Immunohistochemical expression of erythropoietin and erythropoietin receptor in breast carcinoma

Authors

  • Geza Acs M.D., Ph.D.,

    Corresponding author
    1. Department of Pathology and Laboratory Medicine, University of Pennsylvania Medical Center, Philadelphia, Pennsylvania
    • Department of Pathology and Laboratory Medicine, University of Pennsylvania Medical Center, 6 Founders Pavilion, 3400 Spruce Street, Philadelphia, PA 19104
    Search for more papers by this author
    • Fax: (215) 349-5910

  • Paul J. Zhang M.D.,

    1. Department of Pathology and Laboratory Medicine, University of Pennsylvania Medical Center, Philadelphia, Pennsylvania
    Search for more papers by this author
  • Timothy R. Rebbeck Ph.D.,

    1. Center for Clinical Epidemiology and Biostatistics and Cancer Center, University of Pennsylvania, Philadelphia, Pennsylvania
    Search for more papers by this author
  • Peter Acs M.D., Ph.D.,

    1. Department of Medicine, Huron Hospital, Cleveland Clinic Health System, Cleveland, Ohio
    Search for more papers by this author
  • Ajay Verma M.D., Ph.D.

    1. Department of Neurology, Uniformed Services University of the Health Sciences, Bethesda, Maryland
    Search for more papers by this author

Abstract

BACKGROUND

Erythropoietin (Epo), induced by hypoxia, controls the survival, proliferation, and differentiation of Epo receptor (EpoR)-bearing erythroid progenitors and plays a role in the protection of neurons from hypoxic damage. Hypoxia in malignant disease is associated with invasion, metastasis, resistance to therapy, and selection for cells with diminished apoptotic potential. The authors recently demonstrated the basal and hypoxia-stimulated expression of Epo and EpoR in human breast carcinoma cell lines and in breast carcinomas, suggesting a role for autocrine Epo signaling in the hypoxic adaptations of mammary neoplasms.

METHODS

The authors characterized the expression of Epo and EpoR by immunohistochemistry in 184 invasive mammary carcinomas and 158 in situ mammary carcinomas and benign mammary epithelium. They analyzed the correlation of Epo and EpoR immunostaining with clinicopathologic tumor features and the patients' smoking history.

RESULTS

Benign mammary epithelial cells showed weak-to-moderate expression of Epo and EpoR. EpoR immunostaining was increased in carcinomas compared with benign epithelium both in nonsmokers and smokers, and Epo immunostaining was increased in carcinomas compared with benign epithelium in nonsmokers but not in smokers. Prominent Epo staining was seen in tumor cells adjacent to necrotic areas and at the infiltrating edge of tumors. EpoR staining, but not Epo staining, was significantly greater in tumors that showed high histologic grade, tumor necrosis, lymphovascular invasion, lymph node metastases, and loss of hormone receptor expression.

CONCLUSIONS

The current findings suggest that increased EpoR expression may play an important role in breast carcinogenesis. The induction of autocrine or paracrine Epo signaling may represent a novel mechanism by which hypoxia can promote breast carcinoma. Cancer 2002;95:969–81. © 2002 American Cancer Society.

DOI 10.1002/cncr.10787

Erythropoietin (Epo), a glycoprotein hormone stimulator of erythropoiesis,1–3 promotes the proliferation and differentiation of erythroid precursor cells while also inhibiting their apoptosis.1–5 Epo exerts its actions through its specific receptor (EpoR), a member of the cytokine receptor superfamily.1, 6–8 Major signal-transduction pathways activated by Epo include the Jak/signal transducer and activator of transduction (STAT) and Ras/mitogen-activated protein kinase pathways, which are involved in the inhibition of apoptosis and the stimulation of cell proliferation in response to this hormone.4, 7–12

During adult life, Epo normally is produced by the kidney and liver,1, 3, 5 and its gene expression is modulated primarily by tissue hypoxia.13 However, other sites of Epo production have been reported recently, including bone marrow macrophages,14 brain astrocytes,15, 16 and trophoblast cells of the human placenta.17 Considerable amounts of Epo also are present in human milk,18 the source of which is the lactating breast glands, as shown recently.19

In recent years, it has become clear that EpoR is expressed by a variety of nonhematopoietic cell types, including endothelial cells,20 neurons,21 and trophoblast cells.22 Although the specific function(s) of EpoR in these sites is not yet understood, there is increasing evidence that EpoR expressed in these tissues is functional, suggesting a wider biologic role for Epo signaling that is not related to erythropoiesis.23–25 Specifically, the brain appears to have a paracrine Epo/EpoR system that is independent of the erythropoietic system: Neurons express EpoR,15, 21, 26 and astrocytes produce Epo.15, 16 Brain Epo signaling appears to protect neurons from ischemic damage by inhibiting their apoptosis.26, 27 EpoR mRNA also is expressed in endothelial cells, and Epo stimulates the proliferation and migration of human endothelial cells28, 29 and angiogenesis.30 Another paracrine Epo/EpoR system in the uterus may play an important role in uterine angiogenesis through EpoR that is expressed in endometrial vascular endothelial cells.30

