Abstract:
Along with the remarkable success stories in the evolutionary journey of targeted cancer treatment, there are examples where the targeted therapies failed to match the expectations. There have been multiple reports of resistance against such targeted therapies. Thus, it becomes important to understand the underlying evolution that contributes to the developed resistance in the tumor cells. In this study, a comprehensive, high-throughput assessment was performed wherein machine-learning models have been implemented on the cellular expression omics (transcriptomics and proteomics) data of cell lines that are resistant and sensitive to FDA approved targeted drugs. Based on the predictive ability of the models, Axitinib was found to be a drug that showed high predictive accuracy. Thus, various algorithms rooted in explainable AI were implemented to find out the key molecular players that are responsible for mediating resistance. Further, the biological pathways wherein the derived resistance mediators are over-represented were found out and the cross-talk between them were analyzed to understand the affected/perturbed biological processes. Further, it was found out that there isn’t one unique way to deploy resistance but there are multiple mechanisms via which resistance is enabled in the cells. These multiple forms of resistance and their mechanisms were further elucidated. The findings thus gained from the work help us understand the limitations of using Axitinib for cancer treatment and the vulnerabilities present in the cells due to tumor evolution. This study can take us a step further in the domain of precision medicine enabling us to map those vulnerabilities/susceptibilities in the patient genome so as to pre-empt whether a patient will be responsive to Axitinib or not.
