Atrial fibrillatory rate as predictor of recurrence of atrial fibrillation in horses treated medically or with electrical cardioversion

Summary Background The recurrence rate of atrial fibrillation (AF) in horses after cardioversion to sinus rhythm (SR) is relatively high. Atrial fibrillatory rate (AFR) derived from surface ECG is considered a biomarker for electrical remodelling and could potentially be used for the prediction of successful AF cardioversion and AF recurrence. Objectives Evaluate if AFR was associated with successful treatment and could predict AF recurrence in horses. Study design Retrospective multicentre study. Methods Electrocardiograms (ECG) from horses with persistent AF admitted for cardioversion with either medical treatment (quinidine) or transvenous electrical cardioversion (TVEC) were included. Bipolar surface ECG recordings were analysed by spatiotemporal cancellation of QRST complexes and calculation of AFR from the remaining atrial signal. Kaplan‐Meier survival curve and Cox regression analyses were performed to assess the relationship between AFR and the risk of AF recurrence. Results Of the 195 horses included, 74 received quinidine treatment and 121 were treated with TVEC. Ten horses did not cardiovert to SR after quinidine treatment and AFR was higher in these, compared with the horses that successfully cardioverted to SR (median [interquartile range]), (383 [367‐422] vs 351 [332‐389] fibrillations per minute (fpm), P < .01). Within the first 180 days following AF cardioversion, 12% of the quinidine and 34% of TVEC horses had AF recurrence. For the horses successfully cardioverted with TVEC, AFR above 380 fpm was significantly associated with AF recurrence (hazard ratio = 2.4, 95% confidence interval 1.2‐4.8, P = .01). Main limitations The treatment groups were different and not randomly allocated, therefore the two treatments cannot be compared. Medical records and the follow‐up strategy varied between the centres. Conclusions High AFR is associated with failure of quinidine cardioversion and AF recurrence after successful TVEC. As a noninvasive marker that can be retrieved from surface ECG, AFR can be clinically useful in predicting the probability of responding to quinidine treatment as well as maintaining SR after electrical cardioversion.


| INTRODUC TI ON
Atrial fibrillation (AF) is the most common sustained cardiac arrhythmia in horses with an estimated prevalence of 0.1%-2.5% depending on breed and the sample studied. [1][2][3][4] Atrial fibrillation is associated with substantially reduced athletic performance 5,6 and is therefore a clinically important disease. Treatment options include medical treatment and electrical cardioversion. The most used medical treatment is nasogastric administration of quinidine sulphate that successfully restores sinus rhythm (SR) in 80% of the reported cases. The development of severe adverse effects such as colic, and ventricular tachyarrhythmias may necessitate cessation of the treatment before cardioversion has occurred. 7 Other medical treatments have been applied to cases with both induced and spontaneous AF, however, all with lower success rates. [8][9][10][11][12] Transvenous electrical cardioversion (TVEC) has a high success rate, where 96% of the cases reported are successfully cardioverted. 7 The high cardioversion rates are, however, followed by high AF recurrence rates (15%-43%) regardless of treatment method. [13][14][15][16] Atrial fibrillation is a complex disease that remodels the atria, which further promotes AF and the high recurrence rates are believed to be linked to this remodelling. 17,18 Although factors such as breed, AF duration, number of previous AF episodes, valvular regurgitation, atrial size and contractile function as well as increased numbers of atrial premature beats following cardioversion are reported as negative prognostic indicators in horses, 13,15,16 additional biomarkers that are objectively quantified indicating the degree of remodelling and therefore, potential AF recurrence, are warranted. An acknowledged hallmark of atrial remodelling is shortening of the atrial effective refractory period (aERP) 19 and thus shortening of atrial fibrillatory cycle length (AFCL). The mean AFCL can be measured invasively with a catheter placed in the atrium and has been extensively studied in the right atrium of horses with AF. 9,20,21 The atrial fibrillatory rate (AFR) measures the number of fibrillations (f waves) per minute (fpm) and is inversely correlated with AFCL. The AFR can be measured noninvasively from surface ECG recordings and has received considerable attention in horses over the recent years. 12,[22][23][24][25] Both the AFCL and the AFR are generally accepted as surrogate markers for atrial electrical remodelling. 26 Studies in horses have shown the correlation between AF duration and AFR or AFCL, indicating electrical remodelling over time. 20,23,27 Correlation between AFCL measured invasively and AFR derived from the surface ECG has recently been investigated. 25 Also, changes in AFR in response to treatment with antiarrhythmic drugs and towards cardioversion, both with and without drug interventions, have been reported in horses. 22,24 In human AF patients, AFR and AFCL have been proposed as prognostic biomarkers for monitoring the effect of antiarrhythmic drugs [28][29][30][31] and baseline AFR has been proposed as a predictor of spontaneous cardioversion. 32 The objective of this study was to evaluate if AFR derived from surface ECG could predict the outcome of AF treatment in horses. The aim of this study was therefore (1) to evaluate if AFR was associated with successful cardioversion and (2) to evaluate if AFR was associated with AF recurrence after successful treatment. We hypothesised that (1) AFR was higher in horses that did not cardiovert to SR after treatment and (2) horses with high AFR had a higher recurrence rate. The inclusion criteria were as follows: (1) admission for cardioversion using quinidine or TVEC with persistent AF, (2) availability of ECG with AF exported in digital format and (3) ECG recorded at rest prior to any treatment. Also, cardiac auscultation with murmurs graded on a 1-6/6 scale and echocardiographic examination at rest before cardioversion should preferably be available. From the study period, AF episodes were collected from 213 horses. Each horse was only included once in the data set and only for one treatment (either quinidine or TVEC).

