Prostate health index (phi), a measure calculated as p2PSA/fPSA × √tPSA, has shown valuable results in the detection of prostate cancer (PCa), improving the prediction of the aggressiveness of the tumor. The aim of our study was to test whether prostate volume influences phi performance using univariate and multivariate models. 220 patients with PSA<10 μg/L (102 with negative biopsy and 118 with PCa) were included in the study. Serum concentrations of tPSA, fPSA and p2PSA were measured on Access2 analyzer. The higher accuracy was found for phi, obtaining an AUC of 0.748. Bigger AUCs were obtained for phi, %p2PSA, %fPSA and tPSA in patients with small prostate volume (≤35 cc); meanwhile, the lowest AUCs were found in patients with large prostate volume (>50 cc). Including phi and %p2PSA in a multivariable analysis based on patient age, prostate volume, tPSA, and %fPSA accuracy increased from 0.762 to 0.802 (logistic regression model) or 0.815 (artificial neural network). Accuracy excluding prostate volume in these models was 0.762 and 0.775, respectively. The inclusion of phi and %p2PSA in a multivariate model identifies better men with PCa. Prostate volume remains a key factor in the interpretation of biomarkers used to detect PCa.
Keywords: PSA; Prostate cancer; Prostate health index; Prostate volume; p2PSA.
Copyright © 2014 Elsevier B.V. All rights reserved.