Systematic Compared With Targeted Staging With Endobronchial Ultrasound in Patients With Lung Cancer.

Abstract

To evaluate the accuracy of systematic mediastinal staging by endobronchial ultrasound transbronchial needle aspiration (EBUS-TBNA) (sampling of all visible nodes measuring ≥5 mm from stations N3 to N1 regardless of their positron emission tomography/computed tomography [PET/CT] features) and compare this staging approach with targeted EBUS-TBNA staging (sampling only 18F-fluorodeoxyglucose [FDG]-avid nodes) in patients with N2 non-small cell lung cancer on PET/CT. Retrospective study of 107 patients who underwent systematic EBUS-TBNA mediastinal staging. The results were compared with those of a hypothetical scenario where only FDG-avid nodes on PET/CT would be sampled. Systematic EBUS-TBNA sampling demonstrated N3 disease in 3 patients, N2 disease in 60 (42 single-station or N2a, 18 multiple-station or N2b) and N0/N1 disease in 44. Of these 44, 7 underwent mediastinoscopy, which did not show mediastinal disease; 6 of the 7 proceeded to lung resection, which also showed no mediastinal disease. Thirty-four N0/N1 patients after EBUS-TBNA underwent lung resection directly: N0/N1 was found in 30 and N2 in 4 (1 N2b with a PET/CT showing N2a disease, 3 N2a). Sensitivity, specificity, negative predictive value, positive predictive value, and overall accuracy of systematic EBUS-TBNA were 94%, 100%, 90%, 100% and 96%, respectively. Compared with targeted EBUS-TBNA, systematic EBUS-TBNA sampling provided additional important clinical information in 14 cases (13%): 3 N3 cases would have passed unnoticed, and 11 N2b cases would have been staged as N2a. In clinical practice, systematic sampling of the mediastinum by EBUS-TBNA, regardless of PET/CT features, is to be recommended over targeted sampling.