View Abstract

A4686 - Genomic Profiling of Bronchoalveolar Lavage Fluid in Patients with Non-Small Cell Lung Cancer
Author Block: V. S. Nair1, A. Bik-Hu Li2, H. Stehr2, J. Chabon2, A. Chaudhuri3, L. Zhou4, H. Naemi5, K. Ayers6, M. Ramsey7, H. S. Bedi7, R. Van Wert7, A. W. Sung7, N. Lui6, L. Backhus6, M. Berry6, J. B. Shrager6, A. Alizadeh8, M. Diehn9; 1Medicine/Radiology/Canary Center for Cancer Early Detection, Stanford University, Stanford, CA, United States, 2Stanford Cancer Institute, Stanford University, Stanford, CA, United States, 3Radiation Oncology, Stanford University, Stanford, CA, United States, 4Grail Corporation, Menlo Park, CA, United States, 5Radiology, Stanford University, Stanford, CA, United States, 6Cardiothoracic Surgery, Stanford University, Stanford, CA, United States, 7Medicine, Stanford University, Stanford, CA, United States, 8Institute for Stem Cell Biology and Regenerative Medicine/Medical Oncology, Stanford University, Stanford, CA, United States, 9Institute for Stem Cell Biology and Regenerative Medicine/Radiation Oncology, Stanford University, Stanford, CA, United States.
Introduction. Genomic profiling of lung cancer is clinically utilized for treatment decisions but requires tissue obtained through needle or surgical biopsy. Circulating tumor DNA (ctDNA) is a useful method to identify tumor mutations without a biopsy, but not all patients with cancer have detectable levels in blood. We therefore aimed to identify tumor mutations in lung cancer patients by comprehensively profiling cell free DNA isolated from the regional tumor environment using bronchoalveolar lavage (BAL) fluid. Methods. We applied Cancer Personalized Profiling by deep Sequencing (CAPP-Seq) to identify somatic alterations in BAL, blood plasma and primary tumors of patients undergoing bronchoscopy for lung cancer diagnosis, staging, or surveillance. We employed a lung cancer-focused CAPP-Seq panel that covers 252 genes and 302 kilobases. We performed BAL in the segment where the tumor was located if possible, and blood was drawn prior to the procedure. The Stanford pathology lab performed tumor profiling with a CLIA-based, next-generation sequencing, laboratory developed test. Mutations in BAL and plasma were called using the CAPP-Seq bioinformatics pipeline. Calls were then compared by McNemar’s test using tumor mutations as the gold standard. Results. We analyzed 21 lung cancer patients’ BAL, plasma, and tumor. 4.3±0.8 ml of BAL and 5.0±2.3 ml of blood plasma was sequenced with a mean unique read depth for BAL and plasma of 1,761 and 3,450 respectively. Median allele frequency was 0.24% (range 0.05%-6.1%) and 2.64% (range 0.05%-75%) in plasma and BAL respectively. Eleven (52%) plasma and 13 (62%) BAL samples contained matching tumor variant calls (p=0.62). Tumor-derived mutations were found exclusively in plasma and BAL for 1 and 3 patients respectively, and in both sample types for 10 patients. For stage I patients (n=8), tumor mutations were observed in 4 BAL (50%) and 3 (38%) plasma samples. BAL cytology was performed clinically on 12 cancer patients and tumor cells were definitively detected in one (8%). By comparison, tumor-derived mutations were detected in 8 of these 12 BAL samples (67%, p=0.02). Conclusions. In this pilot study, tumor-derived mutations were found at higher concentrations in BAL compared to plasma but detected in a similar fraction of samples. CAPP-Seq of BAL fluid outperformed traditional cytology for identifying lung cancer. These findings suggest that BAL genomic profiling may have clinical utility in patients undergoing bronchoscopy for the diagnosis, staging, or surveillance of lung cancer.