OUP user menu

Metaplastic and Medullary Mammary Carcinomas Do Not Express Mammaglobin

Carolina Reyes MD, Carmen Gomez-Fernández MD, Mehrdad Nadji MD
DOI: http://dx.doi.org/10.1309/AJCP5W5SEZSEHUHE 747-752 First published online: 1 May 2012


Mammaglobin A (MG-A) is purportedly useful for detecting metastatic carcinomas suspected to be of breast origin and has been advocated as a useful marker of micrometastasis in sentinel lymph nodes and minimal residual tumor in bone marrow. Little is known about its expression frequency in histologic subtypes of breast cancer.

Excisional biopsy specimens from 1,079 untreated invasive mammary carcinomas were evaluated for immunohistochemical expression of MG-A. In addition to estrogen (ER) and progesterone receptors (PR) and HER2, staining for p63 and HLA-DR was used to further characterize histologic subtypes.

Of the carcinomas, 36 were classified as metaplastic (based on morphologic features, ER−/PR−/HER2−, p63+), 38 as medullary (ER−/PR−/HER2−, HLA-DR+), and 1,005 as ductal, no special type (NST). All metaplastic and medullary carcinomas were negative for MG-A. Of 1,005 ductal carcinomas, NST, 492 (49.0%) were MG-A+, 62.0% with a reaction in fewer than 25% of the cells.

MG-A immunohistochemical studies failed to detect all medullary and metaplastic cancers and more than 50% of ductal carcinomas, NST. In two thirds of MG-A+ ductal carcinomas, the reaction was only focal and usually in a minority of cells. These findings suggest that MG-A has limited value in identifying the mammary origin of carcinomas, particularly in small biopsy specimens used to detect metastasis or minimal residual disease.

Key Words:
  • Mammaglobin
  • Metaplastic mammary carcinoma
  • Medullary mammary carcinoma
  • Neoplasm metastasis
  • Micrometastasis
  • Minimal residual disease detection

Detection of breast cancer in metastatic sites, including micrometastases in lymph nodes or minimal residual tumor in bone marrow, has gained major importance as a guide to early therapeutic intervention. Various biomolecular markers have been proposed for the detection of breast cancer cells in meta-static sites and in peripheral blood.1 However, the expression of most of these markers is not limited to breast tissue,2 and, thus, the search for truly specific mammary epithelial cell markers has been ongoing. Recent findings point to a number of genes that are preferentially expressed in breast tissue. These include mammaglobin A (MG-A), mammaglobin B (MG-B), and lipophilin B.36 Of these potential mammary cell markers, MG-A has been the most extensively studied.

In 1996, Watson and Fleming3 used reverse transcription–polymerase chain reaction and Northern blot analysis to identify a gene that seemed to be expressed only in breast epithelial cells and overexpressed in breast cancer. This gene was named mammaglobin. Subsequently, Becker et al4 identified a gene highly homologous to mammaglobin and named it mammaglobin B; the original mammaglobin was named mammaglobin A. MG-A and MG-B are members of the secretoglobin superfamily, a group of small secretory dimeric proteins that are mainly expressed in mucosal tissues.7 Although MG-B was similarly proposed to be a marker for detection of breast cancer micrometastases, it was later shown to be expressed by a number of other cancers and, hence, lacked specificity for mammary tumors.8,9

The MG-A gene encodes a 93-amino acid protein with a molecular mass of 10.5 kDa.3 In breast tissue, it exists in 2 forms with molecular masses of 18 and 25 kDa.7 Although MG-A was originally thought to be confined to normal and neoplastic mammary epithelium, it was later shown that it is also expressed by a number of cells and tissues other than breast.3,1013

Expression of MG-A at the RNA and protein levels has been reported in 70% to 80% of primary and metastatic breast cancers.1416 These studies, however, have assessed MG-A expression mainly in ductal and lobular subtypes of breast carcinomas. To our knowledge, there are no studies that have analyzed the immunohistochemical expression of MG-A in histologic subtypes of mammary carcinomas, including metaplastic and medullary types. The purpose of this study was to investigate the immunohistochemical expression of MG-A in ductal and in immunophenotypically proven medullary and metaplastic mammary carcinomas.

