Combining Therapeutic Response Genomic Signatures with Machine Learning to Predict Patient Outcome from Potentially Lethal Prostate Cancer
Abstract
What are the likely contributions of this study to the FY19 PCRP Overarching Challenges? This project will have a direct impact on the Fiscal Year 2019 (FY19) challenge by defining the biology of lethal prostate cancer to reduce death. By testing treatment-based gene expression signatures, we will have the required information to determine whether a particular patient would benefit more from radiation or surgery. By understanding the biology of high-risk prostate cancer, we can reduce the risk of radiation toxicity. Also, we will have ability to incorporate all of our gene expression data with clinical parameters, such as prostate specific antigen (PSA), PSA density, pathology, and magnetic resonance imaging (MRI) score, to generate a novel treatment prediction tool that will improve the outcome for men with potentially lethal prostate cancer. For example, this tool has the capacity to help ensure that patients receive the treatment that is most likely to be efficacious for them and, further, this clinical decision-making tool can potentially accelerate the decision for care, for which time to treatment is critical for long-term prognosis and outcome as well. What types of patients will it help, and how will it help them? This project will directly help active duty Service members, Veterans, military beneficiaries, and the American public that have high-risk prostate cancers in which treatment is imminent. It will help these individuals by possibly avoiding unnecessary radiation, especially those individuals that have a gene signature that demonstrates radiation resistance is highly likely. Further, this project has the capacity to facilitate timely access to effective treatment, leading to an overall decrease in number of doctors’ visits and overall healthcare costs. Importantly, more effective treatment outcomes will result in increased time that active duty Service members, Veterans, and their families have together. What are the potential clinical applications, benefits, and risks? All patients diagnosed with prostate cancer will need to get their pathology confirmed before any treatment is started, which would mean that a prostate biopsy would need to be performed. The tissue core biopsy then would be sent for pathologic grading, and the same tissue could be used to perform a treatment-based gene expression assay. The outcome from this test would inform the patient about whether their prostate tumor is radioresistant or radiosensitive. This indicates that our study is highly beneficial and has high clinical applicability. The risk of this study would be low, since we would rely on our completed decision-making tool and not just one aspect of it. In addition to relying on the tumor biology, we would rely on other clinical parameters. Therefore, to circumvent this potential issue, our prostate cancer treatment prediction tool will be based on machine learning algorithms. These algorithms will take into consideration all available clinical parameters including PSA, PSA density, pathology grading, and MRI score, as well as the gene expression signatures. Also, the algorithm is designed to ensure that, if one clinical parameter were inadvertently omitted, an answer could still be derived. What is the projected time it may take to achieve a patient-related outcome? Since we are dealing directly with prostate cancer patient specimens and data, we expect to achieve a patient-related outcome at the conclusion of this grant (i.e., 3 years). During this time, we will have identified the gene signatures that will provide the most effective assistance in stratifying prostate cancer patients to the most appropriate treatment options. Specifically, we will be able to identify whether a patient is radioresistant, meaning that the patient would not have undergone radiation therapy, but would have the best treatment outcome by undergoing a radical prostatectomy. In addition, we would use the data we g
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Mar 10, 2021
- Source ID
- W81XWH2010297
Entities
People
- Darryl Martin
Organizations
- United States Army
- Yale University