Does the Visibility of Grade Group 1 Prostate Cancer on Baseline Multiparametric Magnetic Resonance Imaging Impact Clinical Outcomes?

PURPOSE: We assessed whether the visibility of Grade Group (GG) 1 prostate cancer on baseline multiparametric magnetic resonance imaging affects clinical outcomes. MATERIALS AND METHODS: We evaluated 454 men who underwent multiparametric magnetic resonance imaging between 2006 and 2018 with maximum GG1 prostate cancer inclusive of magnetic resonance imaging targeted biopsy. Multiparametric magnetic resonance imaging was graded as negative, equivocal or positive. Assessed outcomes were treatment-free survival, biopsy upgrade-free survival and unfavorable disease at radical prostatectomy (pT 3 or greater and/or GG3 or greater). Kaplan-Meier and multivariable Cox proportional hazard analyses were used to estimate the impact of multiparametric magnetic resonance imaging and clinicopathological variables (age, year, prostate specific antigen density and measures of tumor volume on biopsy) on outcomes. RESULTS: During followup (median 45.2 months) 61 men had disease upgraded on followup biopsy and 139 underwent definitive treatment. In men with negative, equivocal and positive baseline multiparametric magnetic resonance imaging at 5 years, treatment-free survival was 79%, 73% and 49% (p <0.0001), treatment-free survival was 89%, 82% and 70% (p=0.002), and survival without unfavorable disease at radical prostatectomy was 98%, 98% and 86% (p=0.007), respectively. At multivariable analysis positive (HR 1.93, 95% CI 1.21-3.09, p=0.006) and equivocal multiparametric magnetic resonance imaging (HR 2.02, 95% CI 1.11-3.68, p=0.02) were associated with shorter treatment-free survival, and positive multiparametric magnetic resonance imaging was a significant prognostic factor for upgrade-free survival (HR 2.03, 95% CI 1.06-3.86, p=0.03) and unfavorable disease at radical prostatectomy (HR 4.45, 95% CI 1.39-18.17, p=0.01). CONCLUSIONS: Men with positive multiparametric magnetic resonance imaging and GG1 prostate cancer on magnetic resonance imaging targeted biopsy are at increased risk for intervention, upgrading and unfavorable disease at radical prostatectomy compared to those with multiparametric magnetic resonance imaging invisible GG1 prostate cancer.

Using decision curve analysis to benchmark performance of a magnetic resonance imaging-based deep learning model for prostate cancer risk assessment.

OBJECTIVES: To benchmark the performance of a calibrated 3D convolutional neural network (CNN) applied to multiparametric MRI (mpMRI) for risk assessment of clinically significant prostate cancer (csPCa) using decision curve analysis (DCA). METHODS: We retrospectively analyzed 499 patients who had positive mpMRI (PI-RADSv2 ≥ 3) and MRI-targeted biopsy. The training cohort comprised 449 men, including a calibration set of 50 men. Biopsy decision strategies included using risk estimates from the CNN (original and calibrated), to perform biopsy in men with PI-RADSv2 ≥ 4 only, or additionally in men with PI-RADSv2 3 and PSA density (PSAd) ≥ 0.15 ng/ml/ml. Discrimination, calibration and clinical usefulness in the unseen test cohort (n = 50) were assessed using C-statistic, calibration plots and DCA, respectively. RESULTS: The calibrated CNN achieved moderate calibration (Hosmer-Lemeshow calibration test, p = 0.41) and good discrimination (C = 0.85). DCA revealed consistently higher net benefit and net reduction in biopsies for the calibrated CNN compared with the original CNN, PI-RADSv2 ≥ 4 and the combined strategy of PI-RADSv2 and PSAd. Original CNN predictions were severely miscalibrated (p < 0.0001) resulting in net harm compared with a 'biopsy all' patients strategy. At-risk thresholds ≥ 10% using the calibrated CNN and the combined strategy reduced the number of biopsies by an estimated 201 and 55 men, respectively, per 1000 men at risk, without missing csPCa, while original CNN and PI-RADSv2 ≥ 4 could not achieve a net reduction in biopsies. CONCLUSIONS: DCA revealed that our calibrated 3D-CNN resulted in fewer unnecessary biopsies compared with using PI-RADSv2 alone or in combination with PSAd. CNN calibration is important in achieving clinical utility. KEY POINTS: • A 3D deep learning model applied to multiparametric MRI may help to prevent unnecessary prostate biopsies in patients eligible for MRI-targeted biopsy. • Owing to miscalibration, original risk estimates by the deep learning model require prior calibration to enable clinical utility. • Decision curve analysis confirmed a net benefit of using our calibrated deep learning model for biopsy decisions compared with alternative strategies, including PI-RADSv2 alone and in combination with prostate-specific antigen density.

Reducing Unnecessary Prostate Multiparametric Magnetic Resonance Imaging by Using Clinical Parameters to Predict Negative and Indeterminate Findings.

PURPOSE: We sought to develop a triage strategy to reduce negative and indeterminate multiparametric magnetic resonance imaging scans in patients at risk for prostate cancer. MATERIALS AND METHODS: In this retrospective study we evaluated 865 patients with no prior prostate cancer diagnosis who underwent prostate multiparametric magnetic resonance imaging between 2009 and 2017. Age, prostate volume, prostate specific antigen and prostate specific antigen density were assessed as predictors of positive multiparametric magnetic resonance imaging, defined as PI-RADS™ (Prostate Imaging Reporting and Data System) version 2/Likert score 4 or greater. The cohort was split into a training cohort of 605 patients and a validation cohort of 260. The optimal threshold to rule out positive multiparametric magnetic resonance imaging was chosen to achieve a negative predictive value greater than 90%. RESULTS: All clinical variables were significant predictors of positive multiparametric magnetic resonance imaging (p <0.05). Prostate specific antigen density outperformed other parameters in diagnostic accuracy and did not significantly differ compared to a multivariate model (AUC=0.74 vs 0.75). At prostate specific antigen density greater than 0.078 ng/ml2 sensitivity, specificity, positive and negative predictive values were 94%, 29%, 22% and 95%, respectively, resulting in 25% fewer scans (64 of 260). In the multivariate model sensitivity, specificity, positive and negative predictive values were 85%, 32%, 22% and 91%, respectively, resulting in 29% fewer scans (75 of 260). Biopsies in men who would not have undergone multiparametric magnetic resonance imaging according to our proposed strategies revealed 2 clinically significant prostate cancers using prostate specific antigen density and 1 using the multivariate model. CONCLUSIONS: In patients at risk for prostate cancer applying a multivariate prediction model or a prostate specific antigen density cutoff of 0.078 ng/ml2 resulted in 25% to 29% fewer multiparametric magnetic resonance imaging scans performed while missing only a minimal number of clinically significant prostate cancers. Further prospective validation is required.