QT interval instability and variability in dogs with naturally-occurring hypercortisolism

Hypercortisolism is one of the most common endocrine diseases in dogs. In humans, it is clearly associated with a higher risk of cardiovascular events, but studies in dogs are scarce. To investigate the arrhythmogenic risk of dogs with naturally-occurring hypercortisolism (NOHC), indices of variability and instability of the QT interval were retrospectively studied in 38 dogs with NOHC and prospectively studied in 12 healthy dogs: variance (QTv), total instability (TI), short-term (STI) and long-term (LTI), and mean (QTm). Except for QTm, all parameters studied were higher in the NOHC group than in the control group. In addition, STI and QTv showed moderate positive correlation with left ventricle wall thickness. The NOHC group was subdivided according to cortisol suppression pattern in the low-dose dexamethasone suppression test. All electrocardiographic indices of partial and absent suppression patterns were numerically higher than healthy dogs. QTv and TI were lower in the control group than in both NOHC subgroups. LTI and STI were lower in the CG than in the group with the partial suppression pattern. There was no statistical difference between sex groups in any of the electrocardiographic parameters studied. This result might indicate that the etiology of NOHC, and its consequent influence on hypothalamus–pituitary–adrenal axis could interfere on the heterogeneity of ventricular repolarization parameters in different ways, especially in the short-term and the long-term stability; however further studies are necessary to understand the role of cortisol on electrical instability in dogs.


Introduction
Hypercortisolism the most common endocrine disorders in dogs, with prevalence ranging from 0.17 to 1.46% (O'Neill et al. 2016;Carotenuto et al. 2019;Schofield et al. 2021). In people, it is clear that chronic excessive secretion of cortisol can lead to several complications such as diabetes mellitus, central obesity, hyperlipidemia, hypercoagulability, and systemic arterial hypertension (Mancini et al. 2004). These abnormalities promote a higher risk of cardiovascular events, such as coronary artery disease, acute myocardial infarction, stroke and heart failure, which may remain high five years after successful treatment of the disease (Colao et al. 1999;Dekkers et al. 2013;Terzolo et al. 2014;Di Dalmazi et al. 2014;Bancos et al. 2016).
Compared to people, veterinary literature has less information related to cardiovascular risk in dogs with NOHC. Nevertheless, cardiac disease has been reported as the main cause of death in dogs with pituitary-dependent hypercortisolism (Chen et al. 2016). Most dogs with hypercortisolism develop systemic arterial hypertension (Vidal et al. 2018;García San José et al. 2020, and are prone to present left ventricular hypertrophy (Chen et al. 2014;Takano et al. 2015), and systolic dysfunction (Chen et al. 2014). Although no association with atherosclerosis was previously reported (Hess et al. 2003), dogs with hypercortisolism are also at risk for thromboembolic disease, with a prevalence of 6.1% in that population (Hoffman et al. 2018).
In addition to functional and structural complications in the cardiovascular system secondary to hypercortisolism, there is evidence of a higher arrhythmogenic risk. In people, several studies have already demonstrated the association between hypercortisolism and atrial fibrillation (Huerta et al. 2005; Van der Hooft et al. 2006;De Caterina et al. 2010;Koracevic et al. 2020), but studies evaluating the risk for arrhythmias in dogs with this condition are lacking. Nonetheless, owing to the clear evidences in people and the high incidence of NOHC in dogs, investigation of markers for cardiovascular events in this species are indeed warranted.
The electrocardiogram is a low-cost exam widely used in veterinary practice, which can provide precious information in these situations. QT interval corresponds to the duration of the ventricular action potential, being one of the most used electrocardiographic parameters for the diagnosis of cardiac abnormalities (Niemeijer et al. 2015). The prolongation of QT interval has been historically associated with a higher frequency of arrhythmic events (Boulaksil et al. 2011;Itoh et al. 2016). Of note, some studies have demonstrated the inaccuracy of the isolated QT interval for stratification of arrhythmogenic risk (Hondeghem 2008a, b). However, parameters derived from QT, such as the instability and variability of QT, have been further studied, and proved more sensitive and specific for predicting the development of arrhythmias (Limprasutr et al. 2018).
In dogs with myxomatous mitral valve disease, QT interval instability increases with disease progression and have predictive value for the development of ventricular arrhythmias (Brüler et al. 2018). While in people with hypercortisolism the prolongation of QT interval corrected by heart rate (QTc) has been confirmed (Pecori Giraldi et al. 2011), to the best of our knowledge no study has investigated the influence of chronic hypercortisolism on myocardial repolarization in dogs. Thus, this study aimed at evaluating QT interval indices in dogs with NOCH. We hypothesized that an increased prolongation and instability of QT interval would be associated with a higher risk of developing arrhythmias.

