Materials and Methods

Case selection—Dogs with pericardial effusion were identified by searching the University of California-Davis Veterinary Medical Teaching Hospital database from 1985 through 2006. All dogs had an echocardiogram performed by a board-certified veterinary cardiologist or a supervised cardiology resident. All dogs were required to have at least mild pericardial effusion identified by use of echocardiography to be included in the study, and dogs with echocardiographic evidence of a cardiac mass without pericardial effusion were not included. Dogs with mild or greater pericardial effusion identified by echocardiography were included in the study when pericardectomy or necropsy was performed during that visit or a subsequent visit to confirm the cause of the pericardial effusion. Dogs undergoing thoracoscopic pericardectomy were excluded because of an inability to view the entire heart during that procedure. Diagnosis of the neoplastic cause in all dogs with cardiac or pericardial masses was made by histologic evaluation of biopsy specimens obtained during surgery or necropsy. Except for dogs with infective pericarditis, diagnosis of nonneoplastic causes of pericardial effusion was also made histologically.

Medical records review—Multiple echocardiographic views were obtained to thoroughly evaluate dogs for cardiac masses; those views included right parasternal short-axis and long-axis views, left apical views, and, most importantly, the left cranial parasternal long-axis view that provides the best images of the right atrium. Three ultrasound machines were used during the study period (machine A,^a 1985 through 1991; machine B,^b 1991 through 2000; and machine C,^c 2000 through 2006). Echocardiographic characterization of the cause of the pericardial effusion included the categories right atrial-auricular mass, heart base mass, pericardial mass, other location of a cardiac mass (if applicable), or no evidence of a mass.

Sensitivity and specificity were calculated on the basis of the first echocardiographic examination for echocardiographic detection of a cardiac mass versus no mass, for detection of a heart base mass versus other causes, and for detection of a right atrial mass versus other causes. To assess whether there was improved diagnostic capability of echocardiography during serial examinations, sensitivity and specificity of echocardiography for detection of a cardiac mass versus another cause were also calculated by use of findings from the final echocardiographic examination in dogs examined multiple times. Subjective evaluation for evidence of cardiac tamponade was performed by use of 2-D echocardiography, and cardiac tamponade was diagnosed when there was diastolic collapse of the right atrium, right ventricle, or both in addition to pericardial effusion.

Clinical characteristics were recorded, which included signalment and body weight. Thoracic radiographs were obtained for all dogs and were characterized subjectively by the presence or absence of a globoid cardiac silhouette, presence or absence of a cardiac mass effect, and presence or absence of metastatic pulmonary disease. Right-sided heart failure was identified and characterized as ascites, pleural effusion, or both and was confirmed by use of ultrasonography. Standard practice included obtaining an ECG when there was an arrhythmia detected during physical examination. Clinicopathologic tests performed on all dogs included a CBC and serum biochemical analysis. Pericardial fluid was analyzed in some dogs, which included cytologic evaluation and possibly aerobic and anaerobic microbial culture.

Statistical analysis—The Fisher exact test was used to compare the rate of metastasis for neoplastic causes and to compare the rates of bicavitary effusion, pleural effusion, or ascites between neoplastic and nonneoplastic causes. Significance was defined as values of P < 0.05. Survival data were also obtained from review of medical records, although many dogs were lost to follow-up monitoring. Dogs that were lost to follow-up monitoring were censored at the date of the last medical entry. Kaplan-Meier curves were generated for the most common causes of pericardial effusion. The Tarone-Ware rank test was used to determine significance. Median survival times were calculated by use of nonparametric survival analysis. Student t tests were used to compare ages of dogs with neoplastic and nonneoplastic causes of pericardial effusion. Metastatic rate was calculated in dogs that underwent complete necropsies.

Results

Review of medical records resulted in the inclusion of 107 dogs in the study. Of these, 48 underwent pericardectomy (via thoracotomy), 44 underwent necropsy, and 15 underwent both procedures. Surgery or necropsy was used to identify 66 dogs with a cardiac mass and 41 dogs without a mass. Location of masses included 38 right atrial masses, 23 heart base masses, 2 concurrent right atrial and heart base masses, 2 pericardial masses, and 1 right ventricular mass (Table 1). Echocardiography identified 53 of 66 dogs with cardiac masses, including 32 of 38 dogs with right atrial masses and 17 of 23 dogs with heart base masses. Sensitivity and specificity of echocardiography for detection of a cardiac mass were 80% and 100%, respectively, and echocardiography had equal sensitivity and specificity for distinguishing right atrial masses from all other causes (84% and 100%, respectively) or for distinguishing heart base masses from all other causes (74% and 98%, respectively).

