Multiparametric Magnetic Resonance Imaging Second Opinion May Reduce the Number of Unnecessary Prostate Biopsies: Time to Improve Radiologists' Training Program?

Abstract

Purpose: To understand the multiparametric magnetic resonance imaging (mpMRI) interreader agreement between radiologists of peripheral and academic centers and the possibility to avoid prostate biopsies according to magnetic resonance imaging second opinion.

Patients and methods: This prospective observational study enrolled 266 patients submitted to mpMRI at nonacademic centers for cancer detection or at active surveillance begin. Images obtained were reviewed by 2 unblinded radiologists with 8 and 5 years' experience on mpMRI, respectively. We recorded Prostate Imaging Reporting and Data System (PI-RADS) v2 categories and management strategy changes after mpMRI rereadings. Interreader agreement was assessed by the Cohen kappa. For mpMRI second opinion, positive predictive value and negative predictive value were calculated.

Results: In the original readings, no lesions (ie, PI-RADS < 2) were observed in 17 cases (6.5%). Reported index lesion (IL) PI-RADS category was 2 in 23 (8.5%), 3 in 85 (32%), 4 in 98 (37%), and 5 in 13 (5%) men, respectively. It is noteworthy that in 30 examinations (11%), an IL was recognized by radiologists, but a suspicious score was not assigned. According to first reading of mpMRI, initial clinical strategy included performing a targeted (226; 85%) or a systematic biopsy (8; 3%), scheduling the patient to an active surveillance program without repeat biopsy (10; 4%), or monitoring prostate-specific antigen without prostate sampling (22; 8%). The mpMRI rereads did not change IL PI-RADS category in 91 cases (38.5%), although in 20 (8.5%) and 125 (53%) IL PI-RADS was upgraded or downgraded, respectively (κ = 0.23). The clinical management changed in 113 patients (48%) (κ = 0.2). Overall, 102 targeted biopsies (51%) were avoided and 72 men (34.5%) were not submitted to biopsy after mpMRI second opinion. Positive predictive value and negative predictive value of the mpMRI rereading were 58% and 91%, respectively. Major limitations of the study are limited-time follow-up and the lack of a standard of reference for some men, who were not submitted to biopsy according to mpMRI second opinion.

Conclusion: There is an important level of discordance between mpMRI reports. According to imaging second opinion, roughly half of targeted biopsies could be avoidable and 34.5% of men could skipped prostate sampling. Prospective randomized trials are needed to confirm our findings.

Keywords: Expert radiologists; Interobserver agreement; Prostate cancer.