Point‐of‐care N‐terminal pro B‐type natriuretic peptide assay to screen apparently healthy cats for cardiac disease in general practice

Abstract Background Point‐of‐care (POC) N‐terminal pro B‐type natriuretic peptide (NT‐proBNP) ELISA test has been evaluated for screening cats for cardiac disease in the referral veterinary setting but less is known about its use in general practice (GP). Objectives To evaluate the diagnostic utility of a POC NT‐proBNP ELISA in cats seen in GPs. Animals Two hundred and seventeen apparently healthy cats from 21 GPs. Methods This was a prospective, cross‐sectional study. Cardiac auscultation and POC NT‐proBNP ELISA were done by veterinarians at their GPs. After enrollment at GPs, cats were sent to a cardiology referral hospital for cardiac auscultation and echocardiographic diagnosis. Results were interpreted based on whether cats had normal or abnormal echocardiographic findings. Results Point‐of‐care NT‐proBNP ELISA results differentiated cats in the abnormal group from those in the normal group with a sensitivity of 43%, specificity of 96%. In cats with a heart murmur at GPs, POC NT‐proBNP ELISA results differentiated cats in the abnormal group from those in the normal group with a sensitivity of 71% and a specificity of 92%. Conclusion and Clinical Importance In apparently healthy cats in GPs, positive POC NT‐proBNP results are associated with heart disease, warranting an echocardiogram, but negative results do not reliably exclude heart disease. These results suggest POC NT‐proBNP is not an effective screening test for apparently healthy cats in GPs, although its performance is improved if it is used only in cats that have a heart murmur.

K E Y W O R D S biomarker, feline, general practice, heart disease, NT-proBNP

| INTRODUCTION
The point-of-care (POC) N-terminal pro B-type natriuretic peptide (NT-proBNP) ELISA test has the potential to be a convenient tool for screening cats for cardiac disease in general practice (GP). Previous studies evaluated its sensitivity and specificity in cats presented to veterinary teaching hospitals, but the samples studied in these investigations were different from feline samples in GP in important ways.
For example, the prevalence of heart disease in past studies of the POC NT-proBNP ELISA (55%, 1 60% 2 ) was higher than the prevalence of hypertrophic cardiomyopathy in the general feline sample (approximately 15%). 3,4 Cats in previous studies underwent POC NT-proBNP testing when they were suspected of having heart disease, but in the GP setting, veterinarians also might use such testing when they want more information on an apparently healthy cat's cardiac status in the absence of auscultatory abnormalities, such as during a preoperative exam. Furthermore, studies have not systematically evaluated the role of physical examination findings, such as the presence or absence of a heart murmur, when considered in tandem with POC NT-proBNP ELISA results.
The primary aims of the present study were: (a) to evaluate the diagnostic utility of a POC NT-proBNP ELISA in a GP cat sample and (b) to evaluate the role of cardiac auscultation when considered in tandem with POC NT-proBNP ELISA results in the same sample. Secondary aims were: (a) to assess the prevalence of heart murmurs and of echocardiographic abnormalities in apparently healthy cats and (b) to compare the number of cats with cardiac auscultatory abnormalities detected by GP veterinarians with the number of cats with such abnormalities detected by a residency-trained investigator in a referral veterinary center.

| Study sample
The study was a prospective, cross-sectional study approved by the Cardiospecial Veterinary Hospital ("referral hospital," RVH) Institutional Animal Care and Use Committee (No. 201701). Written, informed owner consent was obtained from the owner of every cat.
Cats that were apparently healthy were recruited prospectively from 21 GPs from April 2017 to January 2019. Cats were considered apparently healthy if they had no clinical signs of illness, no abnormal physical exam findings (other than a heart murmur, gallop sound, arrhythmia, or both gallop sound and arrhythmia) and no history of a medical concern that had required any treatment or diagnostic intervention within 30 days. Exclusion criteria included age <1 year, serum [creatinine] >2.8 mg/dL, 2,5 systolic arterial blood pressure (SBP) >170 mm Hg, serum total [T4] > 4.0 μg/dL, ongoing medication administration, intolerance to restraint for blood sampling, echocardiography, or both, and >60 days between enrollment at GPs and echocardiography at the RVH. Cardiac auscultatory abnormalities were not an exclusionary criterion.