Elevated Epo levels have long been recognized in patients with renal carcinomas, Wilms tumors, hepatomas, and cerebellar hemangioblastomas, all arising in anatomic sites in which Epo normally is expressed at low levels.13 In addition, symptomatic Epo production has been reported in a patient with breast carcinoma.31 Moreover, it was found that ectopic Epo expression in erythroleukemia cells mediated their autonomous growth.32 EpoR expression also was reported in patients with renal cell carcinoma, and a potential paracrine or autocrine role for Epo signaling for promoting the growth of renal carcinomas has been suggested.33

We recently described human breast carcinoma cell lines and human breast carcinomas that expressed Epo and EpoR mRNA and protein and showed that their expression was enhanced by hypoxia.34 Furthermore, we demonstrated that Epo signaling is active biologically and stimulates tyrosine phosphorylation, DNA synthesis, and proliferation in breast carcinoma cells. We also showed that the vasculature of solid tumors, including breast carcinomas, expresses EpoR. These results suggested a role for autocrine-paracrine Epo signaling in the hypoxic survival and neovascularization of tumors.

In the current study, we analyzed the expression of Epo and EpoR by immunohistochemistry in a large series of patients with primary breast carcinomas and in benign mammary epithelium. We analyzed the correlation between Epo and EpoR expression and various clinicopathologic tumor features. Because the most common reason for elevated serum Epo levels in humans is hypoxia caused by cigarette smoking,35 and because recent studies have suggested that smoking may be associated with an increased risk of breast carcinoma,36–38 we also analyzed the correlation between Epo and EpoR immunostaining and the personal smoking history of patients.

MATERIALS AND METHODS

We selected 184 breast biopsy specimens from the Surgical Pathology files of the University of Pennsylvania Medical Center that included 158 in situ mammary carcinomas and 184 invasive mammary carcinomas. Hematoxylin and eosin-stained slides were reviewed independently by two pathologists (G.A. and P.J.Z.) to confirm the diagnosis, including histologic type and tumor grade, based on established criteria.39–41 Ductal carcinomas in situ (DCIS) were graded as described by Scott et al.,42 using primarily nuclear grade. All invasive carcinomas were graded using the modified, combined histologic grading system described by Elston and Ellis.43 Tumors also were evaluated to determine the presence or absence of tumor necrosis and lymphovascular invasion. Information regarding tumor staging, including tumor size, axillary lymph node involvement, and other therapeutic and prognostic biologic markers, such as estrogen and progesterone receptor status and HER2/neu overexpression, were retrieved from the pathology reports. Smoking history was collected from a standardized questionnaire on 102 individuals who participated in both the current study and in a separate study of breast carcinoma risk factors.

Immunohistochemical assays were performed on formalin fixed, paraffin embedded sections. Five-micrometer-thick sections were cut and deparaffinized in xylene and rehydrated in graded alcohols. Slides were steamed in 0.01 mol/L sodium citrate buffer, pH 6.0, for 20 minutes. Endogenous peroxidase activity was blocked by 3% hydrogen peroxide in methanol for 20 minutes. Slides were incubated with the polyclonal antibody against Epo (rabbit polyclonal, H-162; 1:200 dilution; Santa Cruz Biotechnologies Inc., Santa Cruz, CA) and EpoR (rabbit polyclonal, C-20; 1:200 dilution; Santa Cruz Biotechnologies Inc.) overnight at 4 °C. Slides were then washed five times with Tris-buffered saline containing Tween 20 (TBST), pH 7.6 (DAKO Corp., Carpinteria, CA), and incubated for 30 minutes at room temperature with horseradish peroxidase-labeled dextran polymer coupled to antirabbit antibody (EnVision + System HRP; rabbit; DAKO Corp.). Slides were then washed three times with TBST, developed with diaminobenzidine for 10 minutes, and counterstained with hematoxylin. Slides of fetal liver44 and placenta17, 22 were used as positive controls. A negative control was done in each case by omission of the primary antibody. The specificity of the antibodies were confirmed previously.19, 21, 34, 45, 46 In addition, the specificity of the immunoreactivity also was evaluated with an antibody absorption test: The primary antibody was preincubated with blocking peptide for EpoR (Santa Cruz Biotechnologies Inc.) or human recombinant Epo (rhEpo; 10:1 peptide:antibody ratio; R&D Systems, Minneapolis, MN), which resulted in the complete abolition of immunohistochemical staining. The specificity of the immunostaining reaction is supported further by other experiments using a mouse monoclonal anti-Epo (clone 9C21D11; R&D Systems) and a rabbit polyclonal anti-EpoR antibody (Upstate Biotechnology, Lake Placid, NY),47 which resulted in an immunostaining pattern similar to that obtained with the antibodies used in the current study (our unpublished observation).

