Assessment of circulating N-terminal pro B-type natriuretic peptide concentration to differentiate between cardiac from noncardiac causes of pleural effusion in cats


  • Dr. Schneider is a consultant for IDEXX Laboratories.

  • The authors declare no other conflict of interest.

Address correspondence and reprint requests to Dr. Esther Hassdenteufel, Small Animal Clinic, Frankfurterstr. 126, 35392 Giessen, Germany. Email:



To determine the diagnostic ability of blood N-terminal pro B-type natriuretic peptide (NT-proBNP) measurement to differentiate between congestive heart failure (CHF) and noncardiogenic causes for moderate to severe pleural effusion in cats.


Prospective observational study.


University teaching hospital.


Twenty-one cats with moderate to severe pleural effusion.


Venous blood sampling for NT-proBNP measurement.

Measurement and Results

According to the results of echocardiographic examination, cats were classified in a group with CHF (n = 11) or noncongestive heart failure (N-CHF, n = 10). NT-proBNP was measured via a feline-specific test in EDTA plasma with protease inhibitor. NT-proBNP was significantly (P < 0.0001) higher in the CHF group ( median 982 pmol/L, 355–1,286 pmol/L) than in the N-CHF group (median 69 pmol/L, 26 – 160 pmol/L) and discriminated exactly (area under the curve = 1.0, 95% confidence interval 1.0–1.0) between both groups. Optimum cut-off value considering all samples was 258 pmol/L.


In this small population of cats with pleural effusion, NT-proBNP was able to differentiate between cats with cardiogenic and noncardiogenic causes of effusion. With the currently recommended method of measurement (ie, EDTA plasma with protease inhibitor), a cut-off value of 258 pmol/L discriminates effectively between cats with and without CHF.


area under the curve


congestive heart failure


noncongestive heart failure


N-terminal pro B-type natriuretic peptide


Cats with pleural effusion often present as emergencies in small animal practice. Congestive heart failure (CHF), along with pyothorax, neoplasia, idiopathic chylothorax, and feline infectious peritonitis (FIP), is one of the most common causes of pleural effusion, with a reported incidence rate between 14%–50% of feline patients with pleural effusion when an etiological diagnosis could be

reached.[1-3] Despite different causes, feline patients with pleural effusion due to cardiogenic or noncardiogenic origin present with similar clinical signs, usually with severe dyspnea as the main presenting complaint. Moreover, discrimination between etiological causes based on physical examination alone is challenging. The presence of moderate to large amounts of free fluid in the thorax interferes with the radiographic assessment of the cardiac silhouette,[4] making calculating the vertebral heart size or evaluating distinct cardiac structures (eg, enlarged left atrium) impossible. Definitive diagnosis of CHF in patients with significant pleural effusion is usually obtained by echocardiography. However, this imaging modality requires an experienced sonographer, who might not be readily available. Thus, a simple diagnostic test would be of great utility in diagnosing CHF in patients with pleural effusion. In recent years, N-terminal pro-B-type natriuretic peptide (NT-proBNP), a marker of cardiac wall stress, has frequently been used as a biomarker in cats.[5-13] It represents the inactive counterpart of C-terminal B-type natriuretic peptide, a peptide that is produced mainly by the ventricular myocardium due to increased cardiac wall stress and acts as an antagonist of the renin-angiotensin-aldosterone system, inducing vasodilation and diuresis.[14] Determination of feline NT-proBNP is carried out by a species-specific enzyme-linked immuno sorbent assay (ELISA) and has already been used to detect cardiomyopathy[7, 13] and cardiomegaly[12] in asymptomatic cats. It has also been used to diagnose CHF in cats with dyspnea.[6, 8] Both BNP and NT-proBNP concentration in pleural fluid as well as in blood have been successfully applied to diagnose CHF in human patients with pleural effusion.[15-25] The aim of this study was to investigate NT-proBNP blood concentrations in cats with pleural effusion severe enough to prevent accurate radiographic evaluation of the heart. The hypothesis was that measurement of NT-proBNP facilitates the distinction between cats with CHF and cats with noncardiogenic causes of pleural effusion.

