OUP user menu

ERG Protein Expression in Human Tumors Detected With a Rabbit Monoclonal Antibody

Oksana Yaskiv MD, Brian P. Rubin MD, PhD, Huiying He MD, PhD, Sara Falzarano MD, Cristina Magi-Galluzzi MD, PhD, Ming Zhou MD, PhD
DOI: http://dx.doi.org/10.1309/AJCP3K5VUFALZTKC 803-810 First published online: 1 December 2012

Abstract

Avian v-ets erythroblastosis virus E26 oncogene homolog (ERG) is highly sensitive and specific for endothelial neoplasms and specific for prostate carcinoma. We characterized a rabbit anti-ERG antibody as an immunohistochemical agent to detect ERG expression in various tumors using tissue microarrays with a wide array of epithelial and mesenchymal tumors. ERG was positive in 63 (38%) of 168 prostate carcinomas and negative in all other epithelial tumors. ERG was positive in all 125 vascular lesions. It was also positive in the sarcomatoid component of a high-grade urothelial carcinoma and 6 (40%) of 15 meningiomas. Twelve (80%) of 15 meningiomas were positive for Fli1, including all 6 ERG-positive cases. Positive immunostaining with this antibody is therefore highly specific for prostate carcinoma and vascular lesions, with a few caveats. ERG is rarely detected in nonvascular mesenchymal tumors with this antibody. Furthermore, about 40% of meningiomas are also positive for ERG immunohistochemically, probably because of cross-reactivity with Fli1.

Key Words:
  • ERG
  • Vascular lesion
  • Prostate carcinoma
  • Meningioma
  • Immunohistochemistry

Avian v-ets erythroblastosis virus E26 oncogene homolog (ERG), located on chromosome 21q22.3, belongs to the ETS transcription factor gene family, whose members also include Fli1 and ETS-1.1,2 Recent studies found that ERG also plays an important role in prostate carcinogenesis, as seen by the fact that gene fusions involving androgen-regulated gene transmembrane protease serine 2 (TMPRSS2) and ERG are the most common recurrent genetic alteration in prostatic carcinoma.37 Using fluorescence in situ hybridization (FISH) and reverse transcription polymerase chain reaction (RT-PCR) techniques, the TMPRSS2:ERG gene fusion is highly specific for and detected in approximately 50% of prostate carcinomas in radical prostatectomy cohorts but not in other common human cancers.816 ERG gene fusion leads to high-level production of a truncated ERG protein in prostate cancer cells that harbor ERG gene fusions. Several recent studies demonstrated that positive immunohistochemical detection of ERG protein is highly correlated with the ERG gene fusion determined using FISH and RT-PCR, with sensitivity and specificity both exceeding 95%.10,13,17 These studies have confirmed that antibody-based detection of ERG protein can be used as a surrogate marker for ERG gene status in prostate cancer diagnosis and to study the role of ERG gene fusion in prostate carcinogenesis.

ERG is also strongly expressed in endothelial cells, regulates endothelial apoptosis and angiogenesis,18 and has been used as a highly sensitive and specific marker for vascular tumors.14

Before using antibody-based ERG detection as a specific marker for prostate carcinomas and vascular tumors, the expression of ERG protein in a broad array of human tissues and tumors should be assessed. Currently 2 anti-ERG antibodies are available, a mouse monoclonal (CPDR ERG-Mab)13 and a rabbit monoclonal antibody (clone ID: EPR3864).10,17 These 2 antibodies differ significantly in several aspects. The mouse monoclonal antibody (CPDR ERG-Mab) reacts with the N-terminus of the ERG protein and does not cross-react with Fli1, another ETS gene family member. In contrast, the rabbit monoclonal antibody (clone ID: EPR3864) reacts with the C-terminus of the ERG protein and cross-reacts with Fli1.17,19 The mouse monoclonal antibody (CPDR ERG-Mab) was recently characterized in detail, and using this antibody, ERG protein expression was detected in various fetal and adult tissue and tumors, benign and malignant vascular tumors, blastic extramedullary myeloid tumors, and rarely, Ewing sarcomas, in addition to prostate carcinomas.14

The rabbit anti-ERG monoclonal antibody (clone ID: EPR3864), another antibody currently in clinical use,10,16,17,20 has not been thoroughly investigated for its use as an immunohistochemical reagent to detect ERG protein expression in various epithelial and nonepithelial neoplasms, which was the objective of this study.