Immunohistochemical stains for Epo and EpoR were interpreted semiquantitatively by assessing the intensity of staining according to a four-tiered scale. Cytoplasmic and/or membrane immunoreactivity was considered positive. First, the total percentages of positive tumor cells and benign ductal and lobular epithelial cells were assessed. Then, the percentages of weakly, moderately, and strongly staining cells were determined, so that the sum of these categories equated with the overall percentage of positivity. A staining score was then calculated as follows: score (from a maximum of 300) = sum of 1 × percentage of weak staining cells, 2 × percentage of moderately staining cells, and 3 × percentage of strong staining cells. In addition, a differential tumor score also was calculated by subtracting the staining score of the adjacent benign epithelium from that of the in situ and invasive carcinomas. Immunohistochemical stains were evaluated independently by two pathologists (G.A. and P.J.Z.). Slight differences in interpretation were resolved by simultaneous viewing.

The Wilcoxon signed rank test was used for the comparison of median Epo and EpoR expression levels in carcinomas and in adjacent benign mammary epithelial cells. The correlation between the levels of Epo and EpoR staining in invasive and in situ components of carcinomas was estimated using the Spearman rank correlation test. We examined the correlation between median Epo and EpoR expression levels and tumor type, tumor size, combined tumor grade, lymph node status, estrogen and progesterone receptor status, and HER2/neu overexpression. For the statistical analysis, the Mann–Whitney rank sum test and the Kruskal–Wallis one-way analysis of variance by ranks was used followed by a Dunn multiple-comparison test, when appropriate. Statistical significance was determined if the two-sided P value of a test was < 0.05.

RESULTS

Expression of EpoR

In normal ductal and lobular epithelial cells adjacent to tumor, weak, granular, cytoplasmic EpoR immunostaining was seen in 176 of 184 biopsy specimens (95.6%). Compared with Epo (see below), no increased EpoR staining was seen in lobules that showed secretory changes (not shown). Benign epithelial lesions, including usual hyperplasia, sclerosing adenosis, and papillomas, showed weak EpoR staining similar to that seen in normal epithelial cells (Fig. 1A). Diffuse, moderate-to-strong cytoplasmic and membrane EpoR immunostaining was seen in 121 of 122 DCIS tumors (99.2%), in 36 of 36 lobular carcinomas in situ (LCIS) tumors (100%), and in 183 of 184 invasive carcinomas (99.5%) (Fig. 1B–D). Although EpoR staining usually was uniform throughout the tumor, increased expression was seen in tumor cells adjacent to necrotic areas (Fig. 2A,B). Necrotic tissue itself showed no staining for EpoR. EpoR immunostaining was similar in the in situ and invasive components of the tumors (correlation coefficient [r] = 0.795; P < 0.0001; Spearman rank correlation test). In addition, strong EpoR expression was found in the tumor vasculature (not shown).

Figure 1.

Expression of erythropoietin receptor (EpoR) in benign mammary epithelial cells and invasive breast carcinoma. (A) Weak immunostaining for EpoR is seen in usual ductal hyperplasia without atypia. (B) Strong cytoplasmic and membrane immunostaining for EpoR is seen in invasive ductal carcinoma. Note that the benign duct in the center shows only weak staining for EpoR. (C) Strong EpoR immunostaining is seen in invasive ductal carcinoma. Only weak staining is present in the benign duct in the center. (D) Moderate-to-strong immunostaining for EpoR is seen in invasive lobular carcinoma. The benign duct in the center shows only weak EpoR staining; immunohistochemical staining for EpoR with hematoxylin counterstain). Original magnification ×200 (A); ×400 (B–D).

Figure 2.

Increased erythropoietin receptor (EpoR) expression in necrotic tumor areas. (A) Moderate immunostaining for EpoR is seen in ductal carcinoma in situ. Note the increased EpoR staining in the tumor cells adjacent to the central necrotic area. (B) Strong EpoR immunostaining is seen in cells of invasive ductal carcinoma adjacent to necrotic area (asterisk; immunohistochemical staining for EpoR with hematoxylin counterstain). Original magnification, × 400 (A, B).

Compared with benign epithelial cells, EpoR immunostaining was increased significantly in DCIS (P < 0.0001, Wilcoxon signed rank test; rs = 0.467; P < 0.0001), LCIS (P < 0.0001; Wilcoxon signed rank test; rs = 0.483; P < 0.005), and invasive carcinomas (P < 0.0001; Wilcoxon signed rank test; rs = 0.355; P < 0.0001) (Tables 1, 2; Figs. 1B–D, 3). EpoR immunostaining in DCIS and invasive carcinomas was significantly greater compared with EpoR immunostaining in benign epithelial cells both in nonsmokers (DCIS: n = 30 tumors; P < 0.0001; rs = 0.331; P < 0.05; invasive carcinoma: n = 41 tumors; P < 0.0001; rs = 0.420; P < 0.005; Wilcoxon signed rank test) and in smokers (DCIS: n = 35 tumors; P < 0.0001; rs = 0.608; P < 0.0001; invasive carcinomas: n = 61 tumors; P < 0.0001; rs = 0.494; P < 0.0001; Wilcoxon signed rank test). There was no difference in the level of EpoR immunostaining of benign epithelial cells in nonsmokers (median score, 100; mean ± standard error of the mean [SEM], 85.3 ± 9.9; n = 41 tumors) compared with smokers (median score, 100; mean ± SEM, 95.6 ± 6.6; n = 61 tumors; P > 0.05; Mann–Whitney test).

