Abstract
To develop and validate biopsy-free nomograms to more accurately predict clinically significant prostate cancer (csPCa) in biopsy-naïve men with Prostate Imaging Reporting and Data System (PI-RADS) ≥ 4 lesions. A cohort of 931 patients with PI-RADS ≥ 4 lesions, undergoing prostate biopsies or radical prostatectomy from January 2020 to August 2023, was analyzed. Various clinical variables, including age, prostate-specific antigen (PSA) levels, prostate volume (PV), PSA density (PSAD), prostate health index (PHI), and maximum standardized uptake values (SUVmax) from PSMA PET-CT imaging, were assessed for predicting csPCa. Model performance was evaluated using area under the receiver operating characteristic curve (AUC), calibration plots, and decision-curve analyses, with internal validation. The foundational model (nomogram 1) encompassed the entire cohort, accurately predicting csPCa by incorporating variables such as age, PSAD, PV, PSA ratio variations, suspicious lesion location, and history of acute urinary retention (AUR). The AUC for csPCa prediction achieved by the foundational model was 0.918, with internal validation confirming reliability (AUC: 0.908). Advanced models (nomogram 2 and 3), incorporating PHI and PHI + PSMA SUVmax, achieved AUCs of 0.908 and 0.955 in the training set and 0.847 and 0.949 in the validation set, respectively. Decision analysis indicated enhanced biopsy outcome predictions with the advanced models. Nomogram 3 could potentially reduce biopsies by 92.41%, while missing only 1.53% of csPCa cases. In conclusion, the newly biopsy-free approaches for patients with PI-RADS ≥ 4 lesions represent a significant advancement in csPCa diagnosis in this high-risk population.