This study found the following cut-off values: 100 pg/mL and 172 pg/mL for men aged 45–59 and 60+, respectively, and 164 pg/mL and 225 pg/mL for women aged 45–59 and 60+, respectively. Similar cut-off values were seen in unpublished blood donor data, obtained from the manufacturers, of the NTpBNP assay, finding 97.5th percentile values of 84 pg/mL for men and 178 pg/mL for women aged 40–49 years, 176 pg/mL for men and 185 pg/mL for women aged 50–59 years, and 296 pg/mL for men and women aged 60+ years old combined. The slightly higher values in 60+ years old may have occurred, as echocardiography, ECG, and clinical assessment were not performed. Hence, subjects with asymptomatic disorders would not have been excluded. Johnston etal.¹⁴ using the same NTpBNP assay found 97.5th percentile values of 184 pg/mL and 269 pg/mL for men aged 40–65 and 66–76, respectively, and 268 pg/mL and 391 pg/mL for women aged 40–65 and 66–76, respectively, again higher than in the current study. Possible explanations include the fact that attending subjects did not undergo echocardiography, so that subjects with asymptomatic cardiac structural disease were not excluded, and that attending subjects found to be hypertensive and subjects with minor ECG abnormalities were not excluded. Also, as in the current study, cohort size was small.

Mechanism of age and gender differences

A possible explanation for increased NTpBNP levels with age may be increased age-related fibrosis and thus subtle diastolic dysfunction. Thus, although normal subjects had DHF excluded, normal subjects ≥60 years old still had significantly lower E/A ratios and significantly longer E decels, and IVRTs than normal subjects <60 years old (1.0 vs. 1.2, P<0.0001; 242 vs. 230 ms, P=0.01; and 92 vs. 86 ms, P=0.001, respectively), suggesting greater impairment to diastolic filling. A second explanation may be reduced renal clearance, with the mean estimated creatinine clearance in normal subjects ≥60 years old significantly lower than in normal subjects <60 years old (74 vs. 91 mL/min/1.73 m², respectively, P<0.0001).

The mechanism underlying increased natriuretic peptide levels with female gender is unclear, although Redfield etal.¹⁸ found 21% higher BNP levels in women taking hormone replacement therapy, suggesting a role of oestrogen status.¹⁸

NTpBNP in screening for cardiovascular disease

Although interest in natriuretic peptide measurement was originally as a diagnostic tool to screen for LVSD and heart failure, it soon became apparent that many subjects with raised natriuretic peptide levels had normal systolic function and no overt heart failure,^19–22 with natriuretic peptides rather acting as general indicators of cardiac structural disease.³ The current study has confirmed this further. Only 26% of subjects with raised NTpBNP levels had borderline or worse LVSD, whereas 56% of general population subjects and 62% of high-risk subjects with raised NTpBNP levels had significant cardiovascular disease. Furthermore, 94% of subjects with raised levels had underlying multivariable risk factors and 98% had underlying univariate or multivariable risk factors, suggesting that raised NTpBNP levels represent a pathological state.

As NTpBNP levels rose further beyond their upper reference values, the proportion of subjects with significant cardiovascular conditions increased. Thus, 89% of subjects with more than three times normal levels and 95% of subjects with more than four times normal levels had significant cardiovascular disease. This is the first such assessment for NTpBNP, although similar results have been seen for BNP, confirming these results further. Logeart etal.²³ found that BNP levels >300 pg/mL (more than three times normal) independently predicted heart failure and Maisel etal.²⁴ noted that subjects with BNP levels one to four times normal may have non-cardiac diagnoses, as in the current study for NTpBNP.

