Further characterization of computed tomographic and clinical features for staging and prognosis of idiopathic pulmonary fibrosis in West Highland white terriers

Idiopathic pulmonary fibrosis is an interstitial lung disease of unknown etiology resulting in progressive interstitial fibrosis, with a known predilection in West Highland white terriers. In humans, computed tomography (CT) is a standard method for providing diagnostic and prognostic information, and plays a major role in the idiopathic pulmonary fibrosis staging process. Objectives of this retrospective, analytical, cross-sectional study were to establish descriptive criteria for reporting CT findings and test correlations among CT, clinical findings and survival time in West Highland white terriers with idiopathic pulmonary fibrosis. Inclusion criteria for affected West Highland white terriers were a diagnosis of idiopathic pulmonary fibrosis and available CT, bronchoscopy, bronchoalveolar lavage, echocardiography, and routine blood analysis findings. Clinically normal West Highland white terriers were recruited for the control group. Survival times were recorded for affected dogs. The main CT lung pattern and clinical data were blindly and separately graded as mild, moderate, or severe. Twenty-one West Highland white terriers with idiopathic pulmonary fibrosis and 11 control West Highland white terriers were included. The severity of pulmonary CT findings was positively correlated with severity of clinical signs (ρ = 0.48, P = 0.029) and negatively associated with survival time after diagnosis (ρ = -0.56, P = 0.025). Affected dogs had higher lung attenuation (median: -563 Hounsfield Units (HU)) than control dogs (median: -761 HU), (P < 0.001). The most common CT characteristics were ground-glass pattern (16/21) considered as a mild degree of severity, and focal reticular and mosaic ground-glass patterns (10/21) considered as a moderate degree of severity. Findings supported the use of thoracic CT as a method for characterizing idiopathic pulmonary fibrosis in West Highland white terriers and providing prognostic information for owners.


INTRODUCTION
Idiopathic pulmonary fibrosis has been described as an interstitial lung disease of unknown etiology resulting in progressive and fatal interstitial fibrosis. 1,2 Several factors are known to trigger interstitial lung diseases such as infectious agents, toxin exposure, drug reaction, have a high predisposition. 2,5 The disease is progressive with a protracted history. While definitive diagnosis is achieved on histopathology in human patients, this is rarely undertaken in dogs where idiopathic pulmonary fibrosis is commonly a diagnosis of exclusion. Canine idiopathic pulmonary fibrosis lesions are characterized by an accumulation of collagen in the interstitial space, type II pneumocyte hyperplasia and alveolar septal fibrosis. 5,6 However, the pathological description in the dog does not match that required for a diagnosis of usual interstitial pneumonia in most cases, and is more reminiscent of that seen with nonspecific interstitial pneumonia in humans. 2 The exact relationship, if any, between nonspecific interstitial pneumonia and usual interstitial pneumonia in human patients is unknown, but may represent a spectrum of idiopathic interstitial lung disease that eventually results in end-stage fibrosis. 7 Computed tomography (CT) findings are commonly used as the basis for disease classification and prognosis in people. Human patients with nonspecific interstitial pneumonia have a better prognosis than those with usual interstitial pneumonia. 7 Computed tomography features of canine idiopathic pulmonary fibrosis have been previously described, with common findings including ground-glass pattern, reticular abnormalities, traction bronchiectasis, and honeycombing in the later stage. [8][9][10][11] These characteristics are somewhat similar to high resolution CT findings in human idiopathic pulmonary fibrosis except there appears to be a greater degree of ground-glass attenuation in the dog that is reported to equate with human nonspecific interstitial pneumonia. 7,8,10 Hematology and biochemistry profiles, bronchoscopy, and bronchoalveolar lavage are often unremarkable in affected dogs. Chronic bronchitis can be a common comorbidity that complicates diagnosis, but also the main reason for the presence of a highly cellular bronchoalveolar lavage. 8,9 According to the American Thoracic Society and European Respiratory Society 2011 consensus statement on human idiopathic pulmonary fibrosis the appearance of the lung on CT has both diagnostic and prognostic value that plays a major role in disease staging and decision making for patient care. 12 A correlation between CT findings and clinical signs has not been reported for the dog. An American Thoracic Society workshop report on comparative idiopathic pulmonary fibrosis (2013) highlighted the need for further descriptive research in order to better define the clinical and imaging presentation of dogs with idiopathic pulmonary fibrosis. 13 Aims of the current study were to establish CT descriptive terms for characterizing idiopathic pulmonary fibrosis in West Highland white terriers and determine whether CT characteristics are correlated with clinical signs or survival time. The study hypothesis was that the severity of the lesions on CT would be positively correlated to the severity of clinical signs, thus supporting the use of CT as a prognostic tool for affected dogs.

