Endobronchial Ultrasound-Guided Transbronchial Needle Aspiration (EBUS-TBNA) Cytology: Review of High-Volume Institutional Experience with a Focus on Diagnostic Accuracy and EGFR/KRAS Molecular Testing
Leung Chu Tong, Maria A Friedlander, Atasha Asmat, David J Finley, Andre L Moreira, Natasha Rekhtman. Memorial Sloan-Kettering Cancer Center, New York, NY
Background: Only few large-scale studies have reviewed the performance of EBUS-TBNA cytology in the diagnosis and molecular analysis of lung carcinomas. We retrospectively reviewed our institutional experience with EBUS-TBNA samples, with a focus on cytologic/histologic correlation and sufficiency for EGFR/KRAS molecular testing.
Design: Cytologic/histologic correlation was performed for all EBUS-TBNAs obtained during a 1-year period (Jan – Dec 2009). In addition, results of routine testing for EGFR and KRAS mutations (by Sequenom and fragment analysis) on EBUS-TBNA samples, performed by request of clinicians, during 1.5-year period (Jan 2010 - June 2011) were reviewed, and 70 consecutive samples diagnosed as adenocarcinoma were scored for tumor cellularity to determine potential sufficiency for molecular testing in unselected samples. Based on the requirements in our molecular laboratory, sufficiency was defined as >10% tumor cell proportion and absolute cellularity of >200 cells.
Results: A total of 562 EBUS-TBNAs from 300 patients were performed in 2009. Cytology samples were classified as malignant or suspicious (n=231, 41%), atypical (n=1; 0.2%), negative (n=224; 40%) or non-diagnostic (n=106; 19%). Of 231 malignant diagnoses, a specific tumor type was determined in 191 (83%) cases. 123 of 562 (22%) EBUS-TBNAs had same-site histologic follow-up, revealing sensitivity and specificity for malignancy of 97% and 98%, respectively, and tumor typing accuracy of 100%. A total of 57 samples underwent routine molecular analysis, revealing 10 (18%) EGFR and 19 (33%) KRAS mutations. For 70 consecutive samples, 60 (86%) had sufficient material for molecular testing in the cell block (n=38) or smears (n=22) if cell block was unavailable or insufficient. The mean proportion of tumor cells (relative to benign cells) was 63% for cell blocks and 62% for smears.
Conclusions: Based on the review of our high-volume institutional experience with EBUS-TBNAs, we confirm a high sensitivity and specificity of this modality for the detection of malignancy. Furthermore, we find that EBUS-TBNA samples provide high accuracy of tumor type determination and yield sufficient material for molecular testing in the vast majority of cases.
Monday, March 4, 2013 1:00 PM
Poster Session II # 64, Monday Afternoon