Materials and Methods

Excisional biopsy specimens from 1,079 consecutive cases of untreated invasive breast carcinomas from the files of the Department of Pathology, University of Miami/Jackson Memorial Medical Center and University of Miami/Sylvester Cancer Center, Miami, FL, were evaluated for the immunohistochemical expression of MG-A. All specimens were previously analyzed for estrogen receptor (ER), progesterone receptor (PR), and HER2. We specifically enriched the cohort with mammary carcinomas that, based on morphologic assessment and/or immunohistochemical studies, were suspected or proven to be of metaplastic and medullary phenotypes.

The diagnosis of invasive mammary carcinoma was confirmed in all cases. In the cases in which a metaplastic or medullary phenotype was morphologically suspected, the following immunohistochemical markers were used for confirmation: p63 for metaplastic carcinoma and HLA-DR for medullary carcinoma.

View this table:
Table 1
View this table:
Table 2

We used 4-μm-thick, formalin-fixed, paraffin-embedded sections for MG-A, p63, and HLA-DR immunohistochemical studies following heat-induced antigen retrieval. Table 1 shows the clones and sources of reagents. Normal breast tissue was used as a positive control sample for MG-A and p63. Normal tonsil served as the positive control sample for HLA-DR.

Three pathologists (C.R., C.G.F., and M.N.) independently evaluated the staining results for MG-A and semiquantitatively scored them in the following manner: 1+, 1% to 10% of positive cells; 2+, 11% to 25% of positive cells; and 3+, more than 25% of positive cells.

For metaplastic carcinomas, p63 immunoreaction was seen as dense nuclear staining. In medullary carcinomas, HLA-DR positivity was observed as a cytoplasmic membrane reaction.


Based on histomorphologic features and immunophenotype, 1,005 tumors were classified as invasive ductal carcinoma, no special type (NST). Of the cases, 36 were classified as metaplastic carcinomas and 38 were true medullary carcinomas with pushing borders. All metaplastic and medullary carcinomas were triple-negative for ER, PR, and HER2.

Table 2 summarizes the results of MG-A staining in the histologic types of breast cancer. Briefly, of the 1,005 invasive ductal carcinomas, NST, 492 (49.0%) were positive for MG-A. Among the 492 MG-A+ ductal carcinomas, staining was 3+ in 187 (38.0%) and 2+ in 103 (20.9%) of the cases Image 1. The remaining 202 cases (41.1%) exhibited only a scattered 1+ reaction in a few cells Image 2. Most of the ductal carcinomas that showed a positive reaction for MG-A (425 [86.4%]) were grade 1 and 2 Table 3. The remaining tumors were grade 3 (67 [13.6%]). All metaplastic and medullary carcinomas were negative for MG-A Image 3 and Image 4.


Although MG-A was originally considered a specific marker for benign and malignant breast epithelial cells,17 subsequent studies showed its expression in various nonmammary tissues, including normal cervix, ovary, uterus, skin, sweat glands, kidney, testis, and salivary glands, as well as in tumors of lung, ovaries, and sweat glands.13,14,18 In addition to its lack of specificity, the sensitivity of detection of MG-A in mammary carcinomas was also reported from less than 50% up to 84%. The overall rate of expression of MG-A in our study was 45.6% (492/1,079). Specifically, in ductal carcinomas, NST, the rate was 49.0% (492/1,005). This is lower than the 82% rate reported by Watson et al12 and 84% by Han et al.17 Our results however, agree with the findings of a study by Eiichi et al,19 who found positive MG-A in 49% of 214 ductal carcinomas.

Image 1

Infiltrating ductal carcinoma of no special type (A, H&E, ×50). Positive immunohistochemical reaction for mammaglobin is seen in about 25% of tumor cells (B, ×50).

Image 2

Infiltrating ductal carcinoma of no special type (A, H&E, ×50). Positive immunohistochemical reaction for mammaglobin is seen in fewer than 5% of tumor cells (B, ×50).