Animals
Medical records of dogs diagnosed with NOHC between January 2013 and April 2016 at a veterinary teaching facility were retrospectively analyzed. The inclusion criteria for the hypercortisolism group (HG) were the diagnosis of hypercortisolism through the low-dose dexamethasone suppression test (LDDST) (0.01 mg/kg IV) in dogs with clinical and clinicopathological signs of hypercortisolism, in addition to the presence of an electrocardiographic tracing performed within seven days of confirmation of NOHC. The LDDST in our teaching veterinary facility is performed using the following protocol: collection of the first blood sample (T0h); administration of dexamethasone (0.01 mg/ kg IV); collection of the second blood sample 4 h after the dexamethasone administration (T4h); collection of the third blood sample 8 h after the dexamethasone administration (T8h). The blood samples are centrifuged, their serum is separated, and then subjected to cortisol measurement using the radioimmunoassay method (minimum cortisol concentration measured = 0.01 µg/dL). Due to the retrospective nature of this study, we were unable to classify the etiology of the NOHC as either pituitary-dependent hypercortisolism or adrenal tumor hypercortisolism. Not included in the study were dogs that did not have electrocardiographic tracing, suspected of having iatrogenic hypercortisolism, that were receiving any medication with possible impact on the QT interval, or that were already receiving any treatment for NOHC at the time of electrocardiographic recording. The cortisol concentration results that were available on the medical records of the dogs of HG were included in the analyses. Those LDDST results were classified according to the five patterns of described by Bennaim and colleagues: absence of suppression (T4h and T8h > 1 μg/dl and both > 50% T0h), partial suppression (T4h and T8h > 1 µg/dl, but either < 50% T0h), complete suppression (T4h and T8h < 1 µg/dL), escape (T8h > 1 µg/dL and T4h < 1 µg/dL) or inverse (T4h > 1 µg/ dL and T8h < 1 µg/dL) (Bennaim et al. 2018). To evaluate the presence of left ventricle (LV) concentric hypertrophy, echocardiographic data were also collected from dogs that had this information available in their records. Concentric hypertrophy was defined as diastolic LV free wall (LVFW) and/or interventricular septum (IVS) thickness above the normal range for body weight (Gonçalves et al. 2002). Due to the retrospective nature of the study, the exams were performed by different veterinary cardiologists with equivalent level of training. For the control group (CG), healthy dogs from a Beagle research colony were prospectively enrolled. Only dogs without clinical characteristics compatible with NOHC or any other systemic disease were included in this group. These dogs underwent a thorough clinical examination as well as electrocardiographic evaluation. In both groups, demographic data (breed, sex, age, body weight) was recorded.

