Evaluating the performance of clinical and radiological data in predicting prostate cancer in prostate imaging reporting and data system version 2.1 category 3 lesions of the peripheral and the transition zones

An Apparent Diffusion Coefficient-Based Machine Learning Model Can Improve Prostate Cancer Detection in the Grey Area of the Prostate Imaging Reporting and Data System Category 3: A Single-Centre Experience

The Prostate Imaging and Reporting Data System (PI-RADS) has a key role in the management of prostate cancer (PCa). However, the clinical interpretation of PI-RADS 3 score lesions may be challenging and misleading, thus postponing PCa diagnosis to biopsy outcome. Multiparametric magnetic resonance imaging (mpMRI) radiomic analysis may represent a stand-alone noninvasive tool for PCa diagnosis. Hence, this study aims at developing a mpMRI-based radiomic PCa diagnostic model in a cohort of PI-RADS 3 lesions. We enrolled 133 patients with 155 PI-RADS 3 lesions, 84 of which had PCa confirmation by fusion biopsy. Local radiomic features were generated from apparent diffusion coefficient maps, and the four most informative were selected using LASSO, the Wilcoxon rank-sum test (p < 0.001), and support vector machines (SVMs). The selected features where augmented and used to train an SVM classifier, externally validated on a holdout subset. Linear and second-order polynomial kernels were exploited, and their predictive performance compared through receiver operating characteristics (ROC)-related metrics. On the test set, the highest performance, equally for both kernels, was specificity = 76%, sensitivity = 78%, positive predictive value = 80%, and negative predictive value = 74%. Our findings substantially improve radiologist interpretation of PI-RADS 3 lesions and let us advance towards an image-driven PCa diagnosis.

Transverse prostate maximum sectional area can predict clinically significant prostate cancer in PI-RADS 3 lesions at multiparametric magnetic resonance imaging

Background

To evaluate multiparametric magnetic resonance imaging (mpMRI) parameters, such as TransPA (transverse prostate maximum sectional area), TransCGA (transverse central gland sectional area), TransPZA (transverse peripheral zone sectional area), and TransPAI (TransPZA/TransCGA ratio) in predicting prostate cancer (PCa) in prostate imaging reporting and data system (PI-RADS) 3 lesions.

Methods

Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV), the area under the receiver operating characteristic curve (AUC), and the best cut-off, were calculated. Univariate and multivariate analyses were carried out to evaluate the capability to predict PCa.

Results

Out of 120 PI-RADS 3 lesions, 54 (45.0%) were PCa with 34 (28.3%) csPCas. Median TransPA, TransCGA, TransPZA and TransPAI were 15.4cm², 9.1cm², 5.5cm² and 0.57, respectively. At multivariate analysis, location in the transition zone (OR=7.92, 95% CI: 2.70-23.29, P<0.001) and TransPA (OR=0.83, 95% CI: 0.76-0.92, P<0.001) were independent predictors of PCa. The TransPA (OR=0.90, 95% CI: 0.082-0.99, P=0.022) was an independent predictor of csPCa. The best cut-off of TransPA for csPCa was 18 (Sensitivity 88.2%, Specificity 37.2%, PPV 35.7%, NPV 88.9%). The discrimination (AUC) of the multivariate model was 0.627 (95% CI: 0.519-0.734, P<0.031).

Conclusions

In PI-RADS 3 lesions, the TransPA could be useful in selecting patients requiring biopsy.

New Diagnostic Model for Clinically Significant Prostate Cancer in Biopsy-Naïve Men With PIRADS 3

Purpose

The aim of this study was to explore a new model of clinical decision-making to predict the occurrence of clinically significant prostate cancer (csPCa).

Patients and Methods

The demographic and clinical characteristics of 152 patients were recorded. Prostate-specific antigen (PSA), PSA density (PSAD), adjusted PSAD of peripheral zone (aPSADPZ), and peripheral zone volume ratio (PZ ratio) were calculated and subjected to receiver operating characteristic (ROC) curve analysis. The calibration and discrimination abilities of new nomograms were verified with calibration curve and area under the ROC curve (AUC). The clinical benefits of these models were evaluated by decision curve analysis and clinical impact curves.

Results

The AUCs of PSA, PSAD, aPSADPZ, and PZ ratio were 0.521, 0.645, 0.745, and 0.717 for prostate cancer (PCa) diagnosis, while the corresponding values were 0.590, 0.678, 0.780, and 0.731 for csPCa diagnosis, respectively. All nomograms displayed higher net benefit and better overall calibration than the scenarios for predicting the occurrence of csPCa. The new model significantly improved the diagnostic accuracy of csPCa (0.865 vs. 0.741, p = 0.0284) compared with the base model. In addition, the new model was better than the base model for predicting csPCa in the low or medium probability while the number of patients with csPCa predicted by the new model was in good agreement with the actual number of patients with csPCa in the high-risk threshold.

Conclusions

This study demonstrates that aPSADPZ has a higher predictive accuracy for csPCa diagnosis than the conventional indicators. Including aPSADPZ, PZ ratio, and age can improve csPCa diagnosis and avoid unnecessary biopsies.