There has been no previous study that specifically focused on the immunohistochemical reaction of MG-A in subtypes of mammary carcinoma. Eiichi et al19 reviewed metastatic mammary carcinomas to the brain for the presence of MG-A. Their cohort included 1 case of medullary and 2 cases of metaplastic mammary carcinomas; all 3 were negative for MG-A. One case of metaplastic carcinoma in the series reported by Bhargava et al20 also did not stain for MG-A.

Metaplastic and medullary types of breast cancer are typically negative for ER, PR, and HER2 and, hence, fall into the category of triple-negative mammary carcinomas. As a group, the triple-negative cancers have a poor prognosis but comprise a heterogeneous histomorphologic spectrum. On the basis of gene expression profiles, the majority of triple-negative breast cancers show a basal-like genotype.21,22 Traditionally, basal/myoepithelial cytokeratins have been advocated as surrogate immunohistochemical markers for the basal-like phenotypes.2325 Other studies, however, have questioned the use of molecular profiling as a “gold standard” in general (reviewed by Gusterson26) and the use of basal cytokeratin immunohistochemical studies for identification of basal-like phenotypes in particular.27,28 In this study, we therefore chose not to use the controversial term “basal-like” and instead concentrated on the 2 morphologically distinct types of triple-negative breast cancers, metaplastic and medullary, as they could to some extent, be objectively characterized by histologic and immunohistochemical studies.

View this table:
Table 3
Image 3

Infiltrating mammary carcinoma with a metaplastic (squamous) phenotype (A, H&E, ×50). Tumor cells show positive nuclear staining for p63 (B, ×50). There is no reaction for mammaglobin (C, ×50).

Metaplastic mammary carcinomas are a morphologically heterogeneous group of breast cancers characterized by lack of expression of ER, PR, and HER2 and expression of purported basal/myoepithelial markers. They show a wide spectrum of histomorphologic types, from the classic squamous cell carcinomas to matrix-forming carcinomas, spindle cell carcinomas, and carcinosarcomas.2931 Most, particularly the squamous and spindle cell types, express p63.32,33 None of the p63+ metaplastic carcinomas in our study expressed MG-A. Conversely, none of the ductal carcinomas were positive for p63.

True medullary carcinomas are characterized by syncytial growth patterns composed of large cells with pleomorphic nuclei and large nucleoli. These tumors are, by definition, fairly well circumscribed with a prominent lymphocytic infiltrate.34 Nevertheless, because objective histologic distinction between true medullary and atypical medullary carcinomas of the breast are not universally agreed on, we chose only the tumors that had pushing borders and HLA-DR positivity. It has been shown that most classic medullary carcinomas express HLA-DR, a component of the MHC class II antigenic system.3436 The expression of this marker has also been demonstrated in undifferentiated nasopharyngeal carcinomas and in lymphoepithelioma-like carcinomas of other organs such as skin, gastrointestinal tract, and urinary bladder. Parenthetically, it is not unreasonable, therefore, to consider true medullary carcinomas as lymphoepithelioma-like carcinomas of the breast. None of the 38 true medullary carcinomas showed reactivity for MG-A by immunohistochemical analysis.

Image 4

Infiltrating mammary carcinoma with a medullary phenotype (A, H&E, ×50). Tumor cells are positive for HLA-DR, as are the host lymphocytes (B, ×50). There is no reaction for mammaglobin (C, ×50).

Our study also shows that the expression of MG-A in the group of ductal carcinomas, NST, is focal and only present in fewer than 25% of the cells in approximately two thirds of the cases. This will limit the use of MG-A for detecting the breast origin of metastatic tumors in small biopsy specimens.

The correlation of MG-A positivity with a lower histologic grade of breast cancer has been reported before. Span et al37 demonstrated that MG-A expression is associated with low grade and steroid receptor–positive breast tumors. Our study supports that finding. Of the 49.0% ductal carcinomas that showed a positive reaction for mammaglobin, 86.4% were grades 1 and 2.

We conclude that MG-A is not expressed by metaplastic and medullary cancers of the breast. Furthermore, because this marker is neither completely specific nor sensitive for ductal carcinomas and, when present, is expressed only focally, it should be used only in conjunction with other putative breast markers for the determination of breast origin of metastatic carcinomas. This is especially important when detection of micrometastasis or minimal residual disease in patients with history of mammary cancer is intended.


  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. 28.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
View Abstract