Hemangiosarcoma (HSA) is a malignant neoplasm that occurs more frequently in dogs than in any other species and represents about 2% of all canine tumors.1,2 This neoplasia can affect all breeds; however, German Shepherds, Golden Retrievers, Labrador Retrievers, and other large-breed dogs are overrepresented in many studies.1,3–5
The most common primary site for HSA in dogs is the spleen, while other common anatomic locations are the heart, especially the right atrium or auricle, the liver, and the skin.1 Less common primary sites have also been reported, including but not limited to the kidney, retroperitoneum, urinary bladder, uterus, lung, left ventricle of the heart, region of the great vessels, oral cavity and tongue, bone, spinal canal, muscle, digit, and cornea.6
Due to the intimate association between the blood vessels and this neoplasia, HSA has a typically aggressive biological behavior, with a rapid and widespread metastatic process.6 Metastatic lesions can result either from hematogenous spread or intracavitary implantation following tumor rupture, and the most common sites of dissemination are the liver, peritoneum, and lungs.1,2,6
Staging of patients with confirmed or suspected HSA is crucial; abdominal ultrasound examination (AUS) and 3-view thoracic radiographs are the diagnostic imaging modalities frequently used for this purpose.1 The AUS characteristics of abdominal parenchymal or peritoneal metastases from HSA are variable, with a higher prevalence of multifocal hypoechoic, heterogeneous, or target-like nodules, often with peritoneal effusion.7–9 The sonographic features of muscular metastases have also been described in a dog.10
With radiographs, the prevalence of visible lung metastases at the initial presentation in dogs with HSA is reported to range between 19% and 52%.11 The radiographic features of pulmonary metastases from HSA are variable but are often described as a coalescing unstructured to structured miliary interstitial pattern.12,13 However, survey thoracic radiographs underestimate the prevalence of lung metastases, especially when assessed by nonspecialists.14 Consequently, the fact that there are no radiographically visible pulmonary nodules compatible with metastases does not mean that a metastatic process can be ruled out.15,16
Due to the superior contrast resolution and lack of overlap of the thoracic structures, CT is more sensitive than radiology for identifying pulmonary nodules.15,16 It has also been shown that abdominal CT is more effective than AUS in detecting lesions in dogs heavier than 25 kg or when the abdominal sonographic evaluation is challenging due to the presence of intraluminal gas or food remnants in the gastrointestinal tract.17,18
Several studies have used CT to describe the characteristics of primary nonparenchymal HSA and muscular or brain HSA metastases.4,19–21 However to the best of our knowledge, the CT features of HSA pulmonary metastases have not previously been described, and a detailed description of the enhancement pattern of the extrapulmonary metastases is thus lacking.
The main aim of this study was therefore to evaluate the CT features of pulmonary nodules that were suspected, cytologically compatible, or histologically confirmed as metastases in dogs with histologically confirmed HSA at various sites. In addition, the CT characteristics of extrapulmonary metastases in the same population of dogs were assessed. Our hypotheses were that (1) pulmonary metastases are more frequently numerous and small in size, and (2) extrapulmonary lesions are characterized by areas of intense and heterogeneous intralesional enhancement.
Methods
Case selection
This was a retrospective, observational, and descriptive study. Approval by an IACUC or IRB was not required. Informed owner consent to perform the CT study was obtained for all the dogs.
The electronic medical record database of the Clinica Veterinaria dell'Orologio (Sasso Marconi, Bologna, Italy) was searched from April 2013 to January 2024 for total body pre- and postcontrast CT studies of dogs with histologically confirmed HSA at various sites and with pulmonary nodules compatible as metastases. The search terms used to identify dogs with lung metastases were “pulmonary nodule(s),” “lung nodule(s),” “pulmonary metastases,” and “lung metastases”. After this initial search, only dogs with a confirmed diagnosis of HSA were included and evaluated according to the other inclusion criteria.
In the same population of dogs, the tomographic appearance of extrapulmonary lesions consistent with HSA metastases was also recorded. To be included in the study, at least 1 pulmonary or extrapulmonary metastases had to be confirmed by cytology or histopathology and a high-resolution scan of the thorax had to be available. In our institution, the total-body CT scan includes from the nostril to the stifle joint regions.
All cytological slides were reviewed by an experienced clinical pathologist (RF), and the cytological diagnosis was based on criteria reported in the literature.22
Dogs were excluded if they had a second primary neoplasm in addition to HSA at the time of diagnosis or reported in the medical history and if a single unsampled pulmonary nodule was present.
Furthermore, the cases where the cytological diagnosis of the lung or extrapulmonary metastases was uncertain or diagnosed as a generic mesenchymal neoplasm of nonvascular origin were also excluded.
Decisions regarding case selection were made by a European College of Veterinary Diagnostic Imaging board-certified radiologist (FR). The following data were recorded from the clinical database: sex, breed, and age at the time of imaging.
Computed tomography imaging technique
Two different CT units were used: a 16-slice multidetector CT and a 64- to 128-multidetector CT. The settings for the scans were as follows: helical acquisition; slice thickness range, 0.65 to 1.25 mm; matrix size, 512 X 512; voltage, 100 to 120 kVp; tube setting, 200 to 220 mA; pitch, 0.562:1, and rotation time, 0.7 seconds.
