Prediction Of Local Pathologic Findings Based On MRI-Based Clinical Staging In Prostate Cancer: A Radiation Oncologist Perspective

Abstract

Predicting the risk of extracapsular extension (ECE) and seminal vesicle invasion (SVI) is critical for definitive radiotherapy in prostate cancer. Preoperative magnetic resonance imaging (MRI) is known for its predictive value for those findings. This study aimed to develop a model for predicting pathologic findings integrating MRI-based clinical T-staging (cTMRI, cT1c–cT3b). A total of 1915 treatment-naive patients between 2006–2016 who underwent preoperative MRI before radical prostatectomy met the inclusion/exclusion criteria. We performed a multivariate logistic regression analysis as well as a Bayesian Network (BN) modeling based on possible confounding factors. The BN model was further validated by a 5-fold validation. On multivariate logistic regression analysis, initial PSA (iPSA) (β = 0.050; p<0.001), percentage of positive biopsy cores (PPC) (β = 0.033; p<0.001), both lobe involvement on biopsy (β = 0.359; p-0.009), Gleason score (GS) (β = 0.358; p<0.001), and cTMRI (β = 0.259; p<0.001) were significant factors for ECE. For SVI, iPSA (β = 0.037; p<0.001), PPC (β = 0.024; p<0.001), GS (β = 0.753; p<0.001), and cTMRI (β = 0.507; p<0.001) showed statistical significance. BN models to predict ECE and SVI were also successfully established. PPC (normalized MI, 8.22%; p<0.001) and cTMRI (normalized MI, 9.85%; p<0.001) were the most important factors in each BN model for ECE and SVI, respectively, in G-test. The overall AUC/accuracy of BN models were 0.76/73.0% and 0.88/89.6% for ECE and SVI, respectively. Two models to predict pathologic ECE and SVI integrating cTMRI were established. These models will be installed on a separate website for public access to guide radiation oncologists in the future.