| Experimental protocol
Cats were enrolled when they visited their GPs. The decision to enroll a cat was made by veterinarians at GPs in conjunction with the cats' owner. The stated purpose and benefit of participating in the study was a free heart evaluation at the RVH; no additional incentive was offered. After enrollment, cardiac auscultation was done by veterinarians at GPs. GP veterinarians were asked to identify whether a murmur was present and specifically to choose 1 of the 3 following options: no murmur, systolic murmur, or timing unknown murmur. If a murmur was present, GP veterinarians were asked to note its intensity. 6  After enrolling cats at GPs, veterinarians submitted the signalment information, cardiac auscultation findings, and blood work results of each cat through an online form, and shipped frozen (À20 C) samples consisting of at least 0.5 mL of plasma via freezer truck to the RVH within 2 days (frozen on site at GPs and during transit). After receiving the plasma sample and the cat's baseline information at enrollment from the GP, the RVH contacted the owner and scheduled a cardiac exam date. The plasma sample was stored frozen at À20 C until shipped on dry ice to an external commercial laboratory (IDEXX Reference Laboratories, Westbrook, Maine) for batch measurement of plasma quantitative NT-proBNP by ELISA. 7 Upon presentation at the RVH, cats rested in an exam room for ≥5 minutes for acclimation. Then, a high-definition oscillometric (HDO) blood pressure (BP) monitor (Vet HDO Monitor S + B medVet GmbH, Babenhausen, Germany) was used for performing BP measurements, with the cuff placed around the tail base. The cuff size was selected according to consensus statement guidelines. 8   echocardiographic images beyond those already described, was confirmed via measurement of wall thickness on such images and was recorded. The left atrial to aortic ratio (LA : Ao) was calculated from measurements made on the right parasternal short-axis view images at the end of ventricular systole. 11 Abnormal cats were classified as having hypertrophic cardiomyopathy (HCM); hypertrophic obstructive cardiomyopathy (HOCM); unclassified cardiomyopathy (UCM); or noncardiomyopathic heart disease, which included valvular disease, congenital heart malformations, and primary arrhythmia without echocardiographic abnormalities. HCM was defined as regional or general- In cases where determination, analysis, or both, of echocardiographic findings was/were challenging, a second investigator (YH) reviewed the case and the final result was determined by agreement between the 2 investigators. All investigators were blinded to POC NT-proBNP ELISA results until echocardiographic results had been tabulated and related calculations had been completed.

| RESULTS
Two hundred and forty-five cats initially were enrolled by GP veterinarians. Twenty-eight cats were excluded due to failure to follow with heart rate 91 beats/min due to 2nd degree AV block (n = 1).
The normal group consisted of 168 cats without echocardiographic abnormalities and the abnormal group consisted of 49 cats.

À1-10
Creatinine (mg/dL) .  Similar findings were also noted when comparing POC NT-proBNP ELISA results between cats with normal auscultation findings and cats with abnormal auscultation findings at the RVH (Table 3).