Table 1. Erythropoietin Receptor Expression in In Situ Mammary Carcinomas
In situ carcinomaNo.EpoR staining scoreEpoR DTSa
MedianMean ± SEMP valuebMedianMean ± SEMP valueb
  • EpoR: erythropoietin receptor; DTS: differential tumor score; SEM: standard error of the mean; DCIS: ductal carcinoma in situ; LCIS: lobular carcinoma in situ.

  • a

    EpoR DTS: EpoR staining score of tumor minus EpoR staining score of corresponding benign epithelium.

  • b

    Mann–Whitney or Kruskal–Wallis test.

Type       
 DCIS122200175.7 ± 6.8> 0.0510095.1 ± 6.30.0066
 LCIS36115164.5 ± 10.3 4058.4 ± 10.0
DCIS grade       
 Low2295149.0 ± 16.90.00416063.5 ± 13.10.0111
 Intermediate51105164.4 ± 10.310089.6 ± 9.8
 High49160204.1 ± 9.3100120.8 ± 9.8
Necrosis       
 Absent59115162.4 ± 9.10.0310089.5 ± 7.9> 0.05
 Present63140194.8 ± 9.1100112.3 ± 9.1
Table 2. Erythropoietin Receptor Expression in Invasive Mammary Carcinomas
IMCNo.EpoR staining scoreEpoR DTSa
MedianMean ± SEMP valuebMedianMean ± SEMP valueb
  • EpoR: erythropoietin receptor; DTS: differential tumor score; IMC: invasive mammary carcinoma; SEM: standard error of the mean; N/A: not available; LVI: lymphovascular invasion; ER: estrogen receptor; PR: progesterone receptor.

  • a

    EpoR DTS: EpoR staining score of tumor minus EpoR staining score of corresponding benign epithelium.

  • b

    Mann–Whitney or Kruskal–Wallis test.

Total184190172.5 ± 5.3N/A10093.0 ± 5.3N/A
Type       
 Ductal126185176.7 ± 6.5> 0.05100106.0 ± 5.9< 0.0001
 Mixed27200166.3 ± 15.210086.6 ± 17.4
 Lobular31200161.0 ± 10.34046.0 ± 9.4 
Grade       
 144155149.3 ± 10.60.0266063.9 ± 10.9.0003
 289200177.0 ± 7.410096.4 ± 6.9 
 351195183.6 ± 10.0120117.1 ± 8.3
Size       
 T191180159.2 ± 8.2> 0.059087.0 ± 7.6> 0.05
 T2–T493195178.0 ± 6.810095.6 ± 6.5
Tumor necrosis       
 Absent141180159.2 ± 8.20.0279087.0 ± 7.60.0123
 Present43200193.3 ± 10.3110113.9 ± 9.2
LVI       
 Absent119170161.3 ± 6.90.00698079.9 ± 6.50.0061
 Present65200189.8 ± 7.5110107.1 ± 7.4
Lymph node status       
 Negative95170160.8 ± 7.80.00758580.4 ± 7.6> 0.05
 Positive89200187.5 ± 7.0100101.6 ± 7.0
ER status       
 Negative75200185.7 ± 7.90.0375100108.3 ± 7.20.0115
 Positive109180161.3 ± 6.99080.4 ± 6.7
PR status       
 Negative77200185.9 ± 7.60.0404100101.4 ± 7.3> 0.05
 Positive107170161.1 ± 7.110089.2 ± 6.8
HER2/neu status       
 Negative116180175.3 ± 6.8> 0.0510093.2 ± 6.8> 0.05
 Positive68190164.6 ± 8.610093.4 ± 7.6
Figure 3.

Expression of erythropoietin receptor (EpoR) in benign mammary epithelial cells and invasive mammary carcinoma (IMC). Bars indicate the median values of EpoR immunostaining scores. Asterisks, P < 0.0001 (Wilcoxon signed rank test).

We found no significant difference in EpoR immunostaining based on the EpoR staining scores between either DCIS and LCIS or invasive ductal and lobular carcinomas and carcinomas with mixed ductal and lobular features (P > 0.05; Kruskal–Wallis test) (Table 2). Conversely, based on the differential tumor scores, a highly significant difference in EpoR expression was seen between ductal carcinomas and lobular carcinomas, with mixed tumors showing intermediate levels of EpoR staining (P < 0.0001; Kruskal–Wallis test) (Table 2). A similar difference was found in EpoR staining between DCIS and LCIS as well (P = 0.0066; Mann–Whitney test) (Table 1).

