Introduction

A cavitary pulmonary lesion by definition occurs when there is a mass or nodule that contains a gas-filled space within an area of pulmonary consolidation.¹ These lesions are proposed to occur due to inadequate vascular supply within the center of the structure, which leads to necrosis and expulsion of the necrotic tissue by the bronchial tree leading to cavitation.² Cavitary lesions can be caused by infectious or neoplastic etiologies.^3,4 Other gas-filled lesions such as emphysema, pneumatoceles, bullae, and subpleural blebs can mimic the appearance of cavitary pulmonary lesions; however, these lack surrounding pulmonary consolidation, which prevents their classification as true cavitary pulmonary lesions.^3,4

Primary pulmonary neoplasia in dogs and cats has an incidence rate of approximately 4.2 cases/100,000 dogs and 2.2 cases/100,000 cats.^5,6 The frequency of cavitation in primary pulmonary neoplasia varies in dogs and cats, with one study reporting a prevalence of 11% in dogs and another reporting a prevalence of 63% in cats.^1,7 In humans, the prevalence of cavitation more closely mirrors that noted in the canine studies, with rates ranging between 10% and 15%.^1,8,9

With the increasing availability of CT in veterinary practice, many animals with pulmonary lesions receive CT scans to further characterize these lesions following their discovery on thoracic radiographs. The CT appearance of pulmonary cavitary lesions is well-documented in human medicine and certain parameters, including wall thickness, wall margin characteristics, and the presence of intralesional air bronchograms, can be used to make determinations about the underlying etiology of these lesions. In the human medical field, these characteristics are used to make inferences about the underlying etiology of these cavitary pulmonary lesions based solely on the CT findings. Parry et al¹⁰ described these characteristics in a recent veterinary case series; however, the utility of these characteristics for diagnostic decision-making has not yet been investigated. The objective of the present study was to evaluate and define these CT characteristics in veterinary patients and determine whether they can be used as predictors of malignancy as is described in the human medical literature.

Materials and Methods

This multi-institutional retrospective study was designed to include cases presenting to collaborating institutions between January 1, 2010, and December 31, 2020. At each institution, medical records system and/or imaging database searches were performed to identify dogs and cats with cavitary pulmonary lesions. Search terms included “cavitary,” “cavitated,” and “cavity,” among others. CT studies for candidate cases were submitted from the institutions based on these search terms, and then images were reviewed by a radiology resident (SLJ) and board-certified radiologist (EH) to ensure they met the proper definition of a cavitary pulmonary lesion prior to inclusion in the final study population. Cases were also required to have a confirmation of a diagnosis of the lesion by either cytology or histopathology. Definitive diagnosis was obtained by fine-needle aspirate sample (ultrasound or CT guided) for cytology or histopathology (via surgical excision). Cases were excluded if the diagnosis was not confirmed by histopathology or cytology. Additionally, cases that did not contain gas-filled cavities but rather contained fluid-filled cavities were excluded. A sole air bronchogram without evidence of cavitation, such as is seen in bronchopneumonia, was also a cause for exclusion. Medical records for the included cases were reviewed, and the patient signalment, clinical history, and diagnosis were recorded.

The CT studies were interpreted, and measurements were performed by a third-year radiology resident at The Ohio State University Veterinary Medical Center. These findings were reviewed by a board-certified veterinary radiologist who oversaw data interpretation and was not masked to the resident’s findings. A commercial DICOM viewer (RocketPACS; Vet Rocket LLC) was used for image review and analysis. Display settings were kept constant for pulmonary imaging with a window width of 1,500 HU and a window level of –600 HU. The investigated characteristics were wall thickness at the lesion’s thickest point, wall thickness at the lesion’s thinnest point, lesion diameter (in cm), wall margin characteristics (smooth vs irregular), presence of spiculation, presence of air bronchograms, presence and characteristics of intralesional contrast enhancement (homogenous, heterogenous, or absent), presence of additional nodules, presence of abnormal adjacent pulmonary parenchyma, and presence of pleural tags.^2–24 Lesion wall thickness at the thickest and thinnest points of the lesion were measured using electronic calipers and recorded (in mm). The diameter of the primary lesion was measured at its maximum point (in cm) using electronic calipers.^16,17 The presence and number of additional nodules were recorded, as well as the maximum diameter (in cm) of the largest nodule. Any additional pleural findings were recorded including the presence of a pneumothorax or pleural effusion, but these findings were not statistically analyzed.

