Presence of bronchial nodules, younger age, and heavier body weight are associated with a diagnosis of eosinophilic lung disease in dogs with cough

Enrico Bottero Endovet Professional Association, Rome, Italy

Search for other papers by Enrico Bottero in
Current site
Google Scholar
PubMed
Close
 DMV
,
Pietro Ruggiero Endovet Professional Association, Rome, Italy

Search for other papers by Pietro Ruggiero in
Current site
Google Scholar
PubMed
Close
 DMV, MS
,
Elena Benvenuti Endovet Professional Association, Rome, Italy

Search for other papers by Elena Benvenuti in
Current site
Google Scholar
PubMed
Close
 DMV, PhD
,
Emanuele Mussi Endovet Professional Association, Rome, Italy

Search for other papers by Emanuele Mussi in
Current site
Google Scholar
PubMed
Close
 DMV
,
Daniele Falcioni Endovet Professional Association, Rome, Italy

Search for other papers by Daniele Falcioni in
Current site
Google Scholar
PubMed
Close
 DMV, MS
,
Davide De Lorenzi Veterinary Hospital I Portoni Rossi, Zola Pedrosa, Bologna, Italy

Search for other papers by Davide De Lorenzi in
Current site
Google Scholar
PubMed
Close
 DMV, PhD
, and
Nicola Di Girolamo College of Veterinary Medicine, Oklahoma State University, Stillwater, OK

Search for other papers by Nicola Di Girolamo in
Current site
Google Scholar
PubMed
Close
 DMV, PhD

Click on author name to view affiliation information

Abstract

OBJECTIVE

To describe the association between a diagnosis of eosinophilic lung disease (ELD) in dogs with signalment and bronchoscopic features and evaluate the accuracy of visualization of nodules for the diagnosis of ELD.

ANIMALS

781 dogs with cough that underwent bronchoscopy between 2014 and 2016.

PROCEDURES

Data were extracted from the medical records of each included dog. Multivariable logistic regression was performed to investigate associations between ELD and patient characteristics.

RESULTS

ELD was diagnosed in 113 (14.5%) dogs. More than 3 nodular lesions of the bronchial mucosa were detected in 64 (8.2%) dogs. The odds of having ELD were greater in dogs with nodules (adjusted OR [aOR], 26.0; 95% CI, 13.0 to 52.0) and static bronchial collapse (aOR, 2.3; 95% CI, 1.1 to 4.6), and lower in dogs having focal versus diffuse inflammation (aOR, 0.05; 95% CI, 0.01 to 0.37). The odds of having ELD decreased for each 1-year increase in age (aOR, 0.86; 95% CI, 0.80 to 0.92), and increased for each 1-kg increase in weight (aOR, 1.04; 95% CI, 1.01 to 1.06). Visualization of nodules during bronchoscopy had a overall accuracy of 89.4% (95% CI, 87.0% to 91.4%), sensitivity of 41.6% (32.4% to 51.2%), and specificity of 97.5% (96.0% to 98.5%) for a diagnosis of ELD.

CLINICAL RELEVANCE

On the basis of high specificity and negative predictive value, lack of visualization of bronchial nodules during bronchoscopy can be used to preliminarily rule out ELD. However, visualization of bronchial nodules does not imply presence of ELD. This could be especially relevant when results of BAL cytology are available several days after the actual bronchoscopy.

Abstract

OBJECTIVE

To describe the association between a diagnosis of eosinophilic lung disease (ELD) in dogs with signalment and bronchoscopic features and evaluate the accuracy of visualization of nodules for the diagnosis of ELD.

ANIMALS

781 dogs with cough that underwent bronchoscopy between 2014 and 2016.

PROCEDURES

Data were extracted from the medical records of each included dog. Multivariable logistic regression was performed to investigate associations between ELD and patient characteristics.

RESULTS

ELD was diagnosed in 113 (14.5%) dogs. More than 3 nodular lesions of the bronchial mucosa were detected in 64 (8.2%) dogs. The odds of having ELD were greater in dogs with nodules (adjusted OR [aOR], 26.0; 95% CI, 13.0 to 52.0) and static bronchial collapse (aOR, 2.3; 95% CI, 1.1 to 4.6), and lower in dogs having focal versus diffuse inflammation (aOR, 0.05; 95% CI, 0.01 to 0.37). The odds of having ELD decreased for each 1-year increase in age (aOR, 0.86; 95% CI, 0.80 to 0.92), and increased for each 1-kg increase in weight (aOR, 1.04; 95% CI, 1.01 to 1.06). Visualization of nodules during bronchoscopy had a overall accuracy of 89.4% (95% CI, 87.0% to 91.4%), sensitivity of 41.6% (32.4% to 51.2%), and specificity of 97.5% (96.0% to 98.5%) for a diagnosis of ELD.

CLINICAL RELEVANCE

On the basis of high specificity and negative predictive value, lack of visualization of bronchial nodules during bronchoscopy can be used to preliminarily rule out ELD. However, visualization of bronchial nodules does not imply presence of ELD. This could be especially relevant when results of BAL cytology are available several days after the actual bronchoscopy.

