Introduction

Prostate cancer (PCa) remains the most common non-cutaneous cancer and the most common cause of cancer-related death among men in the USA, with 29 720 deaths estimated for 20131,2. In Brazil, there were 12 778 registered deaths from PCa in 2010 and an estimated 60 180 new cases in 20123.

In 2005, genomic rearrangement between androgen-regulated transmembrane protease (TMPRSS2, 21q22.3) and v-ets erythroblastosis virus E26 oncogene homolog (ERG, 21q22.2) was described as the most common genetic alteration in PCa cells, occurring in approximately 50% of cases4. During chromosomal rearrangement, one TMPRSS2 promoter allele is lost and one ERG allele gain it, which results in the overexpression of ERG protein in tumor cells5.

This fusion can be detected by fluorescence in situ hybridization (FISH) across deletions or chromosomal translocations6,7, real time PCR (qRT-PCR) across fusion transcripts8,9 and immunohistochemistry by ERG protein overexpression10,11,12.

TMPRSS2 promoter analysis revealed the presence of a non-canonical androgen receptor (AR) as a CIS-regulatory target of AR action13. Mutations, amplifications or overexpression of the AR and deletions or inactivation of the tumor suppressor gene PTEN (phosphatase and tensin homolog) are frequently identified in PCa cells14,15. ERG gene rearrangements and PTEN loss often occur concomitantly in PCa cells and promote tumor progression through the PI3K pathway, which increases the advantages of pre-malignant cells compared to normal cells16,17.

This study aimed to evaluate the presence of ERG protein overexpression and TMPRSS2:ERG gene fusion in a cohort of patients with PCa who have undergone radical prostatectomy or radiotherapy and to determine whether there is a correlation between these events. This analysis is particularly interesting in Brazilian patients, where there is a lack of studies evaluating gene fusion in PCa and the population is ethnically heterogeneous, which is a consequence of centuries of miscegenation among Europeans, Africans and Amerindians.

Methods

Study population

The study was designed in accordance with the Guidelines and Standards Regulation Research Involving Human Beings (Resolution 196/1996 of the National Health Council) and was approved by the Ethics Committee of Barretos Cancer Hospital, SP, Brazil (425/2010). In this Institution is requested of all patients completing an informed consent at the first appointment, getting attached to records.

We evaluated 98 patients who underwent radical prostatectomy at the Barretos Cancer Hospital between 01/2006 and 12/2007. Additionally, 27 samples containing cells negative for PCa were evaluated. A review of patient medical records provided the following information: age at the time of treatment, race, pre-treatment PSA level, Gleason score, clinical stage, D'Amico progression risk group, surgical margin status, perineural infiltration of the tumor and time to PSA recurrence or follow-up time after treatment. Biochemical recurrence of PSA was defined as two consecutive PSA tests with levels greater than or equal to 0.2 ng.mL−1 for patients who underwent radical prostatectomy, with follow-up tests occurring for 5 to 6 years. By evaluating PSA levels at the time of treatment, Gleason scores of patient biopsies and clinical stages of PCa, patients were given a low, intermediate, or high risk D'Amico classification18. None of the patients in this study were treated with neoadjuvant therapy.

Tissue microarray (TMA) construction

The pathologist reviewed hematoxylin and eosin stained sections from each patient and identified the most representative regions of tumor samples. To construct a TMA, two representative cores 1 mm in diameter were taken from tissue samples and arrayed on an individual paraffin block. A total of four TMAs were constructed using an MTA-1 Manual Tissue Arrayer (Beecher Instruments, Silver Spring, MD, USA); three of the TMAs contained PCa samples and one contained non-PCa samples.

Evaluation of ERG protein expression by immunohistochemistry

One 4 μm section of each sample was prepared for antibody testing. Immunohistochemistry staining was performed at room temperature using an Autostainer Link 48 (Dako, Copenhagen, DK). Slides were incubated for 20 minutes with Flex Ready-to-Use Monoclonal Rabbit Anti-Human ERG Clone EP111 (Dako, Copenhagen, DK). Slides were incubated with secondary antibody Envision Flex + Mouse ™ (LINKER) (Dako) for 15 minutes. Staining was performed using 3,3′-diaminobenzidine (DAB+, Dako) and the samples were counterstained with Harris' hematoxylin.