Materials and Methods

Between January 2010 and January 2012 feline patients were included into this prospective study that met the following inclusion criteria: (1) Presented with clinical signs leading to admission into the intensive care unit; (2) radiographic confirmation of the presence of pleural effusion severe enough to prevent cardiac evaluation (defined as moderate or severe pleural effusion);[4] and (3) echocardiography performed at the request of the primary clinician.

Radiography was performed either by the referring veterinarian or at presentation at the referral institution. Initial echocardiography was performed either by a board-certified cardiologist or by a specialist-in-training to confirm the presence of pleural effusion and to classify the patient being in CHF or having a noncardiogenic cause for pleural effusion (N-CHF). For patient stratification, the main decision criterion used was the size of the left and right atria.

Therapeutic thoracocentesis was performed with small catheters1 or chest drains,2 and pleural fluid analysis and cytological exam was performed. Pleural fluid was classified as transudate, modified transudate, exudate, chyle, or blood.[26] In addition, a full hematology and biochemical panel were performed in all patients.

After initial stabilization, diagnostic investigation was performed depending on the suspected etiology and therefore differed between the 2 patient groups. In the N-CHF group, diagnostic investigation for the underlying noncardiogenic disease was tailored to the individual patient. This included laboratory investigations (eg, fluid analysis, cytology of fine needle aspirates, hematology and biochemical analysis) and diagnostic imaging modalities (eg, radiography, ultrasonography, computed tomography). Patients with CHF underwent a complete echocardiographic examination (including Doppler-echocardiography and pulsed wave tissue Doppler Imaging), ECG and Doppler-based blood pressure measurement. Underlying heart disease was classified according to the results of echocardiography by a board-certified cardiologist. Cardiomyopathies were divided into 5 possible forms regarding published criteria[27]: hypertrophic cardiomyopathy (HCM, hypertrophy of the left ventricle with M-mode measured diastolic diameter of the septum or left ventricular free wall more than 6 mm, normal or reduced left ventricular diameter), restrictive cardiomyopathy (RCM, left atrial dilation with restrictive filling and normal or near-normal systolic function and wall thickness), dilated cardiomyopathy (DCM, dilation and impaired contractile function of the left ventricle or both ventricles), arrhythmogenic right ventricular cardiomyopathy (ARVC, severe right ventricular and right atrial dilation) and unclassified cardiomyopathy (UCM, cases that did not fit readily into any other group).

Blood for NT-proBNP measurement was taken into EDTA-tubes during venous blood sampling for analysis of hematology and biochemical panel, after client consent was obtained. Within 30 minutes EDTA plasma was separated by centrifugation. EDTA plasma was filled into specific tubes containing a protease inhibitor mixture3 and was subsequently frozen at -20°C. Samples were shipped frozen to a commercial laboratory.4 They were analyzed using a feline specific NT-proBNP test5 by staff blinded to the final diagnosis within 1 week after collection. The performance of this assay has been previously evaluated.[8] Samples with values above the maximum test range (1,285 pmol/L) were reported and analyzed as 1,286 pmol/L.

The following parameters from the physical examination were also collected: heart rate, respiratory rate, temperature and clinical findings during cardiac auscultation (eg, presence of arrhythmias, a heart murmur, gallop rhythm).

Statistical Methods

Continuous variables were inspected visually and tested for normality by D'Agostino and Pearson omnibus test. Descriptive statistics included frequencies for categorical variables and either median and range or mean and SD for continuous variables. Depending on the presence or absence of normal distribution, disease group data (CHF versus N-CHF) were compared by using an unpaired t-test or Mann–Whitney U-test, respectively. Proportions were compared using Fisher's exact test. NT-proBNP values were graphically depicted as scatter plot of individual data points and were analyzed with receiver operator characteristic analysis to determine diagnostic ability of NT-proBNP to diagnose CHF. The area under the curve (AUC) was used as a summary measure and quantification of diagnostic accuracy for NT-proBNP to predict CHF. The cut-off value was chosen based on the highest Youden index (Y = sensitivity + specificity −1).[28] Statistical analyses were performed with a commercially available software program.6 P values < 0.05 were considered significant.


