Comparison of electrocardiographic parameters in dogs with different stages of myxomatous mitral valve disease

Abstract

This study evaluated changes in electrocardiographic (ECG) parameters according to the stage of myxomatous mitral valve disease (MMVD) in dogs, as well as the utility of ECG parameters as prognostic indicators for congestive heart failure (CHF). Medical records of dogs with MMVD were retrospectively searched. Dogs with MMVD (N = 101) were classified into stages B [B1 (n = 52) and B2 (n = 23)] and C (n = 26) according to the American College of Veterinary Internal Medicine guidelines. Baseline variables were collected; these included signalment, radiographic, echocardiographic, and ECG parameters. Corrected QT intervals (QTc) were calculated using the logarithmic (QTc1) and Fridericia (QTc2) formulas. The P wave duration, QTc1, and QTc2 were significantly longer in stage C than in stage B. The P wave duration cutoff of 43.5 ms had a diagnostic accuracy of 65% for differentiating CHF, with a sensitivity of 63% and a specificity of 90%. A cutoff value of 307.8 ms for QTc1 yielded a sensitivity of 62%, a specificity of 76%, and a diagnostic accuracy of 78%, and a cutoff value of 239.2 ms for QTc2 yielded a sensitivity of 62%, a specificity of 83%, and a diagnostic accuracy of 77% for diagnosing CHF. Therefore, prolonged P wave and QTc in dogs with MMVD may facilitate the prediction of CHF. Electrocardiography could provide clinicians with a readily available and cost-effective screening tool for predicting CHF, if the usefulness of ECG parameters can be verified.

Introduction

Myxomatous mitral valve disease (MMVD) is the most prevalent heart disease in the canine population, estimated to account for 75% of all canine heart disease (1). Myxomatous mitral valve disease will lead to heart failure or death in some dogs, whereas others will survive several years without progressing to a more advanced stage or showing any heart disease-related clinical signs. To provide appropriate preventative care to asymptomatic dogs, clinicians should be able to recognize prognostic indicators of heart failure and cardiac death.

Prognostic indicators in dogs with MMVD include age (2), sex (3), heart rate (4), intensity of heart murmur (5), severity of mitral valve regurgitation (MR) (6), left atrial and ventricular enlargement (4,7), and circulating concentration of the N-terminal fragment of probrain natriuretic peptide (8,9). In humans, potential indicators of cardiovascular disease progression observable by electrocardiography (ECG) are well-known. The ECG parameters associated with cardiovascular mortality in humans are prolonged P wave and QRS duration, interatrial and bundle branch block, fragmented QRS, prolonged QT and T_peak to T_end intervals, spatial angles between QRS and T vectors, ST segment depression and elevation, and inverted T wave (10).

It has still not been proven whether ECG parameters can predict exacerbation of MMVD in dogs. Therefore, the aim of the present study was to evaluate changes in ECG parameters according to the canine MMVD stage and to investigate the usefulness of ECG parameters as prognostic indicators for congestive heart failure (CHF).

Materials and methods

Study population

Medical records of dogs diagnosed with MMVD between January 2013 and December 2019 at the Veterinary Teaching Hospital, Chungbuk National University, were retrospectively searched. Diagnosis of MMVD was established based on clinical signs and echocardiographic findings such as left apical systolic murmur and valvular lesions of the mitral valve apparatus (thickening, prolapse, or both) associated with MR (11). Dogs underwent physical examination, thoracic radiography, echocardiography, and ECG. All diagnostic tests were performed within 1 wk.

Dogs were excluded if they had arrhythmia, had been taking antiarrhythmic drugs, or had significant cardiac disease other than MMVD. Among the 114 dogs with MMVD, 13 were excluded. Thus, 101 dogs were included in this study (Figure 1).

Figure 1.

Flowchart for selection and exclusion of dogs with myxomatous mitral valve disease (MMVD) in this study.