A precontrast study followed by a postcontrast study was always performed after the administration of a water-soluble iodinated contrast media at a dosage of 450 to 640 mg L/kg, with an injection rate of 2 to 3 mL/s, and a pressure limit of 325 pounds/sq in. The contrast media was followed by a saline flush with the same injection rate via a dual-syringe injector system. An injection-to-scan delay from the beginning of the injection of approximately 30 to 40 seconds was used. In 8/33 dogs, a standard 3-phase CT study of the abdomen was performed using bolus tracking and a fixed-injection duration (20 seconds) CT angiography protocol.23 If a multiphase angio-CT of the abdomen was done, the second (portal) scan phase was used for the evaluation of the features unrelated to the enhancement progression.
Computed tomography image analysis
All CT studies were reviewed by a second-year European College of Veterinary Diagnostic Imaging resident (MM) who was aware of the final diagnosis. Images were analyzed and measurements were taken using an open-source dedicated DICOM viewer (Horos, version 3.3.6; Horosproject.com). For the evaluation of pulmonary nodules, a lung window (window level, −400; window width, +1,500) was used. Extrapulmonary lesions and the enhancement features of the lung nodules were assessed using a soft tissue window (window level +50; window width +400). Window settings were adjusted as needed.
Primary HSA locations were recorded based on clinical history, including any prior surgeries, or as identified on CT scans, with the largest lesion used to determine the primary location.
In regards lung nodules, the following characteristics were recorded: number (single, between 2 and 10, or over 10), size (miliary [≤ 0.3 cm], subcentimetric [> 0.3 to ≤ 1 cm], or > 1 cm) based on the maximum diameter; margins (well-defined, ill-defined, irregular, or halo sign24); distribution (peripheral, central, or generalized); attenuation (ground glass opacity, soft tissue, or areas of mineralization or gas accumulation); precontrast attenuation and postcontrast enhancement HUs; and pattern of enhancement (homogeneous, heterogeneous, or ring enhancement). In addition, the presence of a feeding vessel, also known as a mass-vessel sign, was recorded and defined as a pulmonary artery branch leading directly to a lung nodule (Figure 1).25 Small (< 0.2 cm) mineralized pulmonary nodules were excluded because they are more suggestive of pulmonary heterotopic bone.11,26
A and B—Transverse CT lung images showed 2 examples of hemangiosarcoma lung metastases with feeding vessels (white arrows) in 2 dogs. In both cases, these nodular lesions appeared well-marginated, peripherally located, and miliary to subcentimetric in size. High-resolution CT images acquired in the transverse plane with 1.25-mm slice thickness and maximum intensity projection reformatted (3 slabs thick) and shown with lung window (window length, −400; window width, +1,500).
Citation: American Journal of Veterinary Research 85, 12; 10.2460/ajvr.24.08.0219
The halo sign is defined as an area of ground-glass attenuation surrounding a pulmonary nodule or mass (Figure 2).24 If a halo sign was present, it was not included in the measured diameter of the nodules, and the margins of these nodules (well or ill defined) were not evaluated.
A and B—Transverse CT lung images showed 2 examples of halo sign surrounding hemangiosarcoma lung metastases in 2 different dogs. In both cases, the nodular lesions appeared peripherally located and subcentimetric in size. High-resolution CT images were acquired in the transverse plane with 1.25-mm slice thickness and shown with lung window (window length, −400; window width, +1,500). The halo sign (white arrows) is defined as an area of ground-glass attenuation surrounding a pulmonary nodule or mass.
Citation: American Journal of Veterinary Research 85, 12; 10.2460/ajvr.24.08.0219
The distribution of pulmonary nodules was evaluated based on a more prevalent subpleural (peripheral), hilar (central), or random (generalized) distribution.
Features such as halo sign, mineralization, gas accumulation, and air bronchograms were considered present if at least 1 of the pulmonary nodules had the described characteristic.
The presence or absence of other thoracic features was evaluated and subjectively categorized as mild, moderate, or marked: bronchial wall thickening, infiltration of the lumen of bronchial and/or vascular structures, pleural effusion, pneumothorax, and thoracic lymphadenomegaly. Thoracic lymph nodes were considered enlarged based on the published reference values.27
As regards extrapulmonary lesions suspected or cytologically compatible as metastases, the following features were evaluated: location of the lesion(s); number (single; between 2 and 10; or over 10); shape (round, oval, or irregular); margins (well-defined, ill-defined, or irregular); precontrast attenuation and postcontrast enhancement in HU; pattern of enhancement (homogeneous, heterogeneous, or ring enhancement); intralesional areas of mineralization (present or absent); and gas accumulation (present or absent). In addition, the presence of peritoneal and/or retroperitoneal effusion was assessed.
These characteristics were evaluated in the precontrast scan and in the single standardized postcontrast scan. However, in cases where a triple-phased angio-CT of the abdomen has been performed, the enhancement pattern progression of the abdominal lesions was compared subjectively in the arterial, portal, and delayed phases.
