A Novel Imaging Based Nomogram for Predicting Post-Surgical Biochemical Recurrence and Adverse Pathology of Prostate Cancer from Pre-Operative Bi-Parametric MRI

Prostate cancer is the most common cancer among men in the United States of America. Even though 1 in 7 men will be diagnosed with prostate cancer in the US in their lifetime, only 1 in 40 will die on account of it. Consequently, there is a need for better approaches to identify which men can be spared radical therapies and instead be candidates for active surveillance. However, annually roughly 29,000 men in the US still die on account of prostate cancer, many of these men have received radical therapy. Hence there is also a need to identify which men following radical therapy (either surgery or radiation) will benefit from additional therapy to further reduce the risk of post-treatment metastasis and disease-specific mortality. While a number of risk calculators for risk stratification of prostate cancer exist, most of these depend on the use of clinical parameters from the surgical specimens for outcome prediction. Similarly, gene expression-based assays have been presented to predict the risk of post-surgical metastasis, however, these assays are expensive and tissue destructive.

In this study, we sought to use computational image analysis to extract textural patterns from diagnostic magnetic resonance imaging (MRI) to evaluate prostate cancer aggressiveness, predict adverse pathology on the surgical specimens, and risk of biochemical recurrence (BCR) following surgery. Specifically, a nomogram, RadClip, was built by integrating texture features from bi-parametric MRI with clinical parameters, such as biopsy Gleason grade group and pre-operative prostate-specific antigen (PSA) level. The texture feature extracted from diagnostic MRI was able to capture subvisual patterns to distinguish more from less aggressive prostate cancer on MRI. The predictions from RadClip were found to be strongly associated with the risk of biochemical recurrence and the presence of adverse pathology on the surgical specimen, a predictor for worse prognosis. RadClip outperformed popular and extant risk calculators for prostate cancer in a multi-site validation set. Our results suggest that with additional validation, RadClip could be used to assess the risk of adverse pathology on surgical specimens and hence could allow surgeons to preemptively adjust their surgical plan. Furthermore, RadClip could also be used to help identify which men with prostate cancer are at increased risk for post-surgical recurrence and hence candidates for additional therapy following prostatectomy.

Written by: Lin Li, Rakesh Shiradkar, Patrick Leo, Ahmad Algohary, Pingfu Fu, Sree Harsha Tirumani, Amr Mahran, Christina Buzzy, Verena C Obmann, Bahar Mansoori, Ayah El-Fahmawi, Mohammed Shahait, Ashutosh Tewari, Cristina Magi-Galluzzi, David Lee, Priti Lal, Lee Ponsky, Eric Klein, Andrei S Purysko, Anant Madabhushi

Center for Computational Imaging and Personalized Diagnostics, Case Western Reserve University, Cleveland, OH, USA., Department of Population and Quantitative Health Sciences, Case Western Reserve University, Cleveland, OH, USA., Department of Radiology, University Hospitals, Cleveland, OH, USA., Urology Institute, University Hospitals Cleveland Medical Center, Cleveland, OH, USA., Department of Radiology, Case Western Reserve University and University Hospitals Cleveland Medical Centers, Cleveland, OH, USA; Department of Diagnostic, Interventional and Pediatric Radiology (DIPR), Inselspital, Bern University Hospital, University of Bern, Switzerland., Department of Radiology, Abdominal Imaging Division, University of Washington, Seattle, WA, USA., Penn Medicine, University of Pennsylvania Health System, PA, USA., Department of Radiology, Icahn School of Medicine at Mount Sinai, New York, New York, USA., Department of Pathology, University of Alabama at Birmingham, AL, USA., Urology Institute, University Hospitals Cleveland Medical Center, Cleveland, OH, USA; Case Western Reserve University School of Medicine, Cleveland, OH, USA., Glickman Urological and Kidney Institute, Cleveland Clinic, Cleveland, OH, USA., Glickman Urological and Kidney Institute, Cleveland Clinic, Cleveland, OH, USA; Imaging Institute, Cleveland Clinic, Cleveland, OH, USA., Center for Computational Imaging and Personalized Diagnostics, Case Western Reserve University, Cleveland, OH, USA; Louis Stokes Cleveland Veterans Administration Medical Center, USA. Electronic address: .

Read the Abstract

Login