This work was performed collaboratively between the Department of Veterinary Medicine, University of Cambridge and the Small Animal Teaching Hospital, University of Liverpool.
The Cardiac Biomarker NT-proBNP Is Increased in Dogs with Azotemia
Version of Record online: 26 AUG 2009
Copyright © 2009 by the American College of Veterinary Internal Medicine
Journal of Veterinary Internal Medicine
Volume 23, Issue 6, pages 1184–1189, November/December 2009
How to Cite
Raffan, E., Loureiro, J., Dukes-McEwan, J., Fonfara, S., James, R., Swift, S., Bexfield, N., Herrtage, M.E. and Archer, J. (2009), The Cardiac Biomarker NT-proBNP Is Increased in Dogs with Azotemia. Journal of Veterinary Internal Medicine, 23: 1184–1189. doi: 10.1111/j.1939-1676.2009.0373.x
- Issue online: 27 OCT 2009
- Version of Record online: 26 AUG 2009
- Submitted November 11, 2008; Revised June 3, 2009; Accepted July 9, 2009.
- Congestive heart failure;
- Kidney failure
Background: Amino-terminal probrain natriuretic peptide (NT-proBNP) has been proposed as a useful biomarker for heart disease in dogs. In humans, decreased glomerular filtration rate (GFR) increases NT-proBNP.
Objective: To investigate whether decreased GFR as indicated by plasma creatinine concentration is associated with increased NT-proBNP in dogs without heart disease.
Animals: Four groups of dogs: healthy (n= 39), azotemic (n= 36), heart disease (n= 37), and congestive heart failure (CHF) (n= 7) presented to 2 teaching hospitals.
Methods: Prospective observational cohort study. Plasma creatinine concentration and NT-proBNP were measured in every dog. Nonparametric tests were used to compare the differences among groups. The median and actual results for each group were compared with the manufacturer's recommended and previously published suggestions for cut-off values for diagnosis of heart disease.
Results: Median (range) plasma creatinine concentration was 1.47 (1.06–1.70), 4.36 (1.74–15.6), 1.22 (0.69–1.91), and 1.45 (0.63–1.64) mg/dL and median (range) NT-proBNP was 118 (2–673), 556 (37–1,819), 929 (212–5,658), and 3,144 (432–5,500) pmol/L for the healthy, azotemic, heart disease, and CHF groups, respectively. Pair-wise comparison indicated a significant difference among all groups for NT-proBNP (P≤ .049). Plasma creatinine concentration was significantly higher in the azotemic group compared with other groups (P < .001) but there was no significant among other groups. Application of 3 recommended cut-off values led to misclassification of dogs with azotemia as having heart disease.
Conclusions: Azotemia results in NT-proBNP being increased to concentrations reported as diagnostic of heart disease or heart failure in dogs. Care should be employed when interpreting the results of NT-proBNP in patients with known or possible increased plasma creatinine concentration.
arrhythmogenic right ventricular cardiomyopathy
brain natriuretic peptide
congestive heart failure
chronic kidney disease
degenerative valvular disease
glomerular filtration rate
immune-mediated hemolytic anemia
multiple organ dysfunction syndrome
amino terminal proBNP
B-type natriuretic peptide (brain natriuretic peptide, BNP) is one of a group of natriuretic peptides and has natriuretic, diuretic, and vasodilatory properties in dogs.1,2 It is released by myocytes in response to myocardial stretch. Very small amounts are present in health, but marked increases occur in the presence of cardiac disease.3
BNP is synthesized as a prohormone, proBNP, which is cleaved into its biologically active moiety BNP and the remaining biologically inactive amino terminal proBNP (NT-proBNP) so that the 2 peptides are released into circulation in equal ratio.4 NT-proBNP offers some advantages as a target for diagnostic testing: NT-proBNP has a longer circulating half-life than BNP (the only previously reported data are from ovine samples),5 has a higher concentration in circulation,5 and is more stable in stored (frozen) plasma.6
As a result, NT-proBNP has been used in human medicine to distinguish between cardiac and noncardiac causes of dyspnea, as a prognostic indicator and to monitor disease progression.7 NT-proBNP recently has been shown to be increased in dogs with heart disease,8–10 to distinguish between cardiac and respiratory causes of dyspnea10 and to assess the severity of cardiac disease.8 Oyama8 recommends that 445 and 1,725 pmol/L be used as cut-off values for diagnosis of heart disease and heart failure, resepectively.
