Aims We evaluated the consistency in ERG protein expression from diagnostic specimens through rebiopsies to radical prostatectomies in patients with clinically localised prostate cancer to investigate the validity of ERG status in biopsies.

Methods ERG expression was assessed by immunohistochemistry (IHC) in 625 biopsy sets and 86 radical prostatectomy specimens from 265 patients with prostate cancer managed on active surveillance. For IHC, a rabbit monoclonal primary antibody was used (clone: EPR3864). TMPRSS2-ERG fluorescence in situ hybridisation (FISH) analyses were performed in 74 biopsies using the FISH ZytoLight TriCheck Probe (SPEC ERG/TMPRSS2). FISH results were correlated with IHC findings.

Results The concordance between FISH and IHC was 97.3% and IHC demonstrated a sensitivity and specificity for ERG rearrangement of 100% and 95.5%, respectively. Applying IHC, 38.1% of patients were ERG-positive, 53.6% were ERG-negative and 8.3% showed both ERG-positive and negative tumour foci (ERG heterogeneous) at diagnosis. When ERG status was dichotomised (ERG-positive or heterogeneous vs ERG-negative), 95.6%–97.1% of patients did not experience ERG reclassification during the first two rounds of rebiopsies. The concordance in ERG status between biopsies and surgical specimen was 89.5%–94.2% depending on the number of rebiopsies included. Sampling bias was assumed to explain most (81.3%) of the mismatches in ERG status.

Conclusions Consistency in ERG status ranged from 90% to 95% for patients undergoing serial biopsies and radical prostatectomy. This indicates that biopsies can be used reliably to investigate ERG's prognostic and predictive value.