Grouping

Dogs with MMVD were divided into different stages according to the American College of Veterinary Internal Medicine (ACVIM) classification (11). Dogs were assigned to stage B if they were diagnosed with asymptomatic MMVD. Stage B was subdivided in accordance with the absence (stage B1) or presence (stage B2) of evidence of cardiac remodeling. Dogs were assigned to stage B2 when all the criteria for cardiac remodeling were met: i) murmur intensity ≥ 3/6, ii) left atrium to aortic root diameter ratio (LA:Ao) ≥ 1.6, iii) left ventricular internal diameter in diastole (LVIDd) normalized for body weight (LVIDDN) ≥ 1.7, and iv) radiographic vertebral heart score (VHS) > 10.5 (11). Dogs were assigned to stage C if they were diagnosed with CHF caused by MMVD. Diagnosis of CHF was based on both clinical signs and radiographic evidence of cardiogenic pulmonary edema, pulmonary venous congestion, or both (Figure 1). Dogs with refractory CHF caused by MMVD (stage D) were not included in this study.

Electrocardiography

Standard 6-lead ECGs [leads I, II, III, augmented vector (aV) right, aV left, and aV foot] were recorded at a paper speed of 50 mm/s and vertical ECG calibration of 20 mm/mV. All ECGs were recorded using a commercially available ECG machine (PageWriter TC70; Philips, Eindhoven, Netherlands). Ten consecutive beats were assessed to reduce the risk of overestimating or underestimating the length of the wave. The following parameters were manually measured on the ECG by using a ruler with 0.5 mm gradations: i) P wave amplitude and duration, ii) PR interval, iii) R wave amplitude, iv) QRS duration, v) QT interval, vi) ST segment, and vii) T wave amplitude.

Generally, the duration of the QT interval depends on the heart rate (12). To take this into account, several QT correction formulas have previously been suggested. Of these, the logarithmic and the Fridericia formulas appear to be reliable for use in dogs and were therefore used in this study (12,13). The 2 formulas for calculating the corrected QT interval (QTc) were:

$$\begin{array}{c}\text{logarithmic\hspace{0.17em}formula\hspace{0.17em}}(\text{QTc}1)=\text{log\hspace{0.17em}}600\times \text{QT}/\text{log\hspace{0.17em}RR},\\ \text{with\hspace{0.17em}QT\hspace{0.17em}and\hspace{0.17em}RR\hspace{0.17em}in\hspace{0.17em}ms\hspace{0.17em}and}\\ \text{Fridericia\hspace{0.17em}formula\hspace{0.17em}}(\text{QTc}2)=\text{QT}/({\text{RR}}^{0.33}),\\ \text{with\hspace{0.17em}QT\hspace{0.17em}in\hspace{0.17em}ms\hspace{0.17em}and\hspace{0.17em}RR\hspace{0.17em}in\hspace{0.17em}s.}\end{array}$$

Echocardiography

Echocardiography was performed to confirm the diagnosis of MMVD and to assess the baseline echocardiographic parameters. The following parameters were measured: i) MR, ii) LA:Ao ratio, iii) LVID in systole and diastole (LVIDs and LVIDd, respectively), iv) LVIDDN (LVIDd/body weight^0.294) (14), and v) fractional shortening [(LVIDd – LVIDs)/LVIDd]. Right parasternal long-axis and left apical 4-chamber views were used to assess the mitral valve apparatus and the degree of MR by color Doppler sonography. The LA:Ao ratio, LVIDs, and LVIDd were measured from the right parasternal short axis view (15).

Radiography

Right lateral and dorsoventral radiographic projections were used to confirm the diagnosis of CHF and assess the cardiac size. The cardiac size was measured using the VHS method (16).

Statistical analyses

Data were analyzed using commercially available statistical software (Prism 6.01; GraphPad Software, San Diego, California, USA). The normality of data distribution was confirmed by the Kolmogorov-Smirnov test. Normally distributed continuous variables were presented as mean ± standard deviation, whereas non-normally distributed continuous variables were presented as median and interquartile range (IQR).