BNP is cleared via receptors that are present in the kidney and vasculature.11 In contrast, NT-proBNP has no known receptors and little is known about its clearance from plasma, although renal metabolism is thought to be responsible.12
Both BNP and NT-proBNP are increased in humans with kidney disease but without heart disease, and their diagnostic value in patients with renal failure has been called into question.13 Modified diagnostic cut-off values have been suggested in order to maintain the diagnostic value of the test in people.14–16 Limited data recently have been presented that a similar increase occurs in dogs with renal disease.17
The aim of this prospective study was to determine whether renal disease increases NT-proBNP in dogs without heart disease and whether this increase limits the diagnostic value of the test.
A prospective, observational cohort study was performed. Data were collected from dogs presented to the Queen's Veterinary School Hospital, University of Cambridge and the Small Animal Teaching Hospital, University of Liverpool. Four groups of dogs were included in the study: healthy, azotemic, heart disease, and congestive heart failure (CHF).
Healthy dogs were presented for routine neutering, minor dental prophylaxis, or orthopedic procedures for which blood was collected for preanesthetic evaluation. Dogs in the azotemic group were selected for having plasma creatinine concentrations outside of the normal reference interval during clinical testing and all underwent clinical evaluation as determined appropriate to investigate their clinical problems by the attending clinician at the Queen's Veterinary School Hospital, University of Cambridge. Azotemic dogs were excluded if any historical, physical examination, or diagnostic test findings indicated the possibility of heart disease at the time of sampling. No dogs in the healthy or azotemic groups underwent echocardiography.
Data were collected sequentially from dogs presented to the cardiology service of the Small Animal Hospital, University of Liverpool. Samples were collected as part of routine clinical investigations and dogs were included if cardiac disease was confirmed. All cases had biochemistry, hematology, blood pressure measurement, 6 lead electrocardiography, Doppler echocardiography, and thoracic radiographs performed at the time of diagnosis. Additional diagnostic tests were performed as deemed necessary by the clinician in charge of the case. The cardiology cases then were divided into 2 groups: those with evidence of heart disease but not CHF (heart disease group) and those with evidence of heart disease and CHF at the time of sampling (CHF group). Left-sided CHF was defined as echocardiographic evidence of severe left-sided heart disease sufficient to cause CHF, radiographic evidence of pulmonary venous congestion and pulmonary edema, together with clinical signs consistent with CHF (eg, tachypnea, dyspnea, or coughing). Right-sided CHF was defined as echocardiographic evidence of severe right-sided heart disease sufficient to cause CHF, together with ultrasonographic evidence of right atrial enlargement, hepatic venous congestion, and ascites. Dogs were deemed to have a primary rhythm disturbance if there was no evidence of underlying systemic or structural cardiac disease on routine or other investigations.
The clinical diagnosis made by the attending clinician was used.
Measurement of Creatinine and NT-proBNP
Blood samples were taken and plasma harvested and frozen at −20°C within an hour of sampling. Plasma creatinine concentration and NT-proBNP were assayed on every sample. Plasma creatinine concentrations were assayed on an automated analyzera by an alkaline picrate method that has an established normal reference interval for dogs of 0.51–1.7 mg/dL. NT-proBNP concentrations were determined by means of a commercially available sandwich enzyme immunoassayb that uses 2 immunoaffinity purified sheep antibodies specific for canine NT-proBNP. Microtiter plates were provided with capture antibody anti-NT-proBNP (25–41) bound to the wells. To each well, 20 μL of sample and 200 μL of tracer consisting of detection antibody (1–22) conjugated to horseradish peroxidase were added. Plates were mixed, covered, and incubated for 16–24 hours in the dark at room temperature. Plates were washed 5 times with 350 μL diluted wash buffer and dried before 200 μL substrate (tetramethylbenzidine) was added. After mixing, plates were incubated in the dark at room temperature for 30 minutes then 50 μL of stop solution was added to each well and mixed. Absorbance was measured immediately at 450 nm on an automated plate reader.c Standard curves were generated using reference standards provided by the manufacturer. All samples were analyzed in duplicate and the mean result used in analysis. The manufacturer reports this test to have an interassay coefficient of variation of 7.1–8.6% and an intra-assay coefficient of variation of 4.7–9.8% (different values reported for low, medium, or high concentration standards). Laboratory personnel were blinded to the disease status of the samples.