Electrocardiographic analysis
In all animals, the electrocardiographic tracing was obtained using computer-based equipment. 1 Bipolar (I, II and III), and unipolar (aVL, aVR and aVF) leads were recorded, with paper velocity adjusted to 50 mm/s and sensitivity calibrated to 1 mV/1 cm. The analysis of the electrocardiographic traces was performed in lead II, and included heart rate, duration of the RR interval, and duration of the QT interval, which corresponds to the beginning of the Q wave to the end of the T wave (Santilli et al. 2019). Fifty consecutive QT intervals were measured and recorded, allowing the calculation of the following parameters: QTm (mean QT intervals); QTc (QT corrected for heart rate); QTv (variance of QT intervals); TI (total instability); LTI (long-term instability); and STI (short-term instability). QTc was calculated according to the formula proposed by Fridericia (Fridericia 1920): The variables TI, LTI and STI were calculated according to the methodology proposed by Van der Linde and colleagues(Van Der Linde et al. 2005), which is based on Poincaré plots ( Fig. 1) constructed from the intersection between QT n (X axis) and QT n+1 (Y axis). TI was calculated by applying the equation described below: Where: The width and length of the plot XY were measured after rotating it in -45° (θ) around its origin. Thus, rotational centers of gravity were obtained, which were later used to calculate LTI and STI. Where:

Statistical analysis
Shapiro-Wilk test was used to investigate normality of the data. Mean, standard deviation, median, interquartiles, minimum and maximum, confidence interval and coefficient of variation values were calculated. Unpaired t test or Mann-Whitney test was used to assess differences between CG and HG, depending on the results of normality test. Chisquare test was used to evaluate difference between both groups regarding sex. To evaluate the correlation between echocardiographic LV measurements and QT mean, variability and instability indices, Spearman test was performed. To minimize body weight bias, systolic and diastolic IVS and LVFW were divided by LV internal diameter (LVD). In addition, diastolic and systolic IVS, LVD and LVFW were normalized for body weight using allometric scalling (Cornell et al. 2004): For the interpretation of the Spearman correlation magnitude, the following classification was adopted: correlation coefficients 0 to 0.1 (negligible), 0.1 to 0.39 (weak), 0.4 to 0.69 (moderate), 0.7 to 0.89 (strong), and 0.9 to 1 (very strong) (Schober et al. 2018).
In addition, depending on the results of normality test, Kruskal-Wallis test followed by Dunn's post hoc test or

Fig. 1
Poincaré plot constructed from the intersection between QTn (X axis) and QTn + 1 (Y axis) in one of the dogs with hyperadrenocorticism included in the study. The circles represent the 49 cartesian combinations formed between 50 sequenced QT intervals (QT1 to QT50), and the black square represents the projection of the gravity center coordinates of the QT intervals on the X and Y axes analysis of variance (ANOVA) followed by Tukey's multiple comparison test was used to check for differences between the suppression patterns in the LDDST. Lastly, Mann-Whitney test was used to evaluate differences attributed to sex in HG. All statistical analyses were performed using the software Graphpad Prism for Windows (v.5.02), considering p < 0.05 as significant.
Regarding the plasma cortisol concentration in the LDDST, 23 (60.53%) dogs had the results of T0h and T8h, and 19 (50%) dogs had the results of T4h (Table 1). Among the dogs with this data available, the complete suppression pattern was observed in one dog (4.35%), the escape pattern in one dog (4.35%), the absent suppression pattern in six dogs (26.09%), and the partial suppression pattern in 12 dogs (52.17%). The three dogs that only had the T0h and the T8h results could not be classified because the absence of the T4h results hinders the differentiation between absence, partial, escape patterns in those cases.

Heart rate, QT mean, variability and instability indices
With the exception of QTm and heart rate, a significant difference was documented between CG and HG in all

