Cardiovascular-renal axis disorder and acute-phase proteins in cats with congestive heart failure caused by primary cardiomyopathy.

Abstract Background Currently, the pathogenesis of congestive heart failure (CHF) in cats is not fully understood. Objective To identify novel biomarkers for CHF in cats caused by primary cardiomyopathy, particularly related to cardiovascular‐renal axis disorder and systemic inflammatory response. Animals Twenty‐five cats in CHF caused by primary cardiomyopathy, 12 cats with preclinical cardiomyopathy, and 20 healthy controls. Methods Case control and observational case series. The following serum biomarkers were compared among the 3 cat groups: a cardiorenal profile that included N‐terminal pro‐brain natriuretic peptide (NT‐proBNP), symmetric dimethylarginine (SDMA), and creatinine and an inflammatory profile that included 7 acute‐phase proteins (APPs). Survival analyses and longitudinal studies were performed in CHF cats. Results All cardiorenal biomarkers were positively correlated and higher in CHF cats, and high NT‐proBNP and SDMA were associated with poor clinical outcome. Cats with CHF had significantly higher leucine‐rich alpha‐2‐glycoprotein 1, serum amyloid A, and ceruloplasmin, and these APPs were positively correlated with NT‐proBNP and left atrial size. In a multivariable survival analysis, alpha‐1‐acid glycoprotein concentration (P = .01), body weight (P = .02) and left atrial‐to‐aortic root ratio (P = .01) were independent prognostic factors for CHF in these cats. Conclusions and Clinical Importance In cats, CHF is an inflammatory disorder and outcome in CHF may be determined by the extent of inflammation and possibly the amount of residual renal function. These novel biomarkers have potential use for the clinical management, prognosis, and future research into CHF and cardiomyopathy in cats.

ceruloplasmin, and these APPs were positively correlated with NT-proBNP and left atrial size. In a multivariable survival analysis, alpha-1-acid glycoprotein concentration (P = .01), body weight (P = .02) and left atrial-to-aortic root ratio (P = .01) were independent prognostic factors for CHF in these cats.
Conclusions and Clinical Importance: In cats, CHF is an inflammatory disorder and outcome in CHF may be determined by the extent of inflammation and possibly the amount of residual renal function. These novel biomarkers have potential use for the clinical management, prognosis, and future research into CHF and cardiomyopathy in cats. provide information on disease processes as well as assist in diagnosis and clinical management. 1,2 To date, commercially available cardiac biomarkers in veterinary medicine are limited to N-terminal pro-brain natriuretic peptide (NT-proBNP) and cardiac troponin I (cTnI). 3 Studies on cardiovascular-renal axis disorder (CvRD) and inflammatory responses have the potential to identify novel biomarkers for cats with congestive heart failure (CHF).
In humans, renal dysfunction is an important predictor of heart failure outcome. [4][5][6] In dogs and cats, azotemia is consistently seen in patients in CHF, 7-11 with a higher incidence observed in more advanced disease. [9][10][11] Although monitoring renal function is considered crucial in management of CHF, 12 the impact of renal function on CHF outcome has not been well documented in cats. Conventionally, serum creatinine concentration is used for monitoring renal function in patients with cardiac disease as a marker of glomerular filtration rate (GFR), but its sensitivity and specificity are inferior to the newer marker of GFR, symmetric dimethylarginine (SDMA). [13][14][15][16] Unlike creatinine, SDMA is not affected by body lean mass, 17 and it can detect renal dysfunction earlier than does serum creatinine concentration. 18,19 Apart from a recent study comparing SDMA in cats with hypertrophic cardiomyopathy (HCM) and primary renal disease, 2 the biomarker role of SDMA in cats with heart disease has not been reported.
Our objective was to identify novel biomarkers of CHF in cats to improve understanding of the pathogenesis of the syndrome of CHF and the implications of GFR and inflammatory status for clinical management. We investigated 2 panels of serum markers: a cardiorenal profile and an APP profile.
We hypothesized that the selected biomarkers would (1) be increased in cats with CHF; (2) be associated with survival in cats with CHF; and (3) correlate with previously established cardiac biomarkers and prognostic indicators in cats with cardiomyopathy.