Definitions of terms not commonly used in the veterinary medical literature were extrapolated from the human medical literature. A spiculation (or the corona radiata) was defined as a linear opacity that extended from the mass outwards. If this opacity extended from the mass to the margin of the visceral pleura, it was defined specifically as a pleural tag.^11,12 Pleural tags are linear in nature and extend from the mass to the visceral pleura and can be representative of fibrotic tissue extending from the lesion to the pleura leading to inward retraction of the visceral pleura.^11,13,14 Air bronchograms are defined as 1 or multiple linear branching hypoattenuations representing bronchi or bronchioles that pass through lung parenchyma, which is densely opacified, or through the mass itself.¹⁵ Wall thickness was measured in a similar fashion as reported in human medicine by determining the diameter of the lesion from the lumen to the outside edge at both the thickest and thinnest points.^16,17

Results

Forty-two animals were identified as having cavitary pulmonary lesions with confirmed histopathological or cytological diagnosis on thoracic CT. Five institutions both from private and academic hospitals participated in this study. They contributed 13 (institution A), 6 (institution B), 5 (institution C), 10 (institution D), and 8 (institution E) animals. Of the included patients, 15 were cats and 27 were dogs. Ages ranged from 3 months to 17 years. The mean age of animals with malignant neoplasia was 11.4 years of age. The mean age of animals with nonneoplastic disease or benign neoplasia was 6.4 years of age. The malignant neoplasia group included 29 of 42 (69%) cases. Diagnoses included pulmonary adenocarcinoma (n = 29) and sarcoma (3). In the included animals, nonneoplastic or benign neoplastic lesions were identified in 13 of 42 (30%) cases. In the included cats, 7 of 15 cases had malignant neoplasia. Of the included dogs, 22 of 27 cases had malignant neoplasia. Diagnoses in this group included pulmonary abscesses or bacterial infections (8/13), parasitic infections (1/13), fungal infection (1/13), pulmonary bullae (1/13), and pulmonary adenoma (1/13).

There was a wide range of clinical signs in these patients. Pulmonary adenocarcinoma was most commonly associated with coughing (in 16/29 cases), with other clinical signs including respiratory distress, weight loss, regurgitation, and anorexia. In patients with histiocytic sarcoma, 2 out of 3 lesions were found incidentally, with only 1 patient presenting with coughing. Pulmonary adenoma was not associated with any clinical signs (incidental finding). Bacterial infections or pulmonary abscesses were associated with coughing in 3 of 8 cases, with the remaining cases having signs including nasal discharge and facial swelling. Both parasitic infections and pulmonary bullae patients presented with recurrent pneumothorax. The patient with fungal disease presented with weight loss and coughing.

Of the imaging parameters evaluated, there was significant overlap between characteristics seen in malignant neoplasia and benign lesions. The presence or absence of these CT characteristics are outlined (Table 1). Seven of 13 descriptors were not statistically associated with the final diagnosis of the lesion (malignant neoplasia vs benign neoplasia or other etiologies). These included irregular versus smooth lesion margins, presence of spiculations, presence of intralesional air bronchograms, having abnormal surrounding pulmonary parenchyma, the presence of pleural tags, or the diameter or the lesion. Six of these 13 parameters were statistically associated with the final diagnosis of the lesion (Figures 1 and 2). These included the presence of intralesional contrast enhancement, type of intralesional contrast enhancement (heterogenous and homogenous studied separately), presence of additional nodules, wall thickness of the lesion at its thickest point, and wall thickness of the lesion at its thinnest point. For continuous variables (wall thickness and lesion diameter), the mean values are listed ± SD. The sensitivity and specificity for each of these values in predicting malignant neoplasia were also identified (Table 2).

Examples from 2 different malignant cases diagnosed with carcinoma with wall thickness > 40 mm, heterogenous contrast enhancement, and evidence of metastatic disease. Patient 1 (A and B) demonstrates a large lobular cavitated mass (outlined with white and black arrowheads) with heterogenous contrast enhancement of the mass. The thickest wall measurement was 42.9 mm in this patient, and this patient has 2 additional metastatic nodules (not pictured). Patient 2 (C and D) demonstrates multiple well-defined soft tissue nodules (white arrows) in the parenchyma surrounding the pulmonary mass. The thickest wall measurement was 51.6 mm in this patient. Soft tissue (A and C) window width and window level 400/40, lung (B and D) window width and level 1,500/–600. The images were acquired with 120 kVp, with the slice thickness for patient 1 being 1.0 mm and the slice thickness for patient 2 being 1.25 mm.