Fig. 2 Poincaré plots showing the dispersion of the intersection points between QTn (X axis) and
QTn + 1 (Y axis) in one of the healthy dogs (a), whose total calculated instability was 6.4, and in three dogs with hyperadrenocorticism, whose calculated total instability was 11.6 (b), 16.9 (c) and 25.8 (d). The black squares represent the projection of the center of gravity coordinates of the QT intervals on the X and Y axes parameters studied. QTv, QTc, LTI, STI and TI were higher in HG than in CG (Table 2). There was no statistical difference of heart rate between both groups. Figure 2 illustrates the greater dispersion of points in the Poincaré plot in HG dogs compared to one of the CG dogs, representing wider total QT instability. There was moderate positive significant correlation between QTv and LVFW/LVDd (p = 0.043; r = 0.46), QTv and LVFWsN (p = 0.019; r = 0.53), and STI and LVFW/ LVDd (p = 0.032; r = 0.4).
Since the partial and the absent suppression patterns were observed in the majority of the HG, the comparison of the electrocardiographic indices was performed between those two HG subgroups and CG. All electrocardiographic indices of both partial and absent suppression patterns were numerically higher than healthy dogs. Except for QTm and QTc, there was significant difference between CG and either or both HG subgroups in all analyzes (Table 3). QTv and TI were lower in the CG than in both HG subgroups. LTI and STI were lower in the CG than in the group with the partial suppression pattern.
When comparing the results according to the sex of HG, there was no difference in any of the electrocardiographic parameters studied (QTm p = 0.5338; QTv p = 0.9113; TI p = 0.7625; LTI p = 0.5349; STI p = 0.7503).

Discussion
In this study, we investigated the influence of NOHC in some QT interval variables in dogs. Although QT interval was not prolonged, the variability and instability of total ventricular electric activity was higher in dogs with NOHC compared to healthy dogs. In addition, also compared to healthy dogs, most indices derived from the QT interval were statistically higher in the partial suppression pattern in the LLDST, and QTv and TI were also higher in the absent suppression pattern.
QT interval reflects the duration of cardiac action potential and its prolongation is considered proarrhythmic. The association between QT prolongation and a higher risk of ventricular arrhythmias and sudden death has been discussed in the literature for more than 40 years (Schwartz and Wolf 1978;Straus et al. 2006;Ahmad and Dorian 2007;Hondeghem 2011;Niemeijer et al. 2015). In dogs, the prolongation of QTc has been demonstrated with the progression of several heart diseases (Koyama et al. 2004;Brüler et al. 2018), which might potentially result in higher arrhythmogenic risk. Nonetheless, in spite of the progressive increase of QT interval along the worsening of mitral valve degeneration in dogs, we recently showed that QTc failed to predict the development of ventricular arrhythmias in those animals (Vila et al. 2021).
Curiously, several studies also suggest that prolongation of the QT interval and QTc are not reliable predictors of ventricular arrhythmia (Spier et al. 2001;Hondeghem 2008b;Leonard et al. 2013). In fact, the prolongation of the duration of the action potential can be antiarrhythmic, since it is the mechanism of class III antiarrhythmic agents (Brendorp et al. 2002). Thus, other factors must coexist with QT interval duration abnormalities for the promotion of proarrhythmic effects, such as spatial or temporal heterogeneity of cardiac repolarization (Schneider et al. 2005). Our results showed no difference between the HG and the CG in the mean duration of QT interval (QTm), but significant difference was found in the parameters that assess repolarization temporal heterogeneity through QT variability (QTv) and instability (TI, LTI and STI).
QT interval exhibits spontaneous fluctuations with each beat, reflecting subtle time variations in ventricular depolarization and repolarization. Because ventricular depolarization is much more stable than repolarization, QT variability may be used as a tool to measure the variance of ventricular Table 3 Mean and standard deviation (median; percentil 25/75) QT interval parameters in healthy dogs and in dogs with NOCH, subdivided according to the cortisol concentration suppression pattern of in the LDDST NOHC, naturally-occurring hypercortisolism; LDDST, low-dose dexamethasone suppression test; QTm, mean QT interval; QTc, QT interval corrected for heart rate; QTv, variance of the QT interval; TI, total instability; LTI, long-term instability; STI, short-term instability; HG, hyperadrenocorticism group; CG, control group. *: ANOVA followed by Tukey's multiple comparison test; **: Kruskal-Wallis test followed by Dunn's post hoc test. Values followed by the same letter do not differ from each other by Tukey's or Dunn's test (p > 0.05)