Introduction

Eosinophilic lung disease (ELD) is a heterogenous group of disorders characterized by eosinophilic infiltration into the airways or the pulmonary parenchyma, previously described as pulmonary infiltration with eosinophils (PIE), pulmonary eosinophilia (PE), eosinophilic pneumonia, and eosinophilic bronchopneumopathy (EBP).14 These diseases have recently been classified as eosinophilic bronchitis (EB), eosinophilic granuloma (EG). and eosinophilic pulmonary disease (EBP) on the basis of hematological, radiographic, and bronchoscopic findings in conjunction with bronchoalveolar (BAL) fluid cytologic assessment.4

Eosinophilic lung disease is considered more common in young adult dogs (mean age at diagnosis, 4.5 to 6 years).1,3 This condition is reported in mixed and pure breed dogs, including Siberian Huskies, Alaskan Malamutes, and Labrador Retrievers.24 The most common clinical sign of ELD is cough, with a reported incidence of 92% to 100%, depending on the study.14 Other clinical signs may include nasal discharge, tachypnea, dyspnea or other abnormal respiratory sounds, and exercise intolerance.14

To date, the etiology of ELD has not yet been determined. A suggested etiology is hypersensitivity to inhaled allergens.1,3,5 Cardiac and respiratory parasites are a common cause of pulmonary eosinophilia and are the major differential consideration in EBP cases,6,7 whereas bacterial or viral infection are unlikely to be involved in the etiopathogenesis of ELD. In a recent study,8 there was no difference in detection rate of Mycoplasma canis, Mycoplasma cynos, and Bordetella bronchiseptica in dogs with eosinophilic infiltration, compared with control dogs.

Depending on the individual study,24,8 clinical diagnosis of ELD has been performed based on the cytologic presence of increased eosinophil counts in BAL fluid in addition to results of other diagnostic tests. Hematological changes can be suggestive of the disorder, with peripheral eosinophilia reported in 36% to 61% of the dogs affected by ELD.1,3,4 Changes visualized upon bronchoscopy may be mild, such as a form of bronchitis. or severe, including polypoid development, bronchiectasis, presence of yellow-green airway exudate, and airway collapse.4 It is currently unclear whether gross bronchoscopic features alone can be indicative of ELD.

For this purpose, we employed a large dataset that included dogs with a diagnosis of BAL fluid eosinophilia and dogs with different results on BAL fluid cytology. The aims of the present study were to describe the prevalence of ELD in dogs with cough undergoing bronchoscopy in Italy, to evaluate the association of a diagnosis of ELD with individual dog characteristics and bronchoscopic findings, and to evaluate the diagnostic accuracy of visualization of bronchial nodules for the diagnosis of ELD.

Materials and Methods

Study design

A retrospective, multicenter, diagnostic accuracy study was designed based on the research question following the PIRT (population, index test, reference standard, and target condition) framework,9 which is a simple tool that help to define a clinical question answerable by a diagnostic accuracy study. For this study, the population consisted of dogs with cough undergoing bronchoscopy and BAL, the index test was gross bronchoscopic features, the reference standard was BAL results, and the target condition was ELD.9 The study was reported following the Standards for Reporting of Diagnostic Accuracy Studies (STARD) statement,10 which was developed to improve the completeness and transparency of reports of diagnostic accuracy studies. The eligibility criteria for the diagnostic accuracy study were purposefully broad to also address the first 2 aims of the study (ie, the prevalence of ELD and the association of ELD with individual patient features).

Eligibility criteria

A search was conducted on the database of Endovet from January 2014 to December 2016. Endovet is a group of associate veterinarians that perform endoscopies and related clinical activity in several veterinary referral centers throughout Italy. All dogs affected by cough and that had undergone a bronchoscopic examination of the lower respiratory tract and BAL fluid cytology in the time period were eligible for inclusion in the study.

Data extraction

The following information was extracted from the medical records of each included dog: breed, age, body weight, sex and reproductive status, duration of the cough (< 1 month or ≥ 1 month), other clinical signs, bronchoscopic alterations (presence or absence of inflammation, tracheal collapse, static bronchial collapse, dynamic bronchial collapse, and nodules), results of BAL fluid cytology, and results of nodule squash preparation cytology. Dogs were classified as having nodules if 3 or more nodules were visualized during bronchoscopy in any explorable area of the airways. Nodules were defined as thickening of the mucosa with a flat base (sessile) and rounded form, white to light red color, and minimal diameter of 3 mm (Figure 1).

Figure 1
Figure 1

Representative bronchoscopic images of dogs with eosinophilic lung disease that were categorized as having bronchial nodules. A—Presence of disseminated, sessile, white to pink nodules in lobar bronchi. B—Visualization of 1 nodule and significant hyperemia, irregularities of the mucosa, and exudate in segmental bronchi. C—Presence of 4 mucosal nodules surrounding the bifurcation of a segmental bronchus. D—Close-up view of an isolated, sessile nodule; notice the irregular mucosa.

Citation: Journal of the American Veterinary Medical Association 260, 4; 10.2460/javma.20.10.0559

Bronchoscopy

The bronchoscopy technique was consistent among the nine contributing veterinarians. Different types of flexible video bronchoscopes (Fujifilm EB 530 S–standard type and Super Slim Pediatric Type) were used depending on the size of the patient. The larynx and trachea were evaluated first. The endoscope was passed in the left cranial and the left caudal lobar bronchi. The endoscope was then retracted and passed in the right cranial, middle, accessory and caudal lobar bronchi, including examination of all explorable segmental and sub-segmental bronchi. After a complete exploration of the bronchial tree in both lungs, bronchoalveolar lavage was performed in all patients using the following standardized technique.11 The instrument was introduced into the smallest accessible bronchus of the pulmonary lobe which presented the most severe alterations. A bolus of warm sterile physiological saline solution (1 to 2 mL/kg) was instilled through the working channel of the bronchoscope and immediately withdrawn with the aid of a surgical aspirator connected to the endoscope. Samples were considered adequate if consisted of at least 50% of the total volume of the instilled saline solution and if presented a cloudy and foamy appearance; these were immediately divided into 2 aliquots: one placed in an EDTA tube for cytological evaluation and the other in a sterile tube for further potential diagnostics.

BAL cytology evaluation

The cytological samples were prepared by means of cytocentrifugation or sedimentation within 30 minutes of collection and stained with Romanowsky-type stains.12 Cytology slides were submitted to a laboratory that at the time had 1 of 3 possible operators reading BAL samples (2 board-certified clinical pathologists and 1 veterinarian with experience in clinical pathology). Specific blinding procedures were not set up for the samples. Each sample was evaluated for presence and relative quantity of neutrophils, neutrophils with intracellular bacteria, eosinophils, macrophages, lymphocytes, parasites (eggs or larvae), malignant cells, and other cells. Cytologic diagnoses were grouped into the following categories: normal, neutrophilic inflammation (> 10% neutrophils), neutrophilic septic inflammation (> 10% neutrophils with intracellular bacteria or predominant karyolitic neutrophils), eosinophilic inflammation (> 25% eosinophils), macrophagic inflammation (> 80% macrophages), lymphocytic inflammation (> 15% small lymphocytes), parasitic infestation (presence of at least 1 egg or 1 larva), neoplastic (prevalence of non-inflammatory cells with cytomorphological atypia), or nondiagnostic (reduced or absent cellularity, inadequate sample). Each sample could have more than 1 cytological diagnosis.

On the basis of previous literature,1316 and to include an unequivocal abnormal increase of eosinophils, the clinical diagnosis of lung eosinophilia in this study was made if the BAL sample contained > 25% eosinophils among all cells retrieved (Figure 2).

Figure 2
Figure 2

Representative bronchoalveolar lavage cytologic image consistent with eosinophilic inflammation. Romanowsky stain; bar = 10 μm.

Citation: Journal of the American Veterinary Medical Association 260, 4; 10.2460/javma.20.10.0559

Squash preparation cytology of bronchial nodules

In some of the dogs with bronchial nodules, squash preparation cytology of the nodules was performed after sampling with 1.2- or 1.8-mm fenestrated oval biopsy forceps (Storz oval biopsy forceps 60175ZE and 60001 KL) using a technique previously described.17 Cytologic findings for nodules were classified as eosinophilic inflammation or noneosinophilic inflammation. All cytology slides of all the nodules were interpreted by the same veterinarian (EB).

Statistical analysis

The Shapiro-Wilk test was used to evaluate continuous data for normality. Continuous variables were nonnormal and were summarized as median and range. For categorical variables, the fraction of patients in each category was calculated. Logistic regression was used to determine the association between individual characteristics (predictors) and the diagnosis of ELD (binary outcome). Individual characteristics included in the initial model were age (years), sex, reproductive status (sexually intact or neutered), body weight (kg), cough duration (≥ 1 month or < 1 month), presence of inflammation (yes or no), distribution of the process (1 lobe or multiple lobes), presence of nodules (yes or no), presence of tracheal collapse (yes or no), presence of static bronchial collapse (yes or no), presence of dynamic bronchial collapse (y/n), and presence of bronchiectasis (yes or no). The variables were first entered in the model using the forward stepwise (conditional) function. The P value cutoff for exclusion of a variable was set at > 0.1. Predictors considered to be potentially clinically relevant confounders (ie, sex and age) were entered in the model regardless of their significance. To avoid overfitting the model, plans were to include a minimum of 10 events per variable. The Hosmer-Lemeshow test was used to assess the goodness of fit of the model, and the Nagelkerke R2 to assess its predictive power. A receiver operating characteristics (ROC) curve was built with the predicted probabilities obtained from the final model, and the area under the curve (AUC) was reported with 95% CIs.

Sensitivity, specificity, positive predictive value, negative predictive value, positive and negative likelihood ratios, and overall accuracy of gross bronchoscopic features to differentiate dogs with EBP from dogs with other conditions were calculated with their respective 95% CIs. Analyses were performed with commercial software (SPSS version 22.0; IBM Corp).

Results

Dogs

Of the 845 dogs that had a bronchoscopy performed during the study period, 64 (7.6%) dogs did not have BAL cytology results in their medical records and were excluded. The remaining 781 (92.4%) dogs fulfilled the inclusion criteria and were included in the study. These dogs ranged in age from 0.4 to 16 years (median 8.0 years), and in body weight from 1.5 to 45.0 kg (median 13.0 kg). A total of 325 (41.7%) dogs were female and 456 (58.3%) were male; 514 (65.8%) dogs were sexually intact, and 267 (34.2%) were neutered. The 5 most represented breeds were mixed-breed (33.9%, 265/781), setter (6.4%, 50/781), Labrador Retriever (5.0%, 39/781), Yorkshire Terrier (4.4%, 35/781), and Poodle (3.8%, 30/781).

The majority of dogs presented with cough for ≥ 1 month (73.1%, 571/781), while about a fourth of the dogs (26.9%, 210/781) had cough for < 1 month. Two-hundred forty-seven dogs had one or more additional clinical signs, including retching (15.9%, 124/781), sneezing (7.3%, 57/781), nasal discharge (2.8%; 22/781), regurgitation (2.4%, 19/781), vomiting (2.0%, 16/781), dyspnea (0.9%, 7/781), itching (0.4%, 3/781), and other respiratory sounds (eg, stertor, stridor, wheezing; 0.4%, 3/781).