| Study design, animals, and clinical records
The study contained both retrospective and prospective components: NT-proBNP, and all had concentrations <100 pmol/L. 47,48 Relevant clinical records of the cardiomyopathy cats were collected from the hospital archives or electronic information system (Excelicare, AxSys Technology, Paisley, UK). The following details were extracted for each cat at entry into the study: signalment, date and age at initial diagnosis, cardiomyopathy diagnosis, CHF diagnosis, and comorbidities. Relevant echocardiographic parameters were examined for diagnosing cardiomyopathies, and 5 of them, which have been reported to be associated with severity and prognosis of cardiomyopathy in cats, were recorded for analyses: left atrial (LA) diameter, LA-to-aortic root ratio (LA/Ao), left ventricular free wall thickness at end diastole (LVFWd), interventricular septum thickness at end diastole (IVSd), and left ventricular fractional shortening (LV FS). 46,49 The following clinical information was recorded at the time of blood sampling: sampling date, age at sampling, body weight, heart rhythm by auscultation, gallop sounds, heart murmur, pulse deficit, heart rate, respiratory rate, ECG findings, CHF severity score based on International Small Animal Cardiac Health Council (ISACHC) classification 50,51 (IIa = 1; IIb = 2; IIIa = 3; IIIb = 4), cTnI (if measured for diagnostic purpose), and cardiac medications.
Survival information of the CHF cats was collected from the hospital record system or by contacting the referring veterinary practice or client. For cats that died, survival in days was defined as the time between initial diagnosis and the date of death. For cats still alive at the end of the study or lost to follow-up, date of last contact was used as the censoring time in the statistical analysis. Causes of mortality were recorded and divided into cardiac and noncardiac causes. The cardiac causes included cats that died or were euthanized because of cardiac disease. Overall CHF stability during the study period was summarized for each cat. A "stable" status was defined as 1 of 2 scenarios: a cat in ISACHC class IIIa that was stable on treatment and did not redevelop heart failure during the study, or a cat that was in ISACHC class IIIb heart failure at the time of inclusion into the study that did not require rehospitalization. A "nonstable" status was defined as a cat that reached a cardiac event. An event was defined as either death (cardiac death or euthanasia prompted by cardiac disease that substantially affected quality of life), progression in ISACHC class, or when hospitalization was required at each visit. Other recorded information included frequency of thoracocentesis, total days in the intensive care unit (ICU), spontaneous echocardiographic contrast (SEC), an intracardiac thrombus seen on echocardiography, arterial thromboembolism (ATE), and cardiac medications received during the study.

| Blood sampling, preparation, and storage
Blood samples were collected from cardiomyopathy cats as soon as possible after initial diagnosis. For the CHF cats enrolled in the longitudinal study, subsequent sample collections were performed at follow-up revisits. Blood was sampled by venipuncture for clinical diagnostic purpose with an additional volume of 0.3-1.5 mL blood obtained for serum biomarker measurements. Following collection, blood samples were allowed to clot and then centrifuged at 9000 rpm for 3 minutes at room temperature. Serum was separated and transferred to a new collection tube and stored at −20 C before transfer to a −80 C freezer for later analysis. Control serum samples from healthy cats were stored using the same protocol.

| Measurement of CvRD biomarkers and APPs
Ten serum markers were examined, including 3 biomarkers associated with CvRD (cardiac marker NT-proBNP, renal markers SDMA, and creatinine) and 7 APPs: alpha-1-acid glycoprotein (AGP), C-reactive protein The NT-proBNP concentration was measured using the Cardiopet proBNP assay, SDMA was measured using IDEXX SDMA (enzyme immunoassay), and creatinine was measured using a kinetic color test (compensated Jaffé reaction). In addition, NT-proBNP concentrations from 9 cats (3 CHF and 6 preclinical cardiomyopathy), sampled and measured in 2014, were included retrospectively. These measurements used the same methodology and the test was performed in the IDEXX Reference Laboratory.

