RESUMO
PURPOSE: About 50% of prostate cancers have TMPRSS2-ERG fusions with concurrent ERG overexpression. The aim of this study was to determine whether clinical differences exist between ERG-positive and ERG-negative cancers in surgically treated patients not exposed to antihormonal therapy. A secondary aim was to search for differences between these tumor classes. EXPERIMENTAL DESIGN: A tissue microarray containing samples from more than 2,800 prostate cancers with clinical data was analyzed for ERG alterations by immunohistochemistry and FISH. Results were compared with tumor phenotype, biochemical recurrence, and molecular features considered important for prostate cancer. The effect of ERG on androgen receptor (AR)-dependent transcription was analyzed in cell lines. RESULTS: ERG expression was found in 52.4% of 2,805 cancers with a 95% concordance between ERG expression and ERG gene rearrangement detected by FISH. ERG expression was unrelated to clinical outcome and tumor phenotype. Differences in AMACR, Annexin A3, Bcl2, CD10, ALCAM, chromogranin A, epidermal growth factor receptor, HER2, mTOR, p53, and synaptophysin status were significant but minimal in absolute numbers. The most striking difference was found for AR expression, which was markedly higher in ERG-positive cancers. In vitro studies showed ERG-dependent impairment of AR-mediated transcriptional activity. CONCLUSIONS: The striking similarities between these two types of prostate cancers rules out a major impact of ERG on tumor aggressiveness in early, not hormonally treated cancer. The marked difference in AR levels between ERG-positive and -negative cancers supports a systematic difference in potential response to hormonal therapy as previously observed in clinical trials.