Serial measurements of cardiac troponin I (cTnI) levels are considered to be better predictors of cardiac death than single-time-point analyses in human medicine. We hypothesised that cTnI levels could reflect the severity of myxomatous mitral valve disease (MMVD), and that serial changes in the cTnI level had a prognostic value in dogs with congestive heart failure (CHF) secondary to MMVD. Seventy-six dogs were initially enrolled and classified by the American College of Veterinary Internal Medicine (ACVIM) staging system. The single-timepoint cTnI concentration in these dogs significantly increased with the ACVIM stage. Twenty-seven dogs with CHF subsequently underwent serial measurement of cTnI levels, and the results showed that those who demonstrated a decrease in cTnI levels from the first to the third visit exhibited a higher risk of cardiac death than did those without such changes (P=0.012). We suspect that the downward trend in cTnI levels may be affected by medical treatment for CHF. In conclusion, although cTnI levels could reflect the severity of MMVD to a certain extent, the serial changes may be affected by medical treatment. Therefore, caution should be exercised when cTnI is used for assessment of the prognosis of CHF secondary to MMVD in dogs.
- cardiac biomarkers
- degenerative mitral valve disease
- tracking measurements
- clinical outcomes
Statistics from Altmetric.com
Myxomatous mitral valve disease (MMVD) is characterised by a degenerative change in mitral valves that causes mitral regurgitation (MR).1 2 MMVD is the most common canine acquired cardiac disease and often occurs in middle-aged and elderly small-breed dogs.1 2 Severe MR results in volume overload and causes left ventricular eccentric hypertrophy, left atrial enlargement and, possibly, pulmonary oedema and subsequent congestive heart failure (CHF).1 2 The duration of survival after the occurrence of CHF is 6–14 months; however, this range is broad due to differences in disease progression.3 Therefore, identification of the prognostic factors of CHF secondary to MMVD is important. Many studies have shown that specific clinical characteristics, imaging findings and cardiac biomarkers are risk factors for cardiac death in MMVD; however, none of them included only dogs with CHF secondary to MMVD that were not receiving diuretic therapy at the time of admission.4–6
N-terminal pro B-type natriuretic peptide (NT-proBNP) and cardiac troponin I (cTnI) have shown a good prognostic value for cardiac diseases in veterinary medicine.6 7 Unlike NT-proBNP, which belongs to the natriuretic peptide system, reflects volume overload and is also affected by other hormone systems like the renin-angiotensin-aldosterone system and sex hormones in human medicine,8 cTnI reflects ongoing myocardial injury secondary to cardiac remodelling in MMVD.9 10 Several previous studies have shown that patients with higher cTnI concentrations at single-time-point measurements showed poor cardiovascular outcomes.5 11 12 However, recent studies have considered that serial changes in cTnI levels over time have a higher prognostic value than single-timepoint values in human medicine.13 14
A previous study showed higher levels of cTnI during serial measurements in severe cases of MMVD, but the study did not evaluate the prognostic value of cTnI levels or its ongoing changes.15 Hezzell et al observed a significantly higher cTnI concentration in the cardiac death group in their study, although their study was not exclusive to dogs with CHF secondary to MMVD.6 Therefore, this study aimed to (1) compare the cTnI concentrations at different severity levels of MMVD and (2) assess the prognostic value of serial changes in the cTnI concentration in dogs with CHF secondary to MMVD.
Materials and methods
Animals and study design
Dogs weighing less than 15 kg with MMVD, and that were not receiving diuretic therapy, were enrolled from the cardiology department of the Veterinary Medical Teaching Hospital of National Chung Hsing University (VMTH-NCHU) in Taiwan from September 2015 to July 2018. Dogs above seven years of age that belonged to the staff and students were recruited as healthy controls. History taking, physical examinations, blood pressure measurements, thoracic radiography and echocardiography were performed at admission. Dogs with other significant systemic diseases, severe respiratory disease, malignant tumours, other congenital or acquired cardiac diseases, heartworm infection and significant cardiac arrhythmia, which requires medication, were excluded from this study. Dogs with severe pulmonary hypertension (peak tricuspid regurgitation above 4.3 m/s measured by spectral Doppler echocardiography) or creatinine levels above 176.8 µmol/l at admission were also excluded.
