Radiographic lung congestion scores in dogs with acute congestive heart failure caused by myxomatous mitral valve disease

Abstract

Background

In humans, lung congestion scores are predictive of recurrence of acute congestive heart failure (CHF) and are superior to cardiac biomarkers in predicting survival.

Objectives

The primary aim of this retrospective study was to determine if a modified lung congestion score (LCS) in dogs diagnosed with acute CHF because of myxomatous mitral valve disease was associated with time until recurrence or death.

Animals

Complete medical records were available for a total of 94 dogs between 2010 and 2019, but only 35 dogs fulfilled the criteria for inclusion.

Methods

This retrospective study used descriptive statistics to describe the cumulative and corrected LCS. Correlations were used to examine the association of the corrected LCS and time until recurrence or death, selected echocardiographic variables, and timing of furosemide administration.

Results

The mean LCS was 8.4 (SD 3.3) and corrected LCS was 0.48 (SD 0.19). The pattern was predominantly symmetric (40% of dogs) and focal (caudal) but more commonly right‐sided when asymmetric (40% vs 20%). The median number of days after initial diagnosis of acute CHF to readmission and death was 150 days (range 4‐572), and 266 days (range 5‐965), respectively. No significant association between the dog's corrected LCS and number of days until readmission (r = .173, P = .42) nor survival (r = .109, P = .56) was found. There was a negative significant correlation (r = −.71, P < .001) between the time interval of furosemide administration and corrected LCS.

Abbreviations

CHF

congestive heart failure

CPE

cardiogenic pulmonary edema

DV

dorsoventral

E wave

early diastolic left ventricular inflow

LA/Ao

left atrial to aortic ratio

LCS

lung congestion score

LVIDDn

normalized left ventricular internal diameter in diastole

MMVD

myxomatous mitral valve disease

VD

ventrodorsal

1. INTRODUCTION

Cardiogenic pulmonary edema (CPE) in the dog is characterized by acute respiratory distress with radiographic evidence of lung infiltrates that resolve with the administration of a diuretic. ¹ , ² Consideration of etiologies that are associated with transient interstitial to alveolar lung pattern in addition to CPE, include but are not limited to, non‐CPE, pulmonary hypertension, pneumonia, and lung contusion. ² , ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ Distinguishing CPE from other causes of abnormal lung patterns on thoracic radiographs is challenging. CPE often has accompanying lobar pulmonary vein dilation reflecting venous hypertension but interpretation could be confounded by recent administration of loop diuretics. The so‐called perihilar anatomical distribution, located caudodorsally on radiographs, is often considered typical for left‐sided congestive heart failure (CHF). Interpretation can be limited by the inability to completely visualize the hilar region as it is obscured by the canine cardiac silhouette on the ventrodorsal and dorsoventral views, and the pulmonary veins and left atrium on lateral views, with a misleading impression as an abnormal lung pattern. ² Lung atelectasia and obesity can further complicate interpretation. The diagnosis of acute CHF in a dog with myxomatous mitral valve disease (MMVD) is a clinical one, made considering the constellation of a typical signalment, a presenting complaint of tachypnea, the presence of a moderate to high‐intensity murmur consistent with severe mitral insufficiency, left atrial enlargement on cardiac imaging, lung infiltrates, and the rapid resolution of clinical signs (and lung infiltrates) in response to furosemide administration. ¹⁸ Attempts at tailored treatment using algorithms based on imaging or biomarker quantification following diagnosis of acute CHF are limited in the veterinary literature. Broad recommendations on dosing strategies of furosemide, based on the clinicians interpretation of respiratory signs, are given in the recent consensus statement on the management of MMVD. ¹⁸ In stable CHF, dose escalation of furosemide or pimobendan, to target a biomarker concentration is an effective management strategy. ¹⁹

Radiological scoring systems of lung infiltrates have been proposed and validated in humans with acute heart failure. ²⁰ , ²¹ , ²² Severity of alveolar edema in dogs with CHF secondary to progressive MMVD have been scored in a previous study, assigned a category of 1 to 4, with 1 having no pattern, 2 and 3 having mild to moderate interstitial edema respectively, and 4 having an alveolar pattern with lung consolidation. ²³ Lung congestion scores (LCS) have been described in dogs, varying from a 4‐quadrant approach, a modified Murray Lung Injury Score, to a clinically relevant approach that divides the canine lung lobes depicted by radiographs to mimic the lung ultrasound regions examined by the Vet BLUE technique. ³ , ⁴ , ⁵ , ²⁴ The approach is that each area examined is assigned a score on a scale from 0 to 3, with 25% increasing increments of quadrant fill.