Clinical data on patient age, sex, breed, and diagnosis was collected from analysis of case records.
Descriptive statistics were generated on patient age, sex, breed, and diagnosis. Age was normally distributed and analysis of variance18 was used to compare age among groups. Because creatinine and NT-proBNP data were not normally distributed, the Kruskal-Wallis test19 was used to identify differences among results for the 4 groups of dogs. When a significant difference was identified, the Mann-Whitney U-test20 was applied sequentially to examine the difference pair-wise between individual groups. Because this method tested multiple hypotheses, the Holm-Bonferroni correction21 was applied to the results of Mann-Whitney U-tests. Spearman's rank-correlation coefficient22 was used to assess correlation between plasma creatinine concentration and NT-proBNP concentration.
For each test, analysis was 2-sided. Results were determined to be significant if P was ≤.05. All statistics were performed by a commercially available software.d
For each group, the numbers of animals that would lie above and below cut-off values recommended by the test kit manufacturer and published previously were reported.
The statistical analysis was repeated on a modified data set that included only 1 sample from any dog and excluded all heart disease or CHF cases that were azotemic and also all dogs for which no structural heart disease was identified (ie, those with primary rhythm disturbances).
Overall, 119 samples were included in the study in the following distribution: 39 healthy, 36 azotemia, 37 heart disease, and 7 CHF. In healthy and heart disease or failure groups only 1 blood sample was taken from each dog. In the azotemic group, 13, 7, and 3 dogs had 1, 2, and 3 blood samples taken at different times, respectively. In the healthy group, 17 dogs were male and 22 were female (neuter status not recorded). Of the remaining samples, 13 were from female entire, 22 female neutered, 26 male entire, 18 male neutered, and 1 unknown sex dogs. Mean age was 6 years and there was no significant difference in age among groups (range, 4 months to 15 years).
Breed was unknown for all 39 healthy dogs and for 10/86 of the remaining dogs. Of 30 breeds recorded, there were 12 boxers (6 azotemia, 6 heart disease), 6 crossbreeds (4 azotemia, 2 heart disease), 6 Cavalier King Charles Spaniels (1 azotemia, 5 heart disease), 5 Cocker Spaniels (4 heart disease, 1 CHF), and 4 Springer Spaniels (2 azotemia, 1 heart disease, 1 CHF). All other breeds were represented by 1 to 3 dogs. Diagnoses are recorded in Table 1.
|Azotemia||Heart Disease||Congestive Heart Failure|
|CKD (juvenile nephropathy)||7||DVD||13||DCM||2|
|CKD||6||ARVC||3||Mitral dysplasia and aortic stenosis||2|
|Acute pancreatitis||2||DVD and atrial fibrillation||2||DVD and atrial fibrillation||1|
|Parathyroid adenoma (hypercalcemia)||2||DVD and chronic bronchitis||2||DCM and ventricular tachycardia||1|
|Hydronephrosis||2||Subaortic stenosis||2||DCM and atrial fibrillation||1|
|Acute renal failure (unknown cause)||2||DCM||1|
|CKD (membranoproliferative glomerulonephritis)||1||DCM and atrial fibrillation||1|
|Lymphoma||1||Mitral dysplasia and aortic stenosis||1|
|Multiple myeloma||1||Patent ductus arteriosus||1|
|IMHA and MODS||1||Tricuspid dysplasia and atrial fibrillation||1|
|No final diagnosis||9||Tetralogy of Fallot||1|
|DVD and high grade 2nd degree AV block||1|
|Primary rhythm disturbances:|
|Supraventricular tachycardia and tachycardiomyopathy (1)|
|Supraventricular bigeminy (1)|
|Ventricular tachycardia (1)|
|SVT and frequent VPC (1)|
|Second degree AV block (1)|
|Third degree AV block (1)|
Median plasma creatinine concentration (range) was 1.47 (1.06–1.70), 4.36 (1.74–15.6), 1.22 (0.69–1.91), and 1.45 (0.63–1.64) mg/dL for healthy, azotemic, heart disease, and CHF groups, respectively (Fig 1). There was a statistically significant difference among the groups when compared overall (P < .001) and among azotemic dogs and healthy, heart disease, and CHF dogs (P < .001) when compared pair-wise. There was no significant difference in plasma creatinine concentration among healthy, heart disease, and CHF dogs when compared pair-wise.