We found no correlation between EpoR immunostaining and tumor size (P > 0.05; Mann–Whitney test). In DCIS, a significant correlation was seen between EpoR staining and tumor grade and the presence of tumor necrosis (Table 1). Similarly, in invasive carcinomas, significantly increased EpoR immunostaining was seen in carcinomas with high histologic grade, in carcinomas that showed tumor necrosis and lymphovascular invasion, and in tumors that were associated with lymph node metastases (Table 2). EpoR staining also was significantly greater in breast tumors with negative hormone receptor expression compared with hormone receptor positive tumors (Table 2). No correlation was found between EpoR expression and HER2/neu overexpression by the tumors (Table 2).

Expression of Epo

In benign mammary epithelial cells, we found weak-to-moderate, granular, cytoplasmic staining for Epo in 169 of 184 specimens (91.8%). Epo staining was most pronounced at the apical part of epithelial cells in the lobules (Fig. 4A). We found prominent, strong Epo immunostaining in lobules that showed secretory changes (Fig. 4B). In benign epithelial lesions (hyperplasia without atypia, sclerosing adenosis, and papilloma), Epo staining was similar to the Epo staining seen in normal epithelial cells (not shown). Epo immunostaining was found in 112 of 122 DCIS tumors (91.8%), in 34 of 36 LCIS tumors (94.4%) (Table 3), and in 174 of 184 invasive breast carcinomas (94.6%) (Table 4). Epo staining was usually weak to moderate and heterogeneous; however, strong, prominent staining was present in viable tumor cells adjacent to necrotic areas (Fig. 5A,B) and at the infiltrating edge of carcinomas (Fig. 5C). Necrotic tissue showed no staining for Epo. Epo expression was similar in the in situ and invasive components of the tumors (r = 0.844; P < 0.0001; Spearman rank correlation test).

Figure 4.

Expression of erythropoietin (Epo) in benign mammary epithelium. (A) Epo immunostaining is seen in benign lobular epithelial cells. Note the more prominent staining at the apical part of the cells. (B) Prominent immunostaining for Epo is seen in a lobule that shows secretory change; immunohistochemical staining for Epo with hematoxylin counterstain). Original magnification ×400 (A); ×200 (B).

Table 3. Erythropoietin Expression in In Situ Mammary Carcinomas
In situ carcinomaNo.Epo staining scoreEpo DTSa
MedianMean ± SEMP valuebMedianMean ± SEMP valueb
  • Epo: erythropoietin; DTS: differential tumor score; SEM: standard error of the mean; DCIS: ductal carcinoma in situ; LCIS: lobular carcinoma in situ.

  • a

    Epo DTS: Epo staining score of tumor minus Epo staining score of corresponding benign epithelium.

  • b

    Mann–Whitney or Kruskal–Wallis test.

Type       
 DCIS1229090.3 ± 6.4> 0.05526.4 ± 4.9> 0.05
 LCIS3610097.9 ± 9.5025.7 ± 8.4
DCIS grade       
 Low226079.6 ± 15.7> 0.057.518.8 ± 11.1> 0.05
 Intermediate518091.5 ± 10.2024.3 ± 8.2
 High4910096.0 ± 9.917.533.4 ± 8.0
Necrosis       
 Absent599591.3 ± 9.4> 0.05527.1 ± 8.2> 0.05
 Present639092.8 ± 8.91027.4 ± 6.4
Table 4. Erythropoietin Expression in Invasive Mammary Carcinomas
IMCNo.Epo staining scoreEpo DTSa
MedianMean ± SEMP valuebMedianMean ± SEMP valueb
  • Epo: erythropoietin; DTS: differential tumor score; IMC: invasive mammary carcinoma; SEM: standard error of the mean; LVI: lymphovascular invasion; ER: estrogen receptor; PR: progesterone receptor.

  • a

    Epo DTS: Epo staining score of tumor minus Epo staining score of corresponding benign epithelium.

  • b

    Mann–Whitney or Kruskal–Wallis test.

Total1848080.8 ± 4.9N/A014.5 ± 3.6N/A
Type       
 Ductal1267078.3 ± 6.3> 0.05013.5 ± 4.5> 0.05
 Mixed278074.4 ± 11.9011.8 ± 10.2
 Lobular3110096.5 ± 9.9020.7 ± 7.5
Grade       
 1449088.2 ± 10.5> 0.0507.3 ± 5.9> 0.05
 2898081.7 ± 6.6213.7 ± 3.8
 3517073.5 ± 10.3020.3 ± 8.9
Size       
 T1913562.4 ± 12.3> 0.0506.2 ± 4.1> 0.05
 T2–T4937077.6 ± 6.9019.3 ± 5.6
Tumor necrosis       
 Absent1418081.7 ± 5.6> 0.05010.6 ± 3.3> 0.05
 Present437077.9 ± 10.7527.2 ± 10.1
LVI       
 Absent1198080.7 ± 6.3> 0.05011.1 ± 4.5> 0.05
 Present658080.7 ± 8.2017.0 ± 4.7
Lymph node status       
 Negative957580.8 ± 7.6> 0.05010.4 ± 4.60.0397
 Positive899084.1 ± 6.91022.6 ± 5.6
ER status       
 Negative758084.4 ± 7.9> 0.05022.8 ± 7.4> 0.05
 Positive1097077.4 ± 6.6010.7 ± 3.2
PR status       
 Negative778080.6 ± 7.7> 0.05013.6 ± 5.0> 0.05
 Positive1077080.4 ± 6.7016.9 ± 5.3
HER2/neu status       
 Negative1168087.8 ± 5.9> .05012.6 ± 4.5> 0.05
 Positive687568.6 ± 8.8 116.8 ± 5.2
Figure 5.