We propose an anatomical approach to LCS, where each lung lobe is given the opportunity to contribute to the final score. In addition, the density of the lung infiltrates in each lobe can be individually scored out of 3, accommodating various degrees of pulmonary edema. The hope is that the distribution and intensity of the lung infiltrates would represent both the severity and distribution of pulmonary edema. This should intuitively reflect the severity of hemodynamic compromise and volume overload and the regurgitant fraction and might have bearing on overall prognosis. If this could be demonstrated, initial radiographs at the time of confirming end‐stage heart disease, that is, CHF (stage C) in dogs with MMVD, would be helpful in providing a prognosis at the onset of heart failure and guide recommendations on intensity of therapeutic dosing.

The primary aims of this study were to (a) describe the distribution of the alveolar‐interstitial pattern according to our lung congestion score (LCS), and (b) determine if there was a significant association between the LCS at initial presentation for acute CHF and time until readmission or death. The secondary outcomes included the examination of the association of 2 additional groups of variables: (1) time interval of thoracic radiographs relative to furosemide administration, and (2) selected echocardiographic variables with established prognostic value, and the severity of the lung congestion score. Lastly, differences between LCS in dogs receiving furosemide before or after thoracic radiographs were compared.

2. ANIMALS, MATERIALS, AND METHODS

This was a retrospective study conducted at the University of Tennessee College of Veterinary Medicine (UTCVM) spanning a timeframe from 2010 to 2019. The database was searched for dogs hospitalized at UTCVM with a diagnosis code of CHF and diagnostic investigation code of thoracic radiographs and echocardiography conducted during their hospitalization period. Inclusion criteria for the study included dogs under 22 kg, with a diagnosis of MMVD, confirmed with echocardiography, resulting in acute CHF. The echocardiographic criteria of MMVD were mitral leaflets that were thickened or prolapsing or both, and mitral regurgitation detected on color‐flow Doppler. ²⁵ Thoracic radiographs and echocardiography had to be completed within 48 hours of each other during hospitalization for CHF. A minimum of 1‐year follow‐up from the date of initial diagnosis had to be available for each dog. This follow‐up period was guided by the median survival of dogs in stage C MMVD receiving pimobendan. ²⁶ Exclusion criteria included congenital heart disease, diffuse pulmonary comorbidities that could not be differentiated from CPE on radiographs, a history of a permanent pacemaker being implanted, primarily disease of the tricuspid valve, pulmonary arterial hypertension, or a previous history of clinical heart failure before admission. Dogs were not eliminated if they had received furosemide before referral. The medical record was reviewed by a board‐certified cardiologist to determine if the final diagnosis was consistent with the diagnosis of acute CHF. This final diagnosis was determined by reviewing the dog's medical history, echocardiographic report, and radiographic findings. A dog was considered to be in acute stage C CHF because of MMVD if the dog presented with dyspnea/tachypnea, had cardiomegaly as defined by the EPIC study, and radiographic lung pattern consistent with lung infiltrates from cardiogenic edema which improved either clinically or radiologically with the administration of furosemide. ²⁷

Data collected for each dog included their signalment, weight (in kg), time and date of admission, diuretic administration, and of initial and recheck thoracic radiographs respectively, if available, if the referring veterinarian (RDVM) gave the dog a diuretic before referral to UTCVM, date of readmission for CHF if applicable, and date and cause of death if available. The number of days until readmission for CHF and survival in days from date of initial diagnosis was then calculated for each applicable dog. This data was collected using the university's medical record system and via phone calls to a dog's primary care hospital to obtain information about outcome if UTCVM's records were incomplete.