Median NT-proBNP (range) was 118 (2–673), 556 (37–1,819), 929 (212–5,658), and 3,144 (432–5,500) pmol/L for healthy, azotemic, heart disease and CHF groups, respectively (Fig 2). There was a significant difference among the groups overall (P < .001) and when compared pair-wise (P≤ .049).
There was no significant correlation between concentrations of creatinine and NT-proBNP (r=−0.089, P= .338). Table 2 shows the number of dogs that would lie above or below previously published or recommended cut-off values.
|Group||Number (%) of Dogs above or below Various NT-proBNP Cut-off Levels (pmol/L)|
|Healthy||7 (18)||32 (82)||3 (8)||36 (92)||0 (0)||39 (100)|
|Azotemic||23 (64)||13 (36)||20 (56)||16 (44)||4 (11)||32 (89)|
|Heart Disease||34 (92)||3 (8)||29 (78)||8 (22)||12 (32)||25 (68)|
|CHF||7 (100)||0 (0)||6 (86)||1(14)||5 (71)||2 (29)|
When analysis was repeated on the modified data set, there was no change in P values denoting significance of any statistical tests except the following: there was a numerically small but statistically significant difference between the plasma creatinine concentration of the healthy dogs (median, 1.47 mg/dL) and those with heart disease (median, 1.17 mg/dL) (P < .001) and there was no statistical difference between the NT-proBNP results of the heart disease and CHF groups.
This study identified that NT-proBNP frequently is increased to concentrations suggestive of heart disease and occasionally to concentrations compatible with CHF in dogs with decreased GFR but no clinical evidence of heart disease. The median NT-proBNP concentration in dogs with azotemia (but no clinical evidence of CHF) was above the manufacturer's recommended cut-off for heart disease (300 pmol/L) and use of this value would have led to misclassification of 23/26 dogs in the azotemic group. Oyama8 suggested 445 pmol/L as a cut-off to distinguish dogs with heart disease from normal dogs and 1,725 pmol/L to distinguish dogs in CHF. Application of those cut-off values to the azotemic group of the present study would have led to misclassification of 16 and 4 of the 26 azotemic dogs, respectively.
Both BNP and NT-proBNP concentrations are increased in humans with decreased GFR (without concurrent cardiac disease), and this increase is more pronounced for NT-proBNP.23 For this reason, the diagnostic value of these tests in patients with renal failure is called into question.13 Recent studies suggest that NT-proBNP assays can be of diagnostic value, but modification of diagnostic cut-off values is suggested by some authors.14–16 These data suggest that a similarly cautious approach should be applied dogs.
Increased NT-proBNP in patients with renal failure is thought to be because of decreased excretion by the kidney which appears to be responsible for plasma clearance.12 Creatinine was used as a marker for glomerular filtration rate (GFR) in our study because it is widely acknowledged to be a reliable and practical marker of decreased GFR in dogs.24 However, there are instances in which plasma creatinine concentration is increased without decreased GFR (eg, recent high protein intake) or plasma creatinine concentration is normal in the presence of decreased GFR (eg, cardiac cachexia and poor muscle mass). Consequently, cases could have been included in the study for which creatinine was not a true reflection of GFR, and the authors acknowledge this limitation. In humans25 and cats,26 hypertension is associated with increased NT-proBNP concentrations. Blood pressure was not measured in the azotemic group as part of this study. Doing so might have shed light on the mechanism of increases observed and as such lack of blood pressure measurement may be seen as a limitation of the study. Another limitation of this study is that the possibility of dogs in the azotemic group having heart disease cannot be completely excluded because cost limitations precluded performance of echocardiographic evaluations in all dogs. However, clinical signs, physical examination, and investigations of renal disease did not identify any signs of heart disease and therefore this is unlikely to account for the consistent increases in NT-proBNP seen in this study.
