Pre-clinical applications of endobronchial ultrasound in lung cancer

© 2018 Dr. Tracy Li-Tsein LeongThe last decade has seen a paradigm shift in the clinical approach to lung cancer. Treatment has evolved from empirical to rational, based on molecular profiling of tissue obtained by minimally invasive techniques such as endobronchial ultrasound (EBUS). However, the stagnantly poor prognosis of lung cancer highlights the ongoing need to improve understanding of the genomic landscape in order to identify novel biomarkers and effective therapeutic strategies. In a series of discrete but inter-related studies, this body of work explored the role that EBUS may play in furthering knowledge in pre-clinical lung cancer research. Parameters of enquiry were established by a systematic review and meta-analysis investigating the ability of systematically performed EBUS to detect occult nodal metastases in patients with a radiologically normal mediastinum. The sensitivity of EBUS for detection of unsuspected nodal disease was 49% (95%CI 41-57%), which was lower than observed for patients with clinically enlarged nodes. Subsequently, work in this thesis focused on lung cancer cases with mediastinal spread where EBUS has established utility. A cohort study addressing the molecular yield of EBUS, demonstrated that a single aspirate returned DNA of sufficient quantity and quality to perform targeted genotyping of lung cancers, including samples with limited tumour cell content. These findings led to a multi-institutional cohort study examining the genomic heterogeneity and evolution of metastatic lung cancer. EBUS was used to sample multiple regions in patients with advanced lung cancer, and whole genome sequencing was performed to generate evolutionary models. Heterogeneity was observed in driver gene mutations between primary and metastatic tumours at single nucleotide variant, insertion/deletion, copy number variant, and structural variant levels. Mutations private to metastatic tumours were associated with a highly diverse somatic mutation signature, most commonly APOBEC. A smoking mutation signature was found highly prevalent in smokers with germline variants in DNA repair genes. Evolutionary modelling suggested that lung cancer precursors accumulated large numbers of mutations before frank malignancy developed. Once established, primary tumours spread to regional lymph nodes rapidly and early in the life of the tumour. While genomic mapping is the first step in development of targeted therapeutics, accurate pre-clinical models are needed to test clinical efficacy. To address deficiencies in existing models of lung cancer, a cohort study was performed to generate patient-derived xenografts of inoperable small cell lung cancer (SCLC) using specimens obtained by EBUS. Viable tumour cells were engrafted into immunodeficient mice with high efficiency, and retained important features of the primary tumour, including mutations characteristic of SCLC. Lastly, ultrasound at low intensities has demonstrated utility as an anti-cancer therapy in a range of cancer cell types cultured in vitro. In a study to explore the potential in lung cancer, both in vitro and in vivo models, including EBUS-generated PDX were subjected to therapeutic ultrasound with cisplatin chemotherapy. In vitro experiments demonstrated increased apoptosis potentiated by chemotherapy combination treatment. However, such findings did not translate to in vivo testing. In summary, this thesis demonstrated that EBUS, a historically diagnostic tool, has value in pre-clinical applications in lung cancer