Twenty-one cats were included into this study. In N-CHF patients (n = 10), the following underlying diseases were diagnosed: pyothorax (n = 2), feline infectious peritonitis (n = 1), lymphoma (n = 2), carcinoma (n = 2), bronchoalveolar adenoma (n = 1), mediastinal sarcoma (n = 1), hypoalbuminemia due to hepatic insufficiency (n = 1). None of the N-CHF cats had echocardiographic changes consistent with concurrent cardiomyopathy. In CHF patients (n = 11), diagnoses were: hypertrophic cardiomyopathy (n = 6), restrictive cardiomyopathy (n = 2), unclassified cardiomyopathy (n = 1), pulmonic stenosis (n = 2). Echocardiographic data that led to disease classification are listed in Table 1. Three patients with N-CHF (30%) and 5 patients with CHF (45%) received pretreatment with furosemide.

Table 1. Echocardiographic findings in cats with pleural effusion
  N-CHF (n = 10#)HCM (n = 6)RCM (n = 2)UCM (n = 1)PS (n = 2)
  1. # Mmode data are available in 7/10 N-CHF cats; LAs = left atrium systolic diameter; RAs = right atrium systolic diameter; LAd = left atrium diastolic diameter; Ao = aortic diastolic diameter; IVSd = interventricular septum diastolic diameter; LVWd = left ventricular wall diastolic diameter; LVDd = left ventricular diastolic diameter; LVDs = left ventricular systolic diameter; FS = fractional shortening.

2D long axisLAs (mm)13.711.8–15.720.418.4–28.122.6; 24.526.513.8; 13.1
 RAs (mm)8.65.6–11.516.913.7–18.412.5; 13.911.819.6; 24.7
2D short axisAo (mm)9.39.0––9.98.4;; 9.4
 LAd (mm)11.98.1–13.218.916.3–26.417.7; 24.420.412.1; 13.9
 LAd/Ao1.30.9––2.52.5;; 1.5
M mode long axisIVSd (mm)4.03.5––7.14.9;; 3.7
 LVWd (mm)4.63.8––8.95.3;; 5.4
 LVDd (mm)15.612.8–20.616.811.9–18.917.1; 14.314.86.8; 9.0
 LVDs (mm)7.95.5–12.911.45.9–16.97.1;; 8.1
 FS (%)4429–653411–5459; 283741; 10

NT-proBNP concentration (Figure 1) in the N-CHF group was found to be median 69 pmol/L (range 26 – 160 pmol/L, n = 10), and was significantly lower (P < 0.0001) than in the group with CHF (median 982 pmol/L, range 355–1,286 pmol/L, n = 11). Receiver operator characteristic analysis calculated an AUC of 1.0 (95% confidence interval 1.0–1.0) with an optimum cut-off value of 258 pmol/L.

Figure 1.

NT-proBNP concentration in 21 cats with pleural effusion. Cats with congestive heart failure (n = 11) had significantly higher NT-proBNP concentrations than cats without congestive heart failure (n = 10; P < 0.0001).

There were no significant differences concerning sex distribution, age, and body weight between both groups (Table 2). Also, no significant differences were found regarding continuous data determined by physical examination (eg, heart rate, respiratory rate, temperature). An arrhythmia was noted in 1 of the cats in the N-CHF group. There were 2 cats with arrhythmia, 2 cats with a gallop rhythm and 2 cats with heart murmur and arrhythmia in the CHF group. In total 6/11 (55%) animals in the CHF group had at least 1 pathological finding on cardiac auscultation. This proportion was not significantly different compared to the N-CHF group (P = 0.0635). Plasma creatinine concentration was not significantly different between groups. Systolic blood pressure was measured in all cats with CHF and none of these cats showed hypertension (>170 mm Hg). Blood pressure measurement was not routinely performed in the N-CHF cats.

Table 2. Summary of the data in two groups of cats with (CHF) or without (N-CHF) congestive heart failure
 (n = 10)(n = 11)P value
  1. DSH = Domestic Short hair; DLH = Domestic Long hair; m = male; f = female.

  2. *Temperature was measured only in 9/11 cats with CHF.

