Introduction
In dogs, variations in the shape and diameter of principal bronchi are reported in bronchomalacia1 and bronchial compression by an enlarged left atrium.2,3 The causes of such changes, especially in those cases without evidence of bronchomalacia or left atrial enlargement, have been the subject of hypotheses,4,5 but important questions remain. Is there a relationship between bronchial morphology and thoracic conformation? When is bronchial narrowing caused by extraluminal pressure from adjacent structures (compression) versus inherent mural weakness (collapse)?
The answers to these and related questions can be important physiologically, epidemiologically during selection for breeding, diagnostically, and therapeutically. For example, the association between tracheal collapse and bronchomalacia6,7 could point to stenting collapsed bronchi as a promising avenue for therapy together with tracheal stenting; however, the influence, if any, of adjacent structures on bronchial morphology deserves further investigation to better understand how interventions might affect these structures. In particular, the role of body conformation (somatotype) has received limited attention,4,5,8 even though thoracic conformation varies across dog breeds.
The purpose of this study was to investigate relationships between principal bronchial dimensions and the following variables in dogs undergoing thoracic CT: thoracic dimensions, dimensions and positions of other intrathoracic structures, and clinical characteristics. The hypotheses of the study were that, in dogs, (1) there is variation in the dimensions and positions of certain intrathoracic structures on CT and this variation would be significantly associated with thoracic conformation, and (2) the way the left principal bronchus (LPB) interacts with adjacent structures would provide information that could be helpful in determining whether bronchial narrowing is due to external compression or intrinsic principal bronchial mural weakness.
Materials and Methods
The imaging archive at a tertiary referral facility (The Animal Medical Center) was searched for dogs that underwent CT and echocardiography within 90 days of CT during the following period covered by the archive at the time of searching: August 1, 2012, to November 15, 2017 (Figure 1). The protocol for performing thoracic CT involved placing dogs in sternal recumbency and imaging during ventilatory pause, without positive-pressure ventilation. Examples of visible CT abnormalities that could be expected to alter results and led to exclusion of cases included moderate or marked pleural effusion, moderate or marked pneumothorax, and space-occupying lesions that were of a size that, in the opinion of the reviewing investigator (EC), could influence the topography of intrathoracic structures. These abnormalities led to culling of a case, whether the investigator noted the abnormality on the radiographic report or the CT image. Mild or trace pleural effusion, mild and sparse nonspecific pulmonary infiltrates, and a single small pulmonary nodule were examples of abnormalities that did not disqualify a case from remaining in the study. Ninety-three studies were retained for further evaluation. Routine owner consent had been obtained at admission to the facility as permission for providing medical care to the animals.
Algorithm depicting the case selection process. DMVD = Degenerative mitral valve disease. dz = Disease.
Citation: Journal of the American Veterinary Medical Association 260, 13; 10.2460/javma.20.12.0716
Measurements were made on a single axial thoracic CT image for each case (reference image), with client name and other identifiers removed. For each dog, the image chosen to be the reference image was the one that was closest to showing the aorta immediately dorsal to the LPB (Figure 2). All measurements were made according to each dog’s dorsoventral (height measurements) and lateral (width measurements) dimensions in the reference image, irrespective of the dog’s degree of rotation in the image.
Example of a reference image of a dog, with measurements applied. For each dog, the image chosen to be the reference image was the one that was closest to showing the aorta immediately dorsal to the left principal bronchus (LPB). 1 = Thoracic height. 2 = Thoracic width. 3 = LPB height. 4 = LPB width. 5 = Right principal bronchus height. 6 = Right principal bronchus width. 7 = Aortic height. 8 = Aortic width. 9 = Aortic-LPB separation.
Citation: Journal of the American Veterinary Medical Association 260, 13; 10.2460/javma.20.12.0716
Prior to making measurements on cases, 2 investigators (EC and CW) evaluated 1 axial thoracic CT image from a dog that was not included in the study to establish a preliminary list of measurements to obtain on all study images. These measurements were then applied by the 2 investigators, independently of each other, to a trial set of CT reference images from 10 dogs that did not meet the criteria for the study group. The purpose was to perform a preliminary analysis and refine measurement parameters. The investigators then reviewed their findings together and established the final measurement parameters and definition of each measurement (Appendix). The investigators performed the measurements on the 93 cases and entered these data into separate spreadsheets, masked to each other’s process and results. Unblinding occurred after all measurements were complete, and no changes were made to data thereafter except to correct 8 errors of data hygiene (5 instances in which paired measurements were inverted, a different limit of precision used by the 2 investigators for 1 measurement, a data point in 1 column entered with a decimal error, and transposed data in 1 pair of columns).
Additionally, the reviewing investigator (EC) subjectively assessed the diameters of the left and right cranial and caudal lobar bronchi and their respective segmental bronchi by scrolling through sequential axial images, from cranial to caudal, for each case. Focal narrowing of a bronchus was defined as a decrease in LPB or right principal bronchus (RPB) diameter in the reference image with subsequent increase in the diameter of that PB distally. When focal bronchial narrowing was apparent, the reviewing investigator captured both the reference image and an additional image in the axial thoracic CT series at the level of the maximal diameter of the bronchus distal to the narrowing (ie, the flared segment representing return to expected diameter) and noted the distance between this image and the reference image. Both investigators measured the PB diameter (height) on the additional image in a masked fashion, as described above.
Information retrieved from medical records consisted of dog sex and breed; age, reproductive status, body weight (BW), and body condition score (BCS; assessed on a scale of 1 to 9 and in tertiles, as follows: 1 to 4, 5, and 6 to 9) at the time of CT; indication for performing CT; whether clinical signs of respiratory dysfunction were noted in the visit history or during the hospital visit for the CT; and echocardiographic findings within 90 days of CT (qualitative echocardiographic diagnosis and 2-D left atrial-to-aortic ratio [LA:Ao]). The investigators were masked to this information during evaluation of CT images.
