Oversampling and overdiagnosis of prostate cancer are significant management and cost issues that burden our health care system and the individual at risk with unnecessary biopsies and potential complications. The proposed studies will validate recent advances in quantitative prostate multiparametric MRI (mpMRI) techniques, blood biomarkers of aggressive prostate cancer and radiogenomics that relate to increased aggressive cancer risk by our group and collaborators. The overarching goal is to increase the negative predictive value (NPV) for significant prostate cancer and consequently reduce unnecessary biopsies. Central to the proposal are key collaborations between investigators from the Consortium for Imaging and Biomarkers (CIB), Early Detection Research Network (EDRN), and Jet Propulsion Laboratories (JPL). Novel automated techniques for quantitative analysis of mpMRI that identify prostate habitats at risk of harboring significant prostate cancer (Gleason score 3+4 and above or Grade Group (GG)2+) will be combined with improvements in mpMRI-ultrasound fusion biopsies. Our automated pixel-by-pixel 3D prostate habitat risk scoring (HRS) system is superior to the standard prostate lesion classification system, PIRADSv2, and is hypothesized to improve the Negative Predictive Value (NPV) for significant GG2+ cancers (Aim 1). Radiomics will be applied in Aim 1 to refine HRS in the University of Miami MDSelect protocol of 250 men (discovery=150; validation=100). Just as PIRADSv2 is suboptimal because it does not incorporate quantitative imaging information in risk stratification, models of risk based only on histopathologic grading ignore the underlying genomic determinants of outcome. We have shown that radiomics features are associated with underlying gene expression markers of adverse outcome. We propose in Aim 2 to apply newer criteria that incorporate Decipher score with clinical-pathologic factors to improve the identification of aggressive prostate cancer. Radiomic features associated with these published criteria, termed the Spratt criteria, will improve the NPV for nonaggressive prostate cancer in the MDSelect cohort. We will also collaborate with investigators involved in the EDRN ID-430 clinical trial to test our models in a cohort (n=200) in a less rigorously controlled multi-institutional group with more variability in imaging techniques, vendors and machines. There is also opportunity to further improve risk classification through the analysis of blood-based markers (Aim 3) such as 4Kscore, circulating tumor cells (CTCs) and circulating cancer associated macrophage like (CAML) cells that are early biomarkers of aggressive cancer. The proposed work will test the incremental benefit of adding these serum-based biomarkers to improve the NPV models for significant prostate cancer.
Oversampling and over-diagnosis of prostate cancer are significant management and cost issues that burden our health care system and the individual at risk with unnecessary biopsies and potential complications. The proposed work will combine quantitative imaging features, informed by tumor grade and gene expression, with blood based biomarkers into an objective, automated, tool optimized to achieve a high negative predictive value for significant cancer in undiagnosed men referred for prostate biopsy. Two cohorts, one from the University of Miami (?MDSelect?) and the other from the EDRN will be used to validate/extend that the developed mpMRI based habitat risk scoring system is superior to the more standard PIRADSv2 in identifying significant cancers in the prostate.