Results of bronchoscopic examination

Bronchoscopic examinations were performed by nine veterinarians and showed mucosal inflammation in 771 (88.5%) dogs, 677 of which had inflammation diffused to multiple lobes (77.7%). Nodular lesions of the bronchial mucosa were detected in 102 (13.1%) dogs, 64 (8.2%) of which had more than 3 nodules. Tracheal collapse was present in 161 (20.6%) dogs, dynamic bronchial collapse in 298 (34.2%) dogs, and static bronchial collapse in 107 (12.3%) dogs. Bronchiectasis was present in 67 (7.7%) dogs. Fifty-four (6.9%) dogs had bronchial foreign bodies.

Results of BAL cytology and squash preparation cytology

Approximately half of BAL fluid cytology samples (53.4%, 417/781) were diagnosed as nonseptic neutrophil inflammation. Septic neutrophil inflammation was diagnosed in 130 (16.6%) samples. Eosinophilic inflammation was diagnosed in 113 (14.5%) samples having > 25% eosinophils (Table 1). A cellular composition considered normal was present in 84 (10.7%) samples. Macrophagic inflammation was diagnosed in 42 (5.4%) samples. Parasites (eggs or larvae) were observed in 30 (3.8%) samples, and were identified as Angiostrongylus vasorum (23), Eucoleus aerophilus (4), Crenosoma vulpis (2), and Filaroides osleri (1). Of these cases, 7 had lung eosinophilia and none had bronchial nodules. Presence of neoplastic cells, plasma cells or lymphocytosis was detected in 20 (2.6%) samples. Four (0.5%) samples were considered nondiagnostic due to the limited cellularity. The BAL cytology results were stratified by age and body weight (Table 2).

Table 1

Clinical and bronchoscopic morphological characteristics of dogs with a proportion of eosinophils among the total of cells retrieved during bronchoalveolar lavage (BAL) ≥ than 25%.

Characteristic No. of dogs No. (%) with ≥ 25% eosinophils
Sex
 Female 325 53 (16.3)
 Male 455 60 (13.2)
Reproductive status
 Sexually intact 514 76 (14.8)
 Neutered 267 37 (13.9)
Cough duration
 ≥ 1 mo 572 60 (10.5)
 < 1 mo 209 53 (25.4)
Inflammation
 Present 771 112 (14.5)
 Absent 9 1 (11.1)
Distribution of inflammation
 Diffuse 677 111 (16.4)
 Focal 95 1 (1.1)
Bronchiectasis
 Absent 714 108 (15.1)
 Present 67 5 (7.5)
Tracheal collapse
 Absent 613 103 (16.8)
 Present 161 9 (5.6)
Static bronchial collapse
 Absent 672 94 (14.0)
 Present 107 18 (16.8)
Dynamic bronchial collapse
 Absent 483 81 (16.8)
 Present 298 32 (10.7)
 > 3 nodules
 Absent 717 66 (9.2)
 Present 64 47 (73.4)

— = Not applicable.

Table 2

Clinical and bronchoscopic morphological characteristics of dogs with the most common types of inflammation identified on BAL fluid cytology.

Characteristic Normal Neutrophilic septic Neutrophilic nonseptic Macrophagic
Age (y) 8.0 (1.0–16.0) 5.0 (0.4–16.0) 10.0 (0.5–16.0) 8.0 (0.4–14.0)
Body weight (kg) 9.5 (3.0–34.0) 17.5 (4.0–45.0) 11.0 (2.0–45.0) 13.0 (2.5–40.0)
Sex
 Female 38 (14.9) 41 (16.1) 137 (53.7) 15 (5.9)
 Male 46 (12.4) 57 (15.4) 215 (58.1) 23 (6.2)
Reproductive status
 Sexually intact 52 (12.4) 75 (17.9) 228 (54.4) 25 (6.0)
 Neutered 32 (15.5) 23 (11.1) 125 (60.4) 13 (6.3)
Cough duration
 ≥ 1 mo 57 (11.9) 67 (13.9) 294 (61.1) 36 (7.5)
 < 1 mo 27 (18.6) 31 (21.4) 59 (40.7) 2 (1.4)
Inflammation
 Present 78 (12.6) 98 (15.9) 351 (56.8) 38 (6.1)
 Absent 6 (85.7) 0 (0.0) 1 (14.3) 0 (0.0)
Distribution
 Diffuse 66 (12.2) 77 (14.2) 309 (56.9) 38 (7.0)
 Focal 12 (15.8) 21 (27.6) 43 (56.6) 0 (0.0)
Nodules
 Absent 84 (14.1) 96 (16.2) 346 (58.2) 38 (6.4)
 Present 0 (0.0) 2 (6.3) 7 (21.9) 0 (0.0)
Tracheal collapse
 Absent 56 (11.6) 86 (17.9) 261 (54.3) 26 (5.4)
 Present 27 (19.0) 11 (7.7) 91 (64.1) 12 (8.5)
Static bronchial collapse
 Absent 77 (14.3) 85 (15.8) 299 (55.7) 32 (6.0)
 Present 7 (8.0) 12 (13.6) 54 (61.4) 6 (6.8)
Dynamic bronchial collapse
 Absent 41 (11.0) 78 (21.0) 189 (50.8) 23 (6.2)
 Present 43 (16.9) 20 (7.9) 164 (64.6) 15 (5.9)
Bronchiectasis
 Absent 77 (13.5) 87 (15.3) 323 (56.7) 33 (5.8)
 Present 7 (12.5) 11 (19.6) 30 (53.6) 5 (8.9)

Values for age and body weight are mean (range). All other values are number (%) of dogs with the indicated characteristic. See Table 1 for key.

