Studying the Utility of Deep Learning Derived MRI Imaging Biomarkers for Localized Prostate Cancer

Prostate cancer remains a significant public health concern and remains the second leading cause of cancer death among men in the United States. The traditional methods for risk stratification of localized prostate cancer using a combination of Gleason grade sum and prostate specific antigen (PSA) levels are limited, resulting in wide variations in outcomes among patients. Although tissue-based genomic biomarkers show promise in improving risk stratification, they often require additional testing and specialized equipment, which may not be scalable in all clinical settings. More importantly, these biomarkers limit their analysis to a small percentage of the prostate and do not evaluate the entire prostate gland. Quantitative imaging has proven prognostic in several diseases and represents a potential low-cost biomarker to overcome challenges associated with tissue-based biomarkers. Multiparametric prostate magnetic resonance imaging (MRI) has improved the diagnosis of clinically significant prostate cancer and represents the ideal imaging type for quantitative image analysis. Our group has developed an MRI biomarker for localized prostate cancer using deep learning, which can identify patients at highest risk for disease recurrence. Previous attempts at developing MRI biomarkers for prostate cancer have shown promise in such applications but have failed to make a clinical impact because they lacked robust external validation and faced informatics challenges in clinical implementation. Lastly, although published MRI biomarkers have shown to be prognostic, there are currently no predictive MRI biomarkers that directly address challenges associated with treatment selection. In Aim 1, we plan to validate the prognostic ability of MRI biomarkers on a multi-institutional cohort of 2,250 prostate cancer patients representing different demographics, genomic/clinical risk groups, and MRI manufacturers. In Aim 2, through unique collaborations with an industry partner, we will address informatics barriers which have previously limited the impact of previous attempts at developing MRI biomarkers for prostate cancer. In Aim 3, we will leverage two recently completed multi-institutional randomized clinical trials to identify the first predictive MRI biomarkers which predict the benefits of various therapeutic strategies for localized prostate cancer and directly address ongoing challenges associated with prostate cancer treatment selection. Upon successful completion of our project, we will have a validated low-cost, non-invasive MRI biomarker, which can aid in treatment selection for localized prostate cancer patients. Project Number: 1R37CA307500-01A1 | Fiscal Year: 2026 | NIH Institute/Center: National Cancer Institute (NCI) | Principal Investigator: Sanjay Aneja | Institution: YALE UNIVERSITY, NEW HAVEN, CT | Award Amount: $519,592 | Activity Code: R37 | Study Section: Clinical Data Management and Analysis Study Section[CDMA] View on NIH RePORTER: https://reporter.nih.gov/project-details/11449993