B-type natriuretic peptide concentrations reliably distinguish between cardiac and respiratory causes of dyspnea, but its utility to detect asymptomatic cats with occult cardiomyopathy (OCM) is unresolved.
B-type natriuretic peptide concentrations reliably distinguish between cardiac and respiratory causes of dyspnea, but its utility to detect asymptomatic cats with occult cardiomyopathy (OCM) is unresolved.
Determine whether plasma N terminal probrain natriuretic peptide (NT-proBNP) concentration can discriminate asymptomatic cats with OCM from normal cats, and whether NT-proBNP concentration correlates with clinical, biochemical, and echocardiographic parameters.
One hundred and fourteen normal, healthy cats; 113 OCM cats.
Prospective, multicenter, case-controlled study. NT-proBNP was prospectively measured and cardiac status was determined from history, physical examination, and M-mode/2D/Doppler echocardiography. Optimal cut-off values were derived using receiver operating characteristic (ROC) curve analysis.
NT-proBNP was higher (median, interquartile range [25th and 75th percentiles]) in (1) OCM (186 pmol/L; 79, 478 pmol/L) versus normal (24 pmol/L; 24, 32 pmol/L) (P < .001); and (2) hypertrophic obstructive cardiomyopathy (396 pmol/L; 205, 685 pmol/L) versus hypertrophic cardiomyopathy (112 pmol/L; 48, 318 pmol/L) (P < .001). In OCM, NT-proBNP correlated (1) positively with LVPWd (ρ = 0.23; P = .01), LA/Ao ratio (ρ = 0.31; P < .001), LVs (ρ = 0.33; P < .001), and troponin-I (ρ = 0.64; P < .001), and (2) negatively with %FS (ρ = −0.27; P = .004). Area under ROC curve was 0.92; >46 pmol/L cut-off distinguished normal from OCM (91.2% specificity, 85.8% sensitivity); >99 pmol/L cut-off was 100% specific, 70.8% sensitive.
Plasma NT-proBNP concentration reliably discriminated normal from OCM cats, and was associated with several echocardiographic markers of disease severity. Further studies are needed to assess test performance in unselected, general feline populations, and evaluate relationships between NT-proBNP concentrations and disease progression.
area under the curve
blood urea nitrogen
ratio of peak early (E) diastolic to peak late (A) diastolic transmitral velocity
ratio of peak early (E) diastolic to peak early mitral annular (Ea) diastolic velocity
left ventricular shortening fraction
hypertrophic obstructive cardiomyopathy
interventricular septal thickness at end-diastole
ratio of LA to Ao dimension
limit of detection
left ventricular chamber dimension at end-diastole
left ventricular chamber dimension at end-systole
ventricular caudal wall thickness at end-diastole
N terminal-probrain natriuretic peptide
receiver operating characteristic curve
Cardiomyopathy is the leading cause of cardiac morbidity and mortality in cats, and hypertrophic cardiomyopathy (HCM) is the most prevalent myocardial disorder.[1-3] Occult cardiomyopathy (OCM) is challenging to diagnose because morphology is variable,[3, 4] and physical examination, electrocardiography, and thoracic radiography have limited sensitivity and specificity.[5, 6] Echocardiography is highly specific for diagnosing myocardial disease, although its sensitivity for detecting HCM can be limited in some circumstances. Doppler echocardiography can add additional information, but it is time consuming and technically demanding. While the number of veterinary practices that perform echocardiographic screening is unknown, relatively few veterinarians are certified in specialties that provide training in echocardiography. At the time of this writing for example, there are 224 cardiologists1 and 381 radiologists.2 By comparison, estimates suggest that 34.8 million American households owned one or more cats, and felines account for 88.3 million pets. These factors underscore limitations of echocardiography for large scale screening, and have stimulated interest to discover new methods that help detect heart disease.