For breed size, American Kennel Club designations were used.9 Cockapoo and Jack Russell Terrier were categorized as small breeds, and Labradoodle and pit bull–type dogs were categorized as large breeds. On the basis of subjective assessment of general breed characteristics, the reviewing investigator classified each dog’s breed as listed in the medical record as being either deep- or barrel-chested10 (or indeterminate for crossbreeds or when no breed was listed) and as typically either brachycephalic or not brachycephalic (or indeterminate for crossbreeds or when no breed was listed). The same investigator classified dogs as being predisposed or not predisposed to collapsing trachea on the basis of clinical experience. Dogs were considered predisposed to degenerative mitral valve disease (DMVD) if they were among the 9 most commonly reported breeds compiled cumulatively from 4 retrospective case series of dogs with this disease.11–14
Statistical methods
Descriptive statistics are reported as mean and SD given that the distribution of residuals was normal as determined by visual inspection followed by the Kolmogorov-Smirnoff test. Between-groups analyses of baseline variables were performed by use of ANOVA. Comparisons of proportions for categorical variables were made with the χ2 or Fisher exact test as appropriate. Date of study entry was defined as the day the thoracic CT was performed, and data obtained at this time were evaluated in baseline analyses.
Logistic regression was carried out by way of a generalized linear logistic model employing receiver operating characteristic curve analysis. Sensitivity analyses were carried out to determine whether a replication effect was present in the model where reviewer was allowed as confounding effect, found to be nonsignificant in all models, and subsequently removed from the model, allowing all analyses to be carried out with observations from both reviewers. All analyses were performed with statistical software (SAS version 9.4; SAS Institute Inc), and values of P < 0.05 were deemed significant.
Results
Dog characteristics
Characteristics of the 93 included dogs were summarized (Supplementary Table S1). The following characteristics were of note: mean ± SD age was 9.7 ± 3.5 years (median, 10 years; range, 1.5 to 16 years; interquartile [25th to 75th percentile] range [IQR], 8 to 12 years); there were 52 males (46 castrated) and 41 females (39 spayed); mean ± SD BW was 18.5 ± 15.7 kg (median, 10.4 kg; range, 1.77 to 82.2 kg; IQR, 6.13 to 29 kg); BCSs were 3/9 (n = 5), 4/9 (21), 5/9 (22), 6/9 (21), 7/9 (5), and 8/9 (5); breed size included toy or small (46), medium (7), large (19), giant (4), and uncertain (17); and breed thoracic conformation included barrel-chested (48), deep-chested (31), or crossbreed or indeterminate (14). Other characteristics of note included brachycephalic breed (yes, 17; no, 60; indeterminate, 16); collapsing trachea breed (yes, 36; no, 36; indeterminate, 21); DMVD breed (yes, 32; no, 42; indeterminate, 19); respiratory signs on day of CT (yes, 22; no, 71); and echocardiographic diagnoses (study within normal limits, 31), including DMVD American College of Veterinary Internal Medicine stage B1 (30) or B2 (11), premature ventricular complexes without echocardiographically apparent lesions (5), moderate or severe pulmonary hypertension (5), DMVD American College of Veterinary Internal Medicine stage C (2), and other (9). Typically deep- or barrel-chested breed conformations were not associated with BCS (neither on a scale of 1 to 9 [P = 0.475] nor by tertile [P = 0.909]).
Imaging: thoracic height-to-width ratio
Examples of CT reference images from dogs with the lowest (0.39), median (0.85), and highest (1.11) thoracic height-to-width ratio (H:W) were obtained (Figure 3). Descriptive statistics for thoracic H:W and associations between thoracic H:W and other variables were tabulated (Tables 1 and 2; Supplementary Tables S2 and S3). The area under the receiver operating characteristic curve for the sensitivity and specificity of thoracic H:W for predicting focal LPB narrowing on the reference image was 0.7604 (Figure 4). The maximal Youden index value was 0.42 at a thoracic H:W of 0.877 (92% sensitivity and 50% specificity). A thoracic H:W > 0.875 was associated with a 90% sensitivity and 50% specificity for detecting focal narrowing of the LPB in the reference image, and a thoracic H:W < 0.735 was associated with a 46% sensitivity and 90% specificity for detecting focal narrowing of the LPB in the reference image.
Computed tomography reference images from dogs with the lowest (A; 0.39), median (B; 0.85), and highest (C; 1.11) thoracic height-to-width ratio (H:W).
Citation: Journal of the American Veterinary Medical Association 260, 13; 10.2460/javma.20.12.0716
Mean and median measurements obtained from CT reference images of dogs (n = 93).
| Variable | Mean | SD | Median | Range | IQR |
|---|---|---|---|---|---|
| Thoracic height (cm) | 9.259 | 2.852 | 8.539 | 2.901–15.345 | 6.894–11.905 |
| Thoracic width (cm) | 11.098 | 3.348 | 10.084 | 5.726–19.495 | 8.113–14.031 |
| Thoracic H:W | 0.840 | 0.140 | 0.850 | 0.39–1.11 | 0.77–0.95 |
| LPB height (cm) | 0.995 | 0.510 | 0.907 | 0.115–2.27 | 0.573–1.397 |
| LPB width (cm) | 1.414 | 0.523 | 1.407 | 0.359–2.731 | 0.984–1.833 |
| LPB H:W | 0.679 | 0.176 | 0.686 | 0.263–1.086 | 0.555–0.806 |
| RPB height (cm) | 1.246 | 0.520 | 1.178 | 0.333–2.523 | 0.8129–1.615 |
| RPB width (cm) | 1.270 | 0.471 | 1.175 | 0.456–2.37 | 0.908–1.64 |
| RPB H:W | 0.970 | 0.140 | 0.979 | 0.529–1.426 | 0.888–1.063 |
| Ao-LPB separation (cm) | 0.120 | 0.190 | 0.000 | 0–0.81 | 0–0.16 |
Ao = Aorta. H:W = Height-to-width ratio. IQR = Interquartile (25th to 75th percentile) range. LPB = Left principal bronchus. RPB = Right principal bronchus.
