A New Risk-Stratified Pathway Based on Internally Validated Pre- and Post- MRI Prediction Models for Clinically Significant Prostate Cancer

Abstract Purpose To develop and internally validate two pre- and post-MRI prediction models (COR-PM1 and COR-PM2) and evaluate a sequential risk-stratified pathway (COR-RSP) for reducing MRI examinations and prostate biopsies during the early detection of clinically significant prostate cancer (csPCa). Methods COR-PM1 and COR-PM2 were developed using logistic regression in 943 men who underwent MRI and biopsy (2016–2023). COR-PM1 incorporated age, serum PSA, DRE findings, prior negative biopsy, and qualitative DRE-derived prostate volume (PV). COR-PM2 incorporated PI-RADS score, with MRI-PV instead of DRE-PV. Calibration, discrimination, net benefit, and clinical performance were assessed after correction by optimism. COR-RSP sequentially applied both models at thresholds corresponding to 95% sensitivity. Results CsPCa was detected in 501 men (53.1%). Optimism-corrected AUC were 0.765 (95% CI 0.737–0.796) for COR-PM and 0.819 (95% CI 0.793–0.845) for COR-PM 2 (p = 0.001). Both models showed good calibration and net benefit. At 95% sensitivity thresholds, COR-PM1 could have avoided 139 MRI examinations (14.7%), whereas COR-PM2 could have avoided 180 biopsies (19.1%), particularly among men with PI-RADS ≤ 3. COR-RSP could have avoided 139 MRI examinations (14.7%), and 240 biopsies (25.5%), while missing 40 csPCa missed (8.0%). Biopsy yield could have increased from 53.1% to 65.6%. Conclusions COR-PM1 and COR-PM2 may help predict csPCa risk before and after MRI. COR-RSP may reduce MRI use and biopsies while increasing biopsy yield, although some csPCa cases may be missed. External validation is needed to confirm its clinical utility.