The evolving PI-RADS paradigm.
Authors
Affiliations (4)
Affiliations (4)
- Department of Advanced Biomedical Sciences, University of Naples Federico II, Naples, Italy.
- Department of Radiology, The Netherlands Cancer Institute, Amsterdam, The Netherlands.
- Department of Radiology and Nuclear Medicine, Erasmus University Medical Center Rotterdam, Rotterdam, The Netherlands.
- Paul Strickland Scanner Centre, Mount Vernon Cancer Centre, Northwood, UK. [email protected].
Abstract
Prostate MRI is central to the diagnostic pathway for prostate cancer (PCa), reducing unnecessary biopsies, improving the detection of clinically significant disease (csPCa), and limiting overdiagnosis of indolent tumours. The Prostate Imaging Reporting and Data System (PIRADS) version 2.1 is the current international standard for MRI data acquisition and interpretation. This review synthesises the evidence for its use, its strengths and limitations, and future developments. PI-RADS v2.1 improved the assessments of multiparametric MRI (mpMRI) by clarifying technical requirements, zone-specific interpretation rules, and structured reporting recommendations. Its diagnostic performance is well established, particularly for ruling out csPCa and guiding risk-adapted biopsy strategies. However, significant gaps remain. MRI image quality varies substantially across centres, and PI-RADS lacks a mechanism to exclude non-diagnostic scans, resulting in inconsistent classification. Several scoring challenges persist, most notably the ambiguous definition and heterogeneous management of PI-RADS 3 lesions, imprecise lesion measurement standards, and limited guidance for central-zone and atypical lesions. Gaps include the absence of strategies for assessing background tissue changes and infiltrative patterns. Cancer detection specificity and inter-reader agreement remain moderate, with significant discrepancies in transition-zone assessments. Artificial intelligence (AI) holds promise for reducing variability, improving lesion detection, and optimising workflow, though rigorous validation and clear human-AI integration frameworks are needed. Advances in deep learning reconstructions improve image quality and enable shorter protocols, thereby supporting the adoption of biparametric MRI. A shift toward risk-based pathways, integrating MRI findings with PSA density and clinical parameters, is reflected in the forthcoming PI-RADS Pathway 2026, which aims to standardize global biopsy practice and reduce unnecessary interventions.