Background: Artificial intelligence ultrasound of prostate (AIUSP)-targeted biopsy has been used for prostate cancer (PCa) diagnosis. The objective of this prospective multi-center head-to-head clinical randomized comparative trail (RCT) is to compare PCa detection rate in the TRUS-guided 12-core standard systematic biopsy (TRUS-SB) group and cognitive fused mpMRI-guided 12-core biopsy (mpMRI) group against AIUSP group.
Methods: Four hundred patients were randomized to three arms and underwent biopsies by TRUS-SB (n = 133), mpMRI (n = 134), and AIUSP (n = 133) between January 2015 and December 2017. In TRUS-SB group, a standard 12-core systematic biopsy was performed. In mpMRI group, mpMRI-suspicious lesions (PI-RADS 3-5) were targeted by 2-core biopsy followed by a 10-core systematic biopsy. Otherwise, 12-core systematic biopsy was performed. In AIUSP group, a 6-core targeted biopsy was performed. The primary endpoint was PCa detection rate.
Results: AIUSP detected the highest rate of PCa (66/133, 49.6%) compared to TRUS-SB (46/133, 34.6%, p = 0.036) and mpMRI (48/134, 35.8%, p = 0.052). Compared to TRUS-SB (35/133, 26.3%) and mpMRI (31/134, 23.1%) groups, clinically significant PCa (csPCa) detection rate was 32.3% (43/133) in AIUSP group. Overall biopsy core positive rate in the TRUS-SB group (11.0%, 176/1598) and in the mpMRI group (12.7%, 204/1608) was significantly lower than that in the AIUSP group (22.7%, 181/798, p < 0.001).
Conclusions: AIUSP detected the highest rate of overall and significant PCa compared to TRUS-SB and mpMRI, and could be used as an alternative to systematic biopsy in the future.
Registration: This trial was registered in ISRCTN (ISRCTN18033113).
Keywords: Artificial intelligence; Prostate cancer; Systematic biopsy; Targeted biopsy; Ultrasound; mpMRI.
© 2022. The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature.