Restriction Spectrum Imaging as a Quantitative Biomarker for Prostate Cancer With Reliable Positive Predictive Value

Area of Science:

• Radiology and Medical Imaging
• Biophysics
• Oncology

Background:

• Prostate Imaging Reporting and Data System (PI-RADS) for clinically significant prostate cancer (csPCa) has variable predictive value.
• Restriction Spectrum Imaging restriction score (RSIrs) is a novel, biophysics-based metric from diffusion MRI.
• RSIrs offers potential as an objective biomarker for csPCa detection.

Purpose of the Study:

• Evaluate the performance of RSIrs for patient-level csPCa detection in a large, heterogeneous dataset.
• Compare RSIrs performance against apparent diffusion coefficient (ADC) and PI-RADS.
• Assess the utility of combining RSIrs with clinical and imaging parameters for enhanced csPCa detection.

Main Methods:

• Prostate MRI data from 1892 patients across 7 centers (January 2016 - March 2024).
• Calculation of patient-level csPCa probability using maximum RSIrs.
• Comparison of detection performance (AUC) of RSIrs, ADC, and PI-RADS for csPCa (GG ≥ 2).
• Multivariable regression analysis combining RSIrs with clinical factors.

Main Results:

• csPCa probability increased with higher RSIrs.
• For biopsy-naïve patients, AUCs were RSIrs=0.73, ADC=0.54, PI-RADS=0.75.
• RSIrs significantly outperformed ADC (P < .01) and was comparable to PI-RADS (P = .31).
• Combined RSIrs and PI-RADS outperformed either alone; the best model included RSIrs, PI-RADS, age, and PSA density.

Conclusions:

• RSIrs is an accurate, reliable quantitative biomarker for csPCa detection.
• RSIrs performs better than ADC and comparably to PI-RADS.
• RSIrs provides objective csPCa probability estimates, reducing reliance on radiologist expertise.