Excellence in non-small cell lung cancer staging by endobronchial-TBNA: Comparison with PET-CT and surgery.

OBJECTIVE: To determine the correlation and/or discrepancies between positron emission tomography (PET-CT) findings, endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) and surgery in the staging of non-small cell lung carcinoma.

MATERIAL AND METHODS: Data were evaluated retrospectively from a prospective interventional endoscopy database. Positive results with EBUS-TBNA was the first end point and all cytology negatives were confirmed with mediastinoscopy/surgery.

RESULTS: Four hundred and eighty three patients were included and 1017 lymph nodes (LNs) were sampled in the study. One hundred and twenty eight LNs were excluded (positive with EBUS-TBNA). Four hundred and sixty five LN (52.3%) were found benign with EBUS-TBNA; however, only 15 of these were confirmed to be malignant by surgery (1.7%). The sensitivity, specificity, PPV, NPV and diagnostic accuracy of EBUS-TBNA were 96.5, 100, 100, 96.7 and 98.3%, respectively. The sensitivity, specificity, PPV, NPV and diagnostic accuracy of PET-CT for maximum standardized uptake value (SUVmax) 2.5 were 90.1, 29.2, 55.3, 75.4, 59.2%, respectively. A cut-off SUVmax of 5.2 was detected with 74.8% sensitivity, 84% specificity, 82.0% PPV, 77.5% NPV and 79.5% accuracy (area under the curve (AUC) of 0.818, 95% CI 0.791-0.843, p<.001).

CONCLUSION: EBUS is a reliable, repeatable and safe technique with a high diagnostic accuracy and should be performed quickly to avoid superfluous time loss in the staging of lung cancer. Abbreviations PET-CT F-18 fluorodeoxyglucose positron emission computed tomography NSCLC Non-small cell lung cancer EBUS-TBNA Endobronchial ultrasound-guided transbronchial needle aspiration SUVmax Maximum standardized uptake value LNs Lymph nodes TTF-1 Thyroid transcription factor-1 H&E Hematoxylin and eosin; Med: Mediastinoscopy VATS Video associated thoracic surgery AUC Area under curve OR Odds ratio CI Confidence intervals.