Immunohistochemical evaluation

Immunoreactions were evaluated semiquantitatively using the criteria previously described19,20. The immunoreaction extent was scored semiquantitatively as follows: 0: 0% presence of immunoreactive cells; 1: <5% presence of immunoreactive cells; 2: 5–50% presence of immunoreactive cells; and 3: >50% presence of immunoreactive cells. In addition, the intensity of staining was scored semiquantitatively as: 0: negative; 1: weak; 2: intermediate; and 3: strong. The final immunoreaction score was defined as the sum of both parameters (extent and intensity) and grouped as negative (score 0 and 1) or positive (3–6).

Immunohistochemistry validation by qRT-PCR

To confirm the findings obtained from immunohistochemistry testing, qRT-PCR was performed in a subset of 45 samples that tested positive or negative for ERG protein expression. Total RNA was extracted from four 10 μm sections of these samples using a RNeasy FFPE kit (Qiagen, Gaithersburg, MD, USA), with 200 ng of RNA converted into cDNA using a SuperScript® VILO™ Master Mix (Invitrogen, Carlsbad, CA, USA), both according to the manufacturer's instructions. We used TaqMan® inventoried assays (Applied Biosystems, Carlsbad, CA, USA) for ERG (Hs03063375_ft) and the endogenous control HPRT1 (hypoxanthine phosphoribosyltransferase 1, Hs99999909_m1). The qRT-PCR reaction was performed using 7900HT System equipment (Applied Biosystems) and a TaqMan® Fast Advanced Master Mix (Applied Biosystems), according to the manufacturer's instructions. To determine the relative expression level, the 2−ΔCT values were calculated.

Statistical analysis

To check the accuracy of immunohistochemistry in determining the presence of TMPRSS2:ERG, gene fusion, sensitivity, sensibility, kappa and ROC curve analyses were performed. The Chi-Square or Fisher exact test was used to compare immunohistochemistry data and clinico-pathological parameters. Joint association of clinical characteristics with ERG expression was performed by multiple logistic regression. A biochemical recurrence-free survival curve was estimated using the Kaplan-Meier method and a Cox regression model was used to identify associated risk factors. For the entire study, a p-value < 0.05 was considered to be statistically significant. All analyses were performed using IBM-SPSS 21.0 (IBM Corporation, New York, NY, USA).

Results

Immunohistochemistry analysis detected ERG protein expression in 46 of the 98 samples (46.9%); all 27 negative controls tested negative for ERG protein. Endothelial tissue usually expresses ERG protein and was used as an internal positive control for all reactions (Figure 1).

Figure 1
figure 1

Immunohistochemistry.

(A) Negative PCa sample ERG−, with endothelium tissue ERG+, serving as internal positive control for immunohistochemistry. (B) PCa sample ERG+.

Full size image

The total RNA quality was assessed using a 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA) and the majority of samples had an RNA integrity number (RIN) of 2.4. Of the 45 samples tested, 21 (46.67%) showed amplification and 24 (53.33%) did not show amplification (Figure 2). There was no significant difference in the detection rates between immunohistochemistry and qRT-PCR (k = 0.643, p < 0.001). Accuracy analysis showed that immunohistochemistry had high sensitivity (81.0%) and high specificity (83.3%). Accuracy analysis showed that immunohistochemistry had high sensitivity (81.0%) and high specificity (83.3%) and technique had in common with qRT-PCR 17 positive and 19 negative cases (Figure 3).

Figure 2
figure 2

Relative TMPRSS2:ERG gene fusion expression, with the normalized distribution of samples by the 2−ΔCT method.

Samples without expression of the gene fusion do not show amplification.

Full size image
Figure 3
figure 3

Venn diagram representing a patient classification according to the IHC and qRT-PCR results.

