Abstract
BACKGROUND AND OBJECTIVES: Since the introduction of endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA), most pulmonary centers use this technique exclusively for mediastinal lymph node (LN) sampling. Conventional "blind" TBNA (cTBNA), however, is cheaper, more accessible, provides more tissue, and requires less training. We evaluated whether sampling of mediastinal LN using EBUS-TBNA or cTBNA according to a predefined set of criteria provides acceptable diagnostic yield. MATERIALS AND METHODS: Sampling method was determined prospectively according to a predefined set of criteria based on LN station, LN size, and presumed diagnosis. Sensitivity, specificity, positive, and negative predictive value were evaluated for each modality. RESULTS: One hundred and eighty-six biopsies were carried out over a 3-year period (86 cTBNA, 100 EBUS-TBNA). Seventy-seven percent of LN biopsied by EBUS-TBNA were <20 mm, while 83% of cTBNA biopsies were ≥20 mm. Most common sites of cTBNA sampling were station 7, 4R, and 11R as opposed to 7, 11R, 4R, and 4 L in the case of EBUS-TBNA. Most common EBUS-TBNA diagnosis was malignancy versus sarcoidosis in cTBNA. EBUS-TBNA and cTBNA both had a true positive yield of 65%, but EBUS-TBNA had a higher true negative rate (21% vs. 2% for cTBNA) and a lower false negative rate (7% vs. 28%). Sensitivity, specificity, positive predictive value, and negative predictive value for EBUS-TBNA were 90%, 100%, 100%, and 75%, respectively, and for cTBNA were 68%, 100%, 100%, and 7%, respectively. CONCLUSION: A. stepwise approach based on LN size, station, and presumed diagnosis may be a reasonable, cost-effective approach in choosing between cTBNA and EBUS-TBNA.