The current study further assessed which factors predict raised NTpBNP levels, finding LVSD, LVH, AF, VHD, IHD, abnormal spirometry, increased alcohol intake and PVD, a condition frequently associated with silent IHD and LVSD²⁵ to be multivariable predictors, and DHF and hypertension to be univariate predictors. All of these conditions could produce subclinical changes in left ventricular structure and function triggering NTpBNP release. The other multivariable predictors were renal dysfunction, increasing NTpBNP levels by reducing renal clearance; increased age, acting via the mechanisms previously discussed; and reduced heart rate, acting via mechanism currently unclear, although shown elsewhere to predict independently raised natriuretic peptide levels, confirming this further.²⁶ One suggested mechanism is that a reduction in heart rate leads to increased left ventricular filling in diastole, which in turn leads to increased left ventricular wall stress, stimulating natriuretic peptide release.²⁶ Similar multivariable predictors have been seen elsewhere, confirming these findings further.^13,14

The current study has further shown that normal NTpBNP levels virtually exclude several important cardiovascular conditions. In general population subjects, normal NTpBNP levels gave a negative-predictive value (NPV) of 99% in ruling out AF, VHD, or significant LVSD, and an NPV of 96% in also ruling out DHF. In high-risk subjects, similar NPVs were also seen, an important finding, as two prior studies have suggested that natriuretic peptides may be less effective in screening high-risk subjects.^27,28 NTpBNP was, however, less effective in ruling out LVH.

Comparison of ROC curves in predicting LVSD

To allow comparisons with other studies not using age and gender cut-offs, AUCs were calculated for NTpBNP to screen for cardiovascular disease irrespective of the developed normal range. This is a retrospective analysis, and thus subject to bias, with poorer screening characteristics expected were best ROC cut-offs to be applied prospectively to a further cohort. Furthermore, it is likely to lead to a lower sensitivity in men, a lower specificity in women, and an increase in apparent best cut-off value as study participant age increases and different best cut-off values for different conditions. None of these problems arises using age- and gender-specific cut-off values. That said, the results compare favourably to other similar studies (Table8), finding AUCs of 0.91 and 0.79 for NTpBNP to screen for LVEF <40% and LVEF <50%, respectively.

Study limitations

This study has used an age cut-off of above or below 60 years to determine NTpBNP normal values, allowing sufficient sample size per group for analysis and being the cut-off used by the manufacturers of the assay allowing comparison. Such an approach may not be ideal, with more elderly subjects probably requiring higher cut-off values. Such an analysis is thus still required. However, the current study has shown in a prospective manner that even in subjects up to 91 years old, such cut-off values give excellent positive and NPVs in ruling in or out cardiovascular disease. A further limitation is the small number of subjects per group, potentially affecting the reliability of the results. However, as similar results have been seen elsewhere, this seems less likely. Furthermore, bootstrap estimates of the 97.5th percentile cut-off values give virtually identical results (100 pg/mL and 168 pg/mL for men aged 45–59 years and 60+ years, respectively, and 164 pg/mL and 220 pg/mL for women aged 45–59 years and 60+ years, respectively), further validating the calculated findings. Finally, despite the fact that a statistically significant difference between attendees and non-attendees could only be found in age, selection bias cannot be fully excluded. The advantage of this study is, however, that upper reference values were only assessed in truly normal subjects, the first time such a study has been performed.

Conclusion

Thus, normal NTpBNP levels should be stratified by age and gender, with this study characterizing age- and gender-stratified normal cut-off values. Normal NTpBNP levels give high NPV in excluding many significant cardiovascular conditions in both general population and high-risk subjects. Raised NTpBNP levels give high positive predictive value in detecting significant cardiovascular disease, especially once levels exceed three or four times normal. As many such conditions go undiagnosed, these findings may help underlie future natriuretic peptide-driven screening programmes.

Acknowledgements

This study was funded by the Northwick Park Hospital Cardiac Research Fund and a grant from West London Research Network (WeLReN). The NTpBNP reagents were supplied free of charge by Roche Diagnostics (Lewes, UK). The Harrow Research Ethics Committee granted ethical permission for this study.

References

© The European Society of Cardiology 2005. All rights reserved. For Permissions, please e-mail: journals.permissions@oxfordjournals.org

Issue Section:

Clinical Research

Email alerts

Citing articles via

More from Oxford Academic