Abstract LB169: Machine learning-based approach for glioblastoma drug repurposing on real-world patient data
Ko‐Hong Lin, Yejin Kim, Dung‐Fang Lee, Xiaoqian Jiang · Cancer Research · 2023
Abstract Glioblastoma (GBM) is the most aggressive and deadly type of brain cancer. The poor survival rate of GBM is often attributed to the low therapeutic efficiency of current front-line therapy, treatment resistance, and high recurrence rate. Sizeable clinical data of cancer patients preserve a great potential for drug discovery and drug repurposing research, enabling comprehensive screening of treatment effects among frequently-prescribed medications. To efficiently characterize potent drug candidates and develop novel treatment strategies for GBM, we designed a machine-learning-based pipeline that predicts repurposable drug combinations from large-scale patient claims data. Our pipeline first identified pseudo-randomized drug-user and non-user cohorts from heterogeneous clinical records and balanced the confounding effects using the Inverse Probability of Treatment Weighting (IPTW) method. Next, our model computed the averaged treatment effect on the treated (ATT) values to determine the treatment effects of our candidates. Our methodology identified several promising drug combinations that have been found to be effective in inhibiting GBM in cell line experiments. This research was conducted by integrating computational and experimental approaches to identify drug combinations, providing a promising strategy for the development of new treatments for GBM. Citation Format: Ko-Hong Lin, Yejin Kim, Dung-Fang Lee, Xiaoqian Jiang. Machine learning-based approach for glioblastoma drug repurposing on real-world patient data [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 2 (Clinical Trials and Late-Breaking Research); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(8_Suppl):Abstract nr LB169.