This study was divided into two parts: (1) comparison at admission and (2) serial analysis. On admission, dogs were assigned to different stages of MMVD according to the American College of Veterinary Internal Medicine (ACVIM) staging system.1 16 Dogs without heart murmurs were allocated to stage A and considered healthy controls. MMVD was diagnosed on the basis of the following criteria: left apical systolic heart murmur, which was identified during physical examination; mitral valve prolapse and/or thickening, which was detected by two-dimensional echocardiography; MR, which was identified using colour Doppler echocardiography; and left ventricular fractional shortening above 20 per cent, which was determined by the M-mode echocardiography.17 If dogs with MMVD had at least two of the following criteria: vertebral heart score (VHS) above 10.5,18 normalised left ventricular internal dimension at end-diastole (LVIDdn) at least 1.7 cm,19 or the ratio of left atrium and aorta dimensions (LA/Ao) at least 1.620 on thoracic radiography and echocardiography, they were considered to have evidence of cardiac remodelling.1 16 Dogs with MMVD but without evidence of cardiac remodelling were classified as stage B1, and dogs with MMVD and with evidence of cardiac remodelling were classified as stage B2. Dogs showing pulmonary oedema on thoracic radiography at admission were classified as stage C (figure 1).
The serial analysis was conducted on dogs classified as stage C with at least two follow-up examinations after admission. Physical examinations, history taking, thoracic radiography and blood sampling were performed at every follow-up examination. All variables obtained at each visit, including variables obtained at admission, were sequentially numbered, and the variables obtained at the last visit were named ‘last.’ For example, the cTnI concentration at admission (first visit) and the second, third and last visits were encoded as cTnI1, cTnI2, cTnI3 and cTnI last, respectively. The scheduled second and third visits were 7 and 28 days after the first visit, respectively. However, if owners were unable to return on these scheduled dates, or if dogs with special conditions had to return early, the alternate date was recorded.
At admission, complete histories were recorded and comprehensive physical examinations were performed. In addition, signalment data (breed, age, sex and neuter status), bodyweight (BW) and the intensity of heart murmur (murmur), which was graded from I to VI using the Levine grading scale,21 were recorded. Blood pressure was continuously measured using high-definition oscillometry (Vet HDO Monitor, DVM Solutions, San Antonio, TX, USA) until at least three acceptable measurements (ie, values did not trend downward and did not exhibit substantial variation) were obtained.22 Heart rate (HR) was also recorded at every visit.
Thoracic radiography and echocardiography
Ventral-dorsal and at least one lateral view (primarily right lateral) thoracic radiographs were taken at admission and at all follow-up visits. The radiographic score (RS) was calculated from 0 to 4 according to the presence of pulmonary distension (distended vein, 1 point) and the opacity of pulmonary parenchyma (normal lung field, 0 point; mild bronchial mixed interstitial pattern, 1 point; moderate interstitial pattern, 2 points; and diffuse interstitial to alveolar pattern, 3 points).23 The VHS was calculated as described previously.24 At admission, echocardiographic examination and measurements were performed with a cardiac ultrasound machine (TOSHIBA SSA-350A Color Doppler Ultrasound with a 7.0 MHz ultrasound transducer) by a single cardiologist (IPC). One case presented with dyspnoea, and echocardiographic examinations were performed during hospitalisation. Every echocardiographic variable was recorded as the average of three cardiac cycle measurements to minimise the effects of the different cardiac cycles and respiration. LVIDdn, normalised left ventricular internal dimension at end-systole (LVIDsn)25 and LA/Ao26 were measured, and the ratio of early transmitral inflow peak velocity to late transmitral inflow peak velocity (E/A), and the ratio of early transmitral inflow peak velocity to the early diastolic mitral annular velocity (E/E′) were also obtained.27
All enrolled dogs were phlebotomised at admission and at least two follow-up assessments during the serial analysis. Serum and plasma were obtained from a gel separator tube and sodium heparin tube, respectively, after centrifugation at 13,100 g for two minutes. Serum samples were stored at −80°C until batch analysis for cTnI measurement, and plasma samples were sent to the clinical pathology department of veterinary medicine at the VMTH-NCHU for measurement of sodium, potassium, chloride, blood urea nitrogen (BUN) and creatinine levels within one hour of collection. Cardiac troponin I levels were measured using a commercially available high-sensitivity indirect ELISA (Canine High Sensitive Cardiac Troponin Elisa Kit, Blue Gene, China), with a limit of detection of 0.001 ng/ml.
