RESUMO
OBJECTIVE: To analyze which factors predict mediastinal N2/N3 lymph node staging and diagnostic accuracy of PET and CT to determine it. PATIENTS AND METHODS: We analyzed data collected prospectively in a database that included patients with non-small cell lung cancer (NSCLC) who underwent EBUS-TBNA. Prior to EBUS-TBNA, CT and PET were used to define the radiographic N stage and lymph nodes with short axis ≥ 1 cm by CT or with ratio between maximum standardized uptake value (maxSUV), by PET, of lymph node and primary tumor greater than 0.56, were considered pathological. Definitive lymph node staging was established through EBUS-TBNA, mediastinoscopy or surgical lymph node dissection. RESULTS: One hundred and thirty four patients were included, in 88 of whom (65.6%), definitive lymph node staging was N2 or N3. Primary tumor of central location, lymph node size, maxSUV of lymph node and radiographic N stage by CT or PET were associated with N2/N3 in univariate analysis, but in logistic regression model it was only independently related with N stage by CT or PET. Negative predictive value and positive predictive value of CT were 0.81 and 0.74, respectively, and for PET 0.78 and 0.68. CONCLUSION: In NSCLC, in locoregional disease radiographic staging by CT or PET predict the existence of N2/N3 mediastinal disease, but negative and positive predictive values of both imaging techniques are not adequate, so EBUS-TBNA samples should be taken in all lymph nodes with a diameter greater than 5 mm, regardless of PET findings.