Nineteen dogs had > 1 echocardiographic evaluation, including 11 dogs with cardiac masses and 8 dogs without cardiac masses. Of the 12 dogs in which use of echocardiography did not identify masses on initial examination, only 4 dogs had repeat echocardiography. Repeat echocardiography on those 4 dogs identified cardiac masses, which increased the sensitivity of echocardiography for detection of a cardiac mass to 88%. The interval from the initial echocardiography to repeat echocardiography ranged from 2 days to 7 months (median, 75 days). Eight of 12 dogs with masses that were not identified during the initial echocardiography had a small volume of pericardial effusion at the time of the initial echocardiographic examination.

Specific causes of pericardial effusion were determined. Hemangiosarcoma was the most common cause of pericardial effusion (n = 36 dogs), followed by idiopathic pericarditis (21), mesothelioma (15), chemodectoma (9), thyroid gland adenocarcinoma (6), infective pericarditis (5), lymphoma (3), sarcoma (2), carcinomatosis (1), ruptured left atrium secondary to severe mitral valve regurgitation (1), sterile foreign body (1), and granuloma (1). Biopsy specimens and histologic evaluation were not performed for 6 dogs with heart base masses that underwent surgery (thoracotomy) for pericardectomy. One dog had a concurrent right atrial hemangiosarcoma and a neuroendocrine heart base mass.

Neoplasia was the cause of pericardial effusion in a majority (76/107 [71%]) of dogs. Five of 15 dogs with mesothelioma had discrete cardiac masses, most often of the heart base (4/5) and rarely of the right atrium (1/5). A majority (35/40 [88%]) of right atrial masses were hemangiosarcoma, followed by 1 (3%) each of neuroendocrine tumor, thyroid gland adenocarcinoma, mesothelioma, lymphoma, and sarcoma. Heart base masses were most often neuroendocrine tumors (9/23 [39.1%] dogs), followed by thyroid gland adenocarcinoma (6/23 [26.1%]), mesothelioma (5/23 [21.7%]), and hemangiosarcoma (3/23 [13%]). Mesothelioma heart base masses were apparent as discrete masses on echocardiography. An undifferentiated sarcoma was the cause of the right ventricular mass. Infective pericarditis in 3 of 5 dogs was caused by foreign bodies (ie, foxtail awns) with secondary bacterial infections.

Complete necropsy was performed on 59 dogs (44 necropsy alone and 15 pericardectomy and necropsy). The metastatic rate did not differ significantly (P = 1.00 for all comparisons) between dogs with hemangiosarcoma (19/28 [67.9%]), mesothelioma (5/9 [55.6%]), thyroid gland adenocarcinoma (2/3 [66.7%]), or chemodectoma (4/6 [66.7%]). Eight of 28 (28.6%) dogs with cardiac hemangiosarcoma also had splenic hemangiosarcoma. All dogs with lymphoma had metastatic disease. The most common site of metastasis for all neoplastic dogs was the lungs (18/59 [30.5%]). In dogs with cardiac hemangiosarcoma, the most common sites of metastasis included the lungs (12/28 [42.9%]), spleen (8/28 [28.6%]), liver (8/28 [28.6%]), and kidneys (4/28 [14.3%]). The most common sites of metastasis in dogs with mesothelioma included the intrathoracic lymph nodes (6/9 [66.7%]), lungs (2/9 [22.2%]), and pleurae (2/9 [22.2%]). In dogs with neuroendocrine tumors, the most common site of metastasis was the lungs (3/6 [50%]), followed by the spleen (1/6 [16.7%]) and liver (1/6 [16.7%]). The most common site of metastasis for dogs with thyroid gland adenocarcinoma was the pericardium (2/3 [66.7%]), followed by metastasis to the lungs (1/3 [33.3%]) and myocardium (1/3 [33.3%]) and transcoelomic metastasis (1/3 [33.3%]).

Most dogs with pericardial effusion were largebreed dogs; body weight ranged from 3.6 to 73.0 kg (7.9 to 160.6 lb), with a median of 31.5 kg (69.3 lb). The most common breeds included Golden Retriever (20/107 [18.7%] dogs), Labrador Retriever (16/107 [14.9%]), and German Shepherd Dog (5/107 [4.7%]). Dogs with cardiac masses were significantly (P = 0.006) older (mean, 9.7 years) than dogs without masses (mean, 7.9 years).