Squash preparation cytology was performed in 66 dogs, with results indicating that 49 bronchial nodules were eosinophilic and 17 were non-eosinophilic. All eosinophilic nodules and 1 of 17 noneosinophilic nodules pertained to dogs with BAL findings consistent with ELD. The remaining 16 noneosinophilic nodules pertained to 6 dogs with aseptic neutrophilic inflammation, 4 dogs with septic neutrophilic inflammation, 3 dogs with lymphocytic inflammation, 2 dogs with macrophagic inflammation, and 1 dog with a non-diagnostic BAL cytology.

Characteristics of dogs with ELD

Dogs with ELD were between 0.6 and 16 years old (median, 5 years), and weighed between 3 and 45 kg (median, 18 kg; Table 1). Sixty-five (57.5%) dogs were males, 48 (42.4%) were females, and 35 (30.9%) were neutered. Within the population of dogs affected by ELD, the 5 most represented breeds were mixed-breed (34), Labrador Retriever, Epagneul Breton (6), Irish Setter, and Golden Retriever (5). Fifty-three (46.9%) dogs had cough for less than 1 month and 60 dogs (53.1%) had cough for ≥ 1 month. Retching was the most common additional clinical sign and was present in 24 (21.2%) dogs. Other less common clinical signs included sneezing in 9 (8.0%) dogs, nasal discharge in 4 (3.5%) dogs, regurgitation in 2 (1.8%) dogs, and vomiting in 1 (0.9%) dog. Dyspnea, pruritus, and other respiratory sounds (eg, stertor, stridor, and wheezing) were not observed in dogs with ELD. Bronchoscopic examination showed mucosal inflammation in 112 (99.1%) dogs, 111 of which had inflammation diffused to multiple lobes. Nodular lesions of the bronchial mucosa were detected in 64 (56.6%) dogs, 47 (41.6%) of which had more than 3 nodules. Tracheal collapse was present in 9 (8.0%) dogs, dynamic bronchial collapse in 32 (28.3%) dogs, and static bronchial collapse in 18 (15.9%) dogs. Bronchiectasis was present in 5 (4.4%) dogs.

Associations between individual factors and ELD

Of the 12 variables that were tested for inclusion in the multivariable model, only age, sex, body weight, cough duration, distribution of inflammation, presence of static bronchial collapse and presence of > 3 nodules were entered in the final model. Regarding bronchoscopic results, the odds of having ELD were 26 times the odds in dogs presenting nodules (aOR, 26.0; 95% CI, 13.0 to 52.0, P < 0.001), 20 times lower in dogs presenting focal rather than diffuse inflammation (aOR, 0.05; 95% CI, 0.01 to 0.37, P = 0.003), and twice the odds in dogs presenting with static bronchial collapse (aOR, 2.3; 95% CI, 1.1 to 4.6; P = 0.025). Regarding dog signalment, for each 1-year increase in age the odds of having ELD decreased by 14% (aOR, 0.86; 95% CI, 0.8 to 0.92, P < 0.001), and for each 1-kg increase in body weight the odds of having ELD increased by 4% (aOR, 1.04; 95% CI, 1.01 to 1.06, P = 0.002) (Figure 3). Dogs with cough for < 1 month had approximately twice the odds of having ELD as dud dogs with cough for longer (aOR, 1.98; 95% CI, 1.10 to 3.54, P = 0.022). Sex had a nonsignificant effect on the odds of having ELD (aOR, 0.78; 95% CI, 0.47 to 1.3, P = 0.34). The final logistic regression model was adequately fitted (Hosmer-Lemeshow test, P = 0.19) and explained 41% of the variability of the dataset (Nagelkerke R2=0.41). The AUC of the ROC curve built from the predicted probabilities of the model was 0.85 (95% CI, 0.82 to 0.89; Figure 4).

Figure 3
Figure 3

Dot plots of body weight (A) and age (B) of individual dogs undergoing bronchoscopy stratified by the presence or absence of eosinophilic bronchopathy. BAL = Bronchoalveolar.

Citation: Journal of the American Veterinary Medical Association 260, 4; 10.2460/javma.20.10.0559

Figure 4
Figure 4

Receiver operating characteristics curve built based on the predicted probabilities of the final multivariable logistic regression model. The area under the curve is 0.85 (95% CI, 0.82 to 0.89).The final logistic regression model included 767 dogs, was adequately fitted (Hosmer-Lemeshow test, P = 0.19), and explained 41% of the variability of the dataset (Nagelkerke R2 = 0.41).

Citation: Journal of the American Veterinary Medical Association 260, 4; 10.2460/javma.20.10.0559

Diagnostic accuracy of nodule visualization for diagnosis of ELD

Considering visualization of > 3 nodules as the index test, 47 dogs would have been correctly diagnosed with ELD, 651 dogs would have been correctly diagnosed without ELD, 66 dogs would have been false negatives, and 17 dogs would have been false positives. Therefore, visualization of nodules during bronchoscopy had a sensitivity of 41.6% (95% CI, 32.4% to 51.2%), specificity of 97.5% (96.0% to 98.5%), positive likelihood ratio of 16.3 (9.7 to 27.4), and negative likelihood ratio of 0.60 (0.51 to 0.70). Based on the prevalence of ELD in this population, the overall accuracy of bronchoscopic visualization of nodules was 89.4% (95% CI, 87.0% to 91.4%), the positive predictive value was 73.4% (62.2% to 82.3%), and the negative predictive value was 90.8% (89.4% to 92.0%).