| MATERIAL S AND ME THODS
Of the 213 horses, ECGs from 18 horses were excluded due to either a poor-quality recording or that ECG recordings were only available during antiarrhythmic treatment, leaving 195 horses to be included in the study (Figure 1). Conclusions: High AFR is associated with failure of quinidine cardioversion and AF recurrence after successful TVEC. As a noninvasive marker that can be retrieved from surface ECG, AFR can be clinically useful in predicting the probability of responding to quinidine treatment as well as maintaining SR after electrical cardioversion.

K E Y W O R D S
AFR, arrhythmia, cardiology, ECG, electrophysiology, horse, surface electrocardiogram regurgitation was classified by the diameter of the regurgitation compared with left ventricular outflow tract from Colour Flow Doppler and classified in the following categories: none, trivial, mild, moderate and severe. The measurements were provided by each referral centre and therefore conducted by several observers. The AF duration was estimated based on the clinical records and anamneses, where the referring veterinarian, owner, trainer or rider estimated AF duration. Previous AF treatment was registered and the duration of the follow-up period as well as eventual AF recurrence within 180 days after treatment were based on information from the medical record.
The following information from the treatment procedures was registered: treatment with quinidine or TVEC. For quinidine treatment (via nasogastric intubation), dose per treatment (mg/kg), number of doses given and total quinidine dose (g) for the treatment period were recorded. For TVEC treatment, the cardioverting energy (J), total energy delivery (J) and the number of shocks were recorded. For all horses, it was noted if additional antiarrhythmic drugs were given before, during or after treatment. Treatment outcome (cardioversion to SR or not) was registered. Modified base-apex ECG recordings were exported as raw data files for off-line processing.
Three 1-minute intervals from the ECG recordings were used for the analyses. 24 If more than three 1-minute ECG recordings were analysed, the first three 1-minute recordings, where >80% of the atrial signals was used for the analyses, were selected. The 1-minute recordings followed each other consecutively with no time span in between. If less than three 1-minute ECG recordings were of acceptable quality for analysis, one 1-minute ECG recording was included.
Analyses and selection of the intervals were performed by one operator without any knowledge of clinical data. The ECG analyses, including QRST cancellation and AFR calculation, were performed using the AFR software (Cardiolund AFR Tracker, Cardiolund) as previously described. 33,34 In short, ECG signals were preprocessed (digital ECG data from different sources adapted to the input format required by the analysis software), filtered, and QRST signals removed by spatiotemporal cancellation. The remaining residual ECG then mainly contained atrial activity and from these, the AFR could be calculated over a specified time period, which is reported as the AFR expressed in fibrillations per minute.
Electrocardiograms from three horses had been used in a previous study, 22 ECGs from another 73 horses have been included in a recent study 25 and 19 of those have also been used in a second study. 20

| Data analysis
Descriptive analysis was used to compare demographics based on age, sex, body weight, breed and use. Categorical variables were expressed as a percentage and continuous variables were presented either by means and standard deviations (SD) for normally distributed data or as median (interquartile range, IQR) for nonparametric data. Normality was checked by visual inspection of the raw data plots and by using the Shapiro-Wilk test of linearity. Friedman's test for comparison of paired groups was used for assessment of AFR stability by testing differences between AFR values calculated from the three 1-minute recordings.