The presence or absence of abdominal lymphadenomegaly was also assessed and was subjectively graded as mild, moderate, or severe. For both pulmonary and extrapulmonary metastases, minimum, maximum, and mean HU values were evaluated. Three measurements were performed by a region of interest encircling the lesion at the level of its maximal diameter, and the average HU values were recorded; only lesions with a diameter greater than 0.5 cm were assessed.
Statistical analysis
Data analysis was performed by a statistician (VM) using GraphPad Prism (version 9.0, GraphPad Software Inc). The Shapiro-Wilk test was performed to assess the normality of data. Descriptive statistics were calculated. In addition, χ2 analysis was used to compare the number of pulmonary lesions with the intraparenchymal distribution; a P value less than .05 was considered statistically significant.
Results
Study population
A total of 33 dogs met the inclusion criteria. Of these, 13/33 were crossbreeds dogs (39.4%), 5 were German Shepherds (15.2%), 2 Boxers (6.1%), 2 Flat Coated Retrievers (6.1%), 2 Jack Russell Terriers (6.1%), and 1 each of German Dachshund, Border Collie, Bernese Mountain Dog, Cane Corso, Golden Retriever, Czechoslovakian Wolfdog, Australian Shepherd, Bergamasque Shepherd, and Rottweiler (3%).
Most of the dogs included were intact males (14/33 [42.4%]), while 10 were spayed females (10/33 [30.3%]), 5 were intact females (5/33 [15.2%]), and 4 were castrated males (4/33 [12.1%]). The median age of the dogs included was 10 years (range, 5 to 14).
Primary tumor localization
The most common location was the spleen (13/33 [39.4%]), followed by liver (6/33 [18.2%]), musculature (5/33 [15.2%]), retroperitoneal space (4/33 [12.1%]), and kidneys (1/33 [3%]). In dogs with splenic HSA, no concomitant right atrial masses were tomographically identified. The primary locations for the remaining 4 dogs (12.1%) were considered as uncertain, because they had 2 or more large lesions in different locations at the same time and with the same tomographic features. In these dogs, all large lesions potentially compatible with a primary location were not considered in the qualitative assessment, and only smaller lesions consistent with metastatic lesions were included in the evaluation.
Distribution of metastases and diagnosis
In line with the study's inclusion criteria, all the dogs included had lung metastases. In 10 dogs (30.3%), only pulmonary metastases were present, while in the remaining 23 (69.7%), there were lesions on 1 or more organs in addition to the lungs. Specifically, in 11/33 dogs (33.3%), metastatic lesions were present at 1 supplementary site besides the lungs, while 4/33 (12.1%), 5/33 (15.2%), 2/33 (6.1%), and 1/33 (3%) of the dogs had metastases in 2, 3, 4, and 5 sites in addition to the lungs, respectively.
Fourteen dogs had metastatic lesions in the liver (42.4%), 11 in the spleen (33.3%), 10 in the muscles (30.3%), 6 in the peritoneal/retroperitoneal space (18.2%), 4 in the kidneys (12.1%), 3 in the bones (9.1%), 3 in the cutaneous/subcutaneous tissue (9.1%), 2 in the brain (6.1%), and 2 in the adrenal glands (6.1%).
In 22/33 (66.7%) dogs, lung nodules were cytologically diagnosed as metastases, while histological examination of lung lesions was available in only 1 dog.
The other 10/33 dogs had at least 1 of the following extrapulmonary lesion cytologically sampled and diagnosed as HSA metastases: muscular (n = 7), splenic lesions (6), hepatic (6), renal (2), peritoneal/retroperitoneal (2), cutaneous/subcutaneous (2), osseous (2), and adrenal (1).
Peritoneal and retroperitoneal effusions were present in 7 (21.2%) and 2 (6.1%) dogs, respectively, at the time of imaging; the analysis of the effusion was conducted on 6/9 dogs, revealing blood in all instances, displaying varying degrees of RBC degeneration, without the presence of neoplastic cells.
Computed tomography features of lung metastases
The CT characteristics of HSA metastases to the lungs are summarized (Table 1).
CT features of lung hemangiosarcoma metastases and other thoracic findings.
| CT features | No. (%) |
|---|---|
| Number | |
| Single | 1/33 (3%) |
| 2 to 10 | 6/33 (18.2%) |
| > 10 | 26/33 (78.8%) |
| Size | |
| Miliary (≤ 0.3 cm) | 29/33 (87.9%) |
| Subcentimetric (> 0.3 to ≤ 1 cm) | 32/33 (97%) |
| > 1 cm | 5/33 (15.1%) |
| Median maximum diameter (cm) | 0.33 (range, 0.13 to 3.2) |
| Margins | |
| Well defined | 29/33 (87.9%) |
| Ill defined | 16/33 (48.5%) |
| Irregular | 11/33 (33.3%) |
| Halo sign | 24/33 (72.7%) |
| Distribution | |
| Peripheral | 10/33 (30.3%) |
| Central | 1/33 (3%) |
| Generalized | 24/33 (72.7%) |
| Median precontrast (HU) | 35 (range, 9 to 62) |
| Median postcontrast (HU) | 65 (range, 22 to 141) |
| Pattern of enhancement | |
| Homogeneous | 26/33 (78.8%) |
| Heterogeneous | 5/33 (15.1%) |
| Ring enhancement | 8/33 (24.2%) |
| Ground glass opacity nodules | 6/33 (18.2%) |
| Mineralization | 0/33 (0%) |
| Gas accumulation | 2/33 (6.1%) |
| Air bronchograms | 3/33 (9.1%) |
| Feeding vessel | 32/33 (97%) |
| Bronchial wall thickening | 9/33 (27.8%) |
| Bronchial lumen invasion | 1/33 (3%) |
| Vascular lumen invasion | 0/33 (0%) |
| Pleural effusion | 1/33 (3%) |
| Pneumothorax | 0/33 (0%) |
| Thoracic lymphadenomegaly | 11/33 (33.3%) |
In the majority of dogs, more than 10 pulmonary nodules were present (26/33 [78.8%]), while in 6 dogs between 2 and 10 nodules were reported (18.2%) and a single nodule was observed in only 1 dog (3%). The distribution of the lesions was most commonly generalized (24/33 [72.7%]), although it was peripheral in 10 (30.3%) and central in 1 dog (3%). A statistically significant correlation was found between the number and distribution of the lung nodules: a smaller number of lesions (between 2 and 10) had a more frequent peripheral distribution, whereas numerous (> 10) lesions had a generalized location (P < .0001).