| Data analyses
Before statistical analysis, data were assessed for normality using an Anderson-Darling test, and natural logarithm, square root, or rank transformations subsequently were used to normalize raw data distributions.
Clinical variables of CHF and preclinical cats at admission were analyzed using 1-way analysis of variance (ANOVA) or t tests for quantitative data comparison and chi-square or Fisher exact tests for categorical data. Baseline biomarker concentrations were analyzed using ANOVA followed by multiple pairwise t tests, with Bonferroni correction of post hoc P values. The correlation between biomarkers and clinical variables was assessed using scatter plots and calculating Spearman's rho.
Candidate prognostic markers were compared between surviving and dead CHF cats using independent t tests. Cox proportional hazards' models were used to estimate the association between biomarkers and other variables with survival time in CHF cats. Univariate screening models were used and variables with significance at P < .20 were selected for multivariable modeling employing a backwards stepwise process. Spearman's rho was used to evaluate collinearity between selected variables. When variables were collinear, the variable with the weaker univariate association with survival was excluded from the multivariable model. Significant variables were analyzed using receiver operating characteristic (ROC) curve analysis to determine the most accurate cutoff for identifying the nonsurviving cats based on the largest Youden index. Variables were removed from multivariable models one-by-one based on the largest Wald P value until all remaining variables were significant at P < .05.
Interaction terms were not evaluated.
The following commercially available software was used for sta- Statistical significance was set at P < .05.

| Study population
In total, 37 cardiomyopathy cats were enrolled in the study after exclusion, with 25 in CHF and 12 in preclinical stage. Signalment, body weight, and clinical information are listed in Table 2. Healthy control cats were significantly younger than the CHF and preclinical cardiomyopathy cats

| Baseline biomarker concentrations in CHF, preclinical cardiomyopathy, and healthy control cats
All CvRD biomarker concentrations were higher in CHF cats compared to healthy controls. The NT-proBNP and SDMA concentrations were higher in cats with CHF compared to those with preclinical cardiomyopathy (Figure 1). The NT-proBNP concentration also significantly differentiated preclinical cats from healthy controls. In the CHF group, 61% of cats had abnormal serum SDMA concentration (>14 μg/ dL), and 44% had abnormal serum creatinine concentration (IDEXX laboratory reference range, <168 μmoL/L).

Concentrations of 3 APPs (LRG1, SAA, and ceruloplasmin)
were significantly higher in CHF cats compared to healthy controls ( Figure 2A-C). Among these, SAA and ceruloplasmin were higher in CHF cats compared to cats with preclinical cardiomyopathies. No significant differences in AGP, CRP, Hp, and PCT were found among the 3 cat groups ( Figure 2D-G). The concentration of PCT was zero in most cats (24 of 37), regardless of cardiac disease status.
The APPs LRG1, SAA, ceruloplasmin, and AGP were positively correlated with the 2 cardiac biomarkers, LA size, and CHF class (P < .05;

| Survival analysis in CHF cats
For all-cause mortality, 5 variables determined at admission were significantly different in cats that died compared with cats that survived (Table 5). These included increased LA diameter and LA/Ao ratio, decreased LV FS, increased NT-proBNP, and increased SDMA.