Prescriptions during each visit were made by a single cardiologist (IPC) and were well recorded during the serial analysis. The furosemide dose (Lasix; Handok) was recorded as mg/kg/day. Doses of the other drugs, including amlodipine (Amndiline; China Chemical & Pharmaceutical), pimobendan (Vetmedin; Boehringer), spironolactone (Aldactone; Pfizer) and enalapril (Sintec; Taiwan Biotech), were recorded as categorical data (prescribed or not prescribed) because of their narrow dose ranges.
The serial analysis was continued until July 2018. Outcome data were obtained mainly through telephone calls. The primary outcome was cardiac death after CHF was diagnosed. Death was considered to be of cardiac-related cause when sudden death occurred, tachypnoea or dyspnoea was observed shortly before death, or euthanasia for refractory CHF was carried out. Survival time was calculated from the onset of CHF to the time of cardiac death. Non-cardiac deaths were excluded from the serial analysis.
Analyses were performed using commercial software (SPSS V.20, IBM) by an investigator (SYW). Normality of all variables was assessed by the Kolmogorov-Smirnov test (n>50) and Shapiro-Wilk test (n<50). Normally distributed data and non-normally distributed data were presented as mean±sd and median (first interquartile-third interquartile), respectively. Categorical data are represented as the count (n). Between the ACVIM stages, analysis of variance was used to compare normally distributed data, and the Kruskal-Wallis test was used to compare non-normally distributed data at admission. Post hoc analyses were conducted using Scheffe’s test and Dunn’s test for normally and non-normally distributed data, respectively. The correlations between other variables and cTnI measurements were analysed using Spearman’s correlation coefficient.
Changes in cTnI levels were calculated using the following formulae:
Absolute cTnI change 2–1 or 3–1=second-visit cTnI (cTnI2) or third-visit cTnI (cTnI3)–first-visit (admission) cTnI (cTnI1)
Relative cTnI change 2–1 or 3–1=(cTnI2 or cTnI3–cTnI1)/cTnI1
For the serial analysis, all variables were initially compared between survivors and non-survivors using Student’s t tests, Mann-Whitney U tests or chi-squared tests, depending on the type of variable. Second, statistically significant variables and significant variables*time (ie, the interaction of significant variables and survival time) were subjected to a univariate Cox regression analysis in order to evaluate the assumption of proportionality. The Cox regression analysis was then performed on all significant variables. For clinical applicability, significant variables that were associated with cardiac death in the univariate Cox regression analysis were converted to binary variables and assessed using receiver operating characteristic (ROC) curves to determine the best cut-off score for predicting cardiac death. Finally, the survival curve was drawn using the Kaplan-Meier method and harzard ratios were calculated. For all analyses, P<0.05 was considered statistically significant.