Plasma concentrations of B-type natriuretic peptide (BNP) have been shown to reflect cardiac dysfunction. BNP is secreted from atrial and ventricular myocytes during myocardial stretch, pressure overload, and neurohormonal stimuli,[11-13] and blood concentrations increase during acute congestive heart failure,[14-16] asymptomatic, or minimally symptomatic diastolic and systolic dysfunction.[17-19] As such, its measurement contributes important information toward reducing cardiac morbidity, mortality, and medical costs. Increased natriuretic peptide concentrations have been detected in cats with symptomatic and asymptomatic heart disease,[14, 15, 20-23] but its reliability to detect subclinical (occult) forms of myocardial diseases in cats remains to be clarified.
Our primary objective was to test the hypothesis that plasma NT-proBNP concentrations discriminate between cats with varying forms of subclinical (occult) cardiomyopathy from healthy, normal cats. We also aimed to identify optimal circulating plasma NT-proBNP cut-off concentrations that distinguished normal from OCM cats. Secondary objectives were to determine whether plasma NT-proBNP concentrations were correlated with age, body weight, blood urea nitrogen (BUN), serum creatinine, serum thyroxine, or cardiac troponin I; if it correlated with echocardiographic measures of left atrial and left ventricular size; with Doppler echocardiographic estimates of left ventricular filling pressure; and whether blood concentrations differed between diverse forms of cardiomyopathy. In addition, we intended to determine whether plasma NT-proBNP concentrations differed between HCM cats whose hypertrophy was confined to either the ventricular septum or the left ventricular free wall, versus HCM cats whose hypertrophy included the ventricular septum and left ventricular free wall.
This prospective, multicenter trial included 18 cardiology specialists from 4 private practices and 8 veterinary medical colleges and universities.
Study protocol was approved by animal care and use committees, and owner consent to participate was obtained for all cases.
Diagnoses were made by cardiology specialists based upon medical history, physical examination, echocardiography, color-flow Doppler echocardiographic examinations, and additional tests as deemed necessary. Healthy cats were solicited from clients, students, veterinarians, and hospital employees that were examined for routine health care, vaccinations, elective surgeries such as dental procedures, or for the incentive of echocardiographic evaluation. Cats were determined to be normal if they had no medical history or examination findings indicative of heart, respiratory, systemic, or metabolic disease; if echocardiographic examination and measured parameters were within normal reference range; and if no substantial valvular regurgitation or gradients were detected by color-flow and spectral Doppler echocardiography. A systolic heart murmur was not an exclusionary criterion, providing that flow disturbances were confined to trivial valvular insufficiency, or mild dynamic right ventricular outflow tract obstruction (<30 mmHg gradient). Cats that were excluded from this normal cohort had gallop rhythm or arrhythmia, or concurrent systemic disease such as renal failure, neoplasia, hyperthyroidism, or systemic hypertension. Cats diagnosed with asymptomatic (occult) cardiomyopathy (OCM) were also recruited from clients through cardiology specialty clinics. All cats were evaluated by echocardiography with simultaneous ECG from standard views according to accepted techniques.[14, 15, 22, 24] Variables of interest included left ventricular caudal wall thickness at end-diastole (LVWd); interventricular septal thickness at end-diastole (IVSd); left ventricular chamber dimension at end-diastole (LVd); left ventricular chamber dimension at end-systole (LVs); and left ventricular shortening fraction (%FS). Left atrial (LA) and aortic root (Ao) dimensions were measured by M-mode echocardiography from the right parasternal, short axis, basilar view; ratio of LA to Ao dimension (LA:Ao) was calculated from these dimensions. Doppler echocardiography (color, pulse wave, and continuous) was performed to characterize flow disturbances and to calculate the ratio of peak early (E) diastolic to peak late (A) diastolic transmitral velocity (E/A) and ratio of peak early (E) diastolic to peak early mitral annular (Ea) diastolic velocity (E/Ea) as previously described.[15, 25, 26] Vagal maneuvers were performed in some cases to separate mitral inflow velocities and mitral annulus motion wave forms. Heart disease classification was specified a priori[3, 4, 27, 28] to identify HCM, hypertrophic obstructive cardiomyopathy (HOCM), restrictive cardiomyopathy (RCM), dilated cardiomyopathy (DCM), arrhythmogenic right ventricular cardiomyopathy, or unclassified cardiomyopathy (UCM). Left ventricular hypertrophy was defined when end-diastolic thickness of the LVW or IVS or both exceeded 6 mm. Investigators were unaware of plasma NT-proBNP concentrations.
