Skip navigation
Link to Back issues listing | Back Issue Listing with content Index | Subject Index

B-type natriuretic peptide (BNP) for CHF

BNP for CHF diagnosis
BNP for prognosis
BNP for monitoring

Diagnostic tests are a problem. The explosion of substances found to influence cellular behaviour seems to be rolling on remorselessly. When we learn how to measure them, we find that most of these substances can be found in body fluids like blood, urine, or cerebrospinal fluid. And because they can be measured, people measure them. They will inevitably find that concentrations are higher in some people than others, perhaps people with different conditions, and a diagnostic test is born.

New-born diagnostic tests are weak and feeble creatures. They will have been tested on a few selected people, often compared in people with the most advanced condition and healthy youngsters in the laboratory. A bit of spin and the weak and feeble test comes roaring though the literature, and is the latest craze.

That overstates it, but it is certainly the case that knowing in whom to use a new diagnostic test is difficult, and the interpretation uncertain. So where are we with B-type natriuretic peptide (BNP), the new kid on the block for use in heart failure? Luckily we already have two systematic reviews [1 2] to help, and their underlying message is that the test may well be helpful, but we need strategies in place to use it properly.


BNP is a cardiac neurohormone secreted from the ventricles in response to volume expansion and pressure overload, though some is also released from atrial tissue. BNP gene expression increases rapidly in response to stimulus. BNP is a 32 amino acid peptide, and is measured using the N-terminal, and assays generally measure both intact and N-terminal BNP, and there is uncertainty about whether measuring both or either is better.

BNP for CHF diagnosis

Four studies of reasonable quality have compared BNP measurements with echocardiogram as a reference diagnostic standard in appropriate patients with New York Heart Association classes I-IV heart failure using independent blind comparison. In all, about 2,200 patients were studied, with the largest group of patients (almost 1,600) presenting to an emergency room with dyspnoea. Some of the studies were using point-of-care BNP tests.

In the largest study about half (49%) did not have congestive heart failure, and most had BNP levels under 100 ng/L. In those with heart failure, concentrations rose with severity (Table 1). Note the very large standard deviations in Table 1, showing great variability.

Table 1: BNP concentrations with increasing CHF severity

NYHA Class Mean SD
I 240 290
II 390 370
III 640 450
IV 820 440

Table 2 shows the diagnostic performance of the four studies. Positive likelihood ratios were as high as 40, and as low as 4. Even so, a likelihood ratio of four for a positive test, starting with a 50% prevalence, would produce a post-test probability of about 80%, and the likelihood ratio of a negative test of 0.1, a post test probability of about 8%. How these figures for the worst result from the largest study look in natural frequencies is shown in Figure 1, using a cut off of 80 ng/L. Below a lower cut-off of 50 ng/L, 96% of all results would have correctly predicted that a patient did not have congestive heart failure, effectively ruling out the diagnosis.

Table 2: Characteristics of BNP testing in four different studies

Study Patients BNP cut-off
Positive likelihood ratio
1 122 76 97 84 6
2 250 80 98 92 12
3 1586 80 93 74 4
4 321 94 86 98 43

Figure 1: Characteristics of BNP test applied to a hypothetical 1,000 people attending an emergency room with dyspnoea and suspected CHF

BNP for prognosis

Two studies of reasonable quality demonstrated that lower BNP concentrations correlated with better outcomes, and higher values with worse outcomes.

BNP for monitoring

Two small studies suggested that better treatment produced lower BNP concentrations, and that BNP-guided therapy resulted in significantly fewer cases of cardiovascular death, admission, or outpatient attendance.


A reader, who wanted to know whether this was a useful test in primary care, alerted Bandolier to BNP as an issue. The straight answer is that as it stands there just isn’t a satisfactory answer, beyond the rather trite one of there not being enough evidence. But a test is just a tactic. What is missing is a strategy.

The two systematic reviews are incredibly helpful in setting out the background. They reflect their origins, one from a department of family and community medicine [1], and the other from a cardiovascular research centre [2]. Both are positive, and the evidence around BNP as a marker of heart failure is good, so far as it goes. But there are unknowns. For instance, is it likely that the test will be done in primary care offices and surgeries where a patient first presents? Unlikely in most places. More likely will be its use in hospital outpatients or emergency rooms, but even there close cooperation between physicians and laboratories will be needed.

Both the reviews tell us how difficult it is to diagnose heart failure, with references to prove it. They hint at strategies involving clinical examination, X-rays, and BNP measurements, and emphasise the other things that can elevate BNP, like myocardial infarction, ventricular hypertrophy, cardiomyopathy, lung cancer, pulmonary embolism, renal failure, and COPD. We can predict that BNP is no holy grail of testing, but it is probably going to be useful. The challenge is to work out how.

And a quick postscript. There is talk of using the test to screen for heart failure. No one has yet tried it, and it will be fraught.


  1. R Cardarelli, TG Lumicao. B-type natriuretic peptide: a review of its diagnostic, prognostic, and therapeutic monitoring value in heart failure for primary care physicians. Journal of the American Board of Family Practice 2003 16: 327-333.
  2. JA de Lemos et al. B-type natriuretic peptide in cardiac disease. Lancet 2003 362: 316-322.

previous story