Electrocardiography parameters were compared between stages B and C using unpaired t-tests. Parameters were compared among dogs with different stages of MMVD (B1, B2, and C) by 1-way analysis of variance (ANOVA) followed by Tukey’s multiple comparison test (for normally distributed variables) or by the Kruskal-Wallis test followed by Dunn’s multiple comparison test (for non-normally distributed variables). Spearman’s correlation coefficient was used to assess the relationship between ECG parameters and other parameters (i.e., echocardiographic and radiographic parameters). The diagnostic utility of ECG to discern between stages B and C was assessed by receiver operating characteristic (ROC) curves. The area under the curve (AUC) of the ROC curve was obtained for each ECG parameter. The AUC value as a criterion of discrimination accuracy was classified as poor (0.5 to 0.6), sufficient (0.6 to 0.7), good (0.7 to 0.8), very good (0.8 to 0.9), or excellent (0.9 to 1.0) (17). Sensitivity and specificity were calculated by using the cutoffs determined from the ROC curves. Diagnostic accuracy was calculated for each ECG parameter by adding the true positive and true negative and dividing this value by the total number of dogs (17). A P-value < 0.05 was considered statistically significant.

Results

Study population

One hundred and one dogs (50 females, 29 of which were spayed; 51 males, 37 of which were neutered) with a median age of 12 y (IQR = 9.8 to 13 y) were included in the study. Fifty-two, 23, and 26 dogs were classified as stage B1, B2, and C, respectively. Several breeds were included, the most common of which were Maltese (n = 52) and Shih Tzu (n = 21). There were no significant differences between dogs in stages B1, B2, and C in terms of age, sex, and breed (P > 0.05, respectively) (Table I).

Table I.

Baseline characteristics of 101 dogs with myxomatous mitral valve disease. Continuous variables are reported as median (interquartile range). Categorical variables are reported as number of dogs (%).

Variables	Stage B (n = 77)	Stage B1 (n = 52)	Stage B2 (n = 25)	Stage C (n = 24)	P-value
Age	12 (9.25 to 13)	12 (9.25 to 13.75)	12 (9.25 to 13)	12.5 (9.25 to 14)	0.47
Sex					0.40
Female	15 (20%)	8 (15%)	7 (28%)	6 (25%)
Spayed female	25 (32%)	19 (37%)	6 (24%)	4 (17%)
Male	9 (12%)	4 (8%)	5 (20%)	5 (21%)
Castrated male	28 (36%)	21 (40%)	7 (28%)	9 (38%)
Breed					0.49
Maltese	36 (47%)	25 (48%)	11 (44%)	16 (67%)
Shih tzu	16 (21%)	10 (19%)	6 (24%)	5 (21%)
Mongrel	6 (8%)	6 (12%)	0 (0%)	0 (0%)
Spitz	4 (5%)	3 (6%)	1 (4%)	0 (0%)
Yorkshire terrier	3 (4%)	2 (4%)	1 (4%)	0 (0%)
Japanese chin	2 (3%)	0 (0%)	2 (8%)	0 (0%)
Miniature pinscher	2 (3%)	2 (4%)	0 (0%)	0 (0%)
Pomeranian	2 (3%)	2 (4%)	0 (0%)	0 (0%)
Poodle	2 (3%)	1 (2%)	1 (4%)	1 (4%)
Cocker spaniel	1 (1%)	0 (0%)	1 (4%)	0 (0%)
Dachshund	1 (1%)	0 (0%)	1 (4%)	0 (0%)
Miniature schnauzer	1 (1%)	1 (2%)	0 (0%)	0 (0%)
Pekingese	1 (1%)	0 (0%)	1 (4%)	2 (8%)

Comparison of ECG parameters among different stages of MMVD

Electrocardiographic parameters obtained from MMVD dogs are presented in Table II. The P wave duration, QT interval, QTc1, and QTc2 in stage C were significantly longer than in stage B, as shown in Figure 2 (P < 0.05, respectively).

Table II.

Selected baseline electrocardiographic variables of 101 dogs with myxomatous mitral valve disease. Continuous variables are reported as mean (standard deviation) when normally distributed and as median (interquartile range) if not normally distributed.