BreedsDSH (6)DSH (6)
 Maine Coon (2)Maine Coon (1) 
 Birma (1)Chartreux (1) 
 Chartreux (1)Russian Blue (1) 
  Crossbreed (1) 
  DLH (1) 
Sex (m/f)6/46/51.0
Age (year)7.4 ± 4.810.4 ± 5.20.1922
Body weight (kg)5.0 ± 1.64.6 ± 1.40.6417
Heart rate (/min)204 ± 18186 ± 380.1814
Respiratory rate (/min)56 ± 2365 ± 180.3231
Temperature (°C)*38.4 ± 0.738.6 ± 0.90.6089
Pathological cardiac auscultation (%)10550.0635
Creatinine (μmol/L)136 (38–325)109 (49–430)0.8053
NT-proBNP (pmol/L)69 (26–160)982 (355–1,286)<0.0001


Veterinary studies in dogs[29-32] as well as in cats[5-8, 13] have typically investigated the relevance of cardiac biomarkers without considering radiographic findings. The present study in cats examined the diagnostic utility of NT-proBNP in a situation of a nondiagnostic radiograph because of pleural effusion. NT-proBNP concentration showed a high diagnostic utility (AUC 1.0) with an optimal cut-off value of 258 pmol/L to distinguish between cardiogenic and noncardiogenic causes for pleural effusion. This value, obtained by measurement in EDTA-plasma with protease inhibitor, was found to be concordant with the published cut-off values for cats with dyspnea and measurement in serum (220 pmol/L)[6] or EDTA plasma (265 pmol/L).[8]

In people with pleural effusion, NT-proBNP measurement in blood samples is associated with similarly good results of diagnostic accuracy (AUC 0.92–1.0).[17, 21, 24] The high diagnostic accuracy in the current study is in agreement with other feline studies, which demonstrated an increasing diagnostic accuracy of NT-proBNP with increasing suspicion of heart disease. This test does not appear to be useful as screening test for breeding examination,[9, 11] but works well in asymptomatic cats with clinical suspicion of heart disease (AUC 0.92)[7] and has the best performance in cats with dyspnea (AUC 0.94[8] and AUC 0.96[6]). One reason for the high accuracy of the test to diagnose CHF in cats with pleural effusion can be assumed to be caused by severe cardiac decompensation of these cats with CHF. Conversely, values in the noncardiogenic group were only mildly increased; this can be explained by the fact that none of cats in this group had concurrent heart disease or other diseases known to cause significant increase in NT-proBNP concentration. It is possible that the accuracy of the test decreases with its application to a broader patient base, as it was recently documented in the largest human study with NT-proBNP measurement in 398 patients with pleural effusion (AUC 0.89).[25]

Originally, it was presumed that pleural effusion by itself might lead to an increase in NT-proBNP due to hemodynamic impairment.[6] But this seems to be unlikely, as one study in people with pleural effusion could not demonstrate a reduction in BNP after thoracocentesis alone.[15] In human patients with pleural effusion, noncardiogenic increase of NT-proBNP was reported in noncardiac transudates and in exudates[25] due to inflammatory or neoplastic diseases[25] as well as in patients with pulmonary thromboembolism.[18] Noncardiogenic pleural transudates in people are usually caused by hepatic cirrhosis, but the mechanism for increases in NT-proBNP concentration in these cases is unclear.[33] In our study, the cat with a noncardiogenic transudate due to hypoalbuminemia caused by hepatic insufficiency had the highest NT-proBNP concentration with 160 pmol/L in the N-CHF group. Increases in NT-proBNP in people with inflammatory exudate can be assumed to be caused by myocardial dysfunction and cytokine release in sepsis.[34, 35] Increases of NT-proBNP concentration in dogs is reported in abstract form regarding patients with systemic inflammatory response syndrome7 and has also been reported in a dog with sepsis.[36] Currently, there is no information available concerning cats, but high concentrations in the current study (120 and 144 pmol/L) in the N-CHF group were found in 2 cats with pyothorax, despite none of them showing echocardiographic signs of myocardial dysfunction. In veterinary medicine, increases in NT-proBNP concentration in neoplastic effusions has not been reported so far and could not be detected in the 6 cats affected by neoplasia of the current study. Severe pulmonary thromboembolism as a cause for increased NT-proBNP concentrations was reported in 3 dogs.[36] It is possible that some of the dogs described with pulmonary hypertension and BNP[37] or increase in NT-proBNP concentration[38] were also affected by smaller pulmonary emboli. Pulmonary thromboembolism or hypertension as reasons for increased NT-proBNP concentration has not been described in cats thus far, but it can be presumed to occur due to right ventricular pressure overload as it is seen in people and dogs. No direct information regarding NT-proBNP measurement in cats with idiopathic chylothorax is available. The current study did not have any cats with this disease, as all cats presented during the study period were not admitted to the intensive care unit because of mild clinical signs after pretreatment by the referring veterinarian. Additional studies including such noncardiac patients are warranted to confirm our preliminary observations. But even if restrictive pericardial disease is presumed to be involved in the pathophysiology of idiopathic chylothorax, this might not be associated with an increase of NT-proBNP concentration, as only mildly increased BNP concentrations are documented in people with constrictive pericarditis.[39]

