Necessity of differentiating small (< 10 mm) and large (≥ 10 mm) PI-RADS 4

Purpose: Prostate Imaging Reporting and Data System version 2 (PI-RADSv2) provides reasonable performance in detecting significant cancers. Still, it is unclear about whether all PI-RADS 4 lesions show the same cancer detection rate (CDR) regardless of tumor size. The aim was to compare the CDRs of small (< 10 mm) and large (≥ 10 mm) PI-RADS 4. Methods: After magnetic resonance imaging (MRI) was performed in 684 men, a radiologist interpreted the MR images and detected 281 index lesions categorized as PI-RADS 4 in 281 men. PI-RADS 4 lesions were divided into small and large groups on size of 10 mm. Overall and significant CDRs were compared between the groups. A significant cancer was defined as one with Gleason score (GS) ≥ 7 or tumor volume ≥ 0.5 ml. Tumor volumes were roughly calculated as πr³4/3 (π = 3.14 and r = a half of tumor size) and were compared between the groups. Standard reference was a biopsy examination. Fisher’s exact and Mann–Whitney tests were used for statistical analysis. Results: The overall CDRs of small and large groups were 39.0% (53/136) and 59.3% (86/145), respectively, (p = 0.0008). The median tumor volumes of cancer-proven small and large groups were 0.18 ml (0.01–0.38 ml) and 0.70 ml (0.52–1.44 ml), respectively (p < 0.0001). Using GS or tumor volume, the significant CDRs of these groups were 26.5% (36/136) and 59.3% (86/145), respectively (p < 0.0001), and using GS alone, 26.5% (36/136) and 39.3% (57/145), respectively (p = 0.0232). Conclusions: PI-RADS 4 lesions should be sub-divided on size of 10 mm because of different significant CDRs.