Materials and Methods

Case Cohort

This study was approved by the authors’ institutional review board. Several multitumor tissue microarrays composed of 539 cases of multiple tumor types, including 335 epithelial tumors, 125 vascular lesions, and 79 nonvascular mesenchymal tumors, were used Table 1.

Tissue Microarray Construction

H&E-stained slides were reviewed to confirm the diagnoses. Tumor regions were marked. Triplicate tissue cores, 0.6 or 2 mm in diameter, were taken from the donor blocks that corresponded with the marked tumor areas on H&E slides and placed into a tissue microarray recipient block using a manual arrayer (Estigen, Tartu, Estonia). Four-micrometer sections were placed onto Superfrost slides for ERG immunohistochemistry.

Evaluation of ERG and Fli1 Protein Expression on Immunohistochemistry

Immunohistochemistry for ERG was performed on a Discovery XT automated stainer (Ventana Medical System, Tucson, AZ). Antigen retrieval was performed in Tris/borate/ethylenediaminetetraacetic acid buffer, pH 8.0 to 8.5, for 8 minutes at 95ºC, 12 minutes at 100ºC, and then 8 minutes at room temperature. Slides were subsequently incubated with an anti-ERG rabbit monoclonal antibody (1:100 dilution, clone ID: EPR3864, Epitomics, San Diego, CA) for 32 minutes at room temperature. A secondary antibody (OmniMap anti-Rabbit HRP, Ventana) was applied for 16 minutes at room temperature. Red chromogen (ultraView, Ventana) was applied for 8 minutes at room temperature. Slides were counterstained with hematoxylin II. For Fli1 immunohistochemistry, the slides were deparaffinized and pretreated on a Bond-MaxTM immunostainer (Leica Microsystems, Buffalo Grove, IL) with ER2 solution (pH 9.0) for 20 minutes, incubated with an anti-Fli1 antibody (clone MRQ-1, dilution 1:200, Cell Marque, Rocklin, CA) for 15 minutes at room temperature, and developed with the Polymer Bond Refine Detection System (Leica Microsystems).

Vascular endothelial cells were used as the internal positive controls that were assigned a staining score of “strongly positive,” and the staining in lymphocytes was assigned a staining score of “weakly positive.” A staining intensity that was between endothelial cells and lymphocytes was assigned a staining score of “moderately positive.” The expression of ERG protein in tumor cells was scored as negative, weak, moderate, and strong, and the expression of Fli1 was scored as positive and negative.

Results

Endothelial cells and lymphocytes, when present in the tissue cores, were always positive for ERG, strongly and weakly, respectively Image 1. Endothelial cells were used as an internal positive control. The results of nuclear ERG staining in various human tumors are summarized in Table 1.

Sixty-three (38%) of 168 cases of prostatic adenocarcinoma were positive for ERG Image 2, with weak staining in 28 (44%), moderate staining in 31 (49%), and strong staining in 4 (6%) positive cases. In a case of high-grade urothelial carcinoma with sarcomatoid differentiation in a female patient Image 3A, ERG was strongly positive in the sarcomatoid component Image 3B composed of nondescript malignant spindle cells and rare cells with cartilaginous differentiation but not in the well-differentiated carcinomatous component. The sarcomatoid component was not angiosarcomatous because it was negative for vascular endothelial markers CD31 and Fli1. ERG was negative in all other epithelial tumors (Table 1). One case each of thyroid medullary carcinoma, parathyroid adenoma, and small cell carcinoma of the lung exhibited a speckled, rather than uniform, nuclear ERG staining pattern Image 4. These cases were scored negative for ERG staining.

ERG was positive in all 125 cases of vascular lesions, including arteriovenous malformation, papillary endothelial hyperplasia, and benign and malignant vascular tumors (Table 1) Image 5A and Image 5B. The staining was weakly positive in 2 hemangiomas (8%) and 1 angiosarcoma (2%). ERG staining was moderate and strong in the remaining vascular lesions.