Blood samples were collected from cats when examined by venipuncture into vacutainer tubes containing K2EDTA. Tubes were centrifuged, plasma was separated, removed, frozen, and stored at −20°C within 60 minutes of collection. Frozen plasma stored for up to 4 weeks was shipped on 2 ice packs to a central laboratory by overnight express mail. Plasma NT-proBNP concentration was measured using a commercially available, horseradish peroxidase, colorimetric end-point assay for the quantitative determination of feline NT-proBNP,3,4 as previously described. The assay uses immunoaffinity purified sheep antibody for feline NT-proBNP. The sandwich comprises a capture antibody (targeted to amino acids 1 through 20 of feline NT-proBNP), and a detection antibody (target to amino acids 60 through 80 of feline NT-proBNP). The latter was conjugated to horseradish peroxidase, and NT-proBNP was detected colorimetrically at a wavelength of 450 mm. Three feline control samples with known NT-proBNP concentrations were used to calculate intra- and inter-assay coefficients of variation. The commercial NT-proBNP control sample concentrations (mean ± range) included low, medium, and high concentrations (low: 111 ± 35 pmol/L; medium: 299 ± 68 pmol/L; and high: 952 ± 166 pmol/L).4 Actual NT-proBNP concentrations (pmol/L) were found to be 126, 277, and 866, respectively. To assess linearity, negative control material was spiked with known concentrations of feline NT-proBNP peptide starting at 2000 pmol/L and after a 1:2 dilution series down to 8 pmol/L (approximately the previously reported [limit of detection] LOD). The analytical sensitivity was determined by reading the +3 standard deviation response from 6 replicate measurements of the zero standard. The LOD of the NT-proBNP assay was 25 pmol/L and a value of 24 pmol/L was assigned to any sample with a concentration measuring <25 pmol/L. For samples with low (126 pmol/L), medium (277 pmol/L), and high (866 pmol/L) NT-proBNP concentrations, the intra-assay coefficient of variation (n = 20) was 9.7, 11.3, and 4.6%, respectively, and the interassay coefficient of variation (n = 3) was 14.2, 14.9, and 13.8%, respectively. The assay was linear in the range 20–2,000 pmol/L when using the spiked controls. The mean recovery (n = 3 experiments) was 144.1%. Additional blood tests were performed by a commercial laboratory5 including creatinine (n = 120), BUN (n = 120), thyroxine (n = 126), and cardiac troponin I6 (n = 107). LOD of the cardiac troponin I immunoassay6 was 0.2 ng/L.
Data were summarized as median and interquartile range (IQR: 25th to 75th percentiles). Proportions were compared by chi-squared test. Patient characteristics and plasma NT-proBNP concentrations between OCM and normal cohorts were compared by Mann–Whitney U-test. Comparison of NT-proBNP in OCM cats with HCM, HOCM, and UCM was made by analysis of variance (anova) with Bonferroni correction, conducted on the natural log of plasma NT-proBNP concentrations. This transformation was utilized to ensure that required assumptions to use anova were satisfied. Spearman's rank order correlation coefficient tested the strength of relationship between plasma NT-proBNP and echocardiographic parameters, age, body weight, and blood biochemical findings in OCM cats. Based on these results, a multiple linear regression was performed to determine whether NT-proBNP levels could be predicted by the study variables which were significantly correlated with it. In particular, the regression considered the dependent study variable, NT-proBNP, and the independent study variables, LA/Ao, LVWd, LVs, and FS% (troponin was not included due to the amount of missing data for this variable). Models were fit for all the cats in the study, and also separately for those with NT-proBNP levels >24 pmol/L. Receiver operating characteristic (ROC) curve analysis was used to assess the capability of plasma NT-proBNP concentrations to predict cardiac status (normal versus OCM), and to help estimate the optimal NT-proBNP cut-off concentration that would best classify cats correctly. We determined that 45 cats were required in each cohort to demonstrate that an area under curve (AUC) of 0.80 was significantly different than the null hypothesis (AUC = 0.50, no discriminating power), with a 0.05 probability of either a type I or type II.7 Results were analyzed using commercially available statistical software.7,8P values <.05 were considered significant.
