Electrical impedance tomography to measure lung ventilation distribution in healthy horses and horses with left‐sided cardiac volume overload

Abstract Background Left‐sided cardiac volume overload (LCVO) can cause fluid accumulation in lung tissue changing the distribution of ventilation, which can be evaluated by electrical impedance tomography (EIT). Objectives To describe and compare EIT variables in horses with naturally occurring compensated and decompensated LCVO and compare them to a healthy cohort. Animals Fourteen adult horses, including university teaching horses and clinical cases (healthy: 8; LCVO: 4 compensated, 2 decompensated). Methods In this prospective cohort study, EIT was used in standing, unsedated horses and analyzed for conventional variables, ventilated right (VAR) and left (VAL) lung area, linear‐plane distribution variables (avg‐max VΔZLine, VΔZLine), global peak flows, inhomogeneity factor, and estimated tidal volume. Horses with decompensated LCVO were assessed before and after administration of furosemide. Variables for healthy and LCVO‐affected horses were compared using a Mann‐Whitney test or unpaired t‐test and observations from compensated and decompensated horses are reported. Results Compared to the healthy horses, the LCVO cohort had significantly less VAL (mean difference 3.02; 95% confidence interval .77‐5.2; P = .02), more VAR (−1.13; −2.18 to −.08; P = .04), smaller avg‐max VΔZL Line (2.54; 1.07‐4.00; P = .003) and VΔZL Line (median difference 5.40; 1.71‐9.09; P = .01). Observation of EIT alterations were reflected by clinical signs in horses with decompensated LCVO and after administration of furosemide. Conclusions and Clinical Importance EIT measurements of ventilation distribution showed less ventilation in the left lung of horses with LCVO and might be useful as an objective assessment of the ventilation effects of cardiogenic pulmonary disease in horses.

Conclusions and Clinical Importance: EIT measurements of ventilation distribution showed less ventilation in the left lung of horses with LCVO and might be useful as an objective assessment of the ventilation effects of cardiogenic pulmonary disease in horses.

K E Y W O R D S
center of ventilation, congestive heart failure, EIT, furosemide, pulmonary edema, pulmonary hypertension

| INTRODUCTION
Left-sided cardiac volume overload (LCVO) causes an increase in pulmonary venous pressure and can lead to fluid accumulation in the interstitial lung tissue and alveolar space in horses. 1 Chronic high intravascular pressure might cause pulmonary vascular remodeling and lead to perivascular inflammation resulting in tissue fibrosis and changes in tissue elasticity as shown in humans. 2 As such, LCVO can impair the respiratory physiology in horses and, when severe, lead to alveolar pulmonary edema (PE) with risk of collapse or sudden death. 3 Stall-side diagnosis of pulmonary disease is often difficult in horses and primarily consists of detection of cough and tachypnoea and the auscultation of loud bronchovesicular sounds or crackles over the thorax. 1,4 Thoracic radiography and ultrasound often lack sensitivity in horses especially for detection of subclinical and deeper parenchymal pulmonary changes, whereas techniques such as computed tomography lack practicality. 5 Electrical impedance tomography (EIT) is a noninvasive radiationfree functional imaging modality for continuous monitoring of pulmonary ventilation. 6 An electrode belt is placed around the thorax of the horse generating alternating currents from sequential pairs of electrodes that then circulate around the thorax. The intrathoracic tissue allows the passage of current with little impedance; however, the inhalation of gas increases intrathoracic resistance. Resulting impedance change is detected by the remaining electrodes, which form a relative image with respect to the reference baseline of the lungs at the start of inspiration. 7,8 This modality has been used to evaluate distribution of ventilation, changes in gas flow in the airways and tidal volume in awake and anesthetized horses. [9][10][11][12][13] Furthermore, noncardiogenic and cardiogenic PE has been assessed with EIT in people, pigs and dogs. [14][15][16][17][18][19] Changes in EIT variables in response to reduction of extravascular lung water in people demonstrated its potential utility in the clinical management of alveolar PE and pleural effusion in this setting. 20,21 The aim of this prospective cohort study was to report EIT variables describing ventilation in standing unsedated horses with LCVO and to compare them to a healthy cohort. For the purpose of this study, LCVO was considered present when there was eccentric hypertrophy of either left atrium or ventricle, which is expected to cause secondary pulmonary venous hypertension. We hypothesized that EIT derived data reflecting ventilation will be different between cohorts. Furthermore, we hypothesized that EIT measurements in decompensated LCVO with clinical signs of alveolar PE will be different to those in compensated LCVO with assumed interstitial accumulation of extravascular lung water and tissue fibrosis.