(A) Positive TMPRSS2:ERG gene fusion expression and ERG+. (B) Negative TMPRSS2:ERG gene fusion expression and ERG−.

Full size image

Patients who had PSA levels less than or equal to 4.0 ng.mL−1 had positive ERG expression compared with patients with PSA levels greater than or equal to 10.0 ng/mL (p = 0.031, Table 1). There was no statistical association between ERG expression and the other clinico-pathologic parameters. Logistic regression was performed by pre-treatment PSA level, clinical stage and surgical margin status and only PSA levels less than or equal to 4.0 ng.mL−1 (OR 4.37; 95% CI: 1.19–16.04; p = 0.026) and PSA levels between 4.0 and 10.0 ng.mL−1 (OR 3.12; 95% CI: 1.15–8.49; p = 0.026) was statistically significant.

Table 1 Association between ERG protein immunoreaction and clinico-pathological characteristics in patients with PCa (Barretos Cancer Hospital, 2006–2007)
Full size table

In Kaplan-Meier analysis, ERG expression was not predictive for biochemical recurrence-free survival (Figure 4). Univariate Cox regression was performed by D'Amico score and surgical margin status and only the patients in the intermediate or high-risk D'Amico groups were further likely to experience biochemical recurrence than the low risk group (HR 5.00; 95% CI: 1.51–16.55; p = 0.008, Table 2).

Table 2 Analysis of biochemical recurrence-free survival of patients with PCa, in relation to clinico-pathological characteristics (Barretos Cancer Hospital, 2006–2007)
Full size table
Figure 4
figure 4

Kaplan-Meir curve for biochemical recurrence-free survival of patients with prostate cancer whose tumors showed positive or negative staining for ERG protein.

Full size image

Discussion

This study evaluated the presence and potential prognostic value of the ERG protein in patients with PCa tumors.

Our study of samples from Brazilian patients with PCa found that immunohistochemistry tests were positive for ERG expression in 45.8% of cases, which is consistent with other studies demonstrating a frequency of approximately 50%4,21,22. Detection of TMPRSS2:ERG gene fusion by immunohistochemistry had a sensitivity of 81.0% and specificity of 83.3%. Chaux et al validated ERG immunohistochemistry using FISH and reported a sensitivity and specificity of 86% and 89%, respectively and an area under the ROC curve of 0.87 (p < 0.00001)23. These results corroborate the fact that TMPRSS2:ERG gene fusion, which leads to overexpression of the ERG protein, is the most frequent oncogene in this type of cancer4.

Several studies have attempted to evaluate TMPRSS2:ERG gene fusion as a prognostic indicator of some risk factors, including the Gleason score, PSA level, metastatic PCa or death by PCa and the results have been variable24,25,26,27,28. Our findings showed no relationship between the presence of the ERG protein and a patient's clinical stage, Gleason score or biochemical recurrence, which corroborates the results of other studies24,28,29,30,31. However, we found an increased likelihood of ERG positive immunoreactions when PSA levels were less than 10.0 ng.ml−1 (p = 0.031)31,32. Sun et al concluded, using cell and animal models, that ERG protein overexpression increases the regulation of the c-MYC oncogene. This occurs because ERG is capable of interacting with ETS binding elements in the oncogene promoter region. This c-MYC upregulation blocks the expression of the MSMB, SLC45A3 and PSA genes, which affect prostate cell differentiation by altering epithelial differentiation. This event might explain the association between the presence of ERG protein and lower levels of PSA33.

As PCa appears years after androgen levels peak, hormone exposure for a long period may be necessary for the development of cancer. This relationship also appears to be associated with the appearance of gene fusion. It has been proposed that AR and TMPRSS2:ERG jointly promote the development of invasive adenocarcinoma because the transcription of the gene TMPRSS2 is dependent on androgen24,34,35.

It is still uncertain whether the presence of ERG protein is a predictor of other risk factors, but the frequency and specificity of ERG in PCa make it a potential adjunct tool for the diagnosis of prostate cancer, regardless of the population type.