Variables	Stage B (n = 77)	Stage B1 (n = 52)	Stage B2 (n = 25)	Stage C (n = 24)
P wave amplitude (mV)	(0.4 (0.2 to 0.45)	0.4 (0.25 to 0.5)	0.3 (0.2 to 0.4)	0.32 (0.2 to 0.525)
P wave duration (ms)	36.18 (7.1)a	34.18 (7.17)a,b	40.34 (4.87)a	49.17 (12.38)
PR interval (ms)	92.73 (19.03)	89.04 (16.24)	100.4 (22.26)	101.25 (20.5)
R wave amplitude (mV)	2.1 (1.7 to 2.7)	2.05 (1.6 to 2.6)	2.2 (1.8 to 2.75)	2.0 (1.44 to 2.64)
QRS duration (ms)	40.39 (13.02)	37.69 (10.22)	46 (16.33)	43.75 (19.07)
QT interval (ms)	173.12 (20.6)a	171.15 (19.77)a	177.2 (22.08)	190.42 (31.96)
ST segment (mV)	−0.02 (0.22)	−0.22 (0.25)	−0.01 (0.14)	0.03 (0.18)
T wave amplitude (mV)	0.2 (0.1 to 0.4)	0.2 (0.1 to 0.4)	0.2 (0.15 to 0.5)	0.35 (0.14 to 0.5)
QTc1 (ms)	290.25 (273.84 to 307.63)a	288.02 (272.28 to 307.8)a	294.34 (282.82 to 307.85)a	319.26 (292.78 to 346.19)
QTc2 (ms)	224.6 (213.03 to 237.28)a	221.74 (212.31 to 238.16)a	226.35 (216.34 to 236.64)a	243.62 (228.87 to 272.29)

QTc1 — corrected QT interval calculated by the logarithmic formula; QTc2 — corrected QT interval calculated by the Fridericia formula.

^aWithin a row, the value differs significantly from that for stage C (P < 0.05).

^bWithin a row, the value differs significantly from that for stage B2 (P < 0.05).

Figure 2.

Scatter plots of P wave duration (A), QT interval (B), corrected QT interval calculated by the logarithmic formula (QTc1) (C), and corrected QT interval calculated by the Fridericia formula QTc2 (D) in 75 dogs with stage B and 26 dogs with stage C myxomatous mitral valve disease (MMVD). The P wave duration, QT interval, QTc1, and QTc2 were significantly longer in stage C in comparison with stage B.

* P < 0.05.

The P wave duration, QT interval, QTc1, and QTc2 were significantly different among stages B1, B2, and C. The P wave for dogs with stage C was significantly longer than that of dogs in stages B1 and B2 (P < 0.05, respectively). The P wave became significantly prolonged with the progression of the MMVD stage (Figure 3 A). The QT interval for dogs with stage C was significantly longer than that of dogs with stage B1 (P < 0.05). There was no significant difference in QT interval between stages B1 and B2 and between stages B2 and C (Figure 3 B). The QTc1 and QTc2 were significantly longer in dogs with stage C than in those with stage B1 and B2 (P < 0.05 and P < 0.05, respectively). However, there were no statistically significant differences in QTc1 and QTc2 between dogs with stages B1 and B2 (Figure 3 C,D).

Figure 3.

Scatter plots of P wave duration (A), QT interval (B), corrected QT interval calculated by the logarithmic formula (QTc1) (C), and corrected QT interval calculated by the Fridericia formula (QTc2) (D) in 52 dogs with stage B1, 23 dogs with stage B2, and 26 dogs with stage C myxomatous mitral valve disease (MMVD). P wave duration exhibited significant prolongation with the progression of the MMVD stage. The QT interval of stage C was significantly longer than that of stage B1. The QTc1 and QTc2 were significantly longer in stage C than in stages B1 and B2.

* P < 0.05.

Correlation of ECG parameters with echocardiographic and radiographic parameters

There was a statistically significant positive correlation between VHS and P wave duration (r = 0.4636, P < 0.05) (Figure 4 A). Similarly, there was a significant positive correlation between the LA:Ao ratio and P wave duration (r = 0.3822, P < 0.05) (Figure 4 B). Moreover, there was a weak, although statistically significant, correlation between VHS and QT interval (r = 0.2518, P < 0.05) (Figure 4 C).