| Healthy cohort
Eight horses from the teaching herd were included in this cohort. Based on long term history, thorough clinical examination and auscultatory findings all horses were deemed free of overt cardiopulmonary disease.
Intermittent informal echocardiography and thoracic ultrasonography, performed as part of standard teaching activities, revealed no evidence of subclinical disease. All body condition scores ranged from 4-6/9.

| Cohort with LCVO
Six horses with LCVO of various severities were included in this cohort.

| Horses with compensated LCVO
Four horses from the same teaching herd with previously diagnosed compensated heart disease causing LCVO as reflected by left atrial or ventricular enlargement, but without clinical signs at rest were recruited to this cohort. Horses had concurrent aortic and tricuspid valve regurgitation (n = 1), aortic, mitral and tricuspid valve regurgitation (n = 2) and restrictive membranous ventricular septal defect (n = 1).
Diagnosis was performed using clinical examination, auscultatory findings and echocardiography. Echocardiographic anatomic and functional measurements confirmed enlargement of 1 or both left cardiac chambers in all 4 horses (Table 1 and Supporting Information

| EIT measurements
After clinical and ultrasound data were recorded, the hair over the thorax directly caudal to the scapula (5th-6th ICS) was moistened with water and ultrasound gel was applied circumferentially in this Briefly, the finite element model was created based on a transverse postmortem section of the thorax obtained from a frozen equine specimen at the level of the 6th ICS. 30  and CoV RL 0% refers to ventilation occurring in the most ventral part of the lung and the right lung respectively, whereas 100% refers to that in the most dorsal part and the left lung, respectively ( Figure 1A). 31 2. Ventilation-induced impedance changes within the lung field below 10% of the maximum value were determined and these regions defined as "silent space." They were divided into dependent (DSS) or nondependent silent space (NSS) and expressed as a percentage of the entire lung region ( Figure 1A). 32 3. The global impedance change of all pixels between the beginning and end of inspiration within the lung field was calculated for each breath and used as a surrogate for tidal volume, expressed by EIT (VΔZ; Figure 1A). 13 4. To describe the ventro-dorsal distribution of ventilation during inspiration regionally on a more detailed pixular level, the entire  c. Graphical illustration of impedance change over 32 matrix lines.
(2) Ventilated area ( Figure 1C): The percentage of pixels of the total image representing the ventilated area of the right (VAR) and left  (VAL) lung field identified by detectable impedance change on an individual pixular level.
(3) Global inhomogeneity index (VGI): The differences in impedance variation between each pixel and the median value of all pixels are calculated and normalized to the sum of impedance values using a previously described formula. 6,33,34 The overall heterogeneity of tidal volume distribution was described as an average over 4 breaths, whereby a smaller VGI represents a more homogeneous distribution, and a larger VGI indicates a more inhomogeneous ventilation associated with altered lung function.
F I G U R E 2 Scatter plot graphs of nonconventional electrical impedance tomography (EIT) variables evaluated from healthy horses (n = 8, black dots) and horses with compensated (blue dots) and decompensated (red dots) left-sided cardiac volume overload (LCVO; n = 6) describing the ventilated area over the right (VAR) and left (VAL) lung field, average maximum impedance change (right: avg-max VΔZR Line ; left: avg-max VΔZL Line ) and amount of ventilated matrix lines (VΔZR Line ; VΔZL Line )

| Statistics
Normally distributed data are summarized as mean ± SD. Results are summarized as median (range) for nonnormal data. As no research was available on which to base the standard deviations of any EIT outcome variable in horses with cardiac disease, no a priori power analysis was performed for this study.  À2.18 to À.08; P = .04) was significantly larger compared to healthy horses (Figures 2 and 3). There were no significant differences between cohorts for other nonconventional EIT variables. Graphical linear-plane distribution of ventilation is shown in Figure 3.

| Observations on compensated and decompensated LCVO horses, and the effect of furosemide administration in decompensated LCVO horses
Horses with compensated LCVO did not show any abnormalities during the clinical examination. Tachycardia (defined as heart rate > 48 bpm) and clinical signs of respiratory disease were found in H1 PE and H2 PE (Supporting Information Table Supp 2). Both horses with F I G U R E 4 Scatter plot graphs of peak inspiratory (PIF EIT ) and expiratory flow (PEF EIT ) normalized by total impedance change (VΔZ) and VΔZ as surrogate for tidal volume measured by electrical impedance tomography evaluated from healthy horses (n = 8, black dots) and horses with compensated (blue dots) and decompensated (red dots) left-sided cardiac volume overload (LCVO; n = 6) decompensated LCVO had pulmonary roots larger than aortic roots, supportive of pulmonary hypertension (Table 1). Further echocardiographic markers of pulmonary hypertension (such as tricuspid regurgitation velocity) were not performed.
Based on observation of the means, VΔZ, PIF EIT , VAL, VΔZL Line were lower, and CoV VD , DSS, and PEF EIT /VΔZ were estimated as higher in the horses with decompensated LCVO than in compensated LCVO (Figures 2 and 4). Furthermore, the ventilation was observed to be estimated lower in the ventral and central-ventral lung regions and higher in the most dorsal lung regions in the horses with clinical signs of alveolar PE than in the horses with compensated LCVO.
Further observation of variables for the H1 PE and H2 PE horses showed that clinical measurements and echocardiography findings improved in response to administration of furosemide (Tables 1 and   2). An improvement toward the values observed in healthy and compensated LCVO horses was seen for VΔZ (18%), DSS (21%), VAL (38%), VΔZL Line (22%), and avg-maxVΔZL Line (62%) after furose- mide. An increased in VΔZ CV and VΔZ CD , but decrease in VΔZ D was observed after administration of furosemide in both horses.