The available lateral and dorsoventral (DV)/ventrodorsal (VD) thoracic radiographs for each dog were examined and scored based on a modified lung congestion score developed for this study. Six total lung lobes were scored: the right cranial, right middle, right caudal, accessory, left cranial, and left caudal. Each lung lobe was identified based on published location criteria (Figure 1). ²⁸ Each lung lobe was assigned a score from 0 to 3. A lobe that was clear of edema was given a score of 0. A score of 1 was given for a lobe that had enlarged pulmonary veins relative to the pulmonary arteries and/or an unstructured interstitial pattern. A score of 2 was given for a lobe that had an interstitial to alveolar pattern. A score of 3 was given to a lobe that had exclusively an alveolar pattern. An unstructured interstitial pattern was defined as an increase in pulmonary opacity with maintained but decreased visualization of pulmonary vascular margins. To qualify as a true unstructured interstitial pattern, the degree of opacification had to exceed expected variations of normal, that is, to avoid overinterpretation of an interstitial pattern, the phase of respiration, age, technical factors, and body habitus were considered when evaluating the radiographs. An alveolar pattern was defined as an increase in pulmonary opacity to the point of loss of visualization of pulmonary vascular margins because of the silhouetting effect, with or without evidence of air bronchograms or a lobar margin sign. Scoring of the radiographs was initially performed independently by a second‐year veterinary student who had undergone training in lung pattern identification and grading. A board‐certified veterinary radiologist subsequently reviewed the scores and adjusted them if needed. Scores for each of the 6 lobes were then added, resulting in a composite range of scores from 0 to 18. If a lung lobe could not be visualized because of positioning, superimposition of the enlarged heart, or lack of a particular radiographic view, this lobe was not scored and the total possible score was adjusted accordingly. The overall score for each set of lungs was expressed as a fraction rather than an integer to reflect the variations in total possible score. This number was referred to as the dog's corrected score. The recheck radiographs, when available, were scored in the same fashion. An example of how scores were derived for a specific dog is shown in Figure 2.

FIGURE 1.

Schematic illustrating the location of the 6 lung lobes used to calculate the modified lung congestion scores based on right lateral (A), left lateral (B), and ventrodorsal/dorsoventral (C) thoracic radiographs. A potential score for each lobe was allocated and summed for a final score or corrected score depending on if all lobes were amenable to scoring. ACC, accessory lung lobe (shaded); LCd, left caudal lung lobe; LCr, left cranial lung lobe; RCd, right caudal lung lobe; RCr, right cranial lung lobe; RM, right middle lung lobe.

FIGURE 2.

Three‐view thoracic radiographs in a 13‐year‐old MC Shih Tzu presented with congestive heart failure secondary to mitral valve disease. The radiographs at presentation (A) were assigned an overall score of 16 (individual scores of “3” for the right cranial, middle, accessory, and the left cranial lung lobes, and “2” for right and left caudal lung lobes). Recheck radiographs 1 day later (B) were assigned an overall score of 7 (individual scores of “2” for the right caudal lung lobe and “1” for all other lung lobes).