Twenty-nine out of 33 dogs had concomitant miliary (87.9%) to subcentimetric (97%) sized nodules, while lesions greater than 1 cm were less common (15.1%); the median diameter of lung nodules was 0.33 cm (range, 0.13 to 3.2 cm).
Twenty-four out of 33 dogs (72.7%) had pulmonary nodules with a halo sign. Excluding nodules with halo signs, 29/33 (87.9%) dogs had pulmonary nodules with well-defined margins.
The median value of precontrast lung nodule attenuation was 35 HU (range, 9 to 62 HU), while in postcontrast scan, this median value increased to 65 HU (range, 22 to 141 HU). Most of the lesions had a homogeneous contrast uptake (78.8%); however, in 8/33 (24.2%), a ring enhancement was visible, while in 5/33 dogs (15.1%) heterogeneously enhancing lesions were reported and characterized by areas of spotty, linear or amorphous intralesional hyperdensity.
Intralesional gas accumulations were present in 2/33 (6.1%) dogs, air bronchograms in 3 (9.1%), while none showed areas of mineralization.
In most dogs 32/33 (97%) a feeding vessel was clearly identified, and in 1 patient (3%), a lung lesion infiltrated a bronchial structure.
Pleural effusion was present in 1 patient (3%), while no dogs had pneumothorax.
In 11/33 dogs (33.3%), sternal lymphadenomegaly was present, graded as mild in 6/11, and moderate in 5/11. No other evidence of intrathoracic lymphadenopathy.
Computed tomography features of extrapulmonary metastases
The CT characteristics of extrapulmonary HSA metastases are summarized (Supplementary Material S1).
Between 2 and 10 or more than 10 extrapulmonary lesions were most frequently reported in the same organ, which were round-oval shaped and subcentimetric to greater than 1 cm in size; median maximal diameter ranged from 0.31 cm and 1.6 cm depending on the organ.
The margins were considered mostly well defined, with the exception of the spleen (72.7%), bones (100%), and brain (100%) where ill-defined margins were most frequently reported.
The median precontrast attenuation value of the metastatic lesions ranged between 27 and 48 HU, while the postcontrast attenuation value was between 44 and 109 HU. The pattern of enhancement was considered heterogeneous in most lesions, with an appearance predominantly characterized by spotty/linear to amorphous intralesional areas of strong hyperdensity, described as the spotty postcontrast linear to amorphous strong hyperdensity (SPLASH) sign (Figure 3), mixed with hypoattenuating intralesional areas compatible with necrosis/hemorrhage. This SPLASH sign was identified in a variable percentage of metastatic HSA lesions in this study, specifically: 12/14 (85.7%) of hepatic (Figure 4), 9/11 (81.8%) of splenic, 8/10 (80%) of muscular, 3/4 (75%) of renal, 4/6 (66.6%) of peritoneal/retroperitoneal, 2/3 (66.6%) of cutaneous/subcutaneous, 1/2 (50%) of adrenal, and 1/2 (50%) of encephalic. A SPLASH sign was not identified in any skeletal lesions.
A–C—Transverse postcontrast CT images showed 3 examples of heterogeneous extrapulmonary hemangiosarcoma metastases with the spotty postcontrast linear or amorphous strong hyperdensity (SPLASH) sign (white arrows) in 3 different dogs, involving the right transverse abdominal muscle (A), the liver (B), and (C) the spleen (C). CT images were acquired with 1.25-mm slice thickness, 100 to 120 kV, 200 to 220 mA, 0.562 pitch, and 512 X 512 matrix and reconstructed with a soft tissue algorithm.
Citation: American Journal of Veterinary Research 85, 12; 10.2460/ajvr.24.08.0219
Transverse postcontrast CT abdominal image showed multiple heterogeneous hemangiosarcoma hepatic metastases in a dog with the SPLASH sign (white arrows). CT images were acquired with 1.25-mm slice thickness, 100 to 120 kV, 200 to 220 mA, 0.562 pitch, and 512 X 512 matrix and reconstructed with a soft tissue algorithm. The black arrowhead indicates the caudal vena cava. E = Esophagus. GB = Gallbladder.