Dogs
The study was a retrospective, analytical, cross-sectional design.
Databases of the following three referral hospitals were searched for West Highland white terriers with a diagnosis of interstitial pulmonary fibrosis: University of Edinburgh, University of Glasgow, and University of Pennsylvania. For purposes of this study, the diagnosis of interstitial pulmonary fibrosis was based on clinical presentation, results from bronchoscopy, and bronchoalveolar lavage with mild changes, and an interstitial pulmonary abnormality on CT images. Inclusion criteria for participation in the study consisted of available thoracic CT images, hematology and biochemistry, and bronchoscopy performed at the time of diagnosis. Decisions for study inclusion for dogs in the affected group were made by the first author (F.T.). The database of the University of Edinburgh teaching hospital was also searched from July 2009 to November 2015 for West Highland white terriers without respiratory disease that had thoracic CT scans. Dogs for this control group were excluded from the study if any pulmonary pathology or significant lung atelectasis was noted on CT, or if there was any history of respiratory disease in the database. Decisions for inclusion or exclusion for the control group were made by the first author (F.T.).

Computed tomography data recorded
All CT studies from dogs with idiopathic pulmonary fibrosis and control patients were randomised and blinded reviewed by a board-certified veterinary radiologist (T.S.) and an imaging resident (F.T.), and scored by consensus. All assessments and measurements were performed  Table 1). The severity of each lung pattern was based on previously published data in human literature. 4,14,15 A grade 0 was equivalent to a normal pulmonary parenchyma, grade 1 (mild severity) equivalent to a ground-glass pattern defined as hazy increased attenuation with preservation of the bronchial and vascular margins. A grade 2 (moderate severity) was given if there was a ground-glass mosaic pattern (patchwork of regions of differing attenuation) or focal reticular pattern (complex network of curvilinear opacities) or any pulmonary consolidation or bronchiectasis defined as a lack of tapering of the bronchi. A grade 3 (marked severity) was equivalent to a generalized reticular pattern or honeycombing (subpleural cystic airspaces) or traction bronchiectasis (irregular bronchial dilatation with abnormal surrounding parenchyma) or nodular pattern. The four scores given for each study were then averaged and named as "CT score". Tracheal or main stem bronchial collapse was also recorded for each tomographic study. A second set of regions of interest placements and Hounsfield unit measurements were performed by the same operator two months after the first set of measures in order to assess the intraobserver variability.

Clinical data recorded
All clinical data were reviewed by a veterinary cardiopulmonary clin- were then taken into account to establish the "clinical score", a whole number from 0 to 3.    and was related to respiratory failure for two of them. Postmortem histology of the lung was available in six dogs. Chronic interstitial fibrosis with type II pneumocyte hyperplasia was noted in three dogs, pleural or subpleural fibrosis in two, and chronic interstitial pneumonia in three.

Descriptive computed tomographic and clinical findings
Alveolar macrophages were described in three dogs and alveolar thickening in two others.