The echocardiography data for each dog was collected from the medical records. All transthoracic echocardiograms were performed by a board‐certified cardiologist or a resident under the supervision of a board‐certified cardiologist. Standard views were obtained as described for the dog. ²⁹ The right parasternal long axis (4 and 5 chambers), and short axis views at the level of the left ventricle (LV), which included the papillary muscles, the mitral valve, and the heart base to include the right ventricular outflow tract, were obtained from the right side of the thorax. Apical (4 and 5 chambers) and left cranial views of the heart base were obtained from the left thorax. Left ventricular outflow tract (LVOT) measurements were obtained from the subxiphoid view. Each measurement was an average of 3 consecutive cardiac cycles. Dimensions of the left ventricle in systole and diastole (LVIDd, LVIDs), the septal (IVSd) and free wall (LVFWd) thickness in diastole, were measured on M‐mode, were measured by the leading‐edge‐to‐leading edge technique as described previously. ³⁰ Electrocardiography assisted in identifying the timing for LV end‐diastolic (the peak of the QRS) measurements, and LV end‐systolic measurements were performed by measuring the distance between interventricular septum and the free wall when structures were at their smallest distance. Fractional shortening (FS) was then calculated and expressed as a percentage. The 2‐D left atrium and aorta were measured on the right parasternal short‐axis view using a common technique described for dogs. ³¹ These values were used to create the left atrial to aortic ratio (LA:Ao). Peak flow velocities (m/s) were measured of the transmitral inflow peak diastolic E waves, using Pulsed wave Doppler (m/s). ³² The left atrial to aortic root ratio (LA/Ao) from 2D echocardiography was considered abnormal if the value was >1.6. Normalized left ventricular internal diastolic diameter (LVIDDn) was considered abnormal if the value is >1.7. ²⁷

Statistical analysis was conducted using SPSS 28. A Friedman's test was used to determine if the LCS differed between lung lobes, followed by a pairwise comparison with a Bonferroni adjustment to compare the 6 lobes. Correlations were used to examine the association of the corrected LCS and time until recurrence or death, selected echocardiographic variables, and timing of furosemide administration. Pearson correlations were used for data that met the assumption of normal distribution otherwise, Spearman correlations were used. Independent samples t‐test was used to test if corrected LCS differed between dogs that received furosemide before or after thoracic radiographs.

3. RESULTS

A total of 177 dogs met the initial search criteria, that is, presentation to the emergency service for cardiac‐related respiratory distress that could be caused by acute CHF. A flow chart depicts the process of eliminating cases and the final number of dogs included in the study (Figure 3). The mean age and body mass of the 35 dogs included in this study was 10.8 years (SD 2.0) and 6.5 kg (SD 4.3) respectively. There were 14 males (13 neutered) and 21 female (20 spayed) dogs. Sixteen breeds were represented including mixed breed dogs (n = 5). The most frequently represented breed was the Chihuahua (n = 7), followed by Shih tzu (n = 4), Miniature schnauzer (n = 4), the toy poodle (n = 2), Papillion (n = 2), and Cavalier King Charles Spaniel (n = 2). The following breeds were represented by 1 dog each: Bichon Frise, Boston terrier, Jack Russel terrier, miniature dachshund, Pekingese, Shetland sheepdog, Whippet, and Yorkshire terrier.

FIGURE. 3.

Flow chart of exclusion of dogs from the study. The medical records were searched by the University of Tennessee College of Veterinary Medicine (UTCVM) spanning a timeframe from 2010 to 2019 for dogs with the diagnosis of congestive heart failure (CHF; n = 177). The diagnosis of “mitral valve disease” (MVD) or “chronic degeneration valve disease” was used to select dogs after dogs that never entered CHF (n = 28) or had represented (n = 27) for a subsequent CHF episode that is, already medically managed for stage C MMVD, were excluded. If clinical signs or severity of disease were predominantly related to tricuspid valvular insufficiency (n = 3), a permanent pacemaker implanted (PPI; n = 2), had congenital heart disease (n = 5), then they were excluded. Dogs with MMVD, and CHF but exceeded a cut‐off weight of 22 kg were also excluded (n = 12). Incomplete records because of death at arrival (n = 10), no echocardiogram (n = 20), and no thoracic radiographs (n = 6) were then excluded. The remaining number of dogs with a diagnosis of acute CHF because of MMVD available for lung congestion scoring was 56 dogs, but 21 dogs had no available follow‐up.

Of the 35 cases that qualified for the study, 24 were readmitted for CHF. Days to readmission for a second event of acute CHF were not normally distributed. The median number of days to readmission was 150 days (range 4‐572, interquartile range [IQR] 302 days). The date of death was known for 31 of dogs. The median survival after initial diagnosis of acute CHF was 266 days (range 5‐965, IQR 288 days).