View this table:
Table 1

ERG was positive in 6 meningiomas (40%), including 5 fibrous meningiomas (83%) Image 6A and Image 6B and 1 transitional-type meningioma (50%). The staining was moderate in 5 cases and weak in 1 case. Other types of meningiomas, including secretory (n = 2), angiomatous (n = 1), and meningothelial (n = 4), showed negative staining. All but 3 cases (12/15, 80%) of meningiomas were positive for Fli1, including all 6 ERG-positive cases Image 6C.

Two lymphoma cases, 1 follicular lymphoma and 1 diffuse large B-cell lymphoma, were both weakly positive for ERG expression.

Discussion

Immunohistochemical detection of ERG protein is considered to be specific for prostate cancer and endothelial lesions. Previous studies have shown that ERG protein expression was detected in vascular endothelial cells 10,13,14,16,17,19,20 and ERG gene fusion–positive prostate carcinoma.10,13,16,17,20 The ERG expression in other types of neoplasms was investigated in only 1 other study using a mouse anti-ERG monoclonal antibody (CPDR ERG-Mab). The rabbit anti-ERG antibody (clone ID: EPR3864)10,14,16,17 used in the current study differs significantly from the mouse monoclonal antibody in the ERG peptide sequences that it recognizes and its cross-reactivity with Fli1. This study, therefore, investigated ERG expression detected with a rabbit anti-ERG antibody (clone ID: EPR3864) in prostate carcinoma, vascular tumors, and other epithelial and mesenchymal tumors and determined the sensitivity and specificity of immunohistochemical detection of ERG protein for prostate carcinoma and vascular tumors.

Image 1

Vascular endothelial cells are strongly positive (arrows) and lymphocytes are weakly positive (arrowheads) for ERG. The endothelial cells were used as internal positive controls.

Image 2

A Gleason score of 4 + 4 = 8 prostate cancer showing moderate positivity for ERG.

Image 3

High-grade urothelial carcinoma with sarcomatoid differentiation. A, The sarcomatoid component comprises nondescript short spindle cells and malignant-appearing cartilage. B, Strong ERG staining was found in the majority of spindle cells but only rarely in cartilaginous cells. C, The sarcomatoid component was negative for the vascular endothelial marker CD31. Note vascular endothelial cells are positive for CD31.

Previous studies found that ERG gene rearrangement is highly specific for prostate carcinoma and is present in 40% to 50% of prostate cancer814,16 and 15% to 29% of high-grade prostatic intraepithelial neoplasia,16,21 which is always in close association with fusion-positive cancer. Scheble et al, using FISH, examined TMPRSS2-ERG gene rearrangement in 54 different tumor types and did not identify any rearrangements in nonprostate tumors.15 Miettinen et al14 studied ERG protein expression detected with immunohistochemistry using a mouse anti-ERG monoclonal antibody (CPDR ERG-Mab) and found that ERG was detected in approximately 45% of prostate carcinoma. Other epithelial tumors, including carcinomas of the breast, gastrointestinal tract, gynecologic system, kidney, lung, ovary, pancreas, salivary glands, skin, thyroid, and testis, were all negative for ERG. The exceptions to this finding were 1 of 42 large cell undifferentiated pulmonary carcinomas and 1 of 27 mesotheliomas, each of which showed focal nuclear ERG positivity. In the current study, ERG protein was detected in 38% of prostate carcinomas, similar to the positive rate of ERG found on immunohistochemistry and FISH in other studies.816 In 1 case of high-grade urothelial carcinoma with sarcomatoid differentiation, the sarcomatoid component comprising nondescript malignant spindle cells was positive for ERG, but the carcinomatous component was negative. Our current study and that of Miettinen et al14 convincingly demonstrate that ERG expression in epithelial neoplasms other than prostate carcinoma is exceedingly rare. Therefore, a positive immunostaining reaction in an epithelial neoplasm confirms its prostatic origin. We also observed speckled, rather than uniform, nuclear ERG staining in 1 case of thyroid medullary carcinoma, parathyroid adenoma, and small cell carcinoma of the lung, all of neuroendocrine origin. These cases were interpreted as negative but merit further investigation.

Image 4

Speckled nuclear ERG staining pattern was observed in a parathyroid adenoma (arrows).

Image 5

An angiosarcoma (A) was strongly positive for ERG (B).

Image 6

A fibrous meningioma (A) was moderately positive for ERG (B) and strongly positive for Fli1 (C).

ERG was diffusely and strongly positive in normal vascular endothelial cells and in all 125 cases of vascular lesions, including arteriovenous malformation, papillary endothelial hyperplasia, and benign and malignant vascular tumors, with predominantly homogenous moderate to strong staining. We can therefore conclude that the sensitivity of ERG for vascular lesions is 100%. In benign vascular tumors strong and moderate staining was present in 84% and 16% of cases, respectively. In the malignant group, strong and moderate staining was present in 85% and 15% of cases, respectively. This finding is in concordance with the study of Miettinen et al.14 ERG compares favorably with other endothelial markers, because its sensitivity for vascular lesions, both benign and malignant ones, is 100%, and superior to the sensitivity of CD31 (approximately 90%),22,23 CD34 (approximately 90%),2325 von Willebrand factor (up to 70%),24 and Fli1 (up to 94%).26,27 Novel vascular markers such as VEGFR-3,24,28,29 podoplanin,30 or CD117 (C-kit)31 appear to be less sensitive and specific.

Miettinen et al14 found that ERG immunohistochemistry using the mouse monoclonal antibody (CPDR ERG-Mab) is specific for vascular lesions and is not found in other mesenchymal tumors. However, in this study, the sarcomatoid component of a high-grade urothelial carcinoma was strongly immunoreactive with the rabbit monoclonal ERG antibody (clone ID: EPR3864). Two other vascular markers, including CD31 and Fli1, were negative, confirming that the sarcomatoid component was not a vascular lesion. This finding suggests that ERG immunohistochemistry using the rabbit monoclonal antibody (clone ID: EPR3864) is not entirely specific for vascular neoplasms and may rarely be detected in other mesenchymal neoplasms.

Miettinen et al14 examined 973 nonepithelial mesenchymal, neuroectodermal, and hematopoietic tumors and found positive ERG (mouse monoclonal, clone CPDR ERG-Mab) staining in 9 hematopoietic tumors but not in other tumors. In the present study, we found that ERG was strongly positive in the sarcomatoid component of a high-grade urothelial carcinoma and weakly positive in 2 lymphomas. ERG is negative in all other epithelial (except prostate carcinoma) and mesenchymal (except meningiomas) tumors.

An interesting finding in this study is that 40% of meningiomas were positive for ERG, mainly in the fibrous type (5 of 6 cases) and the transitional type (1 of 2 cases). Other types of meningiomas, including meningothelial, secretory, and angiomatous, were negative for ERG. Vascular endothelial cells present in these tumors were strongly positive for ERG. In contrast, Miettinen et al14 examined 34 cases of meningiomas using the mouse monoclonal anti-ERG antibody (CPDR ERG-Mab) and found negative ERG staining. Eighty percent of meningiomas, including all 6 ERG-positive meningiomas, were positive for Fli1 in this study. These findings suggest that positive ERG immunoreactivity in meningiomas using the rabbit anti-ERG monoclonal antibody (clone ID: EPR3864) observed in this study likely represents cross-reactivity with Fli1. However, there are no reports of Fli1 expression in meningiomas. Alternatively, our antibody may cross-react with other ETS gene family members. Upregulation of another ETS gene family member, ETS-1, and several genes that ETS-1 controls, such as matrix metalloproteinases,32 has been linked to atypical and anaplastic meningiomas,3336 and ETS-1 expression is associated with an increased risk of recurrence.37 The expression of Fli1 or other ETS gene family members in meningiomas deserves further investigation.

In summary, we characterized ERG protein expression in a wide range of human tumors using a rabbit anti-ERG monoclonal antibody (clone ID: EPR3864). The advantage of this antibody is that it is highly specific for prostate carcinomas and not detected in any other epithelial tumors. A positive ERG immunostain with this antibody supports the prostatic origin of the ERG-positive epithelial tumors. ERG immunostains are also highly sensitive for vascular and lymphatic endothelial cells and can be used in the workup of benign and malignant vascular lesions. However, one needs to be aware of its potential pitfalls in the diagnosis of prostate cancer or vascular lesions. ERG is rarely detected in nonvascular mesenchymal tumors; therefore, a positive ERG staining in the absence of other positive vascular markers may not always support vascular differentiation. Furthermore, about 40% of meningiomas are also positive for ERG on immunohistochemistry, probably because of cross-reactivity with Fli1.

References

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
View Abstract