Comparisons between computed tomographic findings and clinical findings
Nine of the control dogs had mild pulmonary atelectasis on CT evaluation. All dogs of the control group were blindly given a CT score

DISCUSSION
This is the first cross-sectional study demonstrating the correlation between the severity of the clinical signs of canine idiopathic pulmonary fibrosis and the severity of the abnormalities on CT.
We also established that dogs with mild changes on CT are more likely to have a longer survival time. In human patients, the lung patterns on CT have been well classified depending on their specificity for usual interstitial pneumonia. 4,12,14,15 We based our grading scale on these previously published data. The appearance of idiopathic pulmonary fibrosis on CT in the dog is often described as a ground-glass pattern with a generalised hazy pattern. 8,10,11,16 This was confirmed in our study with 16 animal presenting this tomographic feature. Focal reticular pattern and mosaic ground-glass pattern were common findings (10/21) and were considered as a feature of moderate severity (CT score of 2). Mosaic ground-glass pattern has not been previously described for the appearance of idiopathic pulmonary fibrosis on CT in West Highland white terriers and was present in 19% of the affected dogs in this study. Honeycombing is a severe reticular pattern commonly illustrated in advanced cases of human and canine idiopathic pulmonary fibrosis with subpleural location being most commonly described. 8,12,15 Compared to honeycombing observed in humans, the degree was very mild in our study population, consistent with a previous canine idiopathic pulmonary fibrosis study. 8 Nodules scattered throughout the lung parenchyma are described as well with canine idiopathic pulmonary fibrosis although appears less common. 5,8 F I G U R E 3 Clinical score and averaged CT score given for 21 West Highland white terriers with idiopathic pulmonary fibrosis as described in Table 1 F I G U R E 4 Box-plots representing the lung attenuation on CT (in Hounsfield units) in West Highland white terriers with idiopathic pulmonary fibrosis and in unaffected West Highland white terriers (P < 0.001) Parenchymal bands representing atelectasis or fibrosis are common, but nonspecific findings within the pulmonary parenchyma on tomographic images. In a previous report the dorsal aspect of the lung parenchyma was reported to be more affected in dogs with idiopathic pulmonary fibrosis, but the current data identified more diffuse changes. 10 There was a moderate positive association between the severity of the clinical signs and the severity of the lesions within the pulmonary parenchyma on CT, and this finding has not been demonstrated previously. While in human idiopathic pulmonary fibrosis, lung atten-uation measurement is not considered reliable due to its dependence to the respiratory phase, the affected dogs had higher lung attenuation than control dogs. 10 Apnea following manual hyperventilation was induced in all dogs before scan acquisition, which may explain the good reliability of lung attenuation in our study. We chose to establish the cut-off value at -702 HU over which all dogs are considered affected by idiopathic pulmonary fibrosis. A previous study reported lower lung attenuation in dogs with idiopathic pulmonary fibrosis (mean: -735 HU, standard deviation: 55) compared to our data (mean: -563 HU, standard deviation: 74) but the measurement method was not described. 16 High lung attenuation in dogs with idiopathic pulmonary fibrosis is unfortunately not specific for this disorder, but we believe it can be used as an additional tool in the diagnostic process.
In this study, the survival time from diagnosis of dogs with idiopathic pulmonary fibrosis was not correlated to the severity of the clinical signs, but was negatively associated to the severity of the imaging findings. Indeed, advanced lesions on tomographic images tended to occur in dogs with shorter survival time. The median survival time after diagnosis was 8.5 months, which is slightly less than the previously pub- terriers. 4 Further studies with larger sample size are needed to confirm our preliminary findings.
In one-third of the dogs that underwent echocardiography (4/12), there were signs of mild pulmonary hypertension. This proportion is slightly lower than the previously reported prevalence of 44% in West Highland white terriers with interstitial disease. 17 These dogs with mild pulmonary hypertension appeared to have a wide range of severity of clinical signs from mild to severe but all of them had mild changes on tomographic images.