Results of regression analyses concerning measurements obtained from CT reference images of the dogs in Table 1.
| Dependent variable | Independent variable | Estimate | SE | P value |
|---|---|---|---|---|
| Thoracic H:W | Age (y) | –0.005 | 0.003 | 0.080 |
| Weight (kg) | –0.001 | 0.001 | 0.395 | |
| BCS | –0.003 | 0.012 | 0.810 | |
| LPB H:W | 0.342 | 0.051 | < 0.001 | |
| LPB maximum height caudal to focal narrowing | –0.134 | 0.026 | < 0.001 | |
| RPB H:W | 0.375 | 0.066 | < 0.001 | |
| RPB maximum height caudal to focal narrowing (cm) | –0.039 | 0.037 | 0.285 | |
| Ao-LPB separation (cm) | 0.199 | 0.051 | < 0.001 | |
| LA:Ao | 0.036 | 0.034 | 0.280 | |
| LPB H:W | RPB H:W | 0.532 | 0.084 | < 0.001 |
| Ao-LPB separation (cm) | 0.424 | 0.061 | < 0.001 | |
| LPB maximum height caudal to focal narrowing (cm) | –0.269 | 0.034 | < 0.001 | |
| LA:Ao | –0.163 | 0.042 | < 0.001 | |
| LPB height | Ao-LPB separation (cm) | 1.058 | 0.181 | < 0.001 |
| RPB H:W | Ao-LPB separation (cm) | 0.164 | 0.053 | 0.002 |
| LA:Ao | Weight (kg) | –5.383 | 3.832 | 0.162 |
| Thoracic H:W | 0.036 | 0.034 | 0.280 |
BCS = Body condition score. LA:Ao = Echocardiographic left atrial-to-aortic ratio.
See Table 1 for remainder of key.
Receiver operating characteristic (ROC) curve demonstrating the relationship between thoracic H:W and focal LPB narrowing on the reference image.
Citation: Journal of the American Veterinary Medical Association 260, 13; 10.2460/javma.20.12.0716
Imaging: LPB height and H:W
Examples of CT reference images from dogs with the lowest (0.26), median (0.69), and highest (1.0) LPB H:W were obtained (Figure 5). Only 1 (EC) or 2 (CW) dogs had an LPB H:W > 1, indicating a tall, ovoid LPB transverse section, whereas 91 to 92 dogs had an LPB H:W < 1 (range, 0.263 to 0.984), indicating a flatter ovoid LPB transverse section (Tables 1 and 2; Supplementary Tables S2 and S3). The LPB H:W was associated with thoracic H:W and with aortic-LPB separation independently. For a constant aortic-LPB separation, each 1-unit increase in thoracic H:W was associated with an increase in LPB H:W of 0.438 (P < 0.001), and for a constant thoracic H:W, each 1-cm increase in aortic-LPB separation was associated with an increase in LPB H:W of 0.337 (P < 0.001).
Computed tomography reference images from dogs with the lowest (A; 0.26), median (B; 0.69), and highest (C; 1.0) LPB H:W.
Citation: Journal of the American Veterinary Medical Association 260, 13; 10.2460/javma.20.12.0716
To explore the possible role of LPB obliquity altering the LPB H:W, comparisons were made between vertical aortic diameter and variables of interest. Vertical aortic diameter was significantly associated with BW, LPB height, RPB height, LPB H:W, and RPB H:W (all P < 0.001) but not with thoracic H:W (P = 0.172).
Imaging: RPB height and H:W
There were 35 (EC) to 46 (CW) dogs with an RPB H:W > 1 and 46 (CW) to 57 (EC) dogs with an RPB H:W < 1 (P = 0.241; Tables 1 and 2; Supplementary Tables S2 and S3).
Imaging: aortic-LPB separation
Examples of CT reference images from dogs with 0 cm (direct contact between the aorta and LPB) and the highest aortic-LPB separation were obtained (Figure 6), and results for aortic-LPB separation were tabulated (Tables 1 and 2; Supplementary Tables S2 and S3). Fifty-three (EC) or 63 (CW) dogs had an aortic-LPB separation of 0 cm, indicating direct contact between the 2 structures. For each unit increase in thoracic H:W, aortic-LPB separation increased by a mean ± SD of 0.199 ± 0.051 cm (P < 0.001).
Computed tomography reference images from dogs with 0 cm (A; direct contact between aorta and LPB) and the highest (B) aortic-LPB separation (arrows).
Citation: Journal of the American Veterinary Medical Association 260, 13; 10.2460/javma.20.12.0716
Imaging: focal bronchial narrowing on the reference image
In 25 dogs, the LPB height was smaller on the reference image than the LPB height observed on axial CT images caudal to the reference image, indicating focal LPB narrowing on the reference image (Tables 1 and 2; Supplementary Tables S2 and S3). All 25 showed contact between the ventral aspect of the LPB and the dorsal aspect of the cardiac silhouette. In these dogs, the maximal LPB height occurred on an axial image that was a median of 9 mm caudal to the reference image (range, 3 to 25 mm). The greatest LPB height caudal to the reference image for these dogs was a mean ± SD of 0.712 ± 0.332 cm (median, 0.621 cm; range, 0.261 to 1.6 cm; IQR, 0.46 to 0.873 cm). This was a mean of 0.168 ± 0.129 cm (median, 0.118 cm; range, 0.01 to 0.583 cm; IQR, 0.082 to 0.237 cm) smaller on the reference image than the greatest LPB height farther distally, representing a median difference of 22.7% (range, 2.1% to 67.3%).
