Predicting cancer detection rates from multiparametric prostate MRI Beyond the PI-RADS classification system

Area of Science:

• Urology
• Radiology
• Oncology

Background:

• Prostate Imaging-Reporting and Data System (PI-RADS) is standard for prostate cancer risk assessment via MRI.
• Real-world prostate cancer detection rates (CDRs) vary significantly despite PI-RADS.
• This study aimed to identify factors influencing CDRs and develop a predictive model.

Purpose of the Study:

• To evaluate associations between clinical/radiographic features and cancer detection rates (CDRs) in prostate MRI.
• To develop a predictive model for improved clinical management of prostate cancer.

Main Methods:

• Analysis of men (18-89 years) with elevated PSA or active surveillance undergoing MRI-US fusion or in-bore MRI-targeted biopsy.
• Logistic regression examined associations of features with per-lesion CDR (Gleason 6-10) and clinically significant (cs) CDR (Gleason 7-10).
• A predictive model was operationalized based on identified associations.

Main Results:

• 347 lesions in 281 patients underwent targeted biopsy; overall CDR was 49.0%, csCDR was 28.0%.
• Higher PI-RADS category, smaller prostate size, and higher PSA density correlated with increased CDR.
• Prior biopsy was linked to lower CDR; solitary PI-RADS 3-5 lesions and fewer prior biopsies increased csCDR.

Conclusions:

• Clinical and radiographic features independently predict prostate cancer risk in MRI-targeted biopsies.
• A developed predictive model enhances biopsy decisions over the standard PI-RADS 3-5 lesion strategy.
• The model offers improved clinical decision-making for prostate cancer diagnosis.