Echocardiographic predictors of first onset of atrial fibrillation in dogs with myxomatous mitral valve disease

Abstract Background Atrial fibrillation (AF) occurs in dogs with myxomatous mitral valve disease (MMVD) as a consequence of left atrial (LA) dilatation, and it affects survival and quality of life. Objectives To evaluate the usefulness of echocardiography in predicting the first occurrence of AF in dogs with MMVD. Animals Forty‐four client‐owned dogs with MMVD, 22 dogs that developed AF, and 22 dogs that maintained sinus rhythm. Methods Retrospective observational study. Medical databases were reviewed for dogs that developed AF during the year after diagnosis of MMVD (AF group). The last echocardiographic examination obtained while still in sinus rhythm was used to derive selected variables. For each dog with AF, a control dog matched for body weight, class of heart failure, and LA dimension was selected. Echocardiographic results including LA volumes and LA speckle tracking echocardiography (STE)‐derived variables were measured. Results Among the tested echocardiographic variables, only LA diameter (P = .03) and left ventricular internal diameter in diastole (P = .03) differed significantly between groups, whereas body weight‐indexed variables of cardiac dimension as well as LA volumes and volume‐derived functional variables were not different. Among the STE‐derived variables, peak atrial longitudinal strain (PALS) results differed significantly between the AF group (23.8% ± 8.6%) and the control group (30.5% ± 9.6%; P = .03). A value of PALS ≤28% predicted AF occurrence with sensitivity and specificity of 0.80 and 0.65, respectively. Conclusions and Clinical Importance Absolute cardiac diameters and LA STE (in particular, PALS) are useful echocardiographic predictors for the development of AF in dogs with MMVD.

response rate is another pivotal factor affecting clinical presentation and outcome in dogs with AF. Animals with increased heart rate (HR), as measured instantaneously or by calculating the mean HR over 24 hours using a Holter monitor, or poor pharmacological rate control have worse outcome. 1,4,5 Therefore, appropriate control of mean HR using nodal blockers, such as digoxin and diltiazem, should be carried out to increase life expectancy of these patients. 5 Thus, early detection of AF in dogs with cardiac disease is extremely important in order to begin effective treatment and improve survival.
In people, several studies have evaluated the usefulness of surface ECG and transthoracic echocardiography in predicting the future occurrence of AF. [6][7][8][9] Among these, advanced echocardiographic techniques such as LA tissue Doppler imaging and LA speckle tracking echocardiography (STE) have been shown to be valuable tools for predicting AF occurrence in humans with cardiac disease or normallystructured hearts. [10][11][12][13] Left atrial tissue Doppler imaging and STE also have been applied in both clinically healthy dogs and dogs with MMVD. [14][15][16][17][18][19] In particular, LA STE-derived strain has been shown to decrease progressively as the disease advances, 15 and it is a predictor of cardiac death in dogs with MMVD. 17 Our aim was to evaluate the clinical usefulness of several indices of LA and left ventricular (LV) morphology and function to predict the future occurrence of AF in a population of dogs with naturally occurring MMVD.

| Animals
The medical databases of the veterinary teaching hospitals of the Universities of Bologna and Padova were retrospectively reviewed from January 2011 to December 2018 searching for all dogs with MMVD that developed AF during the 12 months after diagnosis of MMVD and inclusion in the study (AF group). Only dogs with at least 1 echocardiographic examination performed before the occurrence of AF that showed normal sinus rhythm (NSR) were considered. The diagnosis of AF was made by analyzing a 6-or 12-lead surface ECG by a board-certified cardiologist. Information regarding clinical variables including breed, sex, BW, and age; presence or absence of CHF; class of heart failure 20 ; drugs received; and the time between the last echocardiographic examination performed in NRS and the diagnosis of AF were recorded. Echocardiographic examinations were retrieved from a dedicated server as digital imaging and communications in medicine (DICOM) files and analyzed by a single operator using dedicated software for image analysis (OsiriX MD Software, Pixeo SARL, Geneva, Switzerland; QLAB quantification software version 9.1, Philips Healthcare, Monza, Italy). A control group was created by selecting dogs with MMVD during the same period of time. These animals matched those of the AF group based on similar signalment, same heart failure class, and similar LA dimension as expressed as the ratio between the diameter of the left atrium and aorta described below.
Dogs in the control group had a follow-up time that was longer than the time between the last echocardiographic examination in NSR and the occurrence of AF for the corresponding patient in the AF group.