Comparisons at admission
Seventy-six dogs, including Maltese Terriers (n=28), Toy Poodles (n=14), Pomeranians (n=7), Chihuahuas (n=7), Miniature Schnauzers (n=7), mixed breeds (n=5), Beagles (n=2), Dachshunds (n=2) and one dog of each of the following breeds: Japanese Chin, Japanese Spitz, Miniature Pinscher and Papillon, were enrolled in this study. There were 28 intact males (M), 6 intact females (F), 23 neutered males (MN) and 19 neutered females (FN). The median age of the dogs was 10 years (8–11 years), and their median BW was 4 kg (3–6 kg). According to the ACVIM staging system, 8 of the enrolled dogs were classified as stage A, 7 as stage B1, 22 as stage B2 and 39 as stage C. There were no significant differences in gender, breed, BW, BUN and blood pressure showed among the ACVIM groups (table 1). Dogs that were classified as stage A were significantly younger than those classified as stages B2 and C and had higher creatinine levels than dogs classified as stages B1 and C (table 1). HR, murmur, VHS, RS, LVIDdn, LA/Ao, E wave, E/A and E/E′ were significantly higher in dogs classified as stage C than in dogs classified as the other stages (table 1). VHS and LVIDdn were significantly higher in dogs classified as stage B2 than in dogs classified as stages A and B1. Murmur was significantly higher in dogs classified as stage B2 than in dogs classified as stage A, and LVIDsn was significantly higher in dogs classified as stage C than in dogs classified as stage B1 (table 1). Cardiac troponin I levels were significantly different among the ACVIM groups, and cTnI levels were significantly lower in the stage A group than in the other groups (figure 2).
At admission, there were significant correlations between cTnI levels and HR, murmur, RS, VHS, LA/Ao, E wave, E/A, E/E′ and LVIDdn. However, there were no significant correlations between cTnI levels and age, BW, BUN and creatinine levels (table 2).
Thirty-nine dogs were diagnosed with pulmonary oedema at admission (first visit). Ten dogs with less than three measurements of cTnI were excluded. Before the end of study, two dogs died of pancreatitis and acute renal failure and were excluded from the serial analysis. A total of 27 dogs were enrolled in the serial analysis; 14 dogs died of cardiac-related causes and 13 dogs were alive at the end of the study. None of the dogs underwent euthanasia. Overall, 14 Maltese Terriers, 5 Pomeranians, 3 Chihuahuas, 2 Miniature Schnauzers, 2 Dachshunds, 2 mixed breeds, 1 Toy Poodle and 1 Miniature Pinscher were included in the serial analysis. There were 12 intact males, 10 neutered males and 7 neutered females. At the time of diagnosis of CHF (first visit), their mean age was 10.19±2.095 years, and their median BW was 3.50 kg (3.0–5.0 kg). All dogs that were enrolled in the serial analysis revisited the hospital at least twice after the first visit. The total number of visits, including the baseline visit, was 4 (range, 3–4 visits). The second, third and last visits were at 7 (range, 7–7 days), 28 (range, 21–45 days) and 33 days (range, 21–59 days) after the first visit, respectively. Serial measurements are shown in table 3. All dogs underwent thoracic radiography during each visit, except for seven dogs that did not undergo thoracic radiography during the third visit. Furosemide was prescribed at the first visit for all dogs enrolled in the serial analysis. Two dogs had received enalapril before the first visit. Three dogs were hospitalised at the first visit and were administered furosemide (1 mg/kg/hour) by continuous rate infusion; therefore, the administration of furosemide1 (the dose of furosemide administered during the first visit) was recorded as 24 mg/kg/day. These dogs subsequently received oral medication and were discharged. The doses of furosemide3 (the dose of furosemide administered during the third visit) and furosemide last were significantly higher in the non-survival group (P=0.029 and P=0.047, respectively; table 3).
Measurements of systolic blood pressure (SBP) were significantly lower in the non-survival group than the survival group at the first visit. Additionally, E/E′ was significantly higher in the non-survival group than the survival group at the first visit (table 4). Absolute cTnI change 3–1 and relative cTnI change 3–1 were significantly lower in the non-survival group than the survival group (table 4). Results from the univariate Cox regression demonstrated that SBP and the absolute cTnI change 3–1 were significantly associated with cardiac death, and the interaction of SBP and survival time was also significantly associated with cardiac death (table 5). Considering the limited sample size, a multivariate Cox regression could not be conducted. To predict cardiac death, an ROC curve was applied and the SBP and the absolute cTnI change 3–1 cut-off points were determined to be 135.5 mmHg and −0.03 ng/ml, respectively (P=0.002 and P=0.026, respectively; figure 3). Moreover, dogs with an absolute cTnI change 3–1 of less than −0.03 ng/ml (those who demonstrated a downward trend in cTnI levels from the first to the third visit) exhibited a significantly higher risk of cardiac death than did those with an absolute change of more than −0.03 ng/ml (P=0.012). The hazard ratio was 6.272 (95% CI 2.052 to 19.174; figure 4).