The time until initial radiographs were performed after admission ranged from 11 minutes to 21.5 hours with a median of 2.1 hours, which excludes 6 dogs who had radiographs performed by their RDVM before they were referred to the ECC service.

The median and range of the lobar LCS and patterns of edema for the lung lobes are depicted in Table S1. The most common patterns of lung edema regarding the caudal lung lobes were symmetric (n = 14, 40%) and asymmetric right‐sided (n = 14, 40%), with left‐sided caudal being less common (n = 7, 20%).

The LCS for each lobe, the cumulative and corrected LCS for each dog is reported in Table S1 as is the median and range of the LCS per lobe. The lung congestion scores (LCSs) were normally distributed with the mean cumulative LCS of 8.4 (SD 3.3) and corrected LCS of 0.48 (SD 0.16). The individual lobe LCSs were not normally distributed. Lung lobes significantly differed [χ²(5) = 43.55, P < .001, W = 0.290]. The right caudal LCS was significantly higher compared to the middle (P < .001), accessory (P = .001), and left cranial (P = .002) and the left caudal lobe was significantly higher than right middle (P = .004). No other differences were found.

On Spearman's rho analysis, there was no significant association between the dog's corrected LCS and number of days until readmission (rho = .173, P = .42). Similarly, no association between the dog's corrected LCS and survival from initial diagnosis was found (rho = .109, P = .56).

Corrected score, LA/Ao, E (m/s), and LVIDDn were all normally distributed. There was no association found between the echocardiographic measurements of left atrial (LA:Ao; r = .246, P = .19) and left ventricle (LVIDDn) size (r = .096, P = .61), as well as early diastolic transmitral flow velocity (E, m/s; r = .124, P = .51), with corrected lung congestion scores. There was no significant relationship between a dog's number of days to readmission and the selected echocardiographic variables: LA/Ao (rho = −.131, P = .55), E velocity (rho = .226, P = .31), or LVIDDn (rho = −.028, P = .90). There was no significant relationship between a dog's survival in days and the selected echocardiographic variables: LA/Ao (rho = −.133, P = .52), E velocity (rho = .069, P = .73), or LVIDDn (rho = −.079, P = .70).

Furosemide administration data was available for 28 of the dogs of which 21 were eligible for statistical analysis. The time interval of 7 dogs was excluded for the following reasons: 1 dog was managed for suspected stable CHF with furosemide at the time of admission, 1 dog the timing of furosemide administration to thoracic radiographs, both of which were performed by the referring veterinarian was estimated times and not thought to be accurate and 5 dogs had an interval of furosemide until thoracic radiographs that exceeded 5 half‐lives of the drug or 6 hours, the reported interval that furosemide is effective. Of those 21 dogs, 13 received furosemide in the hospital before the radiographs were taken (median 96 minutes, range 5‐260 minutes) and 8 received furosemide after radiographs (median 30.5 minutes, 13‐147 minutes). The results of the timing of the furosemide administration before or after radiographs were performed in each dog are detailed in Table S1. The average dog‐corrected LCS if the dog received furosemide in the hospital before radiography was 0.55 (SD = 0.23), while the average corrected LCS if they had not received furosemide was 0.43 (SD = 0.12). No significant difference was found in corrected LCS (t = 1.34, df = 19, P = .20, d = 0.20) between dogs receiving furosemide before radiographs compared to those that did not. A significant, strong, and negative relationship was found between corrected LCS and time since furosemide administration and depicted as a scatterplot in Figure 4 (r = −.77, P < .001).

FIGURE 4.

Scatter plot for corrected lung congestion score (LCS) and Interval between radiographs and furosemide administration (minutes) for 21 dogs. A significant negative relationship was found between corrected LCS and time since furosemide administration (r = −.711, P = .001).