Discussion

Our study demonstrated that there are several factors that are associated with lung eosinophilia and that the lack of visualization of bronchial nodules during bronchoscopy can be used to preliminarily rule out ELD, while waiting for cytological confirmation. This could be especially important in such veterinary settings or in those specific situations in which results of BAL cytology are available several days after the actual bronchoscopy. Our study also found that visualization of bronchial nodules does not imply presence of ELD.

Eosinophilic lung disease was not uncommon in dogs with cough undergoing bronchoscopy, affecting approximately one seventh (14.5%) of the cases in this study. A recent study3 on ELD in dogs had a strikingly similar prevalence, with BAL eosinophilia documented in 14.9% (85/569) dogs undergoing bronchoscopy. In that study, BAL fluid eosinophil percentages exceeding 14% were considered indicative of ELD. In the present study, we employed a higher BAL fluid eosinophil cutoff (25%) as indicative of ELD in dogs. Depending on the source, different percentages of eosinophils are considered normal in BAL fluid cytology, typically between 5 and 10%.13,14 However, in 1 study,15 dogs not raised in a laboratory environment had an average of 24% of eosinophils in BAL fluid cytology. We opted to use a relatively high cutoff to avoid misdiagnosis of ELD in dogs that had only a mild lung eosinophilia. The choice of such a cutoff strengthened the diagnosis of lung eosinophilia in the cases described herein but could have resulted in a potential underestimation of the actual number of ELD cases over the population examined.

Our study identified multiple morphological characteristics associated with a diagnosis of ELD in dogs. The most consistent of these characteristics was the presence of > 3 nodules in the bronchial mucosa; however, noticing > 3 nodules also had a low positive predictive value. In the human medical literature, there are several case reports1820 that document development of bronchial nodules or polyps as a consequence of exposure to inhalation of smoke, toxins or thermal injury. Differently, in allergic reactions such as asthma, diffuse polyposis is uncommon.21 In horses, multinodular pulmonary fibrosis may result in the formation of nodular lesions in the lungs as a consequence of the infection with a γherpesvirus: equine herpesvirus 5.22 However this condition is characterized by a lung neutrophilia rather than eosinophilia.23 Another condition described in horses that could resemble what observed in these cases is idiopathic chronic eosinophilic pneumonia.24 Gross postmortem findings in horses may show 2- to 3-mm, firm, nodules disseminated throughout the lung parenchyma.24 In other species, formation of multiple nodules on specific mucosal surfaces is sporadically reported, such as polypoid enteritis in avian species, polypoid cystitis in rabbits, among others.25,26 In general, nodular or polypoid changes of mucosal surfaces seem to be associated to chronic inflammation or chronic trauma, as observed in case of catheter-associated polypoid urethritis and cystitis.27,28

It is unclear if the prevalence of nodules seen in the dogs in the present study could have been somehow related to specific inhalants present in the region of origins. The group of dogs included in the study was heterogeneous due to the multi-institutional design of the study, originating from diverse urban and suburban areas from southern, central and northern Italy, decreasing the chance of a common regional cause for this finding. Also, it should be noted that in this study an a priori cutoff of 3 nodules was employed to define patients with multiple bronchial nodules. This cutoff was chosen to avoid including in the category of dogs with multiple bronchial nodules those that had only one or two nodules and that did not have a clear clinical significance.

In the present study we found that the odds of having ELD rather than another cytological diagnosis increased as body weight increased. This finding was novel considering that previous studies have not compared characteristics of dogs with and without ELD. Our result confirmed the general assumption that larger dogs are more likely to develop ELD than smaller dogs.1,3,4 The prevalence of ELD could increase for a mechanical factor associated with the larger size of the dogs, or could increase because of an intrinsic higher prevalence in larger dog breeds. However, this latter option seems less likely because the majority of affected dogs in the present study were of mixed breeds, similar to what reported in dogs with eosinophilic granuloma.4 In addition, we found that younger dogs were significantly more at risk of suffering ELD than other diseases. The median age of dogs with ELD was 5 years in our study, and the odds of having ELD decreased by 14% for each 1-year increase in age. Similar young ages (4.5 to 6 years) were reported in previous case series of dogs affected by ELD.1,4 Median age of dogs with neutrophilic nonseptic inflammation in our dataset was 10 years, indicating a possible predisposition to older dogs for chronic bronchitis rather than ELD or septic bronchitis. The readers should be aware that, as shown in Figure 3, there was major overlap of individual weights and ages in dogs with and without ELD. Based on the degree of this overlap, the use of signalment has limited clinical application in the diagnosis of this disease, but knowledge of this factor could be useful as a further insight for future researchers, perhaps to understand the natural history of the disorder and to identify potential causative factors.

The present study was conceived as a diagnostic accuracy study. Diagnostic accuracy studies are characterized by the application of an index test, in this case the visualization of nodules, and a reference test (BAL fluid cytology), in a well-defined population (dogs with cough undergoing bronchoscopy), for the diagnosis of a specific condition (lung eosinophilia).10 Diagnostic accuracy studies allow calculation of statistics such as sensitivity, specificity, predictive values, and likelihood ratios that are considered ideal for evaluation of categorical diagnostic tests and are readily translated into clinical practice, even if they may not always be readily understood by clinicians.29 Bronchoscopy did not identify nodular lesions in 651 dogs unaffected by ELD but identified nodular lesions in 17 dogs unaffected by EBP. This resulted in high specificity of the technique (97.5%), since there were only 17 false positive results among the 781 dogs examined. On the other hand, bronchoscopy identified nodules in 113 dogs, 47 of which had lung eosinophilia, resulting in a much lower sensitivity (41.6%). These results indicated that visualization of nodules during bronchoscopy may or may not be associated with lung eosinophilia, but the lack of visualization of nodules during bronchoscopy could be an important indicator about the lack of lung eosinophilia. In addition to the simple determination of the presence or absence of nodules, potentially the full model could be used for prediction of ELD. Based on the AUC of the ROC curve, the use of that model should result in a correct prediction in 85% of cases.

