Identifying lethal prostate cancer at diagnosis with advanced proteoglycomic, radiomic, and genomic approaches

PROJECT SUMMARY Conventional prostate adenocarcinoma (PCa) is the second leading cause of cancer death in American men. Patients with organ-confined disease are candidates for potentially curative treatment by either radical prostatectomy or radiation therapy. However, 20-40% of patients undergoing radical prostatectomy and 30-50%

of patients undergoing radiation therapy can experience biochemical recurrence within 10-years. These findings indicate that there is suboptimal identification of lethal PCa at the time of diagnosis. Therefore, identification of aggressive disease at the time of diagnosis could stratify patients, develop more effective therapy options, and

extend survival. In the clinical setting, noninvasive imaging biomarkers are routinely measured with multiparametric magnetic resonance imaging (mpMRI). However, mpMRI has multiple limitations that result in reduced sensitivity and specificity for PCa, in part from obscuration from inflammatory or stromal cells in the

prostate. This proposal advances the use of a clinical magnetic resonance imaging (MRI) sequence, diffusion basis spectral imaging (DBSI), that has the ability to detect structural and cellular changes in the PCa microenvironment (e.g., stroma, inflammation, tumor), that cannot otherwise be determined with conventional

mpMRI, a significant advancement. In parallel, our team has discovered a panel of extracellular proteoglycomic biomarkers in lethal forms of PCa (i.e., fucosylated glycans and modified collagens—“FuCol” biomarkers) with Matrix Assisted Laser Desorption Ionization (MALDI) mass spectrometry imaging of histologic specimens. These

molecular markers provide insight into the structural derangements of lethal PCa and because structural changes affect water diffusion, it suggests that these structural changes may actually be detectable with DBSI. We hypothesize that MALDI-detected proteoglycomic markers, expressed as the FuCol score, are associated with

structural and metabolic changes in lethal PCa that can be visualized with DBSI to better identify aggressive, potentially lethal PCa at the time of diagnosis. In the first Aim, we will continue to validate our FuCol score as a predictor of disease recurrence and metastasis in a large institutional biorepository. In this Aim, we will

investigate the effects of race and diet on the FuCol score and its ability to predict poor outcomes. We will also establish the ability to measure a FuCol score as part of a “noninvasive liquid biopsy” to predict outcomes. In Aim 2, we will enroll a prospective cohort of prostatectomy patients to develop “Diffusion Molecular Imaging

(DMI)”; an AI-driven tool that generates in vivo FuCol scores using in vivo DBSI as its input prior to prostatectomy, hence a non-invasive imaging readout of lethal disease. In Aim 3, we will develop an augmented risk prediction model that incorporates novel DBSI imaging, the clinical Decipher genomics platform, and conventional clinical

metrics (grade, stage, PSA) to better predict lethal disease at prostatectomy. In summary, these experiments will result in rapid acceleration of a clinically-ready workflow to detect molecular biomarkers associated with poor outcomes. This will dovetail with parallel strategies that our group is developing to treat these cohorts of patients

with lethal prostate cancer variants.