PI-RADS score v.2 in predicting malignancy in patients undergoing 5α-reductase inhibitor therapy.

Background: To investigate if the PI-RADS score represents an effective tool in detecting prostate cancer in patients treated with a 5α-reductase inhibitor (dutasteride) and to identify dedicated total serum PSA and PSA density thresholds.
Methods: Between April 2015 and March 2018, 75 patients under dutasteride treatment underwent multi-parametric magnetic resonance imaging (mpMRI) scans and US/MRI fusion prostate biopsy. Lesions were classified into two groups: Group 1 included PI-RADS 4 and 5 lesions, whilst Group 2 included PI-RADS 3-4-5 lesions. Lesions in groups 1 and 2 were further divided according to the patients' history of previous prostate biopsy.
Results: Ninety-seven lesions were detected. In PI-RADS 4-5 group, mpMRI showed a sensitivity of 78.0% and a specificity of 78.7%. The positive predictive value (PPV) was 79.6% and the negative predictive value (NPV) 77.1%. In PI-RADS 3-4-5 group, sensitivity was 100%, specificity 21.3%, PPV 57.5%, NPV 100%. In PI-RADS 4-5 biopsy-naive group sensitivity was 71.4%, specificity 75.0%. PPV 78.9% and NPV 66.7%. In PI-RADS 4-5 non-biopsy-naive group sensitivity was 82.8%, specificity 80.6%, PPV 80%, and NPV 83.3%. PI-RADS 3-4-5 biopsy-naive group showed sensitivity 100%, specificity 31.3%, PPV 65.6%, NPV 100%. PI-RADS 3-4-5 non-biopsy-naive group showed sensitivity 100%, specificity 16.1%, PPV 52.7%, and NPV 100%. ROC curve analysis indicated a serum total PSA threshold of 6 ng/ml (AUC: 0.71-95% confidence interval: 0.60-0.81) and a PSA density >0.22 (AUC: 0.70-95% confidence interval: 0.6-0.81) as optimal cut-offs for recommending prostate biopsy.
Conclusions: In our experience the PI-RADS score proved to be sufficiently accurate in predicting prostate cancer in patients under dutasteride therapy.