Cerebrospinal fluid as a liquid biopsy for molecular characterization of brain metastasis in patients with non-small cell lung cancer.

OBJECTIVES:Non-small cell lung cancer (NSCLC) with brain metastases (BM) is a challenging clinical issue with poor prognosis. No data exist regarding extensive genetic analysis of cerebrospinal fluid (CSF) and its correlation to associated tumor compartments. MATERIALS AND METHODS:We designed a study across multiple NSCLC patients with matched material from four compartments; primary tumor, BM, plasma and CSF. We performed enrichment-based targeted next-generation sequencing analysis of ctDNA and exosomal RNA in CSF and plasma and compared the outcome with the solid tumor compartments. RESULTS:An average of 105 million reads per sample was generated with fractions of mapped reads exceeding 99% in all samples and with a mean coverage above 10,000x. We observed a high degree of overlap in variants between primary lung tumor and BM. Variants specific for the BM/CSF compartment included in-frame deletions in AR, FGF10 and TSC1 and missense mutations in HNF1a, CD79B, BCL2, MYC, TSC2, TET2, NRG1, MSH3, NOTCH3, VHL and EGFR. CONCLUSION:Our approach of combining ctDNA and exosomal RNA analyses in CSF presents a potential surrogate for BM biopsy. The specific variants that were only observed in the CNS compartments could serve as targets for individually tailored therapies in NSCLC patients with BM.