CG
HG p Partial suppression pattern Absent suppression pattern QTm 191.6 ± 11.1 (194.5;183.8/197.1) 195.6 ± 20.92 (197.6;183.4/207.4) 207 ± 23.24 (200;190. 0.0008** TI 6.6 ± 1.5 (6.5; 5.6/7.5) A 11.6 ± 5.6 (9.5; 7.5/13.9) B 11.2 ± 4.4 (9.7; 7.5/15.2) B 0.0015** LTI 5.4 ± 0.8 (5.5; 4.8/5.7) A 8.1 ± 3.6 (6.5; 5.5/11.1) B 8.7 ± 3.9 (8.1; 5.6/11.5) AB 0.0376* STI 2.6 ± 0.8 (2.4; 2.2/2.8) A 5.9 ± 3.7 (4.7; 3.1/6.9) B 4.3 ± 2 (4.5; 2.2/5.5) AB 0.0104** repolarization duration (Baumert et al. 2016). Parameters derived from QT variability have already been shown to be useful tools to investigate pro-arrhythmogenic and nonarrhythmogenic effect of drugs in dogs (Schneider et al. 2005), and to be a strong predictor of ventricular fibrillation in rabbits anesthetized with myocardial ischemia, presenting greater sensitivity and specificity than the QT and QTc intervals (Sarusi et al. 2014;Limprasutr et al. 2018). Therefore, the greater variability and instability of QT in dogs with NOHC as compared to healthy dogs (Tables 2 and 3 and Fig. 2) points to a higher risk for cardiovascular events in that population. Higher arrhythmogenic and death risk in people with hypercortisolism has already been associated with cardiovascular complications, such as systemic arterial hypertension, left ventricular hypertrophy, cardiomegaly and myocardial ischemia (Verdecchia et al. 2003;Dekkers et al. 2013;Chatterjee et al. 2014). Systemic arterial hypertension is observed in 85% of people and 80% of dogs with hypercortisolism (Mancini et al. 2004;Vidal et al. 2018), leading to increased afterload and, consequently, LV concentric hypertrophy. In our study, although blood pressure data were not included, LV concentric hypertrophy was observed in 66,67% of dogs with NOHC, which was similar to previous reports (65%) (Takano et al. 2015). Myocardial hypertrophy results in structural disarray, including fibrosis, collagen accumulation, increased interstitial fibroblasts, and diastolic dysfunction (Shenasa et al. 2015), leading to a 3.4-fold and 2.8-fold increase in the risk of developing supraventricular arrhythmias and ventricular tachycardia/fibrillation in humans, respectively (Chatterjee et al. 2014). Indeed, left ventricular thickness variables correlated positively with QT variability and instability in this study, which has also already been shown in humans (Orosz et al. 2015). These results suggest that dogs with NOHC and LV hypertrophy might present greater fluctuation of the QT interval and susceptibility to cardiovascular complications. Dogs with hypercortisolism may exhibit different patterns of suppression of hypothalamic-pituitary-adrenal axis during LDDST, depending on the etiology of the disease (Zeugswetter et al. 2021). Bennaim and colleagues described five cortisol suppression patterns in the LLDST of dogs with NOHC and well as the prevalence of NOHC etiology in each group. All dogs with adrenal tumor hypercortisolism showed absence of suppression pattern, while the dogs with pituitary-dependent hypercortisolism were distributed among five patterns: absence of suppression, partial suppression, complete suppression, escape or inverse (Bennaim et al. 2018). In our study, most electrocardiographic indices derived from QT interval (QTv, TI, LTI, and STI) were significant higher in dogs with partial suppression pattern (i.e., likely to have pituitary-dependent hypercortisolism (Bennaim et al. 2018)) than in the CG (Table 3). In addition, QTv and TI were also higher in dogs with absent suppression pattern (i.e., likely to present adrenal tumor and/or pituitarydependent hypercortisolism (Bennaim et al. 2018)) than in the CG. Although the LDDST result alone is not capable of determining the etiology of NOHC in all cases (Bennaim et al. 2018), the different results in the analyzes cited above might indicate that the etiology of hypercortisolism, and its consequent influence on hypothalamic pituitary-adrenal axis (HPAA), could interfere on the heterogeneity of ventricular repolarization parameters in different ways. Since dogs with the partial suppression pattern might present pituitary-dependent hypercortisolism, it could be hypothesized that the short-term and the long-term stability could have higher specificity and be more influenced by this etiology of NOHC. At least in people with either adrenal tumor or pituitary-dependent hypercortisolism, the risk of death and cardiovascular events was shown to be similar (Dekkers et al. 2013). Nevertheless, to the best of authors' knowledge, there are no studies which investigated such characteristic in dogs. Further studies are necessary to understand the role of NOHC etiology on electrical instability in dogs.
