Poor reproducibility of PIRADS score in two multiparametric MRIs before biopsy in men with elevated PSA

Since January 2015, all men referred to urologists in Norway due to elevated PSA or other suspicion of prostate cancer underwent multiparametric MRI (mpMRI) before prostate biopsy. At our hospital, patients and the initial MRI were assessed by an urologist and if deemed necessary, patients were referred to another institution for MR/US fusion biopsies. Before MR/US biopsy, patients underwent a second mpMRI. Since we noticed disagreement of these two mpMRIs before biopsy, we retrospectively assessed the level of agreement between the two mpMRIs from the two institutions.

During the first 6 months of 2015, 292 patients were referred to our outpatient clinic. We referred 126 patients of these to the other institution for MR/US fusion biopsy. The 2 mpMRIs were performed within 4 weeks. We analyzed MR reports and schematics for number of lesions and highest PIRADS score per side of the prostate and histological result of the biopsies. Bland-Altman's plot was used to compare the level of agreement between the two mpMRIs of the same patient before biopsy.

There was a poor level of agreement between the two mpMRIs and a statistically significant difference in PIRADS scores. Regression analysis showed that there was no proportional or systematic bias.

In unselected patients with elevated PSA, there seems to be a significant variation of mpMRI results across institutions. The PIRADS scoring system needs to be validated with regards to MR equipment, mpMRI protocols and inter-reader variability of radiologists.