Predictors of pleural malignancy in patients with pleural effusion undergoing thoracoscopy.

STUDY OBJECTIVES Thoracoscopic pleural biopsy is highly accurate in the diagnosis of pleural malignancy. However, no scientific evidence is currently available to guide the physician's decision as to when and in which patients with pleural effusion thoracoscopy is indicated. The application of predictive criteria of malignancy might improve the indication of thoracoscopy in patients with undiagnosed pleural effusion. METHODS Prospective study of 93 patients referred for thoracoscopy at a tertiary hospital. Clinical variables were obtained prior to thoracoscopy by clinical history and review of previous data, patient interview, and physical examination. Radiologic variables were obtained by evaluation of chest radiograph and chest CT images by two independent readers. After thoracoscopy, all patients without a diagnosis were sent for long-term follow-up. RESULTS Thoracoscopy demonstrated 94% sensitivity and 100% specificity in the diagnosis of pleural malignancy. Variables, which in a multivariate model are associated with pleural malignancy, include a symptomatic period > 1 month, absence of fever, blood-tinged pleural fluid, and chest CT scan findings suggestive of malignancy. Receiver operating characteristic analysis showed that the use of these four criteria offered adequate classification in 95% of patients. Twenty-eight patients had all four criteria, and all had malignancy; 21 patients had at most one criterion, and none had malignancy. CONCLUSION Clinical and radiologic criteria of patients with pleural effusion permit different risk levels for pleural malignancy to be distinguished. Consequently, application of the four proposed criteria permits better indication of thoracoscopy in patients with undiagnosed pleural effusion.