F I G U R E 1 Schematic flowchart of the horses included in the study and reasons for exclusion of horses
Differences between the horses that were cardioverted with quinidine and by TVEC were examined using Chi-square tests for categorical variables, independent t tests for continuous, normally distributed variables and Mann-Whitney U test for data that were not normally distributed.
For the horses treated with quinidine, differences of AFR between successful and no successful cardioversion was examined using Mann-Whitney U test. Additionally, the value of AFR for prediction of SR restoration following quinidine administration was assessed using ROC curve analysis. As only one horse did not cardiovert to SR with TVEC, these analyses could not be conducted for the TVEC group.
Risk assessment of AF recurrence by 180 days after cardioversion was studied using Kaplan-Meier survival curve and Cox regression analyses in those animals who underwent successful cardioversion and were included in the longitudinal follow-up. A log-rank test was used to compare survival probabilities between the groups. Proportional hazard assumption was tested using Schoenfeld partial residuals method. The horses were divided into groups based on the intervention (quinidine or TVEC) and the analyses were performed separately for the quinidine and TVEC treated animals. The association between AFR and the likelihood of AF recurrence after cardioversion was assessed using Kaplan-Meier survival curve and Cox regression analyses. Horses were censored at the time of the last follow-up if no AF recurrence had occurred until then.
For the Kaplan-Meier analysis, AFR was dichotomised by median values separately for the quinidine (median value 350 fpm) and TVEC (median value 376 fpm, but for simplicity, we chose 380 fpm as the close to median value) treated animals and the significance of differences between the groups depending on the AFR value was assessed using log-rank P value. Cox regression analysis for assessment of the relationship between AFR and the risk of AF recurrence was performed using AFR as a continuous variable and dichotomised by the close to median value as explained above. The results of Cox regression analysis are presented as unadjusted hazard ratio (HR) as well as HR adjusted for clinically relevant covariates (age, breed, duration of AF prior to cardioversion, moderate or severe mitral regurgitation measured by echocardiography, LAA and administration of antiarrhythmic drugs following restoration of SR).
The significance threshold was set at P < .05. Statistical analyses were performed using SPSS version 26.0 and GraphPad Prism,

| Case selection and demographical characteristics
Of the 195 horses included, 167 had no history of previous AF, whereas 28 horses had been treated for AF before and now presented with recurrence. Seventy-four horses were treated with quinidine (two horses received quinidine gluconate intravenously and the specific dose for these two horses were not registered, whereas the remaining horses received 22 mg/kg quinidine sulphate via nasogastric intubation administered with an increasing number of doses over one or more days).
Of the 74 horses receiving quinidine, 21 were treated at Rossdales Equine Hospital, 16 at Swedish University of Agricultural Sciences, 16 at University of Copenhagen, 10 at University of Pennsylvania, 9 at University of Zurich and 2 at Dubai Equine Hospital. One hundred and twenty-one horses were treated with TVEC (85 Ghent University, 15 University of Pennsylvania, eight University of Zurich, seven Dubai Equine Hospital, five Rossdales Equine Hospital, and one at University of Copenhagen). Increasing energy delivery (J) was applied for each shock until cardioversion occurred. Eleven horses did not cardiovert to SR (10 quinidine and 1 TVEC treated horse) and another 10 horses had no follow-up data on AF recurrence. Therefore, in total, 174 horses were subjected to the Kaplan-Meier and Cox regression analyses to investigate the time to AF recurrence (flowchart, see Figure 1). Clinical characteristics for horses successfully cardioverted by quinidine or TVEC including breed, body weight, age, cardiac murmurs, valvular regurgitations, left atrial and ventricular size including the number of animals where the variable had been registered are summarised in Table 1. The majority of horses included in the TVEC group were Warmbloods (84%), whereas racehorses accounted for 58% of the horses included in the quinidine group. Horses treated with TVEC were significantly heavier, older, had a higher prevalence of mitral regurgitation and larger LAA, compared with the horses treated with quinidine (Table 1).
Of the 184 horses successfully cardioverted to SR, 17 had only one 1-minute ECG recording of acceptable quality for AFR analysis. For the remaining horses, no difference was found between the three 1-minute recordings (mean ± SD values; AFR minute 1:368 ± 46 fpm, AFR minute 2:367 ± 46 fpm, AFR minute 3:367 ± 47 fpm, P = .4) and therefore only the first 1-minute AFR recording was used for the following analyses.
The AF characteristics, such as the estimated AF duration before treatment, recurrence rate, AFR, as well as treatment information, are summarised in Table 2. Duration of AF was longer for the successfully cardioverted horses in the TVEC group compared with the quinidine group. The recurrence rate 180 days after cardioversion and the AFR before treatment were lower in the quinidine group compared with horses treated with TVEC ( (P = .03). The ten noncardioverting horses treated with quinidine consisted of seven Warmbloods, one Standardbred and two other breeds, eight were geldings and two mares. Body weight was 624 ± 67 kg, age 9 (8-13) years, three horses had systolic murmur on the left side (grade 2-3/6) and six of the horses had mild or moderate mitral regurgitation.
They had AF for 120 (45-195) days before the cardioversion attempt.

