RESUMO
BACKGROUND AND OBJECTIVE: Medical thoracoscopy (MT) is useful for the management of pleural disease. Rapid on-site evaluation (ROSE) of transbronchial needle aspirates proved to be useful during bronchoscopy. We aimed to evaluate the diagnostic performance of ROSE of MT biopsy specimens and thoracoscopists' impression of the macroscopic appearance and assess the intermodality agreement between ROSE and final histopathologic diagnosis. METHODS: Sixty two patients with exudative pleural effusions further investigated with MT were enrolled. MT was performed under local anaesthesia and conscious sedation, using the rigid pleuroscope. ROSE with the Hemacolor rapid staining method of the biopsy specimens was performed. Thoracoscopists' impression of the macroscopic appearance was recorded. The final diagnosis was established following histopathological examination. RESULTS: Thoracoscopic pleural biopsies were diagnosed in 61 patients (98.4%). Group A (n = 25) consisted of patients with malignancy and group B (n = 37) with benign disorders. Area under the curve of ROSE for the diagnosis of malignancy was 0.86 (95% CI: 0.76-0.96, P < 0.001), with a sensitivity of 79.17%, specificity of 94.59%, diagnostic accuracy of 88.5%, positive predictive value of 90.5% and negative predictive value of 87.5%. Intermodality agreement between ROSE and histopathology was good (κ ± SE = 0.615 ± 0.084, P < 0.001). Area under the curve of the thoracoscopists' impression of macroscopic appearance was 0.72 (95% CI: 0.58-0.85, P = 0.001), with a sensitivity of 100%, specificity of 44.7%, positive predictive value of 53.33% and negative predictive value of 100%. CONCLUSION: Rapid on-site evaluation during MT was found to have high accuracy for predicting malignancy. ROSE can provide the thoracoscopist with an on-site preliminary diagnosis, especially in cases with inconclusive macroscopic appearance.