We propose a novel computational method to identify predictive genetic biomarkers to guide the use of targeted therapies using high-throughput pan-cancer data from primary tumors. The method will be used to identify predictive biomarkers in lung cancer for Sudemycin-D6 (SD6), a novel splicing modulator with potent anti-tumor activity. Currently, new therapies are designed to target genes specifically mutated in cancer, and the corresponding oncogenic mutations are used as biomarkers for the drugs. The use of targeted drugs is limited because the sensitizing oncogenic driver events are absent in many cancers, albeit some cancers have alternative mechanisms of sensitivity. This is particularly true for SD6, which was found to be effective in lung cancer cell lines that do not harbor any mutations in spliceosome genes. Our approach is to use synthetic lethality to identify biomarkers that predict drug sensitivity in cancers that are not mutated for the drug target. In synthetic lethal (SL) interactions, a genetic alteration in one gene leads to dependency on a second gene. Neither alteration by itself is essential for survival, but together they lead to cancer cell death. Our hypothesis is that genetic alterations (mutations/copy number alterations) that are SL partners of drug targets can be used as predictive biomarkers for the therapeutic efficacy of the drug in specific cancer types. For SD6, we will identify genetic alterations that are SL partners of genes constituting the spliceosome.
Two specific aims are proposed:
In Aim 1, we will develop a computational method that can be applied to large-scale primary tumor genomic and transcriptomic datasets to identify candidate predictive biomarkers using SL interactions mined from primary tumor data. This method will be applied to genomic and transcriptomic datasets from The Cancer Genome Atlas (TCGA), European Genome-phenome Archive (EGA), and gene expression data from Veterans Administration (VA) patients to identify candidate predictive biomarkers for SD6 in lung cancer.
In Aim 2, we will experimentally validate the predictive biomarkers identified in Aim 1 in two steps. First, in Aim 2a, we will validate the top 10 candidate biomarkers using genetic knockdown of the biomarker with shRNA and pharmacologic knockdown with SD6 in isogenic lung cancer cell lines in vitro. We will confirm that the mechanism of SD6 sensitivity is via synthetic lethality between the biomarker and splicing factors. Next, in Aim 2b, we will validate the top 3 candidate biomarkers in vivo using pharmacologic knockdown with SD6. We expect the proposed study will find predictive biomarkers for SD6 efficacy and accelerate its clinical development in lung cancer. The long-term objective is to develop a tool for predictive biomarker identification that can be used by the wider research community (physicians, biologists, and medicinal chemists) and have applications in both the clinical development of new targeted agents and in the identification of patients who are likely to respond to them. This tool has the potential to unlock the clinical benefit of the available drug arsenal and greatly improve patient outcomes.
The efficacy of highly selective anti-tumor agents can be greatly enhanced through the use of predictive biomarkers as companion diagnostics to identify responders to a targeted therapy. The goal of this research is to develop a novel computational method to identify predictive biomarkers of targeted drugs using synthetic lethal interactions mined from large-scale primary tumor datasets. In this project, we will focus on splicing modulators, which are novel therapeutic agents with potent antitumor activity in animal models and provide preliminary validation for the predictive biomarkers in lung cancer, an aggressive malignancy for which new therapeutic options are badly needed.