4. DISCUSSION

Our study did not find an association between the severity of cardiogenic edema estimated by a modified LCS and outcome, specifically time until readmission or survival. Antemortem quantification of lung edema is challenging but radiographic perception of lung congestion is considered a surrogate estimate of fluid volume. In the case of acute heart failure secondary to myocardial infarction, radiological scores correlate well with both physical examination findings and lung fluid content. ²⁰ In humans with lung congestion because of heart failure, extravascular lung water, assessed by both ultrasound and radiographs, are prognostic. ²¹ , ³³ , ³⁴ In worsening heart failure in humans, residual radiographic lung congestion assessed by congestion score index, is the best predictor of postdischarge outcome, being superior to biomarkers and other clinical markers of congestion. ²¹ Further, lung congestion assessed radiographically in dogs with chronic heart failure, correlated with restrictive ventilation, increased pulmonary vascular resistance, NTproBNP, and reduced survival. ²² Clinical scoring of ventilation effort was not performed in our study as the availability of data regarding physical examination findings was hampered by the retrospective nature of this study. Similarly, cardiac biomarkers are not routinely performed when echocardiography is available. For these reasons, comparisons between the utility of different diagnostic modalities, namely, imaging, clinical examination, and biomarkers could not be examined. Echocardiographic indices were examined with no association between LCS and the respective echocardiographic variables.

The results of our study not only disagree with findings in the human literature but also with veterinary studies that have examined lung edema indices to predict outcomes in dogs with CHF because of MMVD. A previous study demonstrated a significant difference in admission thoracic radiograph edema score in dogs with CHF between those that had CHF recurrence or short‐term cardiac death and those dogs that did not have short‐term recurrence or cardiac death. ²⁴ This previous study included 25 dogs, all imaged within 0.5 to 6 hours of admission in contrast to our study, which ranged from 11 minutes to 21.5 hours. This same study found the diagnostic utility and predictive value of negative outcomes of serial lung ultrasound in the same dogs diagnosed with CHF, clinically useful. ²⁴ A 4‐quadrant LCS using lung ultrasound, known as the edema score, assigned at admission, was associated with a negative outcome at 90 days, and edema score assigned at the postdischarge appointment, was a significant predictor of death. ²⁴ Radiographs, the classic modality of diagnosing heart failure in humans, are considered, simple, and fast, with good specificity, and moderate sensitivity. ¹⁸ , ³⁵ , ³⁶ The radiological diagnosis of CPE, has long been considered the diagnostic imaging technique of choice in dogs but is fraught with technical implications of maneuvering a critically ill dog, frequent suboptimal technique, radiation exposure for staff, and criticized for its sensitivity and specificity. Lung ultrasound is becoming increasingly recognized as a useful tool in the categorization of alveolar‐interstitial lung patterns in dogs, but in 1 study agreement with radiographs is only fair, although considered more sensitive. ³ False negatives do occur, and in this same study, the specific shortcoming was the failure to detect CPE using lung ultrasound in dogs with mild focal interstitial lung patterns. The modality also lacks specificity, as a number of dogs were misclassified as having CHF where the diagnosis was non‐CPE. Overall, the accuracy of lung ultrasound and radiographs for predicting CPE in dogs is considered similar, and both have limitations in their sensitivity and specificity. ⁴ The results of our study measuring a modified LCS radiographically in dogs with CHF do not support a prognostic benefit in predicting time until readmission nor survival, and does not provide support for the justification of thoracic radiographs in emergency triage for prognosis. Thoracic radiographs and lung ultrasound modalities were not directly compared in this study and specific recommendations cannot be made based solely on the lack of predictive value of radiographs. Reasons for the lack of significant findings in our study that did not find an association between corrected LCS and outcome could include the heterogeneity of the time from admission or receiving furosemide, the time until radiographs were performed, and variation in breed thoracic conformation influencing ability to interpret the modified LCS.

