Development of a Prognostic Prediction Model for Clinically Significant Prostate Cancer Based on Lesion Zone and...

Area of Science:

• Radiology and Imaging Science
• Oncology
• Biostatistics

Background:

• Accurate prediction of clinically significant prostate cancer (csPCa) is crucial for biopsy decisions.
• Multiparametric magnetic resonance imaging (mpMRI) with PI-RADS scoring aids in detecting prostate cancer.
• The predictive value of PI-RADS v2.1 can be enhanced by incorporating quantitative imaging features and lesion characteristics.

Purpose of the Study:

• To develop a predictive model for csPCa in men with PI-RADS 3-5 scores.
• To assess the impact of prostate lesion zone on predictive accuracy.
• To establish an apparent diffusion coefficient (ADC) threshold for csPCa and create a logistic regression model incorporating zone and ADC.

Main Methods:

• Retrospective analysis of 546 lesions from 429 men undergoing prostate mpMRI.
• Lesion-level analysis by fellowship-trained genitourinary radiologists and pathologists.
• Development of a logistic regression model using clinical data, lesion zone, and quantified ADC values.

Main Results:

• Prostate zone (PZ vs. TZ) was a significant predictor of csPCa (OR 2.58, p<0.0005).
• A combined model including zone and prostate-specific antigen density achieved an AUC of 0.78 at an ADC cutoff of 947 µm²/s.
• Simulated scenarios showed probabilities ranging from 16% to 82%, generally above biopsy omission thresholds.

Conclusions:

• Prostate zone and quantified ADC values independently enhance csPCa predictability.
• The developed model shows promise but did not yield biopsy-omission thresholds in simulated cases with MRI-visible lesions.