Sixty-seven (62.6%) dogs with pericardial effusion had evidence of right-sided heart failure. Most dogs (36/107 [33.6%]) had concurrent pleural effusion and ascites, with fewer dogs with isolated ascites (17/107 [15.9%]) or pleural effusion (14/107 [13.1%]). There was no difference between neoplastic and nonneoplastic causes for bicavitary effusion (P = 1.00), pleural effusion (P = 0.84), or ascites (P = 0.93). Cardiac tamponade was subjectively suspected on the basis of results of echocardiography in 42 of 107 (39%) dogs. A globoid-shaped heart was identified in 56 of 107 (52.3%) dogs during examination of thoracic radiographs, which yielded a poor sensitivity for detection of pericardial effusion. Pulmonary metastases were only identified during examination of thoracic radiographs in 7 of 21 (33.3%) dogs with confirmed pulmonary metastases during necropsy or pericardectomy. There was a suspicion of a right atrial or heart base mass during examination of thoracic radiographs in 10 of 63 dogs with masses, with a specificity of 100%.

The most common ECG abnormalities included electrical alternans (30/107 [28.0%] dogs), sinus tachycardia (30/107 [28.0%]), dampened QRS complexes (ie, < 1 mV; 26/107 [24.3%]), and ventricular arrhythmia (14/107 [13.1%]). Other less common abnormalities included supraventricular tachycardia (3/107 [2.8%] dogs), atrial premature complexes (2/107 [1.9%]), atrial fibrillation (2/107 [1.9%]), increase in ST segment (2/107 [1.9%]), high-grade second-degree atrioventricular block (1/107 [0.9%]), and right bundle-branch block (1/107 [0.9%]).

Pericardial effusion was analyzed in 47 of 107 dogs. Fluid analysis identified the cause of the pericardial effusion in 6 of 47 (12.8%) dogs, which included 5 dogs with infective pericarditis and 1 dog with lymphoma. Most pericardial effusion was classified as hemorrhagic fluid (40/47 [85%] dogs), followed by suppurative inflammation (6/47 [12.8%]), pyogranulomatous inflammation (4/47 [8.5%]), a modified transudate (2/47 [4.3%]), and chylous effusion (1/47 [2.1%]). Mesothelial reactivity was identified in 25 of 47 (53.2%) dogs. Neoplastic cells were identified in 1 dog each with lymphoma, mesothelioma, and idiopathic pericarditis.

Cytologic analysis identified Coccidioides organisms in a dog with coccidiodomycosis. Four of 47 (8.5%) dogs had positive results for bacterial culture, including 1 each of Bacteroides spp, Actinomyces spp, Streptococcus canis, and a mixed infection consisting of Pasteurella spp, Peptostreptococcus spp, and Bacteroides spp.

Discordant diagnostic results were evident in several instances. One dog with idiopathic pericarditis was classified with a mesothelioma on the basis of cytologic analysis of pericardial fluid. One dog with a chemodectoma had purulent pericardial effusion. Actinomyces spp were cultured from a dog with histologically confirmed mesothelioma. Histologic examination of the pericardium incorrectly identified 1 dog with mesothelioma that had idiopathic pericarditis and epicarditis identified 2 years later during histologic examination of samples obtained during a second pericardectomy as well as during necropsy shortly thereafter.

Results of hematologic and serum biochemical analysis were available for review in 90 dogs with pericardial effusion. Anemia was the most common abnormality detected (23/90 [25.6%] dogs). Anemia was mild (PCV, 30% to 37%), normochromic, normocytic, and nonregenerative in all dogs, except for 2 in which the PCV was < 30%. Morphological RBC abnormalities were not detected in any of the dogs. Anemia was most commonly detected in dogs with hemangiosarcoma (n = 8 dogs), pericarditis (6), and mesothelioma (4). Increases in activities of liver enzymes were detected in 21 of 90 (23.3%) dogs; those increases were all considered mild. High activities of liver enzymes were detected in dogs with hemangiosarcoma (n = 9), chemodectoma (5), pericarditis (4), nondefined heart base masses (2), and thyroid gland adenocarcinoma (1). Thrombocytopenia (< 200,000 platelets/μL) was the third most common laboratory abnormality evident and was found in 12 of 90 (13.3%) dogs. Thrombocytopenia was detected in dogs with hemangiosarcoma (n = 6), pericarditis (4), and lymphosarcoma (2). Other uncommon abnormalities included mild azotemia (4/90 dogs); slight prolongation of the prothrombin time, partial thromboplastin time, or both (3); mild hypercalcemia (3); and mild hyperkalemia (2). In dogs with the 3 most commonly detected hematologic or biochemical abnormalities (anemia, high activities of liver enzymes, or thrombocytopenia), the most common cause of pericardial effusion was hemangiosarcoma.