In this study, cTnI was higher in dogs with stage B1, stage B2 and stage C MMVD than in dogs with stage A MMVD. The cTnI level was associated with many factors that reflected the severity of MMVD, including HR, murmur, RS, VHS, LA/Ao, E wave, E/A, E/E′ and LVIDdn. In the serial analysis, we found that dogs with an absolute cTnI change 3–1 of less than −0.03 ng/ml exhibited a higher risk of cardiac death than did those with an absolute change of more than −0.03 ng/ml.
Although previous studies have reported that significantly higher cTnI levels are associated with severe MMVD,15 28 29 this was the first study that compared cTnI concentrations across ACVIM stages. Previous studies have shown that age is a confounding factor for MMVD,15 28 and dogs assigned to stage A were younger than dogs assigned to the other stages. Still, age was not significantly associated with cTnI concentrations in the present study, but the small sample size of stage A was a limiting factor in the association between age and cTnI concentration. Similar to previous studies, cTnI was significantly associated with LA/Ao and LVIDdn in the present study.11 28 To the best of our knowledge, this is the first study to demonstrate the association between cTnI and VHS. The VHS, LA/Ao and LVIDdn are important parameters of cardiac remodelling and are highly associated with the severity of MMVD.30 Increased serum concentration of cTnI was considered to be associated with ongoing cardiac injury due to cardiac remodelling in MMVD and has also been associated with the severity of MMVD.9 15 28 In the present study, cTnI was also associated with HR, murmur, RS, E wave, E/A and E/E′. Previous studies have shown that the E wave, E/A and E/E′ may reflect volume overload and were associated with the severity of MR.5 31 Similar to previous studies, the present study excluded significant respiratory diseases23; therefore, RS may reflect the severity of pulmonary oedema and may equal zero before entering the heart failure stage. The intensity of heart murmur increases with the severity of MR in MMVD.32 The sympathetic nervous system is activated during CHF, leading to a significantly higher HR in dogs assigned to stage C than in dogs assigned to the other stages of the ACVIM.33 The cTnI level directly reflects myocardial damage during the progression of MMVD.9 10 In this study, factors such as LA/Ao, E wave, E/A, E/E′ and LVIDdn, which reflect other aspects of the pathophysiology of MMVD and are not directly associated with myocardial damage, were weakly correlated with cTnI. A previous study reported that LVIDsn values were within the normal range in most dogs with CHF secondary to MMVD because of the compensation of the sympathetic nervous system and adrenergic support.34 To the best of our knowledge, we are the first to compare the LVIDsn between the ACVIM stages of MMVD, and LVIDsn was significantly higher in dogs classified as stage C than in dogs classified as stages A and B1, which may have resulted from the process of cardiac remodelling in MMVD.