Citation: American Journal of Veterinary Research 85, 12; 10.2460/ajvr.24.08.0219
In 6/8 cases where a triple-phased angio-CT was performed, the predominant contrastographic features were a prevalent ring enhancement in the arterial phase and a centripetal progression of the enhancing areas in the portal and delayed phases. In the delayed phase, a slightly decreased intensity of these enhancing areas was noted. In these cases, the SPLASH sign was more intense and evident in the portal phase compared to the other phases.
Intralesional areas of mineralization were reported only in 1 patient with muscle metastases and 1 dog with peritoneal/retroperitoneal lesions, while no gas accumulation was reported in any of the metastatic lesions.
Skeletal lesions were identified in only 3/33 dogs, and all of these were exclusively characterized by bone lysis in the absence of clearly visible osteoproduction. In all of these dogs, at least 1 osseous lesion was located in the proximal humerus.
Abdominal lymphadenomegaly was observed in 5/33 dogs, which were considered mild in 4 and moderate in 1 patient. The medial iliac (3/5 [60%]) and portal (3/5 [60%]) lymph nodes were more commonly involved; these lymph nodes were not cytologically assessed.
Discussion
Based on the authors' review of the literature, this is the first study in veterinary medicine describing the CT features of lung metastases from HSA. We confirmed our first initial hypothesis as the majority of pulmonary metastatic lesions from HSA were found to be numerous (> 10) and small in size (miliary-subcentimetric). Moreover, other most commonly identified tomographic features of lung nodules in the present study were generalized distribution, often with halo sign and feeding vessels. In lung lesions in which a halo sign was not identified, the margins of the lesions were mostly well defined. These findings are in agreement with the human literature, where the radiological manifestations of metastatic pulmonary angiosarcoma are typically multiple and generalized nodules with sharp margins.28,29 This is related to the aggressive biological behavior of this neoplasia due to an intimate association with blood vessels, a typical hematogenous metastatic dissemination, and a possible transalveolar spread of neoplastic cells.1,6
Regarding the distribution of lung metastases, a statistically significant correlation between fewer (between 2 and 10) lung lesions and a more frequent peripheral distribution was found. This is also reported in human medicine and has been hypothesized to be a result of the hematogenous spread with the initial formation of neoplastic emboli at the most peripheral portions of the pulmonary vessels.24 In contrast, when numerous pulmonary nodules are present, they tend to have a random distribution, and the size of the nodules in the same patient may vary as a result of repeated episodes of embolization in the lungs or different growth rates.24
Another feature that supports the hematogenous origin of lung metastases from HSA is the presence of the feeding vessel, which is a pulmonary artery branch leading directly to a lung nodule.24 A feeding vessel was noted in the majority of the dogs included in this study (97%); however, it is also commonly reported in other neoplastic or nonneoplastic processes with an hematogenous spread (eg, septic emboli).30
The halo sign, which is an area of ground-glass attenuation surrounding a pulmonary nodule or mass, is a nonspecific feature reported in different types of lung lesions: pulmonary aspergillosis, hemorrhagic nodules, and inflammatory or neoplastic infiltration of the adjacent lung parenchyma.23 In the present study, a 72.7% prevalence of the halo sign was found, similar to findings reported in human patients with angiosarcoma pulmonary metastases (58% to 72.7%).28,29 The histological examination performed in 1 of the dogs included in this study with a halo sign showed only perilesional hemorrhage, without inflammation or neoplastic infiltration. This is in contrast with a previous veterinary study23 in which the halo sign was described and was unrelated to perilesional hemorrhage. However, only 1 dog was assessed, therefore it is impossible to draw any conclusions.
Most of the metastatic lung lesions included in this study were small (< 1 cm) and showed homogeneous enhancement (78.8%), which is in agreement with a recent human study showing that the enhancement pattern of HSA metastases is related to the nodule size. Lesions smaller than 1 cm are more often homogeneous, whereas nodules over 1 cm more frequently have an inhomogeneous enhancement.31
In human patients, another fairly frequently reported tomographic feature of HSA lung metastases is the presence of gaseous collections within the pulmonary nodules, with a prevalence of between 21 and 58%.28,32 This can be secondary to chemotherapy24 and/or associated with pneumothorax.28,32 In our study, only 2/33 (6.1%) dogs had gas-cavitated nodules, and no dogs were under chemotherapy or presented with pneumothorax. Various mechanisms have been proposed to explain the cavitation, including excavation of solid nodules due to tissue necrosis, infiltration of preexisting cavitary lesions, or communication with a bronchial/alveolar structure.32
Sternal lymphadenomegaly was present in 11/33 (33.3%). The majority of these dogs either had an abdominal primary HSA (9/11) or abdominal effusion (8/11). This is not surprising since sternal lymph nodes drain from different thoracic districts but also from the peritoneum,33 and abdominal HSA is reported as the second most frequent cause.34 Sternal node sampling was not performed; therefore, it is impossible to determine its inflammatory or metastatic nature.