| DISCUSSION
The results of our study indicate that both CvRD and an inflammatory response occurred in cats with CHF caused by primary cardiomyopathy. Secondly, several biomarkers were associated with a poor outcome and have the potential to be considered as survival prognosticators in cats with CHF, warranting further investigation.
The presence of CvRD in CHF cats was supported by simultaneous increases and correlations of cardiac biomarker NT-proBNP and renal biomarkers, SDMA, and creatinine. This finding suggests a close functional relationship between the heart and kidney in heart failure, which has been described in other species. 5 Note: Six cats had missing data in one or more of the analyzed variables and therefore could not be included. Abbreviations: AGP, alpha-1-acid glycoprotein; CHF, congestive heart failure; CI, confidence interval; HR, hazard ratio; LA/Ao ratio, left atrial-to-aortic root ratio; N, number; PE, parameter estimate. *Statistical significance at P < .05; **statistical significance at P < .01.
azotemia in CHF is thought to be a reflection of CvRD, which can result either from intrinsic renal insufficiency or the iatrogenic consequences of medical management. 54 In our study we could not differentiate the 2 causes, because all CHF cats were treated at enrollment.
The use of diuretics and angiotensin-converting enzyme inhibitors can cause prerenal azotemia as well as renal injury. 55 It is possible that the increased serum SDMA and creatinine concentrations in CHF cats were mainly caused by higher diuretic doses in more severe CHF cases, but we did not detect any significant correlations between diuretic dose and renal markers in 13 CHF cats in a preliminary study (data not shown). Cardiovascular-renal disorder might not be present in the early stages of cardiomyopathy in cats, at least based on currently used marker panels, because we did not identify significant differences in serum SDMA and creatinine concentrations between preclinical cardiomyopathy cats and healthy controls. Similarly, a previous publication showed no significant increase in serum SDMA concentration in predominantly preclinical HCM cats. 2 Renal injury markers with higher sensitivity for detecting CvRD would be useful to investigate in the preclinical population. 54 Both baseline and longitudinal study results in CHF cats supported SDMA being more sensitive than creatinine, a finding that is consistent with previous publications. 18,19,56 Compared with creatinine, even with less specificity for CKD, 18 SDMA has 2 advantages: (1) it is a more accurate renal function marker in cachexic cardiac patients with low lean body mass and (2) its superior sensitivity would allow earlier detection of changes in kidney function in acute heart failure, when secondary acute renal injury may still be reversible if prompt action is taken. 54 Our findings support the measurement of serum SDMA concentration in cats with CHF for the purpose of early detection of CvRD and appropriate intervention.
Symmetric dimethylarginine also may have prognostic value in cats with CHF, because in our study abnormal serum SDMA concentrations appeared to be associated with shorter survival. Interestingly, multiple studies in humans have demonstrated that SDMA has value as an independent risk predictor for all-cause mortality and cardiovascular disease. 57 Also, SDMA might play an active role in endothelial dysfunction, 58 which also can contribute to cardiorenal disorders. 59,60 Endothelial function was not evaluated in our study and further research is warranted.
Circulating APPs are highly sensitive inflammatory biomarkers but lack specificity, and a multiple APP-based profile has been recommended for optimal assessment of inflammatory conditions. 61 Our study had a relatively small sample size, but using a multiple APP screening approach, we identified 4 of 7 APPs to be associated with CHF in cats.
As a novel APP, the biological functions of LRG1 are not clear. It has been reported to be associated with inflammation and might play a role in myocardial fibrosis by interacting with transforming growth factor beta. 35 In mice, LRG1 has a protective function against adverse cardiac remodeling in experimentally induced myocardial infarction. 53 In our study, LRG1 was significantly increased in CHF cats. The positive correlations with both LA size and NT-proBNP suggest that LRG1 may be associated with cardiac remodeling in cats with CHF. Myocardial and extracellular matrix remodeling are common histopathological findings in cats with cardiomyopathy. 40,62,63  Serum amyloid A protein is a major APP that previously was reported to be increased in sick, hospitalized cats, 25,64 and was an independent prognostic factor in sick cats, regardless of cause. 26 In our study, increases in SAA concentration were associated with advanced CHF but not preclinical cardiomyopathy. A recent report found significant increases of SAA concentration in preclinical HCM cats, particularly in those with generalized left ventricular hypertrophy compared to those with focal hypertrophy, suggesting a possible association of SAA in the early stages of certain types of HCM. 30 In our study, such an association was not found, but it should be noted that our preclinical cardiomyopathies were not restricted to HCM phenotype and more than one third of them did not have generalized LV hypertrophy, which may explain the discrepancy between the 2 studies.
