Evaluation of a multivariate prostate-specific antigen and percentage of free prostate-specific antigen logistic regression model in the diagnosis ofprostate cancer

The use of prostate-specific antigen (PSA) in the diagnosis of prostate can cer is controversial due to false-positive results caused by benign prostat ic hyperplasia. Several groups have suggested the usefulness of the percent age of free PSA (%fPSA) in patients with PSA levels between 4 and 10 mu g/l . Based on previously obtained results, biopsy is carried out in our hospit al if the PSA is greater than 10 mu g/l or if the %fPSA is lower than 20% a nd PSA is between 4-10 mu g/l. In this study, we have compared these result s with those obtained with a logistic regression model based on the determi nation of PSA and %fPSA. The diagnostic efficacy of the logistic regression model is greater than that of the currently used model. The posterior cons truction of a nomogram based on the data obtained greatly facilitates the a pplication of the logistic regression model. Copyright (C) 1999 S.Karger AG , Basel.