| Immediate cardioversion success
Sinus rhythm was restored in 64 (86%) of the horses receiving quinidine, while in 10 horses, cardioversion was unsuccessful. The AFR assessed amongst noncardioverters from the quinidine group was significantly higher than in those who cardioverted successfully (383 In the TVEC group, SR was restored in all (99%) but one horse (AFR 402 fpm) and therefore no statistical analysis was deemed applicable.

| Quinidine group
In total, 57 horses were included in the follow-up after the successful restoration of SR by quinidine. Median AFR at cardioversion in TA B L E 1 Clinical characteristics for horses successfully cardioverted on quinidine (left) and with transvenous electrical cardioversion (TVEC; right). Categorical variables were expressed as a percentage and continuous variables were presented either by means and standard deviations (SD) for normally distributed data or as median (interquartile range, IQR) for nonparametric data. The P value indicates that there is a statistically significant difference when the quinidine and TVEC groups are compared in univariable analysis, and n is the number of animals where the variable has been registered in the clinical records In the Kaplan-Meier curve analysis using AFR value dichotomised by median, no difference was observed between the groups with regard to the likelihood of AF recurrence during the follow up (log-rank P = .608, Figure 2).

| TVEC group
A total of 117 horses were included in the follow-up after the success- Kaplan-Meier survival curve and Cox regression analyses. The Kaplan-Meier analysis (Figure 2) demonstrates that elevated AFR was associated with the risk of AF recurrence. In the Cox regression analysis, increased AFR was significantly associated with AF recurrence both in the univariable analysis and after adjustment for clinically relevant covariates (Table 3).

