p2PSA-associated parameters are significant predictors of reclassification of first-year protocol biopsy on active surveillance for early-stage prostate cancer patients: From the PRIAS-JAPAN study

153 Background: 44 Japanese institutions have joined the Prostate Cancer Research International: Active Surveillance (PRIAS) study as PRIAS-JAPAN. As an side study, PRIAS-JAPAN investigates the usefulness of [-2] proPSA (p2PSA), a precursor of PSA, as a predictive marker for the reclassification during active surveillance patients. In this study, we aimed to evaluate the usefulness of p2PSA related parameters to predict the reclassification of first year protocol biopsy (1yr PBx). Methods: 19 institutions participated in this side study which started in January 2013. All participants were eligible for PRAIS inclusion criteria. These patients were required to receive blood sampling on every visit and protocol biopsy after inclusion. PSA and PSA isoforms (free-PSA, p2PSA) were measured in serum samples and parameters (%free-PSA, %p2PSA, phi) were also calculated. Mann-Whitney U test was used to compare continuous variables between reclassification and non-reclassification groups. Fisher's exact test and Chi-squared test with yates' continuity correction were used for categorical variables. Multivariable logistic regression models were used to predict the risk of reclassification. To assess the predictive power for reclassification, we plotted ROC curves and calculated the AUC. Decision curve analysis (DCA) was used to evaluate whether adding variables to the baseline (age, PSA, prostate volume) model would yield net clinical benefit. Results: From January 2013 to July 2017, this study enrolled 237patients. 99 patients were excluded for the reason of discontinuation of AS, non-adherence of protocol blood sampling and rejection of biopsy. 138patients were able to analyzed PSA, PSA isoforms and its parameters at the time of inclusion and 1yr PBx. Patients who showed reclassification were significantly higher level of %p2PSA at inclusion and %p2PSA and phi just before 1yr PBx. Multivariate analyses showed %p2PSA at inclusion (p= 0.04), %p2PSA (p= 0.01) and Phi (p= 0.008) before 1yr PBx were statistically significant predictors of reclassification at 1yr PBx. ROC analysis revealed that %p2PSA before 1yr PBx (AUC 0.674) and phi before 1yr PBx (AUC 0.669) were a significant discriminant factor in predicting reclassification. DCA analysis showed that phi before 1yr PBx had a highest net benefit in the range of clinically plausible risk threshold (10–40%). Conclusions: %p2PSA and Phi before 1yr PBx have good prediction power and phi is a most useful indicator of clinical decision making about active surveillance. Clinical trial information: UMIN 000009876.