Changes in distribution of causes of pericardial effusion were evaluated by separating the groups into 2 time periods (42 dogs from 1985 through 1995 and 65 dogs from 1996 through 2006). All 5 dogs with infective pericarditis were identified in the first time period, whereas 6 of 42 (14.3%) dogs with idiopathic pericarditis were identified in the first time period and 15 of 65 (23.1%) dogs with idiopathic pericarditis were identified in the second time period. The proportion of dogs with mesothelioma increased from 3 of 42 (7.1%) dogs in the first time period to 12 of 65 (18.5%) dogs in the second time period. In contrast, 6 of 42 (14.3%) dogs in the first time period had chemodectomas, compared with only 3 of 65 (4.6%) dogs in the second time period. The proportions for all other causes were similar between the 2 time periods. Use of the Fisher exact test on a 2-way data contingency table revealed no significant difference in frequency of diagnosis of mesothelioma (P = 0.09), chemodectoma (P = 0.15), or idiopathic pericarditis (P = 0.32) between the 2 time periods. To assess whether there was improvement in diagnosis of cardiac masses in the later time period as a result of improvements in temporal and spatial resolution of ultrasound machines, the number of dogs in which use of echocardiography did not identify the mass was compared between the 2 time periods. There was no difference in the percentage of masses that were not identified between the early period (masses not identified in 5/42 [11.9%] dogs), compared with the percentage for the later period (masses not identified in 7/65 [10.8%] dogs).

On the basis of echocardiographic classification, dogs with no cardiac mass lived significantly (P < 0.001) longer (median survival time, 10.10 months) than did dogs with echocardiographic evidence of a cardiac mass (median survival time, 0.53 months; Figure 1). Dogs with a heart base mass diagnosed by use of echocardiography lived significantly (P < 0.001) longer (median survival time, 5.17 months) than did dogs with a right atrial mass diagnosed by use of echocardiography (median survival time, 0.03 months). All deaths recorded were the result of the specific cause of the pericardial effusion and not the result of other systemic disease. Regarding specific causes of pericardial effusion, dogs with nonneoplastic causes lived significantly (P < 0.001) longer (median survival time, 24.83 months) than did dogs with neoplastic causes (median survival time, 0.63 months). Dogs with hemangiosarcoma had a significantly (P < 0.001) shorter median survival time (0.07 months) than did dogs with all other neoplastic causes combined (5.17 months). Median survival time of dogs with mesothelioma (6.50 months) did not differ significantly (P = 0.51) from that of dogs with heart base masses, which included chemodectoma, ectopic thyroid gland tissue, or nonspecific causes (5.17 months).

View Full Size

Kaplan-Meier survival curves for dogs with pericardial effusion attributable to various causes. A—Dogs with pericardial effusion and no echocardiographic evidence of a cardiac mass (squares) survived significantly (P < 0.001) longer than did dogs with echocardiographic evidence of a cardiac mass (circles). B—Dogs with a heart base mass diagnosed by use of echocardiography (squares) survived significantly (P = 0.002) longer than did dogs with a right atrial mass diagnosed by use of echocardiography (circles). C—Dogs with a nonneoplastic cause of pericardial effusion (as determined on the basis of histologic examination; squares) survived significantly (P < 0.001) longer than did dogs with a neoplastic cause of pericardial effusion (circles). D—Dogs with hemangiosarcoma (as determined on the basis of histologic examination; circles) had a significantly (P < 0.001) shorter median survival time than did dogs with all other neoplastic causes combined (squares). Notice that the scale for the x-axis differs among panels.

Citation: Journal of the American Veterinary Medical Association 235, 12; 10.2460/javma.235.12.1456

• Download Figure
• Download figure as PowerPoint slide

Discussion

In the study reported here, echocardiography performed by a board-certified veterinary cardiologist or supervised cardiology resident had high sensitivity (82%) and specificity (100%) for diagnosis of a cardiac mass in dogs with pericardial effusion. Echocardiography also had high sensitivity and specificity for differentiating heart base masses from other causes of pericardial effusion and right atrial masses from other causes of pericardial effusion. Repeat echocardiographic examinations increased the sensitivity for detection of cardiac masses from 80% to 88%. Of the dogs in which use of echocardiography did not identify a mass, masses were detected in all dogs that had repeat echocardiographic examinations (n = 4). Unfortunately, many dogs did not have repeat echocardiographic examinations, which could have aided in detection of masses in the 12 dogs in which a mass was not echocardiographically identified. Most masses that were not detected by use of echocardiography were in dogs with a small volume of pericardial effusion.