Previous studies have identified a relationship between increased plasma creatinine concentration and increased NT-proBNP in dogs.8,9 Only 1 small study has looked specifically at a cohort of dogs without heart disease but with azotemia17 and reported comparable findings. In contrast to previous studies, the present study demonstrated no correlation between creatinine and NT-proBNP.8,9,17 This is consistent with studies in human patients.14
The azotemic dogs in this study had a range of renal and prerenal causes of azotemia and similar heterogeneity was present in the diagnoses of the heart disease and CHF groups. This range of diseases is representative of that presented to clinicians in practice. The authors suggest this test is likely to be applied to animals presenting with signs suspected to be because of cardiac diseases including primary rhythm disturbances and thus those cases were included in the prospective study. However, such conditions may not cause the myocardial stretch that is the stimulus for NT-proBNP release and so affected patients might not have increased NT-proBNP, thus decreasing the median in the heart disease and CHF groups. For that reason, data analysis was repeated on a modified data set excluding dogs with only primary rhythm disturbances; the significance of the overall differences was not altered.
Azotemia has been shown to be present in as many as 50% of dogs with heart disease9,27 and it is likely that a substantial proportion of cases in which NT-proBNP testing might be used to test for suspected heart failure would have increased plasma creatinine concentration. As a result, those cases were included in our study. However, because of fears that azotemia in dogs in the heart failure groups might have altered NT-proBNP results, the analysis was repeated without including them and this modification did not alter the significance of the overall differences among the groups. Furthermore, many noncardiac diseases result in decreased GFR because of decreased renal perfusion or primary renal disease and these cases could present with signs suggestive of heart failure (eg, tachycardia and tachypnea because of hypovolemia, anemia, and acidosis) that might warrant NT-proBNP testing as part of emergency assessment. Clinicians should therefore be aware that the presence of azotemia may affect NT-proBNP results.
This study did not examine a cohort of dogs with clinical signs suggestive of cardiac disease but rather dogs with a variety of clinical presentations. As a result, generation of sensitivity and specificity data was not deemed appropriate. However, our data clearly demonstrate that indiscriminate use of this test could lead to misdiagnosis, particularly when azotemia is present. As with any diagnostic test, the authors recommend careful consideration of the patient's clinical findings before NT-proBNP assay is used to investigate the presence of heart disease. Although the NT-proBNP assay may provide a minimally invasive, economic aid to diagnosis, it is not a substitute for thorough evaluation of physical examination findings and ancillary diagnostic test results.
In conclusion, this study demonstrated that azotemic dogs can have NT-proBNP concentrations above the previously suggested cut-off values for diagnosis of heart disease and this could to lead misdiagnosis. As a result, the authors recommend that renal status should be known when using this test clinically. If azotemia is present or if renal status is unknown, a higher cut-off value should be used for diagnosis of heart disease in azotemic dogs.
aSynchron CX5, Beckman Coulter, Fullerton, CA
bVetSign Canine CardioSCREEN NT-proBNP, Guildhay, Guildford, UK
cPowerwave 340, Biotek, Winooski, VT
dSPSS 16.0.0, SPSS Inc, Chicago, IL
The study was supported from a grant from the Cambridge Infectious Disease Consortium Clinical Research Outreach Programme. ER's residency is funded by the Alice Noakes Memorial Trust. NT-proBNP assay kits were provided by Guildhay Ltd, Guilford, UK. The authors thank Dr T. J. McKinley for help with statistical analysis.
- 7BNP Consensus Panel 2004: A clinical approach for the diagnostic, prognostic, screening, treatment monitoring, and therapeutic roles of natriuretic peptides in cardiovascular diseases. Congest Heart Fail 2004;10 (Suppl 3):1–30., , , et al.
- 13Sorting out the evidence on natriuretic peptides. Rev Cardiovasc Med 2003;4 (Suppl 4):S13–S19.,
- 17Effect of azotemia on serum N-terminal proBNP concentration in dogs with normal cardiac function: A pilot study. J Vet Cardiol 2009;11:S81–S86., , , et al.
- 18Multiple Classification Analysis, 2 ed. Ann Arbor: University of Michigan; 1973., , ,
- 20Algorithm AS 62: Generator for the sampling distribution of the Mann-Whitney U statistic. Appl Stat 1973;22:269–273.,
- 21A simple sequentially rejective multiple test procedure. Scand J Stat 1979;6:65–70.