S90 A novel organotypic model of bronchial dysplasia for preclinical screening of potential therapeutic agents for early squamous lung cancer (SQC)

Background Squamous lung cancer has 5-year survival rates of less than 15%. Strategies to reduce lung cancer prevalence include smoking cessation, early detection and chemoprevention. There are currently no licensed strategies for the chemoprevention of lung cancer, so this represents a significant unmet need. Bronchial dysplasia precedes the development of invasive SQC. We have developed a novel organotypic (OTC) model of bronchial dysplasia that recapitulates key genetic events in clinical specimens of bronchial dysplasia. Aim Our aim is to screen potential chemoprevention compounds for efficacy in a model of human bronchial dysplasia. Methods We use a novel OTC model incorporating immortalised human bronchial epithelial cells (HBECs) genetically manipulated to reproduce the key genetic lesion reported in the human disease (TP53, and CDKN2A disruption, deregulated SOX2). Importantly the epithelial layer, once confluent, is maintained at the air-liquid interface (ALI) mimicking the physical microenvironment in vivo. Dysplastic lesions develop 3–5 days after inducible deregulation of SOX2. Potential chemoprevention agents are added simultaneously to SOX2 induction to mimic primary chemoprevention, or after the dysplastic phenotype forms in response to SOX2 induction (secondary chemoprevention). We measure phenotypic response using phase-contrast microscopy, histology, immunohistochemistry and western blotting. Results are then corroborated by genetic targeting (shRNA/CRISPR) of drug targets in the OTC and in unrelated squamous carcinoma cell lines and in xenograft models. Results We screened multiple tool compounds or compounds already in late phase clinical trials. All demonstrate therapeutic target engagement in the epithelial layer confirmed by western blotting despite that layer being maintained at ALI. Most screened compounds have negligible impact over a range of doses or lead to generalised toxicity at higher doses. Remarkably, multiple compounds from two therapeutic classes can a) prevent of the emergence of dysplastic lesions thus mimicking ‘primary’ chemoprevention; and b) induce complete resolution of the established dysplastic phenotype with no measurable impact on the intact ‘normal’ epithelial monolayer. CRISPR and shRNA targeting corroborated the therapeutic screening results. Conclusions A rational organotypic model of human bronchial dysplasia can be used to perform preclinical screens for potential efficacy in the chemoprevention/treatment of squamous lung cancer. We are now developing a clinical trial in patients with early stage lung cancer to translate this work.