Heart failure (HF) is a major public health problem worldwide – one in five people aged over 40 will develop HF at some stage – and in Brazil it is the leading cause of hospitalization in those aged over 60.1

Over half of these patients are classified as having HF with preserved ejection fraction (HFPEF)2; this proportion is growing in both developed and developing countries due to ageing populations and increased awareness among physicians of the syndrome, which predominantly affects women and those with multiple comorbidities.3–5 The costs associated with HF are high, for both hospitalization and outpatient care, irrespective of left ventricular ejection fraction (LVEF). A simple exam would thus be useful for screening these patients with a view to more thorough investigation.2–6

B-type natriuretic peptide (BNP) is released in response to increased left ventricular (LV) filling pressure7 and end-diastolic wall stress,8 and plasma BNP levels can help in both diagnosis (confirming or excluding HF) and prognostic assessment of HF.9,10

Paulus et al.6 proposed criteria for the diagnosis of HFPEF that included Doppler echocardiographic parameters and BNP measurement. The BNP cutoff used was based on data from a study by Maisel et al. of patients admitted to the emergency room with acute HF.11

However, BNP assessment in outpatients show lower values than those suggested by Paulus et al. for a diagnosis of HFPEF, although with different Doppler echocardiographic criteria for diagnostic confirmation.12,13

The use of the BNP cutoff proposed by Paulus et al. has not been validated in clinical practice for outpatients, but BNP measurement, an inexpensive exam, may help to exclude HFPEF and thus direct investigation of a patient's symptoms to other areas.

The aim of this study was to determine the diagnostic accuracy of BNP measurement and the best cutoff to confirm or exclude HFPEF in outpatients using the criteria proposed by Paulus et al.

The demographic, clinical, laboratory and echocardiographic characteristics of the study population are shown in Table 1, from which it can be seen that patients with HFPEF were older, and had a higher prevalence of hypertension, diabetes and atrial fibrillation and lower glomerular filtration rate (GFR) than those without HFPEF. LV function assessed by Doppler echocardiography showed more marked systolic and diastolic dysfunction in the group with HFPEF.

BNP levels were significantly different between those with and without HFPEF (mean 144.4 pg/ml, median 113 pg/ml vs. mean 27.6 pg/ml, median 16.7 pg/ml, respectively, p<0.0001).

Table 2 shows the correlations between BNP and clinical and echocardiographic variables. There was a direct correlation between BNP and LAVI, age and E/E’ ratio and an indirect correlation with estimated GFR.

The area under the ROC curve for BNP to detect HFPEF was 0.92 (95% confidence interval [CI] 0.87-0.96, p<0.0001) (Figure 1), and 51 pg/ml was the most appropriate cutoff to detect HFPEF, with sensitivity of 86%, specificity of 86%, positive predictive value of 93%, negative predictive value of 72%, and diagnostic accuracy of 86%.

Figure 1.

Receiver-operator characteristic curve identifying the best cutoff for B-type natriuretic peptide to diagnose heart failure with preserved ejection fraction in outpatients. AUC: area under the curve; CI: confidence interval.

(0.16MB).

This is the first prospective study to use the ESC criteria6 for the diagnosis of HFPEF in outpatients and to demonstrate that HFPEF patients present high BNP values, although lower than those found in emergency room patients. Our results suggest that BNP cutoffs for a diagnosis of HFPEF in outpatients may be lower than those in the ESC consensus statement, which was based on studies in patients presenting with acute HF.6

In the emergency room, signs and symptoms of congestive HF are usually present and since many patients are hospitalized due to acute pulmonary edema, BNP levels are higher than in outpatients, as demonstrated in a study of 1586 emergency room patients, in whom a BNP cutoff of 100 pg/ml had sensitivity of 90% and specificity of 76%, irrespective of LVEF, for differentiating acute HF from other causes of dyspnea.15

Exertional dyspnea may be the only symptom in out-patients with HFPEF, with no sign of congestion, and those most commonly affected are women, the elderly, the obese, and those with multiple comorbidities.16 The diagnosis of HFPEF in such cases is a challenge to physicians and a biomarker such as BNP may be useful in identifying these patients, but based on lower BNP levels than in emergency room patients.

