Multiparametric MRI prostate PI-RAD scoring in a district general hospital: correlating PI-RADS 3 results with histological findings.

Abstract

Prostate cancer is the most common male cancer in the UK. In many hospitals, patients are now being referred for a multiparametric (mp) MRI scan of their prostate as part of an evaluation for the presence of prostate cancer, prior to an ultrasound-guided biopsy. Prostate Imaging Reporting and Data System (PI-RADS) score of 3 are defined as "equivocal" for the presence of prostate cancer. Thus, a PI-RADS 3 lesion does not confidently determine whether there is significant prostate disease or not.Our aim is to determine the correlation of PI-RADS 3 prostatic lesions with histology proven, clinically significant cancer. We performed a retrospective review on a cohort of 143 consecutive patients. Each patient underwent a mp-MRI scan of their prostate given a PI-RADS score. PI-RADS 3 lesions were analysed further based on histology and categorised into malignant and non-malignant lesions. Prostate-specific antigen (PSA) results and prostatic volume of PI-RADS 3 lesions were also analysed. We identified 45 patients with PI-RADS 3 lesions out of 143 patients. 32 patients subsequently underwent transrectal/transperineal ultrasound-guided biopsy. 43% of patients were found to have had a malignant prostatic adenocarcinoma on histology. The remaining 56% had non-malignant findings. Of those with malignant disease, there was a higher median PSA and lower mean prostatic volume. The study confirms that a score of PI-RADS 3 does not accurately differentiate between malignant and non-malignant lesions. Further investigations such as ultrasound-guided prostate biopsy and PSA parameters are required to accurately ascertain the nature of a prostate lesion with PI-RADS score 3. An ultrasound-guided prostate biopsy in patients with PI-RADS 3 remains of paramount importance when distinguishing malignant vs non-malignant lesions. Multicentre data of MRI findings with PI-RADS 3 scores is required to yield a sample size large enough to carry out statistical analysis.