RESUMEN
INTRODUCTION: Endobronchial ultrasound transbronchial needle aspiration (EBUS-TBNA) is an emerging minimally invasive option for pathologic examination of intrathoracic lymphadenopathy as well as for staging lung cancer. OBJECTIVES: To evaluate the diagnostic yield and possible learning curve effects on diagnostic performance using EBUS-TBNA in mediastinal lymphadenopathy. METHODS: A retrospective analysis was performed on 243 consecutive patients who underwent EBUS-TBNA over a 4-year period. Demographic and clinical data and pathology results were analysed for different time frames in order to evaluate potential learning curve effects. The procedures were performed by two experienced bronchoscopists at a single university medical centre. RESULTS: Samples were representative in 83% (200/243) of the patients. The overall diagnostic yield was 66% (n = 161/243). The diagnostic accuracy of EBUS-TBNA for detecting malignancy was 98.0% and for lung cancer 98.5%. The sensitivity, specificity, positive and negative predictive values for lung cancer stage ≥N1 and malignant disease were 100% for the first three studied periods and slightly less favourable in the most recent study period. Representative samples were obtained more frequently in the latter study periods (P < 0.001). CONCLUSION: EBUS-TBNA is a safe method with a learning curve that is easily overcome, although previous experience with ultrasound may be necessary. The diagnostic yield of EBUS-TBNA is in accordance with previously reported yield of standard cervical mediastinoscopy. At present, however, the relationship between EBUS-TBNA and mediastinoscopy appears to be complementary rather than substitutive.