| DISCUSSION
The results of the present study suggest that POC NT-proBNP is not an effective screening test when applied to cats regardless of test on a cat with a heart murmur is more likely to produce diagnostically useful information than performing it on a cat without a heart murmur. Even so, the combination of ausculting a heart murmur and obtaining a positive POC NT-proBNP test result would be expected to detect heart disease in a cat with no overt clinical cardiovascular signs in approximately 7 out of 10 cases. These results signify that using POC NT-proBNP testing in cats without heart murmurs seen in GPs is unlikely to help GP veterinarians distinguish between cats with heart disease and cats without heart disease. When a POC NT-proBNP result is positive, however, it is highly likely that the cat has echocardiographically-identifiable heart disease, which can be of a mild, moderate, or severe degree.
The present results (sensitivity: 43%; specificity: 96%) indicated that POC NT-proBNP ELISA had lower sensitivity and higher specificity for detection of heart disease in apparently healthy cats when compared to 1 earlier study (sensitivity: 84%; specificity: 83%) 2 and similar results compared to another study (sensitivity: 65%; specificity: 100%). 1 The differences between our results and those of previous studies could be due to differences in cat selection. Previous studies 1,2 were performed in teaching hospital settings where cats could have been referred due to a suspicion of heart disease. In order to reflect the clinical scenario of how veterinarians use POC NT-proBNP ELISA in GPs in our region, we designed this study to recruit cats directly at GPs and we asked veterinarians to perform on-site POC NT-proBNP ELISA testing. Thus, we did not select for cats with suspected heart disease. Another possible explanation for the differences between the present results and past results from similar studies could be the selection criteria for normal and abnormal groups. In 1 earlier study, the normal group included cats with echocardiographically normal hearts, equivocal changes, and evidence of mild heart disease, and the abnormal group included cats with echocardiographic evidence of moderate and severe heart disease. 2 In another study, there was no evaluation of disease severity and the normal group consisted of cats with normal echocardiographic findings, whereas the abnormal group included cats with all types of echocardiographic abnormalities. 1 In the present study, in order to reduce misdiagnoses in cases with left ventricular wall measurements that were neither convincingly normal nor convincingly abnormal, we described these cases as equivocal and included them in the normal group for analysis. Despite these differences in study design and cohort composition, our results were similar to those of 1 of the studies. 1 For feline occult cardiac disease, echocardiography currently is the gold-standard diagnostic tool, 14   The prevalence of heart murmurs in overtly healthy cats, and correlation to echocardiographic findings, has been studied in referral veterinary hospitals 3,17-20 and rehoming shelters. 4 The findings of this study provide insights on the existence of heart murmurs and structural heart disease in apparently healthy cats presented to GPs for veterinary care. These findings regarding the prevalence of heart murmurs are comparable to those described in some studies elsewhere in the world 3,17 and not others. 4 Differences also exist in the prevalence of echocardiographically-demonstrated LVH, which has been identified in 14.7% and 15% of apparently healthy cats in previous studies 3,4 compared with 23% (49/217) in the present study. A possible explanation for a higher prevalence of heart disease in the present study could be the increased prevalence of cardiomyopathy in older cats, as noted previously, 4 since cats in the abnormal group in the present study were significantly older than cats in the normal group.
The identification of heart murmurs by specialists examining apparently healthy cats is well-described. 3,4,[17][18][19][20][21][22] However, assessment of the prevalence of heart murmurs in apparently healthy cats by veterinarians in GPs, as described in the present study, has received limited attention 23 despite the primary role of GPs in identifying heart disease in cats. The results of cardiac auscultation performed by GP veterinarians and by the principal investigator showed minimal agreement, with a kappa value of 0.24. These results indicate that some veterinarians in this study were able to identify heart murmurs and these murmurs tended to be of higher intensity. Results also suggested that GP veterinarians were limited in their ability to detect many heart murmurs. However, such generalizations fail to take into account that systolic heart murmurs in cats often are labile (and the intensity -or even presence-of a heart murmur at 1 moment in time in a cat does not determine cardiac auscultation findings in the same cat later on), and that some GP veterinarians might have had stronger, or weaker, cardiac auscultatory skills than others.
Other cardiomyopathies were absent from the cats in this study. This seems unlikely given the stability of NT-proBNP for 2 years when frozen at À20 C, 24 the correlation between quantitative NT-proBNP and POC NT-proBNP ELISA, and the fact that there were more samples with high quantitative NT-proBNP results during the first half of the sampling period (longer storage) than the second half. Another possible source of error could be the imperfection of 2-dimensional and Mmode echocardiography for detection of cardiac disease in cats. Echocardiography is the current gold standard test to diagnose feline heart disease, but there are some limitations of this test including intraobserver variation, 16 the existence of HCM-associated changes that might not be fully expressed as an HCM phenotype, 25,26 and that the severity of abnormalities is quantified in different ways. These limitations of echocardiography could all affect the final diagnosis. In this study, all echocardiographic examinations were performed by a single investigator who was blinded to NT-proBNP test results. This protocol did not formally evaluate intraobserver variation in the performance and interpretation of echocardiograms, however. Finally, the semiquantitative method for categorizing the severity of echocardiographic changes was adapted from a previous study, 12 but did not undergo prospective validation nor was it based on survival data.