External Validation of MRI-Based Prostate Cancer Risk Calculators in an Australian Cohort

ANZ J Surg. 2025 Jun 27. doi: 10.1111/ans.70211. Online ahead of print.

Abstract

Objective: To validate magnetic resonance imaging (MRI)-based prostate cancer risk calculators in an Australian population.

Methods: Data was retrospectively collected from patients biopsied at SALHN from May 2018 to March 2023. Men with MRI-suspicious lesions underwent biopsy. Performance of nine risk calculators was evaluated by discrimination, decision curve analysis (DCA), calibration, and recalibration. Subgroup analysis of biopsy-naïve patients was conducted.

Results: Fifty-two percent had clinically significant cancer (International Society of Urological Pathology [ISUP] Grade Group ≥ 2). Median age of 67 years (IQR 62-67), PSA 7.3 ng/mL (IQR 5.3-11) and prostate volume 45 mL (IQR 32 mL-67 mL) were observed. Abnormal DRE was recorded in 47%, and in 65% a Prostate Imaging-Reporting and Data System (PIRADS) score of > 4 was observed. The RPCRC and Mehralivand risk calculators demonstrated the best performance, with the Radtke risk calculator demonstrating adequate performance in the biopsy naïve subgroup analysis. DCA demonstrated a net benefit of up to 12% of biopsies at a 15% risk threshold prior to recalibration. With recalibration, a net benefit of up to 6% at a 15% risk threshold was demonstrated. Analyses demonstrated a net benefit of up to 12.5% both pre and post recalibration in biopsy naïve patients. Discriminative ability was at least > 0.79, with most demonstrating discriminative ability > 0.82.

Conclusion: Use of prostate biopsy decision-making tools provides avenues for individualized health care based on patient-specific risk factors. This study has successfully completed external validation of nine risk calculators, with one demonstrating adequate discriminatory capacity in all patients and three in biopsy-naïve patients. Systemic use may have a 12% net clinical benefit amongst biopsy-naïve patients, thereby reducing healthcare system burden.

Keywords: biopsy; diagnosis; prediction models; prostate cancer; triage.