Association Between Prostate Imaging Reporting and Data System (PI-RADS) Score for the Index Lesion and Multifocal, Clinically Significant Prostate Cancer

Abstract

BackgroundThe ability to identify clinically significant prostate cancer (csPCa) has dramatically improved with the introduction of multiparametric magnetic resonance imaging (mpMRI). Given the growing interest in targeted biopsy and focal therapy, improving our knowledge on the relationship between mpMRI parameters and the ability to predict csPCa multifocality is mandatory. ObjectiveTo assess whether the Prostate Imaging Reporting and Data System (PI-RADS) score for the index lesion (IL) may predict multifocal csPCa undetected by mpMRI. Design, setting, and participantsThe study included 343 patients who underwent mpMRI of the prostate with subsequent biopsy between 2014 and 2017 at a single tertiary care referral centre. InterventionLesions with a PI-RADS v.2 score ≥2 detected at mpMRI (IL) were targeted with a fusion biopsy (Bx) approach (mpMRI-Bx). Moreover, each patient underwent a random extended transrectal ultrasound-guided biopsy (TRUS-Bx) during the same session. Outcome measurements and statistical analysiscsPCa outside the IL was defined as disease detected at TRUS-Bx with a Gleason score (GS)≥3+4 and equal to or greater than the GS for the IL. The extent of csPCa detected in target and random cores was reported and stratified according to the GS and PI-RADS score for the IL. The probability of diagnosing csPCa outside the IL according to the PI-RADS score was also assessed in multivariable logistic regression analyses (MVA) after accounting for confounders. Results and limitationsThe detection rate for csPCa outside the IL was 30%. The detection rate for csPCa at TRUS-Bx was 8% for PI-RADS 2, 15% for PI-RADS 3, 36% for PI-RADS 4, and 58% for PI-RADS 5 lesions (p=0.03). Overall, the median length of csPCa found at TRUS-Bx and thus missed at mpMRI was 2.6mm. However, the length significantly increased with PI-RADS score for the IL, and was 1.8, 2.3, 2.8, and 3.8mm for PI-RADS 2, 3, 4, and 5 lesions, respectively (p=0.03). On MVA, PI-RADS 4 (odds ratio [OR] 7.6; p=0.008) and PI-RADS 5 scores (OR 17.3; p<0.001) were independent predictors of the presence of csPCa outside the IL. The study is limited by its retrospective design. ConclusionsOverall, the accuracy of mpMRI in identifying multifocal csPCa is poor, missing low-volume csPCa in approximately 30% of patients. Moreover, the rate and the extent of csPCa undetected by mpMRI significantly increased with the PI-RADS score for the IL, which can thus be considered a proxy for tumour multifocality. Patient summaryThe accuracy of multiparametric magnetic resonance imaging in identifying prostate cancer multifocality is poor. False negative findings were highly related to the PI-RADS score of the index lesion. These findings raise concerns about the indication for targeting the index lesion only when considering prostate biopsy and focal approaches.