Precision oncology relies on the hypothesis that further characterizing a patient's tumor will lead to better predictions of treatment response. While this approach effectively breaks diagnoses into smaller and smaller subtypes likely to respond to a given targeted inhibitor, the downside is it results in highly fragmented clinical data spread across multiple treatment arms. The more progress the field makes in assigning new therapies, the harder it will be to accrue adequate sample size to test another therapy. Direct screening of cancer cell lines and primary samples on panels of targeted inhibitors is a uniquely promising approach to this problem, turning every patient sample into a hundred mini experiments, but clinical validation of in vitro drug-response predictions have been hampered by limited numbers of patients who are screened and actually treated with any given drug. Such an evaluation is critical to determine the predictive value gained from drug screening of patient samples. Over the past seven years the Knight Cancer Institute has performed drug screening paired alongside genomic and/or RNA sequencing for over 600 primary leukemic samples. Within two years, we will have accumulated over 200 patients not only screened for in vitro drug response but then treated with matched targeted inhibitors. I will leverage this existing and growing dataset to interrogate the power of in vitro drug screening data to predict clinical response using retrospective data. I will establish a robust framework for primary drug screening and analysis, build interpretable models for clinical decision making, and explore mechanisms controlling drug response. This project will result in improvements to high-throughput drug screening, a thorough accounting of the predictive power of in vitro drug screening, and candidates for treatment combinations in resistant tumors. My goal is to become an independent investigator and cross-disciplinary leader in patient sample multi-omic profiling, targeted therapy selection, and translational oncology. During my mentored phase I will be receiving guidance from Dr. Emek Demir, an expert in computational modeling of systems biology, Dr. Jeffery Tyner, a leader in patient sample drug screening and validation, and Dr. Brian Druker, a pioneer of targeted cancer therapy and director of the Knight Cancer Institute. I will also improve my statistical understanding of complex systems by working with my advisory committee member Dr Tomi Mori, learn to integrate large datasets and predicting patient outcomes from Dr. Shannon McWeeney, and improve upon existing drug screening platforms and analysis methods with Dr. Laura Heiser. I am determined to attain an independent faculty position and my mentors have committed to assisting me in the application and transition process.
A promising field of cancer treatment is targeted therapy, where each patient is prescribed different drugs precisely based on the features of their cancer. We have a large dataset of patients where we tested drugs on their cancer cells in the lab, and later the patient received the same drug as part of their therapy. I will investigate the best methods for testing drugs on each patient's cancer, how effective these tests are for predicting how well a treatment will work in each patient, and why some patients don't see the benefit from targeted therapy we normally expect.