Citation: Journal of the American Veterinary Medical Association 261, 10; 10.2460/javma.23.02.0076

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Examples from 2 different malignant cases diagnosed with carcinoma with wall thickness > 40 mm, heterogenous contrast enhancement, and evidence of metastatic disease. Patient 1 (A and B) demonstrates a large lobular cavitated mass (outlined with white and black arrowheads) with heterogenous contrast enhancement of the mass. The thickest wall measurement was 42.9 mm in this patient, and this patient has 2 additional metastatic nodules (not pictured). Patient 2 (C and D) demonstrates multiple well-defined soft tissue nodules (white arrows) in the parenchyma surrounding the pulmonary mass. The thickest wall measurement was 51.6 mm in this patient. Soft tissue (A and C) window width and window level 400/40, lung (B and D) window width and level 1,500/–600. The images were acquired with 120 kVp, with the slice thickness for patient 1 being 1.0 mm and the slice thickness for patient 2 being 1.25 mm.

Citation: Journal of the American Veterinary Medical Association 261, 10; 10.2460/javma.23.02.0076

• Download Figure
• Download figure as PowerPoint slide

Examples from 2 different malignant cases diagnosed with carcinoma with wall thickness > 40 mm, heterogenous contrast enhancement, and evidence of metastatic disease. Patient 1 (A and B) demonstrates a large lobular cavitated mass (outlined with white and black arrowheads) with heterogenous contrast enhancement of the mass. The thickest wall measurement was 42.9 mm in this patient, and this patient has 2 additional metastatic nodules (not pictured). Patient 2 (C and D) demonstrates multiple well-defined soft tissue nodules (white arrows) in the parenchyma surrounding the pulmonary mass. The thickest wall measurement was 51.6 mm in this patient. Soft tissue (A and C) window width and window level 400/40, lung (B and D) window width and level 1,500/–600. The images were acquired with 120 kVp, with the slice thickness for patient 1 being 1.0 mm and the slice thickness for patient 2 being 1.25 mm.

Citation: Journal of the American Veterinary Medical Association 261, 10; 10.2460/javma.23.02.0076

• Download Figure
• Download figure as PowerPoint slide

These cases represent an example of pulmonary spiculations (white arrowheads), thin linear soft tissue opacities that extend outward from the margins of the mass. Case A represents a 12-year-old male neutered domestic longhair cat that was found to have carcinoma. Case B represents a 12 year old female spayed domestic shorthair cat that was found to have a chronic abscess with necrosis and histiocytosis; no etiological agents were noted. Of note, a pleural tag is along the pleural margin in case A (black arrow), extending from the mass to the pleural margin with evidence of retraction. Lung window width and level 1,500/–600. The images were acquired with 120 kVp, with the slice thickness for A being 2.0 mm and the slice thickness for B being 1.25 mm.

Citation: Journal of the American Veterinary Medical Association 261, 10; 10.2460/javma.23.02.0076

• Download Figure
• Download figure as PowerPoint slide

View Full Size

These cases represent an example of pulmonary spiculations (white arrowheads), thin linear soft tissue opacities that extend outward from the margins of the mass. Case A represents a 12-year-old male neutered domestic longhair cat that was found to have carcinoma. Case B represents a 12 year old female spayed domestic shorthair cat that was found to have a chronic abscess with necrosis and histiocytosis; no etiological agents were noted. Of note, a pleural tag is along the pleural margin in case A (black arrow), extending from the mass to the pleural margin with evidence of retraction. Lung window width and level 1,500/–600. The images were acquired with 120 kVp, with the slice thickness for A being 2.0 mm and the slice thickness for B being 1.25 mm.

Citation: Journal of the American Veterinary Medical Association 261, 10; 10.2460/javma.23.02.0076

• Download Figure
• Download figure as PowerPoint slide

These cases represent an example of pulmonary spiculations (white arrowheads), thin linear soft tissue opacities that extend outward from the margins of the mass. Case A represents a 12-year-old male neutered domestic longhair cat that was found to have carcinoma. Case B represents a 12 year old female spayed domestic shorthair cat that was found to have a chronic abscess with necrosis and histiocytosis; no etiological agents were noted. Of note, a pleural tag is along the pleural margin in case A (black arrow), extending from the mass to the pleural margin with evidence of retraction. Lung window width and level 1,500/–600. The images were acquired with 120 kVp, with the slice thickness for A being 2.0 mm and the slice thickness for B being 1.25 mm.

