A biomarker-based approach for the determination of sample adequacy during endobronchial ultrasound-guided transbronchial needle aspiration

T. Ishiwata^a (Dr), A. Sage^b (Dr), N. Mohammed^b (Ms), T. Inage^a (Dr), N. Bernards^a (Dr), K. Yasufuku*^a (Prof)

^a Division of Thoracic Surgery, Toronto General Hospital, University Health Network, Toronto, CANADA ; ^b Latner Thoracic Research Laboratories, University Health Network, Toronto, CANADA

* Kazuhiro.Yasufuku@uhn.ca

Background: Rapid on-site cytologic evaluation (ROSE) is beneficial in providing an immediate evaluation for sample adequacy on endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA). However, the availability of ROSE is not widespread. We hypothesized that a biomarker-based sample adequacy assessment could provide a simple alternative to ROSE. Methods: To seek biomarkers in lymph nodes, ten potential biomarkers likely to have significantly higher expression levels in lymph node tissue versus whole blood were selected in silico. mRNA expression levels of candidate markers were compared between whole lymph nodes and blood in the pulmonary artery taken from resected lungs. Two biomarkers that had the highest ratio in lymph nodes to blood were selected, and expression levels of the two protein markers in needle rinse solution were measured using EBUS-TBNA samples from lymph nodes. The predicting performance of two biomarkers for the adequate sampling was evaluated compared to ROSE. Results: Quantitative polymerase chain reaction showed that C-X-C motif chemokine ligand 13 (CXCL13) and C-C motif chemokine ligand 21 (CCL21) showed the highest fold increase at 19- and 4- fold increased expression in lymph node tissue (n=10) versus blood samples (n=13). In EBUS-TBNA samples (n=49), both CXCL13 and CCL21 protein levels were significantly elevated in adequate samples compared to in inadequate samples including specimens with low cellularity in the final pathological assessment (CXCL13, 43.7±64.4 vs 1.1±2.0 pg/mL, p=0.0004; CCL21, 513.3±245.4 vs 214.3±101.3 pg/mL, p<0.0001). As univariate predictors of adequate samples, CXCL13 and CCL21 biomarkers performed well, with an area under the receiver operating characteristic curve of 86% and 88%, respectively. ROSE and the biomarker-based approach showed equivalent sensitivity 96% for predicting samples with adequate diagnostic material in the final pathological assessment. Conclusion: Measuring CXCL13 and CCL21 protein levels in EBUS-TBNA samples performs equally well as ROSE in the determination of adequacy yield for diagnostic purposes.

Disclosure of funding source(s):

Funding was provided through the William Coco Chair in Surgical Innovation for Lung Cancer.