Quantitative multi-gene expression profiling of primary prostate cancer

BACKGROUND. This study describes the evaluation of the expression patterns of prostate-related transcripts in 106 matched prostate tissues from prostatectomies as predictors for prostate cancer (PCa). METHODS. Quantitative PCR (QPCR) Assays with site-specific hybridization probes were established for four housekeeping genes (GAPDH, HPRT, PBGD, TBP) and nine prostate-related genes (AibZIP, D-GPCR, EZH2, PCA3, PDEF, prostein, PSA, PSCA, TRPM8). RESULTS. The relative mRNA expression levels of AibZIP, D-GPCR, EZH2, PCA3, PDEF, PSA, TRPM8 (all P < 0.001) and prostein (P = 0.019) normalized to the TBP reference gene were significantly higher in malignant compared to non-malignant prostate tissues. Employing receiver-operating characteristic (ROC) analyses, PCA3 was the best single tumor marker with the highest area-under-the-curve (AUC = 0.85). A multivariate logit model for the predictability of the tumor was developed, which employed the relative expression levels of EZH2, PCA3, prostein, and TRPM8 and yielded an AUC of 0.90. CONCLUSIONS. The transcript marker PCA3 is a powerful predictor of primary PCa but the inclusion of EZH2, prostein, and TRPM8 adds even more to the diagnostic power. The finding of a significantly higher mRNA expression of three different genes (prostein, PSA, TRPM8) in organ-confined tumors compared to non-organ-confined tumors as well as the multi-marker PCa prediction model developed in the retrospective model system on prostatectomies could be of clinical importance for diagnostic purposes, and should be verified in diagnostic biopsies.