Radiomics Features Measured with Multiparametric Magnetic Resonance Imaging Predict Prostate Cancer Aggressiveness.

PURPOSE: We sought to 1) assess the association of radiomics features based on multiparametric magnetic resonance imaging with histopathological Gleason score, gene signatures and gene expression levels in prostate cancer and 2) build machine learning models based on radiomics features to predict adverse histopathological scores and the Decipher® genomics metastasis risk score. MATERIALS AND METHODS: We retrospectively analyzed the records of 64 patients with prostate cancer with a mean age of 64 years (range 41 to 76) who underwent magnetic resonance imaging between January 2016 and January 2017 before radical prostatectomy. A total of 226 magnetic resonance imaging radiomics features, including histogram and texture features in addition to lesion size and the PI-RADS™ (Prostate Imaging Reporting and Data System) score, were extracted from T2-weighted, apparent diffusion coefficient and diffusion kurtosis imaging maps. Radiomics features were correlated with the pathological Gleason score, 40 gene expression signatures, including Decipher, and 698 prostate cancer related gene expression levels. Cross-validated, lasso regularized, logistic regression machine learning models based on radiomics features were built and evaluated for the prediction of Gleason score 8 or greater and Decipher score 0.6 or greater. RESULTS: A total of 14 radiomics features significantly correlated with the Gleason score (highest correlation r = 0.39, p = 0.001). A total of 31 texture and histogram features significantly correlated with 19 gene signatures, particularly with the PORTOS (Post-Operative Radiation Therapy Outcomes Score) signature (strongest correlation r = -0.481, p = 0.002). A total of 40 diffusion-weighted imaging features correlated significantly with 132 gene expression levels. Machine learning prediction models showed fair performance to predict a Gleason score of 8 or greater (AUC 0.72) and excellent performance to predict a Decipher score of 0.6 or greater (AUC 0.84). CONCLUSIONS: Magnetic resonance imaging radiomics features are promising markers of prostate cancer aggressiveness on the histopathological and genomics levels.

Defining Prostate Cancer at Favorable Intermediate Risk: The Potential Utility of Magnetic Resonance Imaging and Genomic Tests.

PURPOSE: We determined whether prostate multiparametric magnetic resonance imaging and genomic biomarkers might help further define patients with favorable intermediate risk prostate cancer which could safely be considered suitable for active surveillance. MATERIALS AND METHODS: From our institutional database we identified 509 patients who underwent radical prostatectomy with preoperative magnetic resonance imaging and a postoperative Decipher® prostate cancer test. According to the NCCN® (National Comprehensive Cancer Network®) risk stratification 125 men had favorable intermediate and 171 had unfavorable intermediate risk disease. Univariable and multivariable binary logistic regression analyses were done to test the utility of different variables in predicting adverse pathology, defined as Gleason Grade Group greater than 2, pT3b or pN1. RESULTS: On univariable analysis favorable intermediate risk, multiparametric magnetic resonance imaging and the prostate cancer test significantly predicted adverse pathology. On multivariable analysis favorable intermediate risk and the prostate cancer test maintained independent predictive value while multiparametric magnetic resonance imaging did not meet statistical significance (p = 0.059). The 19 patients at favorable intermediate risk with high genomic risk had an adverse pathology rate slightly higher than patients at unfavorable intermediate risk (42.1% vs 39.8%, p = 0.56). Those at low genomic risk had an adverse pathology rate slightly lower than patients at very low or low risk (7.5% vs 10.2%, p = 0.84). The 31 patients at favorable intermediate risk but at high multiparametric magnetic resonance imaging and genomic risk had an adverse pathology rate slightly lower than patients at unfavorable intermediate risk (25.8% vs 39.8%, p = 0.14). Those at low multiparametric magnetic resonance imaging and genomic risk had an adverse pathology rate slightly lower than patients at very low or low risk (8.5% vs 10.2%, p = 0.89). CONCLUSIONS: Multiparametric magnetic resonance imaging and the Decipher test allowed us to better define the risk of adverse pathology in patients at favorable intermediate risk who were diagnosed with prostate cancer.

Multiparametric Magnetic Resonance Imaging Features Identify Aggressive Prostate Cancer at the Phenotypic and Transcriptomic Level.

PURPOSE: Multiparametric magnetic resonance imaging is a diagnostic tool for prostate cancer with limited data on prognostic use. We sought to determine whether multiparametric magnetic resonance could predict aggressive prostate cancer features. MATERIALS AND METHODS: We retrospectively analyzed the records of 206 patients who underwent radical prostatectomy between 2013 and 2017. All patients had available RNA expression data on the final pathology specimen obtained from a location corresponding to a lesion location on multiparametric magnetic resonance imaging. The association between the PIRADS™ (Prostate Imaging Reporting and Data System) score and adverse pathology features were analyzed. We also performed differential transcriptomic analysis between the PIRADS groups. Factors associated with adverse pathology were analyzed using a multivariable logistic regression model. RESULTS: Lesion size (p = 0.03), PIRADS score (p = 0.02) and extraprostatic extension (p = 0.01) associated significantly with the Decipher® score. Multivariable analysis showed that the PIRADS score (referent PIRADS 3, OR 8.1, 95% CI 1.2-57.5, p = 0.04), the Gleason Grade Group (referent 3, OR 5.6, 95% CI 1.5-21.1, p = 0.01) and prostate specific antigen (OR 1.103, 95% CI 1.011-1.203) were risk factors for adverse pathology findings. The difference between PIRADS 4 and 5 did not reach significance (OR 1.9, 95% CI 0.8-4.5, p = 0.12). However, the PI3K-AKT-mTOR, WNT-ß and E2F signaling pathways were more active in PIRADS 5 than in PIRADS 4 cases. CONCLUSIONS: The PIRADS score is associated with adverse pathology results, increased metastatic risk and differential genomic pathway activation.