Exclusion criteria were intentionally not defined, so that the actual relevance of the test in veterinary routine diagnostics could be investigated. As in other studies concerning NT-proBNP in cats with dyspnea[6, 8] pretreatment was not an exclusion criteria in the current study. However, certain influences on NT-proBNP concentration cannot be excluded, as medical treatment including furosemide has been shown to decrease NT-proBNP concentrations in people[40] and dogs with CHF.[41] But exclusion of these patients does not seem to be reasonable, as pretreatment is commonly performed for stabilization before referral of a patient, so withdrawal of these patients is not consistent with routine veterinary practice.

Pathological findings on cardiac auscultation were found in 55% of the animals with CHF and in 10% of the animals of the N-CHF group. Both rates are lower than in a retrospective study in cats with dyspnea.[42] This difference can be explained by the influence of pleural effusion on cardiac auscultation. In the current study, the difference in auscultation findings between CHF and N-CHF cats was not significant, presumably because of the small patient number, but the trend is similar to that reported in a larger study in cats with dyspnea.[42] If animals with pathological findings on auscultation were excluded from analysis in the current study, there would be no change in AUC or cut-off value (258 pmol/L). This approach also includes the risk for misclassification of patients with auscultation findings in noncardiac cases[42] and of patients with concurrent subclinical heart disease.[8] Benefits of incorporating NT-proBNP measurement in the diagnostic approach for this patient population may include reaching a more rapid diagnosis and possible cost savings.

Study Limitations

The major study limitation of the current study was the very small number of patients evaluated; in a larger population some false positive or negative results may have been found. Additional limitations included the lack of patients with idiopathic chylothorax, a disease where it would be of special interest to differentiate these patients from cats with cardiogenic chylothorax.

Pleural fluid analysis was not taken into consideration in the current study, because it cannot differentiate between cardiac and noncardiac diseases in some cases.[3] NT-proBNP was not assayed in pleural fluid which might be interesting in cats stressed by venous blood sampling. In people, NT-proBNP concentration in blood correlate well with that found in pleural fluid.[17, 21]

The relevance of the current study to general clinical practice is limited by the lack of a point-of-care test for NTpro-BNP. In people, diuretic therapy may be instituted with a high NT-proBNP concentration and clinically tolerable pleural effusion.[43] Since semiquantitative point-of-care tests are usually optimized to a specific cut-off concentrations, the similarity in cut-off values in cats with pleural effusion with cats with dyspnea are encouraging in terms for a development of feline point-of-care tests for NT-proBNP.


In this small population of cats with marked pleural effusion measurement NT-proBNP effectively distinguished between cardiac and noncardiac underlying diseases. A cut-off value of 258 pmol/L was derived by measurement of NT-proBNP in EDTA plasma with protease inhibition and this must be considered in future development of this diagnostic test.


The authors thank Dr. Silke Schmitz for revising the English manuscript.


  1. 1

    Intravenous cannula, 18 Ga/ 20 Ga, KLINIKA, Usingen, Germany.

  2. 2

    Chest Drain, 6Fr, 50 cm, Walter, Baruth/Mark, Germany.

  3. 3

    Cardiopet proBNP Transport Tube, IDEXX Laboratories, Westbrook, ME.

  4. 4

    Vet Med Labor GmbH, Division of IDEXX Laboratories, Ludwigsburg, Germany.

  5. 5

    Cardiopet proBNP, IDEXX Laboratories.

  6. 6

    GraphPad Prism 5, GraphPad Software, Inc., San Diego, CA.

  7. 7

    Gommeren K, Desmas I, Garcia A, Massart L, Clercx C, McEntee K, Peeters D. Cardiac troponin and natriuretic peptide in canine emergencies with a systemic inflammatory response syndrome (Abstr). Proceedings 21st ECVIM-CA congress 2011:249.