Abstract 1920: Novel molecular targets for chemoprevention of squamous cell carcinoma

INTRODUCTION: It is estimated that cancer prevention efforts can reduce cancer incidence by over 50%. Unlike for advanced disease, effective molecularly-driven interventions and risk assessment are not available for clinically normal tissues that may be have been exposed to carcinogens or for precancerous lesions. Cutaneous squamous cell carcinoma (cuSCC) comprises 15-20% of all skin cancers and has the most accessible and clinically well-characterized progression sequence of any human cancer, from a distinct precancerous lesion, the actinic keratosis (AK), to invasive carcinoma. Thus, it is an ideal model for establishing a paradigm of molecularly targeted cancer chemoprevention. METHODS: Here, we performed next-generation sequencing of total RNA and miRNA (Illumina HiSeq) on matched isogenic samples of human cuSCC, surrounding normal (chronically irradiated) skin (NS), and AK. In parallel, we profiled matched samples from a UV-driven Hairless mouse model of cuSCC for cross-species analysis, to identify the most important drivers of progression from NS to AK to cuSCC. RESULTS: Unsupervised clustering of both mRNA and miRNA expression changes showed that preneoplastic AKs span a continuum indistinguishable from cuSCC or surrounding NS, whereas cuSCC and NS were easily distinguished. Through cross-species computational analysis of mRNA-miRNA functional pairs, we identified miR-21, miR-205, miR-31, let-7B, and miR-497 and their mRNA targets as core drivers of cuSCC development through the AK intermediate. We show that several of these miRNAs are modulated by UV exposure, regulate susceptibility to apoptosis, and regulate cell motility. Several miRNAs are being inhibited in our mouse model to validate them as chemoprevention targets. TRANSFAC analysis identified E2F, SP1, AP1, and TCF3 as key transcriptional regulators of SCC development. We tested the global mRNA and miRNA expression similarities to other tumor types profiled by the NIH TCGA effort, showing that cuSCC is most closely related to head & neck SCC, lung SCC, and basal subtype of breast cancer. Exome analysis confirmed a high frequency of TP53, NOTCH1/2, and CDKN2A mutations in cuSCC, with mutational loads of >30/Mb that are strongly dominated by UVB signature lesions. Surprisingly, NS contains evidence of numerous mutations reflecting UV-induced somatic mosaicism. CONCLUSIONS: We report the first integrated transcriptomic characterization of the development of cuSCC through the preneoplastic AK.