Breeds of dogs with focal LPB narrowing were Yorkshire Terrier (n = 5; P = 0.001); Beagle cross, Cavalier King Charles Spaniel, Chihuahua, Golden Retriever, Havanese or Havanese cross, Maltese (2 each); and Labradoodle, Newfoundland, Norwich Terrier, Old English Mastiff, Pug, Shih Tzu, Soft-Coated Wheaten Terrier, and terrier cross (1 each).
Twenty-two (EC) to 25 (CW) of 25 dogs with focal LPB narrowing had direct contact between the dorsal aspect of the LPB and the ventral aspect of the aorta (ie, aortic-LPB separation = 0 cm). Furthermore, 15 (both EC and CW) of the 22 to 25 dogs with focal LPB narrowing and no aortic-LPB separation had no overlap or contact between the aorta and thoracic vertebra dorsal to it (ie, aortic-vertebral overlap absent; Figures 7–9). Their mean thoracic H:W was 0.811 ± 0.025, which was not significantly (P = 0.178) different from the mean thoracic H:W of the remaining dogs (0.848 ± 0.011). In the other 7 to 10 dogs with focal LPB narrowing and aortic-LPB contact and aortic-vertebral overlap present, the mean thoracic H:W was 0.625 ± 0.029, which was significantly (P < 0.001) different from the mean thoracic H:W of the remaining dogs (0.863 ± 0.009).
Computed tomography images showing focal LPB compression. The reference image (left panel) shows that the LPB is compressed in the dorsoventral dimension (arrow). The right panel is an axial image that was taken 12 mm caudal to the reference image, showing that LPB compression is no longer apparent (arrow).
Citation: Journal of the American Veterinary Medical Association 260, 13; 10.2460/javma.20.12.0716
Computed tomography images showing types of focal LPB compression. A—Aortic constrictive effect; there is narrowing of the LPB (bracket; confirmed by widening of the LPB in axial images caudal to this location [not shown]) and contact between the ventral aspect of the LPB and the heart (arrow) and between the dorsal aspect of the LPB and the ventral aspect of the aorta (arrowhead), but no contact or overlap between the dorsal aspect of the aorta and ventral aspect of the thoracic vertebral body (asterisk). B—Cumulative compression effect, as for Figure 7 panel A, but the ventral border of the thoracic vertebra is at or beyond the level of the dorsal aspect of the aorta (double asterisk), creating a stacking effect of structures associated with LPB compression. See Figure 7 for remainder of key.
Citation: Journal of the American Veterinary Medical Association 260, 13; 10.2460/javma.20.12.0716
Computed tomography reference images from dogs without (A; horizontal line contacts neither the aorta nor the vertebral body) and with (B; horizontal line contacts both the aorta and vertebral body) aortic-vertebral overlap.
Citation: Journal of the American Veterinary Medical Association 260, 13; 10.2460/javma.20.12.0716
In 5 dogs, the RPB height was smaller on the reference image than the RPB height observed on axial CT images caudal to the reference image, indicating focal RPB narrowing on the reference image. This represented significantly fewer dogs than the 25 with focal LPB narrowing on the reference image (P = 0.001). In these 5 dogs, the maximal RPB height occurred on an axial image that was a median 9 mm caudal to the reference image (range, 3 to 18 mm). The greatest RPB height caudal to the reference image for these dogs was a mean of 1.06 ± 0.6 cm (median, 0.869 cm; range, 0.412 to 1.96 cm; IQR, 0.541 to 1.753 cm). This was a mean of 0.113 ± 0.187 cm (median, 0.078 cm; range, –0.131 to 0.412 cm; IQR, –0.054 to 0.32 cm) smaller on the reference image than the greatest RPB height farther caudally, representing a median difference of 22.7% (range, –14.5% to 49.3%). In 2 dogs, direct contact between the dorsal body wall and both the LPB and RPB was apparent in the reference image, whereas in the other 3 dogs, the dorsal RPB was only in contact with lung. In these 3 dogs, the amount of RPB narrowing was 6.4% or 19.6%, 3.6% or 12.6%, and 4.4% or 15.6% (EC or CW, respectively). All 5 cases of focal RPB narrowing also had focal LPB narrowing. The 5 dog breeds with focal RPB narrowing were Chihuahua (n = 2) and Cavalier King Charles Spaniel, Golden Retriever, and Labradoodle (1 each).
Imaging: aortic-vertebral overlap
Data on aortic-vertebral overlap were summarized (Tables 1 and 2; Supplementary Tables S2 and S3). Aortic-vertebral overlap occurred in 15 (EC) to 16 (CW) dogs. Dogs with aortic-LPB separation > 0 cm and absence of aortic-vertebral overlap had a significantly (P < 0.001) higher thoracic H:W (mean, 0.898 ± 0.016) than did dogs with aortic-LPB separation > 0 cm and presence of aortic overlap (0.811 ± 0.012). None of the 1 (EC) or 2 (CW) dogs with LPB H:W ≥ 1 had aortic-vertebral overlap.
Imaging: spondylosis, pectus excavatum
Spondylosis was present in 10 dogs according to one observer and 14 according to the other (Tables 1 and 2; Supplementary Tables S2 and S3). Pectus excavatum was present in 4 dogs according to both observers.