| DISCUSS ION
This study is the first multicentre study to address electrical remodelling measured by AFR as a predictor of cardioversion success and AF recurrence in horses. We found that higher AFR was associated with the failure of quinidine cardioversion and that higher AFR was associated with a higher risk of AF recurrence for horses cardioverted with TVEC. Amongst horses that responded to quinidine treatment, AF recurrence rate during follow-up was low and not associated with AFR.
Atrial fibrillation results in structural, electrical and functional remodelling of the atria, and delayed or insufficient reversibility of these changes will lead to increased inducibility and sustainability of the disease, which may prevent maintenance of SR after cardioversion. 35 Atrial fibrillatory rate is a measure of electrical remodelling and has recently been applied in equine AF research for studying disease development over time in both horses with induced AF 12,23 and horses with spontaneous persistent AF. 22,25 In addition, the effect of antiarrhythmic treatment on the AFR 12,22,24 has been studied.
These studies have shown that AFR increases in experimentally induced acute AF, where the AFR is reported to be approximately 300 fpm, 12,22 increasing to approximately 350 fpm after 30 days of AF 12 and to 376 fpm after 2 months of AF. 23 In human studies, it has been shown that low-frequency fibrillation was more probably to terminate spontaneously or respond to antiarrhythmic therapy, whereas high-frequency fibrillation was more persistent and drug refractory 30,36,37 and therefore AFR has been suggested as a predictor of cardioversion success. Clinical studies in human patients have reported AFR cut-off values predicting spontaneous cardioversion being <355 fpm. 26,32 In horses with induced AF, it was also shown that low AFR is more probably to terminate spontaneously compared with horses with high AFR values. 22 This is similar to the findings of the current study where horses with unsuccessful quinidine cardioversion had higher AFR compared with horses where cardioversion was successful. As almost all horses cardioverted due to the very high success rate of TVEC, we were not able to calculate the predictive value of AFR for TVEC.
In the current study, we found a significant difference in AFR between horses relapsing into AF and those maintaining SR over a 180day follow-up period for the horses treated with TVEC. The findings correspond to studies in human AF patients, where AFR was higher in patients with AF recurrence the following cardioversion. 30,[36][37][38] This has also been observed in a study including 18 horses electrically cardioverted, where the shortest fifth percentile of AFCL (corresponding to high AFR) has been reported as a risk factor of AF recurrence. 20 The above-mentioned findings all indicate that remodelling, whether it is electrical or structural, resulting in higher AFR will increase the risk of AF recurrence after cardioversion. Based on the clinical characteristics, there are noticeable differences between the horses successfully treated with quinidine and with TVEC, and therefore, the outcome of the two groups cannot be directly compared. The TVEC group included larger and older horses (primary Warmbloods), which had a higher prevalence of mitral regurgitation, larger LAA and longer duration of AF. Similar phenotypic characteristics were observed for the ten noncardioverting horses treated with quinidine. The described clinical characteristics of the noncardioverting quinidine treated horses and the TVEC group are in concurrence with risk factors for AF previously reported such as mitral regurgitation, increased LAA and AF duration. 13,15,39 The combination of high AFR and increased LAA has predicted early AF recurrence after successful cardioversion in human patients, providing individual risk estimates in human AF patients. 28 A certain size of the atria is believed to maintain a critical high number of re-entry circuits and large atria therefore can accommodate a higher number of co-existing f waves resulting in increased AFR. 40 The findings in the current study suggest more remodelled atria in the TVEC group and therefore, this group is most probably more prone to AF recurrence compared with the quinidine group.
In human patients, the AF duration affects the significance of the ECG analysis is post-processing, but future ECG recording and analysis systems may be able to calculate AFR instantaneously, which would provide the clinician with information that can assist not only in treatment selection but also in prognostication of the risk of AF recurrence.

| Limitations
Due to the retrospective nature, the data set is based on medical records, and the follow-up strategy varies between the centres where some active follow-up the patients, whereas others have no follow-up strategies. Therefore, the detailed treatment as well as the follow-up protocols may not have been identical between the centres. Duration of AF is difficult to estimate, especially for riding or pleasure horses where AF may have been present for weeks or months unnoticed by the owner or rider. In contrast, AF will seldom be ignored in racehorses due to its performance-limiting effect. These horses are therefore often treated early when the atria are less remodelled and fewer AF recurrences are seen. Finally, the two treatment groups were of variable size, with fewer quinidine cases, and the horses were not randomly allocated to the groups. The low number of horses treated with quinidine affects the strength of the findings in this subgroup and therefore, the findings related to the use of quinidine should be interpreted with caution.

| CON CLUS ION
The current study applied a signal-processing technique, which noninvasively measured the dominant frequency of AF waves from a bipolar surface ECG as a biomarker of electrical remodelling. We included 195 horses from seven different referrals centres making this the largest known AF study in horses. We demonstrated that AFR was associated with cardioversion success following quinidine treatment, and for horses treated with TVEC, higher AFR was associated with an increased risk of AF recurrence. Useful indices to identify candidates with a high risk of AF recurrence after cardioversion are needed, and the AFR could be clinically valuable for predicting the outcome of maintaining SR after cardioversion.

ACK N OWLED G EM ENTS
Ellen Paulussen (DVM), Mélodie Schneider (vet. med., Equine Department, University of Zurich, Switzerland) and veterinary student Mathias A. Dinesen are acknowledged for help with data collection.

CO N FLI C T O F I NTE R E S T S
No competing interests have been declared.

AUTH O R CO NTR I B UTI O N S
R. Buhl and E. Hesselkilde participated in study design, study execution, data analyses and interpretation and preparation of the manuscript. R.
Buhl also has full access to the data and takes responsibility for the Nielsen participated in study execution, and data analyses and interpretation. All authors gave their final approval of the manuscript.

E TH I C A L A N I M A L R E S E A RCH
The local ethical committee at the Department of Veterinary Clinical Sciences, University of Copenhagen approved the study.

I N FO R M ED CO N S ENT
Owner informed consent is not stated for all hospitals involved.

PEER R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1111/evj.13551.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data are available from the corresponding author on reasonable request.