Accuracy, intra-, and inter-radiologist variability of PI-RADS v2.1 scoring for clinically significant prostate...

The Prostate Imaging Reporting and Data System version 2.1 (PI-RADS v2.1) is widely used in prostate assessment, but the intra- and inter-radiologist agreement of the PI-RADS v2.1 sequences score and categories identified by radiologists from different hospitals remain uncertain. This study aimed to explore the accuracy, intra-, and inter-observer consistency of the PI-RADS in identifying clinically significant prostate cancer (PCa) versus non-significant lesions.

A total of 164 patients who underwent prostate magnetic resonance imaging (MRI) and were subsequently confirmed by biopsy pathology were retrospectively analyzed. PCa patients with a Gleason score ≥7 were classified into the clinically significant group, whereas prostatic hyperplasia and PCa with a Gleason score <7 were classified into the non-significant group. The lesions were independently assessed by 6 radiologists from 5 institutions via PI-RADS. The diffusion-weighted imaging (DWI) score, T2-weighted imaging (T2WI) score, and PI-RADS category were recorded. The radiologists re-scored patients after one month. The performance of the PI-RADS category in distinguishing significant between non-significant lesions was evaluated via receiver operating curve (ROC) curve analysis. The kappa coefficient and Kendall's W coefficient (KW value) were used to assess the intra- and inter-observer agreement.

A total of 65 patients with clinically significant lesions and 99 patients with non-significant lesions were included. The area under the curve (AUC) values for the 6 radiologists in identifying clinically significant PCa using PI-RADS category were 0.915, 0.881, 0.863, 0.849, 0.833, and 0.884, respectively. The percentage of non-significant lesions ranged from 61.3% to 91.9% and 18.2% to 45.6% among the PI-RADS 3 and 4 lesions, respectively. The kappa coefficients for the T2WI, DWI score, and PI-RADS category spanned from 0.408 to 0.763. The KW value of the T2WI score in the senior radiologist group was 0.279, whereas the KW values for the T2WI, DWI, and PI-RADS scores among all the readers were 0.199, 0.106, and 0.127, respectively.

Radiologists can effectively identify clinically significant PCa via PI-RADS v2.1. There is a high proportion of non-significant lesions among PI-RADS 3-4. The intra-reproducibility of the scores is moderate to substantial. However, the inter-radiologist consistency is poor except for DWI scores among senior radiologists.