Figure 4.

Correlation of P wave duration with vertebral heart scale (VHS) (A) and left atrium to aortic root diameter (LA:Ao) ratio (B). The P wave duration had a significant and moderate positive correlation with VHS and LA:Ao ratio. There was a weak positive correlation between QT interval and VHS (C).

ROC analysis of ECG parameters

The ROC curves were used to assess the ability of the P wave duration, QTc1, and QTc2 to discern dogs with (stage C) and without (stage B) CHF at the time of diagnosis. The ROC analysis of P wave duration showed an AUC of 0.84, with a 95% confidence interval (CI) of 0.74 to 0.93 (P < 0.05) (Figure 5 A). A P wave duration in dogs with VHS > 10.5 had an AUC of 0.88 (95% CI = 0.79 to 0.97, P < 0.05) and an LA:Ao ratio ≥ 1.6 had an AUC of 0.88 (95% CI = 0.80 to 0.97, P < 0.05) (Figures 5 B,C).

Figure 5.

Receiver operating characteristic curve illustrating sensitivity and specificity of P wave duration (A), P wave duration with vertebral heart scale (VHS) > 10.5 (B), P wave duration with left atrium to aortic root diameter (LA:Ao) ratio ≥ 1.6 (C), corrected QT interval calculated by the logarithmic formula (QTc1) (D), and corrected QT interval calculated by the Fridericia formula (QTc2) (E). The line marked by triangles represents the area under the curve (AUC) that is 50%.

At a cutoff value of 43.5 ms, the diagnostic accuracy for detecting CHF was 65% with a sensitivity of 63% and a specificity of 90%. At the same cutoff value, considering dogs with VHS > 10.5 or LA:Ao ratio ≥ 1.6, the diagnostic accuracy was improved to 79% and 85%, respectively, while the sensitivity increased to 76% and 70%, respectively. However, specificity remained similar (84% and 90%, respectively) (Table III).

Table III.

Cutoff values, sensitivity, specificity, and accuracy of P wave duration, QTc1, and QTc2 for differentiating congestive heart failure in dogs with myxomatous mitral valve disease.

Variables	Cutoff (ms)	Sensitivity (%)	Specificity (%)	Accuracy (%)
P wave duration	43.5	63	90	65
P wave duration (VHS > 10.5)	43.5	76	84	79
P wave duration (LA:Ao ratio ≥ 1.6)	43.5	70	90	85
QTc1	307.8	62	76	73
QTc2	239.2	62	83	77

VHS — vertebral heart scale; LA:Ao — left atrium to aortic root diameter; QTc1 — corrected QT interval calculated by the logarithmic formula; QTc2 — corrected QT interval calculated by the Fridericia formula.

The QTc1 had an AUC of 0.75 (95% CI = 0.62 to 0.87, P < 0.05). A cutoff of 307.8 ms for QTc1 differentiated stage C from stage B with a sensitivity of 62%, a specificity of 76%, and a diagnostic accuracy of 78% (Figure 5 D, Table III). For QTc2, ROC analysis revealed an AUC of 0.78 (95% CI = 0.63 to 0.88, P < 0.05) and showed that a cutoff value of 239.2 ms yielded a sensitivity of 62%, a specificity of 83%, and a diagnostic accuracy of 77% (Figure 5 E, Table III).

Discussion

In this study, we evaluated changes in ECG parameters according to ACVIM stage in dogs with MMVD and investigated the diagnostic performance of these ECG parameters in detecting CHF; this has not been reported previously. We found that P wave duration, QTc1, and QTc2 were significantly longer in dogs with stage C than in dogs with stage B. Our results suggest that P wave duration and QTc prolongation may help to predict CHF in dogs with asymptomatic MMVD.

