Diffusion levels for quantitative assessment of the apparent diffusion coefficient value in prostate MRI: a proof-of-concept bicentric study.

Abstract

Objectives: To investigate the performance of Diffusion levels (DLs) in diagnosing clinically significant prostate cancer (csPCa) when combined with the PI-RADS version 2.1.

Materials and methods: This retrospective, bicentric study included 261 men who underwent 3.0-T prostate MRI between March 2020 and April 2023, receiving systematic and target prostate biopsy on PI-RADS ≥ 3 lesions. Two readers measured the Apparent diffusion coefficient (ADC) of PI-RADS 1-5 findings in the peripheral zone. By plotting the cumulative frequency of csPCa versus ADCs and using ROC analysis, we derived four DLs expressing levels of restricted diffusion, i.e., very low DL (VL-DL), low DL (L-DL), intermediate DL (I-DL), and high DL (H-DL). We compared the per-lesion diagnostic performance in assessing csPCa (grading group ≥ 2 cancer) assuming to biopsy PI-RADS ≥ 3 lesions (strategy 1), PI-RADS ≥ 3 lesions adjusted with ADC values (strategy 2-4), and PI-RADS ≥ 3 lesions adjusted with DLs (strategy 5-7). Net benefit was assessed with decision curve analysis.

Results: csPCa was found in 79/261 men (30.3%) and 152/528 lesions (28.8%). There was a negative correlation (p < 0.0001) between ADC versus malignancy rate (tau -0.970) and DLs versus csPCa grading group (tau -0.614). csPCa prevalence was highest in VL-DL (72.2%) and L-DL (54.4%). Most DLs-based strategies increased specificity, positive predictive value (PPV), and net benefit compared to ADC-based strategies or PI-RADS alone. The best strategy showed 94.7% sensitivity, 82.9% specificity, 69.2% PPV, and 97.5% negative predictive value.

Conclusion: While larger-scale validation is needed, DLs have the potential to improve PI-RADS-based biopsy decisions for detecting csPCa in the peripheral zone.

Key points: Question It is still unclear how to incorporate quantitative information from diffusion-weighted imaging (DWI) into prostate MRI. Findings Combining DWI-derived diffusion levels (DLs) with the PI-RADS version 2.1 categorization reduced false positives while preserving high sensitivity for clinically significant prostate cancer. Clinical relevance DLs permit to easily account for ADC values of prostate lesions and, in turn, refine biopsy decisions.