The lungs were the only site of metastases in 30.3% of the dogs included in this study. In the remaining 23 dogs (69.7%), multiple metastases were found in extrathoracic organs, including the liver (42.4%), spleen (33.3%), muscles (30.3%), and peritoneum/retroperitoneum space (18.2%). This is in agreement with the literature and reflects the typical aggressive biological behavior of this neoplasia, with a rapid and widespread metastatic process.
The prevalence of muscular metastases in this study was slightly higher than previously reported (24.6%).19 This could be due to a selection bias in our study as we included dogs that already had metastatic lung disease, thus with a more advanced neoplastic process.
Similarly to lung metastases, extrapulmonary lesions were also more frequently numerous (2 to 10 or > 10) in the same organ and with well-defined margins. In addition, also our second initial hypothesis was confirmed as extrapulmonary metastases frequently had a heterogeneous appearance in the postcontrast scan, with areas of spotty to linear or more amorphous enhancement, located in the periphery or the center of the lesion, defined as the SPLASH sign. The SPLASH sign was identified with a higher frequency in the liver (85.7%), spleen (81.8%), muscle (80%), and kidney (75%) metastases. These results are aligned with a previous study including dogs with nonparenchymal HSA, where similar features were reported with a prevalence of 76%.4
The presence of these enhancing areas could be secondary to vascular channels resulting from neoangiogenesis, as previously reported.4 An alternative explanation could be that the contrast agent perfuses into the fluid-filled/hemorrhagic intralesional portions, whereas the necrotic or fibrous components correspond to the nonenhanced regions.37
The SPLASH sign, or comparable contrastographic characteristics, has also never been mentioned in relationship with other mesenchymal or nonmesenchymal neoplasia, and therefore, it could be useful to prioritize HSA in the differential diagnoses. Further studies comparing different neoplasia histotypes are needed to assess the actual peculiarity of this tomographic appearance.
In human medicine, the enhancement pattern of angiosarcoma primary and metastatic lesions, and more generally from tumors arising from lymphatics or blood vessels, is well described, and is characterized by evident ring enhancement in the arterial phase, continuous centripetal filling during the portal phase and partial washout in the delay phase.35–37 In the majority of dogs included in this study in which a triple-phased angio-CT was available, a similar perfusion pattern of the extrapulmonary metastases was observed (Figure 5). Secondarily to this enhancing pattern, the SPLASH sign was more evident in the portal phase. Due to the limited number of triple-phase angio-CTs available, a more detailed comparison between standard and triple-phase angio-CTs evaluating the visibility of the SPLASH sign or the progression of the enhancement pattern requires further studies. Furthermore, a clearer correlation between this tomographic appearance and the histological features of the metastases is necessary.
A–C—Transverse postcontrast CT abdominal image showed the enhancement progression of the same peritoneal hemangiosarcoma metastases in arterial (A), portal (B), and delayed (C) phases. Note the prevalent ring enhancement in the arterial phase, which progresses centripetally in the portal phase, and in the delayed phase, with a mild decrease of the intensity in the last phase. CT images were acquired with 1.25-mm slice thickness, 100 to 120 kV, 200 to 220 mA, 0.562 pitch, and 512 X 512 matrix and reconstructed with a soft tissue algorithm.
Citation: American Journal of Veterinary Research 85, 12; 10.2460/ajvr.24.08.0219
Intralesional mineralization was rare and observed only in extrapulmonary metastases, specifically in 1 case of muscular and peritoneal/retroperitoneal nodules. These results agree with previously published studies in veterinary medicine4 and human patients.28,31
Regarding the skeletal metastases, in all 3 dogs at least 1 osseous lesion was located in the proximal humerus. This is in line with a recent paper38 in which humeri were one of the most commonly affected metastatic sites from solid cancers.
This study has some limitations. Considering the inclusion criteria of this study, particularly the enrollment of dogs with pulmonary metastases, it is not possible to apply these prevalence data of pulmonary and extrapulmonary metastases to a general population of dogs with HSA. Including dogs already presenting with pulmonary metastases, and thus with a more advanced neoplastic condition, the prevalence of other metastases may be distorted by the stage of the oncologic disease.
Cytological or histological confirmation was not done in all the pulmonary and extrapulmonary metastatic lesions, especially when the lesion was difficult to sample because of the small size, problematic location, or due to the serious clinical condition of the patient. For this reason, especially some miliary pulmonary nodules may have been overinterpreted as metastases. However, in most cases, lung nodules were associated with other metastases in the same or in another more easily accessible location, which enabled sampling to be performed. In addition, due to its retrospective nature and the long time period in which dogs were included (11 years), a standardized angio-CT protocol was not available for most dogs.
Finally, the absence of a primary cardiac localization, particularly involving the right cardiac atrium or auricle, is attributed to the lack of histological confirmation in this study rather than to the actual low prevalence of this neoplastic localization or to a reduced propensity for distant metastasis.
In conclusion, lung HSA metastases were characterized by numerous, small (miliary-subcentimetric), and generalized nodules, commonly with the halo sign and feeding vessel. Areas with a SPLASH sign were frequently observed in postcontrast scans, especially in extrapulmonary metastases. The features described in this paper could help radiologists and clinicians to orient their diagnosis toward metastatic HSA, which could be useful especially if the lesions cannot be sampled. This study also confirms the important role of a whole-body CT in the staging of dogs with a primary mass compatible with HSA.
