• Stefano Luzzago, Ottavio de Cobelli, Gabriele Cozzi, Giulia Peveri, Vincenzo Bagnardi, Michele Catellani, Ettore Di Trapani, Francesco A Mistretta, Paola Pricolo, Andrea Conti, Sarah Alessi, Giulia Marvaso, Matteo Ferro, Deliu-Victor Matei, Giuseppe Renne, Barbara Alicja Jereczek-Fossa, Giuseppe Petralia, and Gennaro Musi.
• Department of Urology, IEO European Institute of Oncology, IRCCS, Milan, Italy.
• BJU Int. 2020 Jul 1; 126 (1): 104-113.

ObjectivesTo develop a novel nomogram to identify candidates for active surveillance (AS) that combines clinical, biopsy and multiparametric magnetic resonance imaging (mpMRI) findings; and to compare its predictive accuracy to, respectively: (i) Prostate Cancer Research International: Active Surveillance (PRIAS) criteria, (ii) Johns Hopkins (JH) criteria, (iii) European Association of Urology (EAU) low-risk classification, and (iv) EAU low-risk or low-volume with International Society of Urological Pathology (ISUP) Grade Group (GG) 2 classification.Patients And MethodsWe selected 1837 patients with ISUP GG1 or GG2 prostate cancer (PCa), treated with radical prostatectomy (RP) between 2012 and 2018. The outcome of interest was the presence of unfavourable disease (i.e., clinically significant PCa [csPCa]) at RP, defined as: ISUP GG ≥ 3 and/or pathological T stage (pT) ≥3a and/or pathological N stage (pN) 1. First, logistic regression models including PRIAS, JH, EAU low-risk, and EAU low-risk or low-volume ISUP GG2 binary classifications (not eligible vs eligible) were used. Second, a multivariable logistic regression model including age, prostate-specific antigen density (PSA-D), ISUP GG, and the percentage of positive cores (Model 1) was fitted. Third, Prostate Imaging-Reporting and Data System (PI-RADS) score (Model 2), extracapsular extension (ECE) score (Model 3) and PI-RADS + ECE score (Model 4) were added to Model 1. Only variables associated with higher csPCa rates in Model 4 were retained in the final simplified Model 5. The area under the receiver operating characteristic curve (AUC), calibration plots and decision curve analyses were used.ResultsOf the 1837 patients, 775 (42.2%) had csPCa at RP. Overall, 837 (47.5%), 986 (53.7%), 348 (18.9%), and 209 (11.4%) patients were eligible for AS according to, respectively, the EAU low-risk, EAU low-risk or low-volume ISUP GG2, PRIAS, and JH criteria. The proportion of csPCa amongst the EAU low-risk, EAU low-risk or low-volume ISUP GG2, PRIAS and JH candidates was, respectively 28.5%, 29.3%, 25.6% and 17.2%. Model 4 and Model 5 (in which only PSA-D, ISUP GG, PI-RADS and ECE score were retained) had a greater AUC (0.84), compared to the four proposed AS criteria (all P < 0.001). The adoption of a 25% nomogram threshold increased the proportion of AS-eligible patients from 18.9% (PRIAS) and 11.4% (JH) to 44.4%. Moreover, the same 25% nomogram threshold resulted in significantly lower estimated risks of csPCa (11.3%), compared to PRIAS (Δ: -14.3%), JH (Δ: -5.9%), EAU low-risk (Δ: -17.2%), and EAU low-risk or low-volume ISUP GG2 classifications (Δ: -18.0%).ConclusionThe novel nomogram combining clinical, biopsy and mpMRI findings was able to increase by ~25% and 35% the absolute frequency of patients suitable for AS, compared to, respectively, the PRIAS or JH criteria. Moreover, this nomogram significantly reduced the estimated frequency of csPCa that would be recommended for AS compared to, respectively, the PRIAS, JH, EAU low-risk, and EAU low-risk or low-volume ISUP GG2 classifications.© 2020 The Authors BJU International © 2020 BJU International Published by John Wiley & Sons Ltd.

26 keywords...

hide…