Prognostic Radiomic Signatures in Prostate Cancer Patients on Active Surveillance

PROJECT SUMMARY/ABSTRACT With prostate cancer (PCa) being among the most common cancers in men worldwide (estimated 1,600,000 cases, 366,000 deaths annually), the need for new biomarkers for early detection, diagnosis, monitoring, and prognosis remains urgent. Among the available management alternatives for PCa, active surveillance (AS) is

recommended as an initial treatment for males with very low-, low- and favorable intermediate-risk. AS relies on serial monitoring over time to identify progression, so that the patient receives timely curative treatment, while reducing morbidities related to definite treatment delivered at time of diagnosis. However, identifying ideal

candidates for AS is challenging. Despite its limited specificity, the prostate-specific antigen (PSA) is the most used test for early detection of PCa. Other factors based on biopsies such as the Gleason Group (GG), are affected by limited biopsy sampling, while the non-invasive magnetic resonance imaging (MRI) has been

connected to false positives and false negatives. Finally, the implementation of molecular prognostic tests, such as Decipher in AS populations has been limited due to the lack of randomized trials using actual AS patients. Previous work suggests that PCa progression can be dependent on the interactions between extracellular matrix

(ECM) proteins in the stroma with various cell types including the immune system and cancer cells. On this regard, my team of collaborators has developed clinical imaging techniques such as, diffusion basis spectrum imaging (DBSI), and matrix-assisted laser desorption ionization (MALDI) mass spectrometry, that visualize

inflammatory, stromal/ECM, and cancer cell components of the tumor that could be associated with cancer progression. Therefore, I propose to leverage radiomics, a method based on data-characterization algorithms to extract imaging features, to detect patterns in pre-op MRI with the guidance of DBSI and MALDI toward optimally

selecting AS candidates. My central hypothesis is that the spatial analysis of structural components in the ECM extracted from the co-registration of molecular and radiological imaging, accurately predicts tumor upgrading and upstaging in PCa. In Aim 1 I will identify a baseline radiomic signature derived from pre-op MRI to accurately

predict tumor upgrading (i.e., from GG1 to GG2 or higher) by augmenting well-established biomarkers (i.e., PSA, GG, Decipher), with an exploratory SubAim co-registering DBSI and MRI to improve the prediction. In Aim 2, I will use MALDI co-registered with pre-op MRI to guide the derivation of the radiomic signature to accurately

predict tumor upstaging (i.e., from T1/T2 stage to T3 or higher). This research is innovative because, to date, no distinct spatial signatures linked with the ECM and derived from co-registered molecular and radiological imaging have been associated with prediction of tumor progression in PCa. Furthermore, this K22 career transition award

will provide me with the training and resources needed to advance my career as an independent researcher in the field of data sciences applied to cancer research, and also to support my goal of recruiting the nation’s most talented students from backgrounds nationally underrepresented in cancer research.