Abstract
Objective: To determine a predictive model for the primary diagnosis of prostate cancer (PC) based on a multiple serum biomarker assay. Material and Methods: Between August 2011 and February 2013, a total of 387 prostate biopsies were performed. Serum or plasma concentrations of 22 biomarkers (neopterin, IGF-1, IGFBP-2, IGFBP-3, sarcosine, endoglin, TGF-β1, periostin, sPLA2-IIa, chromogranin A, ZAG2, clusterin, PSP94, PSP94bp, leptin, cathepsin D, hepsin, KLK11, PSMA, AMACR, CRISP3 and A1AT) were determined. Biomarker levels were correlated with the prostate biopsy results. Several statistical models for PC detection were created. Results: A total of 167 of the 373 evaluated patients (44.8%) were diagnosed with PC. None of the tested biomarkers reached statistical significance using the univariate analysis. However, the level of serum clusterin was not associated with any other tested parameter. Several basic models showed a higher positive predictive value than individual parameters. Addition of serum clusterin to the base model with prostate-specific antigen, digital rectal exam and prostate size significantly improved the area under curve value (0.723 vs. 0.716). Conclusion: Our findings suggested that multiple serum assays based on some promising markers may only have a limited practical benefit for the prediction of PC in the prostate biopsy.