Weaker ERG expression in patients with ERG-positive prostate cancer is associated with advanced disease and weaker androgen receptor expression: An Indian outlook

Abstract

IntroductionGene fusion between TMPRSS2 and ERG has a causal role in prostate cancer initiation. However, studies evaluating its role clinically have shown conflicting results. We investigated simultaneously multiple aspects of ERG, namely, “presence” and “strength” of ERG expression and “correlation” of ERG with other common clinicopathological parameters. Materials and methodsFrom January 2012 to November 2013, the status of ERG, androgen receptor (AR), and p53 was prospectively determined by immunohistochemistry unselectively in all types of specimens positive for prostate cancer. “Strength” of expression was measured in terms of “intensity” as well as “percentage positivity,” with each parameter given a score from 0 to 3 based on fixed protocol, which was tested for interrater variability as well as test-retest reliability. Data were collected for age, Gleason score, prostate specific antigen levels, presence of perineural invasion and lymphovascular invasion, high-grade prostatic intraepithelial neoplasia, and cancer stage. ResultsIn total, 100 specimens were analyzed, and overall 51 patients had ERG-positive immunostaining. ERG-positive tumors had lower presence of high-grade prostatic intraepithelial neoplasia and p53 positivity, with no significant difference in prostate specific antigen levels, Gleason scores, and presence of lymphovascular invasion. Moreover, 54 patients had complete stage information, and the absolute number of patients with ERG positivity increased with increasing clinical stage. Among these, 30 patients were ERG positive, and ERG score had strong positive correlation with AR expression (Spearman correlation coefficient 0.677). However, median ERG scores showed a significant decline (consistent across percentage positivity and intensity) in patients with stage 4 disease, and score≤2 had 88.2% specificity in identifying patient with stage 4 disease. Cohen׳s κ = 0.81, whereas intraclass correlation coefficient was 0.95, indicating substantial agreement and near-perfect reproducibility of scoring scheme for immunohistochemistry. ConclusionERG-positive tumors increase in proportion with increasing stage of disease, but strength of ERG expression in ERG-positive patients shows a significant decline, or “loss,” in patients with stage 4 disease. This may have potential therapeutic implications as ERG expression score showed strong positive correlation with AR expression score.