Prostate cancer risk stratification via computational 3D pathology

点击翻译按钮获取中文摘要

Summary. Prostate cancer (PCa) treatment management is currently heavily reliant upon slide-based histology of prostate biopsies and surgical specimens (prostatectomies). In particular, Gleason grading of histology sections provides a basis for stratifying patients for clinical management, and can result in dramatically different treatment paths. However, prognostication via Gleason grading suffers from several shortcomings, including subjective visual interpretation of complex 3D glandular morphologies based on 2D images, and analysis of a limited amount of tissue (~1% of the biopsy). These shortcomings contribute to poor inter-observer concordance amongst pathologists and poor stratification of patients with indolent vs. lethal disease. For the clinical management of PCa, two major challenges faced by urologists and oncologists, respectively, are: (1) correctly identifying men with low-risk PCa for active surveillance and (2) identifying men who are likely to have disease recurrence and metastasis after curative therapy (surgery or radiation), and hence would benefit from adjuvant therapy. With our open-top light-sheet (OTLS) microscope technologies, our team at the University of Washington (Liu group) has demonstrated the technical feasibility of achieving high-throughput slide-free 3D histology of biopsy and surgical specimens in a nondestructive and reversible manner that does not interfere with current histology methods. Potential benefits over traditional pathology include: (1) comprehensive imaging of specimens (biopsies and surgical bread loafs) rather than sparse sampling of thin sections on glass slides; (2) volumetric imaging of 3D structures that are prognostic; and (3) non-destructive imaging, which allows valuable biopsy specimens to be used for downstream assays. Our team at Case Western Reserve University (Madabhushi group) has also developed computational pathology classifiers, based on intuitive and interpretable “hand-crafted features,” for characterization of PCa aggressiveness based on 2D whole-slide imaging (WSI). In this R01 project, we seek to combine nondestructive 3D pathology with 3D computational pathology approaches to develop a novel prognostic assay, Prostate cancer Image Risk Score via 3D pathology (ProsIRiS3D), for discriminating between indolent and aggressive PCa. In Aim 1, we will develop the core technologies (hardware and software) for ProsIRiS3D. In particular, the goal of Aim 1a is to develop a “4th-generation” OTLS microscopy system capable of achieving sub-nuclear-resolution to explore the added prognostic benefit provided by such high-resolution features. In Aim 1b, computational imaging tools will be developed for extraction of novel 3D quantitative histomorphometric features for PCa prognostication. Our clinical validation studies will show that ProsIRiS3D is superior to analogous 2D approaches for urologists (Aim 2), to determine which newly biopsied patients should be placed on active surveillance vs. curative therapy, as well as for oncologists (Aim 3), to determine which prostatectomy patients have aggressive disease that may warrant adjuvant therapies.