ABSTRACT
INTRODUCTION: We review endobronchial ultrasound-guided fine-needle aspirate samples and investigate cases with discrepancies between rapid on-site evaluation (ROSE) and final diagnosis in patients with bronchogenic carcinoma. MATERIALS AND METHODS: Endobronchial ultrasound-guided fine-needle aspirates from 2009 to 2010 were studied. On-site adequacy assessments were compared with final diagnoses. Concordant diagnoses showed agreement between ROSE interpretation and final diagnosis. If the initial interpretation differed from the final diagnosis, the case was discordant. Slides from discordant aspirates were reviewed. Discordant results were categorized as sampling error or interpretive/screening error at ROSE. RESULTS: A total of 340 endobronchial ultrasound-guided procedures were performed in 335 patients (168 men, 167 women, median age 65 years). Diagnostic discrepancies between ROSE and final diagnoses occurred in 65 aspirates (11%) from 51 patients with carcinoma. Of the 65 discrepant cases, 52 (83%) were subsequently called positive for carcinoma. Rescreening of slides in 47 available cases with a final positive diagnosis showed insufficient tumor for diagnosis in 28 of 47 cases (60%). The remaining 19 of 47 cases (40%) were classified as interpretive/screening errors at ROSE. Most errors occurred in aspirates called atypical or atypical suspicious, which upon rescreening were considered diagnostic (16 aspirates, 84%). CONCLUSIONS: Initial and final diagnoses were concordant in 89% of aspirates from patients with carcinoma. All aspirates that were positive at ROSE were concordant. In discordant cases, all aspirates deemed "atypical suspicious for malignancy" and 86% of aspirates deemed "atypical cells" on ROSE had a final diagnosis of carcinoma. The majority of discordant cases with a positive final diagnosis were due to sampling (60%).
