In-parallel comparative evaluation between multiparametric magnetic resonance imaging, prostate cancer antigen 3 and the prostate health index in predicting pathologically confirmed significant prostate cancer in men eligible for active surveillance

Objective To assess the performance capabilities of multiparametric magnetic resonance imaging (mpMRI), the prostate health index (PHI) and prostate cancer antigen 3 (PCA3) in predicting the presence of pathologically confirmed significant prostate cancer (PCSPCa), according to the European Randomized Study of Screening Prostate Cancer definition, in a single cohort of patients who underwent radical prostatectomy (RP) but who were eligible for active surveillance (AS). Materials and Methods An observational retrospective study was performed in 120 patients with prostate cancer (PCa), treated with robotassisted RP but eligible for AS according to Prostate Cancer Research International: Active Surveillance criteria. Blood and urine specimens were collected before initial prostate biopsy for PHI and PCA3 measurements, respectively. In addition, all patients underwent mpMRI, preoperatively and 6–8 weeks after biopsy, with a 1.5T scanner using a four-to-five-channel phase array coil combined with an endorectal coin. mpMRI images were assessed and diagrams showing the prostate sextants were used to designate regions of abnormality within the prostate. Prostate findings were assigned to one of five categories according to Prostate Imaging-Reporting and Data System guidelines (PI-RADS) and considered positive for PCa if final PI-RADS score was >3 and negative if ≤3. Results Pathologically confirmed reclassification was observed in 55 patients (45.8%). mpMRI showed good specificity and negative predictive value (0.61 and 0.73, respectively) for excluding PCSPCa compared with the PHI and PCA3. On multivariate analyses and after 1 000 bootstrapping resampling, the inclusion of both mpMRI and the PHI significantly increased the accuracy of the base model in predicting PCSPCa. For the prediction of PCSPCa, in particular, the base model had an area under the curve (AUC) of 0.71 which significantly increased by 4% with the addition of the PHI (AUC = 0.75; P < 0.01) and by 7% with the addition of mpMRI (AUC = 0.78; P < 0.01). Decisioncurve analysis showed that the multivariable model with mpMRI had the highest net benefit. Conclusion In a single cohort of patients who underwent RP but who were eligible for AS, mpMRI and, to a lesser extent, the PHI, had an important role in discriminating the presence of PCSPCa; both measures could therefore be useful in the selection and monitoring of patients undergoing AS.