In the current study, 76% of dogs (16/21) had concurrent signs of chronic bronchitis. This comorbidity has previously been reported and is likely at the origin of the chronic active inflammation of the bronchoalveolar lavage fluid samples. 8,[19][20][21] Determining if abnormalities on CT were attributed to another inflammatory process than idiopathic pulmonary fibrosis is an important issue. Most dogs had mild changes on bronchoscopy (11/16), which alone was not sufficient to explain the clinical signs of the animals of this study, and on that basis it was not unreasonable to decide that idiopathic pulmonary fibrosis was the primary diagnosis of clinical significance. In animals, the most common causes of interstitial lung disease are infectious agents, toxin exposure, high dose irradiation, immunologic, and neoplastic disorders, making tentative diagnosis much easier than for the same class of interstitial lung disease in human patient. 1 Additional risk factors have been described in human idiopathic pulmonary fibrosis such as smoking, or environmental exposures to specific dusts. 12 In the case of idiopathic pulmonary fibrosis-like diseases in people several are described based on histological appearance such as usual interstitial pneumonia, acute interstitial pneumonia, desquamative interstitial pneumonia, or organising pneumonia. 5,15 Inflammation and fibroblast proliferation are common histologic features of these disorders. 5 Human usual interstitial pneumonia has been suggested to resemble idiopathic pulmonary fibrosis in West Highland white terriers, but the preponderance of ground glass opacity on CT, suggests early canine idiopathic pulmonary fibrosis more resembles human nonspecific interstitial pneumonia. 2,3,7 The limited histopathology from cases in this series and other reports would tend to support that assertion. 2,3,7 Histology findings in dogs include accumulation of alveolar macrophages, alveolar luminal changes, honeycombing of the alveolar architecture, thickening of the alveolar septa with oedema, fibroblast proliferation, proliferation of alveolar type II pneumocytes, but no evidence of myofibroblast foci (except in a few cases) that is diagnostic feature of true usual interstitial pneumonia. 2,5 A mosaic ground-glass and laterally a honeycombing pattern is found in moderate and severe forms of the canine disease, and more likely equates with usual interstitial pneumonia in human patients.
In canine idiopathic pulmonary fibrosis there is overexpression of the cytokine chemokine ligand-2 (CCL2). 1,22,23 This cytokine acts on fibroblasts via an increased expression of transforming growth factor TGF-, which then leads to accumulation of extracellular matrix and fibrosis. 2,3,23,24 The retrospective nature of the study was a limitation. Some investigations such as echocardiography or arterial blood gases were not performed in all dogs. The unstable nature of the animals was the main reason for the missing data. High-resolution CT is recommended in human and veterinary literature for diagnosis of idiopathic pulmonary fibrosis. 8,14 This technique does not require specialised CT equipment but is a technique with thin slice thickness, a small field of view and high frequency reconstruction algorithm to maximise spatial resolution. Such tomographic acquisition was not available for all dogs in our study, which represents another limitation. A linear pattern with interlobular septal thickening superimposed on ground-glass opacity (so-called "crazy paving" appearance) has previously been reported on CT in dogs with idiopathic pulmonary fibrosis but was not seen in our study. 8 This type of pattern may only be highlighted by high resolution CT, which would explain its absence in our study. The absence of lung histology in 15 dogs is another limitation. We could not examine the association between pathological and CT features because of the small number of histology results. The nature of the pulmonary nodules detected on CT was not investigated due to their small size precluding a safe sampling. Performing lung biopsy is the gold standard for diagnosis of idiopathic pulmonary fibrosis but the invasiveness of the procedure makes its feasibility difficult. 18 Furthermore, this technique only allows assessing a small portion of the lung parenchyma, which may not be representative. The diagnosis of idiopathic pulmonary fibrosis is challenging. It is common practice to perform a diagnosis of exclusion from other cardiac or respiratory pathology (as has been done in this study). 8 The human 2011 idiopathic pulmonary fibrosis guidelines states that for patients not available for lung biopsies, a diagnosis of idiopathic pulmonary fibrosis only requires the exclusion of other known cause of interstitial lung disease and the presence of usual interstitial pattern on high resolution CT. 12 In conclusion, findings from the current study support the use of thoracic CT as a diagnostic tool for grading of canine idiopathic pul-