Study animals: There were 114 normal, healthy cats and 113 OCM cats (Table 1). Ages (median; IQR) were slightly but significantly less in normal (6.0 years; 3.0, 8.2 years) than OCM (8.7 years; 4.0, 12.0 years) cats (P < .001). More males were in the OCM cohort (P < .001). Body weight was not statistically different in normal (4.9 kg; 4.1, 5.6 kg) versus OCM (5.1 kg; 4.3, 6.2 kg) cats (P = .09). A systolic cardiac murmur was detected in 49 (43%) normal cats (intensity range, 1–3/6; median intensity, 2/6), and in 90 (79.6%) OCM cats (intensity range, 1–5/6; median intensity, 2.5/6). A gallop rhythm was detected in 19 (16.8%) OCM cats. Cats with OCM represented 52 HCM (46.0%), 35 HOCM (31.0%), 22 UCM (19.5%), three RCM (2.6%), and one DCM (0.9%).
|Normal Cats n = 114||Occult Cardiomyopathy Cats n = 113|
|Domestic short hair||77||67.5||80||70.8|
|Domestic long hair||13||11.4||9||8.0|
|Maine Coon cat||4||3.5||5||4.4|
Plasma NT-proBNP concentration (median; IQR) in normal cats was not statistically different between males (24 pmol/L; 24, 32 pmol/L) and females (24 pmol/L; 24, 33 pmol/L) (P = .93), or in OCM cats between males (188 pmol/L; 79, 533 pmol/L) and females (170 pmol/L; 79, 380 pmol/L) (P = .62). Plasma NT-proBNP concentration (median; IQR) was significantly higher in OCM cats (186 pmol/L; 79, 478 pmol/L) versus normal cats (24 pmol/L; 24, 32 pmol/L) (P < .001) (Fig 1); 99 pmol/L was the highest NT-proBNP concentration recorded in the normal cohort. Plasma NT-proBNP (median; IQR) was significantly higher in OCM when LA:Ao ratio ≥1.5 (285 pmol/L; 141, 715 pmol/L) versus when LA:Ao ratio <1.5 (123 pmol/L; 70, 346 pmol/L) (P = .002). Plasma NT-proBNP concentration (median; IQR) was significantly higher for HOCM (396 pmol/L; 205, 685 pmol/L) versus HCM cats (112 pmol/L; 48, 318 pmol/L) (P < .001), but neither group was statistically different from cats with UCM (P = .68). Plasma NT-proBNP concentration did not differ significantly for OCM cats with hypertrophy of both the ventricular septum and left ventricular caudal wall, versus those with only ventricular septum or the left ventricular caudal wall hypertrophy, for cats with HCM (P = .81) or with HOCM (P = .73). Of the 113 OCM cats, 15 had plasma NT-proBNP concentrations that were at or near the LOD (<25 pmol/L, n = 8 versus 28–42 pmol/L, n = 7, respectively). In these 15 cats, the distribution of hypertrophy was confined just to the interventricular septum in 10 (67%); to the left ventricular caudal wall in 1 (7%); and both the interventricular septum and left ventricular caudal in the remaining 4 (27%). The magnitude of maximal diastolic wall thickness in affected regions of these 15 cats was relatively mild (7.0 mm or less) in 9; slightly thicker (7.2 and 7.3 mm) in 2 cats, and in the remaining 4 cats, maximal hypertrophy was severe (8.3 mm) in 1 cat.
In cats with OCM, plasma NT-proBNP was positively correlated (2-tailed test) with cardiac troponin I (ρ = 0.64; P < .001), LVs (ρ = 0.33; P < .001), LVWd (ρ = 0.23; P = .015), and LA/Ao ratio (ρ = 0.31; P < .001) and negatively correlated (2-tailed test) with %FS (ρ = −0.27; P = .004) (these variables also remained statistically significant when NT-proBNP values at the LOD were eliminated from this analysis); and not significantly correlated with age (P = .38), body weight (P = .51), thyroxine (P = .97), creatinine (P = .89), BUN (P = .98), LVd (P = .09), IVSd (P = .27), E/A ratio (P = .71), or E/Ea ratio (P = .56).