The other 2 bronchoscopic findings found to be associated with ELD in the present study were the presence of diffuse rather than focal inflammation and the presence of static bronchial collapse. Dynamic bronchial collapse, tracheal collapse, bronchiectasis, and overall presence of inflammation were not associated with a diagnosis of ELD. Dynamic bronchial collapse was the most common of the airway collapse alterations, affecting over one-third of the dogs in the study, while both static bronchial collapse and tracheal collapse were less common, affecting about one-tenth of dogs. The definition of the different clinical presentations of bronchial collapse is based on a recent classification.30 Static bronchial collapse is characterized by a reduction in luminal diameter during all phases of respiration, while dynamic bronchial collapse is characterized by the reduction in luminal diameter on respiration. Since static bronchial collapse is considered the progression of dynamic bronchial collapse,30 it is possible that dogs with ELD have a higher proportion of static bronchial collapse as a consequence of the chronicity of the disease.

In a previous study,31 bronchiectasis was found to be significantly more common in dogs with EBP (80%) or inflammatory airway disease (74%) than in those with pneumonia (42%). However, in that study only 10 dogs with EBP were included, 8 of which were affected by bronchiectasis. In other studies that evaluated bronchoscopic findings in dogs with ELD or EBP, the proportion with bronchiectasis was definitively lower, of 0%,2 14%,3 26%1 and 48%.4 The 2 articles4,31 that reported the highest proportion of bronchiectasis in dogs with ELD (48% and 80%) included cases from the same institution. In our study, only 4.4% of dogs with ELD had bronchiectasis and our analysis found no association between this alteration and a diagnosis of ELD. It is possible that the difference in the prevalence of bronchiectasis depends on the techniques used for its diagnosis (for example, one study31 included radiographic evaluation) or the way the scoring of bronchoscopy was performed. Based on the larger number of cases included in the present study, it seems unlikely that bronchiectasis and ELD are actually associated.

Other common diagnoses in our dataset included neutrophilic inflammation (septic and aseptic), macrophagic inflammation, normal differential cell count and parasitism. Less commonly, neoplastic cells, small lymphocytes, and plasma cells were found. Lymphocytosis, as defined as an amount of lymphocytes > 20%, was more common in a recently published study,16 accounting for 21.6% (123/569) BAL cytology samples. It is unclear why there is such a substantial difference in the presentation of BAL lymphocytosis in these 2 populations, considering that the proportion of dogs with ELD was instead similar.4

The study reported here presented some limitations. First, it was a retrospective study in which bronchoscopy records were extracted for each included dog, but the video recordings were not examined by a single operator. It is thus possible that different operators could have had different assessments. Second, although all dogs had information about cough duration, owners were inconsistently asked about this variable. Thus, cough duration was categorized by the people extracting the data as < 1 month or ≥ 1 month on the basis of the available information in the records. Third, the pharmacological and clinical case histories were not examined systematically via contact with patient owners. It is thus possible that some patients received medications prior to the bronchoscopic evaluation, thus altering the results of BAL. However, considering the large dataset, it is unlikely that these 2 factors would have had a significant impact on the final results. Fourth, tests other than BAL fluid cytology were performed depending on individual veterinarian recommendations and client preferences and were not analyzed in this study. Consequently, it is possible that some cases of ELD were related to parasitic infestation, fungal infection, or other conditions that can result in lung eosinophilia.

In conclusion, the present study showed that the detection of endobronchial nodules during bronchoscopy is frequent in dogs with ELD. Quantification of eosinophils in BAL fluid remains the reference standard for diagnosis of ELD; however, the absence of nodules could be a useful morphologic criterion for providing a preliminary diagnosis of disorders other than ELD, while awaiting cytological confirmation. In addition, this study highlighted associations between ELD and static bronchial collapse, age, and weight that warrant further investigation.

References

  • 1.

    Clercx C, Peeters D, Snaps F, et al. Eosinophilic bronchopneumopathy in dogs. J Vet Intern Med. 2000;14:282291.

  • 2.

    Corcoran BM, Thoday KL, Henfrey JI, et al. Pulmonary infiltration with eosinophils in 14 dogs. J Small Anim Pract. 1991;32:494502.

  • 3.

    Casamian-Sorrosal D, Silvestrini P, Blake R, et al. Clinical features and long-term follow-up of 70 cases of canine idiopathic eosinophilic lung disease. Vet Rec. 2020;187(8):e65.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 4.

    Johnson LR, Johnson EG, Hulsebosch SE, et al. Eosinophilic bronchitis, eosinophilic granuloma, and eosinophilic bronchopneumopathy in 75 dogs (2006–2016). J Vet Intern Med. 2019;33:22172226.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 5.

    Peeters D, Day MJ, Clercx C. Distribution of leucocyte subsets in bronchial mucosa from dogs with eosinophilic bronchopneumopathy. J Comp Pathol. 2005;133:128135.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 6.

    Traversa D, Di Cesare A, Conboy G. Canine and feline cardiopulmonary parasitic nematodes in Europe: emerging and underestimated. Parasit Vectors. 2010;3:62. doi: 10.1186/1756-3305-3-62

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 7.

