Expected impact of MRI-targeted biopsy interreader variability among uropathologists on ProScreen prostate cancer screening trial: a pre-trial validation study

Abstract

ABSTRACTBackgroundProstate cancer (PCa) histology, particularly the Gleason score, is an independent prognostic predictor in PCa. Little is known about the inter-reader variability in grading of targeted prostate biopsy based on magnetic resonance imaging (MRI).ObjectiveTo assess inter-reader variability in Gleason grading of MRI-targeted biopsy among uropathologists and its potential impact on a population-based randomized PCa screening trial (ProScreen).Design, setting, and participantsFrom June 2014 to May 2018, 100 men with clinically suspected PCa were retrospectively selected. All men underwent prostate MRI and 86 underwent targeted prostate of the prostate.InterventionSix pathologists individually reviewed the pathology slides of the prostate biopsies.Outcome measurements and statistical analysisThe five-tier ISUP (The International Society of Urological Pathology) grade grouping (GG) system was used. Fleiss’ weighted kappa (κ) and Model based kappa for associations were computed to estimate the combined agreement between individual pathologists.Results and limitationsGG reporting of targeted prostate was highly consistent among the trial pathologists. Inter-reader agreement for cancer (GG 1-5) vs. benign was excellent (Model-based kappa 0.90, Fleiss’ kappa κ = 0.90) and for clinically significant prostate cancer (csPCa) (GG 2-5 vs. GG 0 vs GG1) it was good (Model-based kappa 0.70, Fleiss’ kappa κ 0.67).ConclusionsInter-reader agreement in grading of MRI-targeted biopsy was good to excellent, while it was fair to moderate for MRI in the same cohort, as previously shown. Importantly, there was wide consensus by pathologists in assigning the contemporary GG on MRI-targeted biopsy suggesting high reproducibility of pathology reporting in the ProScreen trial.Patient summaryIt is currently unknown to what extent pathologists differ in their evaluation of histopathology in MRI-targeted prostate biopsies. We show that the agreement is good to excellent. We expect individual pathologist to have a minimal impact on MRI-based prostate cancer screening including the ProScreen trial.