Citation: Journal of the American Veterinary Medical Association 261, 10; 10.2460/javma.23.02.0076

• Download Figure
• Download figure as PowerPoint slide

For wall thickness specifically, the ROC was used to determine the best cutoff points for this marker. Using this curve, it was found that at a wall thickness of > 39.6 mm (at the thickest point) there is a 92% specificity for malignant neoplasia and 24% sensitivity. Similarly, a wall thickness > 1.9 mm (at the thinnest point) is 92% specific for malignant neoplasia and 32% sensitive.

Discussion

There have been several studies describing the imaging characteristics of cavitary pulmonary masses in small animals radiographically and 1 study describing the imaging findings on thoracic CT. However, to the authors’ knowledge using an online literature search of PubMed for relevant search terms (spanning from 1966 to present and including CT with terms “cavitary pulmonary lesions,” “pulmonary lesions,” “pulmonary masses,” “cavitary pulmonary masses,” and “pulmonary adenocarcinoma”), this was the first study to look at a larger number of cases and assess the utility of these characteristics in predicting malignant neoplastic disease.^1,3,10,18 In this study population, a statistically significant difference was noted in the wall thickness, type of intralesional contrast enhancement, and presence of additional nodules when comparing malignant neoplasia and nonneoplastic disease or benign neoplasia. The other CT features analyzed in this study were not significantly different between these disease groups.

Irregular wall margins have been described as a finding in neoplastic disease, specifically in pulmonary carcinoma in humans.^11,13,14 This finding is theorized to be due to an uneven rate of growth in cancerous lesions as opposed to benign masses in the lungs.^11,13,14 In a prior veterinary case series, 4 out of 5 cases of cavitary pulmonary adenocarcinoma were noted to have irregular wall margins.¹⁰ In prior studies of pulmonary lesions on CT in humans, 81% of a malignant etiology had irregular wall margins, whereas just 19% of benign lesions had irregular margins.^16,17 In the current study, malignant neoplastic lesions were noted to have irregular margins in 21 of 29 cases; however, 8 of 13 nonneoplastic cases also had irregular margins. The presence of irregular margins was neither sensitive (72%) nor specific (38%) for the identification of malignant neoplastic disease in this study.

Spiculations, also called the corona radiata, are also often discussed in conjunction with irregular wall margins. Spiculations are associated with interlobular septal thickening and fibrosis that occurs from parts of the airway becoming obstructed by either neoplastic cells or, in some cases, benign causes.^11,13,14,19 In humans, fine spiculations have a 90% positive predictive value for pulmonary adenocarcinoma.^7,11,13,20 The presence of spiculations has also been documented in case series of pulmonary adenocarcinoma in dogs and cats.¹⁰ In the present study, spiculations were documented in 21 of 29 malignant neoplastic cases. However, 6 of 13 benign or nonneoplastic cases also contained spiculations and this feature was not statistically significant for discerning between malignant and benign lesions. The presence of spiculations in both a malignant (carcinoma) and benign (chronic abscess) were noted in this study (Figure 2). Spiculations also had relatively low sensitivity and specificity for neoplastic disease at 72% and 53%, respectively.

Although air bronchograms can be seen in many different interstitial pulmonary diseases, it has been noted in prior studies that if they are located within a mass itself and have a more tortuous or ectatic appearance, they are more likely to signal neoplastic disease rather than other etiologies.^21,22 The reason for this is theorized to be that the tumor surrounds the bronchi and invades into the bronchial walls, leading to destruction of bronchial tracts and causing intralesional air bronchograms.²² In pulmonary nodules in humans, the presence of intralesional air bronchograms is noted in approximately 73% of neoplastic lesions.^11,14,23,24 In this study, 75% of malignant neoplastic cases contained intralesional air bronchograms, but 53% of benign lesions also had this finding and there was not a significant difference between the two in this population.

Although the presence of pleural tags, lesion diameter, and the presence of abnormal adjacent pulmonary parenchyma are other CT findings associated with malignant neoplasia, this study did not find a statistically significant difference between neoplastic and nonneoplastic or benign diseases when looking at these characteristics. The presence of interstitial abnormalities in the surrounding pulmonary parenchyma, known as a “ground glass opacity” can be caused by infectious causes, neoplastic disease, and vasculitis.^11,13 In this study, 27 of 29 malignant neoplastic cases had this finding, as did 12 of 13 nonneoplastic cases. The specificity for this feature was 8%, making it a relatively common finding for pulmonary diseases in general but not helpful in discerning between malignant and nonmalignant disease. This is likely due to the fact that this finding is thought to occur as a result of several factors including hemorrhage, inflammation, or direct tissue compression due to tumor growth.¹¹ Similarly, pleural tags have been associated with both metastatic disease as well as granulomatous disease, which likely accounts for its low specificity in our sample population (46%) and lack of significant association between this finding and malignant neoplasia (P = .09).