Supplementary Materials
Supplementary materials are posted online at the journal website:avmajournals.avma.org.
Acknowledgments
None reported.
Disclosures
The authors have nothing to disclose. No AI-assisted technologies were used in the generation of this manuscript.
Funding
The authors have nothing to disclose.
ORCID
M. Mattolini https://orcid.org/0000-0002-0065-2952
S. Citi https://orcid.org/0000-0001-8211-9248
R. Franchi https://orcid.org/0000-0002-6797-397X
V. Meucci https://orcid.org/0000-0001-5672-7856
G. Carozzi https://orcid.org/0009-0003-3281-1414
B. Gianni https://orcid.org/0009-0006-8925-3291
E. Caleri https://orcid.org/0009-0005-7951-3726
References
- 1.↑
Vail DM, Thamm DH, Liptak JM. Withrow and MacEwen's Small Animal Clinical Oncology. Elsevier Health Sciences; 2019.
- 2.↑
Ettinger SJ, Feldman EC, Cote E. Textbook of Veterinary Internal Medicine. Elsevier Health Sciences; 2016.
- 3.↑
Schultheiss PC. A retrospective study of visceral and nonvisceral hemangiosarcoma and hemangiomas in domestic animals. J Vet Diagn Invest. 2004;16(6):522–526. doi:10.1177/104063870401600606
- 4.↑
Fukuda S, Kobayashi T, Robertson ID, et al. Computed tomographic features of canine nonparenchymal hemangiosarcoma. Vet Radiol Ultrasound. 2014;55(4):374–379. doi:10.1111/vru.12136
- 5.↑
Sherwood JM, Haynes AM, Klocke E, et al. Occurrence and clinicopathologic features of splenic neoplasia based on body weight: 325 dogs (2003–2013). J Am Anim Hosp Assoc. 2016;52(4):220–226. doi:10.5326/JAAHA-MS-6346
- 6.↑
Griffin MA, Culp WTN, Rebhun RB. Canine and feline haemangiosarcoma. Vet Rec. 2021;189(9). doi:10.1002/vetr.585
- 7.↑
Mattoon JS, Sellon RK, Berry CR. Small Animal Diagnostic Ultrasound. Elsevier Health Sciences; 2020.
- 8.
Penninck D, d’Anjou MA. Atlas of Small Animal Ultrasonography. John Wiley & Sons; 2015.
- 9.↑
Cuccovillo A, Lamb CR. Cellulare features of sonographic target lesions of the liver and spleen in 21 dogs and a cat. Vet Radiol Ultrasound. 2002;43(3):275–278. doi:10.1111/j.1740-8261.2002.tb01003.x
- 10.↑
Fabbi M, Di Palma S, Manfredi S, et al. Imaging diagnosis—ultrasonographic appearance of skeletal muscle metastases in a dog with hemangiosarcoma. Vet Radiol Ultrasound. 2017;58(6):E64–E67. doi:10.1111/vru.12432
- 11.↑
Lamb CR, Whitlock J, Foster-Yeow ATL. Prevalence of pulmonary nodules in dogs with malignant neoplasia as determined by CT. Vet Radiol Ultrasound. 2019;60(3):300–305. doi:10.1111/vru.12723
- 12.↑
Holt D, Van Winkle T, Schelling C, Prymak C. Correlation between thoracic radiographs and postmortem findings in dogs with hemangiosarcoma: 77 cases (1984–1989). J Am Vet Med Assoc. 1992;200(10):1535–1539.
- 14.↑
Reese DJ, Green EM, Zekas LJ, et al. Intra- and interobserver variability of board-certified veterinary radiologists and veterinary general practitioners for pulmonary nodule detection in standard and inverted display mode images of digital thoracic radiographs of dogs. J Am Vet Med Assoc. 2011;238(8):998–1003. doi:10.2460/javma.238.8.998
- 15.↑
Nemanic S, London CA, Wisner ER. Comparison of thoracic radiographs and single breath-hold helical CT for detection of pulmonary nodules in dogs with metastatic neoplasia. J Vet Intern Med. 2006;20(3):508–515. doi:10.1111/j.1939-1676.2006.tb02889.x
- 16.↑
Armbrust LJ, Biller DS, Bamford A, Chun R, Garrett LD, Sanderson MW. Comparison of three-view thoracic radiography and computed tomography for detection of pulmonary nodules in dogs with neoplasia. J Am Vet Med Assoc. 2012;240(9):1088–1094. doi:10.2460/javma.240.9.1088
- 17.↑
Marolf AJ. Diagnostic imaging of the hepatobiliary system. Vet Clin North Am Small Anim Pract. 2017;47(3):555–568. doi:10.1016/j.cvsm.2016.11.006
- 18.↑
Fields EL, Robertson ID, Osborne JA, Brown JC. Comparison of abdominal computed tomography and abdominal ultrasound in sedated dogs. Vet Radiol Ultrasound. 2012;53(5):513–517. doi:10.1111/j.1740-8261.2012.01949.x
- 19.↑
Carloni A, Terragni R, Morselli-Labate AM, et al. Prevalence, distribution, and clinical characteristics of hemangiosarcoma-associated skeletal muscle metastases in 61 dogs: a whole body computed tomographic study. J Vet Intern Med. 2019;33(2):812–819. doi:10.1111/jvim.15456
- 20.