Lung patterns in dogs with acute CHF are reported as interstitial (67%) and mixed interstitial‐alveolar (33%). ¹ The difference in the distribution is proposed to be related to the severity and temporal progression, either an early‐stage transudate accumulation in the perivascular space or a more severe or advanced alveolar leak respectively. Our study found a significant difference in LCS between lobes (P < .001), with the most common distribution was focal symmetric and when asymmetric, it was right‐sided. The lung lobe with the highest mean corrected score was the right caudal lobe. This is similar to another study of dogs with both MMVD and DCM, where symmetric, focal distribution (caudodorsal) was found to be the most common pattern (65.6%), and approximately one third of dogs with CHF from any cause, had asymmetric focal distribution and consistently on the right caudal lobe. ¹ This study, found that an eccentric mitral valve regurgitant jet, usually from MMVD, was 25.7 times more likely to have an asymmetric pattern as compared to a dog with a centrally directed jet and proposed that this is the likely reason for the asymmetric pattern. Detection of distribution of an alveolar‐interstitial syndrome, differs between radiographs and lung ultrasound in predicting left‐sided CHF in that cranial in addition to caudal quadrants are more likely to score positive for lung ultrasound and caudal quadrants are more likely to score positive for radiographs with lung ultrasound pattern being diffuse and radiographs being more focal. ³ The limitation of this cited study was that it used a 4 quadrant approach, with a quadrant often having overlapping lung lobes, for example, the right middle would be classified as either in the cranial or caudal quadrant. In addition, caudal quadrants, because of the proximity of the diaphragm, are challenging to image with ultrasound. Our study, which allocates a score to each individual lung lobe, found that the caudal lung lobes had the highest lung scores. Further, when the pattern was asymmetric, the right caudal lung lobe was the most commonly affected lung lobe. These results are aligned with those of a previous study in dogs with MMVD. ¹

The effect of intravenous furosemide administration before referral and thoracic radiographs on lung congestion score was investigated and was found to have the impact intuitively predicted. The strong negative correlation between the interval between furosemide treatment to imaging and LCS, is explained in part by the study sample that received furosemide before radiographs with LCS benefiting (lower) with time. The negative association of time interval and LCS in the cohort that had a delay in treatment after radiographs were performed, is likely related to the low LCS with presumably low lung water and less severe clinical signs prompting additional imaging such as echocardiography to support the diagnosis before administration of furosemide.

Our study did not find a correlation between the LCS and echocardiographic measures of left‐sided cardiomegaly or increased filling pressure. One‐ and 2‐dimensional echocardiographic determined left atrial and ventricular size, as well as quantification of mitral valve regurgitation, have been established as prognosticators in dogs with MMVD. ³⁷ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ It is conceivable that the LCS might not be able to accurately estimate lung fluid content in dogs and severity of lung edema as is the case in humans with heart failure. ²⁰ The utility of thoracic radiographs, using gold standard techniques, including catheterization for left ventricular filling pressure, measures of lung compliance loading have not been conducted in dogs and the association is only an assumption.

The established association between transmitral E velocity and a cardiac event and survival, was not found in our study. ³⁷ It is likely that our study lacks a significant correlation due partially to a lack of statistical power. A power analysis using an alpha of .05 and a beta of .80, determined that approximately 500 dogs would be needed in our study to detect a significant correlation of .125 between E velocity and recurrence of CHF.

There are several additional limitations in this study. It suffers from the selection bias inherent in a retrospective study, particularly related to the lack of standardized care and variation in owners' willingness to treat their pets. There was variation in the timing of radiographs performed relative to admission and receiving furosemide. Some dogs that had received furosemide in the past because of a history of a murmur and a cough but had not reached stage C MMVD, were included. These were included as the presence of chronic furosemide treatment was unlikely to lessen the acute decompensation and cardiogenic edema. The role of furosemide in delaying or lessening the acute onset of CHF is unknown but dogs in stage C will typically remain in this stage until the onset of advanced heart failure, that is, the time when recurrent heart failure develops and >4 mg/kg/day of furosemide, or stage D, that is, >6 mg/kg/day is required to manage clinical signs. The furosemide “breaking phenomenon” is a maladaptive process in chronic CHF where natriuresis does not exceed sodium intake and occurs at low extracellular volume. Despite chronic furosemide administration, dogs with CHF will eventually develop worsening congestion unless the deficient natriuresis is therapeutically addressed, irrespective of the progression of the heart disease. Dogs that had received chronic furosemide treatment in ACVIM stage B2 MMVD disease, before presentation for acute CHF, did go on to develop acute clinical CHF requiring hospitalization. The extent to which the onset of stage C was delayed is unknown, although postdiuretic sodium retention occurs within 3 to 6 hours of initial dosing, therefore unlikely to have had a major impact. ¹⁸ , ⁴² The extent to which the prior furosemide treatment tempered the lung congestion score is also unknown.