The expression of erythropoietin (Epo) in invasive mammary carcinoma. (A) Moderate immunostaining for Epo is seen in ductal carcinoma in situ. Note the strong Epo staining in viable tumor cells adjacent to the central necrotic area. (B) Strong immunostaining for Epo is seen in cells of invasive ductal carcinoma adjacent to a necrotic area (asterisk). Note that tumor cells away from the necrotic area show only weak Epo expression (lower part of photomicrograph. (C) Weak Epo immunostaining is seen in invasive ductal carcinoma with strong expression of the hormone in cells at the infiltrating edge of the tumor. (D) Prominent Epo immunostaining is seen in cells of invasive lobular carcinoma. The carcinoma cells infiltrate around a benign duct, which shows only weak staining for Epo; immunohistochemical staining for Epo with hematoxylin counterstain). Original magnification ×400 (A, B); ×200 (C); ×40 (D).

Compared with benign epithelial cells, Epo immunostaining was increased significantly in DCIS (P < 0.0001; Wilcoxon signed rank test; rs = 0.683; P < 0.0001), LCIS (P < 0.005; Wilcoxon signed rank test; rs = 0.477; P < 0.005), and invasive carcinomas (P < 0.0001; Wilcoxon signed rank test; rs = 0.731; P < 0.0001) (Fig. 5D). It is interesting to note that, when smokers (n = 61 patients) and nonsmokers (n = 41 patients) were analyzed separately, we found significantly increased Epo immunostaining in invasive carcinomas compared with benign epithelial cells only in nonsmokers (P = 0.0018; Wilcoxon signed rank test; rs = 0.816; P <0.0001) and not in smokers (P > 0.05; Wilcoxon signed rank test; rs = 0.703; P < 0.0001) (Fig. 6). Similarly, whereas Epo immunostaining was increased significantly in DCIS compared with benign epithelial cells in nonsmokers (P < 0.001; n = 30 specimens; Wilcoxon signed rank test; rs = 0.733; P < 0.0001), the difference was only marginally significant in smokers (P = 0.0444; n = 35 specimens; Wilcoxon signed rank test; rs = 0.687; P < 0.0001). Although Epo immunostaining was increased in benign epithelial cells in smokers (median score, 100; mean ± SEM, 90.2 ± 7.3; n = 61 patients) compared with nonsmokers (median score, 60; mean ± SEM, 79.0 ± 10.1; n = 41 patients), this did not reach statistical significance (P > 0.05; Mann–Whitney test).

Figure 6.

Expression of erythropoietin (Epo) in benign mammary epithelial cells and invasive mammary carcinoma (IMC) in nonsmokers (solid circles) and smokers (open circles). Bars indicate the median values of Epo immunostaining scores. Asterisks, P < 0.005; NS,: not significant (P > 0.05; Wilcoxon signed rank test).

Epo immunostaining was similar in invasive ductal and lobular carcinomas and in carcinomas with mixed ductal and lobular features (P > 0.05; Kruskal–Wallis test) (Table 4). Similarly, no difference was found in Epo immunostaining between DCIS and LCIS (P > 0.05; Mann–Whitney test). We found no correlation between Epo immunostaining and tumor size, tumor grade, presence of necrosis, lymphovascular invasion, lymph node status, hormone receptor status, or HER2/neu overexpression (P > 0.05; Kruskal–Wallis or Mann–Whitney test) (Table 4). Similarly, no correlation was found between Epo immunostaining and grade of DCIS or presence of comedo-type necrosis (P > 0.05; Kruskal–Wallis or Mann–Whitney test) (Table 3).

It is noteworthy that, when the analysis was done using the differential tumor scores for Epo, we found a significant correlation between increased Epo immunostaining in the tumors and the presence of lymph node metastases (P = 0.0397; Mann–Whitney test) (Table 4). When Epo immunostaining in DCIS and invasive carcinomas was compared with the adjacent benign epithelium in the same specimens, we found significantly greater Epo staining in intermediate grade tumors (P < 0.01) and high-grade tumors (P < 0.0001) only, and not in low-grade carcinomas (P > 0.05; Wilcoxon signed rank test).