Lastly, while studies in people have reported that women have higher QTc than men (Linde et al. 2018), other studies have found no difference between sex in QT variability (Bonnemeier et al. 2003;Krauss et al. 2009), which is similar to our findings in dogs.
An important limitation of this study is its retrospective design, which implies data collection carried out by different observers, and limits the acquisition of some data, such as the presence of comorbidities and the results of exams other than the LDDST. Although all the veterinary cardiologists who performed the cardiological exams had an equivalent level of training, echocardiographic measurements can be influenced by interobserver variation (Hsue and Visser 2020). It is known that ventricular repolarization rates may change with heart diseases (Brüler et al. 2018;Vila et al. 2021), with non-cardiac diseases (Armstrong et al. 2017;Kim et al. 2021), electrolyte disturbances (Yelamanchi et al. 2001), as well as with antiarrhythmic drugs administration (Shantsila et al. 2007). Since LDDST cannot distinguish the etiology of the NOCH in all cases (Bennaim et al. 2018), the lack of results of other exams, including adrenal or pituitary imaging impairs the classification of the NOCH and represents the major limitation of this study. This fact does not allow the differentiation between dogs with adrenal tumor hypercortisolism, pituitary-dependent hypercortisolism or both, which may significantly impair the discussion regarding the influence of the origin of the NOHC in repolarization abnormalities. In addition, information such as systemic blood pressure, blood gases, and 24 h Holter recording to thoroughly investigate the presence of arrhythmias could enrich the understanding of the pathophysiological mechanisms involved in the development of cardiovascular complications in dogs with NOHC. Finally, since repolarization indices are influenced by age (Piccirillo et al. 2013) and obesity (Nigro et al. 2010) in people, weight and specially age differences between CG and HG could have induced a bias in the analyses comparing them.
This study demonstrated that NOHC play a role on temporal heterogeneity of the QT interval, since the QT variability and instability indices were higher in dogs with NOHC than in healthy dogs, and showed moderate positive correlation with left ventricle thickness. In addition, fluctuation of ventricular repolarization indices was observed according to the suppression of HPAA during the LDDST. Future studies that focus on the dynamics of cardiac repolarization in dogs with NOHC are encouraged to better understand the relationship between HPAA and cardiac electrical instability.
Authors' contributions Beatriz de Carvalho Pato Vila, Marcela Sigolo Vanhoni, and Marlos Gonçalves Sousa contributed to acquisition of the data, participated in the statistical analysis, and wrote and edited the manuscript.
Funding This research was supported by Coordination of Superior Level Staff Improvement (CAPES), Brazil.
Code or data availability (software application or custom code) Not applicable.

Ethics approval
The study was entirely conducted in a Veterinary Teaching facility following the guidelines outlined in the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Consent to participate
Due to the retrospective nature of this investigation, institutional policies do not require a written consent of animal owners prior to inclusion of data in the study, as well as to publish manuscripts.

Consent for publication
Due to the retrospective nature of this investigation, institutional policies do not require a written consent of animal owners prior to inclusion of data in the study, as well as to publish manuscripts.

Conflict of interest statement
The authors have no conflict of interest to disclose.