Many dogs were lost to follow‐up and censored and likely incorrectly classified skewing the outcome results. The exclusion criteria eliminated many dogs from this study, reducing the numbers to a small cohort with the possibility of a difference being overlooked because of the lack of power in this study (type II error). In humans, association between lung congestion and outcome is established in myocardial infarction heart failure and a previous study examining edema scores in dogs did find a significant association with outcome. ²⁰ , ²⁴ The other canine study recruited similar numbers to our study but prospectively recruited dogs that had either MMVD or dilated cardiomyopathy, the imaging was performed within 6‐hours of admission to hospital. ²⁴ The prescriptive management, stricter criteria for recruitment, earlier time frame of imaging, and greater heterogeneity of acquired disease could account for the differences between this study and ours. Lack of complete data set in each dogs medical records, particularly related to gaps in the referral history and the time of receiving furosemide before referral, compromised the ability to record all data points. Despite the strict exclusion criteria, many dogs could have been misclassified as suffering from left‐sided CHF and might have been suffering from misdiagnosed non‐CPE, chronic bronchitis, and interstitial lung disease, all as comorbidities with MMVD. The inability to score each lung lobe on radiographs because of the severity of cardiomegaly and the cardiac silhouette obscuring the image meant that some lung lobes were not assigned a score and the final score was calculated as the total minus the lobes not examined. The identification and classification of especially an unstructured interstitial pattern is inherently subjective and influenced by other factors including technical parameters and habitus. We tried to address this limitation by using defined criteria when assigning the score. This study utilized a consensus read between a junior investigator (veterinary student) and a board‐certified radiologist. The composite score would allow other causes of interstitial‐alveolar patterns, lobar or diffuse, including noncardiogenic edema and pulmonary parenchymal disease, common as comorbidities in dogs with MMVD, to contribute to the overall score and skew data. The time stamp on radiographs might not have reflected the actual time of the radiograph performed, compromising the calculation of the time interval from receiving furosemide. Some dogs did not receive furosemide until after the radiographs and these could not be included in the analysis for secondary outcomes. It could be argued that not all lung lobes are equal in volume or function and a better score could be created by proportionally assigned fractions dictated by lung volume. Lung ultrasound and B‐lines, considered more sensitive, although lacking specificity, could have added value to this study as is strongly associated with outcome in humans and dogs with heart failure. ²⁴ , ³⁴ Lastly, the survival range in some dogs exceeds the expected survival for stage C MMVD. Although extended survival is reported in dogs with advanced heart failure, it is possible that some dogs were misdiagnosed as stage C rather than stage B2 because of the presence of dyspnea, an alveolar interstitial syndrome, and cardiomegaly. ⁴² In this scenario, the LCS would be related to noncardiogenic edema rather than CHF and could have impacted the predictive value of this tool.

CONFLICT OF INTEREST DECLARATION

Authors have no conflict of interest to report.

OFF‐LABEL ANTIMICROBIAL DECLARATION

Authors declare no off‐label use of antimicrobials.

INSTITUTIONAL ANIMAL CARE AND USE COMMITTEE (IACUC) OR OTHER APPROVAL DECLARATION

Authors declare no IACUC or other approval was needed.

HUMAN ETHICS APPROVAL DECLARATION

Animal owners signs a hospital consent form that the medical record date and imaging studies pertaining to their pet can be used for publications.

Supporting information

Data S1. Supporting Information.

Koster L, Vogel J, Springer CM, Hecht S. Radiographic lung congestion scores in dogs with acute congestive heart failure caused by myxomatous mitral valve disease. J Vet Intern Med. 2023;37(6):1983‐1991. doi: 10.1111/jvim.16850