Imaging: relationship with echocardiographic LA:Ao
Mean LA:Ao was 1.28 ± 0.3 (median, 1.21; range, 0.77 to 2.66; IQR, 1.14 to 1.33; Tables 1 and 2; Supplementary Tables S2 and S3). Coughing was noted in the history and physical examination of 2 of 93 dogs on the day of CT; their LA:Aos were 1.08 and 1.93. There were significant inverse relationships between LA:Ao and LPB: for each unit increase in LA:Ao, LPB height decreased by a mean of 0.31 ± 0.123 cm and LPB H:W decreased by a mean of 0.163 ± 0.042. No such relationship was apparent between LA:Ao and RPB height (P = 0.209) or RPB H:W (P = 0.276). There was an association between thoracic H:W and focal LPB compression that was independent of LA:Ao according to both univariate (–8.8638) and multivariable (–9.9651) models.
Fourteen dogs had an LA:Ao > 1.5 (mean, 1.76 ± 0.34; range, 1.51 to 2.66; IQR, 1.61 to 1.95), which was considered mild (LA:Ao = 1.5 to 1.9; n = 8), moderate (LA:Ao = 1.91 to 2.2; 3), or severe (LA:Ao > 2.2; 3). The median aortic-LPB separation for these 3 groups was 0.12 cm (range, 0 to 0.172 cm), 0 cm (range, 0 to 0.46 cm), and 0 cm (range, 0 to 0.46 cm), respectively. There was no significant difference between these 14 dogs and the 79 dogs with LA:Ao ≤ 1.5 (mean LA:Ao, 1.18 ± 0.14; range, 0.77 to 1.47; IQR, 1.1 to 1.25) in terms of BW (P = 0.069), BCS (P = 0.6), aortic-LPB separation (P = 0.38), or aortic-vertebral overlap (P = 0.71).
Inter-rater variability
A Bland-Altman inter-rater analysis15 was carried out to visually assess agreement between the 2 raters for variables thoracic height, thoracic width, thoracic H:W, aortic height, aortic width, LPB height, LPB width, LPB H:W, RPB height, RPB H:W, and aortic-LPB separation (Figure 10). Excellent agreement existed between raters, where rarely did any of the raters’ assignment of a value fall outside the 95% CI bands generated by the analysis (eg, 1/93 evaluations for both thoracic height and thoracic width).
Bland-Altman plots of inter-rater variability as assessed by 2 observers for thoracic H:W (A), LPB H:W (B), and aortic-LPB separation. See Figure 7 for remainder of key.
Citation: Journal of the American Veterinary Medical Association 260, 13; 10.2460/javma.20.12.0716
Discussion
This study provided new insights into spatial relationships among intrathoracic anatomic structures in dogs. Several important findings emerged in this study of a population of dogs with neither thoracic conformation nor airway disease as a selection criterion. First, results showed substantial intraspecies heterogeneity of dimensions, which was not identified originally in investigations of thoracic morphometry.16,17 For example, across this study population, thoracic H:W varied by 2.58- to 2.82-fold (2 observers), independent of age, sex, breed, BW, or BCS. Similarly, LPB H:W varied by 3.84- to 4.13-fold and RPB H:W varied by 2.31- to 2.48-fold. These ranges were greater than those reported for thoracic H:W previously and for tracheal H:W,8 and they emphasized that generalizations about dogs’ relative thoracic and principal bronchial dimensions cannot be applied uniformly within the species.
Second, certain relationships that were considered intuitive were confirmed and others that had been demonstrated in earlier work were shown to be repeatable. Thoracic H:W was significantly associated with category of breed conformation: dogs of deep-chested breeds had a higher thoracic H:W than did barrel-chested or indeterminate breeds. However, thoracic H:W was not different by breed size category; as shown by Uehara et al,8 smaller dogs were not more likely to have a lower thoracic H:W, contrary to what might have been suspected anecdotally. A lower thoracic H:W was associated with more aortic-vertebral overlap, as might be expected due to reduced space in the dorsoventral dimension with decreasing thoracic H:W and as noted previously.8 There was evidence to associate LA enlargement with LPB narrowing: the higher median LA:Ao in the 46 dogs with LPB H:W less than the median and the inverse relationship between LPB H:W and LA:Ao were expected if LA enlargement is considered to contribute to LPB compression.3 However, the small magnitude of the latter effect in this relationship suggests limited clinical importance, if any. Indeed, dogs with LA:Ao above the reference interval had both aortic-LPB separation and aortic-vertebral overlap that were not different from those of dogs with LA:Ao in the reference interval. This suggested either statistical underpowering or that LA enlargement did not displace the LPB dorsally to a significant degree. Furthermore, the association between thoracic H:W and focal LPB narrowing was independent of LA:Ao, which provided additional support for a limited role of LA enlargement in focal LPB narrowing. For aortic-LPB separation, lower values were associated with lower LPB H:W and RPB H:W and a higher occurrence of aortic-vertebral overlap, which were expected on the basis of the association between thoracic H:W and both LPB H:W and aortic-vertebral overlap. Thoracic H:W was significantly, negatively associated with the presence of pectus excavatum, consistent with the more dorsal position of the sternum in these cases. Similarly, aortic-vertebral overlap was associated with pectus excavatum, which is logical if pectus excavatum displaces the heart, aorta, and bronchi dorsally. Thoracic H:W was not associated with vertical aortic diameter; intuitively, larger aortic dimensions would be expected in larger dogs, and thoracic H:W was not associated with BW. The lack of association between LA:Ao and BW or thoracic H:W was expected. The presence of cough at the time of CT in a very small number of dogs (2/93; one with a normal LA:Ao and the other with LA:Ao > the reference interval) supported the argument that LPB compression does not cause coughing in dogs with DMVD.3
Third, new relationships among intrathoracic structures were demonstrated. A breed-specific finding was the higher prevalence of focal LPB narrowing in Yorkshire Terriers, compared with other breeds.
