Development and External Validation of a Prediction Model to Identify Candidates for Prostate Biopsy

Area of Science:

• Urology
• Oncology
• Medical Informatics

Background:

• Prostate biopsies carry infectious risks and often yield benign or insignificant findings.
• A significant proportion of prostate biopsies detect clinically insignificant cancer, leading to overtreatment.

Purpose of the Study:

• To develop and validate a predictive model for clinically significant prostate cancer (csPCa).
• To reduce the rate of unnecessary prostate biopsies through accurate csPCa prediction.

Main Methods:

• Retrospective analysis of 1632 men from a single center and validation in an external cohort of 622 men.
• Development of a nomogram using multivariable logistic regression based on predictors like PSA density and PI-RADS scores.
• Validation using ROC curves, DCA, and comparison of predicted vs. actual csPCa rates.

Main Results:

• The prediction model achieved an ROC of 0.88 in the external validation cohort.
• PSA density, prior negative biopsy, and PI-RADS scores were significant predictors of csPCa.
• The model could avoid 35% of biopsies and 46% of indolent prostate cancer diagnoses while missing only 5% of csPCa.

Conclusions:

• The developed prediction model effectively reduces unnecessary prostate biopsies.
• The model demonstrates minimal impact on the detection rates of clinically significant prostate cancer.