    Chapman PS, Boag AK, Guitian J, Boswood A. Angiostrongylus vasorum infection in 23 dogs (1999–2002). J Small Anim Pract. 2004;45:435440.

  • 8.

    Canonne AM, Peters I, Roels E, Desquilbet L, Clercx C. Detection of specific bacterial agents by quantitative PCR assays in the bronchoalveolar lavage fluid of dogs with eosinophilic bronchopneumopathy vs. dogs with chronic bronchitis and healthy dogs. Vet J. 2018;232:5256.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 9.

    Mallett S, Halligan S, Thompson M, Collins GS, Altman DG. Interpreting diagnostic accuracy studies for patient care. BMJ. 2012;345:e3999. doi: 10.1136/bmj.e3999

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 10.

    Bossuyt PM, Reitsma JB, Bruns DE, et al. STARD 2015: an updated list of essential items for reporting diagnostic accuracy studies. BMJ. 2015;351:h5527. doi: 10.1136/bmj.h5527

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 11.

    Hawkins EC, DeNicola DB, Kuehn NF. Bronchoalveolar lavage in the evaluation of pulmonary disease in the dog and cat. State of the art. J Vet Intern Med. 1990;4:267274.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 12.

    Marcos R, Santos M, Marrinhas C, et al. Cytocentrifuge preparation in veterinary cytology: a quick, simple, and affordable manual method to concentrate low cellularity fluids. Vet Clin Pathol. 2016;45:725731.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 13.

    Barger AM, MacNeill A. Small Animal Cytologic Diagnosis. CRC Press; 2016.

  • 14.

    Valenciano AC, Cowell RL. Cowell and Tyler's Diagnostic Cytology and Hematology of the Dog and Cat. Elsevier Health Sciences; 2019 Mar 28.

    • Search Google Scholar
    • Export Citation
  • 15.

    Baudendistel LJ, Vogler GA, Frank PA, Zanaboni PB, Dahms TE. Bronchoalveolar eosinophilia in random-source versus purpose-bred dogs. Lab Anim Sci. 1992;42(5):491496.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 16.

    Johnson LR, Vernau W. Bronchoalveolar lavage fluid lymphocytosis in 104 dogs (2006–2016). J Vet Intern Med. 2019;33(3):13151321.

  • 17.

    Bottero E, Melega M, Dimartino ER, et al. Diagnosis of canine gastric adenocarcinoma using squash preparation cytology. Vet Clin Pathol. 2018;47(4):629633. doi: 10.1111/vcp.12667.

    • Search Google Scholar
    • Export Citation
  • 18.

    Adams C, Moisan T, Chandrasekhar AJ, Warpeha R. Endobronchial polyposis secondary to thermal inhalational injury. Chest. 1979;75:643645.

  • 19.

    Glassroth J. Diffuse endobronchial polyposis following a titanium tetrachloride inhalation injury. Am Rev Respir Dis. 1984;130:1189.

  • 20.

    Williams DO, Vanecko RM, Glassroth J. Endobronchial polyposis following smoke inhalation. Chest. 1983;84:774776.

  • 21.

    Niimi A, Amitani R, Ikeda T, Kubo Y, Tanaka E, Kuze F. Inflammatory bronchial polyps associated with asthma: resolution with inhaled corticosteroid. Eur Respir J. 1995;8(7):12371239.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 22.

    Williams KJ. Gammaherpesviruses and pulmonary fibrosis: evidence from humans, horses, and rodents. Vet Pathol. 2014;51:372384.

  • 23.

    Dunkel B. Pulmonary fibrosis and gammaherpesvirus infection in horses. Equine Vet Educ. 2012;24(4):200205.

  • 24.

    Bell SA, Drew CP, Wilson WD, Pusterla N. Idiopathic chronic eosinophilic pneumonia in 7 horses. J Vet Intern Med. 2008;22(3):648653.

  • 25.

    Cohrs P. Textbook of Special Pathological Anatomy of Domestic Animals. Elsevier; 2013.

  • 26.

    Di Girolamo N, Bongiovanni L, Ferro S, et al. Cystoscopic diagnosis of polypoid cystitis in two pet rabbits. J Am Vet Med Assoc. 2017;251(1):8489.

  • 27.

    Norlén LJ, Ekelund P, Hedelin H, Johansson SL. Effects of indwelling catheters on the urethral mucosa (polypoid urethritis). Scand J Urol Nephrol. 1988;22(2):8186.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 28.

    Ekelund P, Johansson S. Polypoid cystitis: a catheter associated lesion of the human bladder. Acta Pathol Microbiol Scand A. 1979;87A(3):179184.

  • 29.

    Whiting PF, Davenport C, Jameson C, et al. How well do health professionals interpret diagnostic information? A systematic review. BMJ Open. 2015;5(7):e008155. doi: 10.1136/bmjopen-2015-008155

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 30.

    Bottero E, Bellino C, De Lorenzi D, et al. Clinical evaluation and endoscopic classification of bronchomalacia in dogs. J Vet Intern Med. 2013;27:840846.

  • 31.

    Johnson LR, Johnson EG, Vernau W, Kass PH, Byrne BA. Bronchoscopy, imaging, and concurrent diseases in dogs with bronchiectasis: (2003–2014). J Vet Intern Med. 2016;30(1):247254.

    • PubMed
    • Search Google Scholar
    • Export Citation
All Time Past Year Past 30 Days
Abstract Views 1210 0 0
Full Text Views 1998 1175 136
PDF Downloads 1620 507 33
Advertisement