A Comparison of Magnetic Resonance Imaging Assessment and Biopsy Outcomes with and Without Central Review in Two Swedish Regional Organized Prostate Cancer Testing Programs

Abstract

Background and objective: The European Council advises evaluating the feasibility of organized prostate cancer testing (OPT) programs, but it is unclear whether results from screening trials can be replicated in population-based testing. The aim of this study is to compare magnetic resonance imaging (MRI) assessments and biopsy outcomes with and without a central review in two Swedish OPT programs.

Methods: Two regional population-based OPT programs invited 65 000 men (2020-2022). MRI scans were read locally, and biopsies followed a strict MRI-based and prostate-specific antigen (PSA) density-based protocol. A blinded central review was done by two radiologists with 8 and 9 yr of experience. Reader agreement was assessed with percentages and kappa scores. Positive predictive values (PPVs) for detecting grade group (GG) 2-5 prostate cancer were calculated with 95% confidence intervals (CIs).

Key findings and limitations: MRI scans for 416 men (median age 52 yr) with PSA ≥3 ng/ml were evaluated. In Skåne, 27% of scans were primarily assigned Prostate Imaging Reporting and Data System (PI-RADS) scores ≥4, compared with 10% in Västra Götaland. At the primary reading, 76 men had PI-RADS ≥4, yielding 43 GG 2-5 prostate cancer cases: PPV 0.57 (95% CI 0.45-0.67). At the central review, 65 men had PI-RADS ≥4. Out of 61 men biopsied, 50 had GG 2-5 prostate cancer: PPV 0.82 (95% CI 0.71-0.90, p < 0.001 for PPV difference). The central review radiologists' kappa score was 0.83. No additional biopsies were taken based on the central review findings.

Conclusions and clinical implications: In population-based screening with local MRI reading, MRI assignment may vary substantially. Centralized reading could reduce these differences and increase the biopsy PPV for GG ≥2 cancer.

Patient summary: In this report, we reviewed local magnetic resonance imaging (MRI) reading in population-based screening. We found that MRI assignment varied between centers. We conclude that centralized reading could reduce differences and improve biopsy outcomes.

Keywords: Early detection of cancer; Magnetic resonance imaging; Observer variation; Prostatic neoplasms.

Fig. 1

Siemens Healthcare AG (Erlangen, Baviaria, Germany) VIDA OPT prostate MRI (3 T) with (A) bilateral T2W hypointensities and (B and C) mildly restricted diffusion with ADC values >1200 × 10^–6 mm²/s. The primary reader assigned bilateral PI-RADS 4 lesions; both central review readers assigned PI-RADS 2. Two rounds of bilateral targeted biopsies (six cognitive biopsies in the first round and eight TRUS/MRI fusion biopsies in the second round) showed benign prostate tissue only. A repeat MRI scan 2 yr later in the OPT program was assessed to be PI-RADS 2. ADC = apparent diffusion coefficient; MRI = magnetic resonance imaging; OPT = prostate cancer testing; PI-RADS = Prostate Imaging Reporting and Data System; TRUS = transrectal ultrasound; T2W = T2 weighted.