DISCUSSION

We characterized the expression of Epo and EpoR by immunohistochemistry in a series of 158 in situ and 184 invasive breast carcinomas and adjacent benign mammary epithelium. We found weak-to-moderate, granular, cytoplasmic Epo and EpoR immunostaining in benign mammary epithelial cells in 91.8% and 95.6% of the specimens, respectively. This pattern of immunostaining is similar to that described previously in breast tissues19 and other tissues.17, 22, 48, 49 In normal ducts and lobules, Epo staining was most prominent in the luminal aspect of lobular epithelial cells. We found strong Epo immunostaining in lobules that showed secretory changes; however, EpoR staining was not increased in such secretory lobules. These findings are similar to those of Juul et al.,19 who recently also reported strong Epo immunoreactivity in mammary epithelial cells from lactating breast tissue, with less reactivity noted during gestation and in nonlactating breast. Those authors also found weak EpoR immunoreactivity in mammary epithelial cells regardless of lactational state. These findings and our results support the concept that Epo present in breast milk is synthesized in the lactating mammary gland epithelium. Although the physiologic role of Epo in breast is not yet clear, and no data are available currently to suggest a possible role for Epo in lactation, the presence of EpoR suggests a specific role for Epo signaling in the breast.19

We found significantly increased Epo immunostaining in both in situ and invasive carcinomas compared with benign ductal and lobular epithelial cells in nonsmokers, but not in smokers. The lack of a difference in Epo staining for smokers appeared to be due to an increased baseline level of Epo expression in smokers in benign epithelium. However, this difference did not reach statistical significance in our series. Conversely, EpoR immunostaining was increased significantly in carcinomas compared with benign epithelial cells both in nonsmokers and in smokers. The baseline level of EpoR expression in benign epithelial cells was similar for the two groups.

Epo immunostaining was heterogeneous in carcinomas, with most prominent expression in tumor cells adjacent to necrotic areas and at the infiltrating edge of the tumors, sites thought to be the most hypoxic parts of tumors.50–52 In contrast, EpoR expression was usually uniform and homogenous, although greater expression was present in tumor cells adjacent to necrotic areas, similar to the expression levels for Epo. These findings support our recent demonstration of hypoxic up-regulation of both Epo and EpoR in breast carcinoma cell lines.34

We found no difference in Epo or EpoR immunostaining between ductal and lobular carcinomas and carcinomas with mixed features. However, compared with benign epithelium present in the specimens, EpoR staining was significantly greater in ductal carcinomas compared with tumors of the lobular type. It is interesting to note that tumors with mixed features demonstrated EpoR immunostaining levels that were intermediate between the levels seen in ductal carcinomas and lobular carcinomas. Epo immunostaining did not show a correlation with any of the clinicohistopathologic features of tumors examined, including tumor size, grade, lymphovascular invasion, lymph node status, hormone receptor status, and HER2/neu expression. In previous studies, these clinicohistopathologic features also were not correlated with intratumoral hypoxia.53 It is noteworthy that Epo immunostaining did not show a statistical correlation with the presence of tumor necrosis despite the prominent Epo staining observed near necrotic areas. This finding likely is due to the fact that necrotic regions usually represented small areas of tumors, and the prominent immunostaining that was restricted to these areas did not increase the overall staining score significantly.

In contrast, we found significantly increased EpoR immunostaining in tumors that showed high combined histologic grade, tumor necrosis, lymphovascular invasion, lymph node metastases, and loss of hormone receptor expression, all features that are considered to be associated with an adverse prognosis.54 No correlation was seen between EpoR expression, tumor size, HER2/neu expression.

Until recently, the action of Epo was thought to be restricted to erythropoiesis.3 There is now evidence that certain nonerythroid cells express EpoR and respond to Epo in vitro and in vivo. The major mechanism by which Epo promotes erythropoiesis is by preventing the programmed cell death of erythroid precursors through up-regulation of the Bcl-2 and Bcl-XL pathways mediated by STAT-5.55, 56 There is also increasing evidence that EpoR expressed in neuronal tissues is functional21 and that Epo may have mechanisms of action in the central nervous system similar to those described in erythroid cells, such as decreasing apoptosis in neurons during normal brain development, or exerting a neuroprotective effect under adverse conditions, such as hypoxia.21, 27

EpoR mRNA also is expressed in endothelial cells, and Epo stimulates the proliferation and migration of human endothelial cells28, 29 and stimulates angiogenesis.30 Recent studies indicate that Epo signaling in endothelial cells is mediated through tyrosine phosphorylation of proteins, including phosphorylation of transcription factor STAT-5, which is similar to that in erythroid cells.57 It was suggested recently that a paracrine mechanism of Epo signaling may play a role in uterine angiogenesis.30 We have found that the vasculature of solid tumors, including breast carcinomas, also expresses EpoR, and a paracrine mechanism of Epo signaling also may play a role in the vascularization of these tumors.34 Our current finding of EpoR expression in tumor vasculature is consistent with this hypothesis.

An autocrine role for Epo in which the hormone contributes to the survival, proliferation, and differentiation of trophoblast cells22 may be analogous to its role in erythroleukemia,58 hepatocellular carcinoma,59 and renal cell carcinoma cells.33 We recently showed that Epo signaling stimulates the proliferation of human breast carcinoma cell lines, suggesting that an autocrine mechanism of Epo signaling also may play a role in the proliferation and hypoxic survival of breast tumors.34 Our current results also support these findings.

