The optimal method for specimen preparation of endobronchial ultrasound-transbronchial needle aspiration (EBUS-TBNA) is still controversial. This study aims to compare several techniques available for EBUS-TBNA specimen acquisition and processing, in order to identify the best performing technique.
We retrospectively reviewed the data of 199 consecutive patients [male, 73%; median age, 64 years (IQR: 52-74 years)] undergoing EBUS-TBNA at our institution from 2012 through 2014 for diagnosis of hilar-mediastinal lymph node enlargement suspect of neoplastic (n=139) or granulomatous (n=60) disease. All procedures were performed by two experienced bronchoscopists, under conscious sedation and local anaesthesia, using 21/22-Gauge (G) needle, without rapid on-site evaluation (ROSE). Five specimen-processing techniques were used: cytology slides in 42 cases (21%); cell-block in 25 (13%); core-tissue in 60 (30%); combination of cytology slides and core-tissue in 51 (26%); combination of cytology slides and cell-block in 21 (10%). To assess the diagnostic accuracy of each tissue-processing technique we compared the EBUS-TBNA results to those obtained with surgical lymphadenectomy, or 1-year follow-up in non-operated patients.
Diagnostic yield, accuracy and area under the curve (AUC) were as follows. Cytology slides: 81%, 80%, 0.90; cell-block: 48%, 33%, 0.67; core-tissue: 87%, 99%, 0.96; cytology slides + core-tissue: 80%, 100%, 1.00; cytology slides + cell-block: 86%, 100%, 1.00. Cytology slides and core-tissue method showed non-significantly different diagnostic yield (P=0.435) and AUC (P=0.152).
In our single-institution experience, cytology slides and core-tissue preparations demonstrated high and similar diagnostic performance. Cytology slides combination with core-tissue or cell-block showed the highest performance, however these combination methods were more resource-consuming.