In this study, P wave duration correlated positively with VHS and the LA:Ao ratio. The VHS has been reliable in showing an increase in cardiac size (16,18) and the LA:Ao ratio is commonly used to detect left atrial enlargement (15,19). Left atrial enlargement is associated with prolongation of left atrial depolarization and corresponds to a prolonged P wave duration (20). The VHS and echocardiographic left atrial and ventricular dimensions are increased prior to the onset of CHF dogs with MMVD (7,21,22). These are predictive factors for CHF in dogs with MMVD.

The P wave duration indicates the time it takes for the right and left atria to depolarize completely. The P wave duration may be increased in atrial enlargement (particularly the left atrium) or interatrial conduction disturbance, such as Bachmann’s bundle block (23). Although interatrial conduction disturbance may be clinically observed (24), the frequency of its occurrence in dogs is unknown. For practical purposes, an extended P wave duration should be considered to indicate LA dilation.

In human medicine, prolonged signal-averaged P wave duration appears to predict hospitalization due to worsening CHF or early death in patients with stable CHF (25). Signal-averaged P wave duration was prolonged in human CHF patients with higher pulmonary capillary wedge pressure (26). The mechanisms underlying the worse prognosis can be explained by the prolongation of the P wave duration due to a hemodynamic overload. At a cutoff of 43.5 ms, the P wave duration with cardiac remodeling (VHS > 10.5 or LA:Ao ratio ≥ 1.6) provided a similar specificity, but a higher sensitivity than the P wave duration without cardiac remodeling. This may be especially significant for identifying the risk in dogs already known to have left atrial pressure or volume overload.

The QT interval indicates the duration for ventricular depolarization and repolarization to occur (23). Koyama et al (13) reported that corrected QT intervals became significantly prolonged with New York Heart Association (NYHA) class progression. Similar to previous studies, we found that QTc was significantly longer in dogs with stage C than those with stage B, but there was no significant difference between stages B1 and B2. This difference in results seems to be due to differences in the MMVD stage classification system. Unlike the NYHA classification, the ACVIM classification more objectively subdivides asymptomatic dogs with MMVD dogs according to the evidence of cardiac remodeling (11,27).

Prolongation of the QT interval has been associated with left ventricular dysfunction (28). Prolonged QTc interval was associated with adverse outcomes in human heart failure patients with elevated intracardiac pressure due to left ventricular failure (29). An increase in intracardiac pressure has been shown to stretch the myocardium, resulting in slowed conduction and enhanced refractoriness. These changes may lead to prolonged ventricular action potentials and instability of ventricular repolarization in human patients with heart failure (30).

Furthermore, prolongation of the QT interval in human heart failure patients was not related to the QRS duration (28). In the present study, the QRS duration did not differ significantly between stages B and C. Hence, it may be that factors other than QRS duration prolongation may have contributed to QT interval prolongation. The QT interval is significantly influenced by changes in sympathetic tone (31,32). Dogs with mitral valve prolapse syndrome have QT interval prolongation and elevated serum catecholamine levels (33). Moreover, advanced heart failure is related to extensive neuroendocrine activation (34), which has been shown to affect the QT interval duration. Therefore, QT prolongation may not only indicate ventricular instability in dogs with heart failure but may also be a marker of advanced heart disease.

This study had some limitations. First, many dogs were taking cardiac medication, such as inotropic drugs, angiotensin-converting enzyme inhibitors, and diuretics at the time of the ECG recording, and each dog had a different drug combination. Further studies regarding the effect of chronic drug treatment on ECG are encouraged. Second, although the ECG parameters were measured by the same investigator to avoid interobserver discrepancies, consistent reproducibility of the method requires further study. Finally, this study confirmed the difference in ECG parameters for different MMVD stages at the time of the ECG recording. Further longitudinal studies are needed to verify the usefulness of ECG parameters as predictors of CHF.

In conclusion, P wave, QTc1, and QTc2 were significantly longer in MMVD dogs with CHF than in such dogs without CHF. The clinical implication of this study is that prolonged P wave and QTc may improve the prediction of CHF in subclinical MMVD dogs. As ECG is a widely available and cost-effective diagnostic tool, our results indicate that this could be used as a screening tool to predict CHF if our results are verified in future studies.