MP75-20: The rate of PIRADS 3 is not a proxy of the quality of MRI of the prostate. Results of a single high volume radiologist series

Armando Stabile, Nicola Fossati, Giorgio Gandaglia, Vito Cucchiara, Simone Scuderi, Francesco Barletta, Carlo Bravi, Shahrokh Shariat, Elio Mazzone, Daniele Robesti, Donato Cannoletta, Aldo Rizzo, Giuseppe Rosiello, Antony Pellegrino, Andrea Gallina, Matteo Menean, Vincenzo Mirone, Giorgio Brembilla, Antonio Esposito, Francesco Montorsi, Francesco De Cobelli, Alberto Briganti

Introduction: The rate and trend over time of PIRADS 3 at MRI has been suggested to represent a proxy of radiologist experience and quality. Our hypothesis is that radiologist experience is not necessarily associated with a decreasing rate of PIRADS 3 over time. Rather, detection of clinically significant prostate cancer (csPCa) should be correlated with radiologist expertise. We tested our hypothesis in a large series of MRI assessed by a single, high volume reader

Methods: We selected 402 men who underwent a systematic plus targeted biopsy (TBx) for a positive MRI (PIRADS=3) reported by a single radiologist at a tertiary referral centre between 2013 and 2019.  The study outcome was to assess the trend of PIRADS score (3 vs 4/5) over time and its association with the trend of csPCa (Gleason score=7) diagnosis at TBx. Multivariable logistic regression analysis (MVA) predicting the presence of PIRADS 3 and 4/5 respectively were performed using age, PSA density, biopsy history and year of biopsy (yob) as covariates. Non-parametric loess function was used to explore the relationship between yob and rate of PI-RADS 3 and 4/5. The same analysis was used to assess the relationship between yob and csPCa detection

Results: Overall 39% and 61% of men had a PIRADS 3 and 4/5, respectively. The detection of csPCa at TBx in PI-RADS 3 and 4/5 was 18% and 61%. At MVA, yob (OR: 1.22), PSA density (OR: 0.04) and age (OR: 0.9) were independent predictors of PIRADS 3 score at MRI (all p<0.04). Moreover, yob (OR: 0.82), PSA density (OR: 15) and age (OR: 1.07) were independent predictors of the presence of PIRADS 4/5 lesions (all p<0.04). Fig 1 shows as the multivariable probability of presence of PIRADS 3 increased from roughly 30 to 50% from 2013 to 2019. However, despite an increase in PIRADS 3 and a decrease in PIRADS 4-5 over time, the csPCa detection rate increased from 38% to 55% from 2013 to 2019. This was confirmed at MVA, where yob (OR: 1.13) was positively associated with csPCa detection at TBx (p=0.02) after accounting for confounders

Conclusions: The rate of PIRADS 3 is not necessarily a proxy of radiologist expertise since it increased over time even in the hands of a single, highly experienced reader. Rather, the detection of csPCa over time should be considered as the main endpoint and a more accurate proxy of radiologist and urologist expertise Source of

Funding: none