Although coughing was not associated with principal bronchial narrowing due to left atrial enlargement in a previous report,3 the prevalence of coughing in dogs with bronchial narrowing due to other causes has not been examined in detail. The present study identified coughing in only 2 dogs, although 25 dogs had focal LPB compression. This suggested a broader conclusion that bronchial compression in dogs is not associated with coughing, whether bronchial compression is caused by an enlarged left atrium or by other adjacent structures. However, retrospective data collection, selection bias for CT, and lack of controlling for treatment were substantial limitations.
The associations between LPB H:W and other measurements suggested either a compressive effect on the LPB (eg, aortic-LPB separation or aortic-vertebral overlap) or an independent relationship (eg, thoracic H:W). Similar associations existed for RPB H:W, and many measurements (thoracic H:W, aortic-LPB separation, or aortic-vertebral overlap) and their lesser magnitude suggested they might be driven by the aforementioned LPB H:W relationships. The higher prevalence of focal LPB compression (25/93) versus focal RPB compression (5/93) was also a new finding. In 2 of the 5 dogs with focal RPB compression, direct contact between the RPB and the dorsal body wall suggested a compressive effect, which was also apparent on the LPB on the same reference image for these 2 dogs, whereas in the other 3 dogs with focal RPB compression, no mechanism could be inferred from CT images. Possible explanations could include bronchomalacia, greater inter-rater variability, and artifact (eg, obliquity of the RPB). In this study population, thoracic H:W was associated with focal LPB narrowing: on the basis of an area under the receiver operating characteristic curve of 0.76, the sensitivity and specificity for predicting focal LPB narrowing based on thoracic H:W were categorized as good (0.7 to 0.8), and a thoracic H:W < 0.73507 indicated a 90% likelihood of focal LPB narrowing, whereas a thoracic H:W > 0.8749 indicated a 90% likelihood of the absence of focal LPB narrowing.
Focal LPB narrowing is well recognized in dogs,3,5 but causative mechanisms are incompletely understood. It has been attributed to bronchiectasis,1,6,7,18,19 abnormal negative airway pressure coupled with the length and few branch points of the LPB in brachycephalic dogs,5 and compression from left atrial enlargement.3,20 The present study, which was not focused on dogs being evaluated for signs related to the respiratory system, provided evidence for 2 other mechanisms contributing to bronchial narrowing: a cumulative compressive effect in which the vertebral column, aorta, LPB, and heart are approximately aligned in a dorsoventral axis and an isolated aortic constrictive effect. The first mechanism, extraluminal LPB compression due to a cumulative dorsoventral compressive effect, was supported by 22 (EC) to 25 (CW) of the dogs with focal LPB narrowing having no aortic-LPB separation, aortic-vertebral overlap being associated with a lower thoracic H:W and a lower LPB H:W, a greater number of dogs with aortic-vertebral overlap having focal LPB narrowing, and a lower thoracic H:W in dogs with focal LPB narrowing but not in dogs with focal RPB narrowing. Consistent with a cumulative compressive effect, thoracic H:W was markedly lower in dogs with focal LPB narrowing and lack of aortic-LPB separation and aortic-vertebral overlap (0.625 ± 0.029 compared to 0.863 ± 0.009). A lower LPB H:W in dogs without aortic-LPB separation and the larger number of dogs with LPB H:W < 1 (89 [EC] or 90 [CW]) compared to those with RPB H:W < 1 (46 [EC] or 57 [CW]) could be explained by LPB compression from either a cumulative compressive effect or by the second putative mechanism, extraluminal constriction from an aortic constrictive effect. Findings in support of this second mechanism included the significantly lower occurrence of aortic-vertebral overlap in dogs with aortic-LPB separation = 0 cm; the small magnitude of effect in the association between thoracic H:W and aortic-LPB separation (0.199), indicating a 1-unit increase in thoracic H:W would increase aortic-LPB separation by only 0.199 cm; and the lack of association between aortic-vertebral overlap and LA:Ao, suggesting that LA enlargement did not displace the aorta dorsally. Most importantly, in contrast to dogs with the same findings but aortic-vertebral overlap present, thoracic H:W was not significantly different in dogs with focal LPB narrowing and lack of aortic-LPB separation and no aortic-vertebral overlap (mean ± SD, 0.811 ± 0.025) compared to the remaining counterparts (0.848 ± 0.011). Thus, the focal LPB narrowing in these dogs would not be attributed to cumulative dorsoventral compression because lack of contact or overlap between the aorta and dorsal body wall (vertebra) would preclude this stacking effect. Rather, the contact between the aorta and LPB seen in the reference image when the aorta was directly dorsal to the LPB in the same axial plane as the focal LPB narrowing made an aortic constrictive effect plausible. Since 7-10 of 25 dogs with focal LPB narrowing showed changes suggesting a cumulative compressive effect and 15 of 25 dogs with focal LPB narrowing showed changes consistent with a constrictive effect, both possible mechanisms should be taken into account when considering the origin of LPB narrowing in dogs. However, these compressive effects would solely explain neither the high prevalence of LPB H:W < 1 (90/93 [EC] or 89/93 [CW]) nor the association of LPB H:W with thoracic H:W independently of aortic-LPB separation.
Fourth, certain relationships were statistically significant but of limited biological significance. For example, for each unit thoracic H:W increased, LPB H:W increased 0.342 units and RPB H:W increased 0.375 units, but an increase of a full unit of thoracic H:W did not occur in this population, and the corresponding LPB H:W and RPB H:W changes would be expected to be of limited importance.