| Echocardiography
Transthoracic echocardiographic examinations were reviewed and analyzed by a board-certified cardiologist to confirm the diagnosis of MMVD and obtain LA and LV morphological and functional variables as well as Doppler-derived variables. All examinations were performed using echocardiographic units (iE33 and CX50 ultrasound systems, Philips Healthcare) from the same manufacturer with similar software and technology, and equipped with a dedicated multifrequency phased array transducer (S5-1 pure-wave phased array transducer, Philips Healthcare). Echocardiographic diagnosis of MMVD was made by observing typical thickening, prolapse of the mitral valve, or insufficiency of the mitral valve on color Doppler examination, or a combination of these. Wall thickness and chamber diameter of the LV were obtained from 2-dimensional guided M-mode images acquired from a right parasternal short axis view at the level of the papillary muscles.
In particular, LV end-diastolic and end-systolic dimensions were normalized to BW and compared to previously published reference intervals. 21 Fractional shortening then was calculated using the appropriate formula. The LA diameter and aortic diameter were measured on 2-dimensional images obtained from a right parasternal short axis view at the basilar level, as previously described. 22 The LV diastolic inflow was recorded from the left apical 4 chamber view by placing the sample volume at the level of the mitral valve on the ventricular side. The peak velocity of the early diastolic flow wave, peak velocity and duration of the late diastolic flow wave, and ratio between the 2 peak velocities were measured.
The LA volumes were measured only from a left apical 4-chamber view using the area-length method, as previously described. 23 The following variables were calculated: maximal LA volume (LAV max); minimal LA volume (LAV min); LA volume at the onset of the P wave  and LAP indexed to BW were calculated by dividing these variables by the BW of the dog expressed in kilograms.
The LA STE was performed as described in a previous study. 15 Briefly All measurements were replicated on 3 consecutive beats, and the mean values then were calculated. The HR provided by the continuous ECG recording performed during each echocardiographic examination was annotated.

| Variability study
Interobserver and intraobserver variability of LA STE already has been studied in dogs with MMVD, and it appeared to be clinically acceptable. 15 Because 2 different echocardiographic units were used at 2 enrollment centers, the effect of the echocardiographic machine was evaluated by selecting 5 dogs not included in the study population. In particular, 5 dogs with or without MMVD were recruited from the university students and hospital staff, and a complete echocardiographic examination was repeated using both machines during the same day in the same dogs, approximately 1 hour apart. At the end of the variability study, the operator performed the STE analysis on the 10 studies that were anonymized to keep the operator blind to the dogs' identities. Variability then was calculated using the coefficient of variation, as previously described. 24  3 | RESULTS

| Demographic variables
Forty-four dogs were included in the study (22 dogs for each group).
No difference was found between groups for BW (P = .8), age (P = .99), and sex (P = .5; Table 1). In the AF group, the most common  , and the optimal cutoff to predict future onset of AF was ≤28%, with sensitivity and specificity of 0.80 and 0.65, respectively ( Figure 1; Table 2).

| Variability study
The variability between the 2 echocardiographic machines was clinically acceptable (<15%) for all STE-derived variables. In particular, the corresponding coefficients of variation were 12.9, 14.8, and 11.0% for PALS, PACS, and the contraction strain index, respectively.

| DISCUSSION
We showed that, in dogs with MMVD, absolute LA and LV dimensions are higher in those subjects that develop AF during the year after the echocardiographic examination. However, only STE-derived PALS represents an independent predictor of AF development in these dogs after univariate analysis, with an increased risk of arrhythmia when PALS ≤28%.
The occurrence of AF in dogs with CHF usually is the result of LA remodeling with chamber dilatation and interstitial fibrosis. 25 Because a critical atrial mass is needed to perpetuate the arrhythmia, 26 dilatation of the left atrium usually is present in small animals. This factor is even more important in small breed dogs, such as those that typically are affected by MMVD, in which LA dilatation usually accompanies the most advanced stages of the disease. Body weight is a risk factor for the development of AF in dogs, 2 and giant breeds usually are more frequently affected by the arrhythmia. 1 In our study, all dogs with MMVD were included, which is a different approach from that of a previous study that evaluated the prognostic relevance of AF in medium-to-large-breed dogs (>15 kg) with MMVD. 4 Therefore, the animals included in our study had severely dilated LA likely associated with relevant atrial remodeling and fibrosis that allowed for maintenance of the arrhythmia. Therefore, studying LA dimension using echocardiography can give relevant information about the stratification of patients with cardiac disease, because several studies in people have indicated that LA enlargement represents a risk factor for development of cardiac events, including first onset of AF. 8,[27][28][29] The dilatation of the LA is not a simple elongation of the myocardial fibers, but an active process of myocardial remodeling. This remodeling implies myofiber hypertrophy, changes in the interstitial matrix, and fibrosis. [30][31][32] In a canine model of CHF induced by tachy-pacing, the heart failure itself promoted atrial interstitial fibrosis, with subsequent abnormal local conduction and occurrence of sustained AF. 25 This process, together with abnormal LA afterload and diastolic dysfunction of the left ventricle, leads not only to enlargement of the chamber, but also to abnormal systolic properties. 33,34 In people, abnormal LA function represents a stronger predictor of AF occurrence when compared to echocardiographic evaluation of LA dimensions alone. 35  Our study has some limitations. First, the number of dogs enrolled was low, which was a consequence of the difficulty of locating patients with MMVD that developed AF during the follow-up time and the low prevalence of arrhythmia in this specific population of animals. Second, although LA STE has been determined to be feasible and reproducible in dogs, 15,18 the technique requires high-quality images, dedicated software, and expertise for the analysis. Moreover, LV STE has been shown to produce different results according to the manufacturer of the machine used for the analysis. 39 Thus far, it is unknown if this phenomenon also is true for LA STE, but it is possible that studies performed using different software will deliver different results.
In conclusion, we showed that, together with absolute LA and LV diameters, the use of LA STE could predict the future development of AF in dogs with MMVD. In particular, PALS was the best predictor of AF in these animals.