Despite some of the CT features of malignant neoplasia not being predictive of neoplasia in our study population, others were noted to have a significant difference between groups. The presence of additional nodules was associated with malignant neoplasia (P = .0015). Interestingly, 5 (38%) of the nonneoplastic cases also had additional nodules, so although this factor is more likely to occur in neoplastic disease, it should not be used as a stand-alone diagnostic marker of malignant neoplasia. The type of intralesional contrast enhancement was also noted to be predictive of malignant neoplastic disease in this study population. In a prior veterinary case series of cavitary pulmonary adenocarcinoma, all 5 cases were noted to have heterogenous contrast enhancement.¹⁰ In human and other veterinary studies, the type of contrast enhancement noted in different neoplastic lesions is variable. Some studies have found associations between hepatocellular carcinoma and the presence of heterogenous contrast enhancement, whereas other studies involving splenic and adrenal masses did not find an association between the type of contrast enhancement and malignancy.^11–15 In this study population, 25 of 29 malignant neoplasia cases had heterogenous contrast enhancement, whereas just 4 of 13 nonneoplastic or benign lesions had this type of contrast enhancement. There was a statistically significant difference between these two in this study; however, this finding should be considered in concert with other possible factors due to its relatively low specificity (69%) and the inconsistent significance of this marker across other studies.

In human medicine, wall thickness of a cavitary pulmonary lesion is noted to be a useful marker in predicting malignancy.^16,17 Woodring et al^16,17 found that a wall thickness of > 24 mm in cavitary pulmonary lesions in humans was 100% specific for malignancy. A wall thickness of < 7 mm was also noted to be 96.7% specific for nonmalignant lesions.^16,17 In our patient population, wall thickness was measured at both the thickest and thinnest points of the wall of the cavity. In this study, the wall thickness was noted to be greater on average for malignant neoplastic lesions at both the thickest and thinnest points than nonneoplastic lesions. At a wall thickness of > 39.6 mm (at the thickest point), there is a 92% specificity for malignant neoplasia. This data point, while being highly specific, is not particularly sensitive with a sensitivity of 24%. Similarly, a wall thickness of > 1.9 mm (at the thinnest point) is 92% specific for malignant neoplasia but just 32% sensitive. These findings suggest that lesions with wall thicknesses > 39.6 mm at their thickest points and 1.9 mm at their thinnest points are more likely to be malignant in origin.

Similar to other retrospective studies, our study had some limitations. The first was the small sample size, which increased the risk of type 2 statistical error and may have been a reason why some markers of malignancy previously noted to be significant in other studies, such as the presence of intralesional air bronchograms, presence of pleural tags, and lesion diameter along with the other 7 identified, were not significant in this study. Although this study evaluated a wide range of cases spanning several institutions, with the relative newness of CT as a diagnostic tool, it may take several more years to accrue a large enough study population to address the problems with sample size. The different institutions also had variations in the exact sedation versus general anesthetic protocols for patients undergoing CT, and there were variations in the exact model of CT used in each institution. These variations could have led to some variability in the imaging results for example in patients in which a specific breath-hold could have been utilized under anesthesia versus in sedated patients for which this was not feasible. Another limitation was the selection bias for cases of cavitary pulmonary lesions that underwent either FNA for cytologic diagnosis or surgical biopsy to confirm their diagnoses. Additionally, there is a selection bias for cases in which CT was able to be performed and many patients may have been excluded after radiographic diagnosis due to inability to pursue a CT. For cases in which CT was possible, a central location of the pulmonary mass or lesion may have precluded the ability to safely obtain a cytological sample without causing possible risk to the patient.

In conclusion, this study has shown that while there are many possible imaging features for identifying malignancy on thoracic CT based on studies in human medicine, many of them are not necessarily more frequent in malignant neoplastic disease as opposed to other causes of cavitary pulmonary lesions included in this study population. Some, including the type of contrast enhancement, presence of additional nodules, and the wall thickness of the lesion, can reasonably be considered supportive for the diagnosis of malignant neoplastic disease. In lesions that have heterogenous contrast enhancement, additional nodules, and wall thickness > 40 mm at its thickest point, it would be reasonable to consider malignant neoplastic disease higher on the list of differentials than other causes. This demonstrates the utility of thoracic CT in cavitary pulmonary masses and how they can be a useful tool in providing clinicians and owners additional information aiding in decision-making prior to possible surgical intervention. Ultimately, histopathologic examination is still considered the gold standard in making a definitive diagnosis in these patients and should be pursued when possible.