Finally, certain relationships that were historically thought to exist were not significant in this study population. Thoracic H:W was not associated with brachycephalic breed, which was counter to anecdotal beliefs. This finding may be explained partly or entirely by the lack of a uniform designation of all currently recognized breeds as brachycephalic, mesaticephalic or normocephalic, or dolichocephalic; the range of body conformations in brachycephalic breeds (eg, deep-chested Boxers vs barrel-chested Pugs); or other variables. The lack of association between brachycephalic breed and aortic-LPB separation was also surprising, since brachycephalic dogs have sometimes been thought to have a greater likelihood of a cumulative intrathoracic dorsoventral compressive effect. However, this finding agreed with the lack of association between thoracic H:W and brachycephalic breed and indicated that thoracic conformation proper, and not brachycephalic breed as a surrogate, is associated with aortic-LPB separation.
The results of this study added to existing knowledge that bronchomalacia is a cause of airway collapse in dogs undergoing diagnostic evaluation due to clinical signs involving the respiratory system.18,19 Since the subjects in the present study were not selected from a pool of patients with respiratory signs, these results represented subclinical findings that could be earlier, lesser abnormalities of bronchial morphology or breed-specific variants. The findings of the present study showed that in a population of dogs without clinical signs referable to the respiratory system, 27% had evidence of focal LPB narrowing, compared to 5% with focal RPB narrowing. The findings suggested that this narrowing occurs due to a compressive effect from structures surrounding the LPB. However, 5 dogs had focal RPB narrowing, and all 5 concurrently had focal LPB narrowing. This finding suggested that PB narrowing may occur in at least 2 ways: one that affects both principal bronchi, which may be a primary bronchial disorder such as bronchiectasis or bronchomalacia, and another that disproportionately affects the LPB and is associated with compression, either cumulatively or via a constrictive effect. These results raised the possibility that, over time, subclinical compression of cartilaginous structures in the respiratory system could lead to material fatigue in these airways such that the static compressions become dynamic and eventually lead to clinical signs during respiration.
Other possibilities could have explained the present findings but were considered less likely. First, the LPB could be imaged more obliquely, or less, in individual dogs depending on its bifurcation angle with the trachea, which would alter the LPB H:W. This possibility was considered less likely because replacing LPB H:W with vertical aortic diameter preserved significant relationships with LPB height, RPB height, LPB H:W, and RPB H:W and because obliquity would not explain the focal LPB narrowing followed by a return to a wider diameter distally in the LPB, as observed in 25 dogs. Second, it is possible that LPB narrowing occurred due to intramural weakness and not compression in these cases and that results suggesting extraluminal compression were coincidental. This was unlikely because the associations found here were statistically significant. Third, forced expiration can cause dynamic narrowing of the bronchi in humans and dogs,21 but the dogs in this study did not undergo forced expiration; rather, the CT images were obtained during brief apnea without positive-pressure ventilation. Nevertheless, mild changes in bronchial dimensions can be expected during intermittent spontaneous respiration under anesthesia for CT, and the retrospective nature of the study precluded definitive elimination of this possibility. The repeatable location of focal narrowing in the reference image, the greater occurrence of LPB versus RPB narrowing, and the other associations described above favored extraluminal compression as a more influential variable than respiratory cycle in focal LPB narrowing.
Supplementary Materials
Supplementary materials are posted online at the journal website: avmajournals.avma.org
Acknowledgments
No third-party funding or support was received in connection with this study or the writing or publication of the manuscript. The authors declare that there were no conflicts of interest.
The authors thank the Oncology, Cardiology, and Internal Medicine Services of The Animal Medical Center for case material used in this study and Katie Lashbrook, LVT, for medical record searching.
References
- 1. ↑
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):247–254. doi:10.1111/jvim.13809
- 2. ↑
Dengate A, Culvenor JA, Graham K, Braddock JA, Churcher RK. Bronchial stent placement in a dog with bronchomalacia and left atrial enlargement. J Small Anim Pract. 2014;55(4):225–228. doi:10.1111/jsap.12183
- 3. ↑
Singh MK, Johnson LR, Kittleson MD, Pollard RE. Bronchomalacia in dogs with myxomatous mitral valve degeneration. J Vet Intern Med. 2012;26(2):312–319. doi:10.1111/j.1939-1676.2012.00887.x
- 4. ↑
Weisse C. Insights in tracheobronchial stenting and a theory of bronchial compression. J Small Anim Pract. 2014;55(4):181–184. doi:10.1111/jsap.12209
- 5. ↑
De Lorenzi D, Bertoncello D, Drigo M. Bronchial abnormalities found in a consecutive series of 40 brachycephalic dogs. J Am Vet Med Assoc. 2009;235(7):835–840. doi:10.2460/javma.235.7.835
- 6. ↑
Marolf A, Blaik M, Specht A. A retrospective study of the relationship between tracheal collapse and bronchiectasis in dogs. Vet Radiol Ultrasound. 2007;48(3):199–203. doi:10.1111/j.1740-8261.2007.00229.x
- 7. ↑
Johnson LR, Pollard RE. Tracheal collapse and bronchomalacia in dogs: 58 cases (7/2001–1/2008). J Vet Intern Med. 2010;24(2):298–305. doi:10.1111/j.1939-1676.2009.0451.x
- 8. ↑
Uehara T, Orito K, Fujii Y. CT-based anatomical features of large airway and heart volume in dogs of different body size. Vet J. 2019;246:21–26. doi:10.1016/j.tvjl.2019.01.014
- 9. ↑
Breed size designations. American Kennel Club. Accessed September 23, 2020. https://www.akc.org/dog-breeds/
- 10. ↑
Definitions of thoracic depth and barrel conformation. American Kennel Club. Accessed September 23, 2020. https://www.akc.org/about/glossary/
- 11. ↑
Borgarelli M, Crosara S, Lamb K, et al. Survival characteristics and prognostic variables of dogs with preclinical chronic degenerative mitral valve disease attributable to myxomatous degeneration. J Vet Intern Med. 2012;26(1):69–75. doi:10.1111/j.1939-1676.2011.00860.x
- 12.
