Imaging features of the PI-RADS for predicting extraprostatic extension of prostate cancer: systematic review and meta-analysis

Abstract

Abstract Objectives To systematically determine the diagnostic performance of each MRI feature of the PI-RADS for predicting extraprostatic extension (EPE) in prostate cancer. Methods A literature search in the MEDLINE and EMBASE databases was conducted to identify original studies reporting the accuracy of each feature on MRI for the dichotomous diagnosis of EPE. The meta-analytic pooled diagnostic odds ratio (DOR), sensitivity, specificity, and their 95% confidence intervals (CIs) were obtained using a bivariate random-effects model. Results After screening 1955 studies, 17 studies with a total of 3062 men were included. All six imaging features, i.e., bulging prostatic contour, irregular or spiculated margin, asymmetry or invasion of neurovascular bundle, obliteration of rectoprostatic angle, tumor-capsule interface > 10 mm, and breach of the capsule with evidence of direct tumor extension, were significantly associated with EPE. Breach of the capsule with direct tumor extension demonstrated the highest pooled DOR (15.6, 95% CI [7.7–31.5]) followed by tumor-capsule interface > 10 mm (10.5 [5.4–20.2]), asymmetry or invasion of neurovascular bundle (7.6 [3.8–15.2]), and obliteration of rectoprostatic angle (6.1 [3.8–9.8]). Irregular or spiculated margin showed the lowest pooled DOR (2.3 [1.3–4.2]). Breach of the capsule with direct tumor extension and tumor-capsule interface > 10 mm showed the highest pooled specificity (98.0% [96.2–99.0]) and sensitivity (86.3% [70.0–94.4]), respectively. Conclusions Among the six MRI features of prostate cancer, breach of the capsule with direct tumor extension and tumor-capsule interface > 10 mm were the most predictive of EPE with the highest specificity and sensitivity, respectively. Graphical abstract