Our observation that non-survivors had lower SBP at first visit enrolment to the study is similar to previous studies that have reported that dogs with severe MMVD had a lower SBP.35 36 It has been speculated that severe MR leads to a decrease in blood pressure because the majority of the stroke volume returns to the left atrium and did not eject to the aorta.37 However, the sympathetic nervous system would be activated during CHF to maintain the SBP within the normal range.33 Moreover, only Boswood et al reported that dogs with a lower SBP demonstrated poorer outcomes even though no dogs were clinically hypotensive, which was similar to our results.38
In veterinary medicine, previous studies have shown good prognostic values of single-time-point cTnI levels in MMVD.6 7 39 These studies demonstrated that higher single-time-point cTnI levels were associated with worse outcomes. Recent studies in human medicine have shown that serial changes in cTnI levels over time have a higher prognostic value than single-time-point cTnI levels.13 Additionally, Kawahara et al showed that stable non-ischaemic heart disease outpatients with cTnI above 0.03 ng/ml at baseline and higher cTnI levels six months later exhibited a greater tendency for cardiac death than did those with cTnI less than 0.03 ng/ml and lower cTnI six months later (hazard ratio=34.9; P=0.0006).14 In our study, dogs with an absolute cTnI change 3–1 of less than −0.03 ng/ml (those who demonstrated a downward trend in cTnI from the first to the third visit) exhibited a higher risk of cardiac death than did dogs with an absolute change of more than −0.03 ng/ml. Interestingly, this result seems to be in discordance with the study of Kawahara et al.14 However, Polizopoulou et al showed that cTnI exhibited a downward trend in the first two weeks after diagnosis of MMVD, although it increased at the eighth visit (16th week) as in dogs with advanced CHF.15 Researchers suggested that the downward trend of cTnI reflected the effect of the treatment of CHF secondary to MMVD. In our study, we used diuretics to control CHF secondary to MMVD, and dogs with an absolute cTnI change 3–1 of less than −0.03 ng/ml were thought to respond well to diuretics according to previous studies. Therefore, we assumed that the downward trend of cTnI in dogs with an absolute cTnI change 3–1 of less than −0.03 ng/ml would eventually shift upward, as observed in the study conducted by Polizopoulou et al.15 Unfortunately, we were unable to conduct a longer or larger study than that conducted by Polizopoulou et al, primarily because the owners refused prolonged participation. Nevertheless, the use of diuretics cannot explain why the cTnI levels at the second and third visits were not lower than the level at the first visit in the survival group in our study. We speculated that this may have occurred because of the dose-dependent effects of the diuretics. In our study, the dogs in the non-survival group were given gradually higher dose of furosemide from the first to third visits to stabilise pulmonary oedema. Moreover, the dogs in the non-survival group were given significantly higher doses of furosemide at the third visit compared with the dogs in the survival group. On the other hand, most dogs in the survival group presented with stable pulmonary oedema, with minimal progression under low doses of furosemide. Moreover, there was no significant decrease in the cTnI level between the first and third visits.
This study had some limitations. First, the sample sizes of the different ACVIM stages were not equal at admission. Secondly, the study population in the serial analysis (27 dogs) was smaller than the population included in previous studies.3–6 However, the statistical power of the study was sufficient to demonstrate that an absolute cTnI change 3–1 of less than −0.03 ng/ml or above −0.03 ng/ml was significantly associated with cardiac death after the onset of CHF. Thirdly, the intervals between the first visit and the third visit were inconsistent, and the observation period may have been too short to predict cardiac death. Fourthly, the outcomes were mainly obtained through telephone follow-ups without any necropsy studies. Fifthly, reference change values (RCV) were not used while evaluating cTnI in the serial analysis. RCV represents the biological variability and is thought to be crucial for interpreting serial biochemistry and biomarker changes during disease monitoring, but previous studies have suggested that these population-based RCVs show limited values for individual dogs during serial cTnI measurements that monitor the severity of MMVD.16 Sixthly, the serum samples, which were obtained from gel separator tubes, may have exhibited artefactually decreased cTnI levels because these circulating biomarkers could be bound in the gel. Finally, the dose and types of medication were not consistent in the serial study. However, at every visit, the cardiologist was unaware of the cTnI levels when the medication was prescribed.
Despite the limitations mentioned above, our findings add to existing literature showing that cTnI levels tend to change over time in dogs with CHF secondary to MMVD. However, our results suggest that the risk of cardiac death is higher in dogs showing a downward trend in the cTnI level after the onset of CHF than in those without a downward trend. Although this result is partially identical to the findings of Polizopoulou et al, it contradicts the findings of most of other studies. We suspect that the downward trend in the cTnI level might be affected by medical treatment for CHF. Therefore, further studies should determine the effects of medication on the cTnI level in dogs with CHF secondary to MMVD and investigate the relevant association with disease progression.
Presented at This study was performed at the Veterinary Medical Teaching Hospital, National Chung Hsing University, Taiwan.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Ethics approval This study was certified by the Institutional Animal Care and Use Committee of National Chung Hsing University, which is a local ethical approval committee.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement All data relevant to the study are included in the article.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.