Boswood A, Häggström J, Gordon SG, et al. Effect of pimobendan in dogs with preclinical myxomatous mitral valve disease and cardiomegaly: the EPIC study—a randomized clinical trial. J Vet Intern Med. 2016;30(6):1765–1779. doi:10.1111/jvim.14586
- 13.
Chetboul V, Pouchelon JL, Menard J, et al. Short-term efficacy and safety of torasemide and furosemide in 366 dogs with degenerative mitral valve disease: the TEST study. J Vet Intern Med. 2017;31(6):1629–1642. doi:10.1111/jvim.14841
- 14. ↑
Mattin MJ, Boswood A, Church DB, Brodbelt DC. Prognostic factors in dogs with presumed degenerative mitral valve disease attending primary-care veterinary practices in the United Kingdom. J Vet Intern Med. 2019;33(2):432–444. doi:10.1111/jvim.15251
- 15. ↑
Bland JM, Altman DG. Measurement in medicine: the analysis of method comparison studies. Statistician. 1983;32(3):307–317. doi:10.2307/2987937
- 16. ↑
Schaible RH, Ziech J, Glickman NW, Schellenberg D, Yi Q, Glickman LT. Predisposition to gastric dilatation-volvulus in relation to genetics of thoracic conformation in Irish Setters. J Am Anim Hosp Assoc. 1997;33(5):379–383. doi:10.5326/15473317-33-5-379
- 17. ↑
De Rycke LM, Gielen IM, Simoens PJ, van Bree H. Computed tomography and cross-sectional anatomy of the thorax in clinically normal dogs. Am J Vet Res. 2005;66(3):512–524. doi:10.2460/ajvr.2005.66.512
- 18. ↑
Johnson LR, Singh MK, Pollard RE. Agreement among radiographs, fluoroscopy and bronchoscopy in documentation of airway collapse in dogs. J Vet Intern Med. 2015;29(6):1619–1626. doi:10.1111/jvim.13612
- 19. ↑
Bottero E, Bellino C, De Lorenzi D, et al. Clinical evaluation and endoscopic classification of bronchomalacia in dogs. J Vet Intern Med. 2013;27(4):840–846. doi:10.1111/jvim.12096
- 20. ↑
Ferasin L, Crews L, Biller DS, Lamb KE, Borgarelli M. Risk factors for coughing in dogs with naturally acquired myxomatous mitral valve disease. J Vet Intern Med. 2013;27(2):286–292. doi:10.1111/jvim.12039
- 21. ↑
Kim H, Kim YJ, Lee H, et al. Computed tomographic and radiographic bronchial collapse may be a normal characteristic of forced expiration in dogs. Vet Radiol Ultrasound. 2018;59(5):551–563. doi:10.1111/vru.12625
Appendix
Measurements and their definitions
| Thoracic height | Distance (cm) from dorsalmost aspect of sternebra to ventralmost aspect of vertebra. |
| Thoracic width | Distance (cm) at widest level of thorax, perpendicular to thoracic height. |
| Thoracic H:W | Thoracic height-to-thoracic width ratio. |
| Spondylosis | In reference to ventral surface of thoracic vertebra in the reference image. Noted as absent or present. |
| Vertical aortic diameter | Distance (cm) from dorsalmost to ventralmost aspects of the descending aorta, including the walls of the aorta. |
| LPB height | Internal distance (cm) from dorsalmost to ventralmost aspect of the LPB, excluding bronchus wall. |
| LPB width | Internal distance (cm) from one lateral wall of the LPB to the other, perpendicular to the LPB height, and excluding the walls of the LPB. In cases of dropout (incomplete bronchial circumference), end of measurement was estimated by visual inspection. |
| LPB H:W | LPB height-to-width ratio. |
| Aortic-LPB separation | Extramural distance (cm) between ventral aspect of aorta and dorsal aspect of LPB. Excluded walls of the aorta and bronchus; if < 1 mm, reported as 0. |
| LPB height at maximum diameter caudal to narrowing | If initial review of CT subjectively indicated a narrowing of the LPB in the reference image, an additional image was measured for that animal. The additional image was also axial, caudal to the reference image, and chosen for showing the maximum LPB height caudal to the reference image. The distance (mm) between the reference image and this additional image was noted as the location of LPB height at maximum diameter caudal to narrowing. |
| Aortic-vertebral overlap | Whether the dorsal aspect of the aorta was more dorsal than the ventral aspect of the vertebral body. Noted as absent or present. |
| RPB height | Internal distance (cm) from dorsalmost to ventralmost aspect of RPB, excluding bronchus wall. |
| RPB width | Internal distance from one lateral wall of the RPB to the other, perpendicular to the RPB height, and excluding the walls of the RPB, in cm; in cases of dropout (incomplete bronchial circumference), end of measurement was estimated by visual inspection. |
| RPB H:W | RPB height-to-width ratio. |
| RPB height at maximum diameter caudal to narrowing | If initial review of CT subjectively indicated a narrowing of the RPB in the reference image, an additional image was measured for that animal. The additional image was also axial, caudal to the reference image, and chosen for showing the maximum RPB height caudal to the reference image. The distance (mm) between the reference image and this additional image was noted as the location of RPB height at maximum diameter caudal to narrowing. |
| Pectus excavatum | Noted as present if the ventralmost surface of the sternebra was more dorsal than the dorsalmost surface of the 2 adjacent ribs. |
LPB = Left principal bronchus. RPB = Right principal bronchus.