Dennler M, Lange EM, Schmied O, Kaser-Hotz B. Imaging diagnosis—metastatic hemangiosarcoma causing cerebral hemorrhage in a dog. Vet Radiol Ultrasound. 2007;48(2):138–140. doi:10.1111/j.1740-8261.2007.00220.x
- 21.↑
Weston PJ, Baines SJ, Finotello R, Mortier JR. Clinical, CT, and ultrasonographic features of canine and feline pleural and peritoneal carcinomatosis and sarcomatosis. Vet Radiol Ultrasound. 2021;62(3):331–341. doi:10.1111/vru.12951
- 22.↑
Bertazzolo W, Dell’Orco M, Bonfanti U, Ghisleni G, Caniatti M, Masserdotti C, Antoniazzi E, Crippa L, Roccabianca P. Canine angiosarcoma: cytologic, histologic, and immunohistochemical correlations. Vet Clin Pathol. 2005;34(1):28–34. doi:10.1111/j.1939-165x.2005.tb00005.x
- 23.↑
Thierry F, Chau J, Makara M, et al. Vascular conspicuity differs among injection protocols and scanner types for canine multiphasic abdominal computed tomographic angiography. Vet Radiol Ultrasound. 2018;59(6):677–686. doi:10.1111/vru.12679
- 24.↑
Secrest S, Sakamoto K. Halo and reverse halo signs in canine pulmonary computed tomography. Vet Radiol Ultrasound. 2014;55(3):272–277. doi:10.1111/vru.12132
- 25.↑
Shroff GS, Benveniste MF, Carter BW, et al. Imaging of metastases in the chest: mechanisms of spread and potential pitfalls. In: Seminars in Ultrasound, CT and MRI. Vol 38. Elsevier; 2017:594–603. doi:10.1053/j.sult.2017.07.007
- 26.↑
Hornby NL, Lamb CR. Does the computed tomographic appearance of the lung differ between young and old dogs? Vet Radiol Ultrasound. 2017;58(6):647–652. doi:10.1111/vru.12532
- 27.↑
Kayanuma H, Yamada K, Maruo T, Kanai E. Computed tomography of thoracic lymph nodes in 100 dogs with no abnormalities in the dominated area. J Vet Med Sci. 2020;82(3):279–285. doi:10.1292/jvms.19-0413
- 28.↑
Yogi A, Miyara T, Ogawa K, et al. Pulmonary metastases from angiosarcoma: a spectrum of CT findings. Acta Radiol. 2016;57(1):41–46. doi:10.1177/0284185115571789
- 29.↑
Wang H, Shi J, Liu H, et al. Clinical and diagnostic features of angiosarcoma with pulmonary metastases: a retrospective observational study. Medicine (Baltimore). 2017;96(36):e8033. doi:10.1097/MD.0000000000008033
- 30.↑
Algin O, Gokalp G, Topal U. Signs in chest imaging: a pictorial review. Diagn Interv Radiol. 2011;17:18–29. doi:10.4261/1305-3825.DIR.2901-09.1
- 31.↑
Chen Y, He X, Shang J, et al. CT findings of pulmonary metastases from primary cardiac angiosarcoma. Curr Med Imaging Former Curr Med Imaging Rev. 2021;17(10):1216–1220. doi:10.2174/1573405617666210521151753
- 32.↑
Tateishi U, Hasegawa T, Kusumoto M, et al. Metastatic angiosarcoma of the lung: spectrum of CT findings. Am J Roentgenol. 2003;180(6):1671–1674. doi:10.2214/ajr.180.6.1801671
- 33.↑
Baum H. The lymphatic system of the dog. University of Saskatchewan. 2022. Accessed July 17, 2023. https://openpress.usask.ca/k9lymphaticsystem/
- 34.↑
Smith K, O’Brien R. Radiographic characterization of enlarged sternal lymph nodes in 71 dogs and 13 cats. J Am Anim Hosp Assoc. 2012;48(3):176–181. doi:10.5326/JAAHA-MS-5750
- 35.↑
Yu JF, Cui H, Ji GM, et al. Clinical and imaging manifestations of primary cardiac angiosarcoma. BMC Med Imaging. 2019;19(1):16. doi:10.1186/s12880-019-0318-4
- 36.
Yi LL, Zhang JX, Zhou SG, et al. CT and MRI studies of hepatic angiosarcoma. Clin Radiol. 2019;74(5):406.e1–406.e8. doi:10.1016/j.crad.2018.12.013
- 37.↑
Zhou Y, Hou P, Wang F, Li B, Gao J. Primary hepatic malignant vascular tumors: a follow-up study of imaging characteristics and clinicopathological features. Cancer Imaging. 2020;20(1):59. doi:10.1186/s40644-020-00336-9
- 38.↑
Agnoli C, Sabattini S, Ubiali A, et al. A retrospective study on bone metastasis in dogs with advanced-stage solid cancer. J Small Anim Pract. 2023;64(9):561–567. doi:10.1111/jsap.13621