Our results show that patients with HFPEF present significant diastolic dysfunction, characterized by alterations in ventricular relaxation (E’) and increased LV filling pressures (as shown by LAVI and E/E’ ratio) but normal LVEF. The correlation observed between BNP values and echocardiographic parameters reflects the association between BNP and LV diastolic function, and BNP measurement is thus a useful and simple exam to assess dyspnea in outpatients with suspected HF.

Arques et al.17 studied 26 outpatients presenting chronic dyspnea as the only manifestation of HF who underwent cardiac catheterization, which is considered the gold standard exam to confirm a diagnosis of HFPEF.18 The study showed that BNP of 31 pg/ml was predictive of HFPEF (p=0.003), with 67% sensitivity and 73% specificity (area under the ROC curve 0.76, 95% CI: 0.55-0.90; p=0.007).17

Penicka et al.19 assessed 30 outpatients (73% female) with unexplained chronic dyspnea, in NYHA functional class II or III and LVEF >50%, who underwent cardiac catheterization. HFPEF was defined as LV end-diastolic pressure of >16 mmHg at rest during hemodynamic study, and HFPEF was confirmed in 20 of these patients (66%). BNP was higher in the HFPEF group than in controls (68 pg/ml vs. 21 pg/ml, p=0.09), but only three patients had BNP levels above the cutoff recommended by Paulus et al. (200 pg/ml) for a diagnosis of HFPEF.19

Kitzman et al.13 assessed 147 outpatients, of whom 59 had HFPEF; mean BNP in this group was 56±30 pg/ml.

Our findings are consistent with the results of the above studies13,17,19 in terms of BNP cutoff points in patients with HFPEF, and confirm that outpatients with HFPEF can have BNP levels well below the cutoffs established by the ESC.6

Determining a more appropriate BNP cutoff than those currently used could help in screening outpatients with suspected HFPEF. Zuber et al. studied 384 primary care patients to assess the accuracy of BNP for diagnosing HF as determined by clinical and echocardiographic data and found that 31 (8%) had HFPEF and 193 (50%) had systolic HF with reduced ejection fraction. The recommended BNP cutoff of 100 pg/ml was used to diagnose or exclude HF, which gave a false negative result in 25% of patients with HF. The study concluded that this cutoff failed to exclude HF in primary care patients.20

The latest ESC guidelines for the diagnosis and treatment of acute and chronic HF21 propose a new BNP cutoff of 35 pg/ml to exclude HF (with reduced or preserved ejection fraction) in outpatients, which may reduce the number of false negatives and unnecessary echocardiographic exams. This cutoff was based on data from eight studies22–29 that analyzed patients with suspected HF or at risk of developing HF due to ventricular dysfunction. The cutoff of 51 pg/ml determined in our study is specific to HFPEF and showed a positive predictive value of 93% and negative predictive value of 72%, making it useful to confirm or exclude HFPEF in outpatients.

In view of the ageing of populations, which will increase the number of cases of HFPEF, BNP measurement could be a less costly and simpler way to screen outpatients for suspected HF than echocardiography.

Certain limitations of our study should be borne in mind. As the absolute number of patients in whom HFPEF was confirmed was small, and given the heterogeneous nature of the syndrome, and the fact that comorbidities can effect BNP levels, further studies involving large samples of outpatients will be required to validate our findings.

BNP levels in outpatients with HFPEF, as diagnosed by ESC criteria, are higher than in those without HFPEF. In the study population, a cutoff of 51 pg/ml had the best diagnostic accuracy to identify HFPEF in outpatients.

The authors have no conflicts of interest to declare.