Echocardiographic and surgical correlation of pericardial effusions in patients with malignant disease

Malignant disease is present in the pericardium of 15% to 20% of patients dying of cancer and examined post mortem, and malignancy is the most common cause of cardiac tamponade in patients without previously suspected cancer. Currently, the most sensitive and the least invasive method for detecting pericardial effusions is echocardiography, which can detect effusions as small as 15 ml, characterize total volume, and demonstrate the presence of fibrin bands, pericardial masses, loculations, thickened pericardial membranes, chamber compression, and dilatation of the inferior vena cava. 3 However, it has not been clearly established whether such echocardiographically obtained information can characterize a pericardial effusion as being malignant or benign. Similarly, although needle pericardiocentesis is effective in draining pericardial effusions, it is unknown whether the additional information gained by histologic examination of pericardial tissue obtained during the performance of a pericardial window will increase the frequency of detection of malignant effusions. Defining the disease processes leading to an effusion is critical, because both overall prognosis and the treatment choice may be altered, and for a marginally stable patient with a potentially limited outlook, extensively invasive intervention may be inappropriate. Therefore, this retrospective study attempts (1) to correlate various echocardiographic signs with pathologic specimens to determine whether echocardiography may reliably characterize an effusion as malignant or benign and (2) to examine the sensitivity and specificity of pericardial fluid cytology and pericardial tissue histology in establishing the presence of intrapericardial malignant disease. The records of 25 patients seen at Memorial SloanKettering Cancer Center during the period of January 1, 1993, to December 30, 1994, with a history of malignant disease, transthoracic echocardiograms demonstrating pericardial fluid, and subsequent pericardial fluid and tissue submitted for analysis were reviewed. All patients were grouped into one of two categories: those with malignant effusions and those with benign effusions. Intrapericardial malignancy was diagnosed when cytologic examination of pericardial fluid and/or pathologic examination of pericardial tissue demonstrated malignant cells. In addition, patient records were reviewed for age, sex, primary cancer diagnosis, prior thoracic irradiation, cardiac function, and relevant laboratory findings. All echocardiograms were again reviewed by one author (D.K.), blinded to cytologic and pathologic data, for the following: size of effusion; degree of collapse of the right atrium; right ventricle, or left atrium, dilatation of the inferior vena cava; and the presence of fibrin bands, loculations, or pericardial masses. Ten men and 15 women, with a median age of 59 years (range 28 to 72 years), were studied. Effusions were confirmed by pericardial sac histology or pericardial fluid cytology as malignant in 13 of 25 patients (52%). The distribution of primary malignant tumors and the causes of the effusions are listed in Table I. Of patients with intrapericardial malignant disease, eight had positive cytology but negative pathology (lung carcinoma, n 5 5; esophageal carcinoma, n 5 1; sarcoma, n 5 1; adenocarcinoma of unknown primary, n 5 1), two had negative cytology but positive pathology (metastatic breast carcinoma, n 5 1; mesothelioma, n 5 1), and three had both positive cytology and pathology (lung carcinoma, n 5 1; breast carcinoma, n 5 1; adenocarcinoma of unknown primary, n 5 1). Only one patient had received prior thoracic irradiation, and this patient was found to have both benign cytology and benign pathology. Cultures were performed on pericardial fluid from five patients with