Mammaglobin vs GCDFP-15: An Immunohistologic Validation Survey for Sensitivity and Specificity

Rohit Bhargava, MD1; Sushil Beriwal, MD2; David J. Dabbs, MD1


Am J Clin Pathol. 2007;127(1):103-113. 

In This Article


Although the mammaglobin gene was discovered almost a decade ago, only a limited number of studies have discussed its clinical usefulness. In the seminal paper by Watson and Fleming,[2] mammaglobin overexpression was identified in breast carcinoma compared with normal breast tissue. Recently, mammaglobin has been identified as a breast cancer–specific gene, and its usefulness as a novel breast cancer marker has been described.[4,8,9] Nunez-Villar et al[10] suggested that elevated mammaglobin (h-mam) expression in breast cancer is associated with clinical and biologic features defining a less aggressive tumor phenotype.

A few other reports have discussed the importance of mammaglobin expression in breast tumors. Ciampa et al[11] studied mammaglobin and CrxA-01 expression in cell block material from malignant pleural effusions. Eighty percent of breast carcinomas were positive for mammaglobin and/or CrxA-01, and none of the nonbreast carcinomas were positive for mammaglobin. Because this study was performed on metastatic tumors in pleural fluid, the numbers of nonlung and nonbreast carcinomas were rather insignificant. Han et al[12] studied mammaglobin and BRST-2 (same as GCDFP-15) expression by immunohistochemical analysis in breast and nonbreast carcinomas. They concluded that mammaglobin has a superior sensitivity to that of BRST-2. Among the nonbreast carcinomas only 1 of 10 urothelial neoplasms and 1 of 10 thyroid carcinomas showed strong reactivity. Our results are very comparable to those of Han et al[12] with respect to breast carcinoma; however, because we analyzed a large number of nonbreast carcinomas, we identified some clinically significant differences.

We compared the immunohistochemical staining pattern of mammaglobin with GCDFP-15 (a sensitive and specific marker of breast carcinoma) in whole tissue sections of 29 primary mammary carcinomas with matched lymph node metastases and 63 randomly selected breast carcinomas represented on a TMA. Higher sensitivity was observed in whole tissue sections (71%) compared with TMAs (41%). Our study shows that lobular carcinomas are more diffusely and strongly stained with mammaglobin compared with ductal carcinomas, which show a more patchy staining pattern. The whole tissue section cases were an equal mix of ductal and lobular carcinomas, whereas the TMA cases were randomly selected and contained predominantly ductal carcinomas. This difference in the case types explains the lower sensitivity of mammaglobin in the breast TMA group than in the whole tissue section group. However, in either group (whole section or TMA), mammaglobin had a higher sensitivity than GCDFP-15. Moreover, the mammaglobin antibody cocktail stained the breast carcinomas more intensely than GCDFP-15, and, among the positive cases, the number of cells stained with mammaglobin is higher than with GCDFP-15.

Previous studies have suggested that mammaglobin expression is mainly seen in well-differentiated, receptor-positive breast carcinomas; however, we failed to show any correlation between mammaglobin expression and different clinicopathologic variables ( Table 5 ). Although many carcinomas would not be included in the differential diagnosis of breast carcinoma, the specificity of the mammaglobin antibody in our study set was 92%.

Among the nonbreast carcinoma group, approximately 40% of the endometrial endometrioid carcinoma showed significant staining. Our findings of mammaglobin expression in endometrial carcinoma are similar to those recently reported by Zafrakas et al.[13] These authors studied more than 300 human tumors and matched normal tissue samples by different techniques and identified significant mammaglobin expression in breast and gynecologic tissues and the absence of mammaglobin expression in prostate, kidney, colon, rectum, small intestine, stomach, pancreas, lung, and thyroid. In this regard, it is important to recognize that mammaglobin is highly related to uteroglobin, a secretory protein of the endometrium induced by progesterone.[14] This probably explains strong staining of endometrial endometrioid carcinomas. Therefore, it is conceivable that other tumors of the reproductive tract or tumors showing progesterone receptor expression should also show mammaglobin immunoreactivity. However, only 1 of 40 ovarian carcinomas included in our study showed significant positivity, but it is important to mention that our set of ovarian carcinomas lacked endometrioid-type cancers.

All invasive endocervical adenocarcinomas studied by whole tissue section immunohistochemical analysis were all negative or showed equivocal staining. In contrast, AIS of the cervix showed significant staining in 55% of cases, although most atypical glands were patchily stained compared with mostly diffuse staining in normal endocervical glands. A review of endocervical cases revealed that mammaglobin staining is mostly present admixed with cytoplasmic mucin. Some of the glands partially involved by AIS showed mammaglobin staining only within the normal mucinous portion of the gland Figure 5. Similarly, in endometrioid carcinomas, mammaglobin staining is often identified within glandular secretions. This staining pattern suggests that mammaglobin is involved in some process of the cell secretory mechanism. This reduced mammaglobin expression in AIS and invasive endocervical carcinoma is worth studying in more detail.

Adenocarcinoma in situ of cervix (A, H&E, ×200) showing reduced mammaglobin expression (B, ×200). Note that intense mammaglobin staining is seen only in a normal mucin-containing gland, and the portion of gland involved by adenocarcinoma in situ demonstrates reduced mammaglobin expression.

Another interesting finding in our study was the significant staining seen in approximately 6% of melanomas. Although it is important to recognize this pitfall, this is unlikely to cause a problem in the differential diagnosis from a breast carcinoma because a panel of immunohistochemical stains is always used in this scenario. Positive mammaglobin staining was also observed in a significant proportion of sweat gland carcinomas. This is not surprising because sweat gland carcinomas with ductal differentiation can also exhibit GCDFP-15 and estrogen receptors.[15] Mammaglobin does not seem to be a useful stain to distinguish breast from sweat gland carcinomas. Positive mammaglobin staining seen in salivary gland tumors was also expected; however, intense staining was not seen in any of the positive salivary gland tumors.

Despite some nonspecificity of the mammaglobin antibody, our data provide compelling evidence for inclusion of mammaglobin in a panel for the workup of carcinoma of an unknown primary site. For diagnosing a breast carcinoma, the sensitivity of mammaglobin is better than that of GCDFP-15 based on our present study and as reported by others.[16,17]

The mammaglobin antibody is a sensitive marker of breast carcinomas. Mammaglobin expression is not altered at the metastatic (lymph node) site. With respect to endometrial carcinoma, it may have a role in the differential diagnosis with an invasive endocervical adenocarcinoma. Mammaglobin antibody can occasionally stain a melanoma, and it is important to recognize this pitfall. Mammaglobin can help, in combination with other markers, to establish the correct diagnosis of metastatic breast carcinoma but per se does not seem to be specific.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.