Ceruloplasmin is a copper transporter involved in iron detoxification and oxidative stress, 22,65 and it was significantly higher in the CHF cats. Both SAA and ceruloplasmin were positively correlated with NT-proBNP, LA size, and CHF grade, which suggests these 2 APPs potentially could be used for assessing disease progression and staging cats with CHF.
To date, AGP has not been linked to cardiac disease in cats. One study of comparative concentrations of AGP in pleural effusions of different etiology in cats identified lower AGP concentrations in cardiogenic compared with infectious or neoplastic effusions, but no healthy controls were included. 29 In our study, AGP appeared to be an independent risk predictor in CHF cats and it was significantly correlated with cTnI and NT-proBNP. Thus, it may have potential as a prognostic cardiac biomarker. In humans, AGP is an independent prognosticator for cardiovascular mortality as well as all causes of mortality. 34 The role of AGP as a prognostic marker in cats with heart disease should be further investigated.
The other 3 APPs evaluated in our study, including CRP, did not show significant differences in study cats. The most commonly studied APP in heart failure is CRP. In dogs, CRP has been shown to be increased in CHF and correlated with disease severity. 33,36,38 It does not appear to be a major APP in cats, 64 but interestingly, we identified significant positive correlations of CRP with renal markers. Our results suggest a link between the cardiorenal axis and an inflammatory response in cats with CHF, which merits further investigation.
One secondary finding of our study is that lower body weight was found to be an independent poor prognosticator in CHF cats.
Weight loss with cardiac cachexia could be a logical explanation for the poorer survival. A previous study reported CHF cats with extreme body weights (either too low or too high) had poorer clinical outcomes. 51 Therefore, balanced nutrition, aiming for a target weight within normal physiological range, is probably most beneficial for CHF cats in long-term management. 66 Another interesting finding was that cardiomyopathy phenotype did not affect the concentration of any investigated biomarker. Thus, these biomarkers potentially can serve as universal tools in studying different cardiomyopathies.
A multiple clinical variables-based scoring approach has been advocated for risk stratification in humans with heart failure and preliminary evaluations look promising. 67 Full clinical records of the healthy control cats were not accessible because their serum samples were from a commercial source and echocardiography examination could not be performed in this group.
Although their low serum NT-proBNP concentrations (<30 pmol/L) suggested that occult heart disease was very unlikely, preclinical cardiomyopathy cannot be completely ruled out in these cats. The healthy controls also were significantly younger than the cardiomyopathy cats, and thus an age effect on biomarker concentration cannot be excluded. However, in the age and biomarker correlation assessment, except for a moderate age correlation observed for serum creatinine concentration and weak age correlations observed for LRG1 and SAA, the remaining biomarkers were not associated with age.
Inadequate serum sample volume was a problem in several cats, and thus not all cats could have all biomarkers measured. Nine serum samples for the APP tests were stored for 3 years and the effect of long-term storage on biomarker concentration was not evaluated.
The NT-proBNP concentrations in our study were limited by the IDEXX Cardiopet assay detection range (ie, 24-1500 pmol/L). Therefore, the median NT-proBNP concentration in the CHF cats was underestimated. Lastly, in the APP component, a PCT assay designed for dogs was used, and species specificity could have influenced measured PCT results.

CONFLICT OF INTEREST DECLARATION
The acute-phase proteins were assayed at Life Diagnostic Inc.; the same company also produces the test assays.

OFF-LABEL ANTIMICROBIAL DECLARATION
Authors declare no off-label use of antimicrobials.

INSTITUTIONAL ANIMAL CARE AND USE COMMITTEE (IACUC) OR OTHER APPROVAL DECLARATION
The study obtained ethical approval from the University of Glasgow Veterinary Research Ethics Committee.