The prognostic and predictive value of TMPRSS2-ERG gene fusion and ERG protein expression in prostate cancer biopsies

The clinical course of prostate carcinoma (PCa) is very heterogeneous. Consequently, a personalised approach for risk stratification and treatment planning is important. Recently, it has become evident that PCa, also at the genomic level, is heterogeneous. An early and common alteration is the gene fusion between the transmembrane protease serine 2 (TMPRSS2) gene and the v-ets avian erythroblastosis virus E26 oncogene homolog (ERG) gene resulting in expression of the oncoprotein ERG. The gene fusion is present in approximately half of PCa patients and the resultant two subgroups demonstrate marked differences in their genomic signatures. It has been hypothesised that genomic alterations can explain some of the observed heterogeneity in the clinical course of PCa. In order to conduct an analysis of the prognostic and predictive value of ERG protein expression in PCa biopsies, the thesis sought to evaluate: 1) the concordance in ERG expression between biopsies and radical prostatectomies: 2) the association between expression of ERG protein and the risk of PCa progression during active surveillance (AS), and 3) the association between ERG protein expression and response to primary castration-based treatment for advanced PCa.

The included patients derived from the institutional AS cohort and an institutional cohort of advanced PCa patients undergoing first line castration-based androgen deprivation therapy (ADT). The 265 patients in the AS cohort were enrolled prospectively between October 2002 and October 2012 and were followed with regular digital rectal examinations, PSA measurements, and repeated biopsies. The advanced PCa cohort comprised of 194 patients diagnosed between January 2000 and December 2011 and was established retrospectively by a standardised extraction of patient data. Immunohistochemical (IHC) assessment for ERG protein expression was performed in all tumours containing diagnostic specimens (AS cohort: n = 459; advanced PCa cohort: n = 968), re-biopsies during AS (n = 402), and deferred radical prostatectomy specimens following AS (n = 86). An anti-ERG rabbit monoclonal primary antibody (clone: EPR3864, dilution 23 μg/ml) was used. Fluorescence in situ hybridisation (FISH) for the TMPRSS2-ERG gene fusion was performed in 76 selected biopsies from the AS cohort using the ZytoLight TriCheck Probe, SPEC ERG/TMPRSS2.

Based on the AS cohort, Study 1 found a 97.3% concordance between the FISH assay and the IHC assay. The IHC assessments of ERG expression in diagnostic biopsies, re-biopsies, and radical prostatectomy specimens demonstrated a low proportion of temporal ERG reclassification. During four rounds of re-biopsies, 6.6% of the patients experienced ERG reclassification, and depending on the number biopsy specimens included 5.8-10.5% of the patients were ERG reclassified after radical prostatectomy. In Study 2, 46.4% of the AS patients were categorised as ERG-positive, whereas 53.6% were categorised as ERG-negative. After median 4.1 years follow-up, a significantly higher risk of disease progression was observed in men with ERG-positive tumours corresponding to a two-year cumulative incidence of 58.6% (95% CI: 48.7-68.5) and 21.7% (95% CI: 14.3-29.1) in the ERG-positive and the ERG-negative group, respectively. In the multiple causespecific Cox analyses, ERG expression was a strong and independent predictor of overall disease progression (HR: 2.45, 95% CI: 1.62-3.72) and histopathological progression in repeated biopsies (HR: 3.06, 95% CI: 1.78-5.26), and ERG status increased the discriminative ability for predicting disease progression significantly. Study 3 included 194 patients with advanced PCa treated with firstline ADT. In total, 54.1% had ERG-positive tumours and 45.9% had ERG-negative tumours. With a median of 6.8 years of follow-up, the risk of developing castration-resistant PCa (CRPC) did not differ between the ERG subgroups (p = 0.51). Finally, inclusion of ERG status in a multiple cause-specific Cox model did not increase the discriminative ability for predicting CRPC development during the first 8 years of ADT.

The thesis has demonstrated that assessment of ERG protein expression is feasible in biopsy specimens, and a high concordance was found between the IHC assay and FISH assessment of ERG rearrangement. The low proportion of ERG reclassification between biopsies and prostatectomies supports the use of ERG assessment in biopsies to characterise the individual patient's ERG status. ERG status harbours important prognostic value in terms of tumour progression for patients managed on AS, whereas ERG expression has no predictive value for ADT response in men with advanced PCa undergoing first-line castration-based ADT. The overall conclusion of the thesis is that ERG protein expression provides valuable prognostic information in low-risk PCa managed observationally, and ERG expression might be used to personalise follow-up regimens in future AS programmes.

Danish medical journal. 2016 Dec [Epub]

Kasper Drimer Berg