Accurate diagnosis of a cardiac mass and defining the location of the cardiac mass can yield important prognostic information and be useful when determining the therapeutic plan. In the study reported here, dogs with echocardiographic evidence of a cardiac mass did not survive as long as dogs without a mass detected by use of echocardiography. More importantly, dogs with an echocardiographic diagnosis of a right atrial mass did not survive as long as dogs with echocardiographic evidence of a heart base mass, irrespective of histologic characterization. It is important to distinguish right atrial masses from heart base masses because dogs with heart base masses may benefit from partial pericardectomy to improve survival time.^2,7,8 Regarding specific causes of pericardial effusion, hemangiosarcoma conferred the gravest prognosis, with a shorter survival time than that for other neoplastic conditions.

The 2 most common causes of pericardial effusion in this study were hemangiosarcoma (36/107 [33.6%] dogs) and idiopathic pericarditis (21/107 [19.6%]), which is consistent with results in other reports.^1,4 Mesothelioma was also an important cause of pericardial effusion in this series of dogs (15/107 [14%]) and was more common than the proportion for chemodectomas (9/107 [8.4%]) or thyroid gland adenocarcinomas (6/107 [5.6%]). In contrast, investigators in an earlier study⁴ in 1984 did not detect any dogs with pericardial effusion caused by mesothelioma and reported that 12% of cases were caused by chemodectoma. Surprisingly, 5 of 15 mesotheliomas caused discrete cardiac masses, with the majority (4/5) being heart base masses and the remaining 1 being a right atrial mass. Another unexpected finding was the infrequency (1 dog) of misdiagnosis of infective pericarditis as mesothelioma on the basis on abnormal mesothelial cell morphology during cytologic analysis of pericardial fluid, compared with results in an earlier study¹² in which investigators found that 13% of nonneoplastic cases were falsely reported as neoplastic. Consistent with results of other reports,^4,7,8 the most common cause of heart base tumors in the dogs of our study was chemodectoma (9/23 [39.1%]), which was followed by thyroid gland adenocarcinoma (6/23 [26.1%]). In addition, 3 of 23 (13%) heart base masses were caused by hemangiosarcoma, which would confer a graver prognosis than for the typical causes of heart base masses.

In contrast to findings in another report,¹ the metastatic rate of the various neoplastic causes (hemangiosarcoma, mesothelioma, chemodectoma, and thyroid gland adenocarcinoma) did not differ significantly. Metastatic rates for chemodectomas, thyroid gland adenocarcinomas, and mesotheliomas were quite high (50% to 66%) in the study reported here. The lungs were the most common site for metastasis of all neoplastic causes combined. Detection of pulmonary metastases during thoracic radiography had low sensitivity because only 7 of 21 (33.3%) dogs with pulmonary metastases were identified by use of radiography. It was surprising that the rate of bicavitary effusion, pleural effusion, or ascites did not differ between neoplastic causes and idiopathic pericarditis. A possible explanation is that both neoplastic and nonneoplastic causes of pericardial effusion often lead to cardiac tamponade and subsequent right-sided heart failure.

Most of the discordant diagnostic test results were for the differentiation of idiopathic pericarditis from mesothelioma. Mesothelial cell reactivity was diagnosed during cytologic analysis of pericardial effusion in approximately half of all dogs, which does not discriminate between idiopathic pericarditis and mesothelioma. Although histologic examination of pericardium is the current standard used for differentiation of idiopathic pericarditis from mesothelioma, there can also be discrepancies in diagnosis with this technique (3 dogs were incorrectly identified). Use of special immunohistochemical stains may help clinicians and researchers better discriminate between these 2 diseases.

Study limitations included the retrospective design, which has inherent limitations. A small number of dogs, in which use of echocardiography did not identify a cardiac mass, did not have pericardectomy or necropsy performed within a short time after the echocardiographic examination and did not have repeat echocardiographic assessment for a mass. Therefore, it is possible that the sensitivity of echocardiography for detection of masses was higher than reflected in the study reported here. Given the retrospective nature of the study, many dogs were lost to follow-up monitoring and were treated as censored variables in the Kaplan-Meier survival analysis. Euthanasia may confound assessment of survival rates because there may be differences in owner commitment for continued treatment of dogs with cardiac masses that have a poor prognosis.

In the study reported here, echocardiography performed by a board-certified veterinary cardiologist or supervised cardiology resident had high sensitivity and specificity for detection of a cardiac mass and for discriminating between heart base masses and right atrial masses in dogs with pericardial effusion. Repeat echocardiographic examinations improved the sensitivity for diagnosis of a cardiac mass. All neoplastic causes of pericardial effusion had high metastatic rates of 50% to 66%, and the metastatic rates did not differ between the specific neoplastic causes.