| DISCUSSION
This study has shown that EIT measurements can be used to detect Clinical signs of primarily interstitial accumulation of extravascular lung water and tissue fibrosis due to high pulmonary pressure in early or mild heart failure might be limited to increased respiratory rate and crackles or moist bronchovesicular sounds. 36 In this study, horses with compensated LCVO did not demonstrate adventitious lung sounds on thoracic auscultation. Only once advanced will alveolar PE cause dyspnea, coughing and profuse frothy nasal discharge, such as seen in H2 PE . Beside clinical signs, thoracic ultrasound can be used to visualize peripheral pleural and parenchymal lesions. Several abnormalities were observed in our horses with LCVO with at least 1 abnormality in every horse confirming the presence of lung changes in this cohort (Table 1). However, with regards to extravascular lung water accumulation, changes are most pronounced at the central hilar area, and could be missed by diagnostic ultrasonography as well as radiology in large animals, especially when interstitial or a low quantity of fluid accumulation in the alveolar space is present. 36 Electrical impedance F I G U R E 5 The "spring model" illustrating the slinky effect of gravity on the distribution of ventilation. The differences in the coil stretch of the spring and the EIT image between a healthy horse (left) and a horse with left-sided cardiac volume overload (LCVO; right) are displayed. The vertically oriented spring is stretched dorsally and compressed at the bottom. The corresponding EIT image demonstrates the areas of detected impedance change in light blue. An increase in extravascular lung water (right) leads to a greater number of coils in the dependent portion of the spring, which is analogous to a greater density of the lung tissue and therefore less ventilation detected in the EIT image tomography has been shown to be able to detect cardiogenic accumulation of extravascular lung water in humans, pigs and dogs, mainly by revealing shifts in the distribution of ventilation. [14][15][16][17][18][19] To inspect the differences in the distribution of ventilation we initially created graphs showing the impedance change for each of the 32 horizontal matrix lines of the EIT functional image for healthy and LCVO horses plus the decompensated horses before and after administration of furosemide ( Figure 3). Additionally, the EIT images generated by the analyzing software were visually inspected. Figures 1 and   5 demonstrate comparisons of the generated EIT images from healthy and horses with LCVO. These illustrations confirmed the hypothesized difference in the distribution of ventilation between cohorts, but also indicated that conventional EIT variables will unlikely be able to verify the alterations in horses with LCVO. Based on these initial observations, nonconventional EIT variables describing the ventilated area and line distribution were constructed to demonstrate significant numerical differences between cohorts, which were present for some of the novel EIT variables.
The observation that the left lung was more affected by LCVO was unexpected. In people, 1-sided PE of the right lung has been reported as a complication of acute mitral valvular insufficiency. 37 This is explained by the vector of the retrograde blood flow across the mitral valve that is directed into 1 of the right superior pulmonary veins in humans. In horses, the anatomy of the pulmonary veins differs 38

| LIMITATIONS
The study is limited to a small number of horses and variability in stages of cardiac disease. Despite our small sample size, we were able to detect significant differences in novel EIT variables. Thus, the results of our study provide motivation to pursue further investigations of EIT to describe pulmonary changes as a sequela of cardiac disease and could direct future, larger scale studies.
In this study, vital variables, thoracic auscultation, and ultrasound were used as inclusion criteria in the LCVO cohort. No radiography was performed despite reports of relevant findings in horses with PE such as prominent pulmonary vascular markings, increased interstitial lung density, and peri-bronchial fluid collection. 1 While this might have given additional information for grading of the extent of pulmonary changes, there was no intent to associate magnitude of changes in EIT variables with extent of disease.

ACKNOWLEDGMENT
No funding was received for this study. The authors thank Dr Guy D Lester, Dr Rosemary S Cuming and Dr Christina Eberhardt for their contribution to ultrasonographic data acquisition.

CONFLICT OF INTEREST DECLARATION
Authors declare no conflict of interest.

OFF-LABEL ANTIMICROBIAL DECLARATION
Authors declare no off-label use of antimicrobials.