It has been demonstrated that Epo expression in the uterus and oviduct is regulated by estrogen in addition to hypoxia.30 It is noteworthy that, although the importance of an increased estrogen effect in breast carcinogenesis is well established,60 we found no correlation of Epo expression with the hormone receptor status of tumors. Conversely, we did find increased EpoR expression by tumors with negative hormone receptor status, a feature that is associated with poor differentiation and an adverse prognosis.

Regions of low oxygen and necrosis are common features of solid tumors.51, 53 Hypoxic conditions in vitro as well as in vivo result in elevated levels of hypoxia-inducible factor 1 (HIF-1), a transcription factor that, in turn, stimulates the expression of a number of genes important for tumor cell survival.61, 62 The overexpression of HIF-1 protein was found in 53% of all primary malignant tumors, including 29% of primary breast tumors and 69% of breast carcinoma metastases.63 In colonic adenocarcinoma, tumor cells at the leading edge of infiltrating carcinoma manifested the most intense HIF-1 expression63; this pattern of staining is similar to the expression of Epo we found in breast carcinomas, supporting the hypothesis that increased Epo expression in the tumors is stimulated by tissue hypoxia.

HIF-1 regulates the hypoxia-driven expression of several other genes in addition to Epo, including vascular endothelial growth factor, glycolytic enzymes, and glucose transporters.13, 64 Thus, it is unlikely that Epo signaling is the only determinant of hypoxic tumor cell survival. It should be noted, however, that, although Epo gene transcription is regulated by HIF-1, Epo mRNA stability, and (hence) translation, is also regulated by RNA-stabilizing mechanisms.65 Thus, elevated Epo protein expression also may result from distinct biochemical alterations in tumor cells.

It is believed that hypoxia mediates the selection of neoplastic cells with diminished apoptotic potential by providing a growth advantage to cells with genetic alterations that impair the process of apoptosis.66 This hypoxia-mediated clonal selection of tumor cells with diminished apoptotic potential has been suggested as an important biologic mechanism of tumor progression. The presence of hypoxia also may be involved in the development of a more aggressive phenotype and may contribute to metastasis67–71 and treatment resistance.72, 73 The potent antiapoptotic actions of Epo signaling may contribute to such effects of hypoxia.55, 74

Recent studies have suggested that both active smoking and passive smoking may be associated with an increased risk of breast carcinoma.36–38 Passive smokers (exposed to 2 hours of smoke for 25 years) had 3.2 times the risk of breast carcinoma compared with women with no such exposure, and active smokers (≥ 20 cigarettes per day) had 4.6 times the risk of breast carcinoma. Our findings of increased Epo immunostaining in the benign mammary epithelium of smokers are consistent with the observation of increased levels of Epo in the breast milk of smokers compared with nonsmokers.18 Cigarette smoking is a common cause of tissue hypoxia, and it is believed that it induces elevated circulating Epo levels in humans.35 Smokers, indeed, have higher serum Epo levels and hematocrits.35 Although it is believed that the major source of elevated circulating Epo in smokers is due to systemic hypoxia and is likely to come from the kidneys, the increased Epo immunostaining in benign mammary epithelium of smokers and in breast carcinomas suggests that increased expression of Epo probably is related to tissue and tumor hypoxia. EpoR-bearing cells, of course, can respond to locally or systemically elevated Epo levels. These findings raise the intriguing possibility that the high expression of EpoR in breast carcinoma cells may represent a mechanism that links smoking, tissue hypoxia, elevated Epo levels, and carcinoma. Clearly, further studies are needed to investigate this possibility.

Hypoxia also can induce EpoR expression,34 and we found increased EpoR staining near necrotic areas of tumors. However, hypoxic regions usually are distributed heterogeneously within tumors.53 Although the heterogeneous pattern of Epo immunostaining is consistent with this phenomenon, the markedly increased expression of EpoR throughout breast carcinoma tissue suggests that it may not be attributed solely to hypoxia. Furthermore, HIF-1 is not known as a regulator of EpoR gene expression. Thus, the biochemical mechanisms underlying elevated EpoR expression in breast carcinoma and its hypoxic enhancement remain unknown. Elevated EpoR expression may increase the sensitivity of the neoplastic cells to available Epo. Elevated systemic Epo levels, thus, may indicate potentially adverse outcomes for patients with breast carcinoma, as proposed for patients with renal carcinoma.33 The routine use of rhEpo in patients with malignancies for the treatment of anemia resulting from chemotherapy, therefore, may need to be re-evaluated for its potential adverse effects. Further studies are needed to examine these hypotheses.

In summary, we have demonstrated that benign mammary epithelial cells and cells of in situ and invasive breast carcinomas express both Epo and EpoR. Increased immunostaining of EpoR in breast carcinomas showed a positive correlation with clinicohistopathologic features suggestive of an adverse prognosis. Our findings suggest that increased Epo signaling may represent a novel mechanism by which hypoxia promotes a malignant phenotype of breast carcinoma.

Ancillary