Background: Performance of linear probe endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) for staging non-small-cell lung cancer has been extensively studied. Alternate indications for its use are less well characterised, and performance in other clinical settings may differ. Methods: We examined a prospectively collected cohort comprising the first 215 patients undergoing EBUS-TBNA at our institution. Patients were analysed according to the clinical and radiological indication for referral. We also examined the effect of the procedural learning curve on diagnostic sensitivity. Results: A total of 215 patients underwent 216 EBUS-TBNA procedures. EBUS-TBNA returned adequate tissue for cytopathological analysis in 202 of 216 procedures (94%). Overall sensitivity for detection of malignancy was 0.92 (95% confidence interval 0.86-0.96); however, this varied according to the primary indication for EBUS-TBNA. Diagnostic sensitivity was high among all sub-groups, but the negative predictive value varied depending on the clinical indication for the procedure. We estimate 104 invasive surgical procedures and 32 inpatient admissions were avoided by use of EBUS-TBNA. Significant improvement in diagnostic performance was seen after 20 procedures were completed, and diagnostic accuracy did not peak until after 50 procedures. Conclusions: EBUS-TBNA is able to confirm accurately histologically a large number of disease processes, both malignant and benign, in all clinical indications studied. The procedure is safe even when carried out by proceduralists with minimal prior experience. Diagnostic performance continues to improve beyond 50 cases carried out.
- Endobronchial ultrasound
- Learning curve
- Mediastinal lymphadenopathy
- Transbronchial needle aspiration