Multiple linear regression was used to determine whether the concentration of NT-proBNP could be predicted by any of the independent study variables (LVs, %FS, LVWd, and LA/Ao ratio) that were shown to be significantly correlated to the dependent variable, plasma NT-proBNP. One model included all diseased cats (112 complete cases) and another model included only diseased cats with NT-proBNP levels >24 pmol/L. In both models, all the independent variables were insignificant (P > .05), although there was evidence of high levels of correlation between %FS and LVs which might have impacted the ability to determine significance.
The ROC curve analysis using all 227 cases revealed that plasma NT-proBNP concentration was useful to distinguish OCM from normal cats (Fig 2). Area under the ROC curve was 0.92. Table 2 lists selected NT-proBNP cutoff values that provide a range of sensitivity and specificity.
|Criterion: NT-proBNP (pmol/L)||Sensitivity (%) (95% CI)||Specificity (%) (95% CI)|
|≥24||100.0 (96.8–100.0)||0.0 (0.0–3.2)|
|>34||88.5 (81.1–93.7)||80.7 (72.3–87.5)|
|>46||85.8 (78.0–91.7)||91.2 (84.5–95.7)|
|>68||80.5 (72.0–87.4)||97.4 (92.5–99.4)|
|>99||70.8 (61.5–79.0)||100.0 (96.8–100.0)|
Plasma NT-proBNP concentrations were significantly increased in cats with asymptomatic (occult) cardiomyopathy compared to normal cats. There was no influence of age, body weight, sex, serum creatinine or thyroxine concentrations on this relationship. Plasma NT-proBNP was independent of the study variables LA/Ao, LVWd, IVSd, LVd, LVs, and age to predict the presence of OCM over normal cats. Cutoff value >46 pmol/L provided approximately 91% specificity and 86% sensitivity to discriminate between these cohorts; >99 pmol/L cutoff value improved specificity to 100%, while reducing sensitivity to 71%. Overall, these results compared favorably to those of a smaller study that reported 100% sensitivity and 89% specificity using >49 pmol/L cutoff value.
While plasma NT-proBNP concentrations correlate with severity of hypertrophy in human HCM,[29, 30] this relationship was comparatively weaker in the present study. Plasma NT-proBNP was likely to be increased in HCM cats when substantial thickening of the interventricular septum or left ventricular caudal wall was present. However, plasma NT-proBNP concentrations were not significantly different in cats that had hypertrophy of both ventricular septum and left ventricular caudal wall, compared with those that had hypertrophy of only the ventricular septum or left ventricular caudal wall. Recently, plasma NT-proBNP was reported to be significantly higher in severely hypertrophied cats compared with HCM cats with lesser left ventricular thickening. However, the majority of these severely affected HCM cats had congestive heart failure. As feline plasma NT-proBNP concentrations can be markedly increased in cats with heart failure,[14, 15] the correlation between left ventricular hypertrophy and BNP concentrations requires further analysis. Others have investigated Maine Coon cats with inherited HCM, reporting that NT-proBNP distinguished severe, but not mild left ventricular hypertrophy. However, the applicability of data from a closed, inbred colony to the greater, worldwide, feline population has been questioned. In the present study plasma NT-proBNP concentrations were at or near the LOD in 15 of the 113 OCM cats and in 14 of these 15 cases, the magnitude of left ventricular hypertrophy was relatively mild. Nevertheless, >46 pmol/L cutoff conferred approximately 86% sensitivity and 91% specificity in this study population, suggesting that overall, false negative results were relatively low.
Cats with HOCM had significantly higher plasma NT-proBNP concentration than cats with the non-obstructive form of HCM. Similar findings have been reported in humans where HOCM patients had higher median NT-proBNP concentrations than those with HCM, and several possible explanations have been proposed. Dynamic obstruction increases wall stress, a stimulus for natriuretic peptide secretion. Furthermore, diastolic dysfunction and left ventricular outflow tract obstruction in HCM also increases NT-proBNP's expression in ventricular myocytes.[29, 30]
Factors other than diastolic wall thickness may have also influenced expression of plasma NT-proBNP in the present study, as circulating peptide concentrations were significantly higher in UCM compared with normal cats. The UCM cohort had left ventricular end-diastolic thickness below the 6-mm cutoff used to designate HCM. Moreover, plasma NT-proBNP concentrations were positively correlated with LA/Ao ratio, LVWd and LVs, and negatively correlated with percent %FS in OCM cats. Together, these observations suggest that circulating NT-proBNP concentrations might reflect disease progression, or point toward other cardiac structural or functional changes. Such potential relationships between plasma NT-proBNP concentrations and echocardiographic variables, and their implications for risk assessment, require further study.
Detection of OCM is challenging in cats.[5, 6] Indeed, a substantial number of asymptomatic cats in the present study had heart murmurs, but cardiac status was judged to be normal after echocardiographic examination. Plasma NT-proBNP concentration reliably discriminated between OCM and normal cohorts with relatively high sensitivity and specificity. Moreover, this assay provided good clinical utility to detect a mixed population of cats with both hypertrophic and non-hypertrophic forms of OCM. As cats with plasma NT-proBNP concentrations >46 pmol/L were likely to have OCM, it is reasonable that the use of this assay might help guide more efficient utilization of echocardiography. An integrated approach using biochemical testing and echocardiography might better characterize affected cats, help formulate monitoring strategies, or identify disease progression. In fact, considerable benefit has been reported for risk stratification across all stages of human heart failure by combining BNP evaluation with echocardiography. Such applications in cats await corroborative studies.
There were several study limitations. Firstly, the clinical utility of a diagnostic test is influenced by disease prevalence. For screening tests, the rate of false positive and negatives are minimized in populations with high disease prevalence. It must be pointed out, therefore, that the present study included cases that were referred to cardiology specialists based upon a degree of suspicion that they might have heart disease, and thus, represent populations in which the clinical utility of NT-proBNP screening is most likely to benefit. Therefore, further studies are needed to assess performance characteristic of this assay when used in general, unselected populations. Secondly, inter-assay coefficient of variation (14.2 and 14.9% at low and moderate NT-proBNP sample concentrations, respectively) could conceivably influence the designation of normal or disease status when NT-proBNP concentrations are close to specific cutoff values. Repeat biochemical assessment or other diagnostic testing may be warranted in these circumstances. Thirdly, echocardiographic data were entered as submitted by participating cardiology specialists, as it was not feasible to utilize a central investigator to confirm echocardiographic measurements. Furthermore, we did not consider whether therapies affected plasma NT-proBNP concentrations.
Nevertheless, the present study demonstrated that statistically higher plasma NT-proBNP concentration were present in cats with OCM compared to normal cats. This supports the concept that plasma NT-proBNP is a useful feline cardiac biomarker, whose measurement may contribute as an adjunctive test to screen for heart disease.
We thank Dr Sam Woolford, Bentley College, for statistical assistance.
Investigators Drs Philip R. Fox, Mark A. Oyama, John E. Rush, Clark E. Atkins and Sonya G. Gordon consult for IDEXX Laboratories.
Dr Mark A. Oyama has received residency training support from IDEXX Laboratories.
Dr Philip R. Fox is on advisory boards for IDEXX Laboratories and Ceva Santé Animale.
American College of Veterinary Internal Medicine, http://www.acvim.org/websites/acvim/index, accessed August 4, 2011
American College of Veterinary Radiology, http://www.acvr.org/members/history/index.html, accessed January 3, 2011
CardiopetTMproBNP, IDEXX Laboratories, Westbrook, ME
Biomedica Gruppe, Vienne, Austria
IDEXX Laboratories, Westbrook, ME
Immulite® 2000 cTnI Assay, Diagnostic Products Corp, Los Angeles, CA
MedCalc Software, Vers